You are on page 1of 237

TRANSACTIONS

OF THE

SOCIETY
PHILOSOPHICAL
AMERICAN
HELD AT PHILADELPHIA

FOR PROMOTING USEFUL KNOWLEDGE

NEW SERIES-VOLUME XXXII, PART III

RESEARCHON EARLY MAN


IN BURMA
HELLMUT DE TERRA and HALLAM L. MOVIUS, JR.

WITH

Supplementary Reports upon the


Pleistocene Vertebrates and Mollusks of the Region
EDWIN H. COLBERT and J. BEQUAERT

AND

Pleistocene Geology and Early Man in Java


HELLMUT DE TERRA

PHILADELPHIA

THE AMERICAN PHILOSOPHICAL SOCIETY

INDEPENDENCE SQUARE

1943
RESULTSOF THE
AMERICAN SOUTHEASTASIATIC EXPEDITION FOR
EARLY MAN

by
Sponsored
THE ACADEMY OF NATURAL SCIENCES OF PHILADELPHIA

THE PEABODY MUSEUM OF HARVARD UNIVERSITY

withassistance
from
THE AMERICAN PHILOSOPHICAL SOCIETY

THE CARNEGIE INSTITUTION OF WASHINGTON

THE MUSEUMOF COMPARATIVE


ZOOLOGY,HARVARDUNIVERSITY
CONTENTS
IPAGE PART II. TIIE STONEAGE OF BURMA.
Introduction........................................ 267 By HALLAML. MOVIUS, JR.
PAGE
Introduction................... .................... 341
PART I. THE PLEISTOCENE
OF BURMA.
By HELLMUTDE TERRA Acknowledgments ..................... ... ........... 342
Geographical Elements of the Region The Palaeolithic Period
Physiography. .................................... 271
The Terraces of the Ancestral Irrawaddy............. 271 Stratigraphy .............. .......................... 343
274 Summary of the Terrace Sequence ................. 343
Primary Land Surface ............................. The Archaeological Succession...................... 346
The Shan Highlands. ............................. 274
The Sites. ......................................... 347
Climatology ........................................ 275
Meteorological Data .............................. 275 Early Anyathian 1................................ 347
Early Anyathian 2................................ 347
Weathering and Soils. ............................. 277
D rainage ........................................... 278 Early Anyathian 3................................ 347
Late Anyathian 1................................. 348
Geological Observations in the Irrawaddy Basin Late Anyathian 2......... ........................ 348
The Irrawaddy Series................................ 280 Typology . ......................................... 348
Definition and Stratigraphy ....................... 280 Raw Material. ................................... 349
The Problem of the Plio-Pleistocene Boundary........ 280 Nom enclature .................................... 350
The Early Anyathian Culture ........................ 353
Sedimentary Characteristics . ...................... 284
Upper Irrawaddian Soils........................... 286 Early Anyathian 1 ............................... 353
Palaeontology .................................... 287 Early Anyathian 2-3 Implements of Fossil Wood..... 353
The Irrawaddy Terraces and Associated Soils........... 289 Early Anyathian 3 Implements of Silicified Tuff...... 362
Previous W ork. ................................... 289 The Late Anyathian Culture ........................ 372
General Aspects. .................................. 290 Late Anyathian 1................................. 372
Terrace Sections.................................. 292 Late Anyathian 2 .............. ................... 372
Soils Associated with the Terraces.... ............... 308 Summary and Conclusions ............ ............. 374
Extra-terrace Soils in the Irrawaddy Basin .......... 310
Remarks on the Soils of the Dry Belt................ 312 The Neolithic Period
Introduction ........................................ 378
Observations in the Northern Shan Highlands The Archaeological Material ......................... 380
Tilted Basin Formations ............................ 313 Kyaukpadaung . ................................. 380
The Structural Origin of the Basins................. 313 M agwe .......................................... 382
The Superposition of Old and Young Pleistocene For- Yenangyaung.................................... 382
m ations........................................ 315 Chauk........... ................................ 384
Pleistocene Fans and Terraces ....................... 315 Pagan and Nyaungu .............................. 386
The Boulder Gravels and Terraces of the Namtu and M inbu........................................... 386
Salween Rivers................................. 316 Summary and Conclusions ......................... 386
The Climatic Nature of the Terraces in the Shan
Highlands...... .. ............................. 318 Reconnaissance in the Southern Shan States
The Pleistocene in the Shan Karst Region.............. 320 Introduction ........................................ 387
General Aspect of the Karst Relief.................. 320 Localities Investigated.............................. 389
The Mogok Karst and its Pleistocene Filling......... 321 Mongta-W a-Ku ................................... 389
Caves and Fossils in the Local Tradition........... 321 Tin-Ain .......................................... 389
Fossil-bearing Caves.. ........................... 323 Ahyu Taung ..................................... 390
Sinks and Fissures .............................. 325 Pa-leng ................... ..................... 390
The Physiographic Cycle of the Karst............... 325 H tam sang ...................................... 390
Sedimentation and Correlations..................... 328 M ong Paw n ..................................... 390
Basin Deposits ................................ 328 Tongt .......................................... 391
The Origin of the Karst Gravel and Loam.......... 329 Kengtung ........................................ 391
Summary and Conclusions ........................ 391
The Pleistocene History of Burma in Relation to that of
Neighboring Regions Bibliography of Parts I and II ...................... 392
Comparisons with India .............................. 330
Comparisons with China . ............................ 330
Comparisons with Malaya and Java................... 332 SUPPLEMENTARY REPORTS
The Cyclic Nature of Pleistocene Stratigraphy
The Origin of River Terraces ........................ 333 PART III. VERTEBRATES
PLEISTOCENE COLLECTED
IN BURMA
BY THE AMERICAN SOUTHEAST ASIATIC EXPEDITION FOR
The Glacial Cycle in Asia and Pleistocene Stratigraphy... 335
EARLY MAN. By EDWIN H. COLBERT
The Influence of Climate on Sedimentation........... 336
The Climatic Conditioning of Soils During the Pleisto- Introduction ................................... . 395
cene ................... ............ 337
The Influence of Climate on Terrace Formation....... 338 The Pleistocene Fossil Beds of Burma .................... 395
265
266 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
PAGE PART IV. FRESH-WATER SHELLS FROM CAVE DEPOSITS IN THE

Collecting Localities in the Pleistocene of Burma ............ 398 SOUTHERNSHAN STATES. By J. BEQUAERT
Lower Pleistocene ................................... 398 PAGE
M iddle Pleistocene .................................. 401 Localities ............................................ 431
Fossils in situ in Terrace T3 ........................ 401 D escription of Shells ................................... 432
Fossils from M ogok Fissures ........................ 401
402 Bibliography of Part IV ............................... 436
Post-Pleistocene .....................................
Post-Terrace Fossils ............................... 402
"Fossils" found in Superficial Cave Deposits .......... 402 PART V. PLEISTOCENEGEOLOGYAND EARLY MAN IN JAVA.
K eng T ung ......................................... 402 By HELLMUTDE TERRA
Fossils of Unknown Age ........................... 402 Review of the Discoveries .............................. 437
The Pleistocene Vertebrate Faunas of Burma Summary of the Data Bearing on the Investigations in
Central Java ....................................... 439
The Upper Irrawaddy Fauna as Found in the Upper Irra-
w addy Beds ...................................... 402 The Beginning of the Pleistocene in Java ................. 439
The Middle Pleistocene Fauna of Burma as Found in the The Lower Pleistocene Site of Modjokerto in Eastern Java. . 441
Caves at M ogok .................................. 417 The Middle Pleistocene Sites of Pithecanthropus in Central
The Middle to Upper Pleistocene Mammals of Burma as Jav a ............................................. 443
Found in situ in the Irrawaddy River Terraces ....... 421 The Solo Plain in the Vicinity of Sangiran .............. 443
Post-Pleistocene Fossils The Geology of the Sangiran Site ..................... 445
The Madioen Plain in the Vicinity of the Trinil Site .... 447
Mogok Surface Deposits in Caves .................... 422 The Geology of the Trinil Site ........................ 447
Post-terrace Deposits ............................... 422 The Age of Pithecanthropus and the Glacial Cycle ....... 450
Fossils of Unknown Age ............................... 423 The Upper Pleistocene Terraces and the Age of Solo Man
The Geology of the Ngandong Site .................... 455
Relationships and Correlation of the Pleistocene Mammalian The Stratigraphic Position of the Palaeolithic Cultures of
Faunas of Burma
Java
Relationships of the Upper Irrawaddy Fauna ........... 423 The Sangiran Industry ......... ...................... 456
Relationships of the Mogok Fauna .................... 424 The Ngandong Bone Industry ........................ 457
Relationships of the Terrace Fossils ................... 425 The Early Palaeolithic of Patjitan ..................... 457
The Correlation of the Pleistocene Mammalian Faunas Java and the Question of Land-bridges .................. 459
of B urma . ....................................... 425
Summary of Prehistoric Migrations to Java .............. 462
Bibliography of Part III ............................... 428 Bibliography of Part V ................................ 463
INTRODUCTION
BY HELLMUT DE TERRA

PROGRAM OF STUDY another source. This was supplied by the Peabody


Museum of Harvard University, which offered not only
At the International Symposium on Early Man, to share the
which was held under the auspices of the Academy of to expenses of the geological party, but also
delegate at their expense Dr. Hallam L. Movius, Jr.
Natural Sciences of Philadelphia in March, 1937, a as
discussion was held on Fossil Man in Asia. It was from archaeologist to the expedition. These grants, one
the Penrose Fund of the American Philosophical
felt at that time that the geological chronology of Early
Society, and the others from the Peabody Museum and
Man, which had previously been worked out in India the Museum of
Comparative Zoology of Harvard Uni-
under my direction, required further elucidation with
versity, as well as from the Carnegie Institution of
special reference to an eastward extension into China Washington, are gratefully acknowledged.
and Java, so that the new data on human origins might The project could not have been realized without
be integrated. The greatest obstacle to such work was leave of
absence, both from the office of the Academy
the lack of data, both stratigraphical and archaeological, of Natural Sciences and from the
Carnegie Institution
in the regions lying between India and China on one of
Washington, with which the expedition leader was
hand and India and Java on the other. It was thought associated. It
gives me great pleasure to express my
that Burma might fill this gap because of its inter- sincerest thanks to Mr. Charles M. B. Cadwalader and
mediate geographical position, and also because of the to Dr.
John C. Merriam for their helpful consideration.
recent finds of Stone Age tools and of Pleistocene ter- A
special grant from the Carnegie Corporation of New
race formations in the Irrawaddy Valley. York through the Carnegie Institution enabled me to
Accordingly a plan was worked out by Pere Teilhard work out the results of this expedition in the course
de Chardin and myself for an expedition to Burma and of 1939-40. An additional
grant was made to cover
Java. Its chief objectives would be: (1) to establish expenses for illustrations in Part I and Appendix IV
a chronology for the alluvial formations of the Irra- of this volume, which is herewith
acknowledged with
waddy Valley, and (2) to collect new data on the Stone many thanks. I feel particularly indebted to Dr. Van-
Age of this tropical land. Like India, Burma had long nevar Bush, President of the Carnegie Institution of
been known for its prehistoric antiquities. Numerous
Washington, whose unfailing support made the prep-
finds of artifacts had been recorded by former members aration of
my part of this manuscript possible.
of the Indian Geological Survey, particularly Noetling, As on a previous occasion, Pere Teilhard de Chardin
Coggin Brown, Cotter and Pascoe. More recently has given generously of his time and energy by joining
AMr.T. 0. Morris, Geologist of Steel Bros., Ltd., had this party for a duration of five months. It is difficult
reported on finds of "Lower Palaeolithic hand-axes" to express in words the admiration and gratitude which
and the occurrence of Pleistocene terrace formations. the expedition members entertain toward their distin-
Hence Burma appeared to be promising in more than guished associate and friend whose encouragement and
one regard, in fact it was to be expected, on the basis vision has been felt as a never-failing source of inspira-
of the literature, that there we would encounter geo- tion. He placed a brief summary of his observations at
logical conditions similar to those which had made the disposal of the expedition leader which he felt should
possible a detailed analysis of the background of Stone be incorporated in any form in this volume. This was
Age man in India. done in the first part where some of the sections are to
be regarded as the joint outcome of our co-operation
ACKNOWLEDGMENTS in the field.
Thanks are also due a number of colleagues both in
The expedition plan when submitted to Mr. Charles this country and abroad who have co-operated in vari-
M. B. Cadwalader, President of the Academy of Nat- ous ways. All the expedition members remember with
ural Sciences of Philadelphia, was received with in- pleasure the aid they received from Dr. M. A. Heron,
terest and approval. Through his sympathetic sup- Director of the Geological Survey of India, as well as
port and that of Dr. E. B. Howard, then Curator of Mr. E. L. C. Clegg and Mr. E. J. Bradshaw of the
the Department of Palaeontology and Geology, appli- Burma Division, who contributed to our undertaking
cation for financial aid was submitted to the American by furnishing information on the region and introduc-
Philosophical Society at Philadelphia. Through its tions to local officials. The Government of Burma most
generosity a grant was made from the Penrose Fund generously permitted travel in frontier regions; their
on the condition that a similar grant be secured from liberal attitude towards our undertaking made it pos-
267
268 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
sible for us to obtain the material which we were after. sections as we could visit within the limited time at our
Various members of the Burmah Oil Company gave disposal. This survey of Pleistocene terraces and older
helpful advice and permitted the use of their bungalows. beds extended from Nyaungu, near Pagan, south to Sale
The co-operation of Dr. Edwin H. Colbert, Associate and east to the foot of the Shan Highlands, south of
Curator, Department of Geology and Paleontology, Mandalay. We made daily excursions by car, there-
Academy of Natural Sciences of Philadelphia, in study- fore we were able to cover most of this territory and
ing the vertebrate fossils secured deserves special see the majority of exposures. Favorable weather
recognition. helped our investigations so that not a single day was
lost. Meanwhile the archaeological party combed the
Itinerary of the Expedition hills and terraces for Palaeoliths, proceeding gradually
The expedition leader, Dr. H. de Terra, and Dr. upstream from Yenangyaung to Chauk and Nyaungu.
Hallam L. Movius, Jr. and Mrs. Movius met in Cal- Dr. and Mrs. Movius continued this work until March
cutta, on November 14, 1937. They then proceeded 8, when they broke camp and proceeded to Mandalay
by steamer to Rangoon, and after a few days' stay the to pack their collections.
party left by rail for Prome where they took the boat. My own geological party left the lowlands on Feb-
The steamer trip up the Irrawaddy to Yenangyaung ruary 24, and traveled by motor car via Mayvmyoto
was completed on November 23. Lashio, capital of the Northern Shan States. From
The first ten days of field work were spent in the here we made daily trips to all exposures of Pleistocene
vicinity of Yenangyaung and from there two excur- formations and Tertiary basin deposits, of which we
sions to the Mt. Popa volcano, near Kyaukpadaung, found many in the Lashio coal fields. Two longer ex-
were made. On December 2 the expedition mem- cursions were conducted into the Salween Valley and
bers left by steamer for Mandalay, where camp was the limestone plateau north of Hsenwi; through these
established on the right bank of the river at Mingun. trips we established the long sought-after contact with
In this region a geological reconnaissance was made the border region of China. The return trip from here
and quite a number of fossils were collected from the was made by car via Hsenwi and Mogok. An addi-
dissected hills west of the river. From here, Dr. and tional stay of two days at the latter town enabled us to
Mrs. Movius left on December 16 for the Southern
study the gem-bearing deposits at the village of Kathe,
Shan States with a plan of searching caves containing near Mogok, where the Pleistocene sequence of the
cultural deposits of the Pleistocene Period. karst region is especially well represented. When we
My own party proceeded to Mandalay and the North- returned to Mandalay, we were warned of the bubonic
ern Shan States on December 18. The trip was made
first by steamer upstream to Thabeitkyin, whence we plague epidemic, which claimed hundreds of victims
every day. We left town that very day for Maymyo,
went by motor car to Mogok in the Ruby Mines Dis- a hill station on the edge of the Shan Highlands. Here
trict. A week later we were joined by Pere Teilhard we packed our collections and expedition baggage. In
de Chardin. who had come directly from China. the meantime Dr. and Mrs. Movius had completed their
Jointly we studied the cave and fissure formations in work and had left for Rangoon, where they embarked
the karst region around Mogok and collected a fair
for Singapore and Batavia on March 17. My own
number of vertebrate fossils in Pleistocene deposits.
party followed one week later.
Very soon we realized that a more detailed investiga-
tion of the caves would not yield sufficient material to
warrant a protracted stay, especially since the main EXCURSIONSIN JAVA
problems had to be solved in the lowlands of the Irra-
Our in were at Bandoeng, capital
waddy. First, we paid a brief visit to the dissected of the headquarters Java and seat of various
slope of the Arakan Yoma Range, west of Minbu, and Preanger Residency gov-
then proceeded to Magwe, south of Yenangyaung. ernment institutions, including the Geological Survey
At Magwe we were joined by Dr. and Mrs. Movius ("Dienst van den Mijnbouw") of the Netherlands East
oil January 15. They had coime from Kyaukpadaung Indies. Here Dr. G. H. R. von Koenigswald intro-
with
where, on their return from the Southern Shan States, duced us to his colleagues, and to the problems
had discovered an extensive Neolithic site. The which he had come to grips during the last few years
they
terrace sequence and soils found on the higher land while searching for Fossil Man. Under his expert
surfaces around Magwe gave the first clear indication guidance we made an extended excursion to various
of sediment and soil forming cycles of the Pleistocene sites where fossil human remains and Palaeolithic arti-
Period. A joint search for Palaeolithic implements facts have been discovered. The expedition members
finally proved successful a few miles south of Yenang- are under a great obligation to Dr. von Koenigswald
yaung, near Sadaing. Further collecting of Palaeo- and various Government officials of the Netherlands
lithic tools made a longer stay for the archaeological East Indies for their arrangements on behalf of our
party necessary, but the geological analysis of the ter- work in Java.
race sequence required the survey of as many transverse The tour led from Bandoeng via Madioen to Trinil
DE TERRA: INTRODUCTION 269
on the Solo River, and here we visited the famous site yielded in 1936 the infant skull of an extinct race-
of the first Pithecanthropus fossil. From Trinil we Homo modjokertensis. This general survey was sup-
proceeded downstream to Ngandong, the second most plemented by additional laboratory studies which Dr.
important Fossil Man locality in the Solo Valley; at and Mrs. Movius carried out in Bandoeng on Dr. von
Ngandong the fragments of eleven skulls and two tibiae Koenigswald's very rich collections of Stone Age imple-
of Hollo soloensis have been discovered. Next we saw ments. To him, Dr. W. C. B. Koolhoven, Director of
the new Pithecanthropus sites near Sangiran on the the Dutch Geological Survey, and Dr. van Bemmelen,
Tjemoro River in the State of Soerakarta. From Soe- we are greatly indebted for the kind co-operation they
rakarta the route led to Patjitan in the Zuider Moun- extended to us while engaged in this work. Dr. Mo-
tains, where Palaeolithic artifacts and Pleistocene fos- vius' studies on the Palaeolithic of Java will be pub-
sils were collected. The last site which we visited lished elsewhere, but a resume of my geological observa-
jointly was near Modjokerto in eastern Java, at which tions on the age of Fossil Man in Java will be given
locality we studied the Early Pleistocene beds that had in Part V of this memoir.
PART I
THE PLEISTOCENE OF BURMA
BY HELLMUT DE TERRA

Plates I-XII

A. GEOGRAPHICAL ELEMENTS OF none of which attain a height in excess of 2500 feet.


THE REGION These smaller folds are elevated to a much greater
1. PHYSIOGRAPHY height wherever volcanic rocks are involved in their
younger structure. The Pegu Yoma is surmounted by
The area with which this report is chiefly concerned a graceful volcanic cone rising approximately 4,000 feet
lies between 20? 10' and 22? 15' N. lat. and 94? 45' to above the adjoining land. Mt. Popa (P1. I, Fig. 1),
96? 15' E. long. (Fig. 1). It coincides largely with and similar landmarks of igneous activity, cluster around
what is known as the "Dry Belt" of Burma, a narrow the various folds of the Pegu Yoma structure, with
strip of country extending from near Minbu in the which they appear to be related in origin. The south-
south to Mandalay on the north, and comprising the ernmost of these volcanic areas falls within our terri-
lowlands of the Irrawaddy Valley. The extraordinary tory, and it is of some importance in view of its silici-
width of these lowlands adjacent to the Irrawaddy fied tuff deposits, which furnished part of the raw mate-
River, which at places amounts to more than 100 miles, rial for the later Stone Age races of Burma. Such
justifies their being called a "basin" in this report. This suitable rocks are encountered in the vicinity of Kyauk-
basin is located between the Burma coastal ranges repre- padaung and Mt. Popa (Fig. 1). To the north, in the
sented by the "Arakan Yoma," and the Shan Highlands. neighborhood of Monywa, in the Shinmadaung and
The latter are commonly referred to in literature as the Shwebo Hills, and in the northern extension of the
"Shan Plateau," but this term seems hardly appropriate Sagaing Hills, volcanic craters and lava flows con-
for an elevated land surface possessing greatly differ- tributed their share to the relief-making of the Irra-
entiated relief. The elevation of the basin is about 300 waddy Basin. The drainage of the basin was in many
feet above sea level; it slopes perceptibly from north instances forced to adjust itself to such new obstacles
to south at the rate of 1.6 foot per mile over a distance and landmarks as the young volcanicity placed in its
of 400 miles. At the mouth of the Irrawaddy delta, way. This is illustrated by the great bend which the
the basin passes into a submerged estuary which is part Irrawaddy stream makes north of the elevated volcanic
of the Gulf of Martaban (Fig. 1). structure of the Pegu Yoma.
A glance at the topographic map of Upper Burma
(Fig. 2) discloses that the Irrawaddy Basin possesses (a) The Terraces of the Ancestral Irrawaddy
a varied relief. A median ridge, the so-called "Pegu
A special feature in the relief of the Irrawaddy Basin
Yoma," divides the basin into a western depression oc- is seen while
travelling by steamer upstream toward
cupied by the Irrawaddy River, and an eastern valley Mandalay. Here one notices the presence of wide, even
drained on the south by the Sittang River, and on the flats some 30 to 400 feet above stream level
north by a small tributary of the Irrawaddy. This (PI. V,
Fig. 1; P1. VII, Figs. 1 and 2). They stretch along
ridge is of great importance and interest because of its the left bank up to the higher slopes of the Pegu Yoma.
comparatively recent origin. Where the basin widens, Downstream, especially in the vicinity of Prome, Thay-
this ridge splits into several groups of hilly ranges, many
etmyo and Magwe, it is difficult to recognize in these
of which coincide with anticlines in Tertiary and Lower wide, level surfaces ancient stream terraces. How-
Pleistocene rocks. Such an anticline can be seen near ever, from Magwe on upstream the river has cut steep
the confluence of the Yaw and Irrawaddy Rivers where bluffs into the half-consolidated sediments of the Plio-
the Thagyi Hills build an asymmetrical anticline of Old cene and Lower Pleistocene rocks, and it is here that
Pleistocene beds (Fig. 3). Farther to the northeast one may clearly recognize a wide terrace, dissected by
the Shinwadaung Hills rise above the Chindwin Valley small tributaries, always gaining in prominence north-
flats, and between them and the Mu River there is an- ward. At Singu, Chauk and Pagan, this surface is
other lesser ridge. Farther north, near Mandalay, the many miles wide, and below it another level appears,
Sagaing Hills accompany the Irrawaddy River from which is built of red gravel and sand, in contrast to the
Sagaing northward to the Katha District. Thus there upper level on which there seems to be only a thin
are four ridges, in addition to the Pegu Yoma proper, veneer of coarse boulder gravel and sand. These two
271
_ _._
94' 95' 96- 97' 98' 99' 100'

~~~~ ~~Hsenwi_~

23-- ^ ^ ^ ^/>7M.ngm-o 'iti c


^O,,,:y
-/
/ThabeitkOin
--oRT7 SHAN TSPATES
* mpong
Kabwet
^Hsipau '
\,ShweboJ|)
1% ^ \I
( y
wvath^ Sine / \ ^ /*
lw :~L ^ /.-A
\ - ) I a -..
,^MnyaiJ
Monywa k~...
227

r~ N.
Mng-N
f '\.
maymyo
NiY?o
' 0-Ff
tb.

'C "?andal. a'"-


5~~~~~~~~~~~~~~~~~~-,
Kaukse
S ^w^^ ) ^r^T^^ r?
n Myingyan w
^. ^^y r^ k
.Pakol Sa,^r 2?{'' SOu H R !':H'_ T G S
~4`
'/'"-Pakokk~ 41c~Kengtung
yaunguu . ,.. - ,, ,
,,
Paz- Lo
M^^ Yenang'yating^
Yenangya
( v , S HAIN \ S T /TkE S
jkpadaug -j Ate iia / ^*

$,lino^~~~ Th e in
Ton IT

-- i
19,~~~~~~~~~~~~~~~~~~~~~~~~~~?
i;gnbaungwe %....,.~.."/%'
'~; G, .' nman )
;4 ,,/'
'~ ~ 'nana c </~9i/
Aingmongoe^
Aingm ~ ~ roe ^ | Cenma

"'i~ii'
~ , . ~I.ARE) i '.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~?'
= ~

-1?
1D'~~!/:~. ~. Toungoo.
- '.v-me ,oChienflmai

4'~~~~~~~~~~~~~~~~~
s
I
:

.......o' ?~o .\ ':'


CEYLON

~ '~~~~~~~3'
PV~~~~~~~~~~~Pg \~ S
~~~ I A~ M
a?~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~na
'-:.', ~~~~~~~Pe?u , '~
":' TIBET C H J N A

~~~~~~~~~~~~~~~~~~~~~~. .
.
i .
?=?~~r Madras~~'"
V\olmin
::':
iifi:
?

~ ~ ~ ~ ~\
? ?.

~ I.D~ ^ 2ZJ
Cluta J ] ,

-LEGEND-
?f
XXXP^
G zzulf o-- Route of the Expedition

,-etc., --Terrace Sections


?-67 ~'3~5~:~:=? ir ~if~:
94^^J^/^^^~~~~~~~~~~
:? :;?:?. itg a r t a b a n a (D, Scale, - 16'
'' : ?'?-. .?.. ^ ? 0 20 40 .60 80 100 Miles
^ -y ^ , , -4 T"""" , '' ^.
94. 95' 96' 97T 98' 99' 100'

FIG. 1. Map of Burma, Showing the Routes Traveled by the Expedition and the Pleistocene
Terrace Sections Investigated in the Irrawaddy Basin.
272
DE TERRA: THE PLEISTOCENE OF BURMA2 273
surfaces can clearly be followed over great distances ent river deposits only fine-grained matter. Hence it
upstream, and at many places a third level appears being seems probable that the lowest terrace is of sub-recent
above the high floodmark of the river. This charac- origin, and that the other two are much older.
teristic is shared by the other two levels, which suggests From above, the upper valley flanks in this region, i.c.
that they also are abandoned stream levels of the ances- the widest of our three levels, present a complex relief
tral Irrawaddy. From the boat one observes how the (Fig. 3). There are isolated and rather prominent hil-
upper two levels are associated with coarse red gravel, locks, as at Chauk (P1. V, Figs. 2 and 3), surrounded
whereas the lowest is built entirely of red and yellow by greatly dissected and often evenly levelled hill coun-
silt. This distinction is important inasmuch as the pres- try. Such hillocks maintain rather even levels, the up-

01020 ) 10 140 160 180 200


40 60 80 100
I-..,- 1 - I= Miles
FIG. 2. Outline Map of Upper Burma and Neighboring Regions, Showing Topographical Features.
274 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

per ones being less uniform than the lower dissected


prominences. One cannot help but feel that this relief
has passed through a prolonged or at least rapid stream
history, which succeeded the deposition of that great
valley formation into which the present river has cut its
meandering course.

(b) Primary Land Slurface


Between the river terraces and the ridges of the Pegu
Yoma there exists in the Dry Belt an intermediate sur-
face, which the casual observer might take for a pene-
plain, although it is by no means of uniform elevation.
In fact, ridges and hills appear to form a part of this
X53 ~ ~ land surface. These are unconsumed portions of re-
.er? ~ sistant rocks rising above this level, called by some a
plateau and by others a peneplain. A simple analysis
shows that it is neither of these. Thinly covered with
u:=rcn~ a sporadic veneer of laterite soil, this hilly upland sur-
*'a^ ~ mounts the terraces by hundreds of feet and near Mt.
o Popa by over a thousand feet. Obviously its origin
'5=~ ~ preceded that of the terraces, and it may well go back
....=1 ~ to the time when the latest folding of the hill ranges took
3|0 ~ place. The underlying Upper Irrawaddian Beds do not
form a part of this surface. They appear to be re-
?.5", ~ stricted to ancient stream channels, a fact suggested by
their restricted presence in synclinal folds. As far as
;>,^ the Irrawaddy Valley is concerned, it can be shown
-c=^ ~ that the topmost beds of this formation occur only in
synclines, which suggests that the upland originated
P. prior to their deposition. Hence the elevated and de-
u formed primary surface must have developed in a time
c=^ ~ span between the Lower and Upper Irrawaddy beds
'g~: ~ (Pontian to Early Pleistocene).
u

?oE~~~ c(c) The Shan Highlands


The eastern flank of the basin bears all the character-
cn
(L)
:o^ ~ ~ istics of a fault-scarp (P1. II, Fig. 2). It rises rather
z
ac0
v) abruptly to a height of three thousand feet, and its
11 -n~g ~ youthful tectonic history is revealed by bold precipitous
? slopes, indicating considerable vertical displacement.
'b~Q ~ Rivers descend over it in gorges, and while their origin
03 is often connected with fault lines, it would appear that
many of them were caused by stream capture in an an-
<c^ cjcientkarst topography developed on marble and lime-
stone rocks. In the valleys, terraces are found which in
many instances appear to correspond both in number
and in composition with those flanking the Irrawaddy
River. Here in the Shan Highlands alluvial fans and
talus formations are frequently encountered, particularly
in depressions in the karst relief. Such formations give
rise to a restless and irregular surface, probably the
most characteristic feature of this mountainous upland.
The undulating relief gives way locally to intermontane
basins, such as at Lashio, and it is here that we may
encounter Late Tertiary and Pleistocene lake and stream
beds (P1. XI, Fig. 4). The Salween-Irrawaddy divide
DE TERRA: THE PLEISTOCENE OF BURMA 275
extends into our region. Its barren summits rise up Now the great advantage connected with working in
to 6000-7000 feet and fall abruptly to the Salween the Dry Belt is the sparse vegetation, allowing favor-
Gorge, through which flows one of Asia's most power- able geological observation and unhampered surface col-
ful rivers. This watershed is not arranged in the same lecting (PI. VIII, Figs. 1-4). It is this feature which
manner as the Arakan or Pegu Yoma, because the Shan accounts largely for the finds of fossil and archaeological
Highlands are only part of a greater landblock extending material that previous students have discovered in this
deep into South China. This ancient land mass is built field. For a tropical land such a dry belt offers a rather
of a basal complex of gneisses and metamorphic rocks, unusual set of circumstances conducive to quick recon-
above which lies a complexly folded series of marine naissance and general survey work, such as we under-
limestones, shales and continental red beds, ranging in took during the winter season of 1937-38.
age from the Permian to Upper Cretaceous. This great
variety of rocks makes for a wide range of landforms. Meteorological Data
Outstanding is the karst, most typically developed in The meteorological conditions found in the Dry Belt
the marble belts of the so-called "Mogok Gneiss Series"
of Upper Burma can be appraised through the records
(P1. XII, Figs. 1 and 2). But karstification has af- of three stations. At Mandalay annual precipitation
fected also the Late Palaeozoic and Cretaceous marine
limestones. The former are apt to stand out in bold amounted to 515.9 mm., at Yenangyaung 571.5 mm.,
and at Minbu 809.3 mm. for the year 1930. These low
escarpments where very impressive gorges are devel- rainfall figures are in marked contrast with those regis-
oped. Dense forests, of both hardwood and bamboo,
cover the Shan Highlands, and accordingly exposures tered from the adjoining Shan Highlands, which rise
are rare and penetration is difficult. This green verdure 2600 feet above the dry belt. During 1930, Mogok
of forests (P1. I, Fig. 2) contrasts remarkably with registered 2541.3 mm. and Lashio, at 2802 feet, 1553.8
the dry grassland vegetation of the Irrawaddy Basin, mm. (Fig. 4). About 80 per cent of the total annual
which reflects the opposition of the two geographical re- rainfall occurs between the months of June and Sep-
gions as determined by climate. tember, as shown in Fig. 5; only 15 per cent falls during
the winter months of December and January. This
means that Burma enjoys a prolonged dry period lasting
2. CLIMATOLOGY
from October until the end of April. This dry period
Burma lies in the path of the Southwest Monsoon, is of great importance for an understanding of certain
which means that for five months of the year heavy soil-making processes in the Dry Belt. It is the season
tropical rains fall over the Irrawaddy Basin and the of gradual desiccation, of large diurnal temperature
adjoining highlands (Fig. 4). The moisture-bearing fluctuations, and of dust storms. All of these factors
storms enter our region via the Irrawaddy delta and are active on the vulnerable shale and silt-sand forma-
also across the Coast Ranges in the southwest. In the tions which underlie the Irrawaddy Basin.
latter instance a good deal of moisture is lost over the The average diurnal temperature changes for the year
wooded mountains that rise up to 10,223 feet in height, 1930 were 11.87? C. at Minbu, 11.83? C. at Mandalay
such as Mt. Victoria (Fig. 2). When the storms reach and 12.56? C. at Lashio (Shan Highlands). Other
the Irrawaddy tract in Upper Burma, there is relatively records for the same stations show the following varia-
little moisture left. The result is a broad stretch of dry tions:
land, reaching from the foothills of the Arakan Yoma
Jan. Febr. July
to the opposite side of the valley at Mandalay. Here Minbu 16.44 16.9 7.83?C (27.44)
the Dry Belt is widest, while southward, about 100 Mandalay 15.61 16.65 8.73" (27.11)
miles from its greatest width, it wedges out in the region Lashio 16.72 16.83 7.94" (21.67)
south of Magwe (Fig. 4). Here it is only about ten
miles wide; farther south at Prome, it is practically non- The mean annual temperature is given in brackets.
existent so that tropical verdure here covers the entire The 0? C. isotherm lies between 13,000 and 16,000 feet,
width of the valley. In other words the Dry Belt makes and the lowest temperature at Lashio was 0.5? C. over a
a triangle, the apex of which is pointed southward. A 25-year period. Inhabitants of this picturesque moun-
glance at the map suffices to understand this peculiar tain town told me of occasional snow flurries during the
position of the Dry Belt. It is widest where the western winter season. These figures suggest that physical
mountain rampart, the Arakan Yoma is highest, and its weathering is most active in the winter. In March,
apex lies near the end of the higher country to the west. temperature rises quickly in the Dry Belt and reaches
Another factor which doubtless accounts for this ar- its maximum in May shortly before the rains come.
rangement is the drying effect which the valley exerts March and February are the driest months, July and
on the monsoon storms as they proceed upstream from August the wettest. Solar radiation is not at all con-
the coast. Thus Burma repeats in all essential features stant over the lowland, as has been stated by Stein
the same type of climatic zoning as that which charac- (1931). In December and during the first half of
terizes the northwestern portion of India. January the sky may be cloudy for a week at a time.
276 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

96

-LEGEND-
240' Altitude in feet b Volcanic regions
-L.- Fault Scarp (Pliocene-Pleist)
1000mm. Annual rainfall (1930)
Scale ? 50 00 Miles

1553mm.
o Lashio
2802'

Co
%,.

oTaunggyi

18

of
Co

Be g al

of ..!

M a.r t a}ll a. :
96

FIG. 4. Map Showing the Contrasts in the Distribution of Rainfall between the Dry Belt and the Shan High-
lands. Arrows Indicate the Direction of Monsoon Storms.
DE TERRA: THE PLEISTOCENE OF BURMA 277
Mandalay
Lashio ..............
100"-

90*-
80*- .. ..
II I
I I I I
I???? ?

70-
60'-
50-
.I I I I I I l I

Jan. June Dec.


Annual Rainfall, 1933 Annual Temperature, 1933

FIG. 5. Annual Rainfall and Temperature Charts for Mandalay (Solid Line) and
Lashio (Dotted Line) during the Year 1933.

January until March is the period when clear skies pre- lack or great scarcity of vegetation, erosion is rapid and
vail and radiation is most intense. soil movement very active. While the silt components
Wind velocities are greatest in the beginning of the are blown from the surface, the coarser talus remains,
monsoon season (June-July). As in northwestern only to be broken up under the impact of the agencies
India, there is a regular period of dust-storms herald- mentioned above. And yet in gullies and dry stream-
ing the onslaught of the monsoon. Storm tracks lie in beds one encounters large quantities of talus, incrus-
the Irrawaddy Valley and move both from north to ted with hydrated ferric oxides and completely leached
south and vice versa. In March and April these sand- inside. In areas with poor drainage and lime-bearing
storms may be so violent that they blot out the landscape rocks, hardpan and concretionary soils are commonly
for days at a time. Then a curtain of suspended silt found. Especially on the silty red soils of Pleistocene
three thousand feet high hangs over the Dry Belt; often age and on such wind-blown deposits as the "Pagan Silt"
this is not precipitated until the rains come. This is the (see p. 309) the slopes are covered with small pellets of
same process which I had previously observed in north- lime, reminiscent of the "loess kindeln" or the Indian
western India, and which in my opinion is one of the "kankar" (P1. VIII, Fig. 3). Generally the pellets are
chief agents for mass precipitation of silt or "pluvial the size of a pea, but locally they may be as large as
loess." I consider that this phenomenon is largely re- walnuts, and they form regular talus fans along the
sponsible for the formation of extensive Pleistocene lower slopes of the gullies. It is doubtful whether they
loess deposits in the marginal zones of the monsoon rain are all of recent age, or whether they had started
belt. In a later section of this report I shall describe to form during a previous period when the Dry Belt
the role which this process played in the deposition of may have been even more arid than it is
today. The
Late Pleistocene and post-Pleistocene silts. post-Pleistocene eolian red silts, which I frequently en-
countered on the higher surfaces of the Pegu Yoma
Weatherilg and Soils slopes, indicate such a drier period during which pan
Periodic rainfall during the summer and intermittent soils may have formed to greater extent than at present.
showers in the winter months, combined with great Such fossil "kankar" soils may be distinguished from
diurnal temperature fluctuations and cool desiccating recent pan soils by the red patination commonly found
mountain winds, are the main factors of weathering in on the pellets.
this country. To this we must add the friable nature On the terraces weathering products vary greatly ac-
of the bedrock, and the reason why so few mature soil cording to the prevalence of silt or gravel components.
profiles are encountered in the Dry Belt can be readily The latter disintegrate on the spot, usually by thermal
understood. At Yenangyaung for instance, one may fracture, leaving a sharp angular talus from which the
walk for hours over perfectly bare rocks with only small finer material is blown off by storm winds. However,
patches of red soil preserved at the highest levels. there are no real "stone-armours,"such as occur in true
These bare rock surfaces resemble the pediments formed semi-arid or desert regions. No ventifacts were found
by a constant lowering of the primary basin surface to although desert patina was seen at several places, invari-
which reference is made on p. 274. ably on silicified tuff. The arid season is presumably
Physical weathering produces cracking and super- not long enough and the vegetation too effective to pre-
ficial crumbling of the solid rocks. The dry topsoil is vent surface deposition of silica gels.
either removed by the first summer rainfall, or it is Lateritic soils are restricted to higher surfaces, and
distributed by winter showers over rills and sloping sur- here it is often impossible to say whether the soil is of
faces. From here the loose surface soil either creeps recent or of older date. Perhaps the paucity of laterites
downhill or is blown off by dust storms, since, with the in the vicinity of the river, and their occurrence on
278 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
higher tracts over Pleistocene terrace formations, indi- peratures for weeks at a time, and their effect is par-
cate that the majority of these soils was formed under ticularly evident on shaly and slaty rocks. Even at
different climates. These may have been as rainy Mogok (3,200 feet) our tents were coated with frost
within the basin as in the Shan Highlands to-day. every night during the second half of December. Here
In the Shan Highlands red earth and laterite are the snow flurries are said to occur every other year.
most common soils. They are very conspicuous in the It is scarcely possible to understand completely the
bamboo-covered regions and in the teak forests. In fact genesis of soils and the former sedimentation processes
on the road from Thabeitkyin to Mogok one does not in these regions until we know more about the activities
encounter lateritic soils until the edge of the forest is of streams in Upper Burma and about the drainage
reached, which is also the border of the rain-belt along pattern of the country.
the slope of the Shan escarpment (Fig. 4). This asso-
ciation between red earths and forest is unquestionably 3. DRAINAGE
one which is connected with the retention of moisture
The masterstream, the Irrawaddy River, or "elephant
through foliage on one hand, and the general increase
of moisture on the other. Dense bamboo jungle and stream," originates from the confluence of two large
rivers which descend from the alpine heights of the
evergreen underbrush tend to increase the capacity of
the foliage to retain rain water temporarily, both in the Tibetan Plateau. They are the Nmai Hka, and the
Mali Hka (compare Bouterwek, 1919, pp. 298-306).
foliage and in the upper soil layers. Thus a fairly even
and long-lasting soaking results which must speed up The confluence occurs approximately 30 miles north of
the formation of red earth. Myitkyina (Fig. 2). The major confluent is the Nmai
Between the Dry Belt of the basin and the forested Hka which takes its source from the Languela glacier
Shan Highlands there is a narrow belt with an inter- in the snow-capped Tila miassif north of Fort Hertz
mediate rainfall. Here hard pan soils and concretion- (27? 24' N. lat.). Kingdon Ward (1924) mentions
that the headwaters of this river drain an area of about
ary limy soils are encountered. During the winter
450 square miles which is over 15,000 feet high. The
they receive little of the moisture that falls in December
and January, because hard, dried leaves of teak, iron- snow-line is at about 15,550 feet, the same level as in
the sub-Himalayas of Kashmir and the Punjab in India.
wood, Bombax and other large trees seal the soil from
above. Winter snow descends here to 6,500 feet and remains
In general, it may be stated that in the Shan area on the ground until April and May. The same ex-
the evergreen forest goes with a red loam type of soil. plorer (1912) maintains that the divide between the
The red loam or red earth is commonly five to seven Irrawaddy and Salween streams receives the heaviest
feet thick, but it may locally accumulate near the valley rainfall of all the mountains in Yunnan, a fact which is
flats to a depth of thirty feet. How much soil-creep important to remember when we come to interpret the
there is on the slopes is difficult to say. Judging from meaning of ancient flood deposits in the Lower Irra-
the scarcity of landslide-scars on hillsides, and taking waddy Basin. Near Hpimaw (26? N. lat.), Kingdon
Ward observed large snowfields at heights of 10-12,000
into account the considerable thickening of these soils
feet during the month of July. In other words the
at the base of the hills, I should think that there is some
headwaters of the Nmai Hka River receive greater
soil-creep, but not much landslide formation. The rainfall and snow water than do the other neighbouring
thickest cover of red earth which I saw was in the streams.
vicinity of Mogok in the Ruby Mines District (P1. XII, The annual discharge of the Irrawaddy River tells
Fig. 3). From here on westward toward the Shan the story of a powerful stream carrying a load of 400
border escarpment, the soil is uniformly thick, this million tons of sediment, with an annual discharge of
being the area of maximum rainfall during the summer 524 million cubic meters, and an extreme high flood
months.
discharge of 2 million cubic feet per second during the
In addition to these recent red soils, there are in
year 1877. In comparison, the Mississippi has an an-
many valleys of the Shan Highlands buried fossil soils. nual discharge of 544 million cubic meters. Gordon
A great many organic soils were encountered near (1885) has stated that during the month of August
Mogok in the karst depressions. In other instances, I 22 per cent of the total discharge occurs, and that 73.11
found lateritized fanglomerates and gravel-fans overlain per cent of the annual sedimentation takes place within
by lake and stream deposits. Again underground, in four months (July to October). He estimated the rate
sink-holes, pits and caves, ocher-colored powdery soils of denudation within the drainage area of the Irrawaddy
are encountered. These will be described in greater at one foot of surface elevation during 412.9 years,
detail in a later section. which is about seven times as fast as the erosion rate
Physical weathering has also played a part in the soil- computed for the Mississippi drainage!
making process of these highlands. Cool nights and hot Such figures illustrate vividly the erosive power in
days make for considerable temperature changes. At the Irrawaddy Basin. Obviously the main force of
altitudes of 5,000 feet or more there are freezing teml- the river is produced by the monsoon rains. In other
DE TERRA: THE PLEISTOCENE OF BURMA 279
words the load which this stream, together with its Significant is the fact that no red gravels or silts are
tributaries, carries to the sea is chiefly produced by a being deposited at present, in marked contrast to pre-
particular type of climate that is conducive to the de- vious times when these colored deposits dominated the
nudation and accumulation of fine-grained sediment in valley floors. When dry, the silt is of light gray or
the lower valley tracts and delta regions. yellowish color, and in this state it is swept up by the
In flood the Irrawaddy rises 15-20 feet in the Man- storm winds when they move upstream.
dalay region, which is some 400 miles north of the The second largest river in our region is the Sal-
delta, and at the so-called "third defile," near Thabeit- ween (Figs. 1 and 2). (In Burmese "Than-lwen" or
kyin, the stream may rise 25 feet above the mean "Dragon Stream.") Like the Irrawaddy, its origin is
average. The bench-marks of these high flood-levels in the snow mountains of Tibet, and it carries a good
can be clearly seen along the river banks. They are portion of the monsoon precipitation back to the sea.
marked by pebble or sand layers full of weeds, woody In the same latitude as the Irrawaddy at Mandalay, the
delris and fresh-water clams. In computing my ob- Salween is a boiling, fast-flowing torrent, some 3,500
servations on the levels of ancient stream terraces, I feet below the Shan Highland divide. Its gradient is
adopted a mean 0-point at 15 feet below such high steeper than that of the Irrawaddy, being 1:500 as
water-marks as I could see on the banks. This task compared to 1:350 for the latter stream. During the
was greatly facilitated by the fact that between Novem- rainy season the Salween is known to rise by 30 feet
ber and April the stream level undergoes very little vari- at a distance of 600 miles from the sea, while a few
ation, permitting one to calculate an average mean for hundred miles from its delta the flood mark is sup-
the area between Magwe and Mandalay. posed to be up to 90 feet above the average (Gazetteer
At present the river meanders along approximately of Upper Burma and the Shan States, Vol. II, Pt. 3,
one hundred feet below the level of the widely dissected pp. 92 ff., Rangoon, 1900-1901). Owing to the steep-
terraces (P1. II, Figs. 1 and 2). In this manner it ness of the valley slopes, the Salween Valley has few
breaks up great portions of the little consolidated sand- terraces, as most of them have been destroyed by vigor-
rock underlying the basin. After each high flood the ous erosion. But terraces exist and are known from
course of the stream changes a little here and a little the Chinese border near Kunl6ng and Bhamo (P1. XI,
there, breaking up older sandbanks and building new Fig. 2).
ones at the next bend downstream. This constant The main tributary of the Irrawaddy is the Chindwin
shifting of river banks has caused widespread denuda- (Figs. 1 and 2), which rises near the Irrawaddy water-
tion of the ancient terraces. As the stream meanders shed in the Kachin Hills (lat. 20? 40' N. and long. 97?
it tends to swing toward the west, following the great E.). It flows due north as far as the Hukawng Valley;
westward bend in its course below Mandalay. Hence it then turns northwest until it reaches the edge of the
very few if any terrace remnants have been preserved valley, where it again turns almost due south. This
on the right bank. general southerly course is followed to the point of its
At present the river carries mainly fine sand and silt confluence with the Irrawaddy, approximately 10 miles
in the region of the Dry Belt, and with it a good deal northeast of Pakokku. Below Homalin the Chindwin
of woody debris, most of which is derived from the receives one of its most important tributaries-the Uru
middle course and its three "defiles." In these the River-which rises in the Myitkyina District of north-
forested hills are undercut by the stream, so that the ern Burma.
natural forests along its banks fall prey to erosion; it The Namtu, another important tributary of the Irra-
is known that floating tree-trunks and branches have waddy, empties into the master stream near Mandalay
been carried out to sea two hundred miles or more. (Figs. 1 and 2). Its headwaters descend from the
The stream-channel varies greatly in width from three Salween-Irrawaddy divide, only 20 miles distant from
and a half miles at Mandalay to five miles at Yenang- the Salween Gorge. From east of Lashio, the Namtu
yaung. Here as elsewhere, it is braided in a wide flood flows across the Shan Highlands in a deeply entrenched
plain strewn with sandbanks on which the Burmese vil- valley. Its chief characteristic is the winding shape
lagers establish temporary winter settlements (P1. II, of its course, with deflections at right angles and can-
Fig. 1). On these sandbanks rice can be planted sev- yons developing within a few miles from mature valleys.
eral times before the flood season begins in July, and it Its stream pattern is complex, and it suggests a very
is this rice which is particularly rich in flavor. complicated history. While the upper course appears
There are three main types of stream sediment: (1) to follow the fault-pattern of the N.E.-S.W. strike, the
micaceous, brown silt mixed with small amounts of middle and more tortuous course is completely dom-
volcanic ash, derived from the eroded tuff and ash layers inated by the sink-hole topography of the limestone
upstream; (2) fine, current-bedded sand containing belt, especially in the "Plateau Limestone" area; the
coarse layers and fine lenses of silt, usually rich in or- Gogteik Gorge (famous for its scenic beauty) is cut
ganic matter; and (3) pebbly sand with individual into such giant fissures and sink-holes. The irregular
pebbles seldom reaching 1 inch in diameter. shape of the stream pattern in the Shan Highlands, with
280 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
its labyrinth of intertwining rivers, is sufficient proof But one wonders whether this considerable pile of
of our contention that this highland drainage is essen- coarse debris is not identical with the tilted. gravels ob-
tially derived from an ancient ill-drained karst land- served by ourselves near Minbu (see p. 284). In this
scape. latter instance the gravel occurs at the base of a layer
On the right bank tributaries like the Yaw and Mu of overlapping Upper Irrawaddies, and such might
Rivers are slope streams of an antecedent type. Their well be the case at other places. In commenting on
lower courses are entrenched in normal folds of Late the composition of this formation, both Pascoe (1912)
Tertiary and Pleistocene origin, while their middle and and Cotter (1918) agree that there is no distinct sedi-
upper portions cut across the strike of the eastward-dip- mentary break within the mass of the Irrawaddian Beds.
ping marine Tertiaries, which form the slopes of the This is indeed strange, considering that the mammalian
Arakan Yoma. None of these streams is of any great fossils indicate a two-fold faunistic division. Pilgrim
importance for our studies except the Yaw River, on (1906) stated that those mammalian fossils found in
whose banks I found a series of terraces reminiscent the lower third represented a type of fauna reminiscent
of those in the Irrawaddy Valley. of, if not identical with, the Middle Siwaliks of India.
Colbert (1938), who recently undertook a new study
of the Irrawaddian fauna with the material collected
B. GEOLOGICAL OBSERVATIONS IN THE by Dr. Barnum Brown, recognized in the Lower Irra-
IRRAWADDY BASIN 1 waddian fauna the Dhok Pathan stage of India, which
I. THE IRRAWADDYSERIES he had on former occasion (1935) identified as Upper
Pliocene. The fossils collected from the upper portion
1. Definition and Stratigraphy of the Irrawaddian Beds he considered as of Upper
Siwalik age closely related to the Tatrot fauna of the
A fundamental and well-known characteristic of the
Late Cenozoic formations in the Irrawaddy Valley is Punjab.
the sharp distinction between a younger sequence of Apart froml such paleontological relationships, there
are other records in the Irrawaddian deposits to indi-
undisturbed terraces and soils, with which the present cate that we have to deal with a complex rather than
drainage is connected, and a very thick older formation a uniform series. In the first place, lithologically the
of tilted and little-consolidated sandstones and silts,
upper beds are less consolidated than the lower ones.
commonly known as the "Irrawaddy Series." Each of a feature which has been noted by many observers. At
these divisions must be considered separately in their
Yenangyaung hard, concretionary sand and siltstones
stratigraphic order. occur above the "red bed," with large chunks of silici-
According to previous observations, chiefly those of fled, limy nodules weathering out on the surface.
Noetling (1895), Pilgrim (1910), Pascoe (1912) and There are several other "red beds" higher up in the
others, the 4,000 to 6,000 feet of the Irrawaddy Series
sequence, all of which are associated with hard shales
ranges in age from the Pontian to the Villafranchian. and sandstones reminiscent of the Dhok Pathan type
The former are represented by consolidated sandstones, of sediments in the Siwalik formation. Such red beds
siltstones and shales containing the so-called Hipparion are not found in the higher division, where sandstone
fauna (Lower Irrawaddies), and the latter by an upper is replaced by sandrock and the conglomerates by loose,
division of conglomeratic sandstones and silts with the or slightly cemented gravels. Hence, we believe that
younger Villafranchian fauna (Upper Irrawaddies). there are at least two different sets of sediments in the
In the central portions of the basin, the contact with
the underlying marine Pegu Series (Oligocene to Mio- sequence, as is the case in India, and also in China.
Furthermore in both the latter regions there is a very
cene) is disconformable, rather than unconformable. marked unconformity between the beds with Pontian
It is marked by a thin "red bed," about 5 feet thick, and Villafranchian mammals.
containing typical Pontian mammal remains. At the
only place where we could study this layer, it gave us 2. The Problem of the Plio-Pleistoccne Boundary
the impression of being derived from an ancient lateritic
soil. It does not differ essentially from similar beds It is to be regretted that in Burma we did not come
marking minor disconformities in higher sections of across any clear section which might have given us a
the series. decisive answer to this important question of Cenozoic
On the eastern border of the basin, near Mezali, 12 stratigraphy in southeastern Asia. However, certain
miles southeast of Pwynbyu (Fig. 1), Cotter (1938, of our observations strongly suggest the presence of a
p. 102) and Clegg (1938, p. 285) have reported the distinct sedimentary and structural break between a
presence of thick gravel beds, almost 4,000 feet thick, lower and an upper division of the Irrawaddian Beds.
which they place at the base of the Irrawaddy Series. One very fine section was observed along the
Pakokku-Pauk road (Fig. 1) on the eastern slope of
1 In this part the brief outline written by P. Teilhard de
Chardin for use in the description of our joint observations is
the Thagyi Hills near milestone 28.4 proceeding west-
included. ward from Pakokku (Fig. 12, p. 287). There is a
DE TERRA: THE PLEISTOCENE OF BURMA 281

deeply stained red gravel (layer 2) full of chunks of overlain by soils and terrace deposits. The sandstone
silicified wood which appears to mark a geological is tilted, whereas the boulder gravel is not. The latter
boundary between a lower division with fine-grained is not consolidated, while the former is at least as hard
sediments and an upper conglomeratic series. Evi- as the Lower Irrawaddian Beds near Yenangyaung. In
dently it represents an erosion interval, or at least a this relationship we believe a structural unconformity
period of soil-wash during which a great amount of is indicated. The horizontal beds are very clearly ex-
fossil wood weathered out of the underlying deposits, posed in a small stream, some 300 yards east of the
forming a bed of resistant talus and residual soil on left bank of the Irrawaddy and due north of the village
the land surface. This section is more fullv discussed of Thabeitkyin (Fig. 6-B). The cliff is high and ex-
on p. 287. poses a sequence of laminated silts, gravels, grit and
Much more suggestive of an unconformity is the curi- reddish concretionary clays. The coarse layers are
ous section which we encountered at Thabeitkyin, on the somewhat indurated; current bedding is dominant.
Irrawaddy River north of Mandalay (Fig. 1). The This sequence is identical with another one which we
section, shown in Fig. 6-A, illustrates in a general way found east of Minbu, at milestone 25 along the road
the relationships between an older boulder gravel and to Ngape (see p. 284), where Pleistocene deposits rest
a tilted series of sandstones, both of which are overlain upon tilted marine Pegu Beds. In this instance the
by red terrace gravels and concretionary soils. The clays and red gravels could be identified as belonging to
following features are significant in this section: (1) the Upper Irrawaddian division, hence we have little
the fact that a bouldery gravel fan covers a dissected doubt that the strata indicated in Figs. 6-A and B repre-
relief of Irrawaddian Beds, and (2) that the former is sent the Upper Irrawaddian Beds.

Eo

Irrawaedda

72' .._-
_
B T\ I/___ ,

W T2-3?I E
Irrawadd T2T4 o o I f.
, - /~.do'o'0 00'8aa
ooooo~$gS~o' I,,3
A ? OOO
I ...I Mi
ilMe.sf.
FIG. 6.Geological Sections near Thabeitkyin, in the Irrawaddy
Valley North of Mandalay.
A. Section on the Road to Mogok.
U.I. Upper Irrawaddian Beds.
F. Border Fault.
BF. Boulder Fan.
Mg. Mogok Gneiss.
T2-4. River Terraces.
B. Section across a Small Valley North of the Thabeitkyin Rest House.
a. Red Clay of T4.
b. Concretionary Clay.
c. Gravel.
d. Concretionary Clay.
e. Gravel.
C. Section near the Public Works Department Bungalow at Thabeitkyin.
a. Concretionary Clay.
b and c. Gravel.
282 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

96

FIG. 7. Map Showing the Location of Fossil-CollectingLocalities-Ma. 3, 4, 5, 8 and 9-


between Mingun and Kodaung, opposite Mandalay.

Such an interpretation is supported by the position of Such relationships as those illustrated in Fig. 6 sug-
the red terrace gravel found near the Public Works De- gest: (1) the cemented Irrawaddian sandstone was
partment bungalow at Thabeitkyin. The section, Fig. eroded; (2) it was covered by fan detritus which re-
6-C, shows that 15 feet of this gravel is overlain by red sulted from repeated faulting along the Shan escarp-
concretionary silty clay. The pebbles consist of quartz, ment to the east; (3) another erosion dissected the fan
granite, silicified tuff, quartzite, etc., which are rather and this time the Irrawaddy River deposited its sedi-
typical components of the Pleistocene gravels along the ments in the new stream channels; (4) the stream low-
Irrawaddy River between Magwe and Chauk. The ered its bed and left the terrace gravel high on the
elevation of this gravel above the river is 60 feet, and banks; and (5) this terrace was cut into by a small
it may correspond with the fourth or youngest Pleisto- tributary.
cene terrace in the main valley, where extensive rem- Corroborative evidence for a major break between
nants can be seen accompanying the river for many the two Irrawaddian divisions was observed opposite
miles. Mandalay. At Mingun (Fig. 7) a thick series of
DE TERRA: THE PLEISTOCENE OF BURMA 283

Upper Irrawaddian Beds is found on the eastern slope they compared favorably with modern fresh-water forms
of the Sagaing Hills (Fig. 8-A). They overlie ancient found along the Irrawaddy, and that they were distinctly
strata of gneiss, marble and schist which form the core different from any of the Tertiary fossils which he had
of an anticlinal ridge, here 700 feet above the river. collected in the lower beds of the Irrawaddian Series.
This narrow strip of crystalline rocks may be pictured Now this suggests that the fanglomerate and the clays
as an upthrust of the basin floor. It is only seven miles are Pleistocene in age, which confirms the evidence of
distant from the Shan escarpment, and the Sagaing anti- the mammalian fossils. The entire section therefore is
cline is faulted on both sides like the Shan Block. The a basal portion of the Upper Irrawaddies, indicating a
eastern fault is exposed one and a half miles west of period of extensive inundation in the basin. Indeed, on
Mingun on the road to Padu (Fig. 7). The section in the basis of the red and ocher-stained beds resting on
the Upper Irrawaddian Beds begins about one mile and top of the clay and fanglomerate beds, one is inclined to
a quarter from Mingun, at a bend in the road marked by consider them as representing a type of climate different
a pagoda to the south (Fig. 8-B; P1. IV, Fig. 1). Here from that of the present.
the stream has exposed a section through a fanglomerate In Fig. 8-B a cross-section is illustrated in which the
composed of angular debris, which is derived from the mass of the Upper Irrawaddian Beds is exposed.
faulted slope of the anticline. The fanglomerate is Above the basal fanglomerate (300 feet thick) rest some
overlain by sandy silt and laminated clay. While the 600 feet of clays intercalated with thin fossiliferous
former is cross-bedded, the clay is rather firm, and its layers. On top of these lie about 2,000 feet of red
extension in the strike can be readily deduced from the ocher-colored sandy grit, with current bedding and
extension of hillocky ground, characteristicof ill-drained indurated layers of coarse gravelly sandstone. Here
clay surfaces. Near the pagoda, at a place called Lo- and there bone fragments may be collected from the
cality Ma. 3, I found mammal remains (see p. 398 of indurated gravel beds. They belong chiefly to primitive
Dr. Colbert's report); these were collected in a tough elephants and cattle, indicating that the mammal fauna
gray clay, containing fossil fresh-water claims and snails was fairly uniform from the basal clay beds through the
near the base of a hill 100 yards distant. The latter overlying sandstones. This coarse mass of alluvial sedi-
occur in a black organic soil, 200 feet below the bone- ments is followed by a clay bed some 80 feet thick, in
bearing clay, and at places this black shell-bed is 40 feet which many leaf impressions were discovered. This
thick. The invertebrate fossils were shown to Mr. plant-bearing silt can also be seen three miles south of
F. E. Eames, paleontologist of the Burmah Oil Com- Mingun, near Letpan village (Fig. 7), where there is a
pany, at his laboratory at Yenangyaung. He stated that steep bluff on the right bank of the Irrawaddy (P1. II,

/
Hilies 700'\7 \' E
W Sasain3g \ E

I ?,
Irrawad-dy
I

A
I .XI .
H.x2Y2
Locality E
Ma.4
ing Hills a b
krrawaddy
I
I_

2500'
~~~B\ 1= !
FIG. 8.
A. Section Showing the Geological Structure through the Sagaing Hills, near Mingun (opposite Mandalay).
F.-Fault; Ls.-Limestone; Gn.-Gneiss.
B. Geological Section through the Dissected Eastern Slope of the Sagaing Hills. At Locality Ma. 4: a = Shell
Bed and b = Fossil Mammal Bed in Basal Upper Irrawaddian Deposit.
284 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

20 1 t1 ilMile
FIG. 9. Geological Cross-section through Tertiary and Pleistocene Beds on the
Road from Singaung to Pyabwe, West of Minbu. The Numbers on the Base-Line
Correspond to Milestones on the Road.

Fig. 1). These leaf impressions were found to be very bow of the dry river channel. We followed the path
friable, so that no good specimens were brought back from the Aingma resthouse, leading westward through
for identification. However, it seemed rather certain the forest and across a spur to some high cliffs. Here
that the plants belonged to an evergreen, tropical savan- the section shown in Fig. 10 was observed. Again we
nah vegetation. The dip of these strata decreases from found a sheet of red gravel overlying a sequence of fine
west to east, from about 30?-35? to 12? at the river gray sand, containing gravel layers reminiscent of tile
bank. Upper Irrawaddies. In the lower two-thirds of the
Another interesting section is shown in Fig. 9 in section a deeply stained fossil soil was seen, and be-
which the contact between the Irrawaddian Beds and neath it lay buff and pink colored clays with brick-red
the Pegu Series can be clearly seen. On the road from speckles and sand lenses. Bed 2 in the section belongs
Minbu to Singaung and Pyawbwe (Fig. 1), in the to a tilted series exposed about 50 yards east of the
Minbu District, a good exposure is found at milestone inspection bungalow at Aingma. The dip here is 25?
14. In a gravel pit north of the road a gray concre- to the southwest. Whatever the age of the underlying
tionary clay was observed disconformably overlain by strata is, it is quite certain that the upper deposits belong
red gravel. While the latter extends rather uniformly to the Upper Irrawaddian Beds. These disconformably
as a sort of soil-sheet over the hills, the underlying overlie layer c in Fig. 10.
formation displays a slight dip, which is exaggerated in
Fig. 9. This becomes more marked farther west. At 0- * o *
-o
0
o
o 0
L-
'
milestone 20 and two furlongs, a titled, whitish-gray, o
"
o000 00

quartz conglomerate appears intercalated with gray /--2


.
'. .' . : ? .
0
'.
g0

clays. Beyond this the ground rises and low hilly ridges .o. o'
0
*
0
o
0

appear. These are built of Pegu rocks, consisting of


alternating greenish micaceous sandstones, shales and
marine clays, and they contain shell-breccias-the guide
fossils of the so-called "Pyawbwe Stage," representing
the Aquitanian or Lower Miocene of Burma. These
FIG. 10. Cliff Section near Aignma, Minbu District.
data indicate that the Sarmatian and Helvetian Stages
1. Upper Irrawaddian Beds.
(Middle and Upper Miocene) are missing, hence the 2. Terrace Gravel (T4).
Irrawaddian Beds have come to lie directly upon marine
Pegu Beds. The gap between the two series may be In view of our observations it would seeml that there
due either to faulting or to a disconformable overlapping actually exists a major break between the Lower and
of the Irrawaddies. The latter is the more likely ex- Upper Irrawaddian divisions. Most probably the lower
planation, in view of the steeper inclination of the older portion wNasdenuded on the flanks of the Arakan Yoma
strata as compared with the gravel-bearing beds. The before the Upper Irrawaddies, containing a Villafranch-
impression gained in this section was that the red gravels ian type of fauna, were laid down. Such a structural
and clays form a uniform sheet over the steeply tilted break corresponds precisely with conditions found in
marine and older Irrawaddian rocks. As to the age of China and India, where the beds with horse, elephant
the latter, it is more likely that they belong to the Upper and primitive cattle lie disconformably on strata yielding
Irrawaddies rather than to the older strata, because no a Pontian fauna. A similar unconformity exists in Eu-
red bed was observed in any of the exposures. This rope, especially at the foot of the Alps, and here the
would possibly indicate a major unconformity in the oldest glacial gravel sheets extend over the tilted Mio-
upper part of the Irrawaddian sequence. cene and Pliocene formations.
At another place-Aingma on the Man River (Fig.
3. Sedimentary Characteristics
1), six miles to the north of the previous section-we
studied several good exposures. They are one-quarter On the whole, the Irrawaddian deposits represent a
of a mile west of the village of Aingma, close to an ox- cross-bedded river formation, covering the entire low-
DE TERRA: THE PLEISTOCENE OF BURMA 285
land tract between the Shan Highlands and the foothills Burmah Oil Company, the total thickness of these
of the coastal ranges (P1. III, Figs. 1 and 2). Its thick- conglomerates reaches 5,000 feet in the area north of
ness suggests that it is the result of a complex drainage Minbu. Such conglomerate beds can be readily dis-
pattern operating on a large deltaic plain. It has tinguished from Pleistocene terrace gravels by the great
counterparts elsewhere in Asia, such as in the Hoangho variety of rock constituents, heavy patination and ir-
Plain of China. As a rule coarseness increases from regularity of pebble size. The patina on the pebbles
bottom to top without ever reaching the degree of com- is mainly composed of silica or iron oxides. Porous
plexity and variation in pebble size found in the Pleisto- substances, such as volcanic tuff, fossil wood, or sand-
cene gravels; Conglomerates occur in lenses rather stone pebbles, show a greater degree of silicification than
than in regular beds. As in the case of the Siwalik do the other less pervious constituents. The con-
formation, such conglomerates are likely to yield abun- glomerate layers are the most heavily stained. No
dant mammalian fossils and silicified wood. The lenses doubt they once constituted the topsoil from which
may locally contain boulders of granite weighing fifty finer particles were blown off; this made for protracted
pounds or more. This coarse debris may have been weathering of the heavier constituents. The surface
carried in the roots of large floating trees, of which we weathering was probably greatly aided by a fluctuating
saw many in silicified condition in the upper portion water table, such as exists nowadays in the broad valley
of the series. flats of the Irrawaddy flood plain. Weathering was so
This silicification of woody debris must have been severe that it caused complete ferritization and silicifica-
completed prior to the tilting and denudation of the Irra- tion of the pebbles. In many instances the fragments
waddian Beds, and certainly prior to the formation of are coated with limonite and are leached inside.
the terraces, because rolled chunks of fossil wood occur No ventifacts were encountered, nor did we notice
abundantly in the Middle and Late Pleistocene sedi- any desert patina or effects of corrosion, such as occur
ments. It would be worth while to make a special study on the gravel surfaces in the Dry Belt. There is, of
of this phenomenon with reference to the age of the course, no reason why the effects of wind action should
silicification process. Previously I have mentioned that not have been preserved if they ever existed. We sug-
in certain horizons, such as in fossil soils (Red Beds, gest that the climate was even less dry during Irra-
etc.), silicified wood is the chief constituent. Judging waddian times than it is nowadays. But the fact is
by our observations on the comparatively recent ac- that the Upper Irrawaddies with their heavy gravel
cumulation of fossil wood in terraces, it is quite evident accumulations require a much greater stream power.
that these ancient fossil wood layers are residual soils. This could hardly have been induced solely by an up-
Such layers are generally associated with large tree lift of the basin flanks, because of the even distribution
trunks, which would suggest that smaller debris is de- of conglomerates in the basin. A rainier type of cli-
rived from the disintegration of large pieces. Chhibber mate for this stage is also suggested by the character
(1934) states that fossil wood is mainly confined to of the Upper Irrawaddian fauna. However, it would
siliceous beds, and that it is conspicuously absent from be rash to attribute all of the Irrawaddian gravels to
the silty clay sediments. W\hile rejecting other people's a rainy or pluvial climate. River action was no doubt
interpretations concerning the origin of this silicification. accentuated on the western flank of the basin by uplift,
Chhibber rightly suggests that it might have been caused and in the coastal ranges on the east by erosion dur-
by an intensification of colloidal action. In my opinion ing the Late Pliocene. This is at least indicated by
this could only have occurred during dry periods when the thick accumulation of conglomerates northwest of
woody debris formed a part of the topsoil on the flood- Minbu. Indeed the-case cited above may be a local
plain. Quite possibly at some places humic acids were facies only, and therefore connected with local uplift.
more active than at others, causing strong solvent action for there are no disconformities to be seen in the upper
in the sandy matrix and leading to the formation of division. Furthermore, had there been continuous up-
silica gels. These did not percolate into the deeper lift from the Late Pliocene on, it would be hard to ac-
layers because of the upward movement of capillary count for the fine-grained sediments, the siltstones and
water under predominantly hot and dry climatic con- clays, which abound in the lower and uppermost por-
ditions. The presence of monocotyledonous fossil tions of the division.
wood, such as palm, proves that the climate then was As regards the petrological nature of the silt all we
very much the same as it is at present. need to emphasize here is its abundance, and its peculiar
Two types of sediment deserve special notice: (1) character. The color varies from ash-gray to green.
gravels and (2) silts and clays. While it is true that but it rarely attains the pink and reddish tints of the
the former are more abundant in the upper division Pleistocene terrace formations. In the lower third of
of the Irrawaddies, it is nevertheless true that gravels the division, silts and clays alternate with marl layers.
occur on the western flank of the basin at irregular East of Yenangyaung, over 1,000 feet of siltstone was
intervals throughout the sequence. According to in- encountered in the lower part of the Upper Irrawaddies.
formation received from the chief geologist of the At Mingun (Fig. 8-B, p. 283) these beds contain fresh-
286 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
water shells, such as Unio and Melania. In the south- This may have been the time of the first glaciation in
western flank of the Thagyi Hills anticline, brown and the highlands of the Irrawaddy and Salween watershed.
gray clays constitute approximately 60 per cent of the and deposition may have taken place in a temporary
entire division. Fig. 11 shows a section from one mile lake. This period was followed by the rapid accumula-
east of Pyinchaung (Fig. 1), where bone-bearing grav- tion of coarse detritus from the mountains. To this
els are intercalated with clays. Here it is obvious that latter stage belong the majority of conglomerates and
the clays are not lake deposits but precipitations of fine sandstones containing the Upper Irrawaddian fossil
river sediments. In fact they may be flood deposits fauna. It was a time of rapid stream accumulation,
such as are formed at present on the Irrawaddy flood reminiscent of the Pinjor Zone in the Upper Siwalik
plain. Considering that this plain must have been sequence of northwestern India. Finally, there was
broader before the stream had entrenched its course, it another period of quiet water sedimentation during
is quite possible that flooding led at times to vast which plant beds were laid down.
inundations and temporary ponding. It is also possible While this division of the Upper Irrawaddies is
that a good deal of the silt may be of eolian origin, for purely tentative, it should be emphasized that the di-
there must have been plenty of loose soil along the flanks verse character of the sediments indicates that they
of the basin during Lower Pleistocene times. cannot very well represent a local facies, since each is
The thickness and regional extension of these Upper found in more or less the same stratigraphic position
Irrawaddian silts argue for a relatively stable position within the sequence at widely scattered places. Such
of the base-level of erosion, a conclusion which confirms a variety of events may well have led to formation of
my previous supposition that the conglomerates require specific soil types, such as are encountered in the Irra-
a rainier climate rather than an unstable crustal condi- waddy Series.
tion. Hence it would seem that the Upper Irrawaddian
sediments reflect various stages in the history of the 4. Upper Irrawaddian Soils
basin. First, there was deposition of fanglomerates
while uplift was active along the basin flanks. Then Although little of an ancient land surface was pre-
there followed a relatively stable period during which served in the basin, certain sediments are encountered in
the compact gray clays and siltstones were laid down, the buried portions which suggest that some sort of
such as those recorded in the section (Fig. 8-B) west ancient soil was formed on the ancestral valley plain.
of Mingun, and east of the Yenangyaung anticline. At The buried position of such soil remnants is very often
this stage a portion of the silt was swept into the basin the only criterion for Upper Irrawaddian sediments as
from the glaciated highlands, because Dr. Krynine's far as age is concerned.
A section showing the superposition of ancient soil
preliminary microscopic studies on our samples from remnants is illustrated in Fig. 12. The western por-
these beds reveal a high percentage of very fine quartz
tion of an asymmetrical anticline, composed of Irra-
splinters (Cayeux's eclats), suggestive of glacial action.2
waddian Beds and located near Pyinchaung in the Yaw
2 Soil samples collected by me in Burma were turned over for Valley (Fig. 1), contains two bone-bearing gravel lay-
petrological study to Dr. Paul D. Krynine, of Pennsylvania State
College, who had previously contributed to the petrology of
ers, designated layer 1 in the section. The lower one
Siwalik and Pleistocene sediments (see de Terra and Paterson, (2) is disconformably overlain by red gravel (3), de-
1939). Unfortunately, by the time this work was submitted for rived from the Irrawaddian rocks. In it only the hard-
publication his report had not yet been completed so that I was est rock constituents are preserved, hence it probably
obliged to base some of my petrological remarks on a brief pre-
liminary study which Dr. Krynine kindly furnished for this pur-
represents a "residual gravel." It forms a thin, ancient
pose. It is to be hoped that his invaluable report on the soils
soil cap in the Thagyi Hills, where it can be seen on
of Burma will be published soon. the road leading from Pyinchaung to Pakokku. Ex-

L IVZMile
I
FIG. 11. Geological Section East of Pyinchaung in the Thagyi Hills, Yaw Valley.
1. Gray Sand and Clay.
2. Sandrock Containing Nodules.
3. Gray Sandy Clay.
4. Brown Clay.
5. Bone-bearing Conglomerate.
DE IERRA: 'HE PLEISTOCENE OF BURMA2 287

.3 1150' M.26.5
w
YawV 4
_,r

in section
FIG. 12. Generalized Cross-Section through the Thagyi Hills Anticline, near Pyinchaung in the Yaw Valley.
1. Ferruginous Gravel Containing Fossil Mammals.Upper Iaad ies.
2. Sterile Ferruginous Gravel. J
3. Red Gravel Found on Higher Slopes.
4. Terrace Soil.

posures are found at milestones 24.3 and 26.5, as well as quent in the upper third of the division, and they are
in the dissected western slope. This residual gravel, in usually restricted to the most pervious beds. To judge
which the diameter of the pebbles varies from 2-3 by the tilted position of these bands, it is certain that
inches, should be younger than the horizons 1 and 2, their origin preceded the post-Upper Irrawaddian fold-
and it may well have succeeded the anticlinal folding. ing and the Middle-Upper Pleistocene terrace forma-
Another such soil was found along the same road be- tion. In other words these zones of weathering may
tween milestone 11 and milestone 7.4. The bed labeled well be contemporaneous with the Upper Irrawaddian
4 in the section, Fig. 12, is a clayey red to purplish earth Stage. The criteria for recognizing these limonite
with quartz pebbles overlying Irrawaddian rocks. This bands as fossil zones of weathering are provided by the
is an eluvial fossil soil which will later be referred to presence of ground-water laterites in certain of the Late
as "Nyaungu Red Earth" (see p. 308). It is a lateritic Pleistocene terraces of the Irrawaddy Valley tract.
soil mixed with volcanic ash, and we found it to be Not only do they occur on top of the most extensive ter-
restricted to a distinct level belonging to Terrace II of race remnant (Terrace III), but they are also encoun-
our sequence. In view of the fact that the age of this tered beneath the "Pagan Silt," a wind-blown deposit of
terrace soil is early Upper Pleistocene, the underlying Late Pleistocene age (see p. 309). There probably is a
residual gravels, etc., must be older. genetic correspondencebetween the two series of weath-
Layer 2 in the section (Fig. 12) is a red gravel and ered profiles. One is still connected with the present
possibly indicates a boundary horizon between the drainage, while the other seems to be associated with the
Lower and Upper divisions of the Irrawaddy Series. much more complex and widespread Upper Irrawaddian
It is only one foot thick and composed of leached limo- stream pattern.
nitic pebbles, limonite pellets and sand concretions. In The precise meaning of these limonitic layers with
this matrix we found large chunks of fossil wood coated reference to former climates is difficult to determine.
with limonite. The other gravel, called 1 in the section, Their repeated appearance in the upper 2,000 feet of
is 6-10 feet thick and remarkably rich in broken bones the Irrawaddian Beds indicates that conditions favor-
of mammals and land reptiles; all of them suggest that able to their formation were present throughout this
the Upper Irrawaddian fauna lived at the time of the period. Such conditions must also have led to the de-
deposition of the gravel. This admixture of a fossil velopment of ground-water laterites, such as those de-
fauna with one of the fossil soil remnants demonstrates scribed by J. Thorpe (1937) in the neighboring prov-
that the entire sequence is of Lower Pleistocene age. inces of South China.
A second locality, where we can be reasonably certain
that we are dealing with Upper Irrawaddian soils, is the 5. Palaeontology
one shown in Fig. 6-B near Thabeitkyin. Layer d is a Invertebrate as well as vertebrate fossils abound in
red concretionary soil buried under river sand and buff- the Upper Irrawaddian Beds, although none of the lat-
colored silty clay. In another section near Kyaukpa- ter ever appear in a complete state of preservation. In-
daung (Fig. 13) green concretionary clay, 2-3 feet vertebrate fossils are especially numerous in the lower
thick, and lateritic gravel are buried under the lava flow clay beds overlying the basal fanglomerates. Usually
from the Mt. Popa volcano. they occur in a dark blackish clay rich in organic mate-
Near Nyaungu and Mingun, the Upper Irrawaddian rial, such as at Locality Ma. 4 (Fig. 8-B). The great
sandrock contains numerous ferruginous layers sug- variety of forms present suggests that swamp conditions
gestive of buried weathering zones. Here coarse sand prevailed for some time, and, since the silt connected
and conglomerate are firmly cemented by limonite; lo- with these shell-bearing layers belongs to the same type
cally hard crusts of limonite, 6 inches thick, make con- as that previously described as derived from glaciated
spicuous layers. Such limonite layers are more fre- highlands (see p. 286), these fossils may be regarded as
288 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
0 0. OC Lower Pleistocene (Period of the First Himalayan
) (<cM
uL Glaciation).
The vertebrate fossils collected by Dr. Teilhard, Dr.
Movius and myself were handed over to Dr. Edwin H.
I?
x
Colbert, Associate Curator of Paleontology of the Acad-
emy of Natural Sciences of Philadelphia. His report
- a)
on the Upper Irrawaddian fauna, as well as on the
Middle Pleistocene and later faunas from Upper
Burma, forms Part III of this report.
As a result of these studies, it appears that the Upper
Irrawaddian fauna is Villafranchian in age. It is
roughly contemporary with the Upper Siwalik (espe-
=b;~o cially the Pinjor) fauna of India, and with the Lower
=( ~ Sanmenian (Nihowan) fauna of China. Following
my previous discussion of the Pliocene-Pleistocene
a*~ ~ boundary in Asia (1939), I consider that this mammal
;>^ ~ assemblage is of Lower Pleistocene age. It would seem
that the presence of Late Tertiary forms, such as Mery-
4?; copotaimus,and Mastodon might contradict such a view-
Ug0~ point. But the occurrence of their skeletal parts does
XS not necessarily mean that the animals actually survived
z into Upper Irrawaddian times. For it should be em-
ir phasized that such fossils were invariably collected in
g>o , limonite layers and conglomerates, whose constituents
4 h may well be derived from the weathering products of
~ u; ^the Lower Irrawaddian Beds. If these bone-bearing
? me conglomerates are fossil soils, as I presume they are,
O gE| then obviously the mammal remains may show a slight
c
ctn? v admixture of older types. In their voluminous reports
<y ~ and discussions concerning the boundaries of fresh-water
X <r o~) formations in Eurasia, palaeontologists have unfortu-
&!
3, C-o? nately not taken sufficient account of these conditions
'
E- . P> of sedimentation and weathering. One thing is certain:
m_
~OQ; in our case the preservation of mammal and reptile
4 4 '
'3 bones was greatly facilitated by a quick burial process,
,, and by subsequent incrustation as a result of percolating
o solutions of hydrated ferric oxides.
Ut;j ~ The generic composition of the fauna leaves little
cI)^ ~ doubt as to the environment in the valley under a warm
U.?< climate. The remains of turtles, crocodiles, buffaloes
0 and elephants are most numerous; with these there
0> roamed on the open valley plain large herds of primi-
tive cattle (Leptobos), horses, gazelles and antelopes.
Xc^ ~ Presumably it was an open plain covered by tropical
savannah vegetation, very similar to the present flora
fc> ~ outside of the Dry Belt. Life no doubt was abundant,
but few examples have been preserved because of the
ever-shifting trend of the large streams that converged
on the great ancestral plain. If Early Man had lived
hereabouts, he would doubtless have found a most favor-
able habitat. But so far no human relics have come to
light in the Irrawaddian Beds, and it is very unlikely
that any will ever be found in these deposits in an
adequate state of preservation. Our collections show
that the far more resistant skeletons of large pachyderms
were widely scattered and badly broken up in the
burial process; therefore there is little chance that suf-
DE TERRA: THE PLEISTOCENE OF BURMA 289
ficient of the more delicate human remains would have, o
been preserved to enable us to recognize the phylo- ''. " "''
genetic status of any new species.

II. THE IRRAWADDY


TERRACESAND ASSOCIATEDSOILS
1. Previous Work
No ancient stream levels were observed in Burma
until recently. This is natural, since they are very in-
completely preserved and a low scrubby vegetation ob- ,
scures many details of the surface relief in the Irra-
waddy Valley. The first mention of stream terraces in -
this region was made by Pascoe (1912, pp. 48-54), who
referred to their existence in the oilfield of Yenaung-
yaung. He does not go into the matter, but simply
mentions the fact that the "Plateau Red Earth" (a
lateritic deposit found on the slopes of the Pegu Yoma) a
might be associated with an ancient stream level in that
vicinity. Cotter (1914, pp. 163-185) observed in the
Yaw Valley (Pakokku District) boulder gravels on ?
terraces 300 feet above the present stream level.
The most recent and interesting work on the Irra-
waddy terraces was done by Mr. T. O. Morris (1932
and 1935); he recognized six different levels each of .
which was named after a type locality in Upper Burma. .
His terraces range from 15 or 20 feet to 350 feet above
the level of the stream. Such a terrace sequence was M
first mentioned in one of his notes (Morris, 1932) con-
cerning the find of a Palaeolithic artifact on what, in his
opinon, is the third terrace level in the oilfield of Singu l
(Pagan District). No detailed description of these- -
features was given, but later Morris (1935) took occa-
sion to refer to these terraces in connection with the de- .>
scription of various Palaeolithic tool types from Upper ---
Burma. On p. 1 of this publication Morris states: "Five : .
distinct cycles of erosion, graded to successively lower
sea-levels, have taken part in the evolution of the physi- ,
ography of the region since the conclusion of the strong
post-Irrawaddian (post-Middle Pliocene) crustal fold-
ing. A sixth cycle is at present in progress." In the \
following year Morris (1936-a) reported on the strati- t ~
graphic composition of one of his terraces from the
neighborhood of Thaytmyo. His fourth terrace was |
found to be composed of 30-40 feet of red gravel at
the base overlain by 50 feet of sand. The formation
was cross-bedded and had subsequently been lateritized. .
This was the first instance which proved the association |~
of ancient river levels with younger alluvium, artifacts
and fossil bones. In a previous work Morris (1935, f
p. 14, P1. I) referred to a general cross-section drawn
across the valley near Seikpyu and Singu (Fig. 20).
This profile, reproduced in Fig. 14, shows a wide gently
sloping surface, some 900 feet above sea-level (stream-
level = ? 160 feet above sea-level), with a steeper slope _
beginning at about 750 feet. On the right bank of the I
Irrawaddy this top level is represented by isolated hil- , . ...,
locks near Petpedaung. The second terrace lies at ?
290 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
about 500 feet above sea-level, and it is found on the waddy terrace sequence, in conjunction with others ob-
right bank only, especially in the vicinity of Sitpin vil- served in the Shan Highlands, shows a close degree of
lage. The third terrace, though present on both banks, correspondence with the terrace systems of both India
is more extensive on the west; it is distinctly more dis- and South China. For this reason our preliminary
sected than the lower levels and occurs at 140 feet above survey of these phenomena needs to be correlated with
stream level. The fourth terrace is the most complete other and more detailed studies undertaken in north-
and extensive. While it is only 40 feet above the river, western India, where correlations between the glaciated
it would seem to be the most prominent in this region. and unglaciated terrains were worked out by Mr. Pater-
The fifth terrace is again restricted to the right bank, son and myself (1939).
and its height is given as about 130 feet above sea-level.
2. General Aspects
Finally, a sixth terrace is inferred on the basis of the
higher sandbanks in the floodplain, and it is this level We entered the field with the expectation of finding
which Morris evidently considered to be "still in the well-preserved terraces between Prome and Mandalay.
making." Instead we encountered local remnants of level surfaces,
It is certain that these studies mark a distinct ad- which, when viewed from a river boat or from a high
vance in our knowledge of the Pleistocene in Burma, vantage point, presented apparent surfaces or an accord-
but they have left open a great many questions pertain- ance of hilltops. While proceeding by steamer from
ing to the origin and age of the terraces. In this report Prome upstream, we observed the first terrace rem-
an attempt will be made to analyze the terrace sequence, nants south of Magwe. Here there are at least two
although it must be admitted that the brief time at our wide surfaces at about 100 and 250 feet above stream
disposal for field work did not allow us to go into the level. At Magwe the former surface is conspicuous.
Pleistocene geology of Burma with as much detail as Hills surmount this terrace suggestive of a higher level,
in the case of India. The explanations given are tenta- which may also be inferred from the small flat benches
tive only and need to be substantiated by correlations that appear here and there along the greatly dissecte(l
with similar formations in areas bordering on the glaci- slope of the oil-bearing anticline, south of Yenangyaung.
ated tracts of Upper Burma. From the following de- On approaching this oilfield, one may observe a third
scriptions, it may appear that there is a perfect corre- level at about 60 feet above the stream, and with it is
spondence between Morris' interpretation and my own, connected a fine red silt. It can be seen to good ad-
but in point of fact this exists only with respect to the vantage at the outlet of smaller tributaries on the left
topographic character of certain terrace levels. From bank. In this region dissection has practically obliter-
Morris' section (Fig. 14) specific inferences may be ated any trace of terraces, at least so it appears from
drawn. The top terrace is a broad sloping surface the river boat. At three or four hundred feet above
which might be anything but a terrace. It can be a streall level flat benches were detected, but they were
pediment level, or it may be a portion of an ancestral neither completely preserved nor very extensive. Far-
surface such as I mentioned previously in connection ther upstream, the 100-foot level and the lower terrace
with the physiographic character of the basin (see p. are conspicuous, but dissection has been severe, leaving
274). As far as our studies go, we have been unable few unconsumed portions of ancient stream levels. At
to corroborate this observation that T1 is a stream level Singu and Yenangyat (Fig. 1) the river cuts through
more than 700 feet above the Irrawaddy. The same another oil-bearing anticline, and in doing so it has de-
holds good for the lowest level, Morris' sixth terrace. nuded practically all older land forms. At Sale the
This is not a terrace because it is part of the recent mlain 100-foot level reappears clearly underlain by a
floodplain; it is subjected annually to widespread inun- thin layer of red gravel and sand.
dation. As for Morris' fifth terrace, it is difficult to see The most perfect sequence of successive levels was
how this could be 130 feet above sea-level when the seen between Sale and Chauk (PI. V, Fig. 1; P1. VII,
present stream level hereabouts is given as 163 feet on Figs. 1 and 2). Here Morris drew his standard sec-
topographic sheet No. 841.3 While it is possible that tion, and here our terrace observations also reached a
the topographers surveyed the region at low stream degree of completeness not attained elsewhere in Burma.
level, it would seem to be improbable that the water- From the river boat it is easy to differentiate among
table stood far below the level cited for the fifth terrace. three levels: (1) a lowest terrace connected with gray
As for the assertion that the terraces owe their forma- laminated silt and sand, (2) another terrace at about 60
tion to eustatic changes of ocean level, I shall refer to feet underlain by deep red gravel with yellowish silt on
the brief discussion of this problem at the close of this top, and (3) a rather extensive surface which dominates
the landscape on the left bank. Beyond one can recog-
report (see p. 335). nize a line of benches running along the higher slopes,
Hence a great many problems were left unsolved,
and while we ourselves claim to give a definite answer and a series of level hilltops which are especially well
to a good many of them, it would seem that the Irra- preserved at the outlet of the Yaw River.
This same type of arrangement of levels continues
3 Sheet numbers refer to the
maps of the Survey of India. upstream to Pagan and Nyaungu, but opposite Pakokku
DE TERRA: THE PLEISTOCENE OF BURMA21 291

7/ ---_ -

*'= -
~ ~- ~~---
/ '
7.1~
,ITITir\ n-

-
,-XV-/ I r r v 1--
-- -
33'2zT3-

-;~= --- KZ--- 0

FIG. 15. Block-Diagram of the Pleistocene Terraces near Chauk, Upper Burma. (Not to Scale.)
(Courtesy of the Geographical Review, published by the American Geogr. Soc. of New York.)

the river has cut away all traces of older alluvium. Terrace 4. More frequently preserved than T,, this
Here and there low ridges and hills appear, none of level is 55-65 feet high, and it is underlain by a red
which, however, bears any trace of gravels. Only op- gravel of medium size (generally walnut-size) and red
posite Mandalay, at the eastern flank of the Sagaing sand. Both of these layers are covered by a fine pinkish
Hills, does the "main level" reappear, and above it one or yellowish silt of eolian origin. The latter is called
notices two higher benches strongly tilted towards the "Pagan Silt," and it also occurs on the next higher level.
river (P1. II, Figs. 1 and 2). These surfaces are de- Limonite and hematite are found associated with an-
veloped on tilted Irrawaddian Beds (PI. IV, Fig. 1). cient lateritic soils in the T4 gravels. The total thick-
Upstream toward the "second defile," vegetation is so ness does not exceed 30 feet (P1. VI, Figs. 3 and 4;
dense that it obscures the relief of the basin, although PI. VII, Fig. 2). Implements of Upper Palaeolithic
the main terrace can be followed almost to Thabeitkyin age were discovered in the gravels of T4.
(Fig. 1). Here, at least two ancient levels were en- Terrace 3. T.: is the widest and most conspicuous of
countered, as illustrated in Fig. 16 (p. 292). This place all levels. This surface, 90-110 feet above stream level,
iarks the northern limit of otir studies in the Irra- is deeply dissected (P1. IV, Fig. 2; P1. VII, Fig. 2).
waddy tract. Its erosional or degradational origin is apparent from
On the whole, it may be said that terrace records are the varying thickness of the deposits found underlying
not well preserved. It is this imperfect state of pres- it. A boulder-bearing gravel at the base (PI. IX,
ervation which compelled us to compute from single
Fig. 1), overlain by red to pink-colored fluvial sand
sections, drawn through the more complete terrace or silt and locally covered by Pagan Silt, are the chief
remnants, the general outline of the stream history. geological characteristics. Rolled fossil bones of Upper
Before we present a description of the single sections. Irrawaddian affinities are frequently encountered in this
it may be well to summarize and enumerate the out- basal gravel, as are Lower Palaeolithic implements. A
standing characteristics of each terrace. firmly cemented gravel or ironstone hardpan with a
As we see it, the Irrawaddy Valley originally con- limy matrix is not uncommon at the base (Pl. III,
tained a rather complete sequence of five terraces, which Fig. 3; P1. VIII, Figs. 2 and 3).
we shall describe briefly in their order of age and state
of preservation (Fig. 15): Terrace 2. Generally this is preserved only in the
form of wide and rather isolated benches some 90 to
Terrace 5. This is the lowest; at Singu it is about 40 140 feet above T, (P1. V, Fig. 2). Nevertheless it is
feet above river level. Its sediments, known as "Singu recognizable by thick eluvial soils of purplish color,
Silt," consist of sand and silt, and they resemble the which will be referred to hereunder as "Nyaungu Red
recent Irrawaddy sediments very closely. The absence Earth." Its composition is a mixture of coarse soil-
of coarse material and of red coloring matter is typical. creep material containing silt and clay, which gives rise
The terrace deposit is found banked up against a steep to miniature badland topography (P1. X, Fig. 1).
slope below the next highest level. Accordingly, it These soils are underlain by river sand and gravels
represents a stage of stream aggradation (P1. VI, Figs. (PI. VII, Fig. 3), apparently of the same type as those
1 and 2). underlying the third terrace. Hence, it may be inferred
292 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
that this second level marks a stage of very thick allu- still form under semi-arid conditions. The soil profile
viation and soil formation. is exposed chiefly along the main road leading toward
Terrace 1. This is a somewhat problematical surface, Mogok. It is hardly more than three feet thick and
is superimposed on a gray, calcareous silt and clay
isolated remnants of which either appear in the form
formation overlying the ancient boulder fan. As one
of flat hilltops, or in the conformity of hills and ridges
approaches the escarpment to the east, the lime nodules
surmounting the former terrace by 100 or 200 feet increase in size, for it is here that schistose marbles of
(P1. V, Figs. 1, 2 and 3). It is associated with very the "Mogok Gneiss" type (see p. 320), abut the valley
coarse red gravel, which differs in certain respects from
those encountered at the lower levels. They are held plain. Hence the lime supply is local, although its
accumulation in the soil sheet can hardly be attributed
to represent what at other places has been called the
to the present drainage, since the streams are deeply
"Uru Boulder Conglomerate" of post-Upper Irrawad-
entrenched and carry the solution of lime directly into
dian age (see pp. 302-303). Quite possibly this high-
the master stream. For this reason the pan-soil is
est level was tilted prior to its dissection.
genetically tied to the underlying silt and clay forma-
As can be seen from this brief review, none of the tion, which may well represent an ancient deposit con-
recent formations within the Irrawaddy flood plain be- nected with a stage of intense soil wash on the ad-
long to any of the stream levels mentioned above. joining highlands. The two levels mentioned may be
followed for some distance downstream on the left bank.
3. Terrace Sections
(b) Section 2 at Ywathit, Singu District (Fig. 17).
Detailed terrace sections were taken between Tha- The map (Sheet No. 84, N/14, C1) shows a very
beitkyin and Magwe (Fig. 1), a distance of some 250 extensive and conspicuous flat level some 9 miles up-
miles, which is about one-sixth of the total known length stream from Singu (Fig. 17). Its elevation is 300 feet
of the Irrawaddy River. above the river, and its width of four miles suggests
the type of terrace designated T:: at other localities.
(a) Section 1 at Thabeitkyin (Fig. 16). Whether this actually is a terrace or whether it is a
The levels referred to as terraces in Fig. 16 do not down-faulted portion of a flat limestone bench, such as
give the impression of having once been river flats be- is illustrated in Fig. 16, remains uncertain. I would
cause of their imperfect state of preservation. The be inclined to adopt the former view, because elsewhere
resthouse stands on a narrow flat, some 60 feet above in this neighborhood terrace remnants are present.
the stream. The latter is underlain by small-sized, red One of them maintains a level of 100 feet and is under-
gravel, 2-3 feet thick, which is overlain by red sand lain by red gravel. Another lower one may be fol-
approximately 10 feet thick. The pebble constituents lowed downstream from Kabwet almost to Singu. It
are foreign and derived from crystalline formations, e.g. is probably one of the lower terraces, very likely the
gneiss, granite, schist, quartzite, quartz, and sandstone; fourth in our sequence, and it seems to continue di-
all of them are much worn, and many are stained with
rectly across the third defile. Below it, some three
limonite. This deposit can be followed upstream for miles upstream from Singu, a bench thirty feet high,
a few hundred yards, while eastward it is covered by with coarse-bedded sand and silt, may be seen, which
pink concretionary clay. This clay seems to be con- bears all the characteristics of our fifth terrace. It also
nected with a higher level as well, since a quarter of a follows the first defile, and being the most recent of
mile eastward it is encountered as a solid sheet over- the terraces, it is particularly prominent along tributary
lying boulder gravel, and here it is 50 to 100 feet higher streams.
than near the resthouse. This soil is a pan-soil charged
with lime concretions and thin irregular bands of marl. (c) Section 3 at Mingun (Figs. 18 and 19).
It may well be of recent origin, as the area falls within As mentioned previously (see p. 281), the eastern
the zone influenced by the Dry Belt where "planosols" slope of the Sagaing Hills, opposite Mandalay (Fig. 7,
1200'

-1000'

- 800'
Thabeitkyin
D.B.
-600'
TT43
, - 400'
~00o o oo 0 0
-
I I I /17 / /-/ i i 200'
0 1 2 3 4 5 6 Miles
FIG. 16. Terrace Section (1) through the Escarpment and Adjoining Basin, near Thabeitkyin. The Numbers on the Base-Line
Correspond to Milestones on the Road to Mogok. D.B. = Rest House.
DE TERRA: THE PLEISTOCENE OF BURMA 293

-650'

T4
/2 -350'
250' /'I I
0 1 2 3 4 5 Miles
FIG. 17. Terrace Section (2) Showing Pleistocene Gravels near Ywathit, Singu District,
Approximately 50 Miles Upstream from Mandalay.

: WA/
Locality 25
Ma. 225'
oo
I 0o.- o ? o o o o0 o 0
0 0o o'oot ~ 00 00
o o6 n
0
ja
0e
7

ol o'~ 700
FIG. 18. Terrace Section (3) at Locality Ma. 3, Near Mingun,
opposite Mandalay.
I
W
,,,I,...11 . . ., _J / __4L
E
:Ly dsanr ca. Iu.- ear n

gravel

l/.

FIG. 19. Geological Section through the Terrace Slope (T3) North of
Kodaung, opposite Mandalay.

p. 282), is broken up into several surfaces. Although from this basal layer. The bones are waterworn and
the slope is much dissected, these surfaces can still be carefully selected as to hardness; the fossils include
reconstructed, especially since there are extensive un- large proboscidean teeth and limb bones, isolated teeth
broken flats, some 100 to 225 feet above the stream. and jaw fragments of bovids and Hippopotamus. On
The 100-foot level is the most extensive. Fig. 18 gives the slopes we found many broken elephant teeth. At
a cross-section, one of many we studied and probably first it was thought that these fossils represented a fauna
the most perfect as far as the preservation of its struc- younger than that of the Upper Irrawaddian Beds, but
ture is concerned (P1. II, Figs. 1 and 2). At the base Dr. Colbert who studied the material assured me that
are tilted Upper Irrawaddy sands and conglomerates this is not the case. The fossils of the basal gravel be-
with mammal bones (see p. 398 of Dr. Colbert's report). long to Lower Pleistocene forms, such as Elephas hy-
It so happens that a great many fossils were collected sludricus, Merycopotalus dissimilis, Stegodon elephan-
from this neighborhood, especially at Locality Ma. 3, toides, and Stegodon insignis birmanicus. All of these
at half a mile southwest of Tanmyin village (Fig. 7), are typical guide fossils for the Upper Irrawaddies, and
and near Letpan (Localities Ma. 8 and 9). Figures since they occur in a terrace gravel of much later origin,
18 and 19 illustrate the manner in which the basal ter- they can only have been washed out of the older forma-
race gravel overlies the tilted beds. The lower two tion and subsequently redeposited in the basal layer of
feet of the gravel are cemented with a limy sand-matrix, the third terrace.
while the upper layers are loose and of filler texture. This presence of rolled fossils of Lower Pleistocene
In the basal gravel boulders up to one foot in diameter age in a terrace formation of much younger date may
occur. Also what fossils we found in this terrace came serve as an instructive example of the inadequacy of
294 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
fossil records for age determinations in Pleistocene allu- remnant of a pediment, I do not know. If it is a ter-
vial formations. In this case the bones are so well race, then it must have been tilted, because it dips 25?
patinated and rolled that there can be no doubt that toward the east. It so happens that this surface lies
they were transported by a later stream. Their affinity close to the fault along which Upper Irrawaddies were
with an older fauna being clearly established, it may be displaced, and this makes it very probable that posthu-
said that a vertebrate fauna in terrace deposits is not mous disturbances led to a tilting of the relief. In this
necessarily contemporaneous with the terrace. In this faulted region the clay hills, as well as those developed
case the third terrace is definitely not of the same age from fanglomerates, are strewn over with limestone
as the fauna, as will be presently explained. detritus. The latter cannot be of recent origin as there
The boulder-bearing gravel is between 6 and 15 feet no longer is a uniform drainage available for the distri-
thick. Where it overlies the Upper Irrawaddian Beds bution of recent fan debris over the hills. One would
without resting on an intermediate indurated quartz rather think that these are residual gravels derived from
gravel, it contains large pebbles of gneiss, quartz, and underlying Upper Irrawaddy Beds, particularly from
slate. The larger boulders are subangular and are ap- the fanglomerates which contain a great deal of lime-
parently all of crystalline rock derived from the Sagaing stone debris.
Hills anticline. Bedding is most irregular with gravel Near Letpan village, south of Mingun, I collected two
lenses wedging out within a few feet and being replaced implements on the slope of T3. One is a flat-based
by sand (Fig. 19). In the higher portions of this scraper of slate rock, and the other a flaked pebble of
terrace cross-bedded red sand and silt appear. Occa- quartz. As the edges are not much worn, I thought
sionally one encounters subangular pebbles; however, that they might have been derived from the red sand
the deposit is generally much finer than the underlying overlying the basal gravel of this terrace.
boulder gravel, and its constituents belong to fewer rock The terrace sequence at Mingun is supplemented by
types. There can be no doubt that this sand is river- a lower level which can be seen in numerous ravines near
laid, but it should be noted that the higher layers have Letpan. This is one of the two lower terraces-either
undergone some diagenetic changes. Concretions of T or T,. The gravels and red silts extend for over
limonitic sand as well as indurated beds testify to twenty miles southward along the slope of the Sagaing
weathering agencies antedating the deposition of a clay Hills, indicating that this formation accompanies the
deposit which blankets the terrace gravel. As one ap- Irrawaddy River on its way southward to the lowlands
proaches the slope to the higher surface, a veneer of of the Pegu Yoma.
brown concretionary clay is found, some 10 feet thick.
Its origin is obscure, but it may be suggested that this (d) Sectionz4 at Nyaungu (Figs. 21 and 22).
silty clay is wind-blown material-a facies of what has From Mandalay on southward the stream swings
been called "Pagan Silt." widely across the basin, branching out into several chan-
The second surface is devoid of coarse sediments. nels and removing effectively whatever remains of the
Its elevation is approximately 225 feet above the stream, Pleistocene terrace formations. As far as our observa-
although there is a perceptible slope toward the river tions permitted us to judge, it seemed that the first good
which brings its outer rim close to 160 feet. On it exposures south of Mandalay occur at Nyaungu (Fig.
we found a brown-reddish silt on the surface of which 20). Between Myingyan and Nyaungu we observed
soils of planosol type are developed. Exposures are few several terrace remnants, and again the most prominent
and do not permit of a clear picture of the sequence. one is the 100-foot level (T::), which begins north of
The abundance of small quartz pebbles on the slopes Nyaungu on the left bank (P1. III, Fig. 3). It can be
and hilltops might indicate that the underlying silt and followed downstream to Pagan and Chauk. As viewed
clay contained thin lenses of gravel. Indeed this is what from the river, this terrace disclosed a thin gravel layer,
we encountered at another section near Nyaungu where often only a foot or two in thickness, capping the tilted
the "Nyaungu Red Earth" is full of irregular pockets strata of the Upper Irrawaddy series. About one mile
and lenses of subangular gravel (see p. 309). As at distant from the river, this wide level is capped by a
Nyaungu and Pagan, this type of sediment is associated very extensive sheet of yellowish silt, which seems to
with the surface directly above the 100-foot terrace continue onto a higher level almost reaching the second
(T2), and it is this evidence which induced me to clas- terrace. Along the bank one cannot fail to notice how
sify the red silt and clay deposit west of Mingun as the small tributaries have cut through a filling of pink
belonging to this second stage of stream aggradation. silt 40 feet thick, which marks the lowest of our levels.
As one approaches the escarpment of the Sagaing This lowest level-terrace five-was most clearly ob-
Hills, the ground gets increasingly more broken up into served at Nyaungu, east of the main pagoda, which
badland forms and steeply sloping stream divides. No stands on a slope behind the village and southeast of
benches or flat surfaces were observed here, but there is the steamer jetty (P1. VIII, Fig. 1). Here there are
nevertheless an evenly slanting line of hills and ridges about 40 feet of gravelly sand, with lenses of silt and
suggestive of an ancient surface (P1. IV, Fig. 1). cross-bedded pink sand. No artifacts or fossils were
Whether this is a higher terrace (T1) or the eroded found. In following the ravine one climbs to the main
U
4: y\n1w
Gwegyo'

.1 C

| \X
C4^^ ils <~a SXag-
^ sNyaungnhla

Th a angyaung

~i~'~~i~
T" .1 at
~~~~~~~~~~htaw-

'^
M
Me-s A__---r
..(?-r-_.,,~t./~ til

FIG. 20. Map Showing Geological and Archaeological Sites Investigated between Pauk and Magwe in Upper
Burma. Arabic Numbers Refer to Terrace Levels; in Circles They Designate Geological Sections Drawn along
the Lines Indicated.
296 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
terrace (T3), which has been cut into the Upper Irra- soil formed during wetter conditions and then later have
waddian sands (Fig. 21). These Lower Pleistocene been transformed (or reduced) to a hardpan under a
beds are tilted 40? N.E., and they are capped by the drier type of climate. Such ironstone hardpans have
horizontal red gravels of T3, approximately 2-6 feet been described by Thorp (1935, p. 140) as occurring in
thick, which form a uniform sheet on an evenly abraded certain Tertiary clay hills of Kwangsi province in South
surface. But they have been largely stripped, so that China. Thorp attributes its formation to podsolization
only an ancient ironstone hardpan or lateritic crust re- "which took place a long time ago, before the present
main. This soil is very extensive, and it everywhere local relief was developed." It is notable that the
underlies the red gravels (P1. VIII, Fig. 2). present rainfall of Kwangsi is four times in excess of
that of the Dry Belt of Burma. In the vicinity of Ny-
Late Anyathian2 (surface) aungu the soil extends under the third terrace for over
Early Anyathian (insitu) // 5 | six miles east of the Irrawaddy River. It is formed
on an irregular surface, chiefly determined by the resist-
ance of Upper Irrawaddy beds. Subsequently, it was
buried under younger alluvium and wind-blown ma-
terial. The red gravel overlying it is of the same type
0o? o 02
o?ooo as the basal gravel of the preceding section (see Fig.
.p~
18). Its fluvial origin becomes apparent from the
.) . * .

FIG. 21. Geological Section through the Slope of Terrace III, perfect wear of its pebbles and the cross-bedding.
near Nyaungu. There can be no doubt that the soil was covered by the
1, 2 and 3. Upper Irrawaddian Beds. (1 = sand; 2 = con- gravels underlying the third terrace. The abundance
glomerate; 3 = indurated sand.) of Paleolithic implements, especially those called "Early
4. Ancient Cemented Crust.
5. Pagan Silt. Anyathian" by Dr. Movius (see p. 341), indicates that
this soil, marking an erosion interval, was a time of
The lateritic crust, an ironstone hardpan underlying human occupation. Early Man must have been at-
tracted by the type of raw material (fossil wood and
T3;,is composed of both large and small pieces of fossil
wood, large limonitic chunks of conglomerate, in addi- silicified tuff), which had accumulated on the surface
tion to quartz and slate pebbles, all of which are firmly as a result of weathering. No doubt the implements
cemented in a sandy matrix. It is a residual soil result- were manufactured on the spot and discarded after use.
The thin formation of red gravel overlying the soil,
ing from the prolonged weathering of the Upper Irra-
waddian Beds under climatic conditions different from may be studied to good advantage in the neighborhood
those prevailing to-day. Desert patina covers much of of the ruined pagodas on the road between Nyaungu
the fossil wood debris; quartz and slate pebbles are and Myingyan. The most complete section seen re-
coated with limonite and surface silica. In the matrix vealed 12 feet of gravel. The latter is well rolled and
silification of sand has taken place, and pure limonite is composed chiefly of quartz, quartzite, fossil wood, silici-
seen wherever the underlying rock is pervious. At one fled tuff and slate. Boulders up to one foot in diameter
locality Dr. Teilhard and I encountered a broken log of are not uncommon. We did not find any fossils here,
a fossil tree 5 feet long, still lying in the same position since the exposures are too incomplete and too weath-
which it apparently occupied when it was stranded on a ered to permit of the preservation of bones. The boul-
sandbank during Upper Irrawaddian times. Here one der gravel seems to have been overlain by red sand
could see how physical weathering had reduced the log for there are places where the latter can still be seen.
to approximately one-half of its original size, and how Near Pagan this gravel contains a few implements, but
the desert patina had been acquired during the period of the exposures are sporadic and the percentage of paleo-
its exposure on the ancient valley floor. For such it liths collected from the gravel patches is very small as
must have been: a wide river plain where the loose compared with the number extracted from the ancient
Upper Irrawaddy rocks underwent prolonged weather- soil.
ing. A third horizon is encountered on the third terrace.
In view of the absence of any such weathering prod- It is a structureless silt of yellow to pinkish color (Fig.
ucts on the present land surface of the Dry Belt, and 21, P1. III, Fig. 3). It blankets the surface for many
also because of the intimate association of this soil with miles and is always associated with pansoils, character-
a stratigraphic hiatus, I suggest that the ironstone hard- ized by lime concretions. As this deposit was first ob-
pan originated during a wet climatic phase, but that it served near Pagan, we have called it the "Pagan Silt."
was subsequently altered in a climate in which rainfall Its composition (PI. VIII, Fig. 3) is as follows: at the
was even less than it is to-day. This does not mean base lie a few inches of quartz gravel, its constituents
that the ironstone hardpan could not have developed not larger than half an inch or so. In this we have
from a red soil antedating a dry period. On the con- found signs.of stratification, though not the type which
trary, it may originally have been composed of lateritic one associates with river-laid formations. Small frag-
DE TERRA: THE PLEISTOCENE OF BURMA 297
ments of indurated sandrock suggest that this is an
alluvial deposit developed from the underlying Irrawad-
dian Beds. The silt itself is some 20-25 feet thick, and
it is structureless. It breaks off in steep walls and is
of a loose, porous texture. Where sufficiently thick,
as for instance one and a half miles south of Nyaungu,
miniature badland canyons have formed. At two places
we found angular, artificially chipped, pieces of fossil
wood, which were heavily encrusted by. lime, not far
from deposits of silt; although lying on the surface,
these may have been derived from the silt. Lime con-
cretions weather out of the slopes and on the surface,
but they rarely exceed a quarter of an inch in diameter.
Their origin here is connected with a pansoil developed
on the Pagan Silt.
According to Dr. Paul D. Krynine's preliminary re-
port, this silt is composed of very fine quartz and vol-
canic ash. It is to be thought of as a wind-blown loessic
type of sediment, whose source material must be looked
for (a) in the silt of the valley flats, and (b) in the 0
drifted ash from the neighboring volcanoes during dust = =
storms. As has been stated in a previous section (see
p. 277), sandstorms are still common in the Dry Belt, - >
and they lead to a large-scale shifting of fine sediment,
derived from the mudflats of the Irrawaddy Valley. 'E '
Just as the recent duststorms deposit their load irre- D

spective of topography, so the silt of the third terrace -


must have been laid down on the then-existing relief.=
Judging from the close association between this terrace ;
and the Pagan Silt, it might seem as though both are of
the same age. However, other sections have revealed
that this formation also covers the fourth terrace, and it T .o ;
is on this lower level (T,) that it really belongs. In
other words, the relief during Upper Pleistocene times u-
' >
must have consisted of a broad valley flat (T)), from Z
Ct
which storms carried and shifted the existing sand and H HHH
silt material over the neighboring banks, thereby blanket-
ing the slopes and surfaces of the most extensive flat/
level, which was the third terrace. The association of ? 6
the silt with the fourth level is not revealed in this sec- ~/
tion, but more important here is the relationship between
the third terrace and the next higher surface (T2).
As one proceeds up the slope in the direction of
Kabani, some five miles east-southeast of Nyaungu, red /
gravel may be seen as high as the 280-foot contour line A
(Fig: 22). North of milestone 25 on the road from
Nyaungu to Kabani, the ancient ferruginous soil is re-
placed by a deep red to purplish soil, which gains rapidly .i
in thickness until at an elevation of 330 feet an entirely
new formation is encountered. Its bright red color,
miniature badland topography, and smooth clay surfaces , / ,
characterize this formation as a thick accumulation of //-
lateritic earth. A quarter of a mile east of Kabani lie the
best exposures of this deposit, known as the "Nyaungu
Red Earth." Except for pockets and lenses of sub- c"\
angular detritus, there is no stratification in the 60 feet H
of Nyaungu Red Earth, which is exposed here. At the " j
base, wherever the Upper Irrawaddian rocks outcrop,
298 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
one notices an indurated ferruginous layer, but no basal such high amounts of rainfall for the formation of red
gravel was seen. Owing to the clay content of the soil, earth fans, it is certainly true that they could not have
peculiar rills and pinnacles are formed by rainwash formed in recent times. Against this latter interpreta-
(P1. X, Fig. 1). In the gullies lie the heavy gravels, tion, one may point not only to the compact structure
containing pieces of fossil wood and quartz, among of the deposit, but also even more conclusively to the
which a good many scrapers and other implements were fact that it is covered by a younger wind-blown silt,
recognized. Since none of these were found in the soil, which at places reaches a thickness of six feet. I should
there can be little doubt that this industry is of later think that during the period of the Nyaungu Red Earth
date-probably Neolithic. rainfall was at least double if not three times as great
As has been mentioned above (see p. 309), Nyaungu as it is to-day. At that time the bamboo and hardwood
Red Earth is connected with a certain surface along the forests probably extended into the basin, covering the
slope of the Pegu Yoma. The elevation at which it is entire Irrawaddy tract and providing large amounts of
generally encountered is 160 feet above T. This is the plant debris. This moist period followed upon a dry
level of the second terrace between Singu and Chauk, interval, during which the ironstone hardpan was formed
therefore it is very probable that this soil accumulated which occurs at the base of the T^ gravel.
on top of a terrace which preceded the erosional stage
to which the slope of To and the third terrace belong. (e) Section 5, south of Pagan (Fig. 23).
The question is what mode of deposition was required The neighborhood of Pagan provides excellent expo-
to form such thick lateritic earth. This may be an- sures in the Upper Irrawaddian Beds and the various
swered by pointing to its peculiar structure: the pockets deposits associated with the third terrace. In fact, it is
of sandy debris, and the irregular lenses of gravel and here that the nature of the soil has determined the
sand in an otherwise fairly homogeneous formation. building of a vast city of tombs and temples, because
Its shape, being that of a flattish fan trailing away from the hard ferruginous soil (ironstone hardpan) at the
the higher slopes of the Pegu Yoma and breaking off base of the red gravel below T3 provides a firm founda-
rather abruptly above the third terrace, indicates a soil- tion for all these buildings. It is said that about 30,000
creep formation. The soils found along the higher pagodas were built over a period of 600 years, and
that all the wealth of the ancient kings of Burma went
slopes, at about 700 to 1100 feet, are very much of the
same type, although they are not recent. The few expo- into the construction of this vast city of shrines and
sures which we saw showed that the red earth was over- memorial monuments. No better foundation could have
lain by a light-colored, silty clay with pansoil on top. been chosen by the ancient architects, for there is no
The underlying Irrawaddian sands produce a deep red other soil in the valley which equals this one in hardness
and extension.
gravelly clay, in which all except the most resistant The section, Fig. 23, was drawn four miles south of
pebbles are preserved. This soil is only 2-3 feet thick,
but it is very extensive. In gullies, it may be found Pagan across the dirt road leading south to Singu. The
road leads over barren gravel-strewn land, well dissected
in a re-washed condition mixed with pebbly sand and
by smaller tributaries of the Irrawaddy River. Along
covered by a later soil. the road and in gullies one notices two prevalent types
The formation of the Nyaungu Red Earth must be of gravels; one is coarse with large boulders capping the
pictured as a process of constant soil-creep down the tilted Upper Irrawaddian Beds, and the other is of
slope to a stream floor, where its movement was checked. smaller size with a high percentage of red sand and silt.
This presupposes that there were thick laterite soils on In the former we recognize the gravel of T,; again
the higher slopes, which may have been set in motion by there is the hard ferruginous crust, containing Early
torrential rainfall. In other words, the formation un-
Anyathian implements, with the Pagan Silt capping ex-
derlying the second terrace must be divided into a lower tensive portions of the terrace deposits. The smaller
gravel and sand division which is of alluvial origin, and gravel is definitely associated with a lower level (about
a higher soil division. While the former does not ap- 65 feet above stream level), which is our fourth ter-
pear in this section, except where the soil was removed, race. The Pagan Silt is here even more extensive than
the upper red earth division is well preserved. on T_. Its color is pink or reddish, but otherwise it
The Nyaungu Red Earth unquestionably required a shows the same structureless composition as in previous
type of climate more humid than the present, and cer- sections. Most characteristic of this T4 gravel are thin
tainly more humiid and tropical than that which pre- bands of indurated gravel or sand displaying hard crusts
vailed during the long dry interval preceding the second of limonite and thin bands of iron concretions (P1. X,
terrace formation. At present no such lateritic fans are Fig. 3).
being formed in the Dry Belt, whereas in the adjoining Such ferruginous crusts are rather common in the
Shan Highlands red earth accumulates on the lower lower parts of the Dry Belt. I presume that they are
slopes. As was previously mentioned (see p. 275), rain- zones of cementation belonging to ancient soils, which
fall here is five times in excess of the amount registered were formed under the influence of a fluctuating ground-
in the Dry Belt, and while it is not necessary to assume water table. One might be tempted to call these
DE TERRA: THE PLEISTOCENE OF BURNMA 299

"ground-water laterites," in spite of the fact that they


lack the structure and composition of true laterites.
Their formation might be pictured as follows: on the
ill-drained surface of the basin, precipitation led to
.
surface concentrates of hydrated ferric oxides. The
moisture, retarded in the capillaries of the sandy gravel,
I mnigrateddownward until it struck the hard impervious
crust which coats the Upper Irrawaddian rocks. Here
the descending water was checked, and in so far as it
did not evaporate, it precipitated its solution over the
old crust. This led to a gradual sealing of capillaries
in the lower parts of the profile. Another portion of
i the rainwater was drawn to the surface during times
of lesser rainfall-dry seasons of monsoon climate-
O
0
causing it to be evaporated, and thereby some of the
metallic oxides were precipitated. This leaching in the
K)
upper layers prepared the top soil for further surface
weathering, during the process of which the silt compo-
nents were removed by wind. While the surface solu-
0
" d tions made for an increase in grain size, they caused
greater imperviousness. Hence circulation of mineral-
laden solutions was augmented, so that more and more
o=^ ~ mineral matter was made available to metamorphose the
SO7~ ~ existing mineral grains. Quartz was replaced by these
czvl^ ~ ferric solutions, which were aided by humic acids pro-
:|X ~ ~ vided by a more luxuriant vegetation. Finally a stage
ES^ ~ ~ was reached when the interstitial space was filled out by
the newly precipitated limonite and silica. At that time
'4,> ~ a hard crust developed at a depth of eight inches to a
'O
foot; compact layers of limonite sealed the way to fur-
ther solutions. But this new crust was eventually
c---?
(
X *0
.o
O- u, I)~1brokenup again, as soon as physical weathering had
o,
v? proceeded further with the removal of the upper soil,
JI
eL,@ ~ and thus the water could once more percolate to deeper
C0
.: ^~H ~ layers. The limonite crust was reduced to angular de-
o
bris of hard concretions. Now, if the then-existing
river meandered across the wide basin floor, it must
have time and again buried such old soil crusts under
'0
~ new gravel layers. In these the same process would
o0
>C4~
start all over again, until such time as the stream ceased
D.
to deposit. As soon as it began to cut into its old
-J
valley floor, no further ferruginous soils could have
been formed, on the contrary they began to break up on
the surface, or they were gradually covered under Pagan
0. Silt. In other words, the sequence of these hard iron-
stained crusts in the fourth terrace required a rainier
climate, more luxurious vegetation and a fluctuating
water-table. The latter can hardly have had anything
to do with seasonal rainfalls, because there are no more
than four or five such superimposed crusts at any one
place in the entire formation. A changing ground-water
table on the other hand must have been related to
changes in stream courses, and these are clearly indi-
_r,
cated by the cross-bedded sediments in the fourth ter-
race. In almost every section which we studied, gravel,
sand and silt occur, so that there must have been a con-
1s stant shifting of the stream course throughout the time
300 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

T4 Ps
C. 60'
Pyinma -3
I T540
I

i CUZ
I
I -20'
I
t00'
I-

FIG. 24. Geological Section through the Right Bank of Pyinma Chaung, near Singu.
1 and 2. Lateritic Crusts.
3. Sand.
4. Pagan Silt.

of Terrace 4. Implements of Late Anyathian type Terraces 5 and 4 may be observed along the river
were found at a few localities; these have been studied where small remnants are found, each of which shows
by Dr. Movius, as described in his section (see p. 372). the characteristic composition described above (P1. VI,
On approaching Singu the fourth terrace becomes Fig. 3). The best preserved surface, T:, lies 100 to 110
more extensive (PI. VI, Fig. 4). Its upper slope shows feet above stream level. Though greatly dissected, it
fine gravel mixed with a slope-wash of lateritic soil. exhibits extensive flat surfaces capped by coarse red
At Pyinma Chaung, a tributary of the Irrawaddy, a gravel (P1. VII, Fig. 2). The latter is only 3-5 feet
very fine terrace section can be studied (Fig. 24, P1. VI, thick, but on approaching the higher slope, it thickens
Figs. 1 and 2). The lowest terrace (T,) exhibits some perceptibly. Large numbers of implements of Early
35 feet of fine cross-bedded sand and silt, intercalated Anyathian type were collected in this horizon (see p.
with thin gravel layers. It is essentially the same type 347 of Dr. Movius' report). No uniform soil crust was
of sediment which the Irrawaddy River deposits at seen here, but there are patches of it preserved at several
present. The nature of this formation is not related to exposures. In wandering across T3, one also encoun-
any of the higher and older alluvial strata, from which ters higher patches of gravel, such as near Zigyobin
we may conclude that Terrace 5 is Post-Pleistocene. where a very coarse boulder gravel was seen to rise 30
It is interesting to note that the Pyinma Chaung has feet above the average surface. These are either un-
cut deeply into this filling, but not sufficiently deep to consumed remnants of the old valley fill, or they are
reveal the total thickness of the deposit (P1. VI, Fig. 1). protected portions resting on higher ground.
No doubt its bed once lay below the level of the present Ascending from this wide third terrace toward the
Irrawaddy, but since the time of its formation, the Buddhist Monastery of Chinaungma, one must climb
river has only slightly lowered its bed, causing a tempo- 120 feet in order to reach the next level. Here the
rary dissection of T,. At present the Irrawaddy has same type of coarse gravel is found, but it is covered
entered upon a new phase of aggradation, which has with deep-red earth, some 12 feet thick, which spreads
in a fanlike fashion above the underlying alluvium (P1.
brought an end to this process. The Singu Silt of
Terrace 5 can be seen to invade the left-bank tributaries VII, Fig. 3). This then is a repetition of the situation,
around Singu (P1. VI, Fig. 1). It is also found three illustrated in Fig. 22 from near Nyaungu, and it is our
hundred feet upstream on the higher slopes, and here second terrace, which extends for many miles all around
we encounter red soils and wind-blown silt, both of the high, prominent Hill of Chinaungma (P1. V, Fig.
which may be regarded as the source of supply for the 2). Composed of Lower Irrawaddian sandstone, this
sediments of the lowest terrace. In fact such a source landmark is capped by another type of alluvium-T,
would explain why the Singu Silt is more deeply colored in the Pleistocene sequence of Upper Burma. Good
than the recent Irrawaddy alluvium. exposures are found on the western face, some 60 feet
below the top of the hill. Here a section, 32 feet high
(f) Section 6 near Chauk (Fig. 25). with boulder gravel at the base and fine small-sized
The lower three terraces may be followed uninter- pebble layers on top, can be seen. Cross-bedding as
ruptedly to Chauk, where they are obliterated by the well as the perfect waterworn shape of the constituents
construction of the oil town of Chauk. About one mile prove the alluvial origin of this formation. Its main
and a quarter south of the town, we studied another constituents are: fossil wood, quartzite, silicified tuff,
section which proved to be the most complete in Upper chert and gneiss. The matrix is sandy to clayey and
Burma with regard to preservation of levels and terrace deep red. It is a type of alluvium such as might result
deposits. One of its landmarks is a Buddhist sanctuary from the redeposition of river gravel under the contem-
called Chinaungma Monastery or Pongyi Kyaung (Fig. porary denudation of lateritic soils. In view of the fact
25; P1. V, Fig. 1). that this formation lies 330 feet above stream level, we
DE TERRA: THE PLEISTOCENE OF BURMA 301
can hardly avoid the conclusion that it is part of an
ancient and very extensive basin fill, of which only iso-
lated remnants are preserved on protected promontories.

(g) Sectioii 7 near Sale (Figs. 26 and 27).


This terrace sequence is rather complete, especially in
its lower portions. At Sale and downstream to a point
one hundred yards south of the pagoda of Uyin, we
observed a good exposure in the fourth terrace (Fig.
26). At the base lies cross-bedded Upper Irrawaddian

60'
PsI. I1.1X

o0 o* . .0 0 0
0. *0 0
E Ooo0 . *0 O QO 0 0

,0
._

u
O~ FIG. 26. Geological Section Half a Mile South of Uyin,
-,,~*5~~~ ~near Sale. Ps.-Pagan Silt.
3
o
>, sand, iron-stained, and overlain by a thin layer of coarse
(0
ui conglomerate. The latter has a loose consistency as
H
.,..o ~ compared with the Irrawaddian conglomerate, and it
Xog ~ may well be a basal terrace gravel. It contains an
abundance of fossil wood debris and rolled fragments
|WCt ~ of a red conglomerate, which can only have been derived
HF^ ~ from the alluvium underlying T,. A similar section
*o3r ~ was seen one mile north of Sale, on the road to Chauk.
Here, Upper Irrawaddian rocks are covered by a hard,
ferruginous, conglomeratic crust, overlain by red con-
0SE ~ cretionary soil and Pagan Silt. The terrace gravel of
cm T4 is seen to grade somewhat into the basal layers of
the Pagan Silt, hence there is no doubt that the latter
is as old as T4. It must be admitted that this gravel can
easily be confused with weathered Upper Irrawaddian
conglomerate, provided it is not associated with the other
two layers.
At another place, 2 miles north of Sale, on the un-
paved highway leading along the stream, the Pagan
Silt was seen to blanket both T3 and T, in one uninter-
rupted sheet. Here also we could see that the Pagan
Silt dissects the gravel of T3 in the manner of a
younger formation.
The most curious geological formation in this neigh-
borhood is the occurrence of coarse boulder gravel on
a hilltop (P. 621), east of Sale, and 450 feet above the
level of the stream. In the section, Fig. 27, its thick-
ness is somewhat exaggerated, for in nature it does not
exceed 12 feet. But there is good reason to believe that
this is only an erosion remnant of a once very extensive
formation, which must have formerly covered large por-
tions of the basin, as is indicated by the occurrence of
isolated patches of this gravel both here and at other
302 TRANSACTIONS OF THE AMERICAN-PHILOSOPHICAL SOCIETY

places. Its reduction in thickness as a result of ero-


sion must be considerable, on account of its exposed
situation on hilltops ald dissected slopes over which the
summer rains strike with full force.
This boulder-gravel (P1. IX, Fig. 2) consists of well-
rounded pebbles, 6-10 inches in diameter, with white
and red vein quartz, black chert, amphibolite. gneiss.
serpentine, greenish-gray quartzite, fossil wood, silici-
fled tuff and gray-brown sandstone as its chief con-
stituents. While the sandstone is either of the Pegu
or Irrawaddian variety, the black chert and quartz are
probably derived from the crystalline rock complex
exposed along the eastern and northern flanks of the
l)asin. Some of the boulders measure up to one foot
in diameter. The matrix is a red sand containinig a
o
considerable admixture of red clay.
This high gravel is referred to an upper terrace (T,).
%O
because as illustrated by Figs. 22 and 23, it is restricted
L
-
to the higher slopes which have undergone such in-
tensive denudation that only isolated hillocks remain.
V)^~t~ We recognized four of these high gravel-capped hills
o ~ l)betweenChauk and Sale, but there may be more, jiudg-
ing from the southward continuation of what appears
XLLu ~ to be the same level. Tentatively we have referred this
,.5^ ~ gravel to a higher terrace chiefly because of its peculiar
^~= ~ position above the horizon bearing the Nyaungu Red
?* Earth, and also because of the difference in the coIl-
OOu^ ~ position of the gravel, as compared to the lower terrace
u formations. The hilltops are by no means at an even
elevation. South of Chauk, the Chinaungma Hill is
H - + 330 feet, east of Sale the gravel is 120 feet higher.
and south of Hill P. 621 there is an even higher hilltol)
which must be close to 500 feet above the level of the
Irrawaddy. While it is possible that such differences
0 in elevation are in part due to young tilting movemenlts
~~^ ~ of the Irrawaddy anticline, it is equally probable that
"'^ ~ the top gravels are the remnants of an ancient basin
Cs-" ~ fill, which once covered our area to a depth of several
hundred feet. There are somlevery significant features
to support the latter view. In the first place, we have
evidence of other flat surfaces corresponding to those
at which the highest gravels are found. For instance,
at the outlet of the Yaw Valley (Fig. 20), such a high
terrace is very conspicuous. Here a coarse boulder
gravel of similar type overlies folded Iower Irrawad-
dian rocks. Also there are high surfaces and benches
in the Upper Yaw Valley, as well as along the eastern
slopes of the Sagaing Hills (P1. IV, Fig. 1). This
would indicate that a high terrace actually exists, but
its elevation differs, due to differential weathering under
impact of the monsoon rains. The second line of evi-
dence for the former existence of a widespread and
thick gravel concerns the appearance of certain con-
glomerates of post-Upper Irrawaddian age.
I i I I I I I Chhibber (1934, p. 275) mentions a boulder con-
glomerate from the Myitkyina District and the Hu-
S o o o go o

kawng Valley in Upper Burma (see p. 273 and Fig. 2).


DE TERRA: THE PLEISTOCENE OF BURMA 303
These localities are about 300 miles north of our region, less, pink and smoky). The colour of smoky quartz was
situated at the northern border of the Irrawaddy basin, observed to be due to inclusions of graphite. Boulders of
near the latitude of the junction of the two headwaters crystalline schists are very common, e.g. mica, quartz,
glaucophane-, graphite-, quartzite-, hornblende- and horn-
-the Mali Hka and the Nmai Hka. In view of the blende-garnet-anthophyllite schists. Boulders of grano-
importance of the stratigraphy and physiography of this diorite, diorite and epidiorite also occur. . . . Small boul-
formation for an understanding of the Pleistocene ter- ders of hematite and limonite are occasionally observed.
race geology, I asked the Director of the Burma Section The occurrence of jadeite boulders, locally called mowagyi,
of the Geological Survey of India, Mr. E. L. G. Clegg, has already been mentioned. . . . Sometimes beds or lenti-
cles of sand-rock are found intercalated in the conglomer-
for an excerpt of Mr. Chhibber's field report dealing ate, and a specimen hammered from about 1/2 a mile
with the "Uru Boulder Conglomerate." I quote from W.N.W. of Mamon (25? 35' 10"; 96? 15' 57") was seen
this report the following information: to contain the following minerals, along with the fragments
of quartz and other schists: quartz, serpentine, chlorite,
It may be statedat the outset that the formationto be de- garnet, hornblende, muscovite, saussurite, etc.
scribedbelow is the source of all jadeite bouldersfound in The following section is generally observed in the Boulder
the neighbourhoodof the Uru River (Fig. 2). So far it Conglomerate workings for jadeite.
has been takento be alluviumof Recentage, and the jadeite
I. Alluvium at the top: thickness variable.
workings in this conglomerateare therefore called alluvium II. A layer of pebbles and gravel, which the miners call
workings. Bleek too mappedit as alluvium,but in reality 'Kadi Kyaw.'
it is not so. It consists of boulders of all sizes, ranging
a
from few inches to several feet in diameter. III. Boulder Conglomerate, which is locally called 'Kyauk
They are
generally embeddedin red earth, but occasionallyin sand- Kyaw.'
rock too. It is noteworthythat some of the boulders are IV. Sand-rock with boulders, locally called 'Thai Kyaw,'
indeed and which is generally gold-bearing, and according to
very big perfectly water-worn, showing that the miners, a better quality of jadeite boulders is
they must have been transportedat least some distance,be- found in this layer. In case this layer is absent, the
fore being depositedin their presentposition. The present local inhabitants state that the place is not worth
Uru Chaung4 does not appearcapableof transportingsuch
huge boulders,but the river, however, is liable to tremen- working and valuable finds are not expected.
dous floodsduringthe rains. Last year, almosthalf the im- V. Bed-rock, locally called 'Phah'; all mining is
portantvillage of Hpakan (25? 36' 38"; 96? 18' 40") was stopped, when this is reached. Jadeite boulders are
washed away, and I was told by the local inhabitantsthat found only in III and IV.
at the time of floods, they could hear a rumblingnoise on Chhibber (1934, p. 276) mentions that such coarse
accountof the rolling of bouldersalong the river bed. The
formation is entirely composedof boulders conglomerated gravels occur also in the Hukawng Valley, and he refers
by red earth, etc., and since it adjoins the Uru, and in all the Gem Gravels of the Ruby Mines District to a similar
probabilitywas depositedby it or its predecessor,the name age. Although no mention is made of a section show-
Uru Boulder Conglomerateis proposedfor it. ing the contact between the Uru Boulder Conglomerate
The thickness of this formationmust be considerablein and the
Irrawaddy Series, it is obvious that these coarse
places, e.g. Balahka (25? 37' 30"; 96? 17' 1") is situated
on this conglomerateat a height of a little over 2000 feet. gravels are younger, not only on account of their dif-
From Balahkathere is a rathersuddendescentto the valley ferent composition, but also because of their slight tilt-
of the Manaung Chaung, and the cliffs overlooking the ing, which contrasts markedly with the folded structures
stream are seen to be entirely composedof this conglomer- found in the Upper Irrawaddian Beds.
ate. All along in the aforementionedstream,or in the cliffs
Now the only deposit which one can compare with
adjoining it, no other rock is exposed till Hpakan (25? 36' the
38"; 96? 18' 40") is reached,which is marked.806 on the Uru Boulder Conglomerate is the highest terrace
map (92 C/6), where the Uru Chaunghas exposed a small gravel between Chauk and Sale. Here we find the
outcropof the crystallineschists. From that it seems obvi- same mixture of greenstone and gneissic rocks and
ous that this conglomeratemust have a thickness of about younger sandstones. As
1000 feet, in places. Cliffs about 350 feet high, solely vious that the regards coarseness, it is ob-
boulders cannot very well be as big as
composed of this conglomerate, are fairly common. A
traverse along any of the importanttributariesof the Uru, those in the Uru Valley, because they have been reduced
e.g. Sabyi or Mamon Chaungs,convinces one of the great in size as a result of longer transportation. But one
thickness of this formation. cannot deny its fluvial origin, and that it was in fact
It attains a maximum width of over four miles in the accumulated
by powerful streams is quite clear. There-
longitudeof Mamon (25? 35' 10"; 96? 15' 57"), which was fore, our tentative conclusion is that we have to deal
once very famous for its jadeite workings, and from there
it continuestowards the north-eastand north on the right with a southern facies of the Uru Boulder Conglom-
bank of the Uru, with an average width of about two erate. Its origin must be pictured as connected with an
miles, while workings on the left bank are confinedto the alluviation subsequent to the folding of the Lower Pleis-
stream bank alone. It would be practicallycorrect to re- tocene basin sediments. Then an ancestral
markthat above the latitudeof Sankywe (25? 35' 25"; 96? Irrawaddv
17' 47"), the Boulder Conglomerateoccurs on the right flowed through a wide valley which was flanked by the
side of the Uru river. two ridges-the Pegu Yoma in the east and the Tertiary
It comprises boulders of altered peridotitesand all va- foothills of the Arakan Yoma to the west. The stream
rieties of serpentine (massive, antigorite, chrysolite, mar- was hemmed in only by the fold-relief of these two anti-
molite,etc.), rhyolite,siliceous brecciaand quartz-(colour- clinal ridges, whose strike is
approximately parallel to
4
Chaungmeans river in Burmese. the present course of the river. The level of this an-
304 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
cient stream must have been 300 or more feet above the
present floor.
(h) Section 8 near Yenangyallng (Fig. 28).
From Sale southward very few of the older and
highest terrace remnants can be seen. Along the river
bank one notices accumulations of silt and red gravel,
which may belong to the fifth and fourth terraces. At ,
Yenangyaung the slope of the Pegu Yoma is much dis-
sected; this and the building activities within the oil-
field make the survey of terrace levels extremely diffi-
cult.
As at other places, the third terrace is most clearly
preserved here, and its deposits yield Early Anyathian
implements. South of Thittabwe (Fig. 20) it makes
a fairly even surface, about one mile and a quarter wide. i /
Its edge along the cliffs of the left bank provided ex- '
cellent exposures (Fig. 28; P1. XIII, Fig. 2). Here
Upper Irrawaddian sand-rock and silt contain a good ,
many fossils (see p. 401 of Dr. Colbert's report), so
that there can be no question concerning the actual age
of the formation capped by the third terrace gravel.
The unconformity is clearly exposed, and generally
there is an indurated boulder bed, two feet thick. at /
the base of 8 to 12 feet of red gravel. At Sadaing, /
where the exposures are especially clear (P1. IV, Fig. .-
i
2), the red gravel is overlain by 7-10 feet of brown .
sand and silt. This is not the Pagan Silt of previous _ /
sections, instead it should be considered as part of the /
terrace deposit underlying T3. The Pagan Silt on the
other hand is found on the fourth terrace, where it con-/
tains concretionary layers. The brown sand is gen-
erally mixed with small gravel, containing a great tX
abundance of fossil wood. These particular strata can V
be seen along the road leading from Yenangyaung to
Thittabwe, and about half a mile from the latter village.
Here two red gravel layers are divided by 8 feet of
brown sand (P1. IX, Fig. 1). Cross-bedding is corn-
mon to all these deposits, as is a ferruginous layer at
the base. This very frequently overlies a two-foot o
layer of cemented conglomerate, containing a hard crust
of limonite 2-3 inches thick. I believe that this basal
crust corresponds in age and origin to the implement-
bearing ironstone hardpan or lateritic crust previously al
described from the vicinity of Nyaungu. It can be seen^
as an extensive sheet in all cliff sections where the T.;
gravels overlie Upper Irrawaddian rocks, such as at X>
Sadaing and farther downstream toward Magwe (P1. :
<
III, Fig. 1). The formation underlying the third ter- t ^
race has already undergone such extensive denudation X l
that the level has been stripped of alluvial sand, leaving
only a few feet of gravel. At other places, especially
northeast of Thittabwe, 20-30 feet of sand are pre- F
served. Such patches occur at various heights, indi-
cating that the terrace floor is somewhat uneven. No ,
doubt the more resistant layers of sandstone and con-
glomerate cause the local promontories, which rise from
the surface of the dissected relief, and quite frequently- -
DE TERRA: THE PLEISTOCENE OF BURMA 305
170'
o o o o0o o o?o red gravel
. -' '
.." '.. :. brown silt
/o?o o-?oo 0 - o o areagravel
o o
/^^ o > o= o^ O O old soil

FIG. 29. Geological Section Showing Terrace Deposits on the Road


from Yenangyaung to Gwegyo.

red gravels are found resting on such protected ground. silt. While the latter exhibits all indications of being
The third terrace and its associated deposits indicate an eolian sediment, the former shows irregular banding
a stream history as follows: (1) formation of an iron- and is mixed with tough red clay. As will presently
stone hardpan on an ancient valley floor which lay some be seen, this clay represents the higher slope facies of
100 to 150 feet above the present stream level. This the Nyaungu Red Earth of the second terrace, so that
was a dry interval with a deficiency of rainfall and it may safely be stated that in the main Pascoe's "Pla-
degradation of the river; (2) aggradation and formation teau Red Earth" is most probably a fossil lateritic gravel
of a thick valley fill composed of red gravel and sand of Middle Pleistocene age.
(Nyaungu Red Earth); (3) degradation and wide-
spread erosion which removed most of the former allu- (i) Section 9 near Magwe (Figs. 30 and 31).
vium; (4) deposition of wind-blown silt (Pagan Silt) This very extensive profile through the left flank of
derived from a lower valley floor which later became the the valley is notable for the completeness of its soils.
fourth terrace. Pagan Silt, Nyaungu Red Earth, Lateritic Gravels and
As to the fourth stage, it must be admitted that not Post-Pleistocene eolian Red Silt are found at their re-
much of the Pagan Silt is preserved in this neighbor- spective levels. Irrawaddian Beds are well exposed in
hood. Only at Sadaing, south of Thittabwe, was the the Kadaung Chaung (P1. III, Fig. 2), and near Magwe
silt observed on the fourth terrace, and farther north (P1. VII, Fig. 4). Here we observed large logs of
along the Pin Chaung. At Thonzechauk village a culti- fossilized trees and a wealth of debris of silicified palm
vated flat may be seen which is underlain by red gravel wood. In the cliff sections a number of Early Pleisto-
and concretionary yellow silt; the latter is the true cene vertebrate fossils were discovered (see p. 400 of
Pagan Silt. This flat is about a mile wide-and rises Dr. Colbert's report). It was at Magwe that we found
gradually toward hill-slopes crossed by the paved high- the hard basal conglomerate, about 3 feet thick, under-
way, which leads to Gwegyo and Chauk. At an ele- lying the T3 gravels, and containing a fair amount of
vation of 170 feet, the section shown in Fig. 29 was lime. From it we extracted a rolled flaked quartz
encountered. Again the Upper Irrawaddy rocks are pebble of Early Anyathian type.
capped by a crust containing limonite concretions and The top soil on T3 presented a difficult problem.
iron-stained pebbles. Thin bands of indurated, fer- Fig. 31 shows the complicated stratigraphy of this de-
ritized sand run in wavy lines across the face of the posit, as it was observed by us about one mile and a
exposure. The brown silt on top, derived from a half east of the ancient gilded pagoda north of Magwe
leached gravel, is of the same type as the T, sand, but and near a foot-path leading from the steep left bank
here it lies 70 feet higher. This may mean that we on to the widest terrace surface northeast to Kadaung
have to deal with another brown sand layer-one which Chaung. Here the T3 gravel is overlain by brown
is associated with the next higher terrace remnant (T2). sand and red concretionary silt, which locally displays
Indeed the top layer in the section, a thin, red gravel, the yellow color of the Pagan Silt. Pan soils cover its
belongs to an outer edge of a very extensive sheet of relief (P1. X, Fig. 2). This was greatly dissected and
red earth, and it may well be that this is a facies of the then filled up again at least to a depth of 35 feet. The
Nyaungu Red Earth, the lateritic soil which we gen- deposits of this fill stage reveal the same type of loose
erally encountered at about this elevation. and ill-stratified shingle as that found on T4. Indeed
This "Plateau Red Earth," as Pascoe (1912) called several implements of Late Anyathian type were found
it, is found chiefly at heights of 300 to 1500 feet above here, indicating that this stage was contemporaneous
the Irrawaddy. Good exposures are found at milestone with the formation of Terrace 4. Where ancient gul-
21 on the Yenangyaung-Gwegyo road, and at mile- lies have cut into the Irrawaddian rocks in this vicinity,
stone 28.3 on the highway south to Magwe. That we they expose the hard ferruginous crust characteristic of
have to deal with a fossil soil is proved by the fact that the base of T3.
this deposit is extensively covered by loose red or pink This terrace rises eastward, because here its mantle
306 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
of silt thickens. But the underlying Irrawaddian sand-
stone slope appears farther on, and some 400 yards east
of the section a new level is reached, which is composed
of purplish and freckled light gravelly clays, immediately
recognized as the Nyaungu Red Earth. Here the mi-
I "! nute badland relief is very pronounced, and its eastern
co boundary is very sharply offset by another soil, RS in
Ai
o
u*"~, ~ Fig. 30. This is a sandy eolian deposit, 4-5 feet thick,
42- 8 E the same formation which we have previously encoun-
o I - -
,q
-

Psi I *II l
h Ips.
' Late Anyathian2 ss
x5 - -'". ' ', 0 0 .' 0
o
0
O
0o oQ ?:~o o o oo o O?0
0
o oo o T2
O 0. . O 0?* O ' .0 .0 ..0 0 .0 T
it ?o
? O o.? , 0 .? 0? *o . o Jo 0
0 *o o 0 ..

USfw 31.
]~~FIG. Geological Section One Mile and a Half
Northeast of Magwe.
Ps. Pagan Silt.
Mo ss. Basal Sand.
C
T2 Red Gravel Underlying both the T3 and the T2 Deposits.
10 ~ U.I. Upper Irrawaddian Sands.
cs
tered in sections 6 (Fig. 25) and 8 (Fig. 28). A good
?1= ~ many implements were collected at this place (Magwe:
u Loc. 1) from the contact between the Nyaungu Red
Earth and the overlying sand, which belong to a Neo-
1. lithic industry.
o ?l:g ~ higher up on the slope, the association of terrace
~~Still
a:I ~ remnants with soils is less clear, and repeated investiga-
C
tions and barometric determinations of levels were
D1^ ~ ~ made, which helped to clarify the relationships of these
^ deposits. Pascoe's "Plateau Red Earth" covers most
cs*N of the higher ground up to elevations of 350 and 400
.? feet above the stream. At one locality, 11.3 miles from
cny~ Magwe on the Yenangyaung road and some 400 yards
,8^ ~ east of the highway (Magwe: Loc. 5), this sediment is
cE? ~ ~ full of iron concretions. The elevation above the Irra-
H?^ ~ waddy is 350 feet, which corresponds roughly with a
-<s ~ terrace level marked by stream gravels south of the
0' Kadaung Chaung. At milestone 13.4 this gravel over-
?'Eg ~ lies Upper Irrawaddian sandstone and is located directly
vo above a terraced slope on which three lower terrace
levels could be clearly recognized. This relationship
? with the younger stream terraces would indicate that
'- the "Plateau Red Earth" is at least as old as the first
r;L,> ~ or second terraces.
At another point (Magwe: Loc. 3), between mile-
stone 9.2 and 9.3 on the same highway, a gravel pit was
found some 200 feet east of the road (Fig. 32). In a
M
coarse basal layer we extracted several Early Anyathian
UJ
F- implements (see p. 347 of Dr. Movius' report). Simi-
lar artifacts were found in a somewhat younger gravel,
I- a few hundred feet to the east, which passes into a
concretionary red sandy silt. About half a mile farther
eastward, the purple-covered Nyaungu Red Earth is
seen. Evidently this is an erosion remnant of the same
ancient lateritic soil sheet, to which we referred, in dis-
DE TERRA: THE PLEISTOCENE OF BURMA 307
cussing the origin of the slope fans under which the
second terrace is buried (Fig. 22).
From these observations it would appear that the soils z
found at lower levels in connection with terraces have uL
their corresponding facies on the adjoining high land
surfaces. However, a detailed petrological analysis of
these deposits would be required in order to determine

Neolithic E
... Early Anyathian1 2

ht32.
Soil 20ydFIG. Ancient f yd 1.
FIG. 32. Ancient Soils on the Higher Slopes West of Magwe
(Magwe: Loc. 3). 4-c
1. Magwe Sand (Eolian and Containing Neolithic Implements).
2. *I
Nyaungu Red Earth. c:
3. Concretionary Red Gravel. C

4. Quartz Gravel Containing Early Anyathian 1 Implements.


5. Red Concretionary Silt.
,:

the specific correlations of each and every terrace soil \ C


1
a C'

3
found. As far as the soil of T2 is concerned, it is now o%r Ct
4c C
rather certain that it is derived from an extensive lat- .-J
eritic soil on the higher relief. Is

(j) Section 10 in the Yaw Valley (Fig. 33).


:>

Previously I mentioned the Yaw Valley (see p. 286)


in connection with the stratigraphy and structure of the o 'U

Upper Irrawaddian anticline in the Thagyi Hills west


of Pakokku (Fig. 20). This town lies 12 miles up- C.,
c
stream from Nyaungu, and to reach the Yaw Valley o. )
one must proceed westward on the road to a village >O
called Pyinchaung. In its vicinity and further up- oL
stream toward Pauk, we observed a good many terraces
U
(Fig. 33-A). The Yaw river, a right tributary of the
o

Irrawaddy, flows in a syncline of Upper Irrawaddian V)


0
Beds (Fig. 3). Both T4 and T5 are present along the 0
c,

left bank of the stream. The former exhibits a layer o


of gray rather structureless silt 6 feet thick, which may
be observed to good advantage on the right bank at
milestone 39 on the road to Pauk. Its basal gravel
contains much vein quartz, greenish quartzite and sand- Ou
stone, but no fossil wood. The silt extends for many v P
miles upstream and gives the relief a smooth surface,
and here again the association with pan-soils is very
conspicuous. T3 is well preserved in this valley, and
it is paved by a layer of tufa or surface limestone. In t,
I-
it we found, at milestone 37.2, teeth of Bos and a broken
jaw of Stegodon, both of which were presumably
washed out of the Upper Irrawaddian Beds and then
re-embedded in a surface soil (see p. 400 of Dr. Col-
bert's report). Vertebrate fossils abound in this neigh-
borhood, in fact bones are still being embedded and
cemented in the loose slope debris. A few implements
were collected in this limestone surface. The total
vertical section here is 30 feet, not including the upper
6 feet of silt.
308 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

Higher up on the slope, there are a few flat-topped curs on the uppermost terrace (T1) and on higher sur-
benches covered by red gravel of a much coarser variety. faces; (3) the Pagan Silt related to terraces 3 and 4;
These boulder deposits were also seen at milestone 40 (4) the Singu Silt found in T,. Samples of each were
on the road to Pauk. Here, about 200 yards up the collected and these were analyzed by Dr. Krynine, who
slope, such formations occur all over the surface of the submitted a brief outline of his study to me. I shall
hill south of the road. Its pebbles are up to one foot in briefly discuss their main characteristics,since they seem
diameter and are much more patinated than in any of to throw some light on the climatic conditions prevailing
the other gravels. Quartzite, amphibolite, quartz and at the time when the terraces were formed.
slate are most frequent, but whether these boulders ac- Nyazmzg Red Earth.-The most striking of all four
tually belong to a terrace or not could not be decided. soil formations is without doubt the Nyaungu Red Earth
Another deposit found in this region is a red gravel with (Fig. 34). Its external characteristics have already
a strong admixture of red clay. It is similar to the been mentioned (see p. 298) ; petrologically it is of varied
"Plateau Red Earth" of Pascoe and may well belong composition. The main mass consists of a brownish-
to the same age. red gravelly sand, which has undergone a certain amount
At Pauk (Fig. 33-B) T, contained rolled implements of lateritization. According to Krynine, approxi-
of Late Anyathian type (see p. 348 of Dr. Movius' re- mately 50 per cent is derived from igneous rocks, and
port). These were found one-quarter of a mile north the balance from metamorphic and subordinate volcanic
of the District Officer's Bungalow, near a gilded pagoda ash material. This is understandable when one con-
which stands on a prominent spur. Most of these arti- siders how varied was the origin of the Irrawaddian
facts were found in the slope-wash, but as the T4 gravel sediments that make up the ancient relief of which this
is the only deposit exposed along the slopes, there can soil is composed. Not only do the Irrawaddian Beds
be little doubt that they came from this layer. contain a considerable amount of fine gravel, but its
Terraces of similar appearance were observed at the constituents come from the more ancient highlands of
outlet of the Yaw Valley, near Seikpyu. In following the eastern and northern basin flanks. The volcanic ash
the path upstream from this village for approximately can only be derived from one of the extinct volcanoes,
two miles, patches of a small-sized, red gravel are pres- either Mt. Popa or the Chindwin craters, most of which
ent, which might well represent the fourth terrace. continued to be active throughout the Pleistocene. The
Some 200 feet above the river, terrace remnants occur, weathering of the particles is not uniform, although the
which are associated with coarse, red gravel similar to sandy matrix shows signs of laterite formation "which
the gravel of T3. A high mesa-like surface is seen on however at a later date suffered an interruption" (Kry-
top of the divide northwest of the Yaw Valley outlet; nine). Presumably this means that the matrix had
this level corresponds well in height with T1 in section 8begun its chemical weathering when it was moved from
its native horizon. This is precisely the impression one
(Fig. 28). It is interesting to note that an ancient
boulder-bearing alluvium extends up to the height of gets from looking at the texture of this peculiar sedi-
360 feet above the stream. ment. It is structureless except for local streamers
and pockets of gravel, and it presents a massive forma-
tion in which coarseness of grain increases towards the
4. Soils Associated with the Terraces
upper portions. At the base we frequently noticed a
A brief review of our observations on soils in connec- layer of round limonite pellets 6 inches thick. All this
tion with the study of the various terraces in Upper points to a short but violent, or at least rapid, transport.
Burma will help to demonstrate the origin and age of It is presumably the result of the slope-wash of an
these deposits. There are four major soil formations eluvial soil, which had undergone lateritization-a proc-
which we must distinguish: (1) the Nyaungu Red ess interrupted by re-deposition on top of the alluvium
Earth associated with terrace 2; (2) the Lateritic Gravel of T2. Whether the slope-wash, or mass transport, was
(Pascoe's "Plateau Red Earth"), which generally oc- assisted by simple soil-creep, or whether this process was

p
T2
T"-
/rrawaddy P.
P
'Ps.
I ~G~
*c-r \-^' . 100'
"

FIG. 34. Section through the Deposits of Nyaungu Red Earth, North of Kabani.
Ps. Pagan Silt.
R. Red Earth.
P. Ferruginous Pellets.
C. Cliffs.
DE TERRA: THE PLEISTOCENE OF BURMA 309
contributed to by supersaturation with rainwater in the proved by the fact that they are buried under extensive
pervious volcanic ash, is very difficult to state. Most red tuffaceous sands containing Neolithic implements
likely it was a time of greater rainfall, and of strong (Fig. 35). That all of the Lateritic Gravel is contem-
volcanic activity. Volcanic ash is known to yield readily poraneous with the Pleistocene alluvium, and that it
to slipping movements on a low-angle slope, and if passes laterally into the red gravels, as Pascoe (1912)
properly saturated with rainwater, it is bound to cause claimed, remains to be proved. We agree with his view
landslides on a large scale. This probably occurred that it is of Pleistocene origin, presumably contempo-
sometime after the second terrace was formed, and raneous with the highest terrace gravels,5 and that it is
prior to the formation of the lower slope of T^, since derived from residual gravels on an ancient land sur-
no such soils were found on any of the lower levels. face.
The examination of various samples has revealed that Pagan Silt.-The other prominent soil is the Pagan
the Nyaungu Red Earth varies in composition. At one Silt, a fine-grained fluvial product reworked by wind,
place it is richer in volcanic ash than at others where and containing a high content of lime (Figs. 36 and 37).
metamorphic and vein quartz constituents may pre-
dominate. This agrees with its general structural char- Jsl' ,/J ^
acteristics; the irregular shape of the gravel pockets, the
V,
00 0 ,0 __ c
varying grain size, etc.-all of which tends to corrobo-
10
000 0o0 Pse
o o 0 o0

rate our explanation. The combination of such lateri-


tized soil with red alluvium suggests that the climate
was rainier at that time; the stream carried more water, o -

and the soil weathered much more quickly than it does


now. For there can be little doubt that this red lateritic FIG. 36. Section through the Deposit of Pagan Silt on Ta,
earth was deposited shortly after the second terrace for- East of Nyaungu.
mation was completed. Ps. Pagan Silt.
a. Pan-soil.
Lateritic Gravel.-Of more regional extent is the b. Pagan Silt Containing Lime Pellets.
Lateritic Gravel or Pascoe's "Plateau Red Earth," be- c. Fine Quartz Gravel.
cause one can see it everywhere on the higher slopes of C. Lateritic Crust Containing Early Anyathian Implements.
the Pegu Yoma at heights between 300 and 1500 feet. U.I. Upper Irrawaddian Beds.
Petrologically this is a coarse, fluvial, gravelly sand, well
rounded and consisting of fragments of schists, quartz- On p. 277 I have referred to an eolian, loess-like product
ites, siltstones, red sandstones and quartz grains of sev- derived from the flood plain of the ancestral Irrawaddy.
eral varieties. All of these components are derived Dr. Krynine's petrological analysis points out the simi-
larity between it and the recent Irrawaddy sand, though
Neolithic it is finer and less well sorted. The high percentage
Implements (up to 7 per cent) of carbonate of lime suggests that
I4 this deposit is somewhat related to a loess. On T, at
,Z rrFriT I
Mingun (see p. 282), its facies is a coarse gravelly sand
~ o Railroad
containing 44 per cent of carbonate of lime mixed with
volcanic ash and metamorphic constituents. Except for
FIG. 35. Section Showing the Ancient Valley Deposits and Soils
in the Railroad Cutting, near Kyaukpadaung. the thin basal gravel layer (Fig. 37-B and C), there is
a. Red Sand (Magwe Sand?). no stratification in this formation, but locally one finds
b. Red Gravel Containing Fossil Logs of Trees (Lateritic pockets of small-sized gravel and slightly facetted peb-
Gravel). bles. All this is indicative of an eolian origin; obvi-
U.I. Upper Irrawaddian Sandstone.
ously, however, the material was derived from a loose
river sand drifted across the fourth and third stream
from the Irrawaddian Beds, and presumably they under- floors.
went a good deal of shifting from one place to another,
Whereas the Pagan Silt is found on T3 (Fig. 36),
for the structure in this formation is very irregular, in- it because of its association with the red alluvium of
is
dicating a rapid change of transporting power. As in that I am inclined to regard this deposit
the preceding case, these gravels are linked to older sur- T, (Fig. 37)
as the precipitate of a pluvial climate. Not unlike the
faces. Some of them may be homotaxial with the
pluvial loess of northwestern India-the so-called "Pot-
Nyaungu Red Earth, others are probably still older. war Loess"-it
In the main, however, they must be older than T, be- may have formed at a time when cy-
cause wherever they occur as a uniform sheet, they are clonic conditions prevailed in the lowlands and when the
connected with surfaces higher than T2. The origin seasonal rainfall was heavier than at present. I have
dates back to a period of intense weathering on the floor discussed this process in detail elsewhere (1939, pp.
of the basin when residual gravel must have covered the 5 In the Irrawaddy Basin these are the equivalent of the Uru
Irrawaddy rocks. That they are not of recent date is Boulder Conglomerate of northern Burma (see p. 303).
310 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

N 5
T3 Pagan Singu
oo. T4 T4 T4 T4
., T5
-A ^"--- 0o..... '" ,.....

TI. T4
w4' 4. t t l .ai a
;i45' T4-LL
o o oo00o o C

B z C V/..
5aL-
FIG. 37.
A. Generalized Section through the Terraces and Soils between Singu and Pagan.
B. The Structure of T4 at Singu.
a. Pagan Silt.
b. Upper Lateritic Crust.
c. Loose Light Sand and Gravel.
d. Lower Lateritic Crust in Conglomerate.
C. The Structure of T4 South of Sale.
a. Pagan Silt.
b. White Sand.
c. Basal Limy Crust.

264-76) and need only mention that such pluvial loess Sand." The type locality (Magwe: Loc. 1) is two
is typical of the monsoon-swept lands south of the high miles and a quarter northeast of Magwe, and it can be
Asiatic massifs. Whereas the loess of Central Europe reached on a path leading from the Buddhist monastery
can be regarded as a precipitate of a glaciation, this north of the town up the slope of T3 toward the village
loessic sand and silt may fall into a late stage of a of Natkan. On this path one first crosses the gravel-
pluvial period when the transporting power of the strewn surface of T3, and when one reaches a somewhat
streams was slackening, and a general silting of the higher level, the purplish-colored Nyaungu Red Earth
valleys was taking place. At such a time silt was plenti- appears. This is overlain by 4-6 feet of red structure-
ful, and cyclonic storms must have still exerted a great less sand (Fig. 38; P1. VIII, Fig. 4). This is the edge
influence over the climates of these lands.
Singu Silt.-The Singu Silt of the fifth terrace (Post-
Pleistocene) does not present any special features of
27' A i
I!III lll11f! a Neolithic
*..* "';-':"'.:'.-:..., Horizon
climatic interest, except that it is most nearly related to
the present Irrawaddy deposits. 6' ' /C
In the foregoing description of our terrace studies, \..* I. .T. *'
*
0'
the impression may have been gained that these re- I (II
spective soils keep strictly to specific levels, but this is
not always so. Figs. 34, 36 and 37 indicate how com- FIG. 38. Section through the Soils at Magwe: Locality 1,
plex is the composition of some of these terrace sections. 2 Miles Northeast of Magwe.
a. Vermillion-colored Sand (Magwe Sand).
b. Nyaungu Red Earth.
5. Extra-Terrace Soils in the Irrawaddy Basin c. Fine Red Gravel.
U.I. Upper Irrawaddian Beds.
Obviously the Pleistocene deposits which we have
called "soils" are not all restricted to the terraces, for
of a very extensive sheet which we observed at many
soil-making was universal over the basin. Also many
soils were covered deeply under the ancient alluvium places between Magwe and Kyaukpadaung (see Fig.
with which they have little or no relationship. 35), always on the higher ground. Though hetero-
The most interesting of these fossil soils are the dark geneous in composition, the Magwe Sand is very uni-
red lateritic crusts or ironstone hardpan beneath the form over this area. Derived from rocks of meta-
gravel of T3, and the loose red sand covering the high morphic origin (such as the sands and gravels which
relief units and containing workshops of Stone Age are abundant in the Irrawaddian Beds), it contains a
people. The lateritic crusts have already been discussed strong admixture of volcanic ash and is locally rich in
(see p. 296), whereas the other soil type, which is dis- carbonaceous matter. In addition to these character-
tributed irrespective of former stream levels, requires istics, pieces of flaked quartz, fossil wood and silicified
a brief explanation. tuff are frequently found. At Magwe we observed
This formation was first observed and studied near regular "workshops" of Stone Age people at the base
Magwe (see p. 306), and hence will be called "Magwe of the sand, yielding large cores of quartzite, flakes,
DE TERRA: THE PLEISTOCENE OF BURMA 311
small pebble choppers, fine scrapers of silicified tuff, etc. the higher slopes of the Pegu Yoma are concerned, there
In fact the archaeological material is so characteristic of are other soils of similar age which appear to be more
this formation that one might well use it as a guide restricted to specific rock types. Such a one was found
fossil for stratigraphic purposes. Equally significant in the valley of the Sai Chaung, near the village of
are the inclusions of charcoal and carbonized fragments Sebauk and south of milestone 39.2 on the paved high-
of grass stalks and roots. Both the grain size and the way leading from Kyaukpadaung to Popaywa, men-
structure of the sand argue for eolian origin. Deposi- tioned above. About 200 feet south of the road the
tion presumably took place at a time when the higher section illustrated in Fig. 39 was observed. The topsoil
land surfaces were populated by prehistoric people who (Layer 1) is 2-7 feet thick and composed of loose
had their camps on a dry grass-covered land. Either brown silt made up exclusively of volcanic material,
they, or natural agencies, were responsible for extensive e.g. large grains of feldspar, pyroxine, hornblende, apa-
fires leading to carbonization of the native vegetation. tite, and a little quartz, all embedded in a dark very fine-
The particular type of silicified tuff pebble from which grained matrix. In it are found angular fragments of
the people manufactured their implements, does not oc- basalt and silicified tuff, the latter being for the most
cur anywhere in the neighborhood. It is not found in part worked into implements. We found scrapers,
the Irrawaddian gravels nor in the Pleistocene terrace numerous waste flakes, and a polished stone celt in this
formations, but it is peculiarly restricted to the Magwe layer; some fragments of pottery were collected in the
Sand, as if it had been gathered by the prehistoric set- alluvium in the floor of the valley (see p. 378 of Dr.
tlers and utilized in their equipment as they roamed over Movius' report). A second brown volcanic ash deposit
the land. Indeed this must have been the case, for the (Layer 2) occurs at a depth of about 6-7 feet, and
only locality where such rocks occur is near Kyauk- here it overlies a sheet of light gray tuff with beds of
padaung. North of the railroad station there is an silicified ash and chalcedony (Layer 3). It is this
ancient crater of andesite partly covered and surrounded material which Neolithic man used for tool-making.
by tuff and agglomerate. These contain patches of On the opposite side of the gully, along which these
white or pink rock, flint-like in appearance and with a strata are exposed, lies a brown sandy silt containing
smooth conchoidal fracture, which is eminently suited three bands of black clay (Layer 4). These soil bands
to artificial flaking. A similar rock was found in large are each 3-5 feet thick and possess a granular structure.
chunks a few hundred yards south of the road leading Below the second soil layer and about 10 feet from the
from Kyaukpadaung to Popaywa, and close to the sec- surface were found half-fossilized teeth and limb-bones
tion shown in Fig. 39. Considering that no such rocks of a horse (or perhaps an ass), of a small deer of the
genus Cervus, of an antelope, and of a bovine animal
W like the modern banting or native cow of Java.6 Quite
possibly these and other animals were hunted by the
prehistoric people in the vicinity when Mt. Popa was
still an active volcano.
Below both the bone- and the implement-bearing soils
lies a mud-flow deposit. In any case this must have
been its origin, in view of the angular debris of lava
FIG. 39. Geological Section through the Soils and Mudflow and the irregular patches of gravelly soil found in a
Formation near Kyaukpadaung.
mass of volcanic ash 60 feet thick. It contains many
1. Topsoil Containing Artifacts.
2. Brown Volcanic Ash. lenses of what seem to have been gravelly surface soils
3. Gray Tuff Containing Beds of Silicified Ash. that were carried along in the mud flow and swept from
4. Mudflow Deposit. the Irrawaddian rocks down the slope of Mt. Popa.
5. Tilted Upper Irrawaddian Beds.
This mud flow apparently advanced through existing
6. Alluvium.
gullies cut into the Irrawaddian formation, as is indi-
occur anywhere south of the volcanic area, we must cated in Fig. 39. It is interesting to note that laterite
assume that the Late Stone Age people provided them- is entirely absent from these volcanic rocks, at least
selves with this raw material when they set out on hunt- as far as the lower slopes are concerned. Not even
ing and food-gathering excursions along the Irrawaddy. incipient lateritization has taken place in contrast to
These flint artifacts were also found near Minbu, Ye- the higher region on Mt. Popa.
As one proceeds up the slope of the volcano, one
nangyaung, Singu, Kyaukpadaung and in the gullies
carved into Nyaungu Red Earth east of Nyaungu (see notices along the road to the village of Popaywa a de-
p. 306). To judge from the character of the Stone Age posit of lateritic gravel. This is younger than the lat-
industry, the Magwe Sand is a Post-Pleistocene prod- eritized Upper Irrawaddian gravel previously men-
uct of a dry period during which sandstorms were fre- tioned. It overlies a basaltic lava flow and appears to
quent and volcanic eruptions active on Mt. Popa. 6 These determinationswere made by Dr. E. H. Colbert (see
While the Magwe Sand is of regional extent as far as p. 402).
312 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

Peneplain
Tz
?
o o.'
T2 T.I... 0
0o 0.

Irrawaddy R. \ . ?.?

TI T4 13

:t
..... S....Si-lt T an n Pe

::~ SinguSilt : Ts and on Peneplain


llLll PaganSilt: T4-T3
.#//// NyaunguRed Earth: Tz and on Peneplain
o.o.o Lateritic Gravel (Plateau Red Earth) : Ti and older

FIG. 40. Schematic Cross-Sectionthrough the IrrawaddyValley, Showing the Distribution of Ancient Soils.

cover the entire surface of the "lava plateau." It is of lateritic type with another group of planosol type.
difficult to tell whether this lateritic gravel is of recent Also there is in the basin sequence a distinct genetic
or Pleistocene age. The latter interpretation is prob- trend, for in practically all sections the lateritic soils
ably the more likely, because similar soils were found decrease as the eolian and more arid soil types gain
at a lower altitude near Kyaukpadaung, where they are in prominence. The palaeoclimatic significance of this
capped by red silt of Post-Pleistocene age. is more fully discussed elsewhere (see p. 337). Here,
I would like to suggest a very tentative schemlefor the
6. Remarks on the Soils of the Dry Belt formation of soils in Burma as follows:
In marked contrast to the present-day weathering Pluvial climate: Lateritic Gravel containing boulders and
products of the Shan Highlands. no typical laterites connected with T, and the higher slopes
were found in this region; the lateritic gravels and of Pegu Yoma.
buried lateritic crusts all seem to be fossil and to have Interpluvial: Ironstonehardpanor lateritic crust below
the T3 gravel (desert patina, ventifacts,
developed under wetter climatic conditions. It is etc.). Erosion.
astonishing to see how abruptly the soil types change Pluvial: Red gravel, sand and red earth (lateritic),
as one descends from the Shan Highlands into the connected with T2, either in fans or as
basin along one of the major roads. The most im- alluvial deposits. (Nyaungu Red Earth.)
Interpluvial: Concretionaryand limoniticsoils on T. or
portant change takes place along the main escarpment on higher surfaces. Erosion.
at an altitude of between 2500 and 1500 feet. Above Pluvial: Red small-sized gravel and sand (T4)
the 2500-foot level, lateritic red earth predominates, with red sand and concretionary silt
while at the lower elevation pan-soils and concretionary (Pagan Silt) on top.
Interpluvial: Concretionary soils. Eolian sand (Magwe
gravels prevail. Sand) and fluviatile silt (Singu Silt).
In the Dry Belt two main chemical processes seem Erosion.
to be responsible for soil formation: (1) concentration
of carbonates of lime by the action of ascending ground- It will be noted that in the above outline of soil-
water and evaporation, and (2) oxidation leading to making events, red lateritic gravel and red soil are as-
the development of red desert soils. Both processes sumed to have been associated with wetter types of
work together, but their individual range is determined climate. Now it may be argued that this need not
by the bedrock. While pan-soils are invariably asso- necessarily be so, since these alluvial formations may
ciated with silt and sand, the red desert soils are com- have acquired their lateritic nature during a succeeding
mon on the drier, ill-drained, gravelly slopes of the stage, possibly even a relatively drier one. But the very
Pegu Yoma. position of the gray hardpan and concretionary soils
As to the fossil soils discussed here, I have tentatively (Magwe Sand, Pagan Silt, and the ironstone hardpan
correlated the various types according to their strati- below the T3 gravel) in the terrace sequence is not in
graphic position (Fig. 40). Herein the soil-sheets of accord with this latter view. These soils are connected
the Irrawaddy Basin, as exposed in the terrace sections, with periods of erosion and a decrease in the water sup-
are connected with those found on the higher slopes ply. The red-colored types of sediments on the other
and highlands of the Pegu Yoma. Judging from the hand are associated with stages of stream aggradation
superposition of the younger pan-soils on the lateritic and an increased water supply. It is this intercalation
gravels in the highland gullies, one can hardly fail to of hardpan and concretionary soils within a sequence of
appreciate these correlations between a group of soils red alluvial deposits which suggests repeated interrup-
DE TERRA: THE PLEISTOCENE OF BURMA 313
tions as a result of climatic changes. If, for instance, ferred to by the last author, we saw only three (Fig.
the gravel of TX should have been weathered into a red 41). These are: the Lashio coalfield in the Upper
soil during the succeeding wet stage (T,), then it Namyau Valley,1 the neighboring Mongyaw Basin,
would be difficult to account for the general waning of and the coalfield of Namma, near Man-Se. In addi-
lateritic soils in the T4 and Post-Pleistocene deposits of tion to these we visited the basin of Mogok in the
T,, as compared with the deeply stained terrace forma- Ruby Mines District. Our visits to these localities
tion underlying T2 and T3. Furthermore if red soils were insufficiently long to permit us to make de-
had formed during the dry intervals, why should they tailed studies of the deposits. All we attempted to
be missing in the most recent alluvium of T, in which learn was the general stratigraphic character of the
we should logically expect the most complete preserva- basin fill, and its relation to the younger Pleistocene
tion of soil profiles? Here it may be well to quote an alluvium. Each of these basins presents a special set
authority on the soils of China, where analogous Pleisto- of questions of both a structural and a stratigraphical
cene formations occur. Thorp (1935-a, p. 366) states nature, which we are unable to answer owing to the
that "laterization at the present time is taking place only limited range of our observations.
in humid tropical and sub-tropical regions, as far as we One outstanding characteristic of all these basin for-
know, and for this reason we feel reasonably safe in mations is their uniform composition. The bulk of the
assuming that fossil laterites and lateritic materials are filling consists of greenish-gray sandstone, shale, silt-
evidence of former tropical or sub-tropical conditions." stone and clay, containing thin gravel beds and promi-
nent lignite deposits. The latter are confined to the
lower portion of the basin deposits. The total thick-
C. OBSERVATIONS IN THE NORTHERN
ness of the basin formation is about 1600 feet. In the
SHAN HIGHLANDS
Lashio Basin, I observed layers of volcanic ash; they
I. TILTED BASIN FORMATIONS may be derived from one of the volcanoes on the Shan
Highlands, of which the volcano Loi Han Hun (Fig.
1. The StrulctulralOrigin of the Basins
41) is the nearest. About 40 miles south of Lashio this
Tilted Plio-Pleistocene formations occur in the de- volcano has broken through the Plio-Pleistocene basin
pressions of the Shan Highlands. They have been filling in a similar manner to that in which Mt. Popa
studied and described at some length by La Touche has erupted through the Upper Irrawaddian Beds of
and Simpson (1906), La Touche (1913), Simpson the lowlands. The tuffaceous beds in the lower part
(1906) and Noetling (1891). The most recent gen- of the basin fill indicate, however, that volcanism began
eral description may be found in Chhibber's Geology 1I suggest that these Plio-Pleistocene basin deposits be re-
of Butrlmla(1934, pp. 259-269); of the six basins re- ferred to as the "Lashio Beds."

^l'~~~~ "/ff
-I^^~~ *Mongyai
Gokteik. l
6 / 1l9 10 20 30 40 50 Miles f
FIG. 41. Map of Part of th 97e orthern S98' .

FIG. 41.Map of Part of the Northern Shan States, Showing the Geological Fault-Pattern and the Principal Basins.
314 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
earlier here, and that it is associated with the formation these and fanglomerates of younger age. The former
of the basins. Except for these tuffaceous beds, all of makes a bold escarpment some 2,000 feet high, while
the other sediments are of fluvio-lacustrine origin. The the second scarp is only a few hundred feet high. It
shales and sandstones contain fresh-water shells and may be observed to good advantage along the highway
plants. The invertebrate fossils, listed by La Touche leading from Hsenwi eastward to Kontap and Holi;
(1913), suggest a Late Pliocene or Lower Pleistocene an especially good exposure occurs at milestone 16
age. Unfortunately no vertebrate fossils have thus far (Fig. 42, and P1. XI, Fig. 3), and here two fans are
been found, although they should occur, to judge by the superimposed. The lower consists of much weathered
presence of a Middle Pleistocene fauna in the sink-hole subangular debris of Namyau sandstone and shale with
and cave topography of the Mogok district. metamorphic rocks. In addition this contains quartz
Petrological analysis of one of the clayey siltstones and quartzite boulders several feet in diameter, which
found in the Lashio coal basin reveals a very close are embedded in a hard, sandy matrix. The cross-
resemblance to the siltstones of the Upper Irrawaddies. bedding is tilted southeastward, and its edges are bev-
The only difference is that in the former region lateriti- elled by a second fanglomerate. This is less weathered
zation is still proceeding. It should be noted that the and of looser consistency, containing much red loam
major portion of these basin deposits were laid down and gravel. The boulders are more waterworn and
less lateritized. No definite age can be ascribed to any
of these fanglomerates, except that the upper one should
be Pleistocene, to judge by its f-resh appearance. The
lower fan must be post-Jurassic, since it contains Nam-
yau shale, and it may well be as late as the Tertiary
in view of its position and state of petrification. Since
I it lies so close to the major escarpment, I suggest that
/F-

it was formed at the time of the basin faulting. The


.1/ A.I
Milestone i6 I IY21vlle fault runs half a mile north of the road and is marked
FIG. 42. Geological Section through the Faulted Basin at Mile- by slickensides and breccias both in the Jurassic shale
stone 16 on the Road from Hsenwi to Kontap and Holi. and in the Plateau limestone. This state of affairs sug-
1. Crystalline Limestone. gests that the basin is due to graben faulting rather than
2. Old Boulder Fan. to warping, as other observers have suggested (La
3. Young Boulder Fan.
F. Fault.
Touche, 1913). The sketch map, Fig. 41, indicates that
this faulting is determined by an old structural pattern.
in sluggish fresh water. At one time the lake level It shows the break-up of the trend-lines of folding of
must have been so low that it gave rise to extensive Late Cretaceous age (Laramide orogeny) into fault
swamps and organic soils. According to our observa- blocks of which we know only the barest outlines. The
tions there is a distinct correspondence between the faults strike from W.S.W. to E.N.E., or from W. to
large structural outlines of the Shan Highlands and E. and are part of a larger system of faults, which ex-
these basins. tends all the way across the Shan Highlands into the
The map, Fig. 41, shows that in some cases the drain- plateau of Yunnan. The main faults are between the
age follows the outline of a basin, as in case of the Plateau limestone (Devonian to Carboniferous) and
Upper Namtu River, near Hsenwi. The basin is here the Jurassic red beds. Others, as at M6ngmit, are
some four miles wide, at a point ten miles east of the between the Upper Irrawaddian formation and the
town. The valley flanks exhibit two fault scarps, one Mogok gneiss or Plateau limestone. Wherever the
between "Plateau Limestone" and red shales of the Plateau limestone is involved, as near Hsenwi and
"Namyau Series" (Jurassic), and a second between Hsipaw, the faults are marked by precipitous cliffs.

+ 200'
.0.
. o 0
.
0' ~o~~~~~~~'
. ~ 0*.0
0

:o

P1
F Ns 'I Y2Mile
FIG. 43. Generalized Cross-Section through the Western Part of the Lashio Coal Field.
PI. Plateau Limestone.
F. Fault.
Ns. Namyau Shale (Jurassic?).
L. Lashio Beds (Plio-Pleistocene).
T. Terraces.
R. Red Loam.
DE TERRA: THE PLEISTOCENE OF BURMA 315

70...
a gd
Nampon Tp
Ta ~~~LL-40
1W' o
Bridge 0? p' b

I --- 200Yards
X_J
FIG.44. Terrace Section near Nampong.
L. LashioBeds.
Ta, Tb. Terraces.

In this connection it is important to notice how the of such stream reversal is unknown, although it must
Shan Highlands border escarpment cuts across this be connected with the cutting of a steep and narrow
West-East alignment of the faults (Fig. 41). This gorge at Mongyaw through which the river breaks
obviously means that the border escarpment is of some- with rapids across the Namyau red beds. From here it
what younger date. Its strike follows the southward issues into a second smaller basin, east of Mongyaw,
bend of the eastern Himalayas and reflects a Mid- or which contains the same Plio-Pleistocene beds. Now,
Late Tertiary fold arrangement, which may actually since the stream terrace mentioned above lies above the
date back to the time when the Irrawaddy Basin was gorge, it would seem as though the two basins made
initiated, i.e., at the close of the Mesozoic. While it is one and the same depression which was once filled by
possible that the basins of the Shan Highlands date back a lake having its outlet in what is now the gorge. Up-
to the time of Laramide movement, it is quite certain lift interrupted this outlet and turned the drainage west-
that they were revived or deepened at the end of the ward into the Namtu River.
Tertiary. More interesting is the superposition of basin and
terrace deposits near Nampong, 25 miles southeast of
2. The Superposition of Old and Young Pleistocene Lashio (Figs. 41 and 44). The Nam Ma River flows
Formations in a syncline composed of Lashio Beds. They are ex-
In contrast to the tilted lacustrine beds, there are posed at the Nampong bridge and from there on down-
stream. Here, plant-bearing lake silts and clays inter-
younger Pleistocene deposits which have not been dis- calated with gravel dip to the southeast and are in turn
turbed by recent mountain-making movements. Direct
horizontally overlain by loose red terrace gravel (Ta).
superposition can rarely be observed, except at such The gravel is mantled by a thin veneer of yellow to
places where stream terraces issue out into a basin. reddish concretionary clay, and the same deposit can
Such a case was observed east of Lashio, at the west-
ern border of the "Lashio coal field." In the neigh- be seen below the rest house. Here a bouldery red
loam forms a steep slope some 30 feet high (Tb). The
borhood of the hot springs, which here mark the
border fault, an interesting section, Fig. 43, was ob- surface on which the resthouse stands is another stream
served. About one mile to the west there is a low ter- terrace, which can be followed for several miles up-
race some 10 to 15 feet above the paddy fields. It stream. There seems to be a still higher terrace, but
the dense jugle obliterates the finer details of the relief.
consists of very fine sand and yellow-gray loam contain-
In all these instances three different stages of stream
ing molluscan shells and fresh-water gastropods. A 25-
foot slope leads up to another surface underlain by red history are recorded: (1) a high terrace with bouldery
loam and a basal boulder gravel. A similar, though red loam; (2) a terrace of medium height underlain by
fine red gravel; (3) a silty to sandy low terrace. Such
possibly older, red loam was found 120 feet above the a succession of ancient stream levels is rather suggestive
rice fields. In it the limestone is completely weathered
and lateritized and only quartz pebbles are left in the of the type of arrangement encountered in the Irra-
lateritic matrix. This lateritic earth is found 200 feet waddy Basin. The superposition of terrace gravels on
above the basin and gives the impression of being a tilted Lower Pleistocene beds, and the existence of at
least three different types of stream deposits, which re-
huge fan deposit issuing from the Plateau limestone to call the sedimentary characteristics of T3, T4 and T,
the north. Here it is obvious that these fans of red
earth and lateritic gravels overlie the basin fill uncon- along the Irrawaddy River, are also analogous.
formably. At the eastern border the same stratigraphic II. PLEISTOCENEFANS AND TERRACES
relationships occur. The Namyau stream has left a
terrace remnant, some 70 feet high, close to milestone 26 One of the most notable features in the valleys and
on the road from Lashio to Mongkyet (Fig. 41). The basins of the Shan Highlands is the preponderance of
gravel is derived from the northwestern flank of the coarse red boulder gravels. Their extension is very
basin, hence the river must have flowed eastward in- great; in almost all cases this formation is found at
stead of to the west as it does at present. The origin levels lying far above the present stream channels. It
316 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
is difficult to appreciate the meaning of these phenomena
unless we attempt to analyse them more closely, and at-
tempt to correlate them .with the Pleistocene formations
of other areas in Upper Burma. To do this we may 2
choose two methods: (1) determine the relationship L

of these red gravels to existing stream terraces, a pro-


cedure fully warranted by the fluviatile nature of these e -
deposits; or (2) correlate these with fossiliferous
formations, since these are represented by cave- and/
fissure-deposits in the karstified limestone regions of
Mogok and Lashio. o
The existing literature mentions the red gravels .
under the heading of either Pleistocene or Recent forma- o
tion. La Touche (1913) as well as Chhibber (1934)? ;
ascribe them to the "Post-Tertiary Period" but refrain 1 a
from detailed classification.

1. The Boulder Gravels and Terraces of the Namtlu X


and Salzeeen Rivers/ !
The relationship of these to each other and to the oo
physiography of the northern Shan Highlands becomes o
obvious in the section shown in Fig. 45. This was sur- .-0
0
-_o
H
veyed on the road leading from Hsipaw northeastward o
to the Tati Ferry on the Namtu River (Fig. 41). o -
From a distance the hills here give the impression of o
an even surface dotted with a few hillocks and terraced
on either side. This ridge is densely forested, and? .
exposures are restricted to the numerous road-cuts C
present at the time of our visit in March 1938.
On the road from the town of Hsipaw, the first ex-
posures are encountered at milestone 2, where a small-
sized gravel is found overlain by a reddish-brown loanm
(T5). As the road climbs to the next level (T4 in/
Fig. 45), a much coarser stream deposit is found; it _
is 8 feet thick with 5 feet of red concretionary clay on i
top. The basal layer is composed of well-worn quartz,
slaty quartzite, and light gray, shaly sandstone, all of
which are members of the crystalline "Mogok Series." \.
This level is 80 feet above the valley floor. At mile-o
stone 3, slope-wash debris overlies upturned edges of 0O
Jurassic Red Beds, and two furlongs farther a boulder 00
gravel appears with an admixture of red clay. Thiso ?
dark red gravel can be seen 60 feet higher up on the
slope, and apparently it belongs to a thick alluvial
formation which underlies To on the corresponding side i
of the ridge. Unfortunately the road-cuts were not X

sufficiently deep to determine the true nature of this


gravel loam. There has been a good deal of slipping,
which may account for the admixture of stream gravel
and red loaml. IH-owever,at mnilestone4 pure red loam
is found, some eight feet thick above the boulder de-
u "
posit. These deposits are separated by a band of hard I
'
lateritic sand. The red loam is leached on top, and
it contains many concretions. From here on the road
follows a flat surface for about two miles, where there
are a few exposures, except the gravel and red loam
that are present everywhere. At places the gravel
DE TERRA: THE PLEISTOCENE OF BURMA 317

components are decomposed into lumps of soft red in Plateau limestone, and may at first sight be explained
grit still showing the original size of the pebbles. To- by a structural break-up of the massive limestone com-
ward the ferry, the same terrace remnants are en- plex along the border fault which runs parallel to the
countered as those existing on the southwestern flank south face of the ridge. However, closer inspection of
of the ridge. Beyond the stream, the road ascends these surfaces disclosed that they are associated with red
steeply and exposes subangular gravel in red loam at alluvium, very similar to the type which we observed at
about the same height as the level marked T4 on the other places where stream terraces are found. Fig. 46
section, Fig. 45. shows what these benches look like, although in nature
The four observed levels are capped by three different they are somewhat less clearly defined owing to the
kinds of terrace deposits. The lowest and youngest is dense jungle vegetation. The upper surface in this sec-
a silty loam full of fine gravel (T,), while the second tion exhibits the same type of red boulder gravel that
(T4) is composed of compact gravel and red loam (P1. was indicated in the preceding section. This is part of a
XII, Fig. 4). Both the second and the third are un- thick valley alluvium which the Namtu River deposited
derlain by red bouldery gravel. Also in this vicinity while flowing along the higher face of the limestone
there are level hilltops and small remnants of flat sur- escarpment. A lower terrace(T3?) was cut into it
faces which may well represent a still higher terrace during a degradational stage, and a still lower terrace
(?T1). Hence it looks as if we have five terraces in remnant (T,?) contains 15 feet of fine silt into which
the region of the Namtu River northeast of Hsipaw. the present stream has cut a meandering pattern. Thus
La Touche (1906, pp. 46-48), in discussing recent in both these sections there is a close measure of corre-
changes in the course of the Namtu River, claims that spondence between the number of terraces and their
they are connected with a successive alluviation in the composition with other Pleistocene deposits.
stream bed at the outlet of the stream from the moun- Another region where we saw ancient alluvium at a
tains. This sedimentation is supposed to have raised great height above the present valley floors was in the
the stream level to such a height that it caused the area between Hsenwi and the Salween River near
stream to cut through the hills along a ravine, which Kunlong. We followed the dirt road from Hsenwi
had previously been excavated by two small side streams eastward via Kontap to Holi, a small village on the
flowing into the Namtu and Nam Ma respectively. The banks of the Salween opposite Kunlong. On this route
map, Fig. 41, makes it evident that the river flowed on one follows the fault-escarpment east of Hsenwi (P1.
a semi-circular course several hundred feet higher than XI, Fig. 1) until one reaches the Salween-Irrawaddy
the present channel. Above the Tati Ferry, Jurassic watershed. This is a ridge, some 4,500 feet above sea
Namyau red beds border on the Plateau limestone. level or almost 2,000 feet higher than the Upper Namtu
This border between the two formations is an overthrust Valley east of Hsenwi. From this ridge one looks east-
along which the stream follows for approximately ward into the Nam Nim Valley, a deep V-shaped side-
forty miles. Now in Pleistocene times the river must valley of the Salween. This tributary follows the great
have followed this fault in a straight line southward, fault which extends from Hsenwi on the northern flank
where it received a tributary from the east. Flow- of the Namtu River into the Nam Nim and across the
ing across soft red shales which dip to the east, the Salween, where it bends slightly to the northeast through
river bed slid gradually in the direction of the dip, until the remarkably straight valley of the Nam Ting toward
it struck the hard limestone which is exposed on the Yunchow in the Yunnan Province of South China (Fig.
left bank opposite the ferry. It is this eastward dis- 41).
placement of the stream channel which accounts for In descending to the Salween through the Nam Nim
the position of our terraces ten miles to the west of the Valley, no clear traces of ancient terraces were seen, but
present gorge. The total amount of vertical erosion in from the road a few miles east of the pass we noticed
Late Pleistocene times amounted to over 300 feet. two high-level benches. These were level surfaces of
Already we may draw certain parallels between this interstream divides on the left bank of the river, and
terrace sequence and the one previously described in they may have been as high as 1,500 to 2,000 feet re-
the basins near Lashio and Nampang. The second and spectively. Such a great elevation would suggest that
third terraces are composed of similar red gravel and they are remnants of an older drainage, presumably of
loam deposits. The latter is lateritic and very similar pre-Pleistocene age. Red gravels of the Pleistocene
to the Nyaungu Red Earth (see p. 308) of the second Period were encountered in the lower course, between
terrace of the Irrawaddy Valley, which we also men- the villages of Man and Holi (Fig. 41). At milestone
tioned above from east of Lashio, where it covers the 48.2 huge boulder fans are cut through by the road, and
medium terrace (Fig. 43). enormous blocks of subangular debris are exposed.
Another section was studied downstream from As a good percentage of this talus material consists of
Hsipaw. About one mile and a half on the road lead- hard rocks (quartzite, sandstone, limestone etc.), it
ing to Bawgyo, one recognizes due north a step-like could not have been transported by the river. At some
arrangement of surfaces (P1. I, Fig. 2). These are places, however, there are pockets of waterworn gravel
318 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
in the talus formation, suggesting that the stream may
have swept past these talus accumulationsat a level some
200 feet above the present valley floor. The talus has
undergone intensive lateritization, and, to judge from
the thick deposit of red earth on top of it, this process
must still continue. The red bouldery formation makes
0 a notable bench on the left bank of the stream, which
becomes rather conspicuous as one approaches the con-
lo fluence with the Salween. Its thickness could not be
<Io determined because of the dense vegetation and the in-
o I?io
accessibly steep precipices above the road.
0O At the confluence point there are two terraces (P1.
I0
XI, Fig. 2). The lower one is 40 feet above the stream.
\oall At its base lies a coarse gravel 10 feet thick, overlain
01
by fine cross-bedded sand with a 2-foot veneer of red
0
loam on top. This is not a high-water terrace of recent
0
ca date because the garrison buildings of the military post
0. of the Burma Frontier Guards are on it. Also the
native village of KunlSng with its ricefields stands on
I0
this lower terrace, which proves that the present high-
o 0 water level is not as high as 40 feet above the low-water
m table. Behind Kunl6ng there is a higher terrace com-
\oo1?I cn

acd posed of coarse red gravel; its height is 70-80 feet


a0 above the Salween. It displays a much thicker deposit
II C)
Cd of red loam, but is extensively overgrown by dense bam-
boo jungle. At milestone 50, a third terrace, at least
I..
150 feet high, can be seen on the left bank of the Sal-
ween, composed of the red talus formation which we
C)
observed in the Lower Nam Nim Valley (Fig. 41).
I
H
cti Still higher, at about 400 to 500 feet high, there are
0 remnants of terraces underlain by red talus. These pre-
sumably constitute the highest of the preserved levels in
0.
0 f
?J0 this neighbourhood.
?3/ 0
0
U,
tct
On the whole it would seem as though we have here
a
4- a sequence of red gravels similar to that of the adjoining
I0 (Oo 4) E- Shan Highlands. First, there is a talus formation of
F- 0- 0
<2;sQ great thickness (at least 500 feet and more), which may
lo Ctj
5
be compared with the red boulder gravel underlying the
second and third terrace remnants of the Namtu valley.
Second, below is a finer red gravel and loam probably
Itt
representing our fourth terrace. Third, beneath this
lies the lowest level on which the Frontier Post of Holi
I is built, and this may be our silty terrace of the Lashio
.1 region (T5).
2. The Climatic Nature of the Terraces in the
Shan Highlands
The above observations lead to the following very ten-
tative outline of the Pleistocene stratigraphy in this
. ' region:
I-
1) In the Namtu Valley and the adjoining Salween
oO I drainage basin three principal types of deposits
are found: (a) boulder gravel or talus formation,
(b) coarse red gravel with sand and loam, and
(c) fine red and grey sand.
2) The thickness of these sediments decreases in the
above-named order, the proportion being approxi-
DE TERRA: THE PLEISTOCENE OF BURMA 319

mately 5:3: 1. There is a general thickening of nomenon can hardly be explained other than by greater
all three divisions toward the Salween drainage. rainfall. It might be argued that the Salween and Irra-
3) The relationship between these deposits and the waddy River levels differ by only a few hundred feet,
terrace flats is such as to suggest a regional and and therefore that erosive power must have been greater
rather uniform history connected with a geological in the Salween drainage, especially as both rivers have
period of post-Upper Irrawaddian age. the same base-level of erosion. But this difference can-
4) This period is characterized by distinct physio- not account for the thickening of the Pleistocene in the
graphical cycles with three stages of stream aggra- Salween drainage. Thus rainier climate, or climates, is
dation. Each of these is separated by periods of indicated by the nature of the terrace sediments.
erosion marking rejuvenation of stream activity The red boulder gravels themselves suggest that they
and cutting of the older alluvium. were laid down under a much wetter type of climate.
5) We believe that all of the terrace deposits reveal At the time of their deposition laterite and red loam
traces of tropical weathering and soil-making. must have formed at a far greater speed than at present,
The soils and sediments are not fundamentally because of the rapid accumulation and great thickness
different from those that are forming to-day, but of these deposits in almost all valleys. Then the land
they are thicker. Furthermore the structure sug- was coated with a thick mantle of red earth which was
gests much more violent forces of deposition than swept into the depressions by rainstorms. This slope-
that which prevails at present. The coarse and wash could not have been kept up for a long time unless
irregular composition of this alluvium perhaps the lateritic soil had rapidly formed again on the higher
indicates two things-(a) the relief may have ground, and this could have been the case only if the
been steeper, causing a faster and thicker filling rate of tropical soil formation had been greater than it
of the newly incised valleys, and (b) the climate is now. Comparative figures on recent soil accumula-
may have been rainier so that the Jurassic Red tion are available only from the Ruby Mines, near
Beds weathered away more rapidly, giving rise to Mogok, where the rainfall is in excess of 2,000 mm.
very thick accumulations of red loam. Therefore per annum. Here lateritic soil had not been formed on
the question arises as to whether these terrace mining debris dating back to the year 1882. There
formations of the Northern Shan Highlands are were signs of incipient lateritization, but no lateritic soil
due to climatic or to physiographic factors or to profile such as is found on the adjoining slopes near
both. Mogok. On slopes with a gradient exceeding 40?,
The problem of the origin of the Shan Highlands and streams of average size have removed large portions of
the red earth, and on the scars where the unweathered
Irrawaddy terraces can hardly be judiciously appraised bedrock is
unless a brief inquiry is made into other indications of exposed, no mature tropical soils were ob-
climatic changes. There is, for instance, the karst or served. From this observation we received the im-
sink-hole topography which must be briefly discussed, pression that formation of red earth today requires a
since it sheds new light on the geological and climatic long time, and that under present rainfall conditions no
thick accumulations of alluvium could take place. Since
history of this region. Before doing this, however, it
these old alluvia fill a deeply dissected relief, it is
might be well to continue with the discussion pertaining
to regional stratigraphic and physiographic correlations quite obvious that streams at that time must have been
of the later Pleistocene alluvium. Such a procedure choked by the load of sediment. The valleys were
will facilitate the study of the karstified regions and will drowned by the tropical wash products of this ancestral
enable us to focus attention on the climatic factors landscape, and the Shan Highlands were then subjected
involved. to a deluge of mud flows, talus debris and landslide ma-
Previously I have stated that the terrace formations terial. Indeed the accumulation of this material was so
show a high degree of uniformity as regards both their rapid that it virtually halted erosion, which must have
geological composition and their elevation above present been after the final uplifting of the Shan Highlands
stream levels. Such uniformity demands a set of condi- block at the end of the Lower Pleistocene.
tions which must have exerted a similar influence both The boulder gravel stage was followed by a period of
on the stream transporting power and on the soil forma- vertical erosion. The process became reversed in as
tion in widely different areas. At this juncture, we much as the streams carried away much of what had
should point out that the transporting power of these been accumulated during the preceding stage. This
streams must have formerly been greater, because in down-cutting may have been caused by either an in-
order to accumulate such heavy boulder gravels, it is crease in the water power, a decrease in the sedimentary
not sufficient to assume steeper gradients. Under pres- supply, or a sudden steepening of stream gradient due
ent conditions a narrow stream or brook, such as the to further uplift. I believe that the latter two factors
Nam Nim River, could not have amassed thick deposits led to rejuvenation, because the karstified relief gives
of alluvium containing boulders up to 20 and 30 feet evidence for a distinct decrease in the sedimentary
in diameter, even with a steeper gradient. This phe- supply at this particular phase. Here the ground-water
320 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
level was lowered, and the karst was dissected. The this type of land form also interesting for studlents of
drop in the water table was presumably caused by a ancient man and ancient animal life. The remains
general lowering of the stream floors, and this even of Peking Man and younger extinct human types in
demanded a stream power which was unhampered by China, as well as a great many fossil anthropoid apes,
excessive sedimentary load. Rivers then undercut their have recently been uncovered in fissures and caves
banks in the old alluvial formations and carried with connected with a karst relief. It was this aspect which
them a good deal of the former fill. Hence we presume made us undertake a brief survey of the Shan karst
that the climate during this particular phase was less with special reference to the finding of palaeontological
rainy, and that uplift led to further steepening of the and prehistoric human records. While it may be said
valley gradients. at the outset that no important finds were made in the
The next stage was again marked by stream aggrada- field of archaeology in this region, it is nevertheless true
tion and the contemporaneous supply of soil material. that we gained a new perspective of Pleistocene geology
The new valleys that had become incised in the boulder and palaeontology through our excursions to the Mogok
or talus formation, were filled with red gravel and loam.District.
In general these deposits are less thick, better stratified,The phenomenon of the karst is linked with solution
and there is no talus mixed with the alluvium. This of limestone by circulating meteoric waters. In our
indicates that the streams had cut down to a fairly con- region it is co-extensive with limestone inlayers of the
solidated formation which resisted any large-scale de- "Mogok Gneiss Series" and of the Late Palaeozoic
nudation. The red lateritic earth found on the medium "Plateau Limestone." The former are crystalline lime-
terrace marks a final stage in the cycle of stream ag- stones of contact-metamorphic origin, which are found
gradation. It corresponds to the valley silts, commonly most frequently along the border of ancient granite
encountered on our American rivers. At this stage the massifs, especially in the area extending from west of
streams were far less powerful than during the pre- Mogok eastward to the Salween River (Fig. 41). A
ceding period, so that when a lower gradient was finally second belt of crystalline limestone occurs southeast of
established, the river carried only the finest sediment Mogok along the Nam Pai River. The former region
in suspension, spreading it across the flat valley floor.has become famous because of its wealth of gem stones,
The climate was still rainier than at present, to judge most of which are connected with the contact between
from the great amount of lateritic earth accumulated in limestone and granitic gneiss. The most precious of
the alluvium of this stage. these-rubies and sapphires-are mined in the karst
Again dissection took place and another (third) fill- near Mogok, and extensive surface operations are still
ing followed. At that time the climate must have been carried out in this neighbourhood despite the decline in
very similar, if not identical, to that obtaining today, output and demand. The crystalline limestone varies
because the sediments found in the lower terrace
greatly as to both mineral composition and texture. It
are of practically the same order as those deposited by
may be associated with thin lenses of graphite schist
present-day streams. The components are well water- and slate, or with clusters of garnets, wollastonite.
worn, but small, cross-bedded sands and laminated silts spinel, tourmaline and epidote. It is massive near M\o-
are encountered everywhere. Incipient lateritization is
seen on the top layers, but for all we know there may gok, but slaty or brecciated as a result of faulting and
sheering movements along the contact.
well be true laterites associated with this formation. The "Plateau Limestone," being the second most im-
In general, therefore, the Pleistocene history of the
portant rock on wliich the karst topography is formed,
Shan Highlands appears to reveal the records of a
distinct climatic cycle, represented by at least three occupies probably more than half of the Northern Shan
States. The middle and upper courses of the Namtu
"pluvial" and two "interpluvial" stages. It remains to River are cut through this rock, and many of the famous
be seen whether or not and in what manner these stages
gorges (e.g. Gogteik Gorge) are developed in it. Al-
correspond to those of the Irrawaddy basin. though the Plateau limestone is full of caves and fis-
sures, it does not lend itself so easily to karstification as
III. THE PLEISTOCENE IN THE SHAN KARSTREGION does the crystalline limestone of the Mogok region.
The cause of this may be structural. Whereas the
1. General Aspect of the Karst Relief
Mogok limestone varies greatly as to both mineral
The Shan karst or sink-hole topography was observed composition and texture, the Plateau limestone is more
by us at many different places, although in the existing uniform and generally more resistant to weathering,
literature it has received little if any mention. This because of its content of dolomite and silica. The
type of solution relief is of unusual interest from a schistose and brecciated varieties of the Mogok lime-
physiographic point of view, especially since its forma- stone lend themselves more readily to physical break-up
tion involved factors of climate and drainage different on the surface and to chemical weathering underground.
from those prevailing to-day. The great number of Hence it is the Mogok crystalline limestone which has
underground shelters found in the karst relief makes produced the most typical karst relief.
DE TERRA: THE PLEISTOCENE OF BURMA 321

. X<BFaC" "v
BF

FIG. 47. Diagram of the North(ernFlankof the MogokBasin.


Ls. MogokCryst:allineLimestone.
Cz. Mineralized
(ContactZone.
Gn. Gneiss.
BF. BoulderFan.

The "Poljen" or sink-hole type of karst relief occurs so that their headwaters have cut open a good many of
northeast of Mogok (Fig. 47). Here a large, oval de- the larger sink-holes. This has given rise to a peculiar
pression is found, half a mile wide, two miles long, and sinuate shape of the valleys, many of which still bear
in places 1,500 feet deep. Its flanks are very precipi- the original form of the karstified landscape. Rivers
tous, especially along the southern slope where schist can be seen to plunge suddenly into a ravine, hidden
and gneiss form bold cliffs. The floor of this giant sink- from view by dense vegetation, and to reappear after
hole is occupied by a lake, the level of which undergoes miles of seemingly underground flow. East of Mogok,
great fluctuations with the seasonal rainfall. In winter "dolinen" or smaller sink-holes dominate the picture.
its level is so low that the ancient workings of the ruby As we are here on higher ground there is less vegeta-
mines and the irregularly pitted surface of the karst tion; thus the topography can be studied to advantage.
are disclosed (Fig. 47; P1. XII, Fig. 1). In the sum- In the vicinity of Pinpyit and Loi-sang villages (Fig.
mer the lake is said to extend for a mile, almost to Leu 48), we observed the best karst forms.
Village (Fig. 48). The karst at the bottom of the The total area we saw covers about 100 square miles.
basin with its intricate pattern of gullies, sinks and East of Lashio there is another karst region. It seems
caverns is best exposed in Mr. Nichol's mine, about one to be coextensive with the "Lashio Basin," but it is
mile and a half northwest of Mogok. At the time of connected with the Plateau limestone which actually
our visit, in December 1937, mining operations had underlies the basin. The southern flank is covered by
proceeded to a depth of 160 feet below the level of the sinks and grooved forms; these appear farther east
basin floor. It is here that the contact with the gneiss along the Namyau River. Here fluvio-lacustrine beds
may be studied to best advantage. Another poljen is cover the underlying karst, but its sink-holes are clearly
seen some 300 feet above the former basin in a small revealed on the pitted topography (P1. XI, Fig. 4).
valley north of the town. This makes it seem probable that the basin itself is a
One of the most typical of karst depressions was giant sink-hole which was subsequently faulted in a
visited near Kathe, west of Mogok (Fig. 48). The graben-like manner. Approximately 22 miles north-
filling of the basin has been removed by extensive placer ward, near Hsenwi, another and even larger cluster of
mining, and hence the irregular bedrock is well exposed. sink-holes is encountered. The Plateau limestone belt
The minutely irregular rib and pinnacle surface is here north of this town is especially rich in elongated drain-
most strikingly shown (P1. XII, Fig. 2), and close to less basins.
the mountain slope giant sinks can be seen with gaping
underground caverns. This basin merges westward 2. The Mogok Karst and Its Pleistocene Filling
into another one at Kyatpyin (Fig. 48); its length is
about three miles and the width three-quarters of a (a) Caves and Fossils in the Local Tradition
mile. As in other basins, the grooved type of karst In the vicinity of the Mogok Caves the inhabitants
formation is restricted to the bottom of the basin, relate many tales of buried dragons and underground
whereas the crevasses and larger caverns occur along spirits, which at one time are supposed to have taken
the slopes. Some five miles west of Kyatpyin, we refuge underground. The association of these beasts
reached the border of the karst. Here slope streams with the cavities presumably traces back to some sort
flowing northward to the Shweli River (Fig. 41), a of worship, but today the people are chiefly after gem-
tributary of the Irrawaddy, have dissected the karst bearing deposits: cave loam and sand. In the course
322 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

FIG. 48. Map of the Mogok Region, Showing the Location of Places Referredto in the Text.

of these mining operations the miners often find fossils, W. Granger (1923) reported from the neighbouring
teeth of elephants and deer, or other bones belonging to province of Szechwan that the Chinese actually mine
animals now extinct. To the local people fossils are fossiliferous sink-hole formations for the sole purpose
known as "naga ajo" or dragon bones. They distin- of extracting "dragon bones." Whereas no such opera-
guish several kinds of dragons, although none of these tions were witnessed by us in the Northern Shan States,
seem to fall within the range of zoological nomenclature. it is quite possible that the villagers occasionally indulge
A miner upon finding a fossil will present his find as a in this operation in order to satisfy the demands of
sort of religious offering to a near-by monastery or Chinese customers.
Buddhist shrine, and here it will be placed before an During my stay at Mogok, it was generally believed
image. In some cases I learned that fossil teeth of by the natives that I had come to search for a special
large size, such as elephant molars, are worshipped as kind of dragon bone. The result was that after a
"Buddha's teeth," but the monks themselves do not week's stay, prices for fossil bones soared, until an
approve of this practice. I presume that they are aware elephant's molar was valued as highly as a five-carat
of the fact that this is a pagan attitude, which has its ruby! This attitude did not make it easy for us to
origin in religious customs foreign to Buddhism, or at acquire much of the cave fauna. At Leu Village, where
least to the Buddhist sections of Burma. Quite possibly I made an attempt to excavate one of the larger caves,
the magic cult came from China where "dragon bones" the headman told me that years ago, near Pinpyit,
continue to play an important role in native pharma- miners had come across large bones. They had been
cology and superstitious customs. W. D. Matthew and so frightened at the sight of the huge animal remains
DE TERRA: THE PLEISTOCENE OF BURMA 323
that they gave up their work, closing the entrance with channel. We received the impression that a great
a stone wall so that the dragon might not walk out and portion of the heavy rock-material-the pebbles of
ravage their village! quartz and other crystalline rocks-was washed out of
Anybody intending to study this problem of the cave the cave loam and assembled at the very bottom of the
fauna in Upper Burma will be well advised to take ad- narrow channels through which the water runs. All
vantage of these legends and local traditions. It is in the heavy material, especially the gems, were thus de-
this manner that the first find of the Burma cave fauna posited in the lowest portions of these underground
was made. I was fortunate enough to profit from the caverns, which makes it easy to understand why the
suggestions of an "old-timer" at Mogok, the late Mr. miners concentrate on these basal sands. The cave,
A. F. Bacon, who had been a surveyor and big-game known as Dato Cave, is apparently different from the
hunter for over thirty years. It was he who had made one explored by Mr. Bacon. At least I was told by
the first finds of a fossil fauna in the Mogok Caves. our guide that the other cave lies several hundred feet
The material was described by Sir Arthur Smith Wood- above our locality. This cave also was visited, but its
ward (1917), who called attention to the presence of a entrance had been sealed by miners.
new genus of an extinct Panda (Aellrceidoplls baconi). A very interesting feature connected with this lime-
stone rock is presented by the formation of several karst
(b) Fossil-Bearing Caves floors. There are at least three at levels of 300, 700 and
My first attempt to excavate one of the caves was 1150 feet above the bottom of the ravine. The signifi-
made at Let Village, some two and a half miles east cance of this lies in the fact that ground-water action in
of Mogok (Fig. 48). The road from Mogok to limestone starts close to the surface, but it very soon
M6ngmit goes through this village, and from here a descends to deeper levels, thereby losing its connection
steep footpath may be taken which leads one 700 feet up with the local level of erosion. In working their way
the northern slope. It took us one and a half hours of downward by solving action, caverns will attain their
climbing through bamboo jungle and thorny thickets greatest size at a level still being fed by surface water.
before we reached the place which our guide called But once this connection is broken, as a result of
"Dato," meaning mercury in Burmese. Here was a gradual evaporation and absorption of the water in the
gaping fissure in the limestone; through it access could cave loam, solving action will quickly be diminished,
be obtained to a series of underground cavities in which and the caverns will no longer be deepened. At this
the people some twenty years ago had searched for gem- stage the underground karst becomes stationary, form-
bearing sand. Since mining operations had stopped for ing what may be called a "floor" or subterranean level,
so many years, the passage had become obstructed by characterized by very large chambers. Subsequently
debris. Thus it became necessary to have the passage these will be connected by anastomosing sink-holes and
cleared, and ladders installed in order to force an fissures with a lower level. To me it seemed that there
entrance to the deeper caverns, reputed to be fabu- were at least three such floors on the northern slope
lously rich in bones. Preparations were completed in near Leu Village.
two days, and we were finally able to descend. A first
A second cluster of caves was found in the immediate
cavern, some 30 feet high and 55 feet long, was struck
at a depth of 21 feet. The soil was a fine silt underlain vicinity of Mogok. Villagers and miners alike fur-
nished information about them, and occasionally they
by yellow micaceous loam with a pebbly brown sand
at its base. We could see pits and various test-diggings brought bones, although these were not always fossil-
ized. Indeed two of the caves yielded only remains of
that, as our guide explained, had been futile owing to
the lack of water, badly needed for the washing and recently slaughtered cattle and sheep. There was one
sifting of the gem-bearing sand. This he claimed was cave, however, which attracted my attention, because in
the reason why the miners had to reach a lower level, its refuse I uncovered a lower human jaw. This cave
in order that a ground-water stream could be used for (Loc. Mogok C 3) is located due east of the "Old
washing operations. A very difficult passage led us Police Lines," in a small attractive valley north of the
some 35 feet deeper, and here a trickle of ground water town of Mogok on the land of a certain Burman, called
was encountered. This seemed to issue from the very Maung Tung, of Mogok (Fig. 47 and 49). In front
fissure through which we had descended. It had been of the cave entrance some 50 feet of deposits had been
sufficiently powerful to wash away part of the cave loam, exposed by a ditch, cut by the miners through the an-
and residual pebbles and lumps of sand had accumulated cient mine refuse. Here various coarse gravels and
along its course. It was here that we collected the talus layers could be seen. These were not cemented,
fauna which Dr. Colbert (see p. 417) has described and they appeared to belong to some sort of alluvial
(Loc. Mogok C 1 of our collection). We found teeth deposit which had preceded the mining operations.
of pig, deer, a medium-sized ruminant, porcupine and They were certainly younger than the Pleistocene for-
rhinoceros. The fossils are scarcely waterworn, prob- mation found in near-by karstified limestones. My
ably because they were first fossilized in the cave guess is that the gravels are of post-Pleistocene age, a
loam and later on washed out into an underground fact also suggested by the fauna from this locality (see
324 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
p. 422 of Dr. Colbert's report). As the jaw was found streams and plenty of game, in addition to fertile loamy
on top of the gravel deposit, it is not certain from where soils covering several square miles of flat ground at the
it actually came, but its position was only 40 feet from valley bottom, would have offered plenty of induce-
the cave entrance. Professor E. A. Hooton of Harvard ment to early settlers. Here, the chase could have been
University informs me that the jaw exhibits no remark- combined either with food-gathering or with argricul-
ably primitive morphological features, and that in his tural practices.
opinion it represents a brachycephalic individual. Ac- The Mogok caves did not yield any Pleistocene fos-
cording to Dr. Hooton, it is a very low jaw and prob- sils, because intensive mining has not even spared the
smallest fissures. Obviously there must be some that
have been left untouched, but these may lie high above
the valley along the limestone cliffs, visible about 1000
to 2000 feet above the town. From such a fissure,
reputed to lie a few miles north of Mogok, villagers
brought perfect molar teeth of Elephas naImadicus,to-
gether with large fragments of well-fossilized leg-bones.
A third and rather more important group of caves
was found in the vicinity of Kathe, a village some nine
miles by road west of Mogok. At Chausong (Fig. 48),
about half a mile north of Kathe, I visited a cave (Loc.
1 140' Mogok C 2) which yielded teeth of Stegodon orientalis,
FIG. 49. Section through the Post-Pleistocene Valley Fill
Hystrix, Elephas namadicus, Rhinoceros, and Bos (see
near Mogok. p. 417 of Dr. Colbert's report). These were uncovered
in a gravelly sand found below reddish cave loam in a
ably belonged to a female. Hence, it may well be the roomy cavern, some 25 feet high and approximately
jaw of a prehistoric cave-dweller, such as must have 32 feet underground, which opened on the valley. The
lived here for many years before the present peoples entrance faced south and led through a wide rocky fis-
settled in this area. In this connection it is interesting sure which still contained patches of the ancient cave
to note that a great many stone implements of Neolithic filling; at one place about 90 feet from the entrance a
type have been collected by local surveyors in the vi- bone breccia was seen. The cave floor consisted of
cinity of Mogok, and most of them are reputed to have reddish loam, 4-6 feet thick, underlain by sandy gravel,
come either from the surface of old lake terraces, ap- and between the two layers there was a thin stratum
proximately 2 miles east of the town, or from the en- with limonite pellets. The gravel is composed of per-
trances of caves. Some of these caves are still inhabited fectly rolled pieces of quartz, gneiss, slate, and mica
by Buddhist hermits, who have installed shrines in them. schist. These pebbles are about a quarter to half an
In one instance I was told of an abandoned cave which inch in diameter, and they belong to formations which
has been used as a cemetery, but whether this was in occur in the immediate vicinity. Fig. 50 illustrates the
recent or in former times I could not determine. In cave's position with relation to the karst topography. I
any case the brief inquiry which I made in this matter was told that another floor of caves existed some 400
resulted in the discovery of an underground burial; feet higher up the slope, which would indicate that here
unfortunately, however, we were forbidden to investi- also several galleries of underground karst exist.
gate it. There is no question that the first people to This suspicion was confirmed by a visit to the ad-
settle in this area took refuge in the caves, because most joining basin near Kaungba. Here a narrow fissure,
of them face a valley that must have offered a most which leads to huge caverns of complicated pattern, was
favorable habitat in prehistoric times. A lake, several seen at a height of 420 feet above the valley floor. Vil-

FIG. 50. Generalized Cross-Section through the Karst Basin of Kathe, West of Mogok.
1. Mogok Crystalline Limestone (Karstified).
2. Karst Filling.
3. Fan Formation.
* 4. Silt and Organic Soils.
DE TERRA: THE PLEISTOCENE OF BURMA 325

lagers told me that they had recovered fossils from these valleys. As regards the age of these sinks, it is obvious
places, but none of them produced any proof of this. that their origin is closely linked with that of the other
at least fourteen caverns existed in that neighbourhood karst formations found associated with them. Judg-
of Kyaukpadaung, north of Kathe. He reported that ing by the great depth to which they descend and by the
at least fourteen caverns existed in that neighborhood, amount of sediment they contain, they would seem
some of which were said to be 1500 feet above the to represent a rather long geological period. Invari-
valley floor. ably they are associated with a mature karst topography,
which indicates that they may well represent a stage
(c) Sinks and Fissures of optimum conditions for karst formation. This does
East of Mogok, on the road to Mongmit, and near not mean that all fissures are of the same age, but it
the village of Pinpyit, sink-holes dominate the relief. makes it appear very probable that the majority of them
Their height is about 1000 feet above Mogok, and they were formed during one of the major pluvial phases of
belong to an undissected karst landscape where sinks the Pleistocene. In view of the fact that the cave fauna
and poljens are still in the making. These sinks can is of Middle Pleistocene type and related, as Teilhard
always be recognized from the distance by clusters of de Chardin (1938) and Colbert (see p. 425), point out,
vegetation, because near them water accumulates in the to the Middle Pleistocene fauna of China, it is very
soft loamy earth (P1. XII, Fig. 1). Thus, while the
likely that the fissures containing the fossiliferous de-
walls of sink-holes are barren and completely devoid of
posits are of a corresponding age.
soil, the floors are filled with thick deposits of terra
rossa and limestone debris. The gaping clefts usually 3. The Physiographic Cyclc of the Karst
lie in an excentric manner and are often not wider than
one foot. Ten feet underground they may widen into The peculiar land forms encountered in the karst of
huge caverns, or into larger fissures through which water the Shan Highlands indicate the impacts of varying geo-
percolates. In some cases such fissures have been cut graphic factors on the relief. In it the large basins
open subsequently by stream erosion or mining opera- represent the most ancient stage. Some of these may
tions. At the surface there is an abundance of grass be pre-Pleistocene, such as the Lashio Basin where the
and humus that is being constantly washed into the fis- sedimentary filling is of Upper Irrawaddian age. Oth-
sure. In former times grazing creatures, such as ele- ers, like the Mogok Basin, may be younger, but still
phants, rhinoceroses and deer were trapped in the clefts, sufficiently old to contain Lower Pleistocene beds.
where they died, and portions of their skeletons found Such considerations of age bring up the question of the
their way underground. In some instances, however, appearance of this land-surface during the Tertiary
it is difficult to see how large limb bones of elephants Period.
could have been swept into such narrow underground Undissected remnants of a pre-Pleistocene relief
passages. At Pinpyit, a local miner showed me a fis- were seen north of Hsenwi. Here the Plateau lime-
sure at the bottom of a large sink-hole from which he stone presents an undulating surface with broad basin-
emerged with large fragments of elephant remains, be- like valleys framed by limestone ridges 1,000 to 3,000
longing to the Pleistocene type: Elephas namadicus. feet above the valley floors. In the Upper Namtu Val-
Other fissures were found some 200 feet above this ley this limestone plateau is broken off by a steep escarp-
sink-hole. They lay on a ridge of limestone, which pre- ment (P1. XI, Fig. 1); this may have been non-existent
viously had separated two big sink-holes from each or only faintly outlined in Tertiary times. Drainage
other. In this case the rock was fissured for an area must have followed such fault-lines, as indicated by the
over 800 feet square, but none of the clefts were larger present course of the Namtu River, which at places
than 10 inches in diameter. keeps close to the boundary between the limestone and
At Mogok, sinks were seen in Mr. Nichol's mine; one the Jurassic red beds. Except for slope streams, re-
of them was 26 feet wide and issued from a karstified stricted to either flank of the Shan Highlands Block,
valley floor. It contained the original fill of gravelly drainage consisted chiefly of two types: an open stream
sand and loam, and I was told that bones had been ex- pattern, and a closed drainage in the karst and faulted
tracted from this place. This fissure served as an depressions. Just as at present, open streams mean-
underground drain for a large basin some 50 square dered across this low relief, but their courses had low
miles in extent. It lay at the contact between marble gradients and no very deep gorges then existed. Even
and gneiss. The fissure deposit consisted of irregular during this early period, the same contest between the
layers of gravel, talus, sand and loam, containing quartz, two types of drainage must have occurred as at present,
slate, gneiss, and other well-rounded pebbles. Similar causing the consumption of an existing karst topography
material was observed in the mines at Kathe. Here a through normal stream work.
surprising amount of coarse gravel has been removed When the Shan Highlands experienced uplift,
from the karstified bottom of the valley by mining opera- streams must have cut vigorously into their channels,
tions, indicating that a conglomeratic alluvium of con- leading to rapid erosion in the headwaters. This
siderable thickness must have once filled some of the brought about widespread stream capture of such closed
326 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
drainage systems as existed in the karstified regions. ing gradually consumed by headward erosion. At this
Slope drainage advanced from the Irrawaddv Valley point, the karst cycle is interrupted by stream erosion.
on the east and through the Shweli River on the north, This is the present stage.
causing intensive dissection of the old upland. Thus The two physiographic cycles preceding the present
the limestone areas were karstified with renewed vigor stage are connected with uplift and pluvial conditions.
during the various pluvial phases of the Pleistocene. Through the uplift of the Shan Highlands Block, normal
Only a few of the highland basins and karst regions rainfall increased. In addition more rain was precipi-
were left untouched, and it is here that the physio- tated during stages of lower temperature, such as are
graphic history can be readily interpreted through the indicated by the lowering of the snow-line in the Hima-
geology of the basin sediments. Obviously, such basins layan regions. Elsewhere (see p. 337 and Fig. 54, p.
registered better than any other regions the various 334), I have discussed the circumstances which led
phases undergone by the Shan Highlands during the to increased precipitation in the lowlands adjoining the
Pleistocene period. Two physiographic chapters are glaciated massifs. There is no doubt in my mind that
of special interest to us: the impact of the Late Tertiary the combined effect of the first broad uplift and the
uplift on the existing relief, and the influence which pluvial climate, prevailing at the beginning of the Pleis-
the Pleistocene climate exerted on the undissected por- tocene, initiated a new cycle of karst formation. To
tions. this first stage may belong the large poljen and basin
The first chapter may be interpreted by the fault- formations. We can say practically nothing about the
escarpments, which we mentioned previously. Along exact age of these features, for no Lower Pleistocene
the larger sink-holes, such as at Mogok, these escarp- deposits have so far been found in the karst.
ments and fault-line scarps are frequently found asso- The next stage was one of internal dissection and
ciated with boulder fans. While the fault pattern may much intensified karstification. I have indicated above
be of earlier date, there is little doubt that faulting that this may have been during the Middle Pleistocene
occurred again along the major formation boundaries Pluvial Stage, as indicated by the fossils thus far re-
when the entire land mass was uplifted. The ensuing covered from caves or fissures in the limestone regions
stream rejuvenation must have affected the shape of (see p. 425 of Dr. Colbert's report). In most cases
the large existing basins and sink-holes. The latter such fossil-bearing formations are associated with the
were then part of the undulating and rather level mature karst. Hence there must be a genetic rela-
topography. Down-warping and accelerated deepening, tionship between this special type of karst and the fossil
as a result of increased ground-water activities, led to fauna.
a kind of internal dissection of the larger sink-holes, In another publication (1939-a) I have pointed out
such as is illustrated by. P1. XII, Fig. 1, taken in the that the greatest climatic pessimum in the Pleistocene
Pinpyit region, near Mogok. At this stage small sink- followed the first period of diastrophism. This was the
holes formed within the larger depressions. They al- phase of tilting in the Upper Irrawaddian formation.
tered and scooped the interbasin divides, destroying Its beds are unconformably overlain by boulder-bearing
whatever was left of the original flat surfaces. The gravels in the Irrawaddy Basin, whose origin we be-
water collected rapidly at the bottom of these narrowed lieve is connected with a rainy climate. This occurred
depressions, leading to an intensified solvent action at in the early part of the Middle Pleistocene period.
their bottoms. This was the period when many of Now, in view of the type of fauna which we find in the
the animals may have been trapped. In other words, karst formations of the Shan Highlands, I do not hesi-
the old sink-hole topography, as contrasted with the tate to correlate the deposits containing the Stegodon-
younger and fresher--lookingpits on undissected ground, Elephas-Hystrix fauna with the coarse red gravels
is of Middle Pleistocene age. At that time the karst underlying the highest of the Irrawaddy terraces. It
"matured,"as a result of another widening of the freshly was then that the existing karst relief was intensified.
formed lower sink-holes which was in progress. This With fluctuating rainfall this process of underground
is the second stage of karst history. solution of the limestone rock may have passed through
The third stage is indicated by a general rejuvenation various stages of karstification. At one stage, with
of the stream pattern, and a widespread capture of the greater rainfall, the process was quickened, and at
closed drainage. This modern dissection works both another it was retarded. This is actually recorded in
in the major stream valleys as well as in the under- the type of sediments and in certain erosion features,
ground channels. Starting from the tributaries, this which I shall presently discuss. Here it is important
erosion leads first to a lowering of the local water-table to note that the main period of deposition, which led
and an intensification of solvent action in the immediate to a virtual choking of karst drainage with soil, took
vicinity of the streams. Underground passages are place in the Middle Pleistocene. This is testified by
being tapped and free circulation of water is initiated the fauna.
where formerly only underground solvent action pre- I hazard the opinion that this was the stage of the
vailed. Some sink-holes have collapsed, others are be- second Himalayan glaciation. At that time the rainfall
V
s^

FIG. 51. Map of Southeastern Asia Showing the Distribution of


Middle Pleistocene Karst Formations.
1. Mogok, Burma. 5. Kwangsi Caves, South China.
2. Tam-Hang, Indo-China. 6. Wanshien Karst-Szechwan Pits-Yangtze Valley.
3. Lang-son, Tonkin, Indo-China. 7. Choukoutien, near Peking.
4. Hoshangtung Cave, Yunnan. 8. Patjitan Karst, Zuider Mountains, South Central Java.
327
328 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
was most plentiful and weathering so intensive that it eritization. Its surface is dissected by a second fan of
caused the formation of sink-holes. This process must deep red color, containing sandy layers and larger
have resulted in the temporary sealing of the karst fis- boulders of subangular shape which are much less
sures with sediments, and it took place at the close of decomposed than in the former case (P1. XII, Fig. 3).
the mature cycle of karst erosion. The event was not The disconformable contact between the two fans is
localized in Burma, for, to judge from the regional ex- exposed at many places in the large surface diggings
tent of fissure formations bearing the same type of south of the road to Leu Village. The best exposures
fauna, it would seem as though similar conditions also are about a quarter of a mile east of Old Mogok. On
prevailed over most of South China. Fig. 51 illus- the basis of the differences of altitude between the upper
trates the distribution of the fossil-bearing, Middle and lower exposures (Fig. 52), it would seem as though
Pleistocene fissure deposits in Southeastern Asia. All this contact was steeply inclined towards the basin. It
of them suggest that the climatic control of karst forma- indicates an erosion interval prior to deposition of the
tion was exercised more or less simultaneously over younger fan, an observation which is significant in con-
the entire area. nection with the question of the climatic control of
It now remains for us to describe briefly the various weathering. The far-reaching decomposition of the
deposits connected with the karst formation in Burma. older fans could not have taken place unless this
Offhand, we might expect that the sink-holes and allied formation had been exposed to agencies of tropical
land forms are filled with the weathering products of weathering for a long time. The old fan must have
the karst landscape. been lateritized before the younger boulder gravel was
formed. Then at a much later time another erosion
4. Sedimentation and Correlations period occurred, which was succeeded by a general
filling of the basin with fine gravelly sand and silt.
(a) Basin Deposits. These deposits are exposed north of the road, and
The basin fillings of the type found at Mogok are they extend for at least one mile east of Mogok. The
most conspicuous. Fig. 52 indicates the following types fine laminae found in them indicate a fluvio-lacustrine
of sediments: (1)' old fanglomerates, (2) young fans, origin. A typical exposure in these sands reveals the
and (3) post-Pleistocene sands. The latter occupy following sequence: gravelly sand at the base, some 4
rather extensive stretches on the northern flank of the to 6 feet thick, overlain by cross-bedded and laminated
basin, while the fan-formations appear on the other side sand, 9 to 12 feet thick, and on top fine gray silt which
at the foot of precipitous cliffs. It is here that mining has undergone varying denudation. This formation
operations have exposed a most complex sequence of contrasts remarkably with the lateritized boulder fans
coarse debris piled up 300 feet high at the foot of the on the southern slope against which the sands rest.
cliffs. As Fig. 52 shows, there are three types of Fig. 50 indicates that it covers the karst and the an-
detritus. In the topmost quarries, there appears a cient red filling.
loamy, reddish boulder deposit in which all the rock This three-fold division of the basin deposits into an
components are completely disintegrated. Boulders of old, coarse boulder fan (highly weathered), a younger
gneiss and quartzite range from 6 to 15 feet in diameter, boulder fan (less weathered), and a loose gray sandy silt
and they are interbedded with sandy red loam. The deposit, is encountered practically everywhere between
shape of the boulders is preserved, but their state of Kathe and Lashio. At Kathe the exposures are espe-
decomposition is such that one can easily cut them with cially clear, because mining operations have left a con-
a knife. It is a fanglomerate in a state of complete lat- siderable portion of the basin filling undisturbed. The

FIG. 52. Geological Cross-Section through the Basin Fillings at Mogok.


1. Old Boulder Fan-Deeply Lateritized.
2. Young Boulder Fan-Less Deeply Lateritized.
3. Fluvio-lacustrine Sand and Silt Underlain by Limestone Karst.
DE TERRA: THE PLEISTOCENE OF BURMA 329
sands here contain organic soils, and a peaty layer with ondly that the karst formation itself is related to the
plant remains some 8 inches thick. This sort of sedi- deposition of boulder fans. That the karst at the bot-
ment would be expected in a fluvio-lacustrine formation, tom of the basins might be contemporaneous with this
which owes its origin to sluggish flowing or ponded formation is indicated by our observations on the cave
water. The total thickness of the karst strata is 90 fillings. These repeat a pattern similar to that found
feet; of these about two-thirds are of fluvio-lacustrine in the karst itself. At the base lies a gravelly sand,
origin. which may well correspond to the boulder fan stage.
The cave loam might represent the younger fan stage
(b) The Origin of the Karst Gravel and Loam of the open karst, but a great many detailed observations
In all the mining pits where the karst filling is well would be required to establish such correlations.
exposed, a great deal of gravel is encountered. Its con- As to the climatic interpretation of the karst deposits,
stituents are well rolled and consist of gneiss, quartz, it should be noted that the deeply weathered boulder
schist, and marble. All of these are derived from the fans surely required intensive slope-wash under a rain-
immediate vicinity, and the miners relate that they are ier type of climate than exists at present. That there
most frequent in the deeper or basal portions of the are two types of fans seems to point to at least two
karst filling. They may be residual gravels washed out such rainy or pluvial stages, especially as both were
of the overlying fans or slope debris. The fans, being interrupted by erosion, which presumably was the re-
rather pervious, may well have acted as percolators in sult of a dry interval. It is suggested that these two
such a manner as to have permitted underground solu- stages may be correlated with the gravels of the first
tion and the removal of silt and clay. Since the and second terraces in the Irrawaddy Basin. In such
karst level was deepened at intervals, fanglomerates a case one could assume that the third type of karst
spreading over the basin from the slopes must have sediment, the gray sand, represents the fourth terrace
frequently suffered collapse, both from the underground gravel of the Irrawaddy. This, however, cannot be
yielding of the karst caverns and the washing out of proved, and it might well be that the lacustrine deposits
the silt and clay particles. This no doubt led to mass containing organic soils belong to a post-Pleistocene
movement and gravitational pull, which forced the stage. In this connection it is interesting to point out
heavy material to the bottom of the sink-holes. an apparent relationship between the finds of polished
The effects of erosion on such fans is so great that a celts and other artifacts and the surfaces of the lacus-
single rainstorm may produce a new surface relief on trine formation, mentioned on p. 324. A thin veneer
the fan. This observation might suggest that the Pleis- of loam on the gravelly sand and silt possibly permitted
tocene gravels had been repeatedly redeposited, espe-
Early Man to practise agriculture.
cially since rainfall must have been heavier during the In conclusion we may say that the formation of the
pluvial stages. However, the subangular state of the karst was largely brought about by the pluvial periods
boulders and the uniform degree of weathering which of Pleistocene Age, and that there were at least two
is hardly conceivable under repeated reshuffling of the
periods of optimum conditions that left visible traces
components, argue against this contention. The fans in the relief and in the sedimentary formation. Such
were static as a mass, but they reacted internally by the
records of Pleistocene pluvial climates surely date back
slipping movements of the heavier material. Such in- to Middle and Late Pleistocene times. Also there
ternal collapse no doubt loosened the structure and
seems to be a direct correlation between a sedimentary
permitted small subterranean rivulets to wash gravelly
sand into the gaping fissures and pits, where ground- cycle in the karst and a physiographic cycle of the ad-
water circulated. Unable to penetrate very deep into joining Irrawaddy Valley, which is little understood
the underlying limestone rock, these rivulets may have at present.
A comparison between the two sets of data obtained
transported a certain amount of gravel underground in the Shan Highlands and in the Irrawaddy Basin re-
through the existing funnels and passages. Such an veals certain analogies and certain differences. Analo-
explanation of karst sedimentation is supported by the
picture presented by the mining operations found in gous, in our opinion, is the stratigraphic composition of
the karst pockets. Here the majority of the heavy Late Cenozoic formations in both areas. This is indi-
cated by: (1) an unconformity between the tilted Lower
minerals, such as garnet, spinel, ruby, hematite, etc.,
is encountered. The miners are quite aware of the fact Pleistocene and the terraced Middle and Late Pleisto-
that nature once washed these minerals out of the cene beds; (2) the presence of weathering and erosion
bedrock in much the same way as it is being done products resulting from tropical pluvial conditions; and
nowadays by placer-washing. (3) similar composition and a seemingly equal number
From these observations two conclusions become of terraces. On the other hand, there are two signifi-
clear, both having to do with the formation of karst. cant differences between the geological data of the two
Firstly, that the irregular rib and pinnacle surfaces (P1. regions. One of these lies in the fact that the Middle
XII, Fig. 2) are intimately connected with the contact Pleistocene fauna of the Shan Highlands is as yet un-
between the limestone and the boulder fans, and sec- recorded in the terrace formations of the adjoining Irra-
330 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

waddy tract. This suggests that possibly the former nant role. The same appears to the true of our region
region constitutes a westernmost outlier of the Pleisto- in Burma.
cene fauna of southeastern Asia. The other difference The analogies between the two areas are of the fol-
is in the geological nature of the soil. In the Dry Belt lowing order.
of the Irrawaddy Basin there are alternating layers of
lateritic and calcareous soils, while in the more humid 1) The Pleistocene is divided into two major divi-
Shan Highlands only lateritic soils are found. This sions; a tilted and folded Lower Pleistocene series
containing a fauna of Villafranchian type, and a
might indicate that the climatic zoning of Upper Burma
was somewhat different during the Pleistocene Period slightly tilted but frequently massive and terraced
Middle and Upper Pleistocene.
from that obtaining today, since at present no lateritic
2) A major angular unconformity between the Lower
soils are being formed in this now arid region. It may
and Middle Pleistocene.
well be that the Pleistocene lateritic gravels of the Irra-
3) An equal number of terraces revealing a similar
waddy Basin are represented by the two boulder fans
in the highlands. These differences are not sufficiently geological composition.
4) The presence of fossil soils and of soil records
great, however, to detract from the essential agreement
between the two regions as far as the Pleistocene rec- suggesting the impact of four major glacial or
ords are concerned. pluvial periods.
5) The first appearance of Lower Palaeolithic imple-
ments in deposits of second glacial or second
D. THE PLEISTOCENE HISTORY OF BURMA pluvial and second interglacial or interpluvial age.
IN RELATION TO THAT OF NEIGH- Such are the major relationships between the Indian
BORING REGIONS and Burmese Pleistocene formations found within or
When we compare the data described above with those near the Siwalik Belt. In detail there are certain differ-
previously obtained in India, we notice a close agree- ences, which we need not discuss here in view of the
ment as far as the climatic and diastrophic records of incompleteness of our studies as far as the geology of
the Pleistocene Period are concerned. As it was one both regions is concerned. But there is one problem
of the chief objectives of this expedition to detect analo- which is of considerable interest: the age of the loess
gies as well as differences in the environmental condi- of northwestern India as compared with the loessic
tions of Early Human cultures in southern Asia, such deposits of Upper Burma and China. In India the so-
comparisons are of special value. This aspect is fur- called "Potwar Loess" was considered to be of third
ther enhanced by supplementary data gained by the ex- glacial age while in Burma the "Pagan Silt" (being
pedition members in Malaya, Java and China. In the the only loess-like formation encountered by us) is defi-
last-named country, Dr. P. Teilhard de Chardin (1937) nitely connected with the fourth terrace, which we hold
had previously attempted correlations which throw a sig- to be of fourth glacial age. Previously (1939) I have
nificant light on the prospects of introducing a new called attention to the possibility that the loess of north-
chronological element in the reconstruction of human western India may have been formed in later times as
prehistory. well. For instance, the fourth terrace in the Potwar
region is composed of reddish, often laminated, silt, the
1. COMPARISONS WITH INDIA origin of which is partly eolian. Also, I have shown
that in the Kashmir Valley loess was formed at the time
By virtue of certain analogies in structure and geo- of prehistoric occupation (1939, pp. 231-235), and a
graphical position, the Dry Belt of Upper Burma pro- much older loess occurs in the Middle Pleistocene "Up-
vides a setting similar to that found in the Siwalik
per Karewa" formation. All this suggests that in In-
region of northwestern India. The two chief structural dia, as in Central Europe, there are a great many dif-
elements are the same. The Shan Highlands, an up-
ferent loess deposits. Although each seems to have spe-
raised faultblock flanked by a geosyncline in the lrra-
cial characteristics, it is as yet not possible to correlate
waddy Basin, would in the case of India find its analogies
in the Salt Range and the Potwar lowlands. In both in- any of the loesses or loess-like formations of eastern
Asia with those of Europe.
stances the highlands receive monsoon rains, while the
The accompanying table of correlations is tentative,
adjoining lowland is comparatively dry, which leads to but in it we have given our interpretation of such cor-
more or less constant erosion in the uplifted block and
relations as can be confidently made with the new in-
to deposition in the adjoining lowland. This geological
formation from Burma.
process of denudation on one hand and deposition on
the other underwent a series of cyclical changes during
2. COMPARISONS WITH CHINA
the Pleistocene, which I have discussed elsewhere
(1939). The nature of the Pleistocene sedimentation The correspondence between the general stratigraphic
was the result of an interference between diastrophic pattern as found in the Middle Yangtze Valley and in
and climatic processes, with the latter playing the donli- the provinces of Yunnan and Kwangsi on one hand, and
DE TERRA: THE PLEISTOCENE OF BURMA 331

TABLE 1
TENTATIVE CORRELATION OF PLEISTOCENE DEPOSITS IN INDIA, BURMA AND CHINA

INDIA BURMA CHINA

PERIOD
Siwalik Hills & Irrawaddy Shan South China North China Cave
Indus Valley Basin Highlands Yangtze Valley Plain Sequence

Silt Terrace (T.) Lowest Terrace Lowest Terrace Lowest Terrace Panchiao Terrace
Post- (T5) (T5)
Pleisto-
cene Potwar Loam Singu Silt Lateritic Loam
Magwe Sand

Gray Gravel and Small Red Gravel Red Gravel and Shiashu Terrace Malan Terrace "Upper Cave" at
Loam (T4) & Pagan Silt (T4) Gray Sand (?T4) (Yellow Loam) with Loess Choukoutien
? Potwar Loess

Main Terrace (T3) Main Terrace (T3) Terrace (Ta) Yiihuatai Terrace Chingshui Terrace
Q
Erosion Erosion Erosion Erosion

a Terrace (T2)
Terrace (T2) Red Gravel and Younger Lamprotula CHOUKOUTIEN
Gray Gravel and Lateritic Soil Boulder Fan? Neichiang Terrace Terrace DEPOSITS:
Potwar Loess (Nyaungu Red Locality 15
Earth)

U
Terrace on Terrace on Red High Terrace of
I) Boulder Boulder Gravel Ichang Locality 1
Fans (T1) (T1)
1-
c4 4->
cn

01 Choukoutien
Red Loams (? or Szechwan-
Boulder Uru Boulder O
bio
Red Loess) and Yunnan Caves
Conglomerate Conglomerate 0 Boulder Fan Fissure Deposits Choukoutien:
(Upper Siwaliks) Upper Sanmenian Loc. 13
Beds
------UN( CONFORMITY- .-- - 4 -?-- -AC

Pinjor Beds
(Upper Siwaliks) Upper Lashio Horse Beds Lower Sanmenian

Irrawaddian of (Nihowan) Choukoutien:

Tatrot Beds Beds Beds Yunnan Beds Loc. 12


(Upper Siwaliks)
--?- - - ,,,,,,,,, CCCCV UNCONFORi VITY-

the Shan Highlands on the other is very striking. In Even in the detailed composition of the single strati-
South China, there exists a series of Pleistocene Beds graphic units and their landscapes, there is close corre-
which can also be divided into two major groups. Dr. spondence. Near Lashio, the rounded hills of tilted Late
Pliocene to Early Pleistocene gravels look strangely simi-
Teilhard de Chardin, in a memorandum submitted to lar to those encountered in the Yungning (Nanning) Basin
me, comments as follows: of Kwangsi. It is here where tilted Pliocene sands and
The Late Cenozoic conditions observed on the Shan clays are covered by a lateritized fan of quartz gravels
Plateau are strikingly similar to those in South China (see Teilhard de Chardin, Young, Pei, and Chang, 1935,
(Yunnan, Kwangsi). Dissection and tilting of Late Plio- Fig. 2). The section given (Fig. 44) of the Mogok basin
cene lake-deposits; subsequent accumulation of thick gravel duplicates exactly the section of a tin mine in northern
formations; extensive lateritization of the soils (at least Kwangsi (Teilhard de Chardin, Young, Pei, and Chang,
up to the close of the Pleistocene); and occurrence of fis- 1935, Fig. 8). Once these regions have been studied in
sure formations with a Stegodon-Ailtropus fauna. greater detail, it may well be that the lateritized fans of
332 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Kwangsi, Hunan, Kiangsi and Hupei will reveal terrace and Granger (1923) and others from sink-hole and
sequences similar to those found in the Namtu Valley of fissure deposits, which was originally interpreted as
the Shan Highlands.
Obviously, the Plio-Pleistocene history is essentially the being a Late Tertiary mammal assemblage. However.
same over wide regions of continental Asia extending east according to Teilhard de Chardin (1937), this view
of the Irrawaddy up to the Middle Yangtze Valley. And must be revised, and Colbert also has pointed out in
we may even observe that detailed analogies exist as far as his report, that the karst fauna is of Middle Pleisto-
the physiography of the two powerful streams, the Ir- cene age. In view of the close relationships existing
rawaddy and the Yangtze, are concerned. As is the case between the fissure fauna of Burma and that of Sze-
of the former stream, the Yangtze, east of Ichang, flows
across a thick series of tilted Cenozoic clays, sands and chwan, and because of the presence of a Middle Pleisto-
gravels, from which are derived a large portion of the cene faunistic element (Elephas namnadicits),it is now
Pleistocene sediments. Furthermore both the Irrawaddy very certain that the major karst filling is of that age.
and the Yangtze are skirted by conspicuous remnants of a In this case the conclusion that Barbour's highest
30 m. terrace (the Yuhuat'ai terrace), and remains of older
river gravels (?T1) are found near Ichang (Teilhard de Yangzte terrace should be of Middle Pleistocene age,
Chardin and Young, 1935). To be sure the Early and provided that its origin can be related to the formation
Middle Pleistocene formations of the Yangtze are more of the Szechwan karst, cannot be avoided. Thus a close
lateritized than are those of the Irrawaddy, but this differ-
ence may be apparent only. As noted above, such strongly correspondence should exist between the terrace se-
lateritized formations as those found near Minbu in Burma quence of the Irrawaddy and Yangtze valleys. Obvi-
are very similar to those occurring between Nanking and ously such a correlation goes a long way toward clarify-
Ichang. On the other hand, true concretionary red clays, ing the climatic factors involved in the Pleistocene his-
presumably identical with those of the Irrawaddy, occur tory of the great belt of lowlands extending from the
along the northern border of the Yangtze (southern foot of Indus Valley in northwestern India eastward to south-
the Tsinling Range). Indeed this is a most curious and
probably not fortuitous parallelism.'
To these remarks we must add certain information
which Barbour (1933) collected in China. He de- \*c _ ^3A60'
A60
scribed from the middle course of the Yangtze, espe-
cially from the Red Basin of Szechwan, three river ter-
races. These lie 12 m., 30 m. and 80 m. respectively
above the river and are associated with gravelly and FIG. 53. Generalized Cross-section through the Pleistocene
loamy sediments. They are so similar to our own se- Valley Fill in the Tin Mine Region East of Port Swettenham,
Malaya.
quence of alluvium in Upper Burma that it would seem G. Granite-Gneiss.
most probable that our second to fourth terraces (T2- S. Schist.
T,) correspond to the levels mentioned by Barbour. 1. Laterite.
Such close correspondence in the stream history of the 2. Blue Clay, OrganicSoil and Silt.
3. Tin-bearingGravel and Sand.
two great rivers is made even more convincing by the
presence of another and higher terrace remnant found ern China. I believe that the stratigraphical relation-
by Barbour on the northwestern rim of the Red Basin.
In one instance, he referred to it as a gravel terrace (300 ships between the Pleistocene of northwestern India
and Burma make it almost certain that the streamnhis-
feet above the stream), and in another as a mesa sur-
face. This highest level may be our first terrace of tory was controlled by the impact of fluctuating Ice
India and Burma, especially since it is associated with Age climates, which conditioned the sedimentation
within this region.
lateritic gravels. Barbour had suggested that the "Red
Earth" formations found associated with the highest
3. COMPARISONS WITH MALAYA AND JAVA
terrace were denudation products from the adjoining
highland of the karst of Szechwan. Little is known of the Pleistocene of the coastal belt
This feature in the Pleistocene geology of the Yangtze of southern Burma, and one has to travel to Malaya in
Valley is of great significance, because it would indicate order to find fossiliferous or implement-bearing forma-
a uniformity of river history over very large areas. I tions of that age. While no attempt will be made to
have pointed out elsewhere (1939-a) that our survey of review the entire mass of scattered information on this
terraces in India and Burma suggests an interpretation subject, it is nevertheless important that a brief account
of the age of these Yangtze terraces different from the be given of our observations at a few isolated localities
one given by Barbour. He argued for a Late Pliocene in the Malayan peninsula.
age of the highest Yangtze level, because it apparently In the region of the tin mines between the coast and
merged with the karst level of Szechwan Province. Kuala Lumpur, in Selangor, the coastal plain is under-
This karst level was held to be of Late Pliocene age on lain by a flat mantle of gravels and red loam. These
the basis of the mammal fauna, described by Matthew deposits form a veneer over a deeply weathered com-
1 For a more detailed discussion of Pleistocene formations in plex of gneiss and other crystalline rocks. Gravel-
China see de Terra, 1941. bearing sand, containing tin minerals and red earth, at-
DE TERRA: THE PLEISTOCENE OF BURMA 333
tain a thickness of 50 to 90 feet. The character of these central Java (Solo Valley), they do not seem to start
deposits resembles that of the Middle Pleistocene of the earlier than the Upper Pleistocene. For this reason
Mogok karst, as illustrated by Fig. 53. A fan forma- alone it is difficult to draw any conclusions as to migra-
tion at the base, overlain by sand and clay with peaty tions of Stone Age peoples between these lands. Also,
layers, form the two outstanding deposits. The laterite in the case of Java, mountain-making has been much
on top of these beds is of recent origin. more intense during the Upper Pleistocene and in sub-
Some three miles north of Singapore, we were shown recent times as compared with Burma. Whether the
a formation of tilted alluvium containing fossil wood terrace system in Java is due to climatic or diastrophic
which reminded us of the Upper Irrawaddian Beds agencies is a problem worth careful and detailed studies.
in Burma. Mr. H. D. Collings, of the Raffles Museum,
Singapore, reported that remains of Elephas nazmadicus E. THE CYCLIC NATURE OF PLEISTOCENE
had been found in Pleistocene gravels of Perak, and
that in this latter neighbourhood a substantial number STRATIGRAPHY
of Early Palaeolithic implements had been extracted I. THE ORIGIN OF RIVER TERRACES
from such deposits. According to Dr. Movius, the im-
plements in question bear a certain resemblance to the One of the greatest problems connected with Pleisto-
Early Anyathian of Burma (see p. 377). In addition, cene geology in Asia is the origin of the terrace forma-
Mr. H. D. Collings reported a number of fissure forma- tions encountered in India, Burma and neighboring
tions from the limestone areas northeast of Singapore. regions. Elsewhere (1939-a), I have pointed out how
This is also suggestive of the presence of those forma- important the method of terrace stratigraphy is for
tions which characterize the upland areas of Burma and prehistoric archaeology in southeastern Asia, especially
South China. Terrace gravels, fossil soils and fissure for the dating of Stone Age cultures. From the fore-
deposits typify the Pleistocene of these lands, and their going descriptions of terraces in Burma, it must be evi-
regional distribution would suggest that their origin is dent that we have to deal with a type of stream history
due to factors exercising regional control over the sedi- analogous to that found in northwestern India and
mentation in these countries. possibly in southern China as well. How are we to
As will be pointed out in the supplementary report explain such a marked correspondenceof alluvial forma-
on my observations in Java (see p. 455), there are some tions, of terraces and soils in regions lying thousands
features in the Pleistocene geology of that island which of miles distant from each other? At first it might well
indicate a certain parallelism in the history of Burma seem that such relationships were merely fortuitous, and
and Java. The related features may be briefly sum- that they had nothing whatever to do with regional
marized as follows: control of stream behavior. However, as I have indi-
cated (1939-a), the correspondence between the various
1) An upward crustal movement causing the folding stream levels is so close as to exclude any but a regional
and warping of inland basins connected with vol- and rather
canic activity. uniformly working agency.
There are two kinds of agencies to which we may at-
2) A terrace sequence along certain streams. tribute the responsibility for this phenomenon of ter-
3) A Lower Pleistocene fauna (Djetis fauna) of race correspondence: climatic and
Villafranchian type, suggestive of faunistic mi- diastrophic. Either
the climate changed and pulsated more or less uniformly
grations between India, Burma and Java, as in these
regions, exercising control over the history of
pointed out by Dr. Colbert. the or else mountain uplifts dictated the cycles
streams,
4) A limestone karst region in southern Java with of
fossiliferous fissure deposits of Middle Pleisto- cutting and filling. We know that uplift occurred in
cene age. almost all mountainous lands adjoining the great alluvial
belt from the Indus eastward to Burma and the Shan
In my opinion these relationships are not sufficient to Highlands. The folding, tilting and faulting of Lower
warrant identical stratigraphy, on the contrary it would and Middle Pleistocene formations prove clearly how
seem as though the bulk of the Pleistocene formation active these upward movements in the foothills were,
in Java cannot be subdivided by pluvial and interpluvial and we even have reason to believe that they continued
stages as has been done in the case of Burma and India. to the close of the Pleistocene. The seismic regions of
The reason for this is that in Java, volcanism has been India and Burma all cluster along the great boundary
active on such a large scale throughout the Pleistocene faults which separate the highlands from the alluvial
that it has obscured those sedimentary records which belt, hence mountain-making is still continuing within
might be compared with the ones formed on the Asiatic these zones. In my memoir on northwestern India (de
mainland. The geological discrepancy between these Terra and Paterson, 1939), I stated that the Pleistocene
regions becomes apparent in the age and nature of the in India was characterized by an interference between
terrace formations. In the case of India and Burma, diastrophic and climatic cycles, but at that time I did
these begin in the late Middle Pleistocene, while in not discuss the manner in which sedimentation was in-
Northern Alps Caucasus I n d i a
Switzerland Pyrenees Carpathians Burma
,Germ,an/ ,S.fr/ance Poland US.S.R. SiwalkI A/slls
ri Irzda aa
I LI ULL
I I I
O*O.O
-0
Lr .0
,T

Erosion Erosion
LU
I II I
1111111
kfo*oO.*H
HT o
00 000ll
0 0
0
0

Erosion Erosion
:o
cU
0 ._ O (3
(U
0 - 0
0000
O -
nho 0
0
0
0
4

osO
o-- 0 o--- 0
0
0
0 O0
U]

Erosion Erosion -J
0 - 0

? ?

FIG. 54. Tentative Correlation of the Pleistocene Sequences in Lowland Regions Adjoining Mountain Ranges in Eurasia
G, M, R, W. Giinz, Mindel, Riss and Wurm Glaciations.
M. Moraines.
HT. High Terrace.
LT. Low Terrace.
G1, G2, G3, G4. Glaciations 1-4.
T1, T2--T. Terraces 1-5.
India: China:
U.S. Upper Siwalik Beds. *S. Sinanthropus.
BF. Boulder Fan. +1. Choukoutienian Culture.
+ Pre-Soan Flakes. +2. Ordos Culture.
+1. Early Soan and Chelleo-Acheul Cultures. +3. Upper Cave of Choukoutien.
+2. Late Soan A Culture. N. Neolithic.
+3. Late Soan B Culture.
Java:
+4, +5. Pindigheb Culture.
N. Neolithic. *P1. Modjokerto Skull.
*P2. Pithecanthropus.
Buirma: *So. Solo Man.
U.I. Upper Irrawaddian Beds. +1. Bone and Stone Industry of the
B.F. Boulder Fan. *W. Wadjak Man.
+ Early Anyathian 1 in situ in the Lateritic Gravel of the Irrawaddy N. Neolithic.
Basin.
+ 1. Early Anyathian 1 in situ in T1.
+2. Early Anyathian 2.
+-3. Early Anyathian 3.
+4. Late Anyathian 1.
+5. Late Anyathian 2.
N. Neolithic.
DE TERRA: THE PLEISTOCENE OF BURMA 335
fluenced by the changes of climate in the non-glaciated we have no such proof in the upper course of the Irra-
tract. With the new information on hand from Burma, waddy. Even if there were marine formations, say
such a relationship between climate and geology would some hundred miles upstream from the mouth of the
seem to warrant a brief discussion, especially as this may river, we could not state with certainty that these were
well be the key for a new approach to Pleistocene due to eustatic changes of sea-level, because of the tec-
stratigraphy. tonic instability of the basin, which at times of rapid
Before entering into a discussion of this problem, it is sinking might have invited the sea to proceed inland.
essential that we answer the question of the origin of Hence it seems to me as though we need to wait for
the Irrawaddy terraces as it was interpreted by Morris further convincing evidence before we can accept the
(1935) and Leicester (1932). Morris claimed that the hypothesis of eustatic control of stream level, as far as
terraces were due to eustatic changes of ocean level (see the Irrawaddy is concerned.
pp. 289, 290, and Fig. 14). This means that the Irra- The origin of these Pleistocene and subrecent ter-
waddy River aggraded its course during times of high races is in my opinion connected with the glacial cycle,
ocean level, while it eroded or degraded its bed in times but in a manner different from that implied by the eu-
of low ocean level. Since the latter state would be static control hypothesis (Fig. 54). On pages 330, 331,
caused by glaciation of the Northern Hemisphere and I have already indicated that the number and composi-
the former by deglaciation, we would have to assume tion of the terraces in Burma correspond to the nature
that each stage of gravel accumulation represents an of the glacial and peri-glacial stream terraces found in
interglacial and each stage of erosion a glacial period. the lowlands adjoining the Kashmir Himalaya. To me
In view of the fact that we have previously assumed this relationship appears to be more reliable, as far as
that a correspondence between glacial or pluvial stages geological correlations are concerned, than the previous
and stages of aggradation exists, such an interpretation view, which attempts to link the stream history with a
of terrace origins as Morris and Leicester have proposed completely unknown process.
would lead us to precisely the reverse picture of terrace
history. Hence the importance which we attach to this II. THE GLACIAL CYCLE IN ASIA AND PLEISTOCENE
speculative approach to solving the origin of the stream STRATIGRAPHY
terraces in Burma.
Let us for a moment assume that the sea-level during It is rather fortunate that our region in Burma falls
Pleistocene times underwent uniform fluctuations. within a piedmont belt adjacent to high mountains,
The maximum rise of level cited by Daly (1934, p. 48) which experienced valley glaciations during the Pleisto-
is about 300 feet, which would mean that the entire cene Period. This makes it possible to relate our geo-
present valley flat between the delta of the Irrawaddy logical data on the Pleistocene in the non-glaciated tracts
and the third defile was submerged under the sea. At of Burma with those from other regions of similar geo-
another phase the ocean level rose to about 200 feet logical structure and history. While in Burma no cor-
above its present level, and in this case the sea would relations between fluvial and glacial gravels are as yet
have invaded the land almost up to Mandalay. It is available, it is important to note that such relationships
of course possible that the basin then was higher, and have been detected in northwestern India. In other
that the sea proceeded only as far as Prome, about two words, what we must try to find out is how the terrace
hundred miles upstream. Even so we cannot imagine formations of Burma are linked with the glacial cycle
that such an event ever occurred, because of the total of India.
lack of Pleistocene marine beds in the Irrawaddy Val- Obviously, an appreciation of the climatic cycle pre-
ley. So far no traces of marine shore-lines have been supposes a conception of temperature fluctuations in the
encountered on the delta flats or farther upstream. Himalayan region. These have been computed, as indi-
While it is probable that the basin floor sank gradually cated in Fig. 54, on the basic assumption of F. Klute
during the Pleistocene and in subrecent time, and that (1928), according to which the present position of the
such marine beds might well have become buried under snow-line 2 corresponds to present climatic conditions,
in the same way as the last glacial snow-line corresponds
younger alluvium, it is nevertheless certain that thus
far we have no reliable geological data to support the to the last glacial climate. The temperature fluctuations
contention for cyclic Pleistocene marine invasions into in the western Himalayas were computed from the
the Irrawaddy Basin. During the Tertiary Period such varying position of the snow-line during the Ice Age,
invasions actually occurred, but here we have indis- as determined by the palaeobotanical, palaeontological
putable records to show in what manner these marine and sedimentary nature of Pleistocene formations. For
invasions took place. In the case of the Pleistocene, we the First Himalayan Glaciation, we calculated a depres-
have only certain deep drill holes and observations on sion of snow-line by 1,600 meters, which corresponds
ground-water lnovements in the deltaic plain near Ran- roughly to a lowering of mean annual temperature of
8? C. (at 33? N. latitude). During the First Inter-
goon. These indicate that the outer fringe of the
growing mouth of the river lay a few miles inland, but 2 The snow-line is the uppermost level at which the snow melts.
336 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

F1G.55. Outline Map of Southern Asia Showing the Main Areas of Investigation.
(Courtesy of the GeographicalReview,published by the American Geogr. Soc. of New York.)

glacial, the climate was warmer in the Kashmir region the relative paucity of grazing animals in the adjoining
than it is today and somewhat more moist, as indicated plains, from which larger ruminants, such as the ele-
by the spread of the pine-oak forest toward the Inner phants and buffalos, had already retreated. In Europe,
Himalaya (de Terra and Paterson, 1939). At that also, the last interglacial climate is considered warmer
time the temperature rose to about 1.5? C. above the than the present, as inferred by. the floras of Pont-ai-
present annual mean. For the Second Glaciation a Mousson (France), Cannstatt (Germany), and Pianico-
depression of snow-line by 1,500 meters has already Sellere (Lake Ivrea, upper Italy). The last glaciation
been calculated by Dainelli (1922), which corresponds in the Himalayas was accompanied by a drop of snow-
to an approximate drop of temperature by 7.5? C., as line of 900-1,000 meters, corresponding to a tempera-
compared with the recent period. During the long Sec- ture drop of 4.5-5? C., and there were three cold sub-
ond Interglacial, the climatic optimum may have been phases, as recorded by terminal moraines. From this
similar to the present, although there were minor fluc- point onward a gradual rise of temperature may be
tuations as recorded by varves in the Upper Karewa assumed, but it must have been interrupted by one or
clays. We know that the beginning of this Interglacial perhaps two brief reverses in post-Glacial times, during
was relatively dry, because dust storms carried great which the snow-line dropped by as much as 600 and 400
amounts of silt from the l)iedmont region into the val- meters respectively, leading to one or two brief ice
leys. The Third Himalayan Glacial indicates a de- advances.
pression of the snow-line by 1,400 to 1,500 meters, or
a temperature drop of 7? C. At that time the advance 1. Tlle Influence of Climiate on Sedimentation
of the valley glaciers was enhanced by a preceding steep-
ening of the valley gradients due to mountain uplift. The first question which arises concerns the nature of
No data are available for the Last Interglacial Period, the sedimentary composition in the plains region, be-
but we may safely assume that it was somewhat drier cause if there is correspondence between glacial and
and warmer than the Second Interglacial, because of non-glacial deposits of the plains, we may expect that
DE TERRA: THE PLEISTOCENE OF BURMA 337
such is practically everywhere the case in lands adjoin- Cold air rested over these highlands, and barometric
ing the Himalayan highlands. From Fig. 55 it seems "highs" must have been more common then. No doubt
obvious that there should be a close correspondence be- the monsoon wind blew against this cold land and cre-
tween glaciation, gravel accumulation and pluvial con- ated cyclonic conditions, which caused a condensation
ditions in the non-glaciated region. of moisture. At such times summer precipitation must
As to the correspondence of glaciation and alluviation, have been greater than nowadays, and, therefore, we
T. T. Paterson and I (1939) have shown that in the may speak of "pluvial periods," as far as these particu-
case of Kashmir, the glacial gravels can be traced for lar piedmont and plains regions are concerned. Slope
long distances into the neighboring lowlands of India. streams must have been choked with debris, because of
Especially the fluvio-glacial gravels of the Second Gla- intensified nivation, frost-action and glacial outwash.
ciation form very conspicuous fans at the outlet of the Frequent and torrential rains carried this excessive
Himalayan tranverse valleys. These fans show them- load into the plains, where it spread out in the shape of
selves to be uniformly dissected, and the valleys are fans which subsequently coalesced, forming wide mar-
filled with a younger gravel, which is capped by loess. ginal gravel belts. Such alluviation was increased by
This deposit can be correlated with the third ice ad- an occasional subsidence of the plains country, since at
vance, and the terrace belonging to this phase was traced places these gravel sheets are several thousand feet thick.
up to the terminal moraines in southern Kashmir. The Another factor that aided in the formation of these
Third Interglacial was a phase of erosion, as was the gravels was the removal and redeposition of great
preceding Second Interglacial, so that when glaciers masses of half-consolidated coarse sediments, such as
later advanced for a fourth time, the melt-waters car- abound in the Siwalik Hills, as well as in the piedmont
ried their load into these younger valley cuts. There is, regions of Central Asia. Obviously, in these regions
therefore, a clear rhythm of deposition to be noted in it should be possible to recognize three to four gravel
this area, which is chiefly a function of climatic changes. zones and the corresponding three erosion periods which
Successive uplifts cannot produce such close corre- mark the interglacial or interpluvial stages. But these
spondence of gravel terraces in the glacial and non- gravels might by chance appear in similar successions in
glacial regions. Furthermore, as mentioned above, the areas lying thousands of miles distant from one an-
same succession of gravels and terraces occurs in other, and still owe their origin by different climatic or
Burma, over 1,500 miles distant from the Kashmir mountain-making conditions. Such uncertainty, how-
region. In the case of the Irrawaddy Valley, I have ever, might be eliminated, if we consider the soils that
shown that this correspondence exists in regions lying go with these gravels.
some 400 miles distant from the glaciated tract. In
such instances it is necessary to ascribe a power of 2. The Climatic Conditioning of Soils During
the Pleistocene
transportation to the streams, descending from the gla-
ciated highlands during the Pleistocene, greater than Vast regions in China, Central Asia and India are
that of the present day. Not only are the gravels covered by Ice Age soils such as loess, loessic siltstones
coarser than any laid down in post-Glacial times, but and red earths, which in most cases are closely associated
they are much thicker and usually connected with soils with Pleistocene gravels. In Kashmir and the Punjab,
suggestive of heavier rainfall. Such a process of allu- as well as in certain regions of Sinkiang (Chinese
viation cannot be explained solely by the melting of Turkestan), there is a close connection between the last
snow or glaciers in the highlands, because at that time glaciation and loess deposition. In India, the bulk of
water was actually still locked up in the form of ice, the Potwar Loess is Third Glacial, and in Burma there
and rainfall may well have been less in the mountains. is a loessic silt-the Pagan Silt-belonging to the fourth
Also, the melting of glaciers is not a catastrophic process terrace, which corresponds probably to the last glacia-
which enhances stream power; it is a very gradual wan- tion. G. B. Barbour has shown that in China loessic
ing which may lead to a local accumulation of debris beds occur in the Sanmenian (Nihowan) Stage (see
(outwash fans), but never to formation of sheets of chart, p. 334). This would clearly argue for an Early
shingle trailing for hundreds of miles away from the Pleistocene Age, a conclusion which is equally justified
ice-bound highlands. The formation of such enormous in the case of Burma and India. Here the Upper Si-
gravel sheets demands specific conditions of climate. walik and Upper Irrawaddian siltstones contain much
Such conditions evidently correspond to glaciations, glacially derived detritus. These earlier loesses are
and they may well be pictured as pluvial periods. In- compact and always in a tilted position, while the
crease of rainfall was a result of the lowering of tem- younger loesses are loose and rarely disturbed by later
perature, because of the refrigerated air masses resting mountain-making movements. In fact, if we take into
over the snow-bound Himalayas and over other ranges account all the evidence on loess and loessic deposits, it
to the east. These uplands comprise some two million would seem as if there were four of these glacial eolian
square miles, of which approximately 60 per cent were soils. In the case of Kashmir the "Upper Karewa
covered by ice and snow during periods of glaciation. Beds," for instance, mark a period of late Second Gla-
338 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
cial and early Second Interglacial dust storms, adding idly, as a result of prevailing aridity. The studies of
another loess to the three kinds already mentioned. Dainelli (1922), however, on the glacial terraces of
A second type of soil which we might use as a climatic the Inner Himalaya, and those made recently by Pater-
indicator is found in a group of tropical and subtropical son and myself (1939) afford proof of the synchronous
soils, such as laterites and red loam. Very often, as in formation of glacial debris with Pleistocene gravels.
Upper Burma and in South China, these are found asso- This holds especially for the aggradational terraces II
ciated with pluvial gravels. They distribute themselves and IV of our system, while terraces I and III are of
in such a fashion as to suggest changing periods of interglacial (degradational) origin.
greater and lesser rainfall. In some cases, e.g. the ter- Under the conception that pluvial stages are concomi-
races of the Irrawaddy Valley, the ground-water later- tant with glacial stages (at least as far as the areas
ites are invariably associated with stages of alluviation, under discussion are concerned), it is not surprising
and since these correspond in type and number to those that the same terraces should be found over regions
found in the glaciated tracts of the Himalayan High- covering thousands of miles in a west-easterly direction.
lands, I suggest that they represent the precipitates of In fact, between the Indus and the Yangtze Rivers this
a pluvial climate. It is to be noted that such fossil relation seems to prevail throughout, though much work
laterites occur down to 20? North latitude in regions remains to be done in intervening areas. The terrace
which are known as "Dry Belts," where the precipita- sequence found here is characterized by five terraces,
tion is less than thirty inches a year. of which four are Pleistocene and one postglacial in
There are other soils in these piedmont lands which age-the oldest and highest being superimposed on
are called "kankar" or limy concretionary soils. In gravel fans which are younger than the tilted beds con-
Burma these soils are conspicuous on surfaces which taining a Villafranchian fauna. In the glaciated re-
have undergone prolonged weathering under drier cli- gions, T1 is of Second Interglacial Age; terraces two
matic conditions. As I have shown, very often they and four mark stages of heavy alluviation corresponding
are deeply buried under younger alluvial and eolian to the third and fourth glacial (or pluvial) stages.
formations. It is possible that a similar system of ancient stream
A special type of soil is found in the more humid levels exists on the Middle Yangtze, as G. B. Bar-
highlands of the Shan States and Yunnan. These are bour's studies suggest (see p. 332). Huntington (1919)
the lateritic fans and red loams that appear associated has described five terraces from the piedmont regions
with the karst relief. These soils, especially the boulder and basins of eastern Persia, which may well corre-
fans, are often connected with cave and fissure deposits, spond to our sequence. Confirmation of this would
containing a Middle Pleistocene type of fauna (Ste- greatly enhance our chances of viewing the Pleistocene
godon orientalis, Elephas namadicus, orang, porcupine, in these areas under the conception presented above.
deer, etc.). The famous gem-bearing gravels and sands But the Pleistocene period reveals other characteristics
of the Ruby Mines District in Upper Burma belong to which are almost equally common to all the regions
this category. Similar soils from the Kwangsi and mentioned; the erosional and structural breaks of the
Szechwan Provinces of China have been described by
sequences. The greatest of these occurs between the
J. Thorp (1937), and quite possibly their extension is beds with a Villafranchian fauna and the Later Pleisto-
greater than we know now. In Malaya, the tin-bearing cene. This angular unconformity is found almost uni-
alluvium is associated with such fossil soils, indicating
versally from the Caucasus to Central and Eastern Asia.
heavy erosion under pluvial conditions. Other breaks are found between the gravel fans and
The correlation of these fossil soils with glacial de-
the loess, and in northwestern India even the loess was
posits makes an interesting and important problem for
Pleistocene geology. At the moment we can say that slightly effected by subrecent mountain-making. Here.
in the case of northern India the glacial gravels merge uplift of the Himalayan front ranges during the Pleisto-
cene amounted to 6,000 feet or more, and the total am-
into pluvial (or fluvial) formations in the piedmont and
plitude of crustal deformation in the sub-Himalayan
plains regions, and that in the case of Burma the glacial
Ranges exceeded 12,000 feet. It is this factor which
cycle is clearly documented in areas lying far beyond
the glaciated tracts. This correlation becomes even may account for the local thickness of the Pleistocene
in the piedmont region.
more significant, if we view the terraces of Central and
From this discussion it is evident that the new ap-
Southern Asia as a whole.
proach to Pleistocene stratigraphy by means of terrace
3. The Influence of Climate on Terrace Formation geology allows for seven subdivisions, as compared to
three which the palaeontological method has suggested.
Previously I have shown that the river terraces of This does not mean that these seven subdivisions could
Southern and Central Asia show a surprisingly uniform be recognized everywhere in Asia. On the contrary.
pattern. As for Central Asia, E. Huntington (1919) the system of the climatic control of stream history, out-
has ascribed the gravel formations to the erosive effects lined above, is applicable only to lowlands adjoining the
of dry periods during which soil-wash proceeded rap- high massives which experienced several glaciations
DE TERRA: THE PLEISTOCENE OF BURMA 339
during the Ice Age. For this reason, it is as yet not outlined in the accompanying chart (Fig. 54). It must
possible to correlate any of our so called "pluvials" with be admitted that such an attempt needs to be substanti-
similar stages such as have been computed from various ated by additional studies, but to me it would seem to
geological records in Africa and North America. How- bear great promise as far as the solution of a great many
ever, I believe it is possible that our stratigraphic ap- geological and archaeological problems is concerned.
proach may improve the status of Pleistocene stratig- I believe that the problem of human evolution cannot
raphy and archaeological chronology within the geo- be viewed in true perspective unless such a multistrati-
graphical realm of the Mediterranean mountain belt, be- graphic method of Pleistocene geology is adopted.
cause here the glacial cycle seems to have been recorded Only in this manner will it be possible to correlate
in a very uniform manner, at least as far as the Eurasi- culture-bearing formations with each other, and to es-
atic mainland is concerned. tablish thereby a firm basis for the chronology of human'
A tentative correlation of glacial formations has been evolution.
PLATE I

FIG. 1. View eastward from near Kyaukpadaung, Upper Burma, across a level surface composed of Irra-
waddian rocks with Mt. Popa volcano (4981 feet) in the background. Note: The sparse xerophytic vegetation
of the "Dry Belt." (Compare with Fig. 2 below.)

FIG. 2. Escarpment and high level terraces (Ti_3) east of Hsenwi, Namtu Valley, Northern
Shan States.
PLATE II

FIG. 1. Right bank of the Irrawaddy River opposite Mandalay with the Sagaing hills in the background.
Right, Low-water sandbanks with temporary huts and rice fields. Left, High-water bench and the third terrace
(T3). M, Mingun Pagoda.

FIG. 2. View from the third terrace near Mingun across the Irrawaddy Valley towards Mandalay (Ma.)
and the Shan Highlands. The fault escarpment separates the Shan block, composed of crystalline rocks, from
the Irrawaddy geosyncline.
PLATE III

FIG. 2. Cross-bedding in Upper Irrawaddian rocks ex


Chaung, between Magwe and Yenangyaung. Note: Indura
and fossil tree log in the foreground.

FIG. 1. Cliff section through tilted Upper Irrawaddian


Sandstone (I) and T3 gravel, south of Yenangyaung on the
left bank of the Irrawaddy River.

FIG. 3. Cliff section through the edge of T3 near Nyaun


waddian conglomerate (I) and the Lateritic crust overlain b
waddy River is on the right.
PLATE IV

FIG. 1. Tilted Upper Irrawaddian rocks (I) exposed on the eastern flank of the Sagaing Hills, west of
Mingun.

FIG. 2. Remnants of the third terrace (T3) near Sadaing, south of Yenangyaung.
PLATE V

FIG. 1. Panorama view of the landscape south of Chauk showing the terrace sequence (T1-T4). The Irrawaddy River is on the extreme left;

FIG. 2. Terrace 1 remnant (?) represented by Hill P. 621, east of Sale, with T2 in FIG. 3. The dissected relief of the Irrawaddy anticli
the foreground. of Chinaungma (T1) at the left and the Chauk oil
PLATE VI

FIG. 1. Terrace 5 in a tributary valley, the Pyinma Chaung, near Singu. Note: FIG. 2. Terraces 4 and 5 in Pyinma Chau
Dissected upland composed of Irrawaddian rocks in the background.

FIG. 3. Terrace 4 on the left bank of the Irrawaddy south of Zigyobin, between FIG. 4. The edge of Terrace 4 on the right bank o
Chauk and Sale. showing red gravel (RG) and Pagan silt (PS) overlying
in the foreground.
PLATE VII

FIG. 1. Dissected surface of the Irrawaddian formation south of Chauk. FIG. 2. View southwestward towards Sale showing
3. Terrace 4 and the Irrawaddy River are i

FIG. 3. Close-up view of Laterite overlain by red gravel on Terrace 2 at the base of FIG. 4. Cliffs of Upper Irrawaddian sands
the hill of Chinaungma, near Chauk.
PLATE VIII

FIG. 1. Terraces 4 and 5 at Nyaungu. FIG. 2. Dissected lateritic crust (LC) on T

FIG. 3. Lateritic crust (LC) overlain by Pagan silt (PS) on T. near Nyaungu. FIG. 4. Red eolian sand (Magwe sand), showing a neo
overlying the Nyaungu red earth, northeast of Ma
PLATE IX

FIG. 1. The T3 gravel and red sand on the road from Yenangyaung to Thittabwe.

FIG. 2. Gravel on Hill P. 621, east of Sale; T3 in the background.


PLATE X

FIG. 1. Nyaungu red earth resembling miniature badlands, east of Nyaungu. FIG. 2. Concretionary soil on the surface of Pa

FIG. 3. Lateritic crusts (LC)-altogether 3-exposed on the edge of T4 near Singu.


PLATE XI

FIG. 1. The fault escarpment between plateau limestone and red beds east of FIG. 2. Terraces 3 and 4 of the Salween Rive
Hsenwi, Namtu Valley, Northern Shan States.

FIG. 3. Fanglomerate superimposed on a tilted boulder fan deposit at milestone 16 on FIG. 4. Sink-hole topography on the basin filling east
the road from Hsenwi to Kontap and Holi. States.
PLATE XII

FIG. 1. Sink-hole topography showing the vegetation cluster in a sink-hole, north- FIG. 2. Karst formation in the crystalline limesto
east of Mogok.

FIG. 3. Lateritized fanglomerate overlain by red fan debris near Mogok. FIG. 4. Terrace 4 gravel and red loam near Hsipa
PART II
THE STONE AGE OF BURMA
BY HALLAM L. MOVIUS, JR.

Plates XIII-XVIII

INTRODUCTION yaung, Mr. Morris (1932; 1936) has reported artifacts


apparently associated with the Pleistocene gravels of
The first stone implements to be recorded from the Irrawaddy. Farther south in the valley, near Tha-
Burma were discovered near Yenangyaung during the yetmyo (Morris, 1936-a), several
alleged human tools
last century by Dr. F. Noetling of the Geological Sur- were recovered from commercial excavations in what
vey of India. It was claimed by Noetling (1894; is stated to be the same horizon. These implements,
1897) that these were found in situ in an Upper Mio- together with a larger series from several new sites at
cene or Early Pliocene deposit, associated with Hippo-
Yenangyaung and from the Paunglaung Valley, near
theriuml antelopinumi and Acerotherium perimense, Pyinmana
(Morris, 1937), have been presented by Mr.
hence the discovery aroused considerable interest (com- Morris to the British Museum.
Through the courtesy
pare Jones, 1894; Blanford, 1895; Cole, 1895). of Mr. C. F. C. Hawkes, F.S.A., Assistant Keeper of
Whereas none of the subsequent investigators of the the Department of British and Mediaeval
Antiquities,
site challenged the human origin of the specimens, it the writer had the
opportunity of examining the material
was felt that Dr. Noetling had been somewhat misled in 1937 while en route for
Rangoon. Whereas the
regarding the stratigraphic horizon of the finds. In collection includes several quite definite man-made, heav-
fact, as early as 1895 Mr. R. D. Oldham pointed out ily rolled choppers and chopping-tools fashioned on large
that identical artifacts occurred in abundanceon the sur- pebbles of silicified tuff, the bulk of the
specimens, if
face of the plateau, ? 100 feet above Noetling's alleged human, have been so
extensively worn by river action
implementiferous stratum. This observation was con- that it is difficult to recognize them as implements. Ac-
firmed by Swinhoe (1902; 1903) and Pascoe (1912), cording to Mr. Morris the specimens may be assigned
but Das Gupta (1923) and Mitra (1927, p. 124-127) to the Chelleo-Acheulean culture, but the writer failed
accepted Noetling's statements and considered that the to recognize a single hand-axe in the entire collection.
evidence demonstrated the presence of Late Tertiary This observation was confirmed later in the field (com-
man in Burma. Mr. J. C. Brown (1931), after review- pare: de Terra, Teilhard, and Movius, 1938; de Terra,
ing the controversy, concluded that in his opinion the 1938; Teilhard, 1939, p. 252), and on the basis of our
matter suffered from too much writing and too little present
knowledge it may be stated with certainty that
field experience. Recently, on the basis of fresh data, hand-axes are absent in the Burmese Palaeolithic.
Mr. T. O. Morris (1935, pp. 2-3) has suggested that Indeed the implements collected during the 1937-1938
the implements were derived from the surface and re- season differ in several fundamental
respects from those
deposited in association with Pliocene beds, a statement of Western Europe, and for this reason the culture has
with which we are in full agreement. However, Mr. been
given a new name. After the colloquial Burmese
Morris' attribution of the material to the Upper Palaeo- for an Upper Burman (an-ya-tha), it is now known as
lithic Period is not supported by the evidence from the the ANYATHIAN. The writer is much indebted to Mr.
localities investigated by the American Southeast Asiatic E. J. Bradshaw, Government Geologist at Yenang-
Expedition during the 1937-38 season. As will be yaung, for suggesting this title. In a chronological
shown in a later section (see p. 380), the material is not sense the Early Anyathian roughly covers the time-span
only associated with the post-Pleistocene Magwe Sand, of the Lower and Middle Palaeolithic Periods of the
but also with pottery and polished stone. It is there- Old World, while the Late Anyathian may be considered
fore Neolithic in age, but whether or not it represents the equivalent of the Upper Palaeolithic Period. In the
a true food-producing culture could not be definitely Irrawaddy Valley there is no stage of development cor-
established. responding to the Middle Palaeolithic of other regions;
Owing solely to the interest taken in the Pleistocene in fact, as will be presently demonstrated, the typology
of Burma by Mr. T. O. Morris, much fresh light has of the Anyathian is remarkably uniform throughout the
been thrown on the question of the antiquity of man in entire Old Stone Age sequence of Burma.
this section of Asia during the last ten years. From the The principal Anyathian sites investigated by the ex-
vicinity of Themathauk Chaung, a small stream which pedition are situated on the left bank of the Irrawaddy
runs through the Singu Oil-Field, and near Yenang- between Magwe on the south and Nyaungu, a small vil-
341
342 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
lage near Pagan, on the north (Fig. 20, p. 295). A few distinguished men in this research. As leader of the
implements were also found at Minbu, opposite Magwe, expedition, Dr. de Terra gave us the benefit of his
and at Pakokku, 12 miles upstream from Nyaungu, on great knowledge of the field of Pleistocene geology, a
the right bank of the river. In addition the Pleistocene prerequisite to the establishment of the archaeological
gravels at Pauk, in the Yaw Valley, yielded a small sequence.
Late Anyathian series. Although the vicinity of Min- Prior to our departure for Burma, information re-
gun in the Mandalay region (Fig. 7, p. 282) was care- ceived in Cambridge from Dr. Gordon T. Bowles of the
fully searched, no very convincing material was found. University of Hawaii, and in London from Mr. T. O.
Our time in this latter area was spent in fossil collecting.
Morris, Geologist of Steele Bros., Ltd., proved ex-
Neolithic sites were found everywhere; the most im- tremely useful. Mr. Morris went over in detail with
portant ones being at Magwe, Minbu, Yenangyaung, us his observations on the Pleistocene of the Irrawaddy
Singu, Nyaungu and Kyaukpadaung in the Mt. Popa Valley, as well as his collections of Stone Age material
region. A short reconnaissance was made to investi- in the British Museum. In Burma we were grateful
gate the possibilities of cave excavation near Taunggyi for the help and co-operation of Mr. E. L. C. Clegg,
in the Southern Shan States (Fig. 1, p. 272), and al- Superintendent, and Mr. E. J. Bradshaw, Resident
though several very fine caves were visited, test pits Geologist, of the Burma Division of the Geological
failed to reveal evidence of occupation during Pleisto- Survey of India; Professor G. H. Luce and Professor
cene times. Early in January 1938 a trip was made to B. R. Pearn of University College, Rangoon; Mr.
Kengtung in the extreme east of Burma. Here a few Austin C. Brady, American Consul, Rangoon; Dr.
Pleistocene fossils were collected, and a small series ofJames H. Telford, Director of the American Baptist
polished stone and bronze implements belonging to Dr. Mission, Pangwe, Kengtung State; Mr. James Ewant,
J. H. Telford, Director of the American Baptist Mis- Headmaster of the Shan Chief's School, Taunggyi; in
sion, Pangwe (near Loimwe), was studied. Lack of addition to various officials of the Burmah Oil Co., at
time prevented us from visiting Dr. Telford's sites, Yenangyaung and Chauk. In Singapore Mr. M. W. F.
since this would have required organizing a pack trip Tweedie and Mr. H. D. Collings of the Raffles Museum
and several days' travel. But the archaeology of this did everything possible to make the important Stone
vast state, which borders on Yunnan, Indo-China and Age collections from Malaya available to us. We wish
Siam, would well repay investigation. to take this opportunity of thanking the Raffles Museum
for contributing to our expenses in Singapore while
ACKNOWLEDGMENTS examining Mr. Collings' new Palaeolithic material from
It is my very pleasant privilege to add my thanks to Perak.
those friends and colleagues mentioned by Dr. de Terra The success of our trip to Java is due solely to the
in the introduction, who contributed in so many vital tireless efforts made by Dr. G. H. R. von Koenigswald
on our behalf. Not only did he prove an expert guide
ways to the success of the American Southeast Asiatic
to the important Pleistocene localities on that island, but
Ekpedition. As previously explained, it was the Pea-
he also placed at our disposal for study his large collec-
body Museum of Harvard University which made
possible the archaeological investigation undertaken by tion of Palaeolithic implements from the vicinity of Pat-
my wife and myself in Burma, under Dr. de Terra's jitan. The results of this study will be reported on else-
direction. Thus at the outset, we wish to acknowledge where. Other scientists in Java whom we wish to thank
the generous financial and moral support received from for their very kind assistance include: Dr. R. W. van
Mr. Donald Scott, Director of the Peabody Museum, Bemmelen and Dr. L. J. C. van Es, of the Geological
not only in the field, but also throughout the prepara- Survey of the Netherlands East Indies; Dr. John van
tion of this archaeological report. From the beginning der Hoop, Director of Archaeology at the Royal Ba-
the project was heartily endorsed by Professor E. A. tavian Society's Museum; Dr. W. F. Stiitterheim, Di-
Hooton, Chairman of the Division of Anthropology at rector of the Archaeological Survey of the Netherlands
Harvard University, Dr. Thomas Barbour, Director of East Indies; Professor A. W. Mijsberg, Department of
the University Museum, and Mr. Lauriston Ward, also Anatomy, Batavia University; and Mr. H. R. van
of the Division of Anthropology, and we are grateful Heekeren of Djember, in eastern Java.
for their advice and wholehearted assistance on many On our return from the Far East, we discussed the
matters connected with the Expedition. We should significance of the Palaeolithic cultures of Burma and
also like to express our sincere thanks to Dr. de Terra Java with various authorities. Among those who saw
and Dr. Teilhard de Chardin. Their help and co- the material and contributed many valuable suggestions,
operation in the field were a constant source of in- we wish to mention the following: Professor H. Breuil,
spiration to both of us; furthermore their understanding Professor M. R. Vaufrey and Mr. Harper Kelley of
of the Cenozoic sequences in northwestern India and Paris; Mr. Miles C. Burkitt, Dr. T. T. Paterson and
China has provided a broader basis for the work in Professor D. A. E. Garrod of Cambridge University;
Burma. It was a pleasure to be affiliated with two such Mr. C. F. C. Hawkes and Mr. E. M. M. Alexander of
MOVIUS: THE STONE AGE OF BURMA 343
the British Museum. In the actual preparation of this the form of remnants connected with T, and the higher
report, we gratefully acknowledge the help of Professor slopes of the Pegu Yoma. According to de Terra, the
Glover M. Allen, Curator of Mammals, and Dr. J. Lateritic Gravel of the Irrawaddy Basin is the counter-
Bequaert, Associate Curator, both of the Museum of part of the post-Upper Irrawaddian Uru Boulder Con-
Comparative Zoology at Harvard University, for iden- glomerate of northern Burma, which in turn may be
tifying the animal bones and shells from sites in the correlated with the Late Upper Siwalik Boulder Con-
Southern Shan States, as well as that of Mr. William glomerate of northwestern India-Early Middle Pleis-
C. Darragh of the Laboratory of Palaeobotany, Harvard tocene (2nd Himalayan Glaciation).
Botanical Museum, and Mrs. M. E. Goodman of the (b) Following this period of alluviation a cycle of
Department of Anthropology, Radcliffe College, for erosion began during a long dry interval. This repre-
notes and tests on the materials used in the manufacture sents a major interpluvial period, which witnessed ex-
of the stone tools from Burma. The excellent drawings tensive degradation and valley cutting by the river-a
of the implements, in addition to several of the maps process which may be considered as a three-fold
and sections, were done by Mr. Elmer Rising of the sequence:
Peabody Museum, Harvard University. (i) The removal of a considerable portion of the
Lateritic Gravel from the peneplain, since only rem-
A. THE PALAEOLITHIC PERIOD nants of it remain at present. In a few instances these
remnants appear as isolated flat hills, capped by very
I. STRATIGRAPHY
coarse red boulder gravels at heights varying from 280
In order to provide a basis for understanding the to 450 feet above river level. These formations have
archaeological sequence in Upper Burma, it seems ad- been identified as representing the highest terrace-T,
visable at the outset to present a brief summary of the --in the Irrawaddy sequence (P1. V, Figs. 2 and 3).
terrace stratigraphy established by Drs. de Terra and Thus the gravels found on T, seem to be residual gravels
Teilhard in the Irrawaddy Valley. The following and apparently date from the previous aggradational
summary is based on Dr. de Terra's report in which a stage when the river flowed at least 300 feet above its
full description of each section investigated by the present level.
American Southeast Asiatic Expedition will be found. (ii) Continued erosion led to the cutting of a wide
In Fig. 56 an attempt has been made to show the various flat valley, which was later partly filled with the de-
levels present at the archaeological sites as well as their posits of To and T3.
associated deposits. (iii) Lastly an extensive and firmly cemented fer-
ruginous crust or ironstone hardpan was formed on the
1. Siummnnaryof the Terrace Sequence ancient valley floor 90 to 110 feet above the level of
the present stream. This lateritic crust is now exposed
There are five terraces (P1. V, Fig. 1) which repre- under the T. deposits at several localities (P1. VIII,
sent six stages in the history of the river, since between Figs. 2 and 3). It was formed during a prolonged dry
the deposition of T1 and T2 a prolonged period of interval when the river was much reduced in size, as
erosion occurred. These may be considered in (lescend- attested by the typical desert varnish displayed by much
ing order, beginning with the highest and oldest terrace, of the rock debris of which the crust is formed (see p.
as follows: 296 of Dr. de Terra's report).
(a) Terrace 1- (c) In the following pluvial period, during which the
Erosion Period between T1 and T,- rainfall in the Dry Belt of Upper Burma was probably 2
(b)
Terrace 2- or 3 times greater than at present, To was formed. This
(c)
Terrace 3: an Erosion Period is a depositional terrace, and it is preserved in the form
(d)
(e) Terrace 4- of wide and rather isolated benches 180 to 250 feet
above the river. The T2 sediments consist of a coarse,
(f) Terrace 5-
reddish, irregularly-bedded gravel, laid down by the
(a) At the close of the Lower Pleistocene the Upper river and containing boulders up to 1 foot in diameter
Irrawaddian Beds (P1. III, Fig. 1; P1. VII, Fig. 4), (P1. VII, Fig. 3). This is overlain by cross-bedded,
which are of Villafranchian age (see p. 427 of Dr. Col- red or pink, fluvial sand. Above occurs the Nyaungu
bert's report), were tilted as a result of diastrophic Red Earth-a deep, reddish-purple soil which attains
movements connected with mountain building, and there a depth of over 50 feet in some places (P1. X, Fig. 1).
followed a period of peneplanation forming the tre- The Nyaungu Red Earth is a lateritic soil which was
mendously extensive piedmont plain of Upper Burma formed on the sides of the valley and which moved into
(P1. VII, Fig. 1). On this plain thick gravel deposits its present position by a process of slope-wash during a
were laid down during a major pluvial period. These period of torrential rainfall, according to de Terra (see
represent an ancient basin fill, called Lateritic Gravel p. 308). The nature of this deposit demonstrates that
by de Terra (see p. 312), and "Plateau Red Earth" by a very humid tropical climate prevailed in Upper Burma
Pascoe (1912). At present the deposit exists only in at the beginning of Upper Pleistocene times.
344 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

(d) Terrace 3, the widest and most conspicuous of a -


all the levels in the Irrawaddy Valley, is a degradational
terrace formed during a period of prolonged and wide- /
spread erosion (PI. II, Fig. 1; P1. IV, Fig. 2; P1. VII,
Fig. 2). T3 occurs at a height of 90 to 110 feet above ., __
the river. Its relation to T, of great importance from e z-
< c
an archaeological point of view, is clearly shown by r
Section 4 at Nyaungu, described by de Terra (see p. 297- /
and Fig. 22). At the base is a layer 1 to 2 feet thick< . z/
<
representing an ancient ironstone hardpan or ferru- z z >
<z <
< < Co
ginous crust which is overlain by alluviuml. The latter
consists of coarse fluvial gravel 8 to 12 feet thick, with -r h o/ o
several feet of cross-bedded sand above (P1. IX, Fig. 1;< ' 0 0 ~~
Pl. XIV, Fig. 1). On the surface is the same struc- \ v
tureless, yellow-gray to buff-colored, loessic silt-the M / n
Pagan Silt of de Terra (see p. 297)-which is found on < /<
T4 (PI. III, Fig. 3; P1. VII, Fig. 3). The age of the- - --- /
silt is therefore subsequent to the aggradational stage ^ *

during which T4 was formed. According to de Terra, /o


the Pagan Silt is very similar to the Potwar Loessic Silt, _j
of Northwestern India: both deposits are typical of the ?o
.
semi-arid sections of the monsoon-swept regions south
of the Himalaya massif. The Pagan Silt found on
T3 in Upper Burma was derived from a lower valley / c
floor which later became T4. It is an eolian silt laid /
down by the monsoon rains, a complex process fully __ / O
described by de Terra in his Indian memoir (with C CD /
Paterson, 1939, pp. 274-276). : x
(e) During the last pluvial period, which occurred at /
the end of the Upper Pleistocene, the T4 deposits were / '
accumulated in the Irrawaddy Valley (P1. VI, Fig. 3). U
This terrace is found at a fairly uniform height of 60 -g j
to 70 feet above the level of the river, and it marks a ~L <L
stage of aggradation. At the base of the T, deposits F
there are 6 to 8 feet of medium-sized red gravel and /
red sand of fluvial origin. On this rests the same fine -
/
silt of eolian origin-Pagan Silt-that is found on T3/ ' (
(P1. VI, Fig. 4; P1. VIII, Fig. 1). According to de /
v
Terra, the latter belongs to a late stage of pluviation
when the transporting power of the streams had dimin- /
ished and a silting up of the valleys was taking place. F p , o
=o
(f) The lowest terrace-T,-is composed of cross- z L

bedded gravelly sand and pink silt: called Singu Silt /) -O o -


>
.(PI. VI, Figs. 1 and 2). It occurs at a height of + 40 >m =" -
feet above stream level and represents a post-Pleistocene / - D C
stage of aggradation by the river. In composition it is< / / l C. o
very like the recent sediments being laid down by the / ,., 0

river. <J - u-'- c


L
roUJ ? m

o- \ U3J lI UJ 2-
The only difficulty in understanding the above se- L - j z u
quence arises from the fact that none of the deposits O O 0o .
exposed on T3 are actually of T. age. As previously ' -
mentioned, T3 is an erosional or degradational terrace, / i
and accordingly its deposits vary greatly in thickness. / a.
On it the following sequence is exposed (see Fig. 56): , z /
a.
WM'-- 0
(i) A basal cemented ferruginous crust or ironstone >e
i',-,\
n rAl-
iictl upctii.Ic .
MOVIUS: THE STONE AGE OF BURMA 345

(ii) Coarse, reddish, irregularly-bedded gravel overlain of T2 and the basal gravel and sand of T2 were laid
by red or pink fluvial sand. down, followed upon a prolonged dry interval when an
(iii) The surface of T3 is locally covered by Pagan Silt. extensive ferruginous crust was formed, now exposed
These deposits belong to three distinct periods, since either on T3 or at the base of the T3 gravel. The third
(i) the basal ferruginous crust was formed at the end terrace itself was cut during the succeeding interpluvial
of the long dry interpluvial period which witnessed the period, and the Pagan Silt belongs to the end of a later
cutting of T_, (ii) the gravels and sands are referable pluvial phase when the river was flowing in a bed now
to the following pluvial stage when the T, deposits-in- represented by T4. As both deposits (i) and (ii) in
cluding the Nyaungu Red Earth-were accumulated, the above sequence contain archaeological material of
and (iii) the Pagan Silt dates from the time of the for- .Lower Palaeolithic (Early Anyathian) affinities, the
mation of T4 during the last pluvial period, which marks geological age of the various T3 horizons is of great
the close of the Pleistocene in Burma. In other words importance. The sequence is shown on the accompany-
the moist period, during which the Nyaungu Red Earth ing chart (Table 2).

BLE 2
CHART SHOWING THE MIDDLE AND LATE PLEISTOCENE STRATIGRAPHYOF THE IRRAWADDY VALLEY, UPPER BURMA. [Based on
the sequence given by Dr. de Terra (pp. 312-313) and his chart (Table 1, p. 331), with the ARCHAEOLOGICAL SUCCESSION ADDED]. Dur-
ing the Pluvial Periods humid tropical conditions pertained; the Interpluvial Periods were characterized by a climate similar to or pos-
sibly drier than the present.

PERIOD GEOLOGICAL FORMATIONS CLIMATE ARCHAEOLOGY

Accumulation of Singu Silt ...... T INTERPLUVIAL


Post- NEOLITHIC
Pleistocene Eolian Deposits ......Magwe Sand Present conditions
established

Deposition of Red Gravel and Sand......T4


PLUVIAL LATE ANYATHIAN 2
Pagan Silt on T3 and T4

Cutting of the Main Terrace ...... T3


INTERPLUVIAL LATE ANYATHIAN 1
Prolonged and widespread Erosion

PLUVIAL
0 Deposition of Basal Red Gravel and EARLY ANYATHIAN
F(- overlying Nyaungu Red Earth...... T2 Rainfall 2-3 times that of
present in Dry Belt
::L

Erosion of T, and the formation of a cemented ferrugi-


nous crust or ironstone hardpan on a former valley INTERPLUVIAL
floor now exposed under T3 at several localities EARLY ANYATHIAN 2
Long dry interval
Erosion of Lateritic Gravel except
for isolated remnants ...... T

Deposition of Lateritic Gravel...... connected with T1


and the higher slopes of the Pegu Yoma
PLUVIAL EARLY ANYATHIAN 1
(This is the equivalent of the Uru Boulder
Conglomerate of Northern Burma)

UNCONFORMITY-
346 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
2. The Archaeological Succession p. 372). Furthermore in this deposit degree of rolling
is an extremely dangerous criterion of age, as the
Our investigations have demonstrated that there are
gravels in question were subjected to considerable re-
three phases belonging to the Early Anyathian whereas
two exist in the Late Anyathian. Each of these phases working during the succeeding period when T3 was cut.
is based on the stratigraphicalsuccession outlined above. (d) Late Anyathian 1.-In spite of the difficulties
However, the only marked typological difference is be- arising with regard to the Early Anyathian implements
from the T. gravels as discussed above, there exists a
tween the Early and the Late Anyathian, since the
.small series from Magwe, Yenangyaung and Chauk
various phases of which each is composed display a sur-
which is only very slightly rolled and which differs in
prising degree of uniformity. The archaeological suc- certain typological respects from the Early Anyathian.
cession may be outlined as follows; the sites are all
shown on the map, Fig. 20 (p. 295). Accordingly, this has been assigned to Late Anyathian
1, as it seems to have more in common with the later
(a) Early Anyathian 1.-This is always very heavily
rolled and is found in situ in the Lateritic Gravel at development than with the earlier one. It appears likely
that the Late Anyathian 1 dates from the erosion period
Magwe: Loc. 3, as well as in the T, remnants of the when T. was cut, exposing the basal gravel deposits laid
same deposit at Chauk: Hill of Chinaungma, and east
down during the preceding interval.
of Sale: Hill P. 621. Only a total of 23 implements
was collected at these sites, but they are very significant, (e) Late Anyathian 2.-This represents the final de-
since they are the oldest human artifacts from the Irra- velopment of the Burmese Palaeolithic and it occurs
slightly rolled in the gravels of T4. Late Anyathian 2
waddy Valley. Thus the Early Anyathian 1 dates from
the pluvial period at the beginning of the Middle Pleis- implements were collected in this horizon one-half mile
east of Magwe Pagoda, at Pauk in the Yaw Valley, and
tocene in Burma, which witnessed the deposition of an
at several localities along the road between Singu and
extensive valley fill now preserved as remnants of La-
teritic Gravel on both T, and the higher slopes of the Pagan. On typological grounds a number of heavily
weathered but unrolled implements found on the surface
Pegu Yoma. of T3 at the Early Anyathian sites probably belong to
(b) Early Anyathian 2.-Nearly 100 implements
this stage. It is impossible to be dogmatic on this point,
found either in or associated with the basal ferrugi-
however, since implements indistinguishable from those
nous crust (ironstone hardpan) exposed under T3 at of the Anyathian appear as well in the Neolithic of Up-
Nyaungu represent the second phase of the Early per Burma. It seems likely that the Late Anyathian 2
Anyathian. Many of these display a typical desert is associated with the Pagan Silt, but this was not defi-
varnish indicating actual occupation contemporary with
the formation of the crust, which occurred during a nitely established. Two artifacts were collected near
outcrops of this deposit, both of which are heavily en-
long interpluvial period in late Middle Pleistocene crusted with silty concretions, but nothing demonstrably
times. Others are rolled and may either be the same human was found in situ. A few extremely heavily
age as the gravels, which were later cemented to form rolled Early Anyathian implements were recovered from
the crust, or derived from older deposits. Although the T4 gravels, which had obviously been derived from
this horizon was observed at several localities in Upper the T3 level.
Burma, the outcrops in the vicinity of Nyaungu were
the only ones that proved to be implementiferous.
In general, Palaeolithic tools at the sites investigated
(c) Early Anyathian 3.-The main development of
the Early Anyathian-Early Anyathian 3-is demon- by the American Southeast Asiatic Expedition are com-
strated by a large series of approximately 400 imple- paratively rare; the total collection made during almost
ments from the gravels which are of T2 age and which four months of intensive work comprises only slightly
are at present exposed on T. at Magwe, Minbu, Yenang- over 650 Anyathian implements. This is doubtless due
in some measure to the enormous extent of the terraces,
yaung, Chauk, Pagan and Pakokku. Of these the lo-
calities in the vicinity of Yenangyaung and Chauk are since T3 is up to one mile and a quarter wide at some
the most important. All the Early Anyathian 3 imple- localities. The result is that the implementiferous hori-
ments are very heavily rolled and are probably contem- zons are very spread out and relatively thin (P1. IX,
porary with the deposition of the gravels-early Upper Fig. 1; P1. XIV, Fig. 1); no thick Pleistocene gravel
Pleistocene. But some of them which display an ex- exposures were found. Now the Irrawaddy reaches a
treme degree of wear may possibly be Early Anyathian width of 3 to 5 miles at present when swollen by the
2 specimens derived from the underlying crust. Typo- floods of the wet season, and it is not improbable that
logically this point is impossible to establish, since, al- during the pluvial periods of the Pleistocene it attained
though definite stratigraphy exists within the T, de- a width greatly in excess of this figure. Even during
posits, there is no discernible difference between the interpluvial periods the seasonal melting of ice on the
implements from the older Early Anyathian 2 phase Tibetan Plateau would insure an abundant water supply
and those from the Early Anyathian 3 horizon. This which would amplify the fairly constant and powerful
question is more fully discussed in a later section (see erosive action of this huge stream. Thus the geological
MOVIUS: THE STONE AGE OF BURMA 347

horizons and the contained archaeological material, de- 2. Early Anyathian 2


rived from contemporary occupation sites in the valley,
have been spread out over an extensive area. Further- (a) Nyalingui (Section 4, Fig. 22, p. 297). The Tn
level-90 to 110 feet above the stream-is extensively
more the country has been greatly dissected since the
time of terrace formation (P1. IV, Fig. 2; PI. VII, Fig. developed east of the village of Nyaungu (P1. XV, Fig.
1), which is 31/2miles upstream from Pagan. On it the
2), which reduces to a minimum the probabilities of basal ferruginous crust, of T3 age and 1 to 2 feet thick,
finding undisturbed occupation sites. If such sites exist is exposed on both sides of the road leading to Kabani
in the Dry Belt, it is unlikely that they will be discovered
in the Irrawaddy Valley itself. Small side streams and (P1. VIII, Fig. 2), and from this ironstone hardpan,
as well as that found within the area marked "Fuel
tributaries of the main river where erosion was not so Reserve" on Sheet 84 K/16 of the 1-inch maps of
severe should be explored, but in the time available to
Burma, 100 heavily weathered and in some cases rolled
the expedition for field work such a plan was not
feasible. implements were recovered (P1. XV, Fig. 2). Many
display the typical "desert varnish" previously men-
II. THE SITES tioned, which demonstrates that dry conditions must
have pertained during the period when the crust was
Palaeolithic implements were collected in Upper formed.
Burma at a total of 12 sites between Magwe and Ny- 3. Early Anyathian 3
aungu, a distance of approximately 70 miles. The loca-
tion of each is shown on the map, Fig. 20. The tools (a) Magwe (Section 9, Fig. 30, p. 306). Here the
are for the most part heavily rolled, and they appear on cemented ferruginous crust is likewise well developed;
the pebble-strewn terrace surfaces where the gravels it underlies approximately 3 feet of typical T3 gravel.
have been exposed by either erosion or ploughing (PI. The crust itself was not implementiferous at this site,
IV, Fig. 2). A fair proportion was also extracted from but north of Magwe Pagoda, near the village of Wadaw-
sections cut through the terrace deposits, but no actual gyaung, six artifacts were found in association with the
digging was done. The sites are fully described and gravels.
listed in chronological order below; in each case refer- (b) Minbui.-Seven implements were collected by
ence is made to Dr. de Terra's geological report. Drs. de Terra and Teilhard from the T3 gravels at
Minbu, which lies directly opposite Magwe on the right
1. Early Anyathian 1 bank of the Irrawaddy.
(c) Yenangyatng (Section 8, Fig. 28, p. 304). The
(a) Magzwe: Loc. 3 (Section 9, Fig. 32, p. 284). vicinity of Yenangyaung (P1. XIII, Figs. 1 and 2)
This locality is a small gravel pit approximately 65 yards proved to be the most important prehistoric locality in
east of the road from Magwe to Yenangyaung between Upper Burma, in spite of the fact that preliminary in-
milestones 9.2 and 9.3. Here a total of nine very rolled vestigations in this area were somewhat disappointing.
implements was found. This deposit is a remnant of Although the region is much dissected, the cross-bedded
the Lateritic Gravel (Pascoe's "Plateau Red Earth"), gravels of T3 are clearly preserved, and there are many
hence the implements are presumably the same age as fairly good exposures. At a few places these reveal up
those found in T1. At several other localities in this to 12 feet of typical fluvial gravels overlain by sandy
vicinity the Lateritic Gravel was observed between 350 layers (P1. IX, Fig. 1; P1. XIV, Fig. 1). However,
and 400 feet above the level of the river. in most cases the T. deposits have been so reduced by
(b) Chauk-Hill of Chinaulngma(Section 6, Fig. 25, erosion that only a thin layer of the original sediments
p. 301). Here fluvial gravels up to 32 feet thick were remain. This gravel, as elsewhere in Upper Burma,
found at a height of 330 feet above stream level (P1. V, contains fossil wood and silicified tuff as well as many
Figs. 1 and 3). These gravels appear to be the same large to medium-sized quartz pebbles. The same ferru-
age as those of Magwe: Loc. 3; at this site however ginous crust, found at Nyaungu and Magwe, occurs here
they occur on T1. Five heavily rolled implements were at the base of T^, but it contains no implements.
collected on this hill. The terrace gravels of the 90 to 110 foot (T.) level
(c) East of Sale-Hill P. 621 (Section 7, Fig. 27, were investigated southward along the left bank of the
p. 302). As in the case of Chinaungma, the 12-foot- river to Thapangyaung, 712 miles from Yenangyaung.
thick deposit of coarse boulder gravel capping Hill P. This area yielded a total of 226 Early Anyathian 3 im-
621 represents an erosion remnant of a very extensive plements. Between Thapangyaung and Magwe no im-
formation with which large portions of the basin were plements were found, which reflects the fact that no
once covered (P1. V, Fig. 2; P1. IX, Fig. 2). At this coarse gravels were observed in this region. But in the
locality, considered the upper terrace (TI) by Dr. de vicinities of Sadaing (P1. IV, Fig. 2) and Yonzeik, 31/
Terra, nine Early Anyathian implements were found. and 5 miles south of Yenangyaung respectively, good
All are rolled to a pronounced degree. The summit of collections were made. The gravels aroulndThittabwe
Hill P. 621 is approximately 450 feet above the river. and Nyaunghla also proved implementiferous; immedi-
348 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
ately north of Yenangyaung itself, however, no finds (a) Magwe-1/2 mile east of Magwe Pagoda (Sec-
were made. Several fossils, derived from the Upper tion 9, Fig. 30, p. 306). At this site Dr. de Terra dis-
Irrawaddian Beds and described by Dr. Colbert (see p. covered 4 artifacts in a fill deposit of loose, ill-stratified
401), were found associated with the T3 gravels near gravel very similar to that of T4. All the implements
Yenangyaung. are slightly rolled and differ typologically from a typical
(d) Chaltk (Section 6, Fig. 25, p. 301). As at Early Anyathian series.
Yenangyaung, the most clearly defined level at Chauk (b) Pauk, Yaw Valley (Section 10, Fig. 33-B, p.
is the dissected third terrace (P1. V, Fig. 1; PI. VII, 307). Seventeen very typical Late Anyathian imple-
Fig. 2). Here the coarse, red, early Upper Pleistocene ments were found by Drs. de Terra and Teilhard both
gravels are 3 to 5 feet thick and they are preserved on in and on the slopes below the fourth terrace 14 mile
flat surfaces 100-110 feet above the river. At Chauk north of the District Officer's Bungalow at Pauk.
this deposit contains a very high percentage of fossil Since the T4 gravels are the only ones exposed in this
wood debris. The richest archaeological site, a gravel vicinity, there seems to be little doubt as to their place
pit on top of a mesa-like terrace remnant, was discovered of origin.
near the village of Nyaungbyubin, on the southern out- (c) Between Singu and Pagan (Section 5, Fig. 23,
skirts of Chauk (P1. XIV, Fig. 2). Implements were p. 299). Numerous exposures of the typical, medium-
also collected in association with T., deposits between sized, red gravels and sands of T4 exist on both sides
Nyaungbyubin and the pagoda south of Zigyobin. The of the road from Singu (just north of Chauk) and
series from this area is composed of 155 specimens. Pagan (P1. VI, Fig. 4). The best archaeological sites
(e) Pagan (near Nyaungu, see Section 4, Fig. 22, were found between the villages of Monatk6n and
p. 297). Only scattered remnants of the T, gravels Ywatha, 4 to 8 miles south of Pagan, in the general
were observed in this area, although the underlying fer- vicinity of Kinka. Here a total collection of 52 Late
ruginous crust is extensively developed, as previously Anyathian implements was made; in addition a small
stated. Probably the gravels, in which six implements series of very heavily rolled Early Anvathian specimens
were found, were for the most part swept away by the was recovered from the same horizon. These had obvi-
river during the erosion period when T. was formed. ously been derived from T3, which is extensively de-
(f) Pakokeku (see p. 295). Pakokku lies approxi- veloped in this region.
mately 12 miles upstream from Nyaungu on the right No human artifacts were found in direct association
bank of the Irrawaddy. Dr. de Terra collected one with the Pagan Loessic Silt on the T4 level, although
Early Anyathian implement here at a typical T3 gravel it is probable that they occur at its base. As mentioned
exposure. above (see p. 297), two implements, both heavily en-
4. Late Anyathian 1 crusted with silt, were collected on T3 near Nyaungu
(P1. VIII, Fig. 3), and, since these are fresh and com-
A few slightly worn implements of Late Anyathian paratively unrolled, it is very likely that they were once
type occur in the T3 gravels at the following sites: covered by this deposit. However, this point was not
(a) Magwee-1 implement. definitely established.
(b) Yenangyaung-5 implements.
(c) Chauk-10 implements. III. TYPOLOGY

As previously stated (see p. -), these artifacts are re- As stated in the introduction, the Burmese Palaeolithic
as
garded dating from the erosion interval when T, was is mainly characterized by various types of choppers
cut and are therefore stratigraphically earlier than the and chopping-tools which are essentially core imple-
Late Anyathian 2 series from T4. The three sites ments. These are accompanied by a few simple flakes
where tools of this period were found are all described and nuclei. For the most part the choppers, and
above. chopping-tools are of primitive type, very similar in
5. Late Anyathian 2 many respects to the Early Soan of the Punjab (Pater-
son, with de Terra, 1939), and the large choppers found
In Burma tools of Anyathian type continued in use associated with Sinanthropus at Choukoutien, near
during Neolithic times, but the Neolithic implements Peking (Pei, 1931; 1937; 1937-a; 1939; Teilhard de
are usually comparatively fresh in appearance and only Chardin and Pei, 1932; Black, Teilhard de Chardin,
very slightly patinated. These artifacts were found on Young, and Pei, 1933). As in India and China, the
the surface of T3 at all the Early Anyathian sites in- primitive aspect of the Anyathian seems to be in part
vestigated in the Irrawaddy Valley. Associated with connected with the siliceous materials used for manu-
them are a number of very heavily patinated, unrolled facture into tools during Early Palaeolithic times, since
specimens which have been assigned to the Late Any- no flint is known to occur in Burma. The rocks em-
athian series. These were found in place and slightly ployed are of such a nature as to have led not only to
rolled in the T4 gravels. In this latter horizon imple- the development of specialized tool types, but also to
ments were collected at the following sites: regional variations, making a comparison even of con-
MOVIUS: THE STONE AGE OF BURMA 349

temporary series from neighboring localities extremely and Chhibber (1927) has made a study of the origin
difficult. and mineral constitution of a large series. According
to the latter authority (1934, p. 256), "the constitution
1. Raw Material of the fossil wood is siliceous, calcareous, ferruginous
There are two materials-silicified tuff and fossil and carbonaceous. Chalcedony and microcrystalline sil-
wood-which were extensively employed by Palaeolithic ica generally and opal occasionally form the bulk of
Man in Burma. As will be presently indicated the in- the specimens, while calcite and siderite have also par-
fluence of each with regard to the typology of the ticipated in the formation of some others. The origin
of fossil wood is due to colloidal material associated with
implements is considerable.
waters laying down the deposits in which it is preserved.
(a) Silicified Tuff.-Several mutually related mate-
The lithology of the Irrawaddy Series supports such an
rials-tuff, rhyolite, etc.-all silicified and of volcanic
origin, are grouped under the general term "silicified origin. Both mechanical and chemical weathering must
tuff." 1 These are all extensively developed in the vol- have helped in the formation of the colloids, some of
canic regions of Upper Burma, such as Mt. Popa in which seem to have been changed into crystalloidal
the vicinity of Kyaukpadaung (see p. 311 of Dr. de forms afterwards. Fresh-water desert (playa) condi-
Terra's report), the Lower Chindwin Valley in the tions are favorable to the formation of fossil wood,
vicinity of Monywa and the Irrawaddy Valley north of especially when a great deal of alkali is present to de-
Mandalay, as shown by Chhibber (1934, Fig. 19). compose the silicates and liberate colloidal silica to be
Siliceous tuffs are also known to occur near Lagwi Pass deposited in the woody tissues." Chhibber concludes
on the Chinese frontier in the extreme north (Chhibber, that the evidence "seems to point to the climate of the
Irrawaddian being a dry one in Central Burma," which
1934, p. 289). Between Magwe and Chauk the Pleisto- is also Dr. de Terra's opinion (see p. 285).
cene terrace gravels of the Irrawaddy contain an abun-
dant supply. As Morris points out (1935, p. 5), rocks For the most part the fossil wood is patinated a rich,
of this material, which have a fine, homogeneous tex- deep brown, but lighter colors, such as light reddish
ture, possess properties little inferior to the Cretaceous brown, russet, and even a dark shade of buff (almost a
flint of Western Europe. Indeed the flake surfaces cream-color), also occur. Fossil palm usually displays
a yellowish-gray or even a bluish-gray patination, while
display points and bulbs of percussion, radial fissures the vascular bundles show up as dark spots on the sur-
and concentric rings or waves that are identical with the
best flint. The coarser specimens frequently contain face. Chhibber (1927, p. 23) states that the latter
impurities; with these the conchoidal fracture is much closely resembles the wood of the toddy palm (Borassits
less perfect. Nevertheless, in spite of its limitations, flabellifer), at present the characteristic tree of the Dry
it is a material which is not difficult to work. Owing to Belt of Upper Burma. The structure of both types of
the presence of a high percentage of iron minerals in wood is clearly apparent, except in those specimens
the Pleistocene deposits, the prevailing patina is a deep which are more highly siliceous than most of the mate-
reddish brown. But the colors range from light buff rial. In marked contrast to the latter the few imple-
ments which are made from highly siliceous wood debris
(almost cream), through various shades of tan and
have been flaked in several directions irrespective of the
russet, to a deep brown. Usually the patination on the
flaked surfaces is of a slightly lighter shade than the growth-rings of the tree. The bulk of the material is
crust, a portion of which appears on almost all speci- extremely friable, however, and controlled flaking is
mens. In a few cases black ferruginous stains, produc- absolutely impossible except when executed along a
ing a somewhat mottled appearance, are present. plane more or less at right angles to the axis of the
growth rings. This factor is of the utmost importance,
(b) Fossil Wood.-As mentioned by de Terra (see since it has exerted a very marked influence on the
p. 285), there is an almost inexhaustible supply of fossil
wood in the Late Tertiary-Early Pleistocene Irrawaddian typology of the fossil wood implements, most of which
are made on tabular fragments of wood derived from
Beds of Upper Burma. Since extensive outcrops of
near the exterior of the fossil logs found in the Irra-
this formation may be found from the Upper Chindwin
waddian beds. These frequently display such irregu-
and Myitkyina Districts in the north, all the way to the larities as knots and deep pits that occur in a normal tree
Gulf of Martaban on the Indian Ocean, this material
must have been always more readily available to Palae- just below the bark. To judge by the closeness of the
growth-rings, almost all of the dicotyledonous forms
olithic Man than was the far more homogeneous silicified used represent hardwoods. Many specimens exhibit
tuff. Both monocotyledonous woods-the remains of
spotty black ferruginous stains on the surface. Those
palms showing clearly the vascular hundles-and di- derived from the basal crust of T. at Nyaungu fre-
cotyledonous woods-which form the bulk of the mate- quently have a portion of black cemented conglomerate
rial-exist. One dicotyledonous form, Dipterocarpoxy-
adhering to them.
lon burmense has been described by Holden (1916), Mr. William C. Darrah of the Laboratory of Palaeo-
1 According to Dr. de Terra this term is preferable to the botany, Harvard Botanic Museum, comments as follows
name "felsite" used by T. O. Morris (1935, p. 5). regarding the fossil woods from Burma:
350 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
This report deals with some of the problems suggested ment at the time of formation, except to point out that there
by the use of petrified wood in the making of implements is a marked degree of similarity between these fossil plants
by early man in Burma. Seven specimens of wood which and living plants in the same vicinity. This argues for a
have not been worked, i.e. which are not artifacts, were cut rather recent geological age and not very great fluctuations
for purposes of investigation. Four of these belong to a in climate, with the exception of a relatively greater hu-
wood called Dipterocarpoxylon burmense by H. S. Holden. midity.
The other three belong to one or two species of palm, which
are placed in the form-genus Palmoxylon. These have not (c) Other Rocks.-A few implements of vein quartz
been determined to species. I have looked over several and a fine-grained quartzite were collected in Upper
hundred specimens which were used in the making of im- Burma. The former is restricted, however, to the Neo-
plements, and I have found a noteworthy consistent rela- lithic series: two Early Anyathian quartzite implements
tion between the quality of petrifaction and the suitability
for flaking: the more complete the degree of mineralization were found in the Pleistocene gravels of the Irrawaddy
the more likely it is that the specimen would be selected for Valley.
use by man. This can be stated in a more logical manner
by saying that the best implements are made of the poorest No true flint was found in Burma, but an attelm)t
examples of petrified wood. The palm woods, although was made to compare silicified tuff and fossil wood with
they are outnumbered approximately 8:1, are much better
and more easily worked than are the dicotyledonous woods. flint on the basis of quantitative measurements. Mrs.
There is a fundamental relationship of structure and mode M. E. Goodman of Radcliffe College tested these rocks
of growth which probably makes this possible: the palm for density, hardness, resiliency and toughness. using a
has many relatively large scattered vascular bundles
piece of East Anglian (Brandon) flint as the "stand-
throughout the cross-sectional view of a piece of wood, so ard." It was thought that by thus analyzing these
that in almost any direction it may be split across what
might be called the grain or structure. This is more true materials from an objective point of view, it might be
of petrifications than of the fresh-cut wood from living possible to throw some light on their workability and
trees, where the long bundles give considerable strength to the possible influence of this on typology. As shown
the shaft. In marked contrast, the dicotyledonous wood
has a fine grain homogeneous in its construction, and it is by the following table, however, the results were not
worked easily only across the grain, but never with it. particularly convincing with one exception: silicified
This naturally raises the question to what extent does tuff appears to be less resilient and at the same time
silicification or any other mode of preservation follow or it is tougher than either flint or fossil wood. On this
alter the original growth form of the wood. In other words basis one would expect that of the samples tested
to what extent does mineralization and so-called mineral re-
silicified tuff would be the more difficult to work.
placement follow the original structure? No complete an-
swer can be given because all degrees of preservation are
possible, but in the average of hundreds of specimens which SAMPLE DENSITY HARDNESS RESILIENCY TOUGHNESS
I have seen from this and other localities, it is apparent that
the planes of weakness always follow along with the grain, (Rockwell
i.e. in a vertical direction throughout the growing plant, Tester)
and there is a close degree of correlation between original Flint 2.57 95 96 14
structure and mode of preservation. I can find no basis SilicifiedTuff 2.47 92 89 19
for distinguishing the kind of wood used, nor the average Fossil Wood 2.53 93 98 16
conditions of preservation in the three localities represented
in the collections, viz. Yenangyaung, Chauk, and Nyaungu.
Aside from the smaller implements from the third locality
no apparent typological difference can be determined in 2. Nomlenclature
those implements made from fossil wood. In a general discussion of the Burmese Stone Age, it
It would appear to me that these pieces of wood were
selected at random by the original artisans perhaps in com- is necessary to consider at the outset the important fac-
plete ignorance of the differences in structure between the tor of the two materials used in the manufacture of
silicified tuffs and the wood. It would require considerable most of the implements-fossil wood and(silicified tuff.
skill for any one unfamiliar with petrifaction to distinguish Whereas the latter afforded an excellent medium for
some of the poorer examples of petrified wood from some Palaeolithic Man, since with it he could make an allost
of the structureless specimens of tuff.
limitless assortment of tools, the former, due to the fri-
My interest in this problem has been whetted by three
observations. (1) Flint, chert and what we have called able nature of the bulk of the material, flakes in only one
silicified tuff can be worked and chipped in any plane. direction: a plane transverse to the axis of the growth-
(2) The fossil palm wood, owing to the peculiarities of its rings, as explained above by Mr. Darrah. Hence nearly
mode of growth, can also be worked in almost any manner. all the implements in fossil wood, from earliest times
(3) The dicotyledonous wood on the other hand can only
be worked satisfactorily across the grain, if the conditions through the Neolithic, are of the same fundamental
of preservation are at all good. type. But, although heavy forms are found associated
Two critical notes are necessary in this report. First, with the Late Anyathian and Neolithic cultures, there
the wood called Dipterocarpoxylon in this paper is not is a decided tendency, in spite of the difficulties imposed
closely identified with the living tree called Dipterocarpus. by the fossil wood, to develop smaller tools. These can
It seems inappropriate at this time to enter into a con-
be correctly classified as scrapers. In the Early Any-
troversy over the identity and affinity of this form. Second,
since the silicified woods are derived from gravels, it seems athian, however, coarse implements of the same basic
inadvisable to enter into a discussion of age or environ- type predominate; in fact they are as common as the
MOVIUS: THE STONE AGE OF BURMA 351
hand-axes associated with the classic Lower Palaeolithic (c) Haud-Adzee.-As explained above, a hand-adze
cultures in other parts of the Old World. Since they is a type of chopping implement of roughly tabular form
apparently relresent a type of chopping implement (Figs. 58 and 59, Nos. 6, 7, 9-15; Fig. 64, Nos. 34 and
rather than a tool intended solely for scraping, and since 35; P1. XVII, Nos. 1 and 2). It is made on a core
they are worked only on the upper surface of either one and has a straight, slightly rounded or even a pointed
(e.g., Fig 59, No. 10) or, in some cases, of both (e.g., cutting edge, which forms a right angle with the long
Fig. 60, No. 18) ends, they clearly belong to the adze axis of the implement. Hand-adzes may be regarded
rather than to the axe family of tools. It is therefore as a special class of chopper, but of square or rectangu-
proposed to define more precisely the much overworked lar rather than of round or oval form. The secondary
term "crude or massive scraper," in order to classify working along the edge is restricted to the upper sur-
these core implements, and the name "hand-adze"is sug- face. It is on this basis that a distinction is made be-
gested for them. In Burma hand-adzes occur in silici- tween hand-adzes and hand-axes, since the latter are
fled tuff (Fig. 64, Nos. 34 and 35) as well as in fossil bifacial implements-i.e. they are flaked on both
wood, and they are true core tools in every respect. surfaces.
Chopping-tools with alternately flaked edges are also (d) Scrapers.-The essential difference between
found in both materials, in addition to choppers flaked choppers and scrapers can be determined only by size,
on one surface only, as well as other types of cutting the term scraper being reserved for small tools made on
and scraping implements. On this basis it seems pref- both cores and flakes that do not fall into the chopper
erable to refer to the Early Anyathian (Lower Palaeo- category (Fig. 61, No. 23; Fig. 68, No. 48; Figs. 71
lithic) of Burma as a chopper-chopping-tool-hand- and 72, Nos. 67-76).
adze complex in which, although hand-axes are con- It will be recognized that no absolutely rigid line can
spicuously absent, core implelmentspredominate. be drawn between the above classes of implements in all
It should be made clear that the criteria used in this cases. The chopping-tools with alternately flaked edges
report for distinguishing the various categories of im- are distinctive, but even here certain crude types may
plements are based on the form and technique of manu- converge on choppers which have been flaked on the
facture of the tools themselves, rather than on their pre- lower as well as the upper surface. Again, small chop-
sumed function. As stated above, practically all the pers and scrapers, especially of the crude variety, cannot
material consists of core implements used for chopping, be differentiated on an objective basis, and gross size
cutting and scraping; indeed it is very likely that many seems to be the only safe criterion. The most easily
tools served all these purposes. It is therefore clear determined category is perhaps that for which the term
that form and technique rather than function must be hand-adze is proposed here; in a general sense this may
used as a basis for classification. In other words, when be considered an uniface implement as opposed to the
the terml chopler is employed, reference is imade to a hand-axe, which is bifacial. As emphasized above,
special shape of tool manufactured in a particular man- these distinctions are based on form and technique of
ner, rather than to an implement intended solely for manufacture, rather than on function, which in the case
chopping. In this sense the following definitions are of prehistoric tools can only be arrived at by inference.
advocated, and an attempt has been made to adhere But for purposes of studying and describing a large
rigorously to thelm throughout. collection it is necessary to subdivide the implements
(a) Choppers.-Admittedly when dealing with those into more or less uniform groups, each of which in-
implements commonly referred to as crude scrapers, cludes one given type of tool. In Burma the bulk of
massive scrapers or choppers, no hard and fast rules the Stone Age material falls into one or the other of the
can be made, since the terms themselves are synony- four categories defined above.
mous. In the interests of uniformity all are considered In dealing with such extremely primitive tools as
choppers in this paper. For the most part choppers some of the single-edged choppers of silicified tuff and
have round, semi-oval or almost straight cutting edges hand-adzes of fossil wood, it is difficult to make a clear
which have been formed by the removal of flakes on the distinction between the work of man and the work of
upper surface of the implement only (Figs. 63 and 64, natural agencies. This applies only to those specimens
Nos. 28-33; PI. XVIII, No. 1), but certain types of in the above categories which display the minimum
choppers are flaked, albeit not extensively, on the lower amount of flaking along a given face or edge. To make
surface as well (Fig. 62, No. 27; P1. XVII, No. 4). the problem even more difficult, as far as material col-
In some cases choppers are also made on flakes (Fig. lected in the T1 and T3 gravels is concerned, many of
68, Nos. 46 and 49). the flake facets have been subjected to such a pro-
(b) Chopping-Tool.-This term is used with refer- nounced degree of rolling that they are often scarcely
ence to a well-known form of implement the cutting discernible. However, certain criteria are helpful, al-
edge of which has been produced by alternate flaking though not by any means conclusive. Among these the
(Fig. 61, No. 24; Figs. 65-67, Nos. 36-45; PI. XVIII, following have been rigorously adhered to: (1) typologi-
No. 4). Chopping-tools are always core implements cal comparison with similar material known to be human
usually made on pebbles, and they have sinuous edges. and from the same deposit; (2) comparison with un-
352 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

1 2 3
Scale
0 s 10 Cm,
I~~~~~~~~~~~~~~~~~~~~~~

4 5
FIG. 57. Early Anyathian 1 Implements. Nos. 1-3 and 5: Steep-ended Hand-adzes; No. 4:
Small Chopping-tool with Alternately Flaked Edge, Made on a Pebble. Nos. 1 and 2: T1, East of
Sale-Hill P. 621; No. 3: Ti, Chauk-Hill of Chinaungma; Nos. 4 and 5: Lateritic Gravel, Magwe-
Loc. 3. Nos. 1, 3 and 5: Fossil Wood; No. 2: Quartzite; and No. 4: Silicified Tuff.
MOVIUS: THE STONE AGE OF BURMA 353
rolled objects of the same type from later sites; (3) tabular or in pebble form, with the minimum amount of
observation of the nature of the flaking on rolled stones primary flaking-two or three such flakes in each case.
on the banks of the Irrawaddy and in the beds of side (d) The hand-adzes are all steep-ended, and the
streams; and (4) actual experimentation on samples of pointed type is predominant.
fossil wood and silicified tuff. All dubious specimens (e) All three materials known to occur in the Any-
were discarded either in the field or during the course of athian: silicified tuff, fossil wood and quartzite, have
laboratory study, and these are not considered in this been utilized.
report. Thus Early Anyathian 1 may be regarded as repre-
senting a rather crude and generalized phase of the
IV. THE EARLY ANYATHIAN CULTURE Anyathian culture which later became somewhat more
In the Early Anyathian large implements are typical; specialized.
in fact almost all of the specimens are over 10 cm. in
their longest dimension. For the most part the speci- 2. Early Anyathian 2-3 Implements of Fossil Wood
mens have been made on natural fragments of rock
Although certain types of tools are common to both
either in roughly tabular or in pebble form, and they the fossil wood and the silicified tuff series, it is neces-
usually display a large area of cortex. Nearly all the sary for purposes of description to consider each series
tools are very heavily rolled, especially those from the
T1 and Lateritic Gravel deposits at Chauk, east of Sale, separately. Fossil wood implements were collected at
all the Early Anyathian 2 and 3 sites, but artifacts of
and near Magwe: Loc. 3., Throughout the entire series silicified tuff occur mainly from Chauk southward in
there is very little typological change in the fundamental the valley.2 Thus all the tools made of this rock from
tool types, although the oldest phase contains very few the T, deposits belong to Phase 3 of the Early Any-
specialized forms. athian. They are in fact more abundant than the fossil
1. Early Anyathian 1 wood artifacts at Yenangyaung, whereas the opposite
is true at Chauk. Before describing the material it
Of the 23 implements from the earliest sites, 17 are should be pointed out that the fossil wood implements
of fossil wood, 5 of silicified tuff and 1 of quartzite. A of Phase 2 (represented
entirely by fossil wood tools)
typical selection is illustrated in Fig. 57. It will be and of Phase 3 (both materials in common use) are
noted that the hand-adzes, Nos. 1-3 and 5, are all of
separable on stratigraphical evidence only. The same
the steep-ended variety including No. 2, which is made forms exist in both
phases; a careful analysis of the
on a flat quartzite pebble. The others are of fossil collection failed to reveal
any very significant typological
wood, and the grain of the wood is plainly visible. difference, in spite of the fact that local factors have
Pointed hand-adzes, Nos. 1-3 are typical of this phase; contributed to the at each site. The silici-
round-ended forms seem to be rare, since only two are fled tuff series willdevelopment
be considered in the next section of
represented in the collection. No. 5 is a steep, square- this
ended type, of which there are five examples. Crude report; here the implements of fossil wood are
discussed.
chopping-tools with alternately flaked edges, such as No. (a) Single-eded Hanld-Ad.ze (Fig. 58, Nos. 6, 7
4, are all made of silicified tuff; three specimens of this and
9; Fig. 59, Nos. 10-12; P1. XVII, No. 1). Over
type were found. Throughout the entire series of chop- one-half of the fossil wood
pers, chopping-tools' and steep-ended hand-adzes the fall into this implements in the collection
category. All are tabular fragments which
cutting edge has been produced with the minimum have been
amount of work by the removal of two or at most three roughly flaked across one end to produce a
flakes. Secondary edge-chipping, the result of use, is sharp adze-like edge, whereas the other five surfaces are
natural. The illustrated specimens show the range in
displayed in all cases, but no retouching occurs. No
flakes or nuclei were found. size from large, massive forms, Nos. 6, 7, 9 and PI.
The general characteristics of the Phase 1 of the XVII, No. 1, through medium-sized tools, such as No.
be summed up as follows: 10, to smaller examples, Nos. 11 and 12. The drawings
Early Anyathian may
(a) Both the fundamental types of implements char- have been somewhat simplified, since actually all display
acteristic of the Burmese Palaeolithic-the hand-adze a pronounced degree of rolling,3 and edge-chipping re-
and the chopping-tool made on a pebble-are present. sulting from use is apparent. Square-ended, No. 6, and
(b) There is no essential difference between the im- round-ended, Nos. 9, 10; 12 and PI. XVII, No. 1, adzes
plements from T1 at Chauk and east of Sale, and those are typical although pointed specimens, Nos. 7 and 11,
from the Lateritic Gravels at Magwe: Loc. 3. This is also occur. The angle of the cutting edge varies be-
in accord with Dr. de Terra's conclusion (see p. 312) tween approximately 45? and 75?
that both deposits are more or less the same age-Sec- 2
Only one implement of silicified tuff was collected at
ond Pluvial Period.
Nyaungu.
(c) In all specimens the working edge has been pro- 3 This comment applies to
practically all the Early Anyathian
duced on a natural fragment of rock, either roughly material illustrated in this report.
354 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

Scale
0o s 10 cm.

8 9
FIG. 58. Early Anyathian 2-3 Implements of Fossil Wood. Nos. 6, 7 and 9: Large Hand-adzes;
No. 8: Pointed Double-sided Chopper. Nos. 6, 7 and 8: Chauk; No. 9: Nvaungu.
MOVIUS: THE STONE AGE OF BURMA 355

/ -~ulc
.1I

10 11 12

Scale
0 . . .... .
5., =
10 cm.
----

I
I
! I

13 14 15

FIG. 59. Early Anyathian 2-3 Implements of Fossil Wood. Nos. 10-12: Medium- and Small-sized
Hand-adzes; Nos. 13-15: Steep-ended Hand-adzes. Nos. 10 and 12: Nyaungu; Nos. 11 and 13: Pagan; Nos.
14 and 15: Chauk.
356 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
(b) Single Steep-ended Hand-Adze (Fig. 59, Nos. type confirms the results of experimentation with fossil
13-15). These are normally made on short thick frag- wood, as it has been found that controlled flaking with
ments of fossil wood debris, and they are very similar the grain of the wood either transversely or longitudi-
to the specimens described above, with the exception nally is virtually impossible. This accounts for the
that the angle of the cutting edge is always over 75?. overwhelming predominance of tools of the adze variety
Frequently, as in the case of No. 13, it forms nearly a in which either one or both of the ends have been sharp-
right angle. This artifact has been used at both ends, ened. Indeed No. 22 is, strictly speaking, a double-
since several flakes have been removed from the butt ended hand-adze roughly flaked along one side in an
to form a double adze-like implement with one very effort to produce a chopper or massive side scraper.
steep edge. For the most part the working edge of the No. 21 is a better example; one end has been fashioned
steep-ended hand-adzes are rounded, but a few pointed into a very crude beaked point formed by a distinct con-
types, identical with those of Early Anyathian 1 also cavity at the lower extremity of the cutting edge. The
occur. No. 15 is unusual in that the grain of the wood opposite end is unworked.
is transverse to the main axis of the implement. (h) Side-and-End Scraper (Fig. 61, No. 23). There
(c) Double-ended Hand-Adze (Fig. 60, No. 18). are very few tools in the Early Anyathian which can
Although only one example of a double-endedhand-adze be classified as true scrapers, and of these No. 23 is a
with the edge produced by flaking from the lower sur- good example. In the case of this specimen, found in
face of the implement is shown, this type is fairly com- the T3 gravels at Pakokku, the flaking is independent
mon. The illustrated specimen is interesting, since one of the grain of the wood. A small notch occurs at one
end is straight whereas the opposite end is round. Both end of the cutting edge, which extends along one side
display characteristic edge chipping-the result of use. and across the end of the implement.
(d) Inverse Double-ended Hand-Adze (Fig. 60, (i) SwmallScraper with Scalloped Edge (Fig. 61,
Nos. 16 and 17). These implements are not as common No. 25). Two examples of these small scrapers were
as the type just described; a total of 15 was found, found-one at Chauk and one at Pagan-and both were
whereas there are 20 of the category (c) specimens. in direct association with the T3 gravels. The ends
The cutting edges are more or less parallel to each have been produced by alternate flaking, but the side is
other at opposite ends of the tabular fossil wood frag- worked on one surface only in order to produce a double
ment from which the tool is fashioned. However, one end-and-side scraper. Where the ridges separating the
cutting edge has been formed by the removal of flakes flake scars intersect the edge there are marked projec-
from the lower surface of the specimen, whereas in the tions which give the implement a scalloped outline. Al-
case of the other, the flakes have been struck from the though made of silicified tuff, the palaeolith described
upper surface downward. In this manner a double- by Mr. T. O. Morris (1936) as a "Chelleo-Acheulean"
ended implement with a roughly parallelogram-shaped hand-axe from Yenangyaung is in reality a small
section has been produced. Size varies from large pointed scraper with an alternately flaked edge very
massive forms, No. 16, to smaller and more nearly similar to the fossil wood example, Fig. 61, No. 25. in
average-sized types, such as No. 17. every respect.4 Mr. Morris' specimen displays only one
(e) Concave-ended Hand-Adze (Fig. 60, No. 19). worked side and is just under 7 cm. long; it is quite
Hand-adzes or large scrapers with concave edges are definitely not a hand-axe.
rare; only 3 were found in Upper Burma. Implements (j) Chopping-Tool with Alternately Flaked Edge
in this category, probably used for scraping rather than (Fig. 61, No. 24). Implements of this type are com-
cutting, differ in no essential particular as to method of monly found in the Early Anyathian 3 silicified tuff
manufacture from the typical hand-adzes, with the series; only 9 fossil wood examples were recovered.
single exception that the cutting edge is concave. The This seems to be directly connected with the nature of
illustrated specimen was found in situ in the gravel pit the fracture of the latter material, and it makes a com-
near Nyaungbyubin, on the southern outskirts of Chauk. parison of the two series very difficult. In No. 24 the
(f) Single-ended and Side Hand-Adze (Fig. 60, cutting edge has been produced by the removal of
No. 20; PI. XVII, No. 2). These implements are like- alternating flakes in such a way that when viewed from
wise variants of the typical hand-adze and differ only in above the edge itself is in the form of a very broad W.
that one side has been partly sharpened to produce a The flake scars are rather deep and they do not carry
second cutting edge. No. 20 is made from a particu- far back from the point of impact. It was found that
larly fine-grained and homogeneous fossil wood frag- it was possible to reproduce these implements by using
ment, and for this reason its flake facets are clearer and a hammerstone and a direct percussion technique on a
less rounded than those displayed by the bulk of the piece of fine-grained, homogeneous fossil wood. Fri-
Early Anyathian material. P1. XVII, No. 2, is a larger able material cannot be worked in this manner. No. 24
example and in it the grain of the wood is coarser, which 4 We wish to take this opportunityof thanking Mr. C. F. C.
explains the primitive aspect of the flaking. Hawkes, F.S.A., Assistant Keeper of British and Mediaeval
(g) Side Hand-Adze or Chopper (Fig. 61, Nos. 21 Antiquities in the British Museum, for kindly furnishing the
and 22). The extreme rarity of implements of this Peabody Museum with a cast of this implement.
MOVIUS: THE STONE AGE OF BURMA 357

17

Scale
16 0{ 5, I 10 cm.
j

19 20

18
FIG. 60. Early Anyathian 2-3 Implements of Fossil Wood. Nos. 16 and 17: Inverse Double-ended Hand-
adzes; No. 18: Double-ended Hand-adze; No. 19: Concave-ended Hand-adze. No. 16: Nyaungu; No. 17:
Yenangyaung; Nos. 18-20: Chauk.
358 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

21 22

Scale
0 5. 10 cm.
L ? ? II

23

I 1
I I

24 25
FIG. 61. Early Anyathian 2-3 Implements of Fossil Wood. Nos. 21-22: Side Hand-adzes or Chop-
pers; No. 23: Side-and-End Scraper; No. 24: Chopping-Tool with Alternately Flaked Edge; No. 25: Small
Scraper with Scalloped Edge. Nos. 21-22: Chauk; No. 23: Pakokku; No. 24: Magwe; No. 25: Pagan.
MOVIUS: THE STONE AGE OF BURMA 359

26

Scale
0 sI 10
I
cm.
I

27
FIG. 62. Early Anyathian 2-3 Implements of Fossil Wood. No. 26: Chopping-Toolwith Rounded Edge;
No. 27: Double-ended
Chopper. All fromChauk.

is made on a well-rolled fragment, an angular edge on gravels. It is really a double-ended hand-adze which
which forms an extension of the old fashioned by al- has been used for chopping, as demonstrated by the
ternate flaking. stepped flakes removed from the lower surface. These
(k) Chopping-Tool with Rounded Edge (Fig. 62, would not have been detached if the artifact had served
No. 26). Only 4 specimens in this category were found only as an adze. Single-ended choppers of this type,
at Chauk, and they represent definite attempts to manu- such as P1. XVII, No. 4, are more common, as 19 ex-
facture typical chopping-tools of fossil wood. As in amples were collected.
Nos. 15 and 20-25 the flaking has been executed irre- (m) Pointed Double-sided Chopper (Fig. 58, No.
spective of the direction of the grain of the wood to 8). If this specimen had been flaked on the lower sur-
form a tool with a convex outline. Although the flaking face, it could be classified as a biface or hand-axe. It is
is continuous around three sides of the upper surface, made of a very fine-grained triangular piece of fossil
it is more or less haphazard on the lower surface and wood, from which broad flat flakes have been detached
is restricted mainly to one side. Nevertheless a fairly along two sides in order to form a double-sided chopper
satisfactory cutting edge has resulted. with a crude point. On the right side extensive edge
(1) Single and Double-ended Chopper (Fig. 62, No. chipping, the result of use, is apparent. This imple-
27; PI. XVII, No. 4). No. 27 is one of the two double- ment, which was found at Chauk, is unique in the collec-
ended choppers, both from Chauk, found in the T3 tion from Upper Burma.
360 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

(n) Pick-like Implement (P1. XVII, No. 3). This when the total collections from Yenangyaung and
implement, found in situ in the ferruginous crust ex- Chauk-both Early Anyathian 3 sites-are considered,
posed on T3 at Nyaungu, is the only one of its type there is a greater degree of local variation displayed,
found in the Irrawaddy Valley. It is a large tool- with regard to implement types, than that which exists
approximately 25 cm. long-and is made on a massive between the fossil wood series from Nyaungu and those
fragment of fossil wood with a smooth "desert varnish" from the T3 gravels at other localities. But certain ob-
on the lower surface. The crudely pointed end, which servations can be made. Thus, as shown on Table 3,
shows signs of use, has been formed by the removal of the more complex and evolved forms are not as common
large flakes extending only a short distance back in the at Nyaungu as at Chauk, which may be regarded as
direction of the butt. This gives the implement a the type site for the Early Anyathian 3 fossil wood
shouldered appearance. series. This is especially apparent in the case of the
In summarizing the main characteristics of the fossil double-ended and inverse double-ended hand-adzes,
wood series, the most important single feature is that it types that are well represented at Chauk. At Nyaungu
is composed almost exclusively of tabular "end" imple- many of the large and medium-sized hand-adzes are
ments of the hand-adze variety, and that chopping-tools, of the steep-ended rather than of the ordinary form;
in addition to forms which have been flaked along the the steep-ended hand-adze is a primitive form charac-
side, are comparatively rare. Hand-axes and true bi- teristic of Early Anyathian 1 (Fig. 57, Nos. 1 and 5)
facial artifacts are completely absent, as are cores and from T1 and the Lateritic Gravel. On the other hand
flakes. Thus the nature of the fossil wood itself seems a greater proportion of the tools from the basal T3 crust
to have been the dominant factor controlling the ty- at Nyaungu are small, in fact the "small hand-adze"
pology of this series. On the whole there are very few category, as shown on the above table, includes very
small implements, as the bulk of the material is more much smaller forms from Nyaungu than from Chauk.
than 10 cm. in the longest dimension. With regard to It is therefore evident that, if size is regarded as im-
the hand-adzes, it is ihteresting to note that round- portant, Nyaungu is somewhat more advanced than
ended types are more common than are those with Chauk. But the fossil wood debris at the latter site is
square ends; pointed forms are comparatively rare ex- larger than at Nyaungu which perhaps explains the size
cept in Early Anyathian 1. Table 3 shows the relative factor. The only really significant difference between
frequency of the various fossil wood tool types of the two localities is that at Nyaungu the heavy, steep-
Phases 2 and 3 at the main sites. ended hand-adzes are fairly common and colmprisejust
Since Early Anyathian 2 is known from only one over one-fifth (21.21 per cent) of the entire series.
site--the basal ferruginous crust of T, at Nyaungu-it This is a characteristic Early Anyathian 1 form, and
can best be compared with Early Anyathian 3 material. on this basis Nyaungu can perhaps be considered as
There is, however, very little typological difference be- representing a transitional phase between Early Any-
tween the two phases, as previously stated. In fact athian 1 and Early Anyathian 3.

TABLE 3
THE RELATIVE FREQUENCY OF THE VARIOUS TYPES OF FOSSIL WOOD IMPLEMENTSFOUND AT THE THREE MAIN EARLY ANYATHIAN
SITES. In each case the percentage frequency of each type has been computed in terms of the total collection from the site; the
percentages in the last column are expressed in terms of the total collection from all three sites.

FOSSIL WOOD IMPLEMENTS YENANGYAUNG CHAUK NYAUNGU TOTALS

3 (8.57%) 8 (6.30%) 3 (3.03%) 14 (5.36%)


Single-ended Hand-Adze: (1) Massive
(2) Large 7 (20.60%) 15 (11.81%) 8 (8.08%) 30 (11.49%)
(3) Medium 8 (22.86%) 22 (17.32%) 10 (10.10%) 40 (15.33%)
(4) Small 4 (11.43%) 26 (20.47%) 29 (29.29%) 59 (22.61%)
5 (14.29%) 9 (7.09%) 21 (21.21%) 35 (13.41%)
Single Steep-ended Hand-Adze
Double-ended Hand-Adze 1 (2.86%) 13 (10.24%) 6 (6.06%) 20 (7.66%)
Inverse Double-ended Hand-Adze 1 (2.86%) 11 (8.66%) 3 (3.03%) 15 (5.75%)
Concave-ended Hand-Adze 1 (0.79%) 2 (2.02%) 3 (1.15%)
1 (2.86%) 2 (1.57%) 3 (1.15%)
Single-ended and Side Hand-Adze
Side Hand-Adze or Chopper 2 (1.57%) 3 (3.03%) 5 (1.92%)
Small Scraper with Scalloped Edge 1 (0.79%) 1 (0.38%)
4 (11.43%) 2 (1.57%) 3 (3.03%) 9 (3.45%)
Chopping-Tool with Alternately Flaked Edge
4 (3.15%) 4 (1.53%)
Chopping-Tool with Rounded Edge
1 (2.86%) 10 (7.87%) 10 (10.10%) 21 (8.05%)
Single and Double-ended Chopper
Pointed Double-sided Chopper 1 (0.79%) 1 (0.38%)
Pick-like Implement 1 (1.01%) 1 (0.38%)

'1rOTALS: 35 127 99 261


MOVIUS: THE STONE AGE OF BURMA 361

28 0
Scale
5 lo cm.

29

30
FIG. 63. Early Anyathian 3 Implements of Silicified Tuff (Nos. 28 and 29) and
Quartzite (No. 30). Large Choppers Flaked Only on the Upper Surface. No. 28:
Yenangyaung. Nos. 29 and 30' Chauk.
362 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
3. Early Anyathian 3 Implements of Silicified Tuff (c) Massive Pick-like Implement (P1. XVIII, No.
The silicified tuff implements are for the most part 2). Only one implement made of silicified tuff which
may be classified as a pick was found in Upper Burma.
larger and more massive than the fossil wood series. It is a far more massive specimen than the fossil wood
Although they differ typologically, both have one essen-
tial feature in common; the desire to produce a working example (P1. XVII, No. 3) described on p. 360, and it
has been made on a large sub-angular pebble of tuff,
edge on either tabular or round pebbles with the mini-
mum amount of effort. Nearly 87 per cent of the silici- approximately 20 cm. long, with a roughly triangular
section. The flaking extends from the point in the
fled tuff implements are from Yenangyaung, where
fossil wood forms occur with almost the same frequency direction of the butt. This implement was found by
as do silicified tuff specimens at Chauk. Thus the two Dr. Teilhard in situ in the T3 gravels immediately south
sites complement each other with regard to the materials of Yenangyaung.
used for manufacture into implements. Only one tool (d) Hand-Adze or Chopper Made on a Pebble (Fig.
of silicified tuff was found at Nyaungu, a medium-sized, 64, Nos. 32 and 33). These thick pebble-tools have
one more or less square cutting edge formed by the
single-ended chopper of the type found in fossil wood detachment of flakes from the upper surface of the
(see category (1) above). In addition a few examples
were found at Magwe and Minbu, but Yenangyaung is implement. At Yenangyaung they form a distinctive
the type locality. category, but they are unknown elsewhere. Away from
the edge, the flaked surface rises steeply, as the result
(a) Chopper Flaked Only on the Upper Surface either of hinge fracture (No. 32) or of the removal of
(Fig. 63, Nos. 28-30; Fig. 64, No. 31; P1. XVIII, No.
short, deeply concave flakes (No. 33), so that the im-
1). As the illustrated specimens clearly show, the
choppers made of silicified tuff display a far greater plements are very thick in proportion to their length.
In No. 32 two flakes have been struck from the butt
degree of variability than does the fossil wood series.
The tools discussed here are all made on either angular end; No. 32 has been sharpened along one side, prob-
(No. 28), flat (No. 29; P1. XVIII, No. 1), or rounded ably for use as a chopper or small adze. Specimens in
this category vary in size; they apparently represent
(Nos. 30 and 31) river pebbles. Actually No. 30 is
of a fine-grained quartzite, which possesses a fracture attempts to fashion hand-adzes or choppers (?) from
river pebbles.
very similar to silicified tuff. The flaking on most of
the implements is restricted to one end or one side, but (e) Tabular Hand-Adze (Fossil Wood Type) (Fig.
on No. 31 it extends around three sides of the specimen, 64, Nos. 34 and 35). With the exception of the mate-
while the butt has had several flakes removed to reduce rial from which they are made, there is a small series
the thickness. Edge chipping as a result of use is ap- of silicified tuff implements which are indistinguishable
from the typical fossil wood hand-adzes. Both square-
parent in all instances, especially along the left side of
No. 31. In spite of the fact that silicified tuff can be ended (No. 34) and round-ended (No. 35) tabular
easily fractured by direct percussion with a hammer- forms are known, on which the cutting edge has been
stone, very few implements have been worked except produced by striking blows from the lower surface. In
along the cutting edge, and elsewhere on the surface the all cases these specimens are made on natural pebbles;
original cortex of the pebble remains. Nos. 30, 31, the angle of the cutting edge varies from approximately
and P1. XVIII, No. 1, are rolled to a pronounceddegree, 45? to 75?. No steep-ended types, with edges at an
whereas in Nos. 28 and 29 the facets are quite clear. angle greater than 75?, were found.
As in the case of the fossil wood implements the draw- (f) Chopping-Tool with Alternately Flaked Edge
ings are a compromise between realistic and diagram- (Fig. 65, No. 36; Fig. 66, Nos. 39-41; Fig. 67, No.
matical representations-i.e., the flake scars have been 44; P1. XVIII, No. 4). Chopping-tools with alter-
made to appear fresher than they actually are for the nately flaked edges may be considered the type imple-
sake of clarification. ments of the Lower Palaeolithic of Burma, as far as
(b) Massive Triangular Chopper (P1. XVIII, No. the Early Anyathian silicified tuff series is concerned.
3). This implement, which is 23.5 cm. long, appears The illustrated specimens, all from Yenangyaung, show
to have been made on a flat river pebble which originally the typological range of this class. They are made on
possessed a roughly triangular outline. No attempt has pebbles, carefully worked by alternate flaking along one
been made to produce a pointed bifacial tool which might edge, to produce a very effective chopping-tool. In
be classed as a hand-axe, but along each of the three most cases angular or sub-angular pebbles were selected,
sides secondary working is apparent. The flaking is and the edge, which has been battered and chipped by
of the stepped variety; as a result of rolling, however, use, has been produced by the removal of two main
the flake scars are extremely indistinct. As the photo- flakes on one side and three on the other side of the
graph shows, numerous small pits, due to the impure implement. Nos. 36, 39 and 41 are typical specimens
nature of the silicified tuff, are present on the surface with cutting edges forming an angle of less than 70?;
of the specimen. This implement, found at Chauk, is the cutting edge of P1. XVIII, No. 4, is approximately
unique in our collections from the Irrawaddy Valley. 80?. In No. 40 the edge is greater than 90? and the
MOVIUS: THE STONE AGE OF BURMA 363

3'
34

Scale
0 c.
cm....o

32

35
33
FIG. 64. Early Anyathian Implements of Silicified Tuff. No. 31: Chopper Flaked Only on the Upper
Surface; Nos. 32 and 33: Hand-Adzes or Choppers Made on a Pebble; Nos. 34 and 35: Tabular Hand-Adzes
(Fossil Wood Type). Nos. 31-34: Yenangyaung: No. 35: Chauk.
364 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
extreme degree of battering which it displays shows that of No. 44, a small chopping-tool made on a flattened
this implement was probably used for breaking and river pebble. all these hand implements are extensively
crushing rather than cutting. A broad, flat flaking tech- rolled.
nique has been employed in the case of No. 41, and, al- (g) Chopping-Tool with "Overthrust" Edge (Fig.
though three flakes have been detached on one side, the 65, Nos. 37 and 38). Characteristicof the Early Any-
opposite surface reveals only one scar. As in the case athian are pebble chopping-tools with high angle (75?-

36

37

Scale
0 5 10 cm.
I

.10

38
FIG. 65. Early Anyathian Implements of Silicified Tuff. No. 36: Chopping-Tool with Alternately Flaked Edge
Nos. 37 and 38: Chopping-Tools with "Overthrust" Edge. All from Yenangyaung.
MOVIUS: THE STONE AGE OF BURMA 365

Scale
o s ~~~~~~~~~~~10
CM.

I ---? I---
~ I
:?i~~~~~~~~

t~~~~:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
i i
I
I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
40
39 II

41~~~~~~~~~~~~~~~~~i
FIG. 6. Erly 3Implmentsof
nyathan Sliciied uff.os.
9-41 Chop;.ngToolwit
(~~~~~lentl lkd de l rmYnnyug

~~~~~~~~~~~~41~~~

FIG. 66. Early Anvathian 3 Implements of Silicified Tuff. Nos. 39-41: Chopping-Tools with
Alterniatelv Flaked Edge. All from Yenangyraung.
366 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

42
43

Scale
0 . I
5 O1cm.
--
- ' =

44 45
FIG. 67. Early Anyathian 3 Implements of Silicified Tuff. Nos. 42 and 45: Chopping-Tools Made on Flat
Pebbles; No. 43: Chopping-Tool or Core (?); No. 44: Chopping-Tool with Alternately Flaked Edge. All from
Yenangyaung.
MOVIUS: THE STONE AGE OF BURMA 367

85?) cutting edges, which when viewed from above secondary working and edge chipping resulting from use
display a marked projection or "overthrust" appearance. along either the side (No. 46) or the end (No. 49). It
This is produced by the removal of two flakes more or seems likely that they were used as rough scrapers or
less at right angles to a previously prepared striking choppers of some sort. Types showing fairly careful
platform. The striking platform was next removed by core preparation prior to detachment (Nos. 47, 48 and
a blow falling exactly at the intersection of the two 51) are not common; but implements produced in this
flakes, so that a sort of "chapeau de gendarme" cutting fashion demonstrate the workability of silicified tuff.
edge was formed, the projection being caused by the In Nos. 47 and 50 the striking platform is at a high
ridge separating two flake scars. Further flaking on angle (greater than 90?) to the long axis of the artifact.
one or both of the sides was employed to extend the This may possibly be connected with the so-called "anvil
edge as far as required. In No. 37, a small example, technique," although the same result can also be obtained
the overthrust portion curves slightly downward and by direct percussion with a hammerstone. The edge
outward, but in No. 38 it is more pointed and beak- chipping is restricted to the broad end of No. 50; how-
shaped. These implements represent a very specialized ever, it extends along both sides as well as the end of
type of chopping-tool thus far known only from Early Nos. 47 and 51. No. 48 is a large, thick, steeply-
Anyathian 3 at Yenangyaung, where 10 specimens were worked, double side and end scraper. Along the left
found, and Chauk, which produced one example. side as well as across the end resolved flaking is present.
(h) Chopping-Tool made on a Flat Pebble (Fig. 67, No true points or perforators were found, No. 51 being
Nos. 42 and 45). Flat tabular or oval pebbles have been the nearest approach to a flake implement of this type
used for making these artifacts, and No. 42 is a typical in the collection. As in the case of the core tools there
example. The edge has been flaked in two directions is abundant evidence of severe water rolling and abra-
in such a way that it extends across the pebble. No sion shown by these artifacts.
definite attempt at alternate flaking has been made, but (k) Cores or Nuclei (Fig. 69, Nos. 52-55). The
simple chopping-tools of this type with bifacially worked Early Anyathian cores all display roughly prepared
edges must have been fully as effective cutting imple- striking platforms formed by the removal of a large
ments as were those made by the more evolved tech- trimming flake from one face of a natural pebble of
nique. The small pointed specimen, No. 45, is unique. silicified tuff. They vary as to both size and the rela-
It is the only implement in the Palaeolithic collection tion of the striking platform to the long axis of the
from Burma which really approaches a hand-axe. But, nodule. This is clearly shown by Nos. 52 and 53; in
although the flaking extends on both sides of the edge No. 52 the striking platform forms a right angle with
for a short distance near the point, it is not a true biface. the nodule's long axis, whereas in No. 53 these planes
In fact the point seems to be an original feature of the are parallel to each other. Although a series of long,
pebble rather than a result of purposeful working, as more or less straight-sided blades have been struck from
signs of use are apparent only along the flaked sides. No. 52, no implements of this type were found.
(i) Chopping-Tool or Core (?) (Fig. 67, No. 43). Neither do the T3 gravels contain small blades of the
South of Sadaing, near Yenangyaung, two pebble im- type that have been detached from No. 55 at any of the
plements of the type illustrated by No. 43 were found. localities investigated by the expedition, in fact imple-
They seem to be cores with roughly prepared striking ments less than 6 cm. in the longest dimension were
platforms which also served as chopping tools, since found to be altogether lacking at the Lower Palaeolithic
the edges, which form almost a right angle, appear to sites in the Irrawaddy Valley. No. 53 is an excellent
have been used. It is evident that flakes detached from example of an Early Anyathian core from which three
a nucleus of this sort would have facetted butts; how- broad flakes have been struck. Specimens with only a
ever, none of the flakes collected in the T3 gravels in single flake scar, such as No. 54, are fairly common, but
this vicinity display this feature. Thus it seems very no parent cores or flakes with high-angle striking plat-
probable that No. 43 may be considered a chopping-tool forms, Nos. 47 and 50, were found. On the whole
of the type with a slightly "overthrust" edge, made on a there is nothing particularly distinctive about the Early
well-rolled pebble. Each of the flake scars on this Anyathian cores. The simple crude flakes, character-
specimen displays a well-defined negative bulb of per- istic of the industry, seem to have been struck by direct
cussion, and, as the implement is made of an especially percussion from large rounded to subangular pebbles
homogeneous and fine-grained fragment, the facets are with rough striking platforms. A few types reveal that
also clear. a fairly advanced blade technique was likewise em-
(j) Flake Implements (Fig. 68, Nos. 46-51). Com- ployed, but no blade implements or small flakes were
pared to the core tools, flakes are extremely rare in the found in Upper Burma.
Early Anyathian 3 silicified tuff series; only 33, or 15 The most significant feature of the Anyathian culture
per cent of the total collection, consist of flakes. Large is the complete absence of hand-axes. This cannot be
trimming flakes displaying cortex on the upper surface, explained by the scarcity of good raw material, since the
and with plain unfacetted striking platforms, such as silicified tuff occurs in large boulders. Thus the only
Nos. 46 and 49, are the prevailing type. These reveal logical conclusion seems to be that the people responsi-
368 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

46 47

49

Scale
0 5 10 cm.
TvI

48

:
!I

.\ II4

50 51
FIG. 68. Early Anyathian 3 Implements of Silicified Tuff. Nos. 46, 47 and 49: Large Flake Implements;
No. 48: Double Side-and-End Scraper Made on a Flake; Nos. 50 and 51: Small Flake Implements. All from
Yenangyaung.
MOVIUS: THE STONE AGE OF BURMA 369

52

53
Scale
0 5 10 cm.
I
I tr

55
54
FIG. 69. Early Anyathian 3 Implements of Silicified Tuff. Nos. 52-55: Various Types of Cores or Nuclei.
Nos. 52-54: Yenangyaung; No. 55: Magwe.
370 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
ble for the Burmese Palaeolithic were not users of hand- In a general consideration of the Early Anyathian 3
axes. Instead chopping-tools and tools of the adze or implements of silicified tuff, it may be noted that chop-
chopper variety comprise the bulk of the stone imple- ping-tools with alternately flaked edges occur with ap-
ment equipment of the ancient settlers in the Irrawaddy proximately the same frequency at Chauk as they do at
Valley. On the basis of the fossil wood series alone Yenangyaung (see Table 4), although the total series
it might be argued that hand-axes are absent because from Chauk is very much smaller. At the latter site
this material does not lend itself to their manufacture. flakes and cores are very rare, but with this exception
However, such an argument cannot be applied to the the typology agrees very well with Yenangyaung. Table
silicified tuff for the reason given above (see p. 367). 4 shows the relative frequency of the various forms of
As previously stated the influence of raw material is silicified tuff implements; a comparison of Tables 3 (see
considerable, and, although the two series of Early p. 360) and 4 reveals that, whereas the typology of the
Anyathian implements exhibit certain differences from two series overlaps to a certain extent, the silicified tuff
a typological point of view, there can be no question that range is far greater. In the case of the fossil wood the
each is a contemporaryindustrial expression of the same apparent conservatism as regards the typological range
culture. This is made evident by the fact that hand- of the material is, as previously stated, due to the nature
adzes, the type fossil wood implement, occur also in of the fossil wood itself. The tendency in each phase
silicified tuff, whereas the chopping-tool with alternately of the Early Anyathian to develop very similar fossil
flaked edge, the type silicified tuff implement, is like- wood implements has already been commented on. The
wise found made of fossil wood. Furthermore each same observation applies to the silicified tuff series, but
series is characterized by essentially the same types of here one can definitely assert that Early Anyathian 1
tools which persist with very little typological change chopping-tools with alternately flaked edges are very
throughout. This is perhaps to be expected, since on much cruder and less evolved than are those associated
the basis of the primitive forms displayed by most of with Phase 3. Implements of this type are unknown
the Early Anyathian tools, it seems unlikely that much in Phase 2, since the only silicified tuff implement from
change would occur. Such evolved implements as hand- Nyaungu is a small single-ended chopper of the type
axes can and do cover a wide typological range, but in described in category (1) of the fossil wood series (see
dealing with choppers and adzes of various categories p. 353). The small size of the collection and the limited
one can expect only a certain technological refinement area which it represents makes further comment dif-
so long as the same fundamental types remained in com- cult. It appears that such variables as raw mlaterialand
mon use. The present collection from Burma is far too local specialization have played an extremely important
limited to reveal any very marked developmental se- role in the development of the Lower Palaeolithic of
quence, but local differences, as previously suggested, are Burma, and more work must be done in the field before
apparent. the answer to these problems will be clear.

TABLE 4
THE RELATIVE FREQUENCY OF THE VARIOUS TYPES OF SILICIFIED TUFF IMPLEMENTS FOUND AT THE THREE MAIN EARLY
ANYATHIAN SITES. The percentages have been computed in the same manner as those shown in Table 3 (see p. 360).

SILICIFIEDTUFF IMPLEMENTS YENANGYAUNG CHAUK NYAUNGU TOTALS

Choppers Flaked only on the Upper Surface: (1) Massive 16 (8.37%) 1 (3.57%) 17 (7.73%)
(2) Large 15 (7.85%) 6 (21.43%) 21 (9.55%)
(3) Medium 17 (8.90%) 4 (14.29%) (100.00%) 22 (10.00%)
Massive Triangular Chopper 1 (3.57%) 1 (0.45%)
Massive Pick-like Implement 1 (0.53%) 1 (0.45%)
Hand-Adze or Chopper Made on a Pebble: (1) Large 6 (3.14%) 6 (2.73%)
(2) Medium 7 (3.66%) 7 (3.18%)
Tabular Hand-Adze (Fossil Wood Type) 10 (5.24%) 5 (17.86%) 15 (6.82%)
Chopping-Tool with Alternately Flaked Edge 14 (7.33%) 3 (10.71%) 17 (7.73%)
Chopping-Tool with "Overthrust" Edge 10 (5.24%) 1 (3.57%) 11 (5.00%)
Chopping-Tool Made on a Flat Pebble 10 (5.24%) 3 (10.71%) 13 (5.91%)
Chopping-Tool or Core (?) 2 (1.05%) 2 (0.91%)
Flake Implements: (1) Massive 11 (5.76%) 11 (5.00%)
(2) Large 12 (6.28%) 1 (3.57%) 13 (5.91%)
(3) Medium 9 (4.71%) 9 (4.09%)
Cores or Nuclei: (1) Large-with prepared striking platform 6 (3.14%) 6 (2.73%)
(2) Large-with plain striking platform 15 (7.85%) 15 (6.8Z%)
(3) Medium-with plain striking platform 19 (9.95%) 2 (7.14%) 21 (9.55%)
(4) Small-with plain striking platform 11 (5.76%) 1 (3.57%) 12 (5.45%)
28 1
TOTALS:
TOTALS: 191 28 1 220
MOVIUS: THE STONE AGE OF BURMA 371

56

Scale
5I
0 - s~~~~~~~~~~
? ? V
10 cm.

58
57

59
60
FIG. 70. Late Anyathian 1 Implements of Silicified Tuff. No. 56: Small Bifacial Chopper; No. 57:
Hand-adze or Chopper Made on a Flake; Nos. 58 and 60: Cores or Nuclei; No. 59: Small Flake Implement.
Nos. 56, 59 and 60: Yenangyaung; Nos. 57 and 58: Chauk.
372 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
V. THE LATEANYATHIANCULTURE hand-adzes of fossil wood were found, but two large
silicified tuff flakes, retouched along the end, Fig. 70,
As compared with other areas of the Old World there No. 57, seem to have been used as choppers or adzes.
is very little typological difference in the Palaeolithic of Thus the Late Anyathian 1 implements display a more
Burma, between the "early" and "late" cultures. In advanced flaking technique in addition to being smaller
many respects the situation is analogous to that which than the Early Anyathian 3 material. Possibly this may
exists in northwestern India when the Early and Late be correlated with the fact that at this time only very
Soan are considered, but the Late Soan development ap-
fine-grained siliceous rocks, conducive to controlled flak-
parently begins earlier than does the Late Anyathian of ing, were used.
Burma (see Fig. 54, p. 334). Whereas in India Late
Soan A is characterized by the development of a Leval-
2. Late Anyathian 2
loisian technique, which becomes even more marked in
Late Soan B, in Burma no such definite change occurs. Actually the typology of Late Anyathian 2 is very
Although the Late Anyathian is typologically more ad- limited, a fact which is doubtless to be correlated with
vanced than the Early Anyathian, it contains essentially the raw material employed for the manufacture of most
the same basic core implements that have already been of the implements. With only three exceptions they
considered, with the exception of chopping-tools with are of fossil wood, so that from a typological point of
alternately flaked edges. In fact no true blade industry view the same general observations may be made as in
appears in Burma until post-Palaeolithic times. But in the case of the Early Anyathian 2-3 series of this ma-
the Tg gravels there is a small series of slightly rolled terial from the T. gravels. Including the heavily weath-
material, apparently dating from the erosion period when ered series from the surface of the T:; sites at Chauk,
T:, was cut, which reveals a more advanced flaking tech- Pagan and Nyaungu, a total of approximately 150 Late
nique than that characteristic of the Early Anyathian. Anyathian 2 implements was collected. Of these only
This material has been assigned to Late Anyathian 1, 73 were found in association with T4 deposits at Magwe.
and Phase 2 of this culture is well documented by imple- Pauk, and between Singu and Pagan (see Section 5 and
ments from the late Upper Pleistocene gravels of T4. Fig. 23, p. 299 of Dr. de Terra's report). As stated
It is not altogether clear if the Late Anyathian repre- above, a few specimens were found at Pagan and
sents an indigenous development in response to a new
Nyaungu which may have been derived from the Pagan
stimulus or if it is due to the introduction of new ele- Loessic Silt. These are comparatively fresh, and al-
ments from outside. Before this point can be deter-
though they are coated with a thin film or crust of the
mined it will be necessary to study a larger collection of
silt, they were not found in place in the deposit. In
material than is now available, and covering a wider
addition, both in T4 and on the surface of T. a few rolled
geographical area. Early Anyathian implements were found which dis-
played fresh flaking, the result of reworking during Late
1. Late Anyathian 1
Anyathian or Neolithic times. As in the case of Phase
The tools considered here all display a very much 1, the Late Anyathian 2 series is not as heavily patinated
lighter patination-light shades of gray-blue, tan and as the older material. Nevertheless it is easily dis-
brown-than the Early Anyathian series from the same tinguished from the Neolithic, which is sharp and fresh,
horizon. Furthermore they are only very slightly rolled in spite of the fact that many of the types are identical.
as compared with the earlier material. Of the total of (a) Hand-Adze zith Scalloped Edge (Fig. 71, Nos.
16 Early Anyathian 1 implements, 14 are of silicified 61 and 62). These are small, somewhat foreshortened
tuff; the 2 fossil wood specimens include a core and a hand-adzes fashioned by the removal of three primary
flake, types unknown in the Early Anyathian collections flakes parallel to each other and struck from the lower
of fossil wood. Since both cores and flakes of fossil surface of the implement. Since each of these flakes
wood do occur in the Late Anyathian 2 of T,, it was has been detached in such a way as to leave a concave
decided that the slightly rolled and lightly patinated ma- scar, the sharpened edge thus produced has a scalloped
terial from T3 was in certain respects more closely appearance; the ridges which separate the flake scars
related to the later development than to the Early form small points or projections. No. 62 is made of
Anyathian. In addition there are other distinguishing a siliceous fragment of fossil palm in which the vascular
features, such as the small oval bifacial chopper, Fig. bundles are clearly apparent.
70, No. 45, flaked on both surfaces, and the 3-sided core, (b) SmzallSteep-ended Adze (Fig. 71, Nos. 63 and
Fig. 70, No. 58. Of the latter type there are several 66). Tools of this type may have served also as planes
examples; the fossil wood core referred to above is of or steep scrapers. They are made on short thick frag-
the same form as the small core of silicified tuff, Fig. ments of fossil wood debris worked in characteristic
70, No. 60. Several small to medium-sized flakes, fashion across one end. In Late Anyathian 2, imple-
similar to Fig. 70, No. 59, which has been retouched on ments of this type are not common; the flaking is defi-
the lower surface, are likewise characteristic. No true nitely better than that found on the earlier material.
MOVIUS: THE STONE AGE OF BURMA 373

"WI

-1

61 62 63

65 66

64 0I
Scale
sI 10 cm.
4

68 69 70
67
6?'
FIG. 71. Late Anyathian 2 Implements of Fossil Wood. Nos. 61 and 62: Hand-Adzes with Scalloped
Edge; Nos. 63 and 66: Small Steep-ended Hand-Adzes; Nos. 64, 65, 69 and 70: End-Scrapers or Small Hand-
Adzes; Nos. 67 and 68: Double-ended Scrapers. Nos. 61 and 62: Chauk; Nos. 63-66, 68, 69 and 70: between
Singu and Pagan; No. 67: Nyaungu.
374 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

(c) End Scraperlor SmlallHand-Adze (Fig. 71, Nos. been retouched. All three implelnents display edge
64, 65, 69, and 70; Fig. 72, No. 75). Although smaller chipping resulting from scraping and cutting.
and more carefully flaked, the tabular artifacts in this (i) Core (Fig. 72, No. 80). This is the only arti-
category are typologically identical with the larger forms fact of silicified tuff found in the T4 gravels. It is
found in the Early Anyathian. It may be noted that patinated a dull pink and is rolled to the same degree as
square and round-ended types prevail; pointed speci- the fossil wood specimens from this horizon. Typo-
mens, such as No. 70, are rare. On the basis of size, logically this core, a small pebble from which a tria-
the series considered here are probably to be regarded as angular point has been struck, is very similar to Fig.
scrapers, in spite of the fact that they are core imple- 70, No. 60, a Late Anyathian 1 implement found in the
ments. Strictly speaking they belong to the hand-adze T3 horizon.
family of tools, but they probably served a variety of The most significant fact about the typology of the
purposes. No. 75 is interesting in that it is one of the Upper Palaeolithic of Burma is its essential similarity
few examples found of an artifact made on a segment of to the Lower Palaeolithic when the fossil wood imple-
the trunk or branch of a small fossilized tree with an ments of the two cultures are considered. In both, the
oval section. hand-adze class of tool predominates, although in the
(d) Double-ended Scraper (Fig. 71, Nos. 67 and case of the Late Anyathian, implements of this type are
68). Completely absent in Late Anyathian 2 are double- considerably reduced in size and display a marked re-
ended implements of the hand-adze or scraper variety finement of technique. It is, however, difficult to decide
with inversely sharpened ends. As in the case of Nos. whether this is entirely due to a survival of tradition or
67 and 68, the flakes have been removed only from the whether the inherent nature of the material itself has
lower surface to produce these composite tools. Speci- been the dominating influence. Various types of scrap-
mens of this type are not as common as in the Early ers, implements with scalloped edges, as well as flakes
Anyathian. of fossil wood appear in Late Anyathian 2 for the first
(e) Side-and-End Scraper (Fig. 72, Nos. 71-73). time. Furthermore choppers with alternately flaked
Nos. 71 and 73 are worked on one end and along the edges, concave hand-adzes and inverse double-ended
adjacent side in such a way as to form a semi-oval cut- hand-adzes of the Early Anyathian are absent. Thus
ting edge. This type is the nearest approach to a true Late Anyathian 2 is in some respects distinctive, but it
discoidal scraper found in the Burmese Upper Palaeo- is practically impossible to compare it with Late Any-
lithic. Although No. 72 has a scalloped edge and is not athian 1 which is composed almost wholly of silicified
as finely worked, it belongs in this category; it was found tuff artifacts. This difficulty is enhanced by the limited
at Pagan on T3 lying near deposits of Pagan Silt. size of the collections. As to the Late Anyathian 1
Since it was heavily encrusted with silt when found, it series, certain forms, especially the small, oval, semi-
seems very probable that it was derived from this de- bifacial chopper, Fig. 70, No. 56, in addition to 3-sided
posit. If so it is presumably contemporary with the cores, Fig. 70, No. 58, represent types unknown in the
Late Anyathian 2 material from T4. Early Anyathian. Thus the possibility that new in-
(f) Steep Pyramidal Core-Scraper (Fig. 72, No. 75). fluences penetrated Burma at this time cannot be over-
The illustrated specimen is the only implement of this looked, although this point cannot be definitely estab-
type found. It is made of very fine-grained fossil wood, lished. Regarding the continuity of type between the
and the flake scars, from which small blades were ap- Early and Late Anyathian fossil wood implements, a
parently struck, converge at the top to form a rough direct parallel may be drawn with northwestern India.
cone or pyramidal-shaped core, subsequently used as a According to Paterson (with de Terra, 1939, p. 308),
scraper. The patination, a rich dark brown color, is the pebble tools of the Early Soan undergo a certain
unusual in the Late Anyathian. degree of refinement in the Late Soan phases, but. as
(g) Side Scraper (Fig. 72, No. 76). Side scrapers Paterson points out, the influence of the material is
are very uncommon in both the Lower and the Upper considerable. Until more is known about the Palaeo-
Palaeolithic of the Irrawaddy Valley. In the T4 gravels lithic of Eastern and Central Asia in general, this ques-
between Singu and Pagan a few crude examples, of the tion of continuity must be left open.
type illustrated by No. 76, were found. This specimen
is made on a rough flake pointed at one end. VI. SUMMARY AND CONCLUSIONS
(h) Flake Implements (Fig. 72, Nos. 77-79). These
are all made of fine-grained fossil wood. The striking That the primitive nature of the Palaeolithic chopping-
platforms are unfacetted, the bulbs clearly defined, and tool-hand-adze complex of Burma is not due entirely to
in the case of Nos. 78 and 79 the core was carefully the limitations imposed by the nature of the raw material
prepared prior to detachment. No. 79, which is made is conclusively demonstrated by the fact that similar
of fossil palm, is extensively worked on the under sur- developments are found in other regions of southeast-
face of the left edge. The upper surface of the bulbar ern Asia, northern India and China. On the basis of
end of No. 77 has been chipped in an effort to reduce the pebble chopping-tools found in the Soan and Indus
the thickness, whereas the projection on the end has Valleys of northwestern India, Dr. de Terra and Pat-
MOVIUS: THE STONE AGE OF BURMA 375

71 72 73

Scale
0 - I s I 10 cm.

76

74 75
77

78 79 80
FIG. 72. Late Anyathian 2 Implements of Fossil Wood. Nos. 71-73: Side-and-End Scrapers; No. 74:
Steep Pyramidal Core-Scraper; No. 75: End-Scraper; No. 76: Side-Scraper; Nos. 78 and 79: Flake Implements;
No. 80: Core or Nucleus. No. 71: Chauk; Nos. 72 and 77: Pagan; No. 75: Nyaungu; Nos. 73, 74 and 79:
Pauk, Yaw Valley; Nos. 76, 78 and 80: between Singu and Pagan.

erson (1939) have identified a new Palaeolithic cul- tools do not form part of the Soan development proper,
ture-the Soan, and influences from this source seem to which is fundamentally a chopping-tool culture. The
have penetrated southward into the Narbada Valley, as stratigraphic evidence proves that the Early Soan may
well as to other parts of peninsular India. At the same be assigned to the Middle Pleistocene of northern
time true hand-axes appear in the Punjab, but bifacial India, which in turn can be directly correlated with the
376 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Middle Pleistocene of Burma, according to de Terra site are a few crude hand-axes similar to those from
(see p. 334). At Choukoutien, near Peiping in north- Patjitan. On the basis of Mr. Collings' collection it
eastern China, the heavy stone industry, known from the seems very probable that in Java and Malaya these im-
excavations at Locality 1 (Sinanthropus deposits) and plements developed from pointed chopping-tools, which
Locality 15, is basically a chopping-tool-crude scraper approach hand-axes in that they are worked on both
culture in which sandstone, chert, flint, volcanic rocks, faces. This view is strengthened by the fact that the
as well as limestone, were used for raw material (com- Levalloisean technique, which accompanies the hand-
pare: Pei, 1931; Teilhard de Chardin and Pei, 1932; axe cultures in other sections of the Old World, is
Black, Teilhard de Chardin, Young and Pei, 1933; Pei, absent in the Far East. Whereas it appears in north-
1937, 1937-a, 1939). Associated with the choppers, ern India, it seems to indicate an intrusive influence,
which are of the same basic type as in the Soan and the probably also responsible for the Abbevilleo-Acheulean
Anyathian, is a small flake industry in quartz which in the Punjab, rather than an independent develop-
represents a special facies. As in India and Burma, ment.
the Lower Palaeolithic chopping-tool development be- The fact that hand-axes do not exist in Burma and
gins at Choukoutien in Middle Pleistocene times, as China6 cannot be explained as due to the absence of
indicated by the stratigraphic scheme proposed in this raw material suitable for their manufacture. It is true
report (compare Table 1, p. 331). Farther south near that, because of its friable nature, it is virtually impos-
Patjitan in Java, Dr. von Koenigswald has discovered a sible to miakean implement of this type from the fossil
rich Lower Palaeolithic site, which has been briefly wood of Burma. But, as already pointed out, there is
described (von Koenigswald, 1936; 1937, pp. 29-30; an abundant supply of silicified tuff in the Irrawaddy
1939, pp. 43-45). A large collection of approximately Valley, a rock with a conchoidal fracture which was ex-
2500 implements was studied by the writer in Bandoeng tensively used for making hand-axe-like tools in Java.
in 1938, and it was found to consist predominantly of Furthermlore at Choukoutien both sandstone and ig-
chopping-tools of Anyathian type, in addition to many neous rock were used for the manufacture of the large
cores and flakes. Hand-axes occur, but they are not hand cleavers from Locality 15 (Pei, 1939-a, pp. 173-
typical; in Dr. von Koenigswald's collection, they coln- 176). These implements are actually made on flakes.
prise only 6.2 per cent of the entire series. Further- They not only display a highly developed flaking tech-
more they seem to represent a special type of pointed
nique, but they also indicate that the material of which
bifacial chopper with longitudinal flaking, rather than
they are made would be eminently suitable for making
true Acheulean "coups-de-poing." The Levalloisean is
hand-axes, had the occupants of Locality 15 desired to
absent in Java, as in Burma and China. This flake make them. The samle argument applies to the Soan
culture is well developed in southern India (Paterson,
with de Terra, 1939, pp. 327-30; Cammiade and Bur- choppers and scrapers, most of which are Imade of
kitt, 1930; Krishnaswami, 1938, 1938-a), where it seems quartzite pebbles. Thus the Soan culture cannot be
to be intimately associated with the Madrasian (Acheu- explained solely as a result of the nature of the available
raw material, and it is now evident that it belongs to a
lean). In the Punjab it exerted a very strong influence
on the Late Soan development. Thus the Patjitan im- great Lower Palaeolithic province documented by the
Choukoutienian of China, the Anyathian of Burma, the
plements appear to be distinctive in comparison with the
classical Lower Palaeolithic sequence of the Old World. Tampanian of Malaya and the Patjitanian of Java.
Although direct proof is still lacking, the chopping-tool Although from widely separated areas, they may be re-
culture of Patjitan probably dates from Late Trinil garded as local cultures, which collectively form part
of the Lower Palaeolithic chopping-tool complex of
(Middle Pleistocene) times. In Malaya, H. D. Col-
southeastern Asia, northern India and China. The
lings (1939) has found implements at Kota Tampan,
near Lengong in the valley of the Perak River, which recognition of this comlplex as distinct from the classic
are very similar to the chopping-tools of Burma, Java, development found in other areas of the Old World has
China and northwestern India. However, the Pleisto- been the most significant contribution of our research
cene of Malaya has never been intensively studied, hence in Burma from an archaeological point of view.
the material from Kota Tampan cannot as yet be ac- Dr. de Terra has clearly shown in the form of a cor-
curately dated.5 Associated with the choppers at this relation table (Fig. 54, p. 334) that the earliest humian
industries in northwestern India, Burma and China are
5This comment likewise applies to the chopper-like imple-
ments found by J. H. Edgar in the Yangtze Valley between all found in deposits referable to the Middle Pleistocene
Ichang and Chungking and illustrated by Dr. D. C. Graham (compare Table 1, p. 331). The evidence may le sum-
(1935, Pls. I and II). Two are stated to have been found in marized as follows:
situ at the base of a deposit of loess 100 feet thick at Sharatang,
while several others were discovered in a cemented conglomerate 6 Although Dr. Pei (1934) has described one artifact from
containing river pebbles. This layer was apparently exposed by Choukoutien, Loc. 13, as a small hand-axe, the excellent cast of
erosion, but since its height above the river is not given, it is not this specimen in the Peabody Museum quite definitely does not
certain whether or not it is associated with one of the Yangtze represent a true coup-de-poing. The writer suggests that the
terraces. implement is a pointed bifacial chopping-tool.
MOVIUS: THE STONE AGE OF BURMA 377
1. Northwestern India.-The oldest archaeological India, and that they had a characteristic culture which
material is represented in this region by the so-called was identical, or nearly so, throughout this vast area."
"Pre-Soan," an industry of which large crude flakes are Perhaps this may be the explanation of the negative
typical. These are found in situ in the late Upper Si- results of previous attempts to fit the Far Eastern Stone
walik Boulder Conglomerate (early Middle Pleisto- Age development into that established for Western
cene). The true Early Soan development dates from Europe.
the deposition of T1, which has been assigned to late It is interesting to note that the Lower Palaeolithic
Middle Pleistocene times. first appears in southeastern Asia during a period
2. Burma.-Early Anyathian 1 (see p. 347) is found (Middle Pleistocene) which witnessed the dispersal of
in TI and the Lateritic Gravel of the Irrawaddy Basin, what von Koenigswald (1937, p. 28) has called the
both of which are of Middle Pleistocene age. The T, "Sino-Malayan" fauna. As Dr. Colbert points out (see
gravels appear to be erosion remnants of the Lateritic p. 427), this fauna seems to have spread from Southern
Gravel, a deposit corresponding in age to the Uru China. It is unknown in northern India. Here the
Boulder Conglomerate of northern Burma. Soan culture of the Punjab is probably to be regarded
3. China.-The implementiferous deposits at Chou- as intrusive from those regions of Central Asia to the
koutien-Localities 13, 1 and 15-all date from the north and east of India, since contemporary with its de-
Middle Pleistocene on the basis of the stratigraphic velopment a typical Lower Palaeolithic (Abbevilleo-
scheme advocated throughout this report. Locality 1, Acheulean) hand-axe-Levallois flake complex is also
where the Sinanthropuis deposits were discovered, is present. It is a regrettable fact that the only fossils
younger than Locality 13 and older than Locality 15 as found in association with the Pleistocene terraces of the
demonstrated by the fauna (Pei, 1939-b, p. 7). Irrawaddy Basin-Elephas hysudricus and Hemzibostri-
In Malaya the dating evidence is still unclear, as quetricornis (see p. 424 of Dr. Colbert's report)-may
stated above, but if the gravels of the Perak Valley are both be considered survivals of the so-called "Siva-
the age of the tin gravels observed by Dr. de Terra Malayan" group of the Lower Pleistocene (Villafranch-
between Port Swettenham and Kuala Lumpur, in Se- ian). Nevertheless an abundant and rich Middle Pleis-
langor (see p. 332 and Fig. 53), it seems likely that Dr. tocene fauna is represented by the collection from
Collings' Tampanian may also be assigned to the Middle Mogok in the Northern Shan States of Burma, which
Pleistocene. The situation in Java is less obscure. has been described by Dr. Colbert. According to de
Whereas it is difficult to fit the Cenozoic sequence of Terra (see p. 329), the deposits in which they occur are
this latter region into that established for the mainland contemporary with the Irrawaddy terraces. In Java
of southern and eastern Asia, the evidence of the fauna the "Sino-Malayan" group includes many of the type
indicates that the Trinil Zone belongs to the same Middle fossils of the Trinil Zone. Thus in Burma and Java
Pleistocene horizon under discussion, according to Dr. the development of a Lower Palaeolithic of "eastern"
Colbert (see p. 426). Now Pithecanthropus erectus, affinities begins shortly after the introduction of a new
originally discovered by Professor Dubois and of which "eastern" faunal assemblage.
additional examples have recently been found by Dr. That we are dealing with a separate and distinctive
von Koenigswald, dates from the early portion of the Old Stone Age complex in southeastern Asia, northern
Trinil Zone. Furthermore, Dr. de Terra has demon- India and China is emphasized not only by the presence
strated that the Patjitanian chopping-tool development of various primitive types of choppers, but, as has been
probablybegan at least as early as Late Trinil times (see stated above, by the absence of those flake and core tools
Part V, p. 458). Thus it appears probable that the which characterize the Lower Palaeolithic horizons in
Patjitanian is the culture of Pithecanthropus, or at least other sections of the Old World. These yield both true
of his direct descendants. However, the evidence from hand-axes ("coups-de-poing" or "bifaces") and Leval-
China, where Sinanithroplus was found at Choukoutien lois flake tools. Except on the extreme northwestern
in direct association with chopping-tools and other im- periphery of their distribution-i.e., France and Eng-
plements, is far more convincing. And the Choukoutien land-these two types of implements, sometimes con-
deposits have been shown to be comparable in age with sidered to be mutually distinct industries, cannot be
the Trinil Zone in Java. Thus it may well be that one segregated as separate and independent developments.
of the most vital reasons why the Lower Palaeolithic of Indeed the distribution of those which are in the so-
southeastern Asia, northern India and China differs called Levallois tradition closely follows that of true
from that of the rest of the Old World lies in the fact hand-axe complexes of developed Acheulean type. This
that we are also dealing with men of a different evolu- distribution extends as a huge triangle from Western
tionary status from that found elsewhere. This point Europe to southern India to the Cape of Good Hope,
has recently been stressed by Professor Hooton (1940, embracing the Mediterranean Basin and the Near East
p. 73), who states that in his opinion "it seems probable as well as the entire continent of Africa. In certain
that men at the Pithecanthropus-Sinanthropus stage of parts of this area, especially in South and East Africa
physical evolution extended down the eastern half of and peninsular India, chopping-tools occur. These im-
Asia from Peking to Java and westward to Northern plements seem to be fairly typical of the pre-Stellenbosch
378 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
of the Vaal Valley (van Riet Lowe, 1937, pp. 75-76), in time, the culture may be basically a survival of the
the Oldowan of Tanganyika (Leakey, 1934, pp. 144- old food-gathering economy, which was in receipt of
146), the Kafuan and Oldowan of Uganda (Wayland, higher influences from true Neolithic sources. This
1934; O'Brien, 1939, pp. 69-94), as well as the Lower latter alternative seems the more likely on the basis of
Palaeolithic of the Narbada Valley of central India (de present evidence.
Terra, with Paterson, 1939, pp. 321-326) and elsewhere The type Neolithic station is immediately east of the
in the peninsula, according to information received Kyaukpadaung-Popaywa road at milestone 39.2 (see p.
from Mr. E. C. Worman, Jr. It therefore appears 311 of Dr. de Terra's report). Both here and in the
probablethat primitive tools of the chopper variety were vicinity of the Kyaukpadaung railroad station (see p.
more or less widespread at an early date, but that with 311, as well as Fig. 35 of Dr. de Terra's report), there
the gradual evolution of pointed, bifacial, core tools of is an abundant supply of a very fine-grained, white or
true hand-axe form, choppers became of secondary im- pink silicified tuff, which has a flint-like appearance and
portance. Nevertheless in southeastern Asia and China which possesses a smooth conchoidal fracture. These
this ancient tradition persisted and continued to develop outcrops were an important source of supply for raw
uninfluenced by the contemporary innovations found material for the manufacture of implements during Neo-
elsewhere. In northern India, however, where the lithic times. Although this type of silicified tuff does
same complex occurs, it was strongly modified as a result not occur in a native state at any of the localities in-
of contact with influences from other sources. The vestigated along the Irrawaddy, between Magwe on the
presence of pointed bifacial choppers or crude hand- south and Nyaungu on the north, artifacts made from it
axes in Malaya and Java demonstrates, however, that are found here in profusion. Especially prolific are the
even in the Far East an evolution comparable with that sites near Magwe. Here implements and charcoal were
which led to the Acheulean of other regions was under found at the base of a deposit of loose red sand, which
way. Thus the archaeological material discovered dur- may attain a thickness of 4-5 feet. According to de
ing the last ten years in Asia has proved the obvious Terra (see p. 310), this sand, because of its structure
fallacy of attempting to classify the Palaeolithic of the and grain size, is of eolian origin, and it has been called
Old World on the basis of a single taxonomic scheme. the Magwe Sand. Not only is it found on the surface
The recognition of the significance of the Anyathian of of the plateau, but in places it overlies the Nyaungu Red
Burma has thrown new light on such questions as the Earth. In addition to Magwe, Neolithic material was
affinities of the Soan and Choukoutienian, hitherto re- found at Minbu, Yenangyaung, Chauk, Pagan and
garded as unrelated. It is now clear that the classic Nyaungu; in the vicinity of Kabani, east of Nyaungu,
Western European sequence is absent in southeastern several typical implements were collected at the base of
Asia, and that already during Lower Palaeolithic times small gullies carved into the Nyaungu Red Earth (P1.
we have to deal with independent complexes of cultures, X, Fig. 1; see pp. 298 and 311 of Dr. de Terra's re-
which seem to have followed very different patterns of port). These had apparently been washed down from
growth. the surface of the deposit. At the Kyaukpadaung site
(Ky. 1), which is at the base of Mt. Popa (Fig. 20 and
B. THE NEOLITHIC PERIOD PI. I, Fig. 1) near the village of Sebauk in the valley of
a small stream, the Sai Chaung, a tributary of the
I. INTRODUCTION Taungzin Chaung, the archaeological horizon is at the
base of a soil similar to the Magwe Sand and of similar
The archaeological material discussed in this section
age (PI. XVI, Figs. 1 and 2; Fig. 39, p. 311). It is
has been the subject of considerable controversy, as 2-7 feet thick and composed of loose brown silt con-
stated on p. 341. Recently Mr. T. 0. Morris (1935, taining volcanic ash. Below is a second layer of brown
pp. 5-7) has assigned it to the Upper Palaeolithic Pe- ash from Mt. Popa overlying a bed of light gray vol-
riod on the basis of his typological study of a series of canic material containing silicified tuff-one source of
implements from 11 surface sites in the Thayetmyo and the raw material used by Neolithic man. These deposits
Magwe Districts. But data secured by the American are clearly exposed along the eastern side of the Sai
Southeast Asiatic Expedition demonstrate a post- Chaung Valley. On the opposite side of the valley
Pleistocene antiquity for the material in question as there is a deposit of brown sandy silt containing bands
originally suggested by Swinhoe (1903, p. 256), in spite of black clay, where we collected bones and teeth of
of the fact that the typology of certain of the tools is Horse (? Ass), Deer, Antelope, and Bos, possibly of
in some respects reminiscent of the Upper Palaeolithic. banting type (see p. 402 of Dr. Colbert's report).
The stratigraphic position of the implements is con- Although most of the implements are made of silicified
firmed by the fact that at several localities pottery and tuff, other rocks were used during post-Pleistocene times
polished stone tools were found, suggestive of Neolithic for the manufacture of tools. Artifacts of fossil wood
affinities. Whether or not a true Neolithic food- were found at all the sites, and many of the types rep-
producing culture is represented, however, could not be resented are identical with Late Anyathian forms. As
established. Thus it is possible that, while Neolithic explained on p. 360, this is due to the structure of the
MOVIUS: THE STONE AGE OF BURMA 379

82

83

C- --4-
bypScale a0 1 2 3 4 scm.
81 84
L-8 '--

85 8687 88 89

---am~~~~~~~~~9
~i 94-
92~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~":?

90 91 92 93 94

FIG. 73. Neolithic Implements from Locality Ky. 1, near Kyaukpadaung in the Vicinity of Mt. Popa.
380 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
wood. In addition large hand-adzes reminiscent of the choppers, as well as one pick, roughly flaked and with a
Early Anyathian occur, not only at the actual Neolithic triangular section (Fig. 73, No. 81). Fossil wood im-
stations, but also on the surface of the Pleistocene ter- plements (Fig. 73, No. 83) also occur. \Whether or
races of the Irrawaddy, especially T::. These fresh not this type should be classified as a small hand-adze
unpatinated hand-adzes of fossil wood from the T3 level or a large scraper is immaterial, since the important fact
are probably Neolithic in age, but as they cannot be to be observed is that it is very similar to Anyathian
dated, no discussion of them, beyond the fact that they forms (compare: Fig. 59, Nos. 10 and 12). That it
do occur, is included here. Typologically they display appears in a Neolithic context does not seem to demon-
no variation from the commlonforms of the Early Any- strate survival, but rather what our own experiments
athian. A few small implements of vein quartz, of have shown, i.e., that this type of fossil wood can only
similar type to those of silicified tuff, are commonly be flaked successfully in one direction-at right angles
found. The polished stone tools are mainly of meta- to the tree growth-rings and across the grain of the
morphic rocks-fine-grained schist and slate-as well wood. Small flakes and blades showing use (Fig. 73,
as basalt, which have been heavily weathered. As to Nos. 85-87) are typical; the flake scars on the upper
the patination of the series from Kyaukpadaung, white surface indicate careful core preparation prior to de-
and light shades of gray and pink prevail, but a few tachment. Scrapers comprise by far the largest category
specimens display a deep brick-red color. At the sites of implements from this site, and for the most part
along the river where the artifacts are associated with they are small. Nevertheless a few large forms, such
the Magwe Sand, a dull chalky white patina is char- as Fig. 73, No. 82, which is extensively worked on one
acteristic, and many of the implements have been sub- side and across one end, are present. The prevalent
sequently stained a dull red color by the ferruginous types are included in Figs. 73 and 74. No. 84 is a
material in the deposit. As the result of sand particles small steep scraper; Nos. 88-92 are various small end-
carried by the wind and blasted against the worked scrapers made on flakes, the bulbar portion of which is
stones lying on the surface, silica polish is displayed by usually broken off; Nos. 93 and 94 are roughly dis-
many of the specimens from the Neolithic sites. This coidal; Nos. 95-98 are side scrapers on which the work-
condition is also met with in the case of many of the ing is frequently executed along more than one edge, a
potsherds and polished stone axes, which have been ex- common feature as shown by the illustrated specimens;
posed, along with the silicified tuff, vein quartz and and Nos. 99-101 are concave scrapers probably used for
fossil wood implements, by erosion and drifting sand. fashioning wooden or bone shafts. Borers or perfora-
Although the Singu Silt of T,, the lowest terrace in the tors are not common, but two quite definite forms are
Irrawaddy sequence, contained no artifacts, it is prob- represented; a more or less triangular-shapedtype (Fig.
ably in part at least of the same age as the Magwe Sand. 74, No. 102), and one fashioned on the projecting
corner of a small rectangular blade (Fig. 74, No. 103).
II. THE ARCHAEOLOGICAL MATERIAL The cores, of which Fig. 74, No. 104, is a typical ex-
ample, all have plain striking platforms, and, although
With the exception of the Kyaukpadaung locality, the there is an abundant supply of raw-materials, they are
sites are all situated on the surface of the peneplain be- not large. To judge by the small number of chips and
tween Magwe and Nyaungu along the Irrawaddy River. waste flakes found at this locality, it is probably to be
In all nine stations, including Noetling's original locality regarded as an occupation site rather than as a workshop.
near Beme, 11/2miles east of the town of Yenangyaung, In association with the material described above, a
were investigated. As previously stated, the most im- small chipped and polished stone celt, Fig. 74, No. 105,
portant one for establishing the range of the implement was found. It is made of a fine-grained, meta-
types is at Kyaukpadaung. The evidence obtained at morphosed, light green stone, of the slate variety, and
this site will be considered first. the length of the specimen is 4.2 cm. The flake facets,
which have not been completely removed by grinding,
1. Kyaukpadaung are apparent along the sides. This implement is the
only one of its type actually found at the Kyaukpadaung
The stratification at Kyaukpadaung is shown in Fig.
locality, although ground and polished celts are known
39 (p. 311), and the section is fully described by Dr. to occur elsewhere in this region (Swinhoe, 1903, p.
de Terra. It was determined by careful investigation 256). Potsherds, of which typical examples are shown,
that the main archaeological horizon consists of an Fig. 74, Nos. 106-108, were also collected at Kyauk-
old land surface which lies at the base of the topsoil padaung. Many examples were discovered in situ in
(P1. XVI, Fig. 2). This latter deposit also yielded the alluvium (Layer 6 in Fig. 39, p. 311) which covers
a few implements, as did the alluvium covering the the floor of the valley, and they are therefore pre-
floor of the Sai Chaung Valley (P1. XVI, Fig. 1). sumably of alout the same age as the stone tools, since
Although most of the tools are small-consisting of a few scrapers and small blade implements were also
scrapers, points and blades-a few heavy core imple- found in the same deposit. The ware is of a coarse
ments were found. These include several very crude texture, comprises numerous grits and is a light brown
MOVIUS: THE STONE AGE OF BURMA 381

96
95
97

98 99 100

101 102
103
Scale
o 1 2'
3I 4 5 cm.
I ii =....
........... I

104

105

....~ -.~.-:- ;~ t --C1?~~~~~~~~~~~7-


7-'"
;.
I . .
?-~cu-r\* -- '"-s ;.E-r?
.
107-

106

107 108
lOG:

FIG. 74. Neolithic Implements from Locality Ky. 1, near Kyaukpadaung in the Vicinity of Mt. Popa.
382 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
or reddish color. It was identified by Messrs. Collings No. 116) than the cord-marked ware described above
and Tweedie of the Raffles Museum, Singapore, as from the alluvium of the Sai Chaung Valley. Perhaps
"matt- or cord-marked ware," but it is hitherto unre- this indicates that a late survival of the Stone Age in
corded in Burma. Admittedly its association with the Burma is to be expected; however, none of these sherds
material from the old land surface at the base of the top- can be dated with certainty.
soil is somewhat vague. However, the polished stone At Magwe: Loc. 3, which is opposite the Early Any-
celt found at the site suggests that the industry cannot athian 1 site discussed on p. 347, 100 yards east of the
be placed earlier than the Neolithic Period, although it Magwe-Yenangyaung road between milestones 9.2 and
possibly represents the survival of an older tradition. 9.3 (Fig. 32, p. 307), and at Magwe: Loc. 5, near mile-
Since Kyaukpadaung is not the only locality where stone 11.3 on the same road and 400 yards east of it, a
pottery was found together with this industry, it seems small series of typical implements was found (see pp.
probable that the latter appeared in Burma at a time 306 and 307 of Dr. de Terra's report). Both these lo-
chronologically not far removed from the introduction calities afford further proof that the culture layer is
of polished stone tools. But this suggestion must be situated at the very base of the Magwe Sand. The
regarded in a purely tentative light until more is known artifacts of silicified tuff from these sites include typical
about the Neolithic Period as a whole in southeastern scrapers, points, flakes and small blades, which in some
Asia. instances are made of vein quartz. Loc. 5 yielded one
small, celtiform core implement (Fig. 75, No. 120),
2. Magwe apparently used as a side-and-end scraper. Fossil wood
At three sites-Magwe: Loc. 1, Loc. 3 and Loc. 5- tools are numerous at Loc. 3. Fig. 75, No. 121, is a
all in the general vicinity of Magwe (see Fig. 20, p. large inverse double-ended hand-adze, a form common
in Early Anyathian 3 (see p. 356). Fig. 75, No. 118, is
295), implements of the same type as those described
above from Kyaukpadaung were found in .direct as- a steep scraper of fossil wood of the type generally
sociation with the Magwe Sand. By far the richest of found in the Neolithic sites of Upper Burma. Whereas
these from an archaeological point of view was Loc. 1 thick flakes of this material are not of infrequent oc-
(P1. VIII, Fig. 4), situated on the peneplain 21/4 miles currence, they are usually large and display no second-
northeast of the town of Magwe on the path leading to ary working. Small well-made implements, such as
the village of Natkin. The geological situation at this Fig. 75, No. 119-a square-ended scraper with a hook-
site is discussed by Dr. de Terra on pp. 306 and 307 shaped perforator projection-are rare. This artifact is
made from an exceptionally fine-grained and homo-
(see Section 9, Figs. 30 and 38). Our investigations
revealed that the culture layer is at the base of the sand, geneous fragment in which the wood structure has al-
which overlies what appears to be an old land surface most completely disappeared.
composed of Nyaungu Red Earth. -Some of the sand Approximately a quarter of a mile east of Magwe
has been blown away by the wind exposing an old camp Pagoda a collection of surface material was made (Fig.
site, where we collected a large series of scraper types 76, Nos. 129-135). No deposits of Magwe Sand were
made of flakes of silicified tuff (Fig. 75, 113-115) as observed at this locality, but there is no difference with
well as fossil wood fragments (Fig. 75, Nos. 110 and regard to patination and degree of silica polish between
this series and that found at Loc. 1. The fossil wood
111). Typical of this locality are small chopping-tools material is made up of small end-scrapers, No. 19,
of flat silicified tuff pebbles obtained from the Pleisto-
cene gravels of the valley (Fig. 75, No. 112). Three steep scrapers, No. 133, and small hand-adzes prob-
polished stone implements, two polishers and one small ably used for scraping, No. 135. The silicified tuff
axe or celt (Fig. 75, No. 109) were found associated implements are all small. Characteristic types include
with the chipped stone tools. The last is made of the thumb-nail scraper, No. 130, the square-ended
basalt, is 11.8 cm. long and is heavily weathered. Its scraper and perforator, No. 131, and the rectangular
rounded edge as well as one side have been sharpened perforator or borer, No. 132 (compare Fig. 75, No.
by the removal of short flakes subsequent to the original 119). Associated with these chipped stone artifacts was
a polished stone spindle whorl, Fig. 76, No. 134. This
polishing. Whereas it was not found in situ at the base
of the Magwe Sand there seems to be no valid reason specimen is 4.3 cm. in diameter and is made of a light
for doubting that it is the same age as the objects of gray, fine-grained metalnorphic rock.
chipped stone from the site. The same applies to 3. Yenangyaung
several of the potsherds that were found under similar
circumstances, and two examples of them are shown East of Yenangyaung, near the southwestern corner
here (Fig. 75, Nos. 116 and 117). The ware is a dark of the Beme Reserve, shown on the map (Fig. 20), a
brick-red color. It is well fired and has a very much small series of typical implements was found in the
finer texture than the pottery from Kyaukpadaung. It vicinity of Dr. Noetling's original site. Only two speci-
is likewise better finished and has a somewhat more mens are illustrated here. Fig. 76, No. 125, is a small
sophisticated decoration (note the diagonal grooves on steep side-and-end scraper, while Fig. 76, No. 126, is a
MOVIUS: THE STONE AGE OF BURMA 383

1110

111
109
Scale
0 1 2 3 4 5 cm.

113 114

112

IG
115

117

118
119 120 121
FIG. 75. Neolithic Implements from Magwe: Loc. 1 (Nos. 109-117); Magwe: Loc. 3 (Nos. 118, 119 and 121);
and Magwe: Loc. 5 (No. 120).
384 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

124 125
123

126 127 128

129

122

130
131
Scale
i
2 3 4 cm.
I~-

132 133134 134- 135


FIG. 76. Neolithic Implements from Chauk (Nos. 122, 124 and 127); Pagan (No. 123); Beme, near
Yenangyaung (Nos. 125 and 126); Nyaungu (No. 128); and Magwe (Nos. 129-135).

similar type of implement of roughly discoidal outline implements associated with deposits of Magwe Sand
and made on a short thin flake of vein quartz. Mr. were observed; the archaeological material collected by
Morris, who has also visited the site, mentions the pres- us was all picked up on the surface of the peneplain.
ence of gravers comparable with "Tardenoisean" forms
of central India (Morris, 1935, p. 6). Whereas a few 4. Chauk
"burins de fortune" were collected by the American
Southeast Asiatic Expedition, no true burins or gravers Mr. Morris reports that in 1930 he found a hand-axe
were noted at any of the localities investigated in Upper of "Chelles type" approximately 130 yards northwest
Burma. Nor do any traces of microlithic (Tardenoi- of the cairn which marks the northeast corner of Block
sian) influences, such as are found in India and Ceylon, 50 N of the Singu Oilfield (Morris, 1930, pp. 19-20).
exist in the Irrawaddy Valley. At Yenangyaung no This locality was carefully investigated by Dr. de Terra
MOVIUS: THE STONE AGE OF BURMA 385
and the writer in November, 1937. According to Mr. that the specimen illustrated here is Neolithic rather
Morris' statement, this section of the oilfield is supposed than Palaeolithic in age. Indeed as far as artifacts of
to mark a river terrace, but we are unable to agree with fossil wood from Burma are concerned, it must first be
this observation (see p. 290 of Dr. de Terra's report). demonstrated that they are associated with Pleistocene
However, on the opposite bank of the Themathauk gravels before remote antiquity for them can be claimed.
Chaung we found the three implements illustrated in Owing to the fact that fossil wood implements display
Fig. 76, Nos. 122, 124 and 127. No. 122 is a typical very little typological variation from Palaeolithic times
hand-adze of fossil wood, similar in many respects to down to that part of the Recent Period when stone tools
the so-called Chellean hand-axe discovered by Morris. still remained in common use, it is obviously dangerous
It is sharp and unrolled, the flake scars are unpatinated, to assign a definite age to surface material. At the pres-
and it was found lying on the surface. Since similar ent stage of our knowledge regarding the prehistory of
implements occur in the Neolithic sites at Kyaukpa- Burma, however, it seems justifiable to consider that
daung and Magwe, and since two small scrapers, Nos. unrolled and only slightly patinated objects picked up
124 and 127, were found near by, we venture to suggest
on the surface belong to the Late Stone Age. It is still
7The Pleistocene sequencein this region as interpretedby Mr. an open question how long this period lasted in the Irra-
Morris (1935) is discussed by Dr. de Terra on pp. 289 and 290
of this report; Mr. Morris' section is reproduced in Fig. 14, waddy Valley; in the extreme north of the country there
p. 289. are people still using stone tools.

I I^ ^ I

138
136 137

Scale
0 1 2
!
3 4
t__
s cm.
[1 ?-
1 II --

142
139 140 141

FIG. 77. Neolithic Implements from Minbu.


386 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
5. Pagan and Nyaungu Magwe Sand on the left bank of the Irrawaddy, marks
At the localities investigated in the vicinity of Pagan the first really significant typological break in the Stone
and Nyaungu a few scattered artifacts of post-Palaeo- Age sequence of Upper Burma. Whereas during
lithic type were collected. Two of these are illustrated Palaeolithic (Anyathian) times large tools, generally
made on cores, are typical, the post-Pleistocene develop-
(Fig. 76, Nos. 123 and 128); No. 123 is a double ment witnessed the introduction of a blade and flake
side scraper and perforator made on a leaf-shaped
flake, which has been worked on the lower surface technique, with the result that most of the implements
of the pointed end. The small steep, single end-and- are small. It therefore seems evident that a new cul-
side scraper, No. 128, is an excellent example of the ture replaced the Anyathian. Now for the first time
the extensive outcrops of fine-grained silicified tuff in
type implement of the culture described in this part of the Mt. Popa region were exploited, and other materials
the report and attributed to the Neolithic Period.
such as vein quartz were used as well. Of the latter
6. Minbu rock there is an abundant supply in the Irrawaddian
Beds (see p. 285), but the fine-grained silicified rock
At Minbu, on the right bank of the Irrawaddy oppo- was brought from the vicinity of Mt. Popa. Fossil
site Magwe (see Fig. 20), Drs. de Terra and Teilhard wood continued in use. Owing apparently to its inher-
found a Neolithic surface site containing chipped tools ent properties rather than to a continuity of tradition,
of fossil wood and silicified tuff in association with ma-
terial of polished stone (Fig. 77, Nos. 136-142). We types of hand-adzes and scrapers are found similar to
those of the Anyathian. Indeed, as has been pointed
therefore have further evidence here that the small
out, implements made of fossil wood are of the same
scraper-perforator complex of Upper Burma is com- fundamental type in Upper Burma from Lower Palaeo-
paratively late. From the collection a small series has lithic times down to the period when stone tools went
been selected for illustration, including a small, inverse, out of common usage. The fact of their late survival is
double-ended scraper, No. 137, and a steep scraper, No. demonstrated by the presence of pottery and polished
138-both of fossil wood-in addition to a blade imple- stone at several of the sites.
ment with a deep notch on the right side, No. 140, a When the data obtained are considered as a whole
perforator with inverse retouch, No. '139, and a small they point fairly conclusively to the fact that the in-
end scraper, No. 141. Nos. 139-141 are of silicified
tuff. The polished stone axe (Fig. 77, No. 136) is 8.6 dustry, originally assigned to the Late Tertiary and
more recently to the Upper Palaeolithic, is not only
cm. long; it is made of schist. This material seems to
post-Pleistocene in age, but also in part contemporary
have been in fairly common use during the full Neo- with traits generally indicative of a Neolithic stage of
lithic Period in Upper Burma, since Morris states
development. At both Kyaukpadaung and Magwe
(1935, p. 9) that in a series of over 300 polished stone (Loc. 1) implements of polished stone occur along with
implements examined by him 10 per cent were of schist. core tools of fossil wood, in addition to flake artifacts
As the broken edge on the right side of the present speci- of silicified tuff and rarely of vein quartz. Pottery like-
men indicates, schist is too brittle to make a really ef-
wise occurs at both sites, but the association of the
fective cutting implement. According to Morris (1935,
sherds with the chipped stone artifacts is not as defi-
p. 11), only three ring stones are at present known from nite as in the case of the polished axes and celts. But.
Burma, one from the Lower Chindwin District and two on the basis of the evidence from Indo-China and Ma-
from Shwebo. Thus the present specimen from Minbu,
laya, it seems clear that the so-called "matt- or cord-
Fig. 77, No. 142, is of interest. Actually it is only a marked ware" from Kyaukpadaung is the sort of thing
small segment of a ring stone, the original diameter of
which was 11.8 cm., the diameter of the hole being 5.3 one would normally expect to find appearing in Burma
cm. The outer edge has been retouched for use as a along with or very shortly after the earliest tools of
scraper. It is made of an exceptionally fine-grained polished stone. The French archaeologists, Mansuy
and Collani, have made extensive excavations in pre-
piece of silt-stone, and it is considerably thicker than any
of the specimens recorded by Morris. In addition to historic sites in Tonkin, Annam and Cambodia (see
Burma, ring stones also occur in Indo-China, China 8 papers by Mansuy and Collani, Mc1i. du Service Geol.
and Manchuria; they are very characteristicof the Neo- de l'Indochine and Bull. de l'Ecole Franc. d'Extrlnie
lithic and later periods. Orient; also: Menghin, 1928 and Heine-Geldern, 1932;
1936). Although the stone implements of the Indo-
III. SUMMARYAND CONCLUSIONS Chinese sites display considerable typological variation
The material associated with the old land surface at as compared with those of Burma, matt-marked pottery
the base of the topsoil at Kyaukpadaung, and with the appears, along with completely polished stone artifacts,
in the late phases of the Hoabinian and Bacsonian Cul-
8 At T'ai P'ing Ch'ang, near Hanchow in Szechwan Province,
tures as well as in the more advanced Somrong-Sen Cul-
several of these objects ("symbols of Heaven") have been found
in a Chou Dynasty (ca. 1100-225 B.C.) grave, according to ture of the full Neolithic Period. The same is true of
Dr. D. C. Graham (1934, plate facing p. 118). Malaya, where cord- or matt-decorated pottery is also
MOVIUS: THE STONE AGE OF BURMA 387
characteristic of the Neolithic. This has been clearly tung, in the extreme east of Burma (Fig. 78). This
demonstrated by the recent careful excavations con- region has never been explored from an archaeological
ducted by Messrs. Collings and Tweedie of the Raffles point of view, and it was hoped that possibly some trace
Museum, Singapore (Collins, 1936, pp. 11-14, Pls. XI, of Early Man would be discovered. However, the re-
XII and XV; Tweedie, 1936, pp. 20-22, Pls. XXIV sults of our investigations were in this respect negative,
and XXV; 1940, p. 16, Pls. VIII-XI). Thus with re- although several sites revealed traces of occupation by
gard to Burma it may be suggested that pottery and man during Mesolithic (?) or later times.
polished stone are probably to be considered as having The structure of the Southern Shan States is in all
appeared at more or less the same time, on the basis essential respects similar to that described by Dr. de
of analogy from elsewhere in southeastern Asia. Ad- Terra for the region farther north (compare Coggin
mittedly, the present data throw very little specific light Brown and Sondhi, 1934; 1934-a). The area is an ele-
on the problem of the chronology of the New Stone vated tableland rising -- 4000 feet above sea-level, and
Age in Upper Burma, but one fact seems fairly evi- separated from the plains of Upper Burma by the Shan
dent: at the sites investigated by the expedition the Border Fault (Fig. 4). Parallel north-south ridges, ris-
presence or absence of pottery and/or polished stone ing 5000-7000 feet above the sea and separated by inter-
cannot be taken as proof that the culture is truly Neo- mountain valleys, traverse the region. For the most
lithic in the sense that the economy was based on the part these ridges are composed of "Plateau Limestone."
production of food. It seems very much more likely Just west of Kalaw (Fig. 1), near the village of Myin-
that in Burma, as in Indo-China and Malaya, a funda- daik, is the divide between the watersheds of the two
mental Mesolithic food-gathering complex persisted great rivers of Burma-the Irrawaddy and the Salween.
relatively late-perhaps as late as the Buddhist Period- The region east of Taunggyi (Fig. 78) has been only
and that it received influences from higher sources. superficially investigated by Coggin Brown and Sondhi
This problem, however, can be solved only by future (1934, see PI. V), but it is obvious that this limestone
excavations. country exhibits the same typical sink-hole topography,
giving rise to karst relief, that is described in detail by
Dr. de Terra for the Northern Shan States (compare
C. RECONNAISSANCE IN THE SOUTHERN
SHAN STATES Coggin Brown and Sondhi, 1934, P1. III, Fig. 2).
West of Taunggyi, the geology of the area is better
I. INTRODUCTION known (Middlemiss, 1900, p. 147; Coggin Brown,
1932, p. 406; Coggin Brown and Sondhi, 1934-a), espe-
In the limestone country of the Southern States and the
cially Valley area, which lies immediately
Tenasserim caves are abundant.9 In the latter region west of Yawnghwe
the Taunggyi Range (Fig. 78). Along the
a number have been reported from the vicinity of base
of this range there is a high-level terrace 300 feet
Moulmein (Fig. 1), which have been described by
above the Yawnghwe Plain, which occurs at a height of
Chhibber (1933, pp. 120-124). Not much is known +
3000 feet above the sea (see Coggin Brown and
about caves in the north of Burma, with the exception
of one recorded by Chhibber (1933, p. 126), near the Sondhi, 1934-a, PI. X, Fig. 1). This terrace is com-
posed of lacustrine deposits, and dates from a period
village of Sankywe in the valley of the Uru River, Myit-
(Late Pliocene-Pleistocene) when the Yawnghwe Val-
kyina District. Approximately 8 miles east of Singti was
and some 50 miles north of Mandalay, Captain Henry ley occupied by an extensive lake of which Lake
Yule and Dr. T. Oldham visited a cave known as Inle (Fig. 78) is a shrunken and shallow remnant. Ac-
Shwe Male in 1855 (Yule, 1858, pp. 177 and 330- cording to Annandale (1918-a, p. 6), Lake Inle is a
solution lake-i.e., one whose bottom has been hollowed
31). This cave, 90-100 yards deep, contained a Bud-
dhist shrine. Caves are very numerous in the region out of the limestone by the dissolving action of water
of Mogok, Northern Shan States, and several of them (compare Chhibber, 1934, pp. 40-43). Our observa-
were explored by Dr. de Terra, as described on pp. 323, tions bear out Annandale's statement that this lake is not
324. Since Middlemiss (1900, p. 131). Annandale the only one that once existed in the area, since the
(1918, p. 143), and Coggin Brown and Sondhi (1934, superficial deposits found in other now empty basins in
p. 139) all report caves in the Southern Shan States, it other sections of the Southern Shan States make it seem
was decided that a short reconnaissance in this region evident that in Plio-Pleistocene times the region was a
should be carried out for the purpose of determining regular lake country. Indeed the local people of Keng-
whether or not any of these localities contained Pleisto- tung have a tradition that formerly a lake filled this
cene deposits. Furthermore Professor G. H. Luce of depression, which at present is a fertile valley.
University College, Rangoon, informed us that fossil Around Taunggyi, where we made our headquarters
bones had recently been found in the vicinity of Keng- during the reconnaissance trip to the Southern Shan
9 For a comprehensive account of the caves of Burma, Siam States, red clays everywhere overlie the Plateau Lime-
and the Malay Peninsula, see Annandale, Coggin Brown and stone. These represent the same lateritized soils which
Gravely (1913). are typical of the Northern Shan States (see p. 319 of
FIG. 78. Map of Part of the Southern Shan States, Showing the Location of Places Referred to in the Text. Dashed Line: Route Travelled
MOVIUS: THE STONE AGE OF BURMA 389
Dr. de Terra's report). Occasionally the red clays at- on a refuse heap outside the near-by village of Wabya.
tain a thickness of 40-50 feet, but owing to the dense Our guide informed us that he could remember when
jungle vegetation exposures are difficult to find. No his mother used to make this pottery, and he asserted
river terraces were noted by us in this region, nor did that wheel-made pottery only came to be extensively
we observe any silicified rocks suitable for use in the used in the small villages in the vicinity of Taunggyi
manufacture of implements. In addition to the caves during the last few years. Nevertheless some of the
to be described presently, several fissures in the lime- sherds from M6ngta-wa-ku are almost certainly older
stones were explored. However, no true breccia or than the Buddhist Period, on the basis of analogies with
travertine, where one might expect to discover the re- other regions of southeastern Asia. But it is impossi-
mains of fossil animals, were found in any of them. ble to be dogmatic on this point in an area where there
Instead they contain only lateritic earth, which is not has apparently been so little change in pottery style
conducive to the preservation of bone. It seems ap- during the last 2000-3000 years. That the upper cave
parent that if such deposits exist in the vicinity of earth containing the pottery represents a deposit which,
Taunggyi, the natives would have discovered them long in part at least, is not older than the Neolithic Period, is
ago, since bones are regarded as having such strong confirmed by the fact that at a depth of 25 cm. the last
magical powers that the owner is immune from disaster. upper molar of a domestic sheep was recovered. This
One man we talked to was convinced that carrying a was identified by Professor Glover M. Allen of the
portion of a modern jaw in a small leather pouch pro- Museum of Comparative Zoology, Harvard University.
tected him from snake bites. The fact that he had trav- At a depth of 12 cm. a small greenstone chip was found,
elled over 70 miles to obtain this object suggests that one surface of which appeared to have been polished.
inhumed bones are scarce in this part of the world where However, whether this was a fragment of a polished
cremation is the rule. Thus our search for fossils in stone axe or of a smooth river pebble could not be de-
the Taunggyi region yielded only negative results. termined. In addition, two species of shells represent-
ing modern forms-Melanoides tuberculata and Brotia
II. LOCALITIES INVESTIGATED variabilis-were collected here, as described by Dr.
Bequaert (see Part IV, p. 431); these are edible
Although nothing significant with regard to the pres- mollusks
ence of Early Man, or of localities containing a Pleisto- unquestionably brought into the cave by man
for food. Small fragments of charcoal and bone were
cene fauna, resulted from our archaeological recon-
noted throughout. Below, at a depth varying from 10-
naissance in the Southern Shan States, the following
12 cm. to 30 cm., a heavily cemented breccia was en-
brief account of the sites which we visited is included.
countered overlying the cave floor. Several exposures
Undoubtedy further excavations at several of them of this deposit were
would yield good results in the form of occupation lay- investigated, but no bones or cul-
tural material were found in it. It is our opinion that
ers of the Mesolithic and later periods, but the time at
the deposits in the first chamber of Mongta-wa-ku have
our disposal did not permit of more than a superficial
been but as there are indications that
examination of each place. As the main object of the the badly disturbed,
site may have been occupied as early as Neolithic
expedition was to search for Early Man in Burma-i.e., excavations in the second chamber might yield
Palaeolithic material-we did not feel justified in con- times,
significant results.
ducting excavations in obviously later deposits. The
sites, which are described below, are all shown on the 2. Tin-A il
map, Fig. 78.
1. Mongta-wa-ku Approximately three miles east of Yawnghwe and one
mile north of the vilage of Le-pin, we explored two small
This is a very large cave, not far from the village of limestone caves of the eastern slope of the valley of the
M6ngta-wa and situated just below the summit of the Hsin-Dawng stream. These caves, Tin-Ain North and
Taunggyi Range, ? 4500 feet above sea-level. It is Tin-Ain South, lie at the base of the range of moun-
approximately two and a half miles southwest of tains forming the eastern edge of the Yawnghwe Valley,
Taunggyi, near the top of the 1500-foot escarpment and they face westward towards Lake Inle. Both are
overlooking the Yawnghwe Valley. As the large second ? 350 feet above the level of the lake, which in turn is
chamber of the cave serves as an important shrine con- 2915 feet above the sea. We were unable to dig in
taining three large statues of Buddha, we had to con- the first chamber of either of the Tin-Ain Caves, as each
fine our test pits to the first chamber, where the deposits contains a pagoda. A small test pit in the second
were somewhat disturbed. These yielded both wheel- chamber of the North Cave yielded negative results:
made and glazed sherds of modern Burmese pottery, only two small sherds of modern Burmese pottery were
in addition to sherds of a hand-made ware decorated in obtained. No pottery or artifacts were found in the
some instances with incision or cord impressions. This South Cave; however, in a small chamber to the right of
latter material was all considered to be prehistoric until the entrance we recovered a number of animal bones in
we found broken pots decorated in an identical manner addition to some shells and charcoal fragments. The
390 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
facts that all the bones, with the exception of the teeth, plain the presence of this very clearly defined low hill
were broken and that ash and charcoal occurred through- or mound. Possibly it represents a deltaic deposit
out the cave earth, led us to the conclusion that this site formed where a stream debouched into the lake; in any
had been occupied by man. The shells, which have been case the hill itself is not artificial. The remarkablething
described by Dr. Bequaert (see Part IV, p. 431), about Ahyu Taung is that the surface deposit, which is
belong to two species-Taia intermedia and Brotia per- 15 to 18 cm. thick, is full of the broken bones and teeth
sculpta-representing variants of types now living in of animals. Many of these are burned, and fragments
Lake Inle. Both of these molluscs are edible. Accord- of charcoal as well as angular quartz pebbles, apparently
ing to Professor Allen, the animal bones include the fol- split and cracked as the result of heat, also occur in
lowing forms: Hog Deer (Cervus porcinus), Eld's Deer profusion. Most of the bones, which were heavily
or Thamin (Cervus eldi), Sambar Deer (Rusa uni- weathered, have been identified by Professor Allen as
color), Rhinoceros (1 lower molar fragment), and Cattle belonging to Cattle, but the material is too fragmentary
(the size of domestic Cattle, but also similar to Wild to determine whether they represent the wild (Bibos
Cattle or Banting-Bibos sondaicus). Several of the sondaicus) or the domestic (Bos taurus) form. A few
bones are burned, and this supports the view that we can probably be attributed to Eld's Deer or Thamin
are dealing with a culture deposit. (Cermvs eldi). The condition of the bones suggests that
Now these are the only caves in this vicinity, therefore this place has a considerable antiquity, which is borne
they are almost certainly the ones visited by Annandale out by the fact that there is a local tradition about it. A
and Gravely in 1917 (Annandale, 1918, p. 143; 1919, Buddhist monk stated that Ahyu Taung was formerly
pp. 217-18). Although Annandale mentions two lime- inhabited by a demon known as Balu, who used to cap-
stone caves in the Hsin-Dawng Valley, + 3 miles east ture men and take them to the hill where he devoured
of Yawnghwe and 300-500 feet above the level of Lake them. When we suggested that possibly the site repre-
Inle, he does not state from which of the caves he ob- sented the remains of a prehistoric feast, he calmly
tained his collections of shells. Nor does he record any stated that such barbaric customs were unknown to the
of the mammal fauna with the single exception of teeth Shans, but that in the days of Balu men lived in terror
of Cervus eldi. The shells on the basis of his inter- of this frightful monster!
pretation all represent extinct species, washed into the
cave by floods during a period when the Yawnghwe 4. Pd-Leng
Valley was filled by a lake, of which the Hsin-Dawng
basin was a small but deep subsidiary. Later (Annan- Just below the summit of the large hill (4643 feet
dale and Rao, 1925, p. 102), after a second visit to the above sea-level) that rises to the north of the village of
Pa-Leng, there is a cave, the mouth of which faces west.
locality, this view was somewhat modified, since Annan-
dale asserts that "the shells had evidently been washed This site can be seen on the left of the road ? 10 miles
into the caves mainly through holes in the roof." The from Taunggyi, on the edge of the Hopong Plain. Al-
makes it seem much though the entrance is imposing, the cave itself is a very
archaeological evidence, however,
more likely that the shells were brought here by man for shallow affair, and it is filled with Buddhist statuary.
No deposits overlie the cave floor; furthermore there
food, which is also Dr. Bequaert's opinion (see p. 432).
Neither of the caves contains lacustrine deposits; fur- was no evidence in the talus in front suggesting that this
thermore the mammlalfauna includes only post-Pleisto- site had ever been occupied during prehistoric times.
cene forms. This agrees with Dr. Bequaert's conclusion,
based on the conchological evidence, that the supposedly 5. Htamsdng
"extinct" species of Taia from Tin-Ain are merely Near milestone 25.6 and immediately north of the
variants of a species now living in Lake Inle. If the road between
Hopong and M6ng Pawn, there is a small
pagoda that occupies the main chamber of Tin-Ain valley near the village of Htamsang. This cave was
South could be removed, a complete archaeological in- visited
by Coggin Brown and Sondhi (1934, p. 139 and
vestigation of this site would be well worth while. PI. IV). It is filled with very beautiful formations of
stalagmites and stalactites, but we could find no indica-
3. Ahyli Taung tions that it had ever been occupied. Presumably dur-
This locality is situated near the village of Kunlon in ing Pleistocene or even later times, this cave was partly
the Yawnghwe Valley, 33 miles north of the Kalaw- filled with water; there are many large pools-in it at
Taunggyi road on the way to Lawksawk. The site of present, and parts of it are still flooded during the wet
Ahyfi Taung, which means "Bone Hill" in Burmese, season.
consists of a low hill approximately 8 feet high and 100
feet in diameter which rises from the Kunlon Plain. 6. Mong Pawn
As this plain is composed of alluvium, laid down when This place is situated ? 40 miles on the road east of
the valley was occupied by a lake (Coggin Brown and Taunggyi, on the west side of the M6ng Pawn Plain.
Sondhi, 1934-a, pp. 190-193), we were unable to ex- Here to the south of the road and overlooking the Nam
MOVIUS: THE STONE AGE OF BURMA 391
Pawn River, there is a small cave facing east, approxi- fact the presence of caves near Lawksawk, Mongnai,
mately half a mile west of the town of Mong Pawn. At Mawkmai and Loilem were reported to us, but unfortu-
present this cave is occupied by a Buddhist monk, there- nately we were not able to explore them. Of the locali-
fore no excavations in it were possible. Nearby, how- ties listed above, the rock-shelter near Mong Pawn
ever, there is a rock-shelter where we dug a small pit seems to us the best site for excavation. A superficial
1.00 m. deep and 75 cm. square. This yielded: 8 broken examination of the deposits revealed that this place had
river pebbles (probably artificially broken), 1 pointed been occupied by a shell-eating people. Although no
limestone flake, 2 pieces of quartz, numerous shells rep-actual implements were found, the presence of broken
resenting modern forms--Brotia baccata and Sulcospira river pebbles, a limestone flake, and shells in profusion
praemordica (see p. 434 of Dr. Bequaert's report), bits suggests that a Mesolithic or Early Neolithic (?) in-
of charcoal, in addition to numerous fragments of bone dustry similar to that found in the Caves of Kwangsi-
too small to identify. The evidence secured leads to the Kweilin, near Hsingan, and Wuming, northwest of
conclusion that the locality was occupied by prehistoric Yungning-investigated by Drs. Teilhard de Chardin,
man, probably during the post-Pleistocene Period. Had Young, Pei and Chang (1935, pp. 192-98; see also Pei,
time permitted, we would have excavated the area more 1935)-is also present in Burma. Comparable deposits
extensively, since this site was the most promising one have likewise been reported from the Heichinglungtsun
which we found in the Southern Shan States from an rock-shelter, near Chiupei, Yunnan, by M. N. Bien and
archaeological point of view. L. P. Chia (1938, pp. 345-7). The archaeological evi-
dence from South China has recently been summarized
7. Tongta by Dr. W. C. Pei (1939, p. 126), who suggests the
possibility of a late Upper Palaeolithic dating on the
The village of Tongta is situated in Kengtung State, basis of the degree of mineralization of the fossils.
east of the Salween River and approximately 40 miles Nevertheless, in spite of the absence of pottery, it seems
from the town of Kengtung. Just north of Tongta on to us much more
likely that the cave deposits of both
the left bank of a small stream, a tributary of the Nam South China and eastern Burma are the age of the
Hsim, there is a small cave, formed by a cleft in the Bacsonian (Early Neolithic) of Indo-China, especially
limestone which has been enlarged by weathering. This since many of the cultural features of the Kwangsi
site had only a very shallow deposit containing no cul- caves are
closely akin to those reported from the Tonkin
tural refuse. localities (see Pei, 1935, pp. 407-8). Thus present
indications point to the probability that the post-
8. Kengtung Pleistocene Stone Age sequence of Indo-China and
From Kengtung in the extreme east of Burma, several Malaya will some day be extended to include southern
China and eastern Burma. This problem, however,
important trade routes continue eastward into south-
western Yunnan and Indo-China, as well as southward can be solved only by further reconnaissance and exca-
into Siam. In fact it is possible to reach Siam by car; vation in the sites of the Southern Shan States.
otherwise Kengtung is the terminus of the main 300-mile In Burma the presence of Buddhist shrines and pa-
road which starts at Taunggyi and traverses the South- godas in the limestone caves is bound to be a really
ern Shan States from east to west. As stated above, the serious handicap to the archaeologist. For instance, our
observations at the cave of Tin-Ain South demonstrate
trip to Kengtung was made in order to investigate the
fossiliferous fissures reported from this region. On that this place was occupied during post-Pleistocene
arrival we learned that this would have required or- times, but in order to excavate the main deposit, one
ganizing a pack trip and travelling several (lays south- would have to remove the medium-sized pagoda which
ward to the valley of the Nam Hkok River, a journey occupies practically the whole cave. At Mongta-wa-ku
which we were not prepared to make. However, two part of the second chamber could be dug without inter-
fossil elephant teeth from the Nam Hkok locality were fering with the Buddhist statuary; according to our
presented to the expedition by Dr. James H. Telford, investigations the Neolithic or later deposits of the first
Director of the American Baptist Mission, Pangwe chamber are quite badly disturbed. As regards Pleisto-
(near Loimwe). These have been identified by Dr. cene fossils, the gravel and pebble beds in the high-level
Colbert (see p. 423) as possibly a survival of Elephas terraces found in the intermountain valleys might yield
some material. Furthermore the southern part of
hysudricuzsbut more probablyPalaeoloxodon namzadiclts.
Kengtung State (Nam Hkok Valley) should be investi-
III. SUMMARY AND CONCLUSIONS gated. The fossil elephant teeth collected in this region
by Dr. Telford presumably come from a fossiliferous
None of these sites which we investigated during the fissure or cave deposit of the Pleistocene Period. Even
limited time at our disposal for reconnaissance in the so nothing approaching the rich fossil-localities of the
Southern Shan States offered any promise of containing Mogok District, described by Dr. de Terra (see p.
evidence of occupation by Palaeolithic man. But caves 323), is suggested by the present evidence from the
abound in the region, and these should be visited. In region of the Southern Shan States.
392 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

BIBLIOGRAPHY OF PARTS I AND II


ANNANDALE,N., J. COGGINBROWN,and P. H. GRAVELY.1913. DAINELLI, G. 1922. Studi Sul Glaciale nel Himalaya e Cara-
The limestone caves of Burma and the Malay Peninsula. corum. Bologna, 1922.
Jour. Asiatic Soc. Bengal, IX, pp. 391-424. DALTON,L. V. 1908. Notes on the Geology of Burma. Quart.
ANNANDALE, N. 1918. Aquatic Molluscs of the InIe Lake and Jour. Geol. Soc. London, LXIV (1908) 604-644.
Connected Waters. Rec. Indian Mus. Calcutta, XIV DALY, R. A. 1934. The Changing World of the Ice Age.
(1918) 103-182. New Haven, 1934.
. 1918-a. Introductory -Account of the Inle Lake. Rec. DAS GUPTA, H. C. 1923. Indian Prehistory. Jour. Dept. of
Indian Mus. Calcutta, XIV (1918) 1-7. Science, Calcutta Univ., V (1923) 10-15.
. 1919. The Gastropod Fauna of Old Lake-Beds in Upper DE TERRA,H. 1938. Preliminary Report on Recent Geological
Burma. Rec. Geol. Surv. India, L (1919) 209-240. and Archaeological Discoveries Relating to Early Man in
ANNANDALE, N., and H. S. RAO. 1926. Further Observations Southeast Asia. Proc. Nat. Acad. Sci., XXIV (1938)
on the Aquatic Gastropods of the Inle Watershed. Rec. 407-413.
Indian Mus. Calcutta, XXVII (1925-26) 101-127. DE TERRA,H., P. TEILHARDDE CHARDIN, and H. L. MovIus,
BARBOUR, G. B. 1935. Physiographic History of the Yangtze. JR. 1938. Geological and Archaeological Aspects of South-
Mem. Geol. Surv. China, XIV (1935) 1-112. Eastern Asia. Nature, 142 (1938) 275.
BIEN, M. N., and L. P. CHIA. 1938. Cave and Rock-Shelter DE TERRA, H., and T. T. PATERSON. 1939. Studies on the
Deposits in Yunnan. Bull. Geol. Soc. China, XVIII (1938) Ice Age in India and Associated Human Cultures. Car-
325-347. negie Inst. Washington Pub. No. 493 (1939), 1-354. (Also
BLACK,DAVIDSON,P. TEILHARDDE CHARDIN,C. C. YOUNG,and quoted as de Terra, 1939; Paterson, 1939; Paterson, with
W. C. PEI. 1933. Fossil Man in China. Mem. Geol. de Terra, 1939; and de Terra, with Paterson, 1939.)
Surv. China, Series A. No. 11 (1933) 1-166. DE TERRA, H. 1939-a. The Quaternary Terrace System of
BLANFORD,W. T. 1895. The Burmese Chipped Flints, Plio- Southern Asia and and the Age of Man. Geog. Review,
cene and Miocene. Nature, LI (1895) 608. XXIX (1939) 101-118.
BOUTERWEK,K. 1919. Das Land der Meridionalen Strom- DE TERRA, H. 1941. Pleistocene formations and Stone Age
furchen im Indo-Chinesischen Grenzgebiet. Mitt. d. Geo- Man in China. Institut de Geo-Biologie, No. 6, Peking,
graph. Gesell. Miinchen, XIII (1918-19) 161-341. p. 54, 5 figs.
BROWN, J. COGGIN. 1931. Relics of the Stone Age in Burma. GORDON,R. 1882. The Irrawaddy and the Sanpo. Proc. Roy.
Jour. Burma Research Soc., XXI (1931) 33-43. Geog. Soc., VI (1882) 559-563.
. 1932. The Geology and Lead-Ore Deposits of Maws6n, 1885. The Irrawaddy River. Geog. Jour. VII (1885)
Federated Shan States. Rec. Geol. Surv. India, LXV 292-325.
(1931-32) 394-433. GRAHAM,D. C. 1934. A Preliminary Report of the Hanchow
BROWN,J. COGGIN,and V. P. SONDHI. 1934. Geological Re- Excavation. Jour. West China Border Res. Soc., VI
connaissance in the Southern Shan States. Rec. Geol. Surv. (1933-34) 114-131.
India, LXVII (1933-34) 135-165. .1935. Implements of Prehistoric Man in the West China
.1934-a. The Geology of the Country between Kalaw Union University Museum of Archaeology. Jour. West
and Taunggyi, Southern Shan States. Rec. Geol. Surv. China Border Res. Soc., VII (1935) 47-56.
India, LXVII (1933-34) 166-248. HEINE-GELDERN, R. 1932. Urheimat und friiheste Wanderun-
CAMMIADE,L. A., and M. C. BURKITT. 1930. Fresh Light on gen der Austronesier. Anthropos, XXVII (1932) 543-619.
the Stone Ages of Southeast India. Antiquity, IV (1930) .1936. Prehistoric Research in Indonesia. Ann. Bibliog.
327-339. Indian Arch., IX (1936) 26-38.
CHHIBBER,H. L. 1927. The Origin and Mineral Constitution HOLDEN,(Miss) R. 1916. A Fossil Wood from Burma. Rec.
of the Late Tertiary Fossil Wood of Burma. Jour. Proc. Geol. Surv. India, XLVII (1916) 267-272.
Asiatic Soc. Bengal, XXIII (1927) 13-26. HOOTON,E. A. 1940. Why Men Behave like Apes and Vice
1933. The Physiography of Burma. London, 1933. Versa. Princeton, 1940.
1934. The Geology of Burma. London, 1934. HUNTINGTON, E. 1919. The Pulse of Asia. New York, 1919
CLEGG,E. L. C. 1938. The Geology of the Minbu and Tha- (2nd edition).
yetmyo Districts, Burma. Mem. Geol. Surv. India, LXXII JONES, T. R. 1894. Miocene Man in Burma. Natural Science,
(1938) 137-317. V (1894) 345-349.
E. H. 1935. Siwalik Mammals in the American Mu-
COLBERT, 1928. Die Bedeutung der Depression der Schnee-
KLUTE, F.
seum of Natural History. Trans. Amer. Phil. Soc., XXVI
grenze fur eiszeitliche Probleme. Zeit. Gesell. Erdkunde,
(1935) 1-401. XVI (1928) 70-93.
.1938. Fossil Mammals from Burma in the American
Museum of Natural History. Bull. Amer. Mus. Nat. Hist., KRISHNASWAMI, V. D. 1938. Environmental and Cultural
LXXIV (1938) 255-436. Changes of Prehistoric Man near Madras. Jour. Madras
1895. Miocene Man in Burma. Natural Sci- Geog. Assoc., XIII (1938) 58-90.
COLE,G. A. J. . 1938-a. Prehistoric Man Round Madras. Indian Acad.
ence, VII (1895) 295.
H. D. 1936. Report of an Archaeological Excava- Sci., Madras Meeting, 1938.
COLLINGS,
tion in Kedah, Malay Peninsula. Bull. Raffles Museum, I LA TOUCHE,T. D. H. 1906. On Recent Changes in the Course
of the Namtu River, Northern Shan States. Rec. Geol.
(1936) 3-16.
1938. Pleistocene Site in the Malay Peninsula. Nature, Surv. India, XXXIII (1906) 46-48.
CXLII (1938) 575-576. LA TOUCHE, T. D. H., and R. R. SIMPSON. 1906. The Lashio
COTTER,G. DE P. 1914. Some Newly Discovered Coal-Seams Coalfield, Northern Shan States. Rec. Geol. Surv. India,
near the Yaw River, Pakokku District, Upper Burma. XXXIII (1906) 117-124.
Rec. Geol. Survey India, XLIV (1914) 163-185. LA TOUCHE,T. D. H. 1913. The Geology of the Northern
.1918. The Geotectonics of the Irrawaddy Basin. Jour. Shan States. Mem. Geol. Surv. India, XXXIX (1913)
Proc. Asiatic Soc. Bengal, XIV (1918) 409-420. 1-379.
.1938. The Geology of Parts of the Minbu, Pakokku, LEAKEY,L. S. B. 1934. The Sequence of Stone Age Cultures
and Lower Chindwin Districts. Mem. Geol. Surv. India, in East Africa. Article in "Essays Presented to C. G.
LXXII (1938) 1-136. Seligman, 1934." London, 1934, pp. 143-147.
MOVIUS: THE STONE AGE OF BURMA 393
LEICESTER,P. 1932. The Geology and Underground Water SIMPSON,R. R. 1906. The Namma, Man-sang and Man-se-le
Supply of Rangoon. Government Press, Rangoon (1932). Coalfields, Northern Shan States. Rec. Geol. Surv. India,
MATTHEW,W. D., and W. GRANGER.1923. New Fossil Mam- XXXIII (1906) 125-156.
mals from the Pliocene of Sze-Chuan, China. Bull. Amer. STAMP,D. L. 1922. An Outline of the Tertiary Geology of
Mus. Nat. Hist., XLVIII (1923) 563-598. Burma. Geol. Mag., LIX (1922) 481-501.
MENGHIN, OSWALD. 1928. Zur Steinzeit Ostasiens. P. W. . 1924. Notes on the Vegetation of Burma. Geog. Jour.,
Schmidt--Festschrift: Publication d'Hommage. Wien, LXIV (1924) 231-237.
1928, pp. 908-942. STEIN, K. 1931. Birma: Das Stromgebiet des Irawadi. Mitt.
MIDDLEMISS,C. S. 1900. Report on a Geological Reconnais- d. Geograph. Gesell. Miinchen, XXIV (1931) 1-101.
sance in parts of the Southern Shan States and Karenni. SWINHOE, R. J. C. 1902. Prehistoric Man in Burma. Zool-
General Report of the Geol. Surv. of India for 1899-1900, ogist, VI (1902) 321-336.
pp. 122-153. . 1903. Some Further Notes on Chipped Flints at Ye-
MITRA,PANCHANAN. 1927. Prehistoric India: its place in the nangyaung,Upper Burma. Zoologist, VII (1903) 254-259.
world's cultures. Calcutta, 1927 (2nd edition). TEILHARDDE CHARDIN, P., and C. C. YOUNG. 1935. The Ceno-
MORRIS, T. 0. 1932. A Palaeolith from Upper Burma. Jour. zoic Sequence in the Yangtse Valley. Bull. Geol. Soc.
Burma Research Soc., XXII (1932) 19-20. China, XIV (1935) 161-178.
. 1935. The Prehistoric Stone Implements of Burma. TEILHARDDE CHARDIN, P., C. C. YOUNG,W. C. PEI, and H. C.
Jour. Burma Research Soc., XXV (1935) 1-39. CHANG. 1935. On the Cenozoic Formations of Kwangsi
. 1936. A Palaeolith from Yenangyaung. Jour. Burma and Kwangtung. Bull. Geol. Soc. China, XIV (1935)
Research Soc., XXVI (1936) 119-121. 179-205.
. 1936-a. The Konbyin Terrace of the Irrawaddyat Tha- TEILHARDDE CHARDIN,P. 1936. Notes on Continental Geol-
yetmyo. Jour. Burma Research Soc., XXVI (1936) 163- ogy. Bull. Geol. Soc. China, XVI (1936) 195-220.
169. . 1937. The Post-Villafranchian Interval in North China.
-. 1937. Prehistoric Stone Implements from the Konbyin- Bull. Geol. Soc. China, XVII (1937) 169-176.
myint of the Irrawaddy and Paunglaung Rivers. Jour. .1938. See De Terra, H., P. Teilhard de Chardin, and
Burma Research Soc., XXVII (1937) 74. H. L. Movius, Jr.
NOETLING,F. 1891. Report on the Coalfields in the Northern . 1939. Les recherches prehistoriques en Extreme Orient.
Shan States. Rec. Geol. Surv. India, XXIV (1891) 99- Anthropologie, XLIX (1939) 251-252.
119. TEILHARD DE CHARDIN, P., and W. C. PEI. 1932. The Lithic
.1894. On the Occurrenceof Chipped (?) Flints in the Industry of the Sinanthropus Deposits in Choukoutien.
Upper Miocene of Burma. Rec. Geol. Surv. India, XXVII Bull. Geol. Soc. China, XI (1932) 315-358.
(1894) 101-103. THEOBALD, W. 1870. On the Alluvial Deposits of the Irawadi,
. 1895. The Development and Sub-Division of the Ter- more particularly as contrasted with those of the Ganges.
tiary System in Burma. Rec. Geol. Surv. India, XXVIII Rec. Geol. Surv. India, III (1870) 17-27.
(1895) 59-86. THORP,JAMES, 1935. Geographic Distribution of the Important
1897. On the Discovery of Chipped Flint Flakes in the Soils of China. Bull. Geol. Soc. China, XIV (1935) 119-
Pliocene of Burma. Natural Science, X (1897) 233-241. 146.
O'BRIEN,T. P. 1939. The Prehistory of Uganda Protectorate. 1935-a. Soil Profile Studies as an Aid to Understanding
Cambridge,England, 1939. Recent Geology. Bull. Geol. Soc. China, XIV (1935) 359-
OLDHAM, R. D. 1895. The Alleged Miocene Man in Burma. 381.
Natural Science, VII (1895) 201. 1937. The Soils of China. Nanking, 1937.
PASCOE, E. H. 1912. The Oil-Fields of Burma. Mem. Geol. TWEEDIE, M. W. F. 1936. Report on Cave Excavations car-
Surv. India, XL (1912) 1-269. ried out in Bukit Chintamani,near Bentong, Pahang. Bull.
PATERSON,T. T. See De Terra, H., and T. T. Paterson. Raffles Museum, I (1936) 17-25.
PEI, W. C. 1931. Notice of the Discovery of Quartz and .1940. Report on Excavations in Kelantan. Jour. Ma-
Other Stone Artifacts in the Lower Pleistocene Hominid- layan Branch Roy. Asiatic Soc., XVIII (1940) 1-22.
Bearing Sedimentsof the ChoukoutienCave Deposits. Bull. VAN RIET LOWE,C. (with P. G. S6HNGE, and D. J. L. VISSER).
Geol. Soc. China, XI (1931) 109-139. 1937. The Geology and Archaeology of the Vaal River
. 1934. Report on the Excavation of the Locality 13 in Basin. Dept. of Mines: Geol. Surv. of the Union of South
Choukoutien. Bull. Geol. Soc. China, XIII (1934) 359-367. Africa. Memoir 35. Pretoria, 1937.
. 1935. On a Mesolithic (?) Industry of the Caves of VON KOENIGSWALD, G. H. R. 1936. Early Palaeolithic Stone
Kwangsi. Bull. Geol. Soc. China, XIV (1935) 393-408. Implements from Java. Bull. Raffles Museum I (1936)
.1937. Palaeolithic Industries in China. Article in 52-60.
"Early Man" (edited by G. G. MacCurdy). Philadelphia, . 1937. A Review of the Stratigraphy of Java and Its
1937, pp. 221-232. Relation to Early Man. Article in "Early Man" (edited
.1937-a. Les Fouilles de Choukoutien en Chine. Bull.
Soc. Prehist. France. XXXIV (1937) 354-366. by G. G. MacCurdy). Philadelphia, 1937, pp. 23-32.
.1939. Das Pleistocan Javas. Quartir, II (1939) 28-53.
. 1939. The Recent Progress of Quaternary Study in
China. Quartar, II (1939) 120-132. WARD, F. KINGDON. 1912. Through the Lutzu Country to
. 1939-a. A Preliminary Study on a New Palaeolithic Menkong. Geog. Jour., XXXIX (1912) 582-592.
Station known as Locality 15 within the ChoukoutienRe- .1924. The Snow Mountains of Yunnan. Geog. Jour.,
Bull. Geol. Soc. XIX 147-187. LXIV (1924) 222-231.
gion. China, (1939)
. 1939-b. An Attempted Correlationof QuaternaryGeol- WAYLAND, E. J. 1934. Rifts, Rivers, Rains and Early Man
in Uganda. Jour. Roy. Anthropol. Inst., LXIV (1934)
ogy, Palaeontology and Prehistory in Europe and China.
Inst. of Arch., Univ. of London, Occ. Paper No. 2. Geo- 333-352.
WOODWARD, SIR A. SMITH, 1915. On the Skull of an Extinct
chronological Table No. 1. London, 1939.
PILGRIM,G. E. 1906. Fossils of the Irrawaddy Series from Mammal Related to Aeluropus from a Cave in the Ruby
Rangoon. Rec. Geol. Surv. India, XXXIII (1906) 157- Mines at Mogok, Burma. Proc. Zool. Soc. London, 1915,
158. pp. 425-428.
. 1910. Preliminary Notes on a Revised Classificationof YULE, CAPTAIN HENRY. 1858. A Narrative of the Mission to
the Tertiary Freshwater Deposits of India. Rec. Geol. the Court of Ava in 1855 (with an Appendix by Dr. T.
Surv. India, XL (1910) 185-205. Oldham). London, 1858.
PLATE XIII

FIG. 1. The clearly-defined T3 level immediately south of Yenangyaung. View looking from Thittabwe
towards Nyaunghla.

FIG. 2. Dissected T3 remnants between Sadaing and Yonzeik, south of Yenangyaung. Irrawaddy River
on the left.
PLATE XIV

FIG. 1. Exposure of T3 gravels south of Sadaing.

FIG. 2. The gravel pit on T3 at Nyaungbyubin, immediately south of Chauk.


PLATE XV

FIG. 1. T3 east of Nyaungu on the road to Kabani. Note the pagodas of Pagan in the distance.

FIG. 2. Collecting implements in the basal ferruginous crust exposed on '3east of Nyaungu. PS, Pagan silt.
PLATE XVI

FIG. 1. The valley of the Sai Chaung near the village of Sebauk, at the base of Mt. Popa, northeast of
Kyaukpadaung.

FIG. 2. Neolithic site (Ky. 1) in the valley of the Sai Chaung near the village of Sebauk. The dashed line
indicates the archaeological horizon at the base of the topsoil.
PLATE XVII

Early Anyathian 2-3 implements of fossil wood. 1, Single-ended hand-adze; 2, single-ended and side hand-adze; 3, pick-like im-
plement; 4, single-ended chopper. All from Nyaungu.
PLATE XVIII

Scale
aI_$I _ = Jocm
* W

3
Early Anyathian 3 implements of silicified tuff. 1, Chopper flaked only on the upper surface; 2, massive pick-like implement; 3,
massive triangular chopper; 4, chopping-tool with alternately flaked edge. 1, 2 and 4 from Yenangyaung; 3 from Chauk.
PART III

PLEISTOCENE VERTEBRATES COLLECTED IN BURMA BY


THE AMERICAN SOUTHEAST ASIATIC EXPEDITION
BY EDWIN H. COLBERT

Plates XIX-XXXII

I. INTRODUCTION lishing the correct localities for some of the fossils col-
lected by him.
In the fall of 1937 an expedition sponsored by the I am particularly indebted to Mr. Robert G. Chaffee,
Academy of Natural Sciences of Philadelphia, the Pea- of the Academy of Natural Sciences of Philadelphia,
body Museum of Archaeology and Ethnology of Har- for much help in the prosecution of this study. Mr.
vard University, the Carnegie Institution of Washington, Chaffee prepared many of the fossils that needed such
and the American Philosophical Society, and designated attention, and he supervised the numbering of the speci-
as "The American Southeast Asiatic Expedition," was mens. He has been instrumental in sorting and ar-
sent to Burma for the purpose of studying the Pleisto-
ranging the collection, and in keeping straight the rec-
cene geology of that region in its relation to Early Man. ords of these fossils.
This expedition was under the leadership of Dr. H. de Miss Virginia W. Tomlin (now Mrs. John L. Cotter),
Terra, then associated with the Academy of Natural formerly of the Academy of Natural Sciences, typed the
Sciences and the Carnegie Institution, and included
manuscript and assisted in reading proof. The pen
among its personnel, Dr. Hallam L. Movius, Jr., of and ink illustrations were made by Mr. John Sheldon,
Harvard University, and Dr. P. Teilhard de Chardin, the maps by Mrs. Elsa Arnoux. The photographs
of the National Geological Survey of China. In con- were made by Mr. A. Delwin Warden and retouched
nection with their work of studying the sequence of
by Mr. Sheldon.
Pleistocene deposits, as an interpretation of the Ceno-
zoic history of southeastern Asia, the members of the ex- II. THE PLEISTOCENE FOSSIL BEDS
pedition collected all available fossils of Pleistocene age. OF BURMA
In this manner a small but important series of Pleisto-
cene vertebrates, consisting for the most part of mam- Before presenting descriptions of the fossils found by
malian remains, was collected from along the upper members of the American Southeast Asiatic Expedi-
reaches of the Irrawaddy River, between Magwe and tion, it may be well to review briefly certain facts con-
Mogok. These fossils, which are to be divided equally cerning the Pleistocene geology of Burma, as we now
between the Academy of Natural Sciences of Philadel- know them (see Dr. de Terra's report). This brief
phia and the Museum of Comparative Zoology of Har- elucidation of some of the more significant points as to
vard University, were turned over to the writer for the Pleistocene sequence in Burma, while not in the
identification and description, with the particular object least definitive, may serve to a certain extent as a basis
of ascertaining what additional evidence they might for the better understanding of the fossils themselves.
offer as to the sequence and relationships of the Qua- Pleistocene fossils in Burma have been found in two
ternary vertebrate faunas of Burma. distinct physiographic provinces: namely (a) the Shan
This investigation logically follows a recently pub- Plateau and (b) the Central Belt of the Irrawaddy
lished study of all of the extinct mammals from Burma River Valley.
hitherto known (Colbert, 1938), and may be considered (a) The Shan Plateau occupies the eastern half of
as supplementary to the paper cited herewith. the country and is a table-land composed of folded pre-
I wish at this place to acknowledge the help that I Palaeozoic, Palaeozoic and Mesozoic rocks, limited to
have had from various persons in the preparation of the west by the Shan boundary fault. In the northern
this paper. To Drs. P. Teilhard de Chardin and H. de portion of the plateau, particularly in the vicinity of
Terra I am greatly indebted for much advice as to the Mogok, north of Mandalay, is an extensive develop-
stratigraphic and geographic relationships of the fos- ment of karst topography, which has resulted in the
sils herein described. These gentlemen have also ex- formation of numerous fissures and caves. It is within
pressed to me quite freely their views, in some cases as the fillings of these fissures that Pleistocene fossils are
yet unpublished, concerning the Pleistocene sequence found in this plateau area of Burma.
in southeastern Asia. To Dr. Hallam L. Movius, Jr., of (b) The Central Belt of the Irrawaddy River Val-
Harvard University, I am indebted for help in estab- ley lies between the Shan Plateau on the east and the
395
396 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
folded Arakan Yoma on the west. It is a long, nar- to note in this connection that practically all of these
row strip of lowlands about 600 miles in length in a fossils are derived and not in situ, for they commonly
north to south direction, and averaging about 130 miles show evidences of a considerable amount of rolling and
in width. Pleistocene fossils in this region are found weathering. Moreover, these fossils are in almost all
most abundantly between the vicinity of Yenangyaung cases of Upper Irrawaddy relationships, and thus are
on the south and Mandalay on the north, contained of Lower Pleistocene age. Occasionally a fossil is
within thick sandstones of the Irrawaddy series and found in situ in the third terrace, in which case it is a
within a series of post-Irrawaddian River Terraces. later Pleistocene form, not an animal of true Irrawad-
The Pleistocene of Burma is inaugurated by the Ir- dian age.
rawaddian sequence of deposits, from which fossils The lowest terraces along the Irrawaddy River sys-
have been known for more than a hundred years. In- tem are the fourth and fifth of the series, designated T4
deed, the first fossil vertebrates collected in Burma and and T5 respectively by de Terra. Of these, the fourth
described in 1828 were obtained from the Upper Ir- is probably of late Pleistocene affinities, while the fifth
rawaddy beds along the shores of the Irrawaddy River. is post-Pleistocene in age. Some fossils are found in
As has been shown by several authors, the Irrawaddy or below the fifth terrace, and these are the latest re-
beds as a whole contain both Pliocene and Pleistocene mains occurring in the Central Irrawaddy River Valley
deposits, with definitely separable faunas. Thus, in the province.
lower sections of the Irrawaddy series are fossils of The terraces as developed along the Irrawaddy River
Pliocene age, comparable to the assemblage comprising have been briefly described by de Terra (1939, pp. 110-
the Dhok Pathan fauna of northern India. Above this
faunal horizon, separated from it by a distance of some 111) as follows:
4500 feet, according to Stamp, is the zone wherein oc- On the left bank of the Irrawaddy between Mandalay
curs the typical Upper Irrawaddy fauna, a mammalian and YenangyaungI observedfive terraces, three of which
are built of various gravels suggesting three major stages
assemblage of definite Lower Pleistocene affinities. of aggradation. The topmost is built of bouldery red
This fauna in many cases lies within a limited stratum gravels, forming the oldest river drift, and lies 350 feet
or bone bed, which, though in the upper part of the abovethe presentstream-an indicationof a stage of lateral
Irrawaddy sequence as a whole, still lies some distance erosion that succeededa period of great water supply and
below the top of the series. of laterite formation. As in the Narbadaregion, no later-
ite is being formedhere at present,becauseof scanty rain-
After the deposition of the Irrawaddy series there fall and prolonged dry seasons. Hence we conclude that
was evidently a period of uplift, during which a certain the oldest river gravel representsa pluvial corresponding
amount of tilting and erosion of the Irrawaddies took to the great second pluvial of India. The next younger
place. As a result of this phase of geological history terrace is also built of red gravel, which fills the valley to
in Burma there was formed the first and highest of the a depth of 90 feet above stream level. This is the major
terrace gravel, and in it were found vertebratefossils that
several river terraces along the Irrawaddy drainage sys- are characteristicof the Middle Pleistocene beds of the
tem. This is T1, as designated by de Terra and Teil- NarbadaValley. Also, the early Paleolithictools found in
hard de Chardin, probably formed by the degradation both these formationsbelong to all appearancesto the same
of a formerly aggraded "boulder gravel"-this latter cultural level. Of special significancehere is the associa-
tion of "Red Earth,"a clayey, weatheringproductcontain-
perhaps an equivalent of the well-known Boulder Con- ing iron-stainedconcretions. It is found also on top of
glomerate of India. It is an interesting fact, accord- lateritic deposits in the hills, whence it was washed down
ing to the observations of de Terra and Teilhard, that to the second terrace. The third terrace is of very wide
there are no thick boulder fans in the Irrawaddy Valley extent. As in northern India, the third gravel underlies
terrace 4, and it also is associated with a special soil-
comparable to the Boulder Conglomerate of northern Pagan silt, reminiscentof the Potwar loess in India. But
India. Therefore it would seem that if such gravels whereas the Potwar loess dates from the third glacial, the
were at one time present, as is probable, they were Pagan silt clearly belongs to a younger stage of aggrada-
subsequently degraded to form the high terrace, T1. tion and wind deposition, which we correlate with the
fourth pluvial. The fifth terrace, finally, is composedof
Consequently T1 is to be regarded as nearly equivalent silt and clay, and, as in India, its thicknessin places is con-
to, but slightly later than, the Boulder Conglomerate siderable.
of northern India, perhaps equivalent in part to the
Narbada beds of central India and to the Mogok fis- In the Shan Plateau province, near Mogok, are nu-
sure deposits, described below. T1 in India is con- merous fissures due to karst development, and these
sidered as of late Middle Pleistocene age because it is contain deposits in which are found remains of fossil
later than the Upper Siwaliks and the Boulder Con- mammals. The Mogok fissure deposits are regarded
glomerate-the latter belonging to the second glaciation. by Teilhard as equivalent in age to the fissure de-
Following the high terrace are the two successively posits of south China, found in Yunnan, Kwangsi, and
lower terraces, designated by de Terra as T. and T3 Szechwan, and specifically to those in the Choukoutien
respectively. These terraces are also of Pleistocene region of north China. This would make the Burma
age. In the third terrace, T., are found numerous cave deposits of general Middle Pleistocene age. Need-
vertebrate fossils, mostly mammals, but it is important less to say, if these cave deposits are correlative with
COLBERT: PLEISTOCENE VERTEBRATES 397
the Pleistocene cave deposits of China, they are of the
utmost importance, for they connect directly the human Irrawaddy River Shan Plateau
prehistory of Burma with that of the definitive Chinese Valley
area.
Post-Pleistocene Post-Terrace fossils Fossils from super-
Dr. P. Teilhard de Chardin . . . has diagnosed the fis- ficial cave deposits
sure fauna [at Mogok] as Middle Pleistocene, the same age
as the Stegodon-Orang-fauna of south China. Since the Upper Pleistocene Terrace Deposits,
latter presumably represents the Choukoutien stage, it be- fossils in situ (also
came imperative to investigate its relationship to the Ir- rewashed Upper
rawaddy Quaternary. [Meaning here, the Quaternary of Irrawaddy fossils)
the Irrawaddy River, not the Upper Irrawaddy beds.]
Our joint excursion from Mogok resulted in a direct Middle Pleistocene Terrace Deposits, Mogok Caves
correlation of the fissure formation with the boulder fans fossils in situ (also
which veil the hillslopes near the caverns. The former rewashed Upper
apparently merge with the latter, and subsequently the Irrawaddy fossils)
boulder fans underwent lateritisation. This feature added
another element for our stratigraphic correlations with Lower Pleistocene Upper Irrawaddy
south China. Fissure deposit and boulder fan probably beds
belong to that same stage during which Stegodon, El. Upper Irrawaddy
namadicus and Panda roamed the highlands of Burma and fauna
south China. Judging from the nature of the sediment,
this was a period of rapid accumulation of rock debris
which evidently required a more pluvial climate and a THE PLEISTOCENE FAUNAS OF BURMA
different physiographic aspect than prevails in the Shan
Plateau nowadays. (De Terra, 1938a.) Upper Pleistocene-Terrace Deposits (Fossils in situ)

Finally, in some of the Mogok fissures or caves are Elephas hysudricus Falconer and Cautley
found superficial deposits containing mammalian re- Hemibos triqettricornis Riitimeyer
mains of a relatively recent character. These remains,
comparatively scarce, are undoubtedly post-Pleistocene Middle Pleistocene-Mogok Caves
in age, and in some cases they may be more or less cor-
relative with the very late mammalian remains found in Ailuropoda baconi (Woodward)
or below the fifth terrace of the Irrawaddy River. Stegodon orientalis (Clift)
Palaeoloxodon anamadicus(Falconer and Cautley)
From the foregoing discussion, it is evident that
Pleistocene vertebrates, particularly mammals, are Rhinoceros sp.
found in Burma under three principal sets of condi- Sus sp.
tions, namely: Cervus sp.
Bovine
1. In the heavy sandstones making up the upper por-
tion of the Irrawaddy series, along the Irrawaddy River. Hystrix sp.
2. In the terraces along the Irrawaddy River, over-
Lower Pleistocene-Irrawaddy beds
lying the Irrawaddy beds. The fossils in these terraces
may be derived, or they may be contemporaneous with Stegolophodon latidens (Clift)
the terraces and in sitiu. Stegodon elephantoides (Clift)
3. In the fissures or caves of the Shan Plateau karst Stegodon insignis birmanicus Osborn
Elephas hysudricus Falconer and Cautley
region, particularly in the vicinity of Mogok. Most of
the fissure fossils are undoubtedly of Pleistocene age. Rhinoceros sivalensis Falconer and Cautley
Hipparion cf. antelopinum (Falconer and Cautley)
In addition, scattered animal remains of post-Pleisto- Equus yunnancnsis Colbert
cene or Recent age are found in Burma under two sets Potamochoerus sp.
of conditions: Hexaprotodon iravaticls Falconer and Cautley
Hexaprotodon cf. sivalensis Falconer and Cautley
(a) In the superficial beds along the Irrawaddy Merycopotamus dissimlilis (Falconer and Cautley)
Cervus sp.
River, subsequent in age to the last Irrawaddy terrace. Hemibos triquetricornis Riitimeyer
(b) In superficial cave deposits, at Mogok. Proleptobos birmanicus Pilgrim
Bibos (?) cf. sondaicus
The relationship of these fossil assemblages in geo- Hippotragine
logic age and to the physiographic regions in which Gazella sp.
they are found may be shown in the following tabular Caprine
manner. The geologic ages here expressed are more Boselaphine
fully discussed and explained in subsequent pages, near Gavialis sp.
the end of this present contribution. Chelonian
398 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
III. COLLECTING LOCALITIES IN THE "Broken and rolled teeth of Elephas and Stegodon in
PLEISTOCENE OF BURMA recent slope wash below outcrop of terrace gravel un-
In the following pages are listed all of the localities derlying third terrace near limestone ridge.
"Remarks Possibility of mixture between Upper
at which Pleistocene vertebrate fossils were collected
Irrawaddy fossils and terrace fauna. Perhaps to be
by the members of the Southeast Asiatic Expedition. decided by wear of bones or by their patination."
The description of each locality is given verbatim, as it
was written by Dr. de Terra in a personal communica- ANSP No. 14629, Elephas hysudricus. Right DM,
tion to the author. The fossils, as identified and num- ANSP No. 14655, Elephas hysudricus. Right ramus and
bered, are listed for each locality. The localities are symph.
ANSP No. 14656, Elephas hysudricus. Tooth frags. (6)
shown on the accompanying map. The locality designa- MCZ No. 6290, Elephas hysudricus. Palate frag.
tions are those used by Drs. de Terra and Movius. MCZ No. 6293, Elephas hysudricus. Tooth frags. (7)
The following abbreviations are used in citing museum MCZ No. 6284, Stegodon elephantoides. Lower molar
numbers: frag.
MCZ No. 6285, Stegodon insignis birmanicus. Tooth
ANSP-Academy of Natural Sciences, Philadelphia. frag.
MCZ-Museum of Comparative Zoology at Harvard MCZ No. 6286, Stegodon insignis birmanicus. Tooth
ANSP No. 14634, Merycopotamuitsdissimilis. Cranium
College, Cambridge.
AMNH-American Museum of Natural History, New
MINGUN, IRRAWADDY
RIVER. LOCALITYMA. 4
York.
"112 furlongs from fault escarpment on limestone
LOWERPLEISTOCENE
ridge west of Mingun and identical with the clayey
Derived Fossils of Upper Irrawaddy Age from the bone bed mentioned under Ma. 3. 11/1 miles from
Irrawaddy River Terraces Mingun, on the road to Padu."
RIVER. LOCALITYPAGAN
PAGAN, IRRAWADDY MCZ No. 6294, Elephas hysudrictus. Tooth frag.
ANSP No. 14639, Cervid (?). Antler frag. (?)
"Famous temple city in Upper Burma on the Ir-
rawaddy south of Mandalay. RIVER. LOCALITYMA. 5
MINGUN, IRRAWADDY
"Fossils apparently all derived from the Upper Ir-
rawaddies. Their state of incrustation and waterwear "11/4 miles south of Ma. 4 in same bone bed of Upper
and their occurrence in conglomeratic layers suggest Irrawaddy age.
that they were swept together by floods of the ancient "To this number belong some teeth of Equus which
Irrawaddy. All specimens except a few collected from I found myself in sitt in brown clays of Upper Ir-
hillslopes. Beds are sandy throughout with conglomer- rawaddy type. The section here was well exposed and
ate layers. studied in some detail. The ossiferous clays make a
"Remarks: In the Upper Irrawaddies a so-called regular bone bed below the sandy and conglomeratic
'fauna' may be composed of types that lived at the upper portion of this zone. Reminiscent of Pinjor-
time of the sediment and of others that died and be- Tatrot clays of India but less bright in color."
came fossilized prior to this time-their remains hav-
ANSP No. 14632, Equus sp. Left [M2]
ing been washed out by the meandering river at a sub- MCZ No. 6258, Equus sp. Right [M2]; radius
sequent period. This goes for all Upper Irrawaddy
and Upper Siwalik sediments. Hence one may ex- RIVER. LOCALITYMA. 9
MINGUN, IRRAWADDY
pect a mixture of late Pliocene and early Pleistocene
"Bone Bed in Upper Irrawaddy clays. Various out-
types in a similar way as Hopwood suggested for the
Eoanthropus gravels of the Thames Valley." crops northeast of Ma. 8, 31/2 miles southwest of
Mingun. 1/2 mile south of Letpan."
ANSP No. 14623, Stegodon insignis birmanicus. Frag.
of ramus with RM3 AMCZ No. 6275, Rhinoceros sivaleusis. Molar frag.
ANSP No. 14652, Stegodoninsignis birmanicus. Left ra- MCZ No. 6274, Hexaprotodon sp. Frag. left lower ca-
mus frag. nine
ANSP No. 14653, Stegodoninsignis birmanicus. Left ra- ANSP No. 14643, Cervid. Base of antler
mus with M2and symph. MCZ No. 6271, Cervid (?). Antler frags.
ANSP No. 14650, Stegodon elephantoides. U. molar ANSP No. 14638, Gazella sp. Horn core
frags. (2) MCZ No. 6263, Hemibos triqutetricornis. Two upper
MCZ No. 6283, Stegodon elephantoides. U. molar molars
frags. (3) MCZ No. 6310, Hemibos triqulctricornis. Distal end
ANSP No. 14630, Rhinoceros sivalensis. Left P4 of metapodial
MCZ No. 6276, Bibos (?), cf. sondaicus. Teeth
RIVER. LOCALITYMA. 3
MINGUN, IRRAWADDY ANSP No. 14659, Elephas hysudricus. Tooth frag.
"On ridge 1/2 mile southwest of Tanmyin village. ANSP No. 14660, Elephas hysudricus. Tooth frag.
Collected and purchased from villagers. Derived from
MCZ No. 6295, Elephas hysudricus. Tooth frag.
MCZ No. 6297. Elephas hysudricus. Tooth frags. (2)
Upper Irrawaddies (sandy grit and conglomerate). MCZ No. 6304, Gavialis sp. Vert.
Legend ++++ ++++'++
alluvium +
,,: Recent _KATHEoIA0^

Pliocene and Pleis+o- ++++++


+
cenre-Irrawaddy ! + /+
Lower and rriddlei +
Ter+iary-Pequ u[ ++

v v^ Recen + and \, +|
upper
v Ter."iary volcanic rocK,

,cryS+all1 es ,' R I,-


EI / ?KY

PAUK PAD^UN6 -" *? l


<
^V^e----
Di g?? ?NUK . ?\
KYAUKPADAUN?-

/~~ : ~? I - 1e,? ? ?

-- .
J'.'-^^^^YENANGY^UNy^^^

I ' tiMAGWE '

Scale I t i M iles
0 52.
FIG. 79. Geological map of a portion of Upper Burma, showing localities at which fossils were collected
by the American Southeast Asiatic Expedition.
399
400 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
RIVER. LOCALITYMINGUN TER-
MINGUN, IRRAWADDY calated with a thick clay series presumably marking a
RACE lower portion of the Upper Irrawaddies. Lithological
similarities with Pinjor zone of Punjab Upper Siwaliks
"Third Terrace gravel."
rather striking."
ANSP No. 14654, Elephas hysudricus. Left ramus
MCZ No. 6291, ANSP No. 14649, Stegodon elephantoides. Left ramus
Elephas hysudricus. Ramus
MCZ No. 6292, Elephas hysudricus. Ramus frag. (erupting molar)
ANSP No. 14663, Proboscidean. Ramus frag. MCZ. No. 6289, Stegodon sp. Tooth frag.
MCZ No. 6264, Bibos (?), cf. sondaicus. Mand. frag. MCZ No. 6315, Hexaprotodon iravaticus. Frag. of
mandible
Right M2-3 MCZ No. 6316, Hexaprotodon iravaticus. Frag. of
mandible
MAGWE, IRRAWADDY
RIVER. LOCALITYMAGWE U. I. ANSP No. 14717, Boselaphine (?). Tooth
MCZ No. 6266, Caprine (?). Right horn core
"Various localities along bank of Irrawaddy up- ANSP No. 14642, Bibos (?), cf. sondaicus. Teeth
stream from Magwe. Cliffs along river in Upper Ir- MCZ No. 6265, Hemibos triquetricornis. Back of skull
rawaddy sands." MCZ No. 6311, Hemibos triquetricornis. Frag. meta-
podial
ANSP No. 14622, Stegodon elephantoides. Left M2 MCZ No. 6312, Hemiibos triquetricornis. Frag. meta-
ANSP No. 14624, Stegodon elephantoides. Frag. of ra- podial
mus, Right DM3-4 MCZ No. 6269, Hippotragine (?). Horn core
MCZ No. 6251, Stegodon elephantoides. Left M' MCZ No. 6272, Hippotragine (?). Frag. horn core
MCZ No. 6254, Stegodon elephantoides. Ramus, Right MCZ No. 6314, Cervid (?). Astragalus
M2 ANSP No. 14664, Gavialis sp. Ramus frag.
MCZ No. 6267, Stegodon elephantoides. U. molar MCZ No. 6305, Chelonian. Shell frags.
MCZ No. 6282, Stegodon elephantoides. Le ft ramus, ANSP No. 14644, Chelonian. Part of Plastron
molar frag.
MCZ No. 6253, Stegodon insignis birmanicus. Left
molar SEIKPYU, YAW VALLEY. LOCALITY LOWER YAW
MCZ No. 6287, Stegodon sp. Right ramus, DM VALLEY
MCZ No. 6298, Elephas hysudricus. Tusk frag.
ANSP No. 14662, Proboscidean. Ramus frag. "Opposite Chauk at Seikpyu, about 21/%miles up-
MCZ No. 6308, Rhinoceros sivalensis. Ectoloph Left stream. Bone bed in gravelly sandstone of Upper Ir-
M. rawaddy age ( ?)."
ANSP No. 14633, Potamochoerus sp. Frag. of Ramus
with Right M2 MCZ No. 6288, Stegodon sp. Tooth frag.
MCZ No. 6259, Hexaprotodon iravaticus. Right M2 ANSP No. 14640, Caprine. Frag. horn core
ANSP No. 14637, Hemibos triquetricornis. Mand. frag. ANSP No. 14666, Gavialis sp. Scute
Right M1-3 ANSP No. 14667, Gavialis sp. Vertebra
MCZ No. 6260, Hemibos triquetricornis. Portion of ANSP No. 14670, Chelonian. Frags. of carapace
skull
MCZ No. 6262, Hemibos triquetricornis. U. molar
MCZ No. 6313, Hemibos triquetricornis. Atlas verte- CHAUK, IRRAWADDY
RIVER. LOCALITYIRR. CHAUK
bra "On left bank of river in Upper, possibly ? Middle
MCZ No. 6270, Hippotragine (?). Horn core
MCZ No. 6273, Bibos (?), cf. sondaicus. Horn core Irrawaddy Series."
ANSP No. 14668, Gavialis sp. Vertebra ANSP No. 14621, Stegolophodon latidens. Left lower
ANSP No. 14669, Gavialis sp. Tooth molar
MCZ No. 6301, Gavialis sp. Frag. of symphysis
MCZ No. 6306, Chelonian. Frags. of shell
PAUK, PAKOKKU DISTRICT, UPPER IRRAWADDYBEDS,
PYINCHAUNG,YAW VALLEY. LOCALITYYAW VALLEY 491 feet above sea-level.
"A side valley of the Irrawaddy with its junction op- ANSP No. 14631, Hipparion cf. antelopinum. Left P2
posite Chauk. Fossils collected in Upper Irrawaddies
on left bank of river near Pyinchaung. "[The Yenangyaung localities] all are similar, con-
"A single tooth of Bos from a younger tufa deposit at sisting of remnants of Terrace 3, which is the main
mile stone 37, furlong 2. Beneath this place was found terrace in the Irrawaddy sequence. The localities are
a broken jaw of Stegodon in Upper Irrawaddy gravels. all places where we found human implements either in
All others collected in dissected western flank of Thagyi the terrace gravels or in derived position on the dissected
Hills, an anticline of Upper Irrawaddy rocks west of slopes along the Irrawaddy. It is, therefore, possible
Pakkoku. I have noted: Stegodon-tooth, Rhino-tooth that some of the fossils, which are all surface finds, have
fragment, antelope broken horn core, Hexaprotodon jaw been derived from older beds into which the terraces
fragment, turtle bones. have been cut. Unfortunately I did not find one single
"Remarks: Searched for two days for a decent bone fossil in situ in the terrace directly associated with a
bed but none was found despite excellent exposures. human artifact. In other words I cannot definitely
Fossils chiefly in ferruginous sandstone which is inter- state that any of the palaeontological material which I
COLBERT: PLEISTOCENE VERTEBRATES 401
collected from this region comes from the implementif- Fossils from Mogok Fissures
erous horizon in the basal gravels of Terrace 3." (H.
MOGOK, UPPER BURMA. LOCALITY MOGOK C1.
L. Movius, Jr., letter to author, 1939.)
"Dato Cave ("Mercury" cave) 11%hours' climb from
YENANGYAUNG, IRRAWADDYRIVER. LOCALITYYG. 4 Leu village near Mogok (Upper Burma). The locality
is not identical with the one mentioned by Smith-
"? 3 miles south of Yenangyaung, between Ny- Woodward in his description of Aelureidopus baconi but
aunghla and Sadaing." lies some 300 feet below Col. Bacon's cave on the same
MCZ No. 6302, Gavialis sp. Ramus frag., vert. ridge. Cave was visited several times in the course of
MCZ No. 6312, Hemibos triquetricornis. Distal end of our two weeks' stay at Mogok. Bones were collected
metapodial from ossiferous sand beneath cave loam-Stegodon,
ANSP No. 14651, Stegodonelephantoides. U. molar frag. Cervus,
Sus, Bos, Rhinoceros, rodent. All these are
represented by single teeth or antler fragments. Some
YENANGYAUNG,IRRAWADDYRIVER. LOCALITYYG. 8 bones show effects of gnawing by rodents.
"? 4 miles south of Yenangyaung, and immediately "Remarks: If one can assume that Stegodon became
south of Sadaing." extinct during the Middle Pleistocene in this region
then the cave fauna must be of that age. Villagers re-
MCZ No. 6307, Chelonian. Shell frag. that large skeletons had been seen by them in
ANSP No. 14635, Hemibos triquetricornis. Portion of ported caves but they had buried them for fear the
cranium neighboring
dragon demons might haunt their village. The occur-
rence of partial skeletons of Panda, Elephas namadicus
YENANGYAUNG,IRRAWADDYRIVER. LOCALITYYG. 9 and Rhinoceros in these caves is indicated by the find of
"? 5 miles south of Yenangyaung, and immediately a Panda skull by Col. Bacon and also by the fact that
north of Yonzeik. (Locality 9 is just south of Locality villagers brought us large limb bones from sink-hole
8.)" deposits nearby."
MCZ No. 6261, Hemibos triquetricornis. Frag. horn MCZ No. 6279, Sus sp. Right M3
core ANSP No. 14647, Cervus sp. Antler frags. (2)
ANSP No. 14648, Medium-sizedruminant. Right upper
YENANGYAUNG,IRRAWADDY RIVER. LOCALITYYG. 13 MCZ M; Right Ms; Left Me. III-IV; Right ast.
No. 6280, Medium-sized ruminant. Right lower
"%1mile south of Yenangyaung, between Yenang- third molar; Right ast.; frags.
yaung and Thittabwe." MOGOK,UPPER BURMA. LOCALITYMOGOKC2
ANSP No. 14641, Hexaprotodon. Left half of symph.
MCZ No. 6303, Gavialis. Ramus frag. "Cave near Kathe village (Mogok district) at a ham-
let called Chausong. It is situated 2 miles north-
MIDDLE PLEISTOCENE
northeast of Kathe village which itself is 9 miles west of
Mogok. A villager from Chausong brought teeth of:
Fossils in situ in Terrace T3 "Stegodon, Bos, Elephas namadicus, and a piece of
bone breccia which supposedly was found at the en-
MINGUN, IRRAWADDY
RIVER. LOCALITYMA. 3*.
trance. I verified this statement at a later occasion and
"Rolled molar of Elephas was found 10 feet below my assistant entered the deepest parts of the cave where
edge of terrace III. Derived from bouldery gravel he found at a depth of about 80 feet a pseudo-artifact
at base of terrace and should be Middle or Upper of brown patinated flint. I saw the main cave room
Pleistocene." and found that gem-diggers had washed its filling for
rubies and other precious stones. At the inner end of
ANSP No. 14657, Elephas hysudricus. Tooth frag.
the cave entrance I found a bone breccia but since it was
IRRAWADDY RIVER. LOCALITY MA. 8*
only a small remnant I did not start with excavation.
MINGUN, It contained ribs of smaller mammals and broken leg
TERRACE bones of artiodactyls. Here also coarse sand with peb-
"On terrace III, 1/2 mile south of Letpan village bles underlies the cave loam.
downstream from Mingun and 150 feet above river. "Remarks: Chausong is one of 14 caves which we
Brown stained gravel with Elephas molar fragment, searched. Most common fossil in these are molars of
teeth of Bos in a jaw fragment. Certainly from basal immature Stegodon and Elephas, Rhino, and deer ant-
gravel below third terrace." lers. Of the first genus there seems to be a selection of
immature types. No beasts of prey were found. Ani-
ANSP No. 14658, Elephas hysudricus. Tooth frag. mals may have been trapped in superficial pits of the
MCZ No. 6296, Elephas hysudricus. Tooth frag.
ANSP No. 14636, Hemibos triquetricornis. Max. frag. karst and their remains subsequently washed down by
Right M'3 underground water."
402 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
ANSP No. 14646, Hystrix sp. Left P4 part of Kengtung State. Apparently derived from a
ANSP No. 14625, Stegodon orientalis. Right DM4 cave as at Mogok.
ANSP No. 14626, Stegodon orientalis. Left DM4 "I obtained this material from Dr. J. H. Telford, Di-
MCZ No. 6252, Stegodon orientalis. Right DM4
MCZ No. 6277, Stegodon orientalis. Tooth fragments rector of the American Baptist Mission (Pangwe, near
ANSP No. 14627, Palaeoloxodon namadicus. Right M2 Kengtung), in the far east of Burma, near the Yunnan,
MCZ No. 6255, Palaeoloxodon namadicus. Left M2 Indo-Chinese and Siamese frontiers. He collected it at
MCZ No. 6256, Palaeoloxodon namsadicus. Left M2 various times during trips in the Nam Hkok Valley in
ANSP No. 14628, Stegodon (?) or Palaeoloxodon (?).
Tusk the south of Kengtung State, but I was unable to de-
ANSP No. 14645, Rhinoceros sp. Right P3 termine the precise locality." (H. L. Movius, Jr., letter
MCZ No. 6278, Rhinoceros sp. Right lower molar to author, 1939.)
MCZ No. 6281, Large bovid. Left P4, left M:, part of
atlas MCZ No. 6257, Eleplias sp. Left M2
MCZ No. 6268, Elephas sp. Upper molar
POST-PLEISTOCENE LOCALITY UNCERTAIN

Post-Terrace Fossils ANSP No. 14639, Bovid. Frag. horn core


ANSP No. 14661, Proboscidean. Ramus frag.
KYAUKPADAUNG, IRRAWADDY RIVER. LOCALITY LOC. MCZ No. 6299, Proboscidean. Ramus frag.
1, KY. MCZ No. 6300, Proboscidean. Limb frag.
ANSP No. 14665, Gavialis sp. Occiput frag.
"At milestone 39, furlongs 2, on road from Yenang-
yaung to Popaywa. Fossils in situ, 10 feet below sur- IV. THE PLEISTOCENE VERTEBRATE
face, in a dark bank of fossil soil overlying a volcanic FAUNAS OF BURMA
mudflow deposit.
THE UPPER IRRAWADDY FAUNA AS FOUND IN THE
"Remarks: It is important to ascertain the species
UPPER IRRAWADDY BEDS
because this is the only place known so far where verte-
brate fossils of Post-Terrace age have been located in MAMMALIA
Burma. Also there were some artifacts associated with
the bones." PROBOSCIDEA

MCZ No. 6317, Equus sp. Right M3 Stegolophodon latidens (Clift)


MCZ No. 6318, Cervts sp. Teeth, jaw frags., bones Mastodoiilatidens,Clift, 1828. Trans.Geol.Soc. London,(2),
MCZ No. 6319, Antelope ?. Teeth, jaw frags., bones II, Pt. 3, pp.371,372,P1.XXXVII, figs. 1, 4; P1.XXXVIII,
MCZ No. 6320, Bos sp. Teeth, jaw frags., bones fig. 1; P1.XXXIX, figs. 1, 2, 3.
Stegolophodon latidens,Schlesinger,1917. Dellkschr.des K. K.
"Fossils" Found in Superficial Cave Deposits Naturhist.Hofmuseums,Band I, Geol.-Pal.Reihe I, p. 115.

MOGOK, UPPER BURMA. LOCALITY MOGOK C3 Type (Lectotype).-"An upper jaw containing RM2,
RM3. Original in Geol. Soc. Coll.; Brit. Mus. cast.
"Cave-like fissure (old ruby mine) owned by Maung B. M. No. M2888-9. P1. XXXVII, fig. 1, Clift."-
Tung of Mogok. Located about 2 miles north of Colbert, 1938.
Mogok in a small valley containing European cemetery Cotypes.-Brit. Mus. No. 7391 (cast) fragments of
and 'old Police lines.' Two furlongs east of old Police tooth; No. 7394 (cast), right lower jaw. Both figured
lines. by Clift.
"In youngest cave loam a jaw of small Bos and one Horizon.-Upper Irrawaddy beds, Lower Pleistocene.
vertebra. Locality.-Irrawaddy River, in the vicinity of Ye-
"Remarks: A non-fossilized human jaw of peculiar nangyaung.
shape was found in cave detritus in front of entrance. Diagnosis.-A proboscidean intermediate between the
Bones also little fossilized. Suggestive of historic mastodonts and stegodonts, with low-crowned cheek
retreat." teeth in which the ridge crests are limited in numbers
and composed in each case of a few large, heavy cones.
ANSP No. 14620, Elephas indicus. Left M1
MCZ No. 6309, Bos. Mandible Specimens under Consideration.-ANSP No. 14621,
portion of a left lower molar. From the Upper Irra-
KENG TUNG waddy beds, near Chauk.
ANSP No. 14623, portion of right mandibular ramus
Fossils of Unknown Age with the third molar. From the Upper Irrawaddy beds,
near Pagan.
EASTERN BURMA, NEAR YUNNAN FRONTIER. LOCAL-
MCZ No. 6267, upper molar. From the Upper Irra-
ITY KENGTUNG.
waddy beds, near Magwe.
Collected by Dr. H. L. Movius, Jr., from a mission- Stegolophodon is a genus occupying a position more
ary who found the material in a valley in the southern or less intermediate between the mastodonts and the
COLBERT: PLEISTOCENE VERTEBRATES 403
true stegodonts. Consequently it is oftentimes a diffi- The worn upper molar, MCZ No. 6267, is provision-
cult form to deal with, either when one is discussing its ally referred to this species. Because of its seeming
probable larger relationships, or determining its specific true shortness and low number of ridge crests, its thick
identity on the basis of incomplete and worn fossil enamel and brachyodonty, it is considered as of the
specimens. genus Stegolophodon. However, it may be a Stegodon
The mandibular ramus listed above is identified as tooth of deceiving appearance.
belonging to the genus Stegolophodon because its con-
tained molar tooth, although much worn, would seem to Stegodon elephantoides (Clift)
show characters less advanced than those of the true Mastodon elephantoides,Clift, 1828. Trans. Geol. Soc. London,
Stegodon group. Thus the posterior portion of this (2), II, Pt. 3, pp. 372-373, PI. XXXVI, fig. 1; P1. XXXVII,
tooth indicates that the ridge crests were much more figs. 1, 2; P1. XXXVIII, fig. 2; P1. XXXIX, fig. 6.
brachyodont, and less regular than is typical of Stego- Elephas cliftii, Falconer and Cautley, 1846. "Fauna Antiqua
Sivalensis," P1. XXX, figs. 1-5 (a synonym).
don, and each crest is made up of a few very large Elephas insignis, (in parte), Falconer and Cautley, 1846.
conelets, rather than of the numerous mammillae that "Fauna Antiqua Sivalensis," PI. XX, figs. 9, 9a.
constitute the ridge crests in a Stegodon tooth. There Elephas cliftii, Falconer, 1868. Pal. Mem., I, pp. 81-82, 113-
are seven crests and a small heel, a number that would 114,461-462.
correspond with the type lower third molar of Stegolo- Lectotype.-"Left lower jaw figured by Clift, 1828,
phodon latidens, in which there are six full crests, with P1. XXXVIII, fig. 2"-Colbert, 1938.
lesser crests anteriorly and posteriorly. This low num- Cotype.-Brit. Mus. No. 7388 (cast), upper molar.
ber of crests may be contrasted with the condition in Horizon.-Upper Irradaddy beds, Lower Pleistocene.
Stegodon elephantoides, in which there are ten full Locality.-Irrawaddy River, in the vicinity of Ye-
crests in the third lower molar, and with Stegodon nangyaung.
insignis birnimanicus,possessing at least thirteen crests in Diagnosis.-A true Stegodon, closely comparable to
the same tooth. The back of the tooth in the specimen Stegodon orientalis but distinguished by the relatively
under consideration tapers markedly, so that it is rather low-crowned cheek teeth, in which each ridge crest is
pointed behind-a resemblance to Stegolophodon lati- composed of rather heavy mamillae.
dcnls and a rather distinct contrast to the abruptly termi- Specimens under Consideration.-ANSP, No. 14622,
nated tooth of Stegodon. The molar is characterized, a left upper second molar. From the Upper Irrawaddy
too, by the large marginal buccal cuspules that block the beds near Magwe.
external limits of the transverse valleys between the ANSP No. 14624, fragment of a ramus with right
ridge crests-a feature often found in Stegolophodon third and fourth deciduous molars. From the Upper
but not seen in Stegodon. The mandibularramus is not Irrawaddy beds near Magwe.
so deep nor so heavy as is the case in Stegodon from ANSP No. 14649, fragment of lower ramus (erupt-
the same beds. ing molar). From the Upper Irrawaddy beds. Lo-
The fragmentary lower tooth, No. 16421, is much cality Yaw Valley.
worn, and not very well preserved. Four worn, pos- ANSP No. 14650, two upper molar fragments.
terior ridge crests are observable-perhaps the tooth From the Upper Irrawaddy beds. Found at Pagan.
did not have any more crests than those preserved, as ANSP No. 14651, upper molar fragment. From the
might be possible in an intermediate molar of this Upper Irrawaddy beds. Found at Yenangyaung. Lo-
genus. The last ridge crest is definitely made up of cality Yg. 4.
two large conelets, one on either side of the median line MCZ No. 6251, left upper second molar. From the
of the tooth. The occlusal surface is concave. Upper Irrawaddy beds, near Magwe.
These specimens are valuable in that they give addi- MCZ No. 6254, ramus with right lower second molar.
tional evidence for the definite occurrence of Stegolo- From the Upper Irrawaddy beds, near Magwe.
phodon latidens in the Upper Irrawaddy beds of Lower MCZ No. 6282, lower ramus, molar fragment. From
Pleistocene age. the Upper Irrawaddy beds, near Magwe.

MEASUREMENTS
ANSP No. 14623, ANSP No. 14621, MCZ No. 6267,
M3and ramus Lower molar Upper molar
Length ................................. 237 mm. 134 mm. 145 mm.
Breadth . . .............................. 93 77 92
Index .................................. 39 58 63
Height ................................. 40+
Ridge crests per 100 mm.................. 3 3 312
No. of ridge crests....................... 712 4+ 6+
Depth of ramus ........................ 150 mm.
404 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
MCZ No. 6283, three upper molar fragments. From homologous as to its position in the jaw with the above-
the Upper Irrawaddy beds. Found at Pagan. described specimen.
MCZ No. 6284, lower molar fragment. From the The lower molar, No. 6254, contains seven ridge
Upper Irrawaddy beds at Mingun. Locality Ma. 3. crests, and is without doubt the second molar. It is
The specimens herein described probably offer the large and heavy, and less curved laterally than similar
best supplementary evidence as to the characters of the teeth in Stegodon orientalis. The fragment of ramus
genotypic species that has been secured since the original containing it is very strong and heavy.
discovery and naming of this form, more than a hundred The juvenile mandible containing milk teeth is an
years ago. It will be remembered that Stegodon ele- exceptionally fine specimen-the first published record
phantoides was based by Clift upon two specimens, a of a milk dentition in this species. The third and
mandibular ramus containing a complete third molar, fourth deciduous molars are present, the former con-
which subsequently has been chosen as the lectotype, taining four crests and a large heel, the latter with six
and an upper molar tooth. crests and a heel. In these teeth the enamel, particu-
The original lower molar is rather long, with rela- larly on the side of the teeth, is strongly rugose, and
tively low ridge crests, and there are ten of these com- each ridge crest consists of about eight principal conelets.
posing the tooth. The upper molar, probably a second On the basis of these specimens, together with the in-
molar, is made up of six ridge crests. formation already extant as to this species, the ridge
In the new material at hand, the two upper teeth are formulae for Stegodon elephantoidcs may be written
very well preserved and fairly complete. One of these tentatively as:
specimens, ANSP No. 14622, is large and broad, with
a strongly convex occlusal surface. It is made up of DM2-, DM3 4/4, DM4 61/2/6 1/, Ml 6 1/2/6 1/2 M2
seven well-developed ridge crests, none of which has 6-7/7, A3 8-9/10.

MEASUREMENTS

Stegodon elephantoides
a) b) c) d)
ANSP No. 14622 MCZ No. 6251 MCZ No. 6254 ANSP No. 14624 d)
M3 M3 M2 DM3 DM4
Length .................... 242 mm 1. 226+ mm. 231 mm. 60 mm. 119 mm.
Breadth ................... 111 108 96 45 58
Index ..................... 46 42 75 49
Height (4th crest) .......... 53 51 55 22
Ridge crests ............... 7 6+ 7 412 /2 6/2
Crests per 100 mm.......... 2'2 212 3 6

been touched by wear. These crests are rather widely Stegodon insignis birmanicus Osborn
spaced, and each is composed of about eight very heavy, Stegodon insignis birmanicus,Osborn, 1929. Amer. Mus. Novi-
robust conelets-a distinct contrast to the more numer- tates, No. 393, pp. 15, 16, fig. 16.
ous and much more delicate conelets of other stegodonts,
such as S. orientalis. It appears as though a partially Type.-Amer. Mus. No. 20002, a mandible, of which
the left ramus and molar are complete.
formed crest may have been contiguous to the last plate
Horizon.-Upper Irrawaddy beds, Pleistocene.
of this tooth, so that the ridge crest formula was very
Locality.-Mingun, opposite Mandalay, Burma.
probably eight-instead of seven. Diagnosis.-A large Stegodon with very massive
Is this a second or a third upper molar? From its mandible and elongated molar teeth. "The jaw is more
strongly convex occlusal surface, one might suspect it massive and the inferior grinding teeth surpass in length
of being the last tooth in the series, in which case it measurement those of any other stegodont type known."
might be thought of as having opposed a longer third (Osborn, 1929.) Conelets of teeth stout, cement pres-
lower molar, containing more ridge crests. On the ent along the length of each tooth. About thirteen
other hand, there is a strong possibility that the second ridge crests in the third inferior molar.
upper molar in this species might have had seven or Specilmens under Consideration.-ANSP No. 14652,
eight crests, and the third molar ten. A comparison fragment of left ramus. From the Upper Irrawaddy
with other related species would favor the former beds, near Pagan.
interpretation. ANSP No. 14653, left ramus with M2, and symphy-
Another tooth, MCZ No. 6251, is similar to the molar sis. From the Upper Irrawaddy beds, near Pagan.
just described except that its occlusal surface is not so MCZ No. 6253, fragment of a lower molar tooth.
strongly convex. Six ridge crests are present, but the From the Upper Irrawaddy beds, near Magwe.
front of the tooth is broken, so there might have been MCZ No. 6285, tooth fragment. From the Upper
an additional crest, or two. This tooth is probably Irrawaddy beds, Mingun. Locality Ma. 3.
COLBERT: PLEISTOCENE VERTEBRATES 405
MCZ No. 6286, lower molar. From the Upper Ir- MCZ No. 6288, tooth fragment. From the Upper
rawaddy beds. Found at Mingun. Locality Ma. 3. Irrawaddy beds. Found at Seikpyu, Yaw Valley.
In this present Burma collection, the Stegodon jaw, MCZ No. 6289, tooth fragment. From the Upper
ANSP No. 14653, is perhaps one of the best specimens Irrawaddy beds in the Yaw Valley.
to be collected. This jaw consists of a complete man-
dibular ramus of the left side, with the symphyseal re- Elephas hysudricus Falconer and Cautley
gion well preserved and a molar tooth in place. The
molar, since it has ten ridge crests and is rather blunt Elephas hysudricus, Falconer and Cautley, 1846. "Fauna An-
tiqua Sivalensis," P1. I, fig. 3; Pl. IV; PI. V; Pl. VI, figs.
at its posterior end, is identified as the second member 1--3; P1. VII; P1. VIII. Letterpress, p. 41.
of the molar series. This tooth is to be compared with Euelephas hysudricus, Falconer, 1857. Quar. Jour. Geol. Soc.
the third molar in other specimens (notably Amer. Mus. London, XIII, Pt. 4, pp. 315, 317-318.
No. 20002) in which there are thirteen or more ridge Elephas meridionalis, Pohlig, 1884. Sitzung. Niederrhein Gesell-
schaft, Bonn, pp. 47-61.
crests, with the tooth tapering somewhat at its posterior
Hypselephas hysudricus, Osborn, 1936. "Proboscidea," I, p. 12
end. The tooth is elongated, as is characteristic of this (Special Publ., Amer. Mus. Nat. Hist.).
species of Stegodon, and owing to its curvature to fit the
horizontal ramus, is concave externally along the long Cotypes.-"The numerous specimens figured by Fal-
axis of the tooth. Each individual crest is mammillated coner and Cautley in the above-cited plates of the Fauna
so that there are about eight to eleven conelets making Antiqua Sivalensis"-Colbert, 1938.
up the ridge. Horizon.-Upper Siwaliks, India. Upper Irrawad-
The ramus is moderately deep, perhaps not so deep dies, Burma. Lower Pleistocene.
in the symphyseal region as the ramus of the American Locality.-Siwalik Hills and adjacent regions.
Museum specimen cited above. The symphysis is long, Northern Burma, along the Irrawaddy River and its
protruding far beyond the anterior border of the molar, tributaries.
and it shows a deep trough dorsally, for the reception Diagnosis-An Elephas similar to the modern Asiatic
of the anterior part of the tongue. There are three elephant in size and structure. Ridge formula of mo-
openings for the mental foramen below and in front of lars, 9-12, 10-12, 13-17/9-12, 12-13, 14-18. Ridge
the anterior border of the molar. The ascending ramus plates tall and rather thin; enamel of medium thick-
is broad. ness, and plicated.
MCZ No. 6253, consisting of the two most posterior Specimens under Consideration.-ANSP No. 14629,
crests of a lower molar, is referred to Stegodon insignis a right lower fourth milk molar. From the Upper Ir-
birmanicus, because of its general resemblance to the rawaddy beds, at Mingun. Locality Ma. 3.
type. The conelets making up each crest are large and ANSP No. 14654, left ramus. From the Upper Ir-
robust, as compared with the numerous small mamillae rawaddy beds on the Mingun terrace.
that constitute the ridge crests in Stegodon elephan- ANSP No. 14655, right ramus and symphysis.
toides. From the Upper Irrawaddy beds, at Mingun. Lo-
Comparative measurements are given below. cality Ma. 3.
MEASUREMENTS
ANSP No. 14653 Amer. Mus. No. 20002 Amer. Mus. No. 20001
Length of jaw, condyle-symphysis
(diagonal) ................. 655 mm.
Length of jaw, condyle-symphysis
(parallel to lower border) .... 575
Height of jaw, lower border-con-
dyle ...................... 395
Depth of ramus, near middle..... 137 222 mm. 189 mm.
Width (thickness) of ramus...... 195 160
Breadth of ascending ramus...... 161
Length of molar................ (M2) 229 (M3) 359
Breadth of molar............... 81 98 (M3) 99
Height at sixth crest............ 72 49
Ridge crests per 100 mm......... 5 32 33A

Stegodon sp. ANSP No. 14656, six tooth fragments. From the
The following specimens of Stegodon from the Up- Upper Irrawaddy beds, at Mingun. Locality Ma. 3.
ANSP No. 14659, tooth fragment. From the Upper
per Irrawaddies were not regarded as sufficiently well
preserved to be specifically identifiable. Irrawaddy beds. Found at Mingun. Locality Ma. 9.
MCZ No. 6287, right ramus, DM. From the Upper ANSP No. 14660, tooth fragment. From the Upper
Irrawaddy beds, near Magwe. Irrawaddy beds. Found at Mingun. Locality Ma. 9.
406 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
MCZ No. 6290, fragment of palate. From the Upper ANSP No. 14663, fragment of ramus. From the
Irrawaddy beds, at Mingun. Locality Ma. 3. Upper Irrawaddy beds, on Mingun terrace.
MCZ No. 6291, ramus. From the Upper Irrawaddy MCZ No. 6299, fragment of ramus. Of unknown
beds. Found on the Mingun terrace. age. Locality uncertain.
MCZ No. 6292, ramus. From the Upper Irrawaddy MCZ No. 6300, limb fragment. Of unknown age.
beds. Found on the Mingun terrace. Locality uncertain.
MCZ No. 6293, seven tooth fragments. From the
Upper Irrawaddy beds, at Mingun. Locality Ma. 3. PERISSODACTYLA
MCZ No. 6294, tooth fragment. From the Upper
Irrawaddy beds, at Mingun. Locality Ma. 4. Rhinoceros sivalensis Falconer and Cautley
MCZ No. 6295, tooth fragment. From the Upper Rhinoceros sivalensis, Falconer and Cautley, 1847. Fauna An-
Irrawaddy beds. Found at Mingun. Locality Ma. 9. tiqua Sivalensis, P1. LXXIII, figs. 2, 3; P1. LXXIV, figs.
MCZ No. 6297, two tooth .fragments. From the Up- 5, 6; P1. XXV, figs. 5, 6.
per Irrawaddy beds. Found at Mingun. Locality Ma. Type (Lectotype).-Brit. Mus. No. 39626, part of a
9.
skull, from the Upper Siwalik beds in the Siwalik Hills.
MCZ No. 6298, fragment of tusk. From the Upper
Cotypes.-Brit. Mus. Nos. 39325, a skull; 39646, a
Irrawaddy beds, near Magwe. mandibular symphysis; 39647, portion of a skull.
The small milk molar, ANSP No. 14629, is referable
to Elephas hysudricus, typical of the Upper Siwalik
beds in India, and of the correlative Upper Irrawaddies
of Burma. It is characterized by its thin, crinkled
enamel. There are seven full ridge plates.

MEASUREMENTS
ANSP No. 14629
Length ................. ........... . 94 mm.
W idth ................... ................ 53
In dex . .. . . ............................... 56
No. of ridge plates ......................... 2 71

The palate, MCZ No. 6290, shows portions of the


right and left molars, having characters typical of
Elephas hysudricus. The portion of the lower jaw,
ANSP No. 14654, is very heavy, with an extremely
thick horizontal ramus. The molar, broken at the
front, contains fifteen plates-probably three plates are FIG. 80. Rhinoceros sivalensis Falconer and Cautlev. ANSP
No. 14630, left P4. Crownview, natural size.
missing. The enamel is rather heavy and crenulated.
MEASUREMENTS Horizon.-Upper Siwaliks, Lower Pleistocene. Also
Lower jaw ANSP No. 14654 Upper Irrawaddies.
Depth of ramus ......................... 167 mm.
Thickness of ramus ...................... 180 Locality.-Siwalik Hills, northern India. Irrawaddy
Length of M3 (as preserved) .............. 245 River, Burma.
Length of complete tooth (estimated) ...... 284 Diagnosis.-A large species, with a strong parastyle
Breadth of tooth, anteriorly .............. 83 buttress on the upper molars, a distinct crochet but no
Ridge plates per 100 mm .................. 6 2-7 crista.
Palate MCZ No. 6290
Width of molar .......................... 92.5 mm. Specimens lunder Consideration.-ANSP No. 14630,
Ridge plates per 100 mm .................. 62-7 a left upper fourth premolar. From the Upper Ir-
Left M3 MCZ No. 6257 rawaddy beds at Pagan.
Length ................................. 204 mm . MCZ No. 6275, ectoloph of a right upper molar.
W idth .................................. 67
126
From the Upper Irrawaddy beds, Mingun. Locality
Height .................................
No. of ridge plates ....................... +14 Ma. 9.
Ridge plates per 100 mm.................. 6 MCZ No. 6308, ectoloph of a left upper molar.
From the Upper Irrawaddy beds, near Magwe.
Proboscideans indet.
This (no. 14630) is a characteristic Rhinoceros
ANSP No. 14661, fragment of ramus. Of unknown tooth, with a strong parastyle buttress, a well-developed
age. Locality uncertain. crochet reaching almost to the protoloph at its base and
ANSP No. 14662, fragment of ramus. From the a strong anterior cingulum. There is no crista nor is
Upper Irrawaddy beds, near Magwe. there an antecrochet. Of rather small size, especially
COLBERT: PLEISTOCENE VERTEBRATES 407
for Rhinoceros sivalensis, but this might be expected Specimens under Consideration.-ANSP No. 14631,
within the limits of individual variation. This tooth is a left upper second premolar. From the Pleistocene
closely comparable to the upper premolar of Rhinoceros gravels at Pauk. Probably rewashed from beds of
sivalensis figured by Lydekker (1881, P1. V, fig. 6) Lower Irrawaddy age.
except that the crochet is single rather than double. A typical Hipparion, undoubtedly redeposited in
In both teeth there is a strong vertical fold on the Pleistocene sediments from a lower horizon. Hip-
posterior part of the ectoloph. Comparable in size to parion has been known before from the Lower Ir-
the same tooth in Rhinoceros sinensis, but differing rawaddy beds of Burma (see Colbert, 1938, p. 402).
from this latter form by reason of the fold just men-
tioned. Equus yunnanensis Colbert
MEASUREMENTS Equus yunnanensis, Colbert, 1940. Amer. Mus. Novitates, No.
ANSP No. 14630 1099, pp. 1-4.
Length ................................... 43.5 mm .
W idth .................................... 58
Type.-Amer. Mus. No. 38960, nine associated up-
per cheek teeth.
Lydekker first described fossil rhinoceros remains Paratypes.-Amer. Mus. Nos. 38961-38964, inclu-
from Burma under the name of Rhinoceros iravaticus. sive. Upper and lower dentitions and teeth.
The specimens so named were subsequently regarded Horizon.-Lower Pleistocene.
by this author to be more properly included in the spe- Locality.-Makai Valley, Yunnan, China.
cies Aceratherium perimense, and were thereby shifted, Diagnosis.-A rather small equid, characterized in
the name R. iravaticus being regarded as synonymous the dentition by the moderately elongated protocone,
with A. perimense. If such is the case, it would seem the well-developed pli caballin and the crenulated
enamel of the fossette border.
probable that these specimens came from the Lower
Specimens under Consideration.-ANSP No. 14632,
Irrawaddy beds-correlative with the Dhok Pathan of a left upper cheek tooth-probably M2. From the Up-
India. This was evidently Pilgrim's opinion, for
(1910b, p. 196) he listed Aceratherium lydekkeri (to per Irrawaddy beds, at Mingun. Locality Ma. 5.
which species he assigned most of the specimens previ-
ously designated A. perimense) in the Lower Ir-
rawaddy fauna.
In this same publication Pilgrim recorded Rlhinoceros
sivalensis as from the Upper Irrawaddies of Burma-
correlative with the Upper Siwaliks of India. This ex-
tension of the range for this typical Upper Siwalik
rhinoceros was confirmed by the present author (1938),
on the basis of rather tenuous evidence. The present
find gives much stronger grounds for regarding the
Pleistocene extension of Rhinoceros sivalensis from 14631
A.N.S.R
the Siwaliks into Burma as valid.

Hipparion cf. antelopinum (Falconer and Cautley) I


Hippotherium antelopinum, Falconer and Cautley, 1849. "Fauna
Antiqua Sivalensis," Pls. LXXXII-LXXXV.
Hipparion antelopinum, Lydekker, 1885. Cat. Siw. Vert. Ind. M.C.Z.6258
Mus., pp. 57-58.
Hippodactylus antelopizlnui, Pilgrim, 1910. Rec. Geol. Surv. A.N.5.P
14632
India, XL, p. 201.

Type (Lectotype).-Brit. Mus. No. M2647, a right


maxilla with P2-M3.
Cotypes.-Brit. Mus. Nos. 16710, portion of a skull,
M2652, a mandible, M2653, a mandible, M2648, frag-
ment of an upper molar; also various limb bones.
Horizon.-Middle Siwaliks, Pliocene. Also Lower
Irrawaddies.
Locality.-Siwalik Hills and Salt Range, India. Ir- FIG. 81. Equid teeth from the Upper Irrawaddy beds.
rawaddy River, Burma. Hipparion cf. antelopinum (Falconer and Cautley). Left, above.
ANSP No. 14631, left P2. Crown view, natural size.-Equus
Diagnosis.-A large species with oval protocone, and
crenulated fossette borders in upper cheek teeth. yunnanensis Colbert. Left, below, and right. MCZ No. 6258,
right M2. Crown view, natural size. ANSP No. 14632, left M2.
Slender median metapodials in feet. External lateral view above, crown view below; natural size.
408 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
MCZ No. 6258, a right upper cheek tooth-probably ARTIODACTYLA
M2. Associated is a badly-worn distal portion of a
Potamochoerus sp.
radius. From the Upper Irrawaddy beds at Mingun.
Locality Ma. 5. Specimen under Consideration.-ANSP No. 14633,
These teeth, definitely of Upper Irrawaddy age, con- a fragment of the right mandibular ramus, containing
stitute the first valid record of Equus in the Pleistocene M2. From the Upper Irrawaddy beds, near Magwe.
of Burma. They belong to a medium-sized equid, with
rather square molar crowns, a protocone of medium
length, a small but distinct pli caballin and fairly com-
plex folding of the enamel on the fossette borders.
Thus these teeth show resemblances to the teeth in cer-
tain Pleistocene horses of China and North America,
those equids distinguished by a protocone of medium
length and folded enamel, and are distinct from the
more primitive horses of basal Pleistocene age with a
round to oval protocone and rather simple enamel,
exemplified in Europe by Equus stenonis. It is an in-
teresting fact that the two teeth now under considera- FIG. 82. Potamochoerus sp. ANSP No. 14633, right M2.
tion are very close to similar teeth recently described Crown view, natural size.
from Yunnan, China (Colbert, 1940). Therefore, the
Burmese specimens may be considered to be cospecific Because of its robustness, and its rather simple,
with the form from Yunnan. heavy enamel, this specimen is referred, on the basis of

MEASUREMENTS

ANSP No. 14632 MCZ No. 6258 AMNH No. 38960, type

Antero-posterior diameter of crown ....... 26.5 mm. 27.5 mm. 25.5 mm.
Transverse diameter of crown ........... 28 26.5

The extraordinarily close resemblances between the its tooth, to the genus Potamochoerus rather than to
Pleistocene horses of Burma (Upper Irrawaddy) and Sus. The enamel of the four principal cusps is heavy
of Yunnan, point to a close faunal relationship between and simple, a distinct contrast to the condition in a
these two areas in late Cenozoic times similar to the typical Sits molar, and the tooth is rather broad, as
present-day condition. This close relationship is shown contrasted with the relatively narrow tooth in Sus.
by other elements in the separate Pleistocene faunas of Potaimochoerusand related suids were in the Pleisto-
the two regions, but in no case is it so strikingly dis- cene beds of India-hence the present specimen may be
played as it is with the horses. This may have been considered as indicative of an eastward extension of this
due, in part, to the nature of the animals involved. group of pigs into the Burmese area.
Wild ungulates are roving creatures, and it must be
MEASUREMENTS
supposed that during the Pleistocene horses wandered ANSP No. 14633
freely back and forth between the lowlands of the Ir- M 2, length ................................ 22.5 mm .
rawaddy and the mountainous region of western China, breadth ............................... 18.0
just as did the stegodonts, elephants and certain others
of the larger elements in the faunas. Hexaprotodon iravaticus Falconer and Cautley
As has been shown in the description of the equid
Hexaprotodon iravaticus, Falconer and Cautley, 1847. Fauna
from Yunnan, this was a small horse, seemingly related Antiqua Sivalensis, P1. LVII, figs. 10, 11.
to the true horses rather than to the asses. It may
Type (Lectotype).-Brit. Mus. No. 14771, a man-
very possibly have been close to the ancestor of the dibular symphysis.
steppe horse of modern times, Equus przewalskii, al- Cotype.-The mandibular symphysis figured in the
though it probably was not directly ancestral to this Fauna Antiqua Sivalensis, P1. LVII, fig. 11.
modern horse. Certainly it would seem to be distinct Horizon.-Upper Irrawaddy beds, Lower Pleisto-
from the ancestry of the Asiatic asses, such as Equus cene.
hemionus, a relationship assumed by Lydekker to be Locality.-Irrawaddy River Valley, Burma.
probable for the Pleistocene horse of India, Equus Diagnosis.-A true Hexaprotodon; distinguished by
sivalensis. its small size.
COLBERT: PLEISTOCENE VERTEBRATES 409

Specimens under Consideration.-ANSP No. 14641,


left half of mandibular symphysis. From the Upper
Irrawaddy beds, near Yenangyaung. Locality Yg. 13.
MCZ No. 6259, a right lower second molar. From
the Upper Irrawaddy beds, Magwe.
MCZ No. 6315, fragment of mandible. From the
Upper Irrawaddy beds, in the Yaw Valley.
MCZ No. 6316, fragment of mandible. From the
Upper Irrawaddy beds, in the Yaw Valley.
The well-preserved single tooth is perhaps somewhat
larger than average for this species, but it is definitely
smaller than typical H. sivalensis molars. It supple-
ments the material already known of the Burmese spe-
cies-the type mandibular symphyses in the British
Museum, the fragmentary mandibles in'the Indian Mu-
seum and the partial skull and canine tooth in the
American Museum.
MEASUREMENTS

MCZ No. 6259 AMNH No. 20037


M2 M2
Length ................ 47 mm. 36 mm.
Breadth ............... 31 33.5
Height ................ 32
The mandibular symphysis is particularly interesting
because it shows very plainly the alveoli of the left
incisors. The three incisors were large, but of these
the middle one was appreciably smaller than the ones
placed internally and externally to it. Moreover, this
second incisor was pushed up by the first and third FIG. 83. Hexaprotodon iravaticus Falconer and Cautley.
incisors so that it occupied a position dorsal to them- ANSP No. 14641, left half of mandibular symphysis, anterior
a condition noted by Falconer and Cautley and by view, three-fourths natural size, showing the alveoli of the three
incisors, and the manner in which the second incisor is pushed
Lydekker in the case of Hexaprotodon sivalensis from out of alignment and reduced by the development of the central
the Siwaliks. and lateral teeth.

MEASUREMENTS
ANSP No. 14641 BM No. 147711
Width, symphysis to external border of ramus ...................... 78 mm. 84 mm.
Diameter, alveolus of first incisor ................................. 26 21
Diameter, alveolus of second incisor ............................... 23 18
Diameter, alveolus of third incisor ................................ 25 15
Length of first premolar (roots) ................................... 31 20
Length of second premolar (roots) ................................. 37
1 Measured from
figures-approximate.

Hexaprotodon sp., cf. sivalensis Type (Lectotype).-Brit. Mus. No. 18441, a skull.
Specimen under Consideration.-MCZ No. 6274, Cotype.-Brit. Mus. No. 18442, right mandibular
ramus. From the Upper Siwalik beds of India.
fragment of a left lower canine. From the Upper Ir-
Horizon.-Middle and Upper Siwaliks, Pliocene and
rawaddy beds near Mingun. Locality Ma. 9.
This specimen is indicative of a large type of Hexa- Lower Pleistocene. Upper Irrawaddy beds, Burma.
protodon in the Upper Irrawaddy sediments, perhaps Locality.-Siwalik Hills and Salt Range, India. Ir-
comparable to the Siwalik form, Hexaprotodon sival- rawaddy River, Burma.
ensis. Diagnosis.-A large specialized anthracothere, with
elevated orbits, broad mandibular symphysis, deep man-
Merycopotamus dissimilis (Falconer and Cautley) dibular angle and selenodont, quadricuspid molars.
Specimen under Consideration.-ANSP No. 14634,
Hippopotamus dissimilis, Falconer and Cautley, 1836. Asiatic a cranium, of which the post-orbital portion is fairly
Researches, XIX, pp. 49-61.
Merycopotainus dissimilis, Falconer and Cautley, 1847. Fauna
well preserved. From the Upper Irrawaddy beds, near
Antiqua Sivalensis, Pls. LXVII-LXVIII. Mingun. Locality Ma. 3.
410 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Falconer, in 1868, described and figured in his "Palae- is a high, strong sagittal crest, merging anteriorly with
ontological Memoirs" the front part of a skull of a the strong frontal crests that run out to the supraorbital
young Merycopotamus dissimilis, which was found in processes.
Burma and sent to him by Professor Oldham. At the Posteriorly the occiput is seen to be rather broad in
time this specimen was collected, there naturally was no comparison to its height-perhaps not quite so broad
distinction made between the different levels of the Irra- relatively as in Hippopotamius, but certainly more so
waddy series, so that we do not know from what geo- than in the occiput of Merycopotalmusfigured by Miss
logic horizon the fossil was collected. In 1910 Pilgrim Pearson. The occiput is marked by the very prominent
listed the genus as constituting a member of the Lower central supraoccipitalregion, strongly separated by deep
Irrawaddy fauna, an occurrence that was questioned by depressions from the laterally expanded exoccipitals and
the present author in 1938. It was my contention that squamosals. These depressions were quite obviously
the form was probably of Upper Irrawaddy relation- for the insertion of strong neck muscles. There is also
ships, occurring in association with Stegodon and Hexa- a strong surface of attachment for the ligamentum
protodon. By reason of an inexcusable oversight nuchae. The exoccipitals extend laterally from the re-
whereby Falconer's description, mentioned above, was gion of the condyles, and then expand dorsally and later-
overlooked, I made the statement at that time that "to ally, to form on either side a crescent-shaped plate,
the best of my knowledge the specimen or specimens closely attached to the post-tympanic process of the
from Burma were never described." This statement is, squamosal. As seen in side view, the occiput is vertical
of course, quite erroneous. above the condyles.
In 1929 Miss Pearson mentioned a specimen from The basicranial region is compressed antero-poste-
Yenangyaung (Brit. Mus. no. 8583) as being among riorly, as noted by Miss Pearson. The post-tympanic
the series on which she based her reconstructions of the and the postglenoid processes are so closely appressed
occiput and basicranium in this genus. Evidently this that the tympanic "neck" is squeezed into a very thin
was the skull previously described by Falconer. plate between them, and they form a virtual closing of
The specimen that forms the subject of the present the external auditory meatus below. It would appear
description is but partially preserved. It consists of the that the paroccipital processes were small and short.
postorbital portion of a cranium, with one condyle The tympanic bulla, missing in this specimen, probably
broken and the basicranium partially destroyed. was small, as shown by Miss Pearson in her recon-
It is representative of a medium-sized ungulate- struction of the braincase in Merycopotamus. The
seemingly about the same size as the Merycopotazimus glenoids are shallow, and they are but slightly raised
skulls from the Siwaliks, described by Falconer and above the level of the basicranium and the basioccipital
Cautley and by Lydekker. The brain case is somewhat is heavy.
expanded, and terminated anteriorly by flat frontals, ex- In 1935 I presented a detailed argument in which I
tending laterally to the high, prominent orbits. There tried to show that Hippopotamulsis probably a late de-

FIG. 84. Merycopotamus dissimilis (Falconer and Cautley). ANSP No. 14634, cranium.
Lateral view of right side, three-fourths natural size.
COLBERT: PLEISTOCENE VERTEBRATES 411

FIG. 85. Merycopotamus dissimilis (Falconer and Cautley). ANSP No. 14634, cranium.
Dorsal view, three-fourths natural size.

14634
ANS.P\
14634
A.N.S.P 43
FIG. 86. Merycopotamus dissimilis (Falconer and Cautley). ANSP No.
14634, cranium. Occipital view, three-fourths natural size.

scendant from an advanced anthracothere, such as of the Hippopotamidae,appears to me to offer additional


Merycopotamus, rather than the culmination of a long support to the theory of their anthracothere ancestry.
separate line of evolutionary development having its ulti- The evidence for this is twofold. In the first place, the
mate origin in an ancestor close to the ancestral suids or skull under consideration shows, in occipital view, a
to Cebochoerus. The specimen described above, in so striking resemblance to the skull in occipital aspect of
far as it has any bearing on this question of the origin certain of the more primitive hippopotami. Thus, as
412 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

compared with Hippopotamus minutus this present skull The base of a rather large antler is the only definitive
shows many suggestive similarities, in its general evidence of a cervid in the Upper Irrawaddies. This
breadth, in the lateral expansion of the supraoccipital was probably a form comparableto the modern Rusa.
region, and in the shortness of the paroccipitalprocesses.
Of course there are the characteristic depressions be- MEASUREMENTS
ANSP No. 14643
tween the supraoccipital and the exoccipitals, as shown
also by Miss Pearson's reconstruction, yet these depres- Ant.-post. diameter at base ................. 48 mm.
Height of fork above base .................. 49
sions are bridged at their outer borders by the lateral
reaches of the lambdoidal crest, so that all in all the occi-
Cervid (?)
put has a very hippopotamine appearance, more so than
is the case with the Pearson restoration. Specinens under Consideration.-ANSP No. 14639,
In the second place, a strong similarity between this a fragment presumably of an antler. From the Upper
skull and that of the Hippopotamidae is to be seen in Irrawaddy beds, Mingun. Locality Ma. 4.
the structure of the basicranial region, for in both cases MCZ No. 6271, two fragments, similar to the above-
there is a considerable degree of compression in the listed specimen. From the Upper Irrawaddy beds,
glenoid region, squeezing the tympanic neck and vir- Mingun. Locality Ma. 9.
tually closing the external auditory meatus below, as MCZ No. 6314, an astragalus. From Pyinchaung,
was pointed out in a preceding paragraph. Yaw Valley. May be cervine.
It has been argued by certain students of mammalian The supposed antlers are at best problematical. They
evolution that the resemblances between such advanced would appear to be of probable cervid relationships.
anthracotheres as Merycopotamus and the Hippopotam- They are solid, as one might expect an antler to be,
idae are due to parallelism, and that the Hippopotam- roughly rounded in cross section, and have heavy fold-
idae are in fact of ancient lineage. But if this is so, ings or striations parallel to the long axis of the bone.
why should it be that there is marked "parallelism"be- This last feature is especially pronounced in ANSP No.
tween two artiodactyl groups, of which the ancestry of 14639.
one may be traced back in great detail to Eocene times,
MEASUREMENTS
whereas nothing is known as to the ancestry of the Diameter
other beyond the Upper Pliocene? In other words, why
ANSP No. 14639 ................................. 37 m m .
is it necessary to evoke parallelism to account for the M CZ No. 6271 .................. ................ 31
similarities between Merycopotamulsand the Hippopo-
tamidae in view of the fact that no ancestral primitive Hemibos triquetricornis Riitimeyer, ex Falconer ms.
hippos have been found in early or even middle Tertiary
sediments? Doesn't it seem more logical, on the basis Hemiibos triquetricornis, Rutimeyer, 1865. Verh. Naturf. Ges.
of the many similarities between Merycopotaimus and Basel, IV, (2), p. 330.
Probubalus sivalensis, Riitimeyer, 1865. Idem, p. 331.
Hippopotamus, to assume that the latter is a very late Heimibos triquiticeras, Falconer, 1868. Pal. Mem., I, p. 546.
Tertiary descendant of the former? Is it not probable (Description of unpublished plate H of the Fauna Antiqua
that we have never found primitive ancestral hippopo- Sivalensis).
tamids at a time when the phylogenetic history of most Peribos occipitalis, Lydekker, 1878. Pal. Ind. (X), I, p. 141.
Hemibos occipitalis, Lydekker, 1880. Pal. Ind. (X), I, p. 174.
mammalian families is pretty well known, for the simple Bubalus occipitalis, Lydekker, 1885. Cat. Foss. Mam. Brit.
reason that there are none? The ancestors of the hip- Mus., I, p. 30.
popotamids were specialized anthracotheres. Ainoa triquetricornis, Trouessart, 1898. Cat. Mammalium, p. 985.
MEASUREMENTS

Merycopotanmusdissimilis
o. 14634 BM No. 165511 BM No. 184411 BM No. 16552
Length, condyle-front bord. orbit ........ 175 nnm. 138 mm. 190 mm. 126 mm.
Length, condyle-postorb. proc........... 135 102 150 90
Breadth, glenoids ...................... 130 1602 144+ 150
Breadth, parietals ..................... 89 102 81
Height of occiput above for. mag......... 77 71e3 782 69
1 Measured from figures-measurements approximate.
2 Crushed
(?).
3 Estimated.

Cervus sp. Type.-Brit. Mus. No. 39565, a skull. From the


Specimien under Consideration.-ANSP No. 14643, Upper Siwalik beds of the Siwalik Hills, India.
base of an antler. From the Upper Irrawaddy beds, Horizon.-Upper Siwalik beds, Lower Pleistocene.
near Mingun. Locality Ma. 9. Also Upper Irrawaddies.
COLBERT: PLEISTOCENE VERTEBRATES 413

FIG. 87. Hemibos triquetricornis Rtitimeyer, ex Falconer ms. ANSP No. 14637, right mandibular ramus
with M1_3. External lateral view, natural size.

FIG. 88. Hemibos triquetricornis Riitimeyer, ex Falconer ms. ANSP No. 14637, right mandibular ramus
with M_-3. Crown view, natural size.

Locality.-Siwalik Hills and Salt Range, India. Also MCZ No. 6261, fragment of horn core. From Up-
Irrawaddy River, Burma. per Irrawaddies, Yenangyaung. Locality Yg. 9.
Diagnosis.-A broad-skulled bovine, with relatively MCZ No. 6262, upper molar. From Upper Irra-
large parietals, and moderately divergent horn cores of waddies. Locality Magwe.
triangular cross section, horn cores arising below oc- MCZ No. 6263, two upper molars. From Upper
cipital plane and tilted backward. Upper molars square. Irrawaddies, Mingun. Locality Ma. 9.
MCZ No. 6265, occiput. From Upper Irrawaddies,
Specimiens under Consideration.-ANSP No. 14635, at Pyinchaung, Yaw Valley.
back portion of a skull of a male of the "normal" form,
MCZ No. 6310, distal end of metapodial. From Up-
with the occiput and cranium preserved, as well as the
base of the right horn core. From Upper Irrawaddy per Irrawaddies, Mingun. Locality Ma. 9.
MCZ No. 6311, metapodial fragment. From Upper
beds, Yenangyaung. Locality Yg. 8. Irrawaddies at Pyinchaung, Yaw Valley.
ANSP No. 14637, right ramus with M_1. From MCZ No. 6312, distal end of metapodial. From
Upper Irrawaddies. Locality Magwe. Upper Irrawaddies at Pyinchaung, Yaw Valley.
MCZ No. 6260, occiput. From Upper Irrawaddies. MCZ No. 6313, atlas vertebra. From Upper Irra-
Locality Magwe. waddies. Locality Magwe.
414 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Henmibos triquetricornis, a characteristic Upper Si- Besides the material definitely identifiableas Hem?ibos,
walik bovid, is shown by the above-listed specimens to there are other fragmentary bovid remains from the
have extended its range into Burma during Lower Pleis- Upper Irrawaddies, the relationships of which are prob-
tocene times, another instance of the close relationship lematical. In this connection it might be said that most,
between the Lower Pleistocene faunas of India and of
if not all, of the teeth identified above as Hemiibos are
Burma.
The best example of this species in the present collec- placed in this genus only in a most provisional way. It
tion is the partial skull, consisting of cranium and oc- seems very likely that the upper and lower teeth, ANSP
ciput, and the base of the right horn core. The horn Nos. 14636 and 14637, belong to Hemiibos. The other
core is large and strongly triangular, with an internal, teeth, however, may very well be referable to Bibos
an external and an anterior keel, indicative of the fact rather than to Hemibos, for these two genera are diffi-
that the skull probably represents a male of the "normal" cult to distinguish from each other on the basis of dental
form of this species. The bases of the two horn cores characters. As for the remaining bovid material, short
are fairly close together, and the frontals are concave
discussions are presented below.
between them-a character typical of Hemibos. The
horn cores diverged each from the other at slightly more
than a right angle, and as seen from the side the right
Bibos (?), cf. sondaicus
horn core is tilted backwardly to a pronounced degree-
again a characteristic feature of Hewibos. On the top Specimens ziuder Consideration.-ANSP No. 14642,
of the cranium are strong parietal crests, bounding the
upper and lower teeth, from the Upper Irrawaddy beds.
temporal openings and marking the upper limits of the Yaw Valley.
temporal muscles.
The occiput is projected posteriorly, so that the large MCZ No. 6264, fragment of ramus with right M,_::,
condyles are situated far behind the strong lambdoidal from the Upper Irrawaddies, Mingun Terrace.
crest. As seen from the back, the occiput is broad and MCZ No. 6273, fragment of a horn core, from the
relatively low. The basicranium is partially destroyed Upper Irrawaddies, Magwe.
but it would appear that the paroccipital processes were
MCZ No. 6276, upper molars and fragment of can-
strong and the bullae large and well formed, as is typical
of the large bovids. non bone, from the Upper Irrawaddies, at Mingun.
An occiput, MCZ No. 6260, is definitely of this same Locality Ma. 9.
species. This specimen, because of the lack of horn The teeth listed above are not like those identified as
cores, is probably a female individual. A third oc- Henmibos,but they do seem to show some resemblance
ciput, MCZ No. 6265, is less certainly referable to the to the teeth of a modern banteng. The section of horn
form under discussion, although it is a bovid.
A number of isolated teeth in the collection may be core is large, with a round cross section; it is slightly
considered as belonging to the genus Hemibos, and their curved, and it tapers rapidly. It is comparable in these
specific identity with the skull described above is respects to the basal portion of a large banteng horn
probable. core.

MEASUREMENTS

ANSP No. 14635 MCZ No. 6260 MCZ No. 6265


Length, ant. bord. orbit-occ. cond............... 225 mm.
Breadth, occiput ............................. 154 161 mm.
Breadth, cranium ............................. 103 102
Breadth, occ. condyles ........................ 92 89 90 mm.
Height, occiput above for. mag ................. 68 71
Ant.-post. diam. horn core, at base.............. 70
Transv. diam. horn core, at base............... 64
ANSP No. 14636 MCZ No. 6263 MCZ No. 6262
M1, length X width ...........................
M 2, length X width ........................... 31 X 27 31.5 X 26.5 1
M 3, length X width ........................... 33 X 26 35 X - 36 X 271
ANSP No. 14637
Depth of ramus at M ......................... 50
M1, length X width ........................... 25 X 20
M 2, length X width . ......................... 31 X 21.5
Ma, length X width ........................... 46 X 21.5
1 Position not certain (isolated molars).
COLBERT: PLEISTOCENE VERTEBRATES 415
MEASUREMENTS
ANSP No. 14642 MCZ No. 6276 MCZ No. 6264
Upper molars, length X width ............... 21 X 22 mm.
Upper molars, length X width ............... 28 X 24 mm.
Upper molars, length X width ............. 28.5 X 22
Upper molars, length X width .............. 29 X 23

M2, length X width ........................ 29 X 17 mm.


M3, length X width ......................... 36.5 X 15 43 X 17
MCZ No. 6273
Horn core, ant.-post. diam................... 69 mm.

Hippotragine (?)
Specimtens unllderConsideration.-MCZ No. 6269,
basal portion of a horn core, from the Upper Irra-
waddy beds, Yaw Valley.
MCZ No. 6270, basal portion of a horn core, from
the Upper Irrawaddies, Magwe.
MCZ No. 6272, fragment of a horn core, from the
Upper Irrawaddies, Yaw Valley.
These horn cores are rather large, somewhat elliptical
in cross section, and appear to have been rather straight.
The closest comparisons would seem to be with certain AN.5.P14717
of the modern hippotragines.
This identification, obviously of a most provisional
nature, would, if correct, accord with previous descrip-
tions by Lydekker and by Pilgrim of hippotragine
antelopes in the Upper Siwalik beds.
MEASUREMENTS
FIG. 89. Boselaphine. ANSP No. 14717, lower molar. Exter-
MCZNo. 6270 MCZ No. 6269 nal lateral view above, crown view below, natural size.
Ant.-post. diam. horn core, at
base ................... 51 m m . 46 mm.
Transverse diam. horn core, at
Comparableto the horn cores in some of the modern
base ................... 45 36 gazelles. Small, roundish to elliptical cross section and
very rugose.
Boselaphine (?) MEASUREMENTS
ANSP No. 14638
Specimen tunder Consideration.-ANSP No. 14717, Ant.-post. diam. horn core .................. 27.5 mm.
a lower third molar, from the Upper Irrawaddy beds, Transverse diam. horn core ................. 22
Yaw Valley.
Characterized by the relatively short crown, rather Caprine (?)
strongly curved laterally, and the smooth inner surface, Specimen under Consideration.-ANSP No. 14640,
on which there are no prominent styles. Because of
these charcters, the specimen is considered as of pos- portion of a horn core, from Upper Irrawaddy beds,
Lower Yaw Valley.
sible boselaphine relationships. A fair-sized horn core, flat, with blunt anterior and
MEASUREMENTS posterior keels. Seemingly very much like the horn
ANSP No. 14717 cores of certain Caprinae.
Lower molar
Length (lacking talonid) .................. 27 mm. MEASUREMENTS
B readth ................................ 14 ANSP No. 14640
H eight .................................. 35 Ant.-post. diam. horn core ................. 40 mm.
Transverse diam. horn core ................. 27
Gazella (?)
Caprine (?)
Specimen unIder Consideration.-ANSP No. 14638, Specimen under Consideration.-MCZ No. 6266,
part of a horn core, from the Upper Irrawaddy beds, at fragment of a horn core, from the Upper Irrawaddy
Mingun. Locality Ma. 9. beds, Yaw Valley.
416 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

FIG. 90.
a. Cervus sp. ANSP No. 14643, base of antler. Lateral view.
b. Cervid (?). ANSP No. 14639, fragment of antler. Lateral view.
c. Caprine (?). ANSP No. 14640, fragment of horn core. Lateral view.
d. Caprine (?). MCZ No. 6266, fragment of horn core. Lateral view.
e. Bibos (?). MCZ No. 6273, fragment of horn core. Lateral view.
f. Hippotragine (?). MCZ No. 6270, fragment of horn core. Lateral view.
g. Hippotragine (?). MCZ No. 6269, fragment of horn core. Lateral view.
h. Gazelle (?). ANSP No. 14638, fragment of horn core. Lateral view.
All figures natural size.

This appears to be a caprine, because of its shape and REPTILIA


cross section. The latter is roughly triangular near the CROCODILIA
base, with a blunt anterior keel. The horn core shows
a considerable degree of twisting in the portion pre- Gavialis sp.
served. It cannot be correlated definitely with any
Specimens under Consideration.-ANSP No. 14664,
modern caprine at hand, yet it does not appear to be
fragment of ramus, from the Upper Irrawaddy beds,
of antelopine relationships. Yaw Valley.
MEASUREMENTS ANSP No. 14665, occiput fragment, of unknown age
MCZ No. 6266 and locality.
Ant.-post. diam. horn core .................... 34 mm. ANSP No. 14666, scute, from the Upper Irrawad-
Transverse diam. horn core. .............. .. 26 dies, Lower Yaw Valley.
COLBERT: PLEISTOCENE VERTEBRATES 417
ANSP No. 14667, vertebra, from the Upper Irra- widely spread and coexistent. This problem will be
waddies, Lower Yaw Valley. more fully discussed in subsequent pages of the present
ANSP No. 14668, vertebra, from the Upper Irra- work.
waddies, Magwe. MAMMALIA
ANSP No. 14669, tooth, from the Upper Irrawad-
dies, Magwe. PROBOSCIDEA
MCZ No. 6301, fragment of symphysis, from the Stegodon orientalis Owen
Upper Irrawaddies, Magwe.
MCZ No. 6302, fragment of ramus and vertebra, Stegodon orientalis, Owen, 1870. Quar. Jour. Geol. Soc., XXVI,
p. 421, P1. XXVIII, figs. 1-4.
from the Upper Irrawaddies, Yenangyaung. Locality
Yg. 4. Type (Cotypes).-"Two fragments regd. 41926,
MCZ No. 6303, fragment of ramus from Upper Irra- 41927, Brit. Mus. Geol. Dept."-Hopwood, A. T., 1935.
waddies, Yenangyaung. Locality Yg. 13. Horizon.-Pleistocene.
MCZ No. 6304, vertebra, from the Upper Irrawad- Localitv.-Szechwan, northwest China (for the
dies. Locality Ma. 9. types).
An assortment of jaw fragments, teeth and vertebrae Diagnosis.-"A Stegodon which resembles S. insignis
from various localities, indicates the presence of a very Falconer and Cautley, but differs from it in having the
large gavial in the Upper Irrawaddy beds. ridges more widely spaced, in the lesser development of
the cement, and in the greater lateral curvature of the
Chelonians lower molars, as well as the greater convexity of the oc-
clusal surface of the upper molars." (Hopwood, 1935,
Specimens under Consideration.-ANSP No. 14644,
p. 77.)
part of plastron, from the Upper Irrawaddy beds, Yaw
Valley. Specimens under Consideration.-ANSP No. 14625,
ANSP No. 14670, five fragments of carapace, from a right DM^, the crown complete but the roots some-
the Upper Irrawaddies, Lower Yaw Valley. what broken. From Pleistocene cave deposits. Lo-
MCZ No. 6305, shell fragments, from Upper Irra- cality Mogok C2.
ANSP No. 14626, a left DM4, the crown complete
waddies, Yaw Valley.
MCZ No. 6306, shell fragments, from Upper Irra- but the roots somewhat broken. From Pleistocene cave
waddies, Magwe. deposits. Locality Mogok C2.
MCZ No. 6307, shell fragments, from Upper Irra- MCZ No. 6252, a right DM4, the crown complete but
the roots eaten away (after death but before fossiliza-
waddies, Yenangyaung. Locality Yg. 8.
rodents. From Pleistocene cave deposits.
Represented by fragments of carapace and plastron. tion) by
Some of these fragments, especially ANSP No. 14644, Locality Mogok C2.
be indicative of a form of MCZ No. 6277, tooth fragments. From Pleistocene
may trionychid relationships.
cave deposits. Locality Mogok C2.
THE MIDDLE PLEISTOCENEFAUNA OF BURMA AS These very important specimens may be identified,
FOUND IN THE CAVES AT MOGOK
without much doubt, as Stegodon orientalis, for they are
quite similar to comparable teeth from the Szechwan
At Mogok, in Upper Burma, some fifty miles north- fissures. They may be described briefly as follows.
east of Mandalay, are numerous caves and fissures, The upper tooth consists of seven mammillated crests,
which during many years past have been extensively each crest having on an average about eleven conelets,
excavated by the natives for the rubies that are con- or mammillae. The crests are sharp, and the valleys
tained in the cave deposits. In the course of these ex- are deeply V-shaped. The last crest, or heel, is but half
cavations, the remains of fossil mammals are occasion- the size of the others, and might properly be considered
ally encountered, and these are sometimes saved. The as a half-crest-thereby making the formula for the
expedition, during a two-weeks' stay at Mogok, was tooth 61/2, instead of 7. The occlusal surface is con-
able to collect and purchase a number of such fossils. vex, a character cited by Hopwood as distinctive for
In 1915 Smith Woodward described the skull of a this species.
giant panda from these cave deposits in Upper Burma. The two lower teeth each consist likewise of seven
This specimen has caused some authorities to wonder mammillated crests, in which the number of conelets or
if the deposits in which it was found may not be more or mammillae is essentially the same as for the upper tooth.
less correlative with the fissure deposits of China, in In these teeth, as distinguished from the upper tooth,
which the giant panda is a prominent faunal element. It the seventh crest is virtually of full size, and behind it
is an interesting fact, therefore, that the fossils collected is a very small accessory crest or heel. These teeth are
at Mogok by the Southeast Asiatic expedition lend particularly characterized by their strong lateral curva-
weight to this view-and they go a long way towards ture, a feature designated by Hopwood as distinctive of
linking throughout east Asia that portion of the Pleisto- this species, and certainly quite characteristic of the
cene during which Stegodon and Ailuropoda were Chinese forms.
418 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Two points with regard to the lower teeth deserve MCZ No. 6255, a left M2, with thirteen ridge plates.
special mention. In ANSP No. 14625, there is an ad- (Possibly a left M2, although the grinding surface is
ditional half-crest on the inner side of the tooth-a vari- concave, as in a lower molar, and the median axis of
ation occasionally seen in proboscidean molars, so that the tooth seems to be considerably curved, laterally.)
there are eight crests along the external surface of the From Pleistocene cave deposits. Locality Mogok C2.
tooth, corresponding to seven along the outer side. MCZ No. 6256, a left M2, with thirteen and a half
This half-crest is intercalated between the second and ridge plates. From Pleistocene cave deposits. Local-
third full crests. ity Mogok C2.
In MCZ No. 6252, the roots are completely missing, The upper molar has about eleven full ridge plates,
and the base of the tooth is extensively scored as the and is therefore identified as the penultimate member
result of gnawing by rodents, probably Hystrix, after of the series. It is broad and robust, with tall ridge
the death of the Stegodon, but before it had been fossil- plates. The enamel is strongly crinkled, as is charac-
ized. In this respect the tooth from Mogok may be teristic of Palaeoloxodon.
compared with fossils from caves in China, especially The two lower teeth are typical Palaeoloxodon teeth,
from a cave at Hoshangtung in Yunnan, recently de- characterized especially by their crinkled enamel. Both
scribed by Bien and Chia. These authors attribute the have been identified as lower second left molars, for in
gnawing of the Chinese fossils to Hystrix. both the grinding surfaces are concave, as is charac-

MEASUREMENTS
ANSP No. 14626 ANSP No. 14625 MCZ No. 6252 AMNH No. 18630a
DM4 DM4 DM4 DM4
Length ................ 101 mm. 122 mm. 117 mm. 130 mm.
Breadth............... 60 55 56 62
Index ................ 59 45 48 48
Height ................ 20e 27 27 32
(4th crest)
No. crests............. 62 712 7'2 7
Crests per 100 mm...... 6 6 612 6
Average no. mamillae per
crest.............. 12 11-12 10-11
1An additional half-crest on the lingual side of the tooth.

Palaeoloxodon namadicus (Falconer and Cautley) teristic of lower teeth, and the ridge plates agree in
number with those of the second molar in P. namadicus.
Elephas namadicus,Falconer and Cautley, 1846. Fauna Antiqua
Sivalensis, Pt. 2, Pls. XII A-D, XIII.
Because of this low ridge plate count-as compared with
Elephas (Elasmodon) namadicus, Falconer and Cautley, 1847. eighteen to twenty plates in the third molar-and be-
Fauna Antiqua Sivalensis, PI. XLII. cause of the configuration of the posterior portion of
Elephas (Euelephas) namadicus, Falconer, 1857. Quar. Jour. each tooth, these specimens would seem to be precluded
Geol. Soc. London, XIII, Pt. 4, pp. 315, 317-318. from identification as third molars.
Elephas antiquus (namadicus), Pilgrim, 1905. Rec. Geol. Surv.
India, XXXII, pp. 199-218, Pls. IX-XIII. There is a great disparity in size between the two
Elephas (Palaeoloxodon) namnadicus,Matsumoto, 1924. Jour. specimens, the larger of the two teeth, No. 6255, being
Geol. Soc. Tokyo, XXXI, No. 371, pp. 257, 260. twenty percent longer than the smaller one, almost
Sivalikia namadica, Osborn, 1924. Amer. Mus. Novitates, No.
thirty percent broader, and twenty-five percent deeper.
152, p. 2. This means that the cubic content of the larger tooth is
Type.-Brit. Mus. No. M3092, a skull lacking the more than twice that of the smaller one. In the larger
alveoli for the tusks, but with portions of both third tooth, there are approximately 5/2 ridge plates per 100
molars. millimeters along the worn occlusal surface of the tooth
Horizon.-Middle Pleistocene. as compared with 71/2 plates per 100 millimeters in the
Locality.-Narbada Valley, central India. (For the smaller tooth.
type.) Several interpretations might be given for the dis-
Diagnosis.-A large elephantine with divergent tusks, parity in size between the two molars, MCZ Nos. 6255
and a very strong frontal crest, due to a forward over- and 6256. In the first place, this difference may indi-
growth of the frontal bones. Cheek teeth hypsodont, cate that they represent two extreme variates in one
with wrinkled enamel. Loxodont sinus rudimentary species. Or it is quite possible that the difference in
or absent. About eighteen to twenty ridge plates in size is due to a considerable degree of sexual dimor-
the third molars. phism-which on the basis of our knowledge of modern
Specimens lunder Consideration.-ANSP No. 14627, proboscideans doesn't seem probable. In the third
a right upper second molar. From Pleistocene cave place, it might be supposed that the one is a dwarf
deposits. Locality Mogok C2. individual belonging to a species of which the other
COLBERT: PLEISTOCENE VERTEBRATES 419

specimen represents a normal-sized individual. Finally, wear and of individual variation. Consequently, identi-
the two teeth might be considered as representative of fications within the genus, based upon teeth alone, are
two species or subspecies-an interpretation that doesn't of little value.
seem particularly valid in view of the fact that the speci- As to the present specimen, it is representative of a
mens came from the same locality and are of the same very large individual-perhaps somewhat larger than
geologic age. Taking all factors into consideration, it most large individuals of the modern African Hystrix,
would seem most likely that the two specimens repre- which in turn would seem to be on the average some-
sent extreme variates in one species, even though this what larger than the Oriental forms. Except for its
supposes a very considerable difference in the size of large size, there is little to be said concerning this
adult individual animals. tooth. It is considerably abraded, so that the re-entrant
Measurements of the specimens are given below. folds have been completely eliminated from the outer
ANSP No. 14627 MCZ No. 6255 MCZ No. 6256
M2 M2 M2
Len g th . . ................................. 210 mm. 236 mm. 196 mm.
Breadth ................................... 86 85 62
Index .................................... 41 36 32
H eig ht . . . . .. . .. . . . . . . . . . . . . . . . . . . . . . . . . .. . 163 130 119
Index (height/length)........................ 77 55 61
Approx. volume (cc.). .... ............. 2587 1404
N o. plates ................................. 11Y2-12'2 13 Y2 13Y
Plates per 100 mm.......................... 52 5'2 712

Stegodon (?) or Palaeoloxodon (?) enamel borders, to form "lakes" within the tooth crown.
There are strong roots, anteriorly and posteriorly.
Specimen under Consideration.-ANSP No. 14628,
a portion of a small tusk. From the Pleistocene cave
MEASUREMENTS
deposits. Locality Mogok C2. ANSP No. 14646
This specimen is not certainly identifiable as to genus
or species. Length ................................... 10.5 mm .
W idth .................................... 8.5
RODENTIA
The association of Hystri. with giant panda, Stego-
Hystrix sp.
don, deer, pig, bovids and other mammals character-
Specizmenunder Consideration.-ANSP No. 14646, izes the Pleistocene cave faunas of China, as has been
a left lower fourth premolar. From Pleistocene cave noted by various authors. Indeed, in some of the
deposits. Locality Mogok C2. Chinese caves, as described recently by Bien and Chia
In a single modern species of Hystrix, the cheek
(1938), almost all of the bones found were damaged
teeth are extraordinarily variable, owing to factors of
to a greater or lesser degree by having been gnawed by
Hystrix. Probably the porcupines were attracted to
7: N
the caves, where they fulfilled the role of scavengers to
a certain extent-at least by cleaning up all of the softer
or less resistant portions of skeletons. As has been
noted above, some of the Stegodon teeth from Mogok
have had the roots completely gnawed away, so that
only the hard, resistant crowns remain. Perhaps the
porcupines were not so numerous nor so destructive in
the Burma caves as in the Chinese fissures, especially
those of Yunnan, but it is quite evident that the same
set of conditions existed in both areas so that the en-
vironment and the association of mammals must have
been almost identical. These facts lend weight to the
data supporting a geologic correlation between the fis-
sures of Choukoutien, Szechwan, Yunnan and Mogok.
The deposits must have been virtually identical in age,
FIG. 91. Hystrix sp. ANSP No. 14646, left P4. Crown view
and the time of their deposition was approximately in
above, external lateral view below, twice natural size. the Middle Pleistocene.
420 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
PERISSODACTYLA These antler fragments are indicative of a very large
deer, comparable to the modern sambar (Cervus
Rhinoceros sp.
(Rusa) unicolor) or barasingha (Cervus duvauceli) in
Specimiens under Consideration.-ANSP No. 14645, size. They might belong to a species comparable to
a right upper third premolar. From Pleistocene cave the Lower Pleistocene form, Cercus elegans, from
deposits. Locality Mogok C2. Nihowan.
MCZ No. 6278, a right lower cheek tooth. Froml
Pleistocene cave deposits. Locality Mogok C2.
These are typical Rhinoceros teeth of medium size.
The upper premolar is seemingly rather closely com-
parable to a P3 of Rhinoceros sivalensis of the Upper
Siwaliks. It has two strong vertical folds on the
M.CZ.
6280
ectoloph, and a prominent "parastyle." The crochet is
doubled, while a crista and an antecrochet are lacking.
The posterior fossette is closed posteriorly. In these
respects it resembles the P3 of Rhinoceros sivalensis
figured by Lydekker in 1881.
MEASUREMENTS
ANSP No. 14645 MCZNo. 6278
RP3 Lower tooth
Length .................. 38.5 mm. 47.5 mm.
Breadth ................. 49 27

ARTIODACTYLA FIG. 93. Cervus sp. MCZ No. 6280, right M3. Crown view
above, external lateral view below, natural size.
Sus sp.

Specimen under Consideration.-MCZ No. 6279, a The tooth represents an animal of slightly smaller
much worn right upper third molar. From Pleisto- size, but still close to the Rusa type. Perhaps the dif-
cene cave deposits. Locality Mogok C1. ference is ascribable to individual variation in size.
This tooth, a lower molar, is hypsodont, with fairly
well-developed styles.
MEASUREMENTS
ANSP No. 14647
Diameter (at base) ........................ 77 mm.
Diameter (beam) .......................... 44

Sus sp. MCZ No. 6279, right M3. MCZ No. 6280
FIG. 92.
Crown view, natural size. Length (without talonid) ................... 22 mm.
W idth ............... .................... 13
H eight ................................... 33
Very little can be said about this specimen except
that it affords some evidence of rather modernized pigs
in the Middle Pleistocene of Burma. The tooth re- Medium-sized ruminant
sembles in its general appearance similar teeth from the Specimens under Consideration.-ANSP No. 14648,
Szechwan caves. a right upper molar and a right lower third molar.
MEASUREMENTS
From Pleistocene cave deposits. Locality Mogok C1.
MCZ No. 6279 These teeth represent a ruminant of medium to large
Length ..................................... 37 m m. size, apparently of bovid relationships. It has not been
B readth .................................... 20.5 considered advisable to attempt a generic identification.

Cervus sp. Bovine


Specimens under Consideration.-ANSP No. 14647, Speciiitens under Consideration.-MCZ No. 6281, a
base of an antler; portion of an antler, gnawed by left P4 of a large bovine; also a fragment of a left M3.
rodents. From Pleistocene cave deposits. Locality From Pleistocene cave deposits. Locality Mogok C2.
Mogok C1. These specimens represent a large bovine, probably
MCZ No. 6280, a right third lower molar; also a Bibos. The lower molar fragment resembles to some
right astragalus and some broken rib fragments. From extent comparable elements in the third molar of the
Pleistocene cave deposits. Locality Mogok Cl. banteng.
COLBERT: PLEISTOCENE VERTEBRATES 421

along the Irrawaddy River do not constitute a true


fauna. In the first place, most of the fossils found in
the terraces were not in situ, but were derived speci-
mens, quite obviously washed into the terraces from
the older Upper Irrawaddy beds. These derived fos-
A.N.S.P sils have been described in preceding pages of this
14648
paper, included within that section dealing with the
Upper Irrawaddy fauna, and they have been distin-
guished in their locality descriptions from those Upper
Irrawaddy fossils found in place in the Irrawaddy beds.
It so happens, that in the collections described in this
present contribution, only four specimens were found
in situ in the Irrawaddy River terraces. All of these
fossils were found in the third terrace, T3, as described
by de Terra, and may, therefore, be considered as of
late Middle or early Upper Pleistocene age.
FIG. 94. Bovine. ANSP No. 14648, right upper molar and Unfortunately, these terrace fossils offer little, if any,
right M3. External lateral view of upper molar, left above; additional information as to the sequence of fossil
crown view, left below. Crown view of lower molar, right above; faunas in Burma. As identified, they would seem to
external lateral view, right below. All natural size. be specifically the same as like fossils in the Upper Ir-
rawaddy fauna. Consequently, it would appear as if
there had been a continuation in the Irrawaddy River
Valley of the characteristic Lower Pleistocene faunal
facies of Siwalik relationships from Lower into Middle
and Upper Pleistocene times. Such an occurrence is
neither unreasonable nor unusual-for careful studies
in various parts of the world, based upon the best faunal
|l M.C.Z.6281 and stratigraphic evidence, would indicate that Pleisto-
cene faunas generally continued without much change
I through the duration of the epoch-the differences be-
tween the mammalian assemblages of successive de-
posits in any one area being due more to facies differ-
ences, dependent upon local extinctions, emigrations
I, and immigrations, than to evolutionary changes within
the faunas. This explanation, for instance, is the logi-
cal one to account for the differences between the Up-
per Irrawaddy and the Mogok faunas, of Lower and
Middle Pleistocene age respectively, as will be shown
in subsequent pages of this present work.

PROBOSCIDEA

Elephas hysudricus Falconer and Cautley


FIG. 95. Large bovine, cf. Bibos. MCZ No. 6281, left P4 (For synonymy, types, horizon, locality and diagnosis of this
and a fragment of left M3. External lateral view of P4, left species, see page 405)
above; crown view, left below. Crown view of M3, right above;
external lateral view, right below. Natural size. Specinens under Consideration.-ANSP No. 14657,
tooth fragment. Found in situ in Irrawaddy River
THE MIDDLE TO UPPER PLEISTOCENE MAMMALS OF Terrace III. Locality Ma. 3*.
BURMA AS FOUND IN SITU IN THE IRRAWADDY ANSP No. 14658, tooth fragment. Found in sitlt
RIVER TERRACES in Irrawaddy River Terrace III. Locality Ma. 8* ter-
race.
FOSSILS OF THE IRRAWADDY RIVER TERRACES MCZ No. 6296, tooth fragment. Found in situ in
In contradistinction to the fossils of the Upper Ir- Irrawaddy River Terrace III. Locality Ma. 8* terrace.
rawaddy beds and to those of the Mogok Caves, the Characterized by the rather thin and much-crenulated
specimens discovered in the five successive terraces enamel. Therefore identified as E. hysldriclts.
422 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
ARTIODACTYLA Speciimen unlder Consideration.-ANSP No. 14620,
Hemibos
Hemibos triquetricornis Ruitimeyer, Fal r ms.
Riitimey, ex Falconer S. left upper first kmolar. From post-Pleistocene deposits.
triquetricornis
Locality Mogok C3.
(For synonymy, types, horizon, locality and diagnosis of the A characteristic Indian elephant molar, which shows
this species, see page 412)i, if a r o i
little, if any, trace of mineral replacement.
'pecimen under Consideration.-ANSP No. 14636,
right M2-3 in palatal fragment. Found in situ in Ir- ARTIODACTYLA
rawaddy River Terrace III. Locality Ma. 8* terrace.
Seemingly identical with dentitions of Hemibos from Bos sp.
the Upper Irrawaddies and the Upper Siwaliks.
Specimenun1der Consideration.-MCZ No. 6309, a
left mandibular ramus, containing P4-M:,. From post-
Pleistocene deposits. Locality Mogok C3.
This jaw is representative of a medium-sized Bos-
perhaps it belonged to a domesticated animal. It is
noteworthy in that it shows little, if any, fossilization,
and moreover, in that it has been extensively gnawed
by rodents-presumably Hystrix.

POST-TERRACEDEPOSITS
A few scattered teeth and broken bone fragments
were found near Kyaukpadaung, at Locality K1 and
K2, Ky., in deposits of post-Terrace age. Thus these
specimens must be of very late Pleistocene or post-
Pleistocene affinities. They are interesting in that they
were associated with stone implements.

PERISSODACTYLA

Equus sp.
Specimen under Consideration.-MCZ No. 6317, a
right upper third molar, crown broken. From post-
Terrace deposits, near Kyaukpadaung, Locality Ky-2.

FIG. 96. Hemibos triquetricornis Riitimeyer, ex Falconer


ms. ANSP No. 14636, right maxilla with M2-3. External
lateral view above, crown view below, natural size.

V. POST-PLEISTOCENE FOSSILS
A few specimens were collected from deposits of post- I
Pleistocene age. They are described briefly below.
MOGOK, SURFACE DEPOSITS IN CAVES
PROBOSCIDEA

Elephas maximus Linnaeus


Elephas maximus, Linnaeus, 1766. Syst. Nat., I, p. 48. FIG. 97. Equus sp. MCZ No. 6317, right M3. External lateral
Elephas indicus, Cuvier, 1817. Regne an. I, p. 231. view above, crown view below, natural size
COLBERT: PLEISTOCENE VERTEBRATES 423
This specimen represents a rather small equid, char- if they are from cave deposits, as suggested by de Terra.
acterized .in the upper molar by the fairly simple fos- Unfortunately, the development of the worn enamel
sette borders, the lack of a pli caballin and the very on the lower molar, MCZ No. 6257, offers little aid in
long protocone. From these characters, particularly an interpretation of relationships.
the lack of the pli caballin, it is possible that the tooth According to a personal communication from Dr.
may be that of an ass. H. L. Movius, Jr., who secured the specimens:
"I obtained this material from a missionary (Dr. J.
MEASUREMENTS
H. Telford) in the far east of Burma, near the Yunnan,
Length .......................................... 25 mm . Indo-Chinese and Siamese frontiers. He collected it at
Width . .. . ....................................... 20 various times during trips in the Nam Hkok Valley in
the south of Kengtung State, but I was unable to de-
ARTIODACTYLA termine the precise locality."
Cervus sp. MIEASUREMENTS
Lower molar
Specimens under Consideration.-MCZ No. 6318, MCZ No. 6257
teeth, jaw fragments, bones. From post-Terrace de- 204 m m.
Length .....................................
posits, near Kyaukpadaung, Locality K1-Ky. W idth ..................................... 67
A few jaw fragments and teeth show the presence in Height ..................................... 126
the post-Terrace deposits of a small Cervus with hypso- No. ridge plates ............................. 14+
dont teeth. Ridge plates per 100 mm.................... 6

Antelope ?
VII. RELATIONSHIPS AND CORRELATION
Specimens under Consideration.-MCZ No. 6319, OF THE PLEISTOCENE MAM-
teeth, jaw fragments, bones. From the post-Terrace MALIAN FAUNAS
deposits, near Kyaukpadaung, Locality K1-Ky. OF BURMA
The specimens listed under this classification repre-
RELATIONSHIPS OF THE UPPER IRRAWADDY FAUNA
sent a rather large antelope or gazelle. Perhaps more
than one type is present. It has been a recognized fact from the days of the
earliest work on fossil vertebrates in the Orient that
Bos sp. the fossil mammalian faunas of Burma are closely re-
lated to those of India itself. This fact was made ap-
Specimens under Consideration.-MCZ No. 6320,
teeth, jaw fragments, bones. From the post-Terrace parent by the studies of the earlier workers-Falconer
and Cautley, and Lydekker, but it remained for Pilgrim
deposits, near Kyaukpadaung, Locality K1-Ky.
Teeth of a large bovid, closely comparable to homolo- (1910) to emphasize the close relationship of the Bur-
teeth in Bibos. These may be indicative of a mese and Indian faunas. This author pointed out the
gous
banteng. close correspondence of the Lower Irrawaddy fauna to
In addition to the listed material, is a certain amount the Middle Siwalik fauna of Dhok Pathan. Somewhat
of broken limb bones, vertebrae, etc., of ruminant rela- later Stamp (1922) demonstrated the resemblance of
tionships but not further identified. the Upper Irrawaddy fauna to the lower part of the
Upper Siwaliks (Tatrot) and suggested a close genetic
VI. FOSSILS OF UNKNOWN AGE relationship between them. This correlation has been
confirmed recently by Colbert (1938) and by Pilgrim
PROBOSCIDEA (1939).
In spite of the understood relationship between the
Elephas sp. Upper Irrawaddy fauna and that of the Upper Siwaliks,
no very detailed arguments have been presented to af-
Specimens under Consideration.-MCZ No. 6257, a
left lower second (?) molar. From eastern ford a background for the conclusions reached. In
Kengtung,
Burma, near the Yunnan border. Horizon unknown. view of the several additional elements added to the
MCZ No. 6268, an upper molar. From Kengtung. Upper Irrawaddy fauna by the collections described on
Horizon unknown. preceding pages of this paper, thereby affording us a
From the nature of the material it is very difficult to more complete knowledge as to the composition of the
make a specific identification of the teeth from Keng- fauna than had hitherto been obtained, it is proposed
tung. They may be Elephas hysudricus-if so, they to determine as far as possible by detailed comparisons
possibly represent a mid-Pleistocene survival of this the stratigraphic and geographic affinities of the Upper
species. On.the other hand, the rather tall ridge plates Irrawaddy mammalian assemlblage.
suggest the possibility of their identification as Palae- One evident fact is that, so far as our knowledge is
oloxodon nanimadicus-a possibility made very probable based upon described species, the number of Upper Si-
424 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
walik elements in the Upper Irrawaddy fauna is over- semblages of Java are closely related to the Upper
whelmingly preponderant. Of thirteen species of mam- Siwalik fauna or faunas of India. These are specifically
mals known from the Upper Irrawaddy beds, eight are the Tji Djoelang fauna, which he correlates with the
present also in the Upper Siwaliks. In addition, there Indian Tatrot, and the Kali Glagah fauna, which is
are three other species seemingly confined to Burma, placed by him between the Tatrot and the Pinjor of
but very closely related to Siwalik forms. Only one of India, in which there are definite Siwalik elements and
the described species is seemingly related to a form from a noticeable lack of eastern Asiatic forms, such as are
outside India-namely Equus yunnanensis, a type of- found in the early Pleistocene of southern China.
fering a close tie-up between Burma and Yunnan. A Therefore, the broad faunal assemblage stretching from
number of genera are known from the Upper Irra- India through Burma into the East Indian Islands is
waddy fauna, of which the material is too poorly pre- designated by that author as the "Siva-Malayan Fauna,"
served to afford specific identifications. Practically all representing the earliest stage of the Pleistocene (Upper
of these genera are also present in the Upper Siwaliks- Pliocene of Koenigswald) in the Oriental region. The
although in this case comparisons are difficult and in- collections made by Teilhard and de Terra in Burma
conclusive, because the forms in question are for the confirm the reality of this Siva-Malayan fauna, seem-
most part widely-spread Eurasiatic or Oriental types. ingly originating in a northern Indian center, pushing to
Perhaps the above facts may be brought out to addi- the southeast through Burma-in which region it is but
tional advantage by a chart or diagram, as follows: little changed from its original aspect, and finally arriv-
Pr esent in Confined to East of Widely Spread
Upper Irrawaddy Mammalian Fauna Siiwaliks Burma Burma in Asia
Stegolophodon latidens ........................ x
Stegodon elephantoides ....................... x
Stegodon insignis birmanicus .................. x
Elephas hysudricus .......................... x x

Rhinoceros sivalensis ......................... x


Hipparion cf. antelopinum.................... x
Equus yunnanensis .......................... x

Potamochoerus sp............................ x
Hexaprotodon iravaticus ......................
Hexaprotodon cf. sivalensis.................... x
llerycopotamus dissimilis..................... x x x

Cervus sp................................... x
H ippotragine ............................... x
Gazella ..................................... x
Caprine .................................... x
Boselaphine ................................ x
Hemibos triquetricornis ....................... x
Proleptobos birmanicus ....................... x
Bibos cf. sondaicus .......................... x

From these data it seems reasonable to conclude that ing, at least in part, in the westernmost islands of the
the Upper Irrawaddy fauna is essentially an eastwardly East Indies, which were at that time undoubtedly con-
extension of the Upper Siwalik fauna. Of course we nected with the mainland.
know that certain elements in the Siwalik fauna ex-
tended over wide areas in eastern Asia; Archidiskodon RELATIONSHIPS OF THE MOGOK FAUNA
planifrons and Elephas hysudricus, for example. But
it appears probable that Burma represents an area near When we turn to a consideration of the Mogok Cave
the eastern border of the Siwalik faunas as a whole, fauna, we find a picture considerably different from
while the faunas inhabiting the regions to the east of that shown by the Upper Irrawaddy fauna. For the
Burma, though often containing certain Siwalik ele- Mogok fossils, even though scanty and insufficiently
ments, were essentially independent complexes, having known, would seem to show affinities with similar cave
definitive characters of their own. The Burma Equus faunas to the east-rather than with the faunas of the
described above is a local element entering the charac- Indian region to the west. The best documented of the
teristic Upper Siwalik assemblage from the southeast. Mogok species is Stegodon orientalis, known from a
All of these faunas are essentially of Lower Pleistocene number of teeth and jaw fragments, both upper and
age. lower, and these are seemingly identical with similar fos-
In recent years (1934, 1935, 1939) von Koenigswald sils found in various cavern deposits throughout south-
has shown that the earliest Pleistocene mammalian as- ern and western China. Also, Elephas nanmadicusis a
COLBERT: PLEISTOCENE VERTEBRATES 425
characteristic element in the Mogok fauna, and while archipelago, but it seems to the present author more
this species is quite characteristic of the Middle Pleisto- likely that the spread of the Sino-Malayan fauna to the
cene of central India, the type locality, it is nevertheless East Indies was by way of the Malay Peninsula-the
a widespread form throughout eastern and southern natural route for all mammalian migrations from the
Asia. Of course, Elephas namadicus alone would be mainland to the islands during Pleistocene times. The
insufficient to establish the geographic and faunal rela- Mogok fauna, being close to the cave faunas of Yunnan.
tionships of the Mogok fauna, but the fact that it is Kwangsi and Szechwan, may be considered as an ele-
found associated with Stegodon orientalis increases con- ment in this large faunal complex.
siderably the probability that the Burma fauna is re-
lated to cave faunas to the east in China, and this proba- RELATIONSHIPS OF THE TERRACE FOSSILS
bility is made much more certain by the occurrence of
two other fossil forms in the Mogok fauna, namely the It has already been shown that the terrace fossils
giant panda, Aeelreidopus (= Ailuropoda) baconi, and contain a mixture of rewashed Upper Irrawaddy speci-
the porcupine, Hystrix. It might be added, also, that mens and fossils in situ, these latter the remains of ani-
the other Mogok fossils, although too fragmentary to be mals that were contemporaneous with the terrace in
truly definitive, would seem to show resemblances to which they are deposited. Owing to the small number
similar fossils from the Chinese caves. of these terrace elements in place, no very definite idea
If we analyze the Mogok fauna, as the Upper Irra- as to the characteristics of the "fauna," if so it may be
waddy fauna was analyzed, a chart may be prepared called, is to be had. The evidence would seem to point,
somewhat as follows. Since Elephas namadicus is the however, to the idea that these terrace animals are
elephantine at Mogok, the comparison with faunas to more or less continuations of the Upper Irrawaddv
the west is made with the Narbada fauna. mammalian assemblage-as was suggested on a preced-
Present in Confined to East of Burma Widely Spread
Mogok Fauna India Burma (China) in Asia
Ailuropoda baconi ..................... x (x)
Stegodon orientalis ..................... x
Elephas namadicus........ ............ x
Rhinoceros sp ......................... x x
Sus sp ............................. x x
Cervus sp ............................. x x
IHystrix sp .......................... x
Bovine .............................. x

From this it may be seen that the truly definitive spe- ing page of this work. Consequently there might have
cies in the Mogok fauna are essentially southeastern been two faunal facies in Burma during at least a part
Asiatic forms. And while no one of these species in of the Pleistocene, the Irrawaddy facies, having its
itself would be particularly definitive in settling the origin in and relationships with mammalian faunas in
affinities of the Mogok fauna, the combination of forms India, and the Mogok facies, having its origin and re-
is such that the Mogok fauna resembles closely the cave lationships with mammalian blocks in China and ad-
faunas of southeastern China. jacent regions. It must be remembered in this connec-
The weight of evidence founded upon recent work in tion, and this point is worthy of emphasis, that any
southeastern Asia would seem to indicate that these cave conclusions based upon the fossils found in situ in the
faunas are later than the Upper Siwalik complex, and Irrawaddy River Terraces must of necessity be very
therefore might be regarded as of late Lower Pleistocene tentative at the present time, because of the scarcity of
or as of Middle Pleistocene age. They are correlative such fossils.
with the Narbada fauna of India, and to the south, with
the Trinil fauna of Java (Fig. 99). THE CORRELATION OF THE PLEISTOCENE MAMMALIAN
Von Koenigswald has recognized this fact, for he FAUNAS OF BURMA

has shown that whereas the earlier Pleistocene faunas


Perhaps some words of explanation are necessary as
of the East Indies (considered by him to be of Upper to the correlations adopted in the foregoing discussions,
Pliocene age) are of Siwalik affinities, constituting his and particularly in the chart, as presented in Fig. 98.
Siva-Malayan complex, the succeeding faunas (the It will be noted that the Upper Irrawaddy fauna is cor-
Trinil and associated assemblages, placed by him in the related with the Pinjor fauna to the west, and with the
Lower Pleistocene) are related to the cave faunas of Ma-Kai fauna of Yunnan and roughly the Djetis fauna
south China. Consequently he has named this group of Java; all of which are placed in the Lower Pleisto-
of mammalian faunas the Sino-Malayan fauna, having cene. Again, the fauna of the Mogok caves is made
its origin in China. According to von Koenigswald correlative with the Boulder Conglomerate deposits of
this fauna reached the Javan region through the eastern India, and roughly with the Narbada fauna, with the
426 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

FIG. 98. The relationships of the Pleistocene mammalian faunas of southeastern Asia.
The lower arrow shows the southeastern spread of the Indian fauna in Lower Pleistocene
times; the upper arrow shows the southwestern spread of the Chinese fauna in lower Middle
Pleistocene times.

faunas of the Chinese caves to the east and of Trinil to c. That the Pinjor fauna is equivalent to the Euro-
the southeast; all of these being regarded as more or pean faunas found in the Villafranchian stage.
less of Middle Pleistocene age. d. That both the Astian and the Villafranchian are
These correlations differ somewhat from the plans of Pliocene age.
adopted separately by Pilgrim and by von Koenigswald. Many of the differences of opinion regarding the cor-
Pilgrim regards the Pinjor, with which the Upper Ir- relative relationships of the Indian faunas with faunas
rawaddy is correlative, as of Villafranchian age, which in other parts of the world, are due to differences in the
stage he places in the Upper Pliocene. The underly- definition of the Pliocene-Pleistocene boundary. It is
ing Tatrot of India is regarded by Pilgrim as of Astian my contention that the Villafranchian of Europe, con-
age, while the overlying Boulder Conglomerate and the taining true Equus, Elephas (Archidiskodon) and Bos,
Narbada, with which latter the Mogok may be cor- represents the beginning of the Pleistocene, while Pil-
related, is placed by this author in the Lower Pleisto- grim would place this same stage at the top of the
cene. Similarly, in correlating the Javanese faunas, Pliocene.
von Koenigswald places the Tji Djoelang and the Kali Let us assume that the Villafranchian does represent
Glagah, the former of which he would correlate with the beginning of the Pleistocene, since it is here that
the Tatrot, in the Pliocene. The Djetis fauna of Java,
Equlits, Archidiskodon and Bos appear, along with nu-
which is regarded as perhaps somewhat later than the merous other "modern" mammalian types. What,
Pinjor of India, is placed by this author in the Lower then, are the relationships of the Tatrot and Pinjor of
Pleistocene, while the Trinil fauna above it is placed India ?
higher up in the Pleistocene, more or less equal to the Unfortunately, the Tatrot fauna is rather scanty, as
Boulder Conglomerate and the Narbada-these latter
compared with the rich Dhok Pathan and Pinjor faunas
being regarded by von Koenigswald as of Middle below and above it. Even so there is a considerable
Pleistocene age. list of genera and species found in the Tatrot, of which
Pilgrim's arguments as to the age of the Upper many are holdovers from the Dhok Pathan. Some of
Siwalik faunas-as recently presented (1938)-are these persistent types give a Pliocene aspect to the
based upon the following considerations: Tatrot fauna, particularly such genera as Sivaonyx,
a. That the Tatrot fauna is quite separate and more Agriotherium, Hipparion, Pachyportax and Seleno-
primitive than the Pinjor fauna. portax. On the other hand, as Pilgrim points out, only
b. That the Tatrot fauna is generally equivalent to one Tatrot genus, Proamphibos, fails to persist into the
the Astian stage, as represented by the Rous- Pinjor; so that the Pleistocene or "Villafranchian"
sillon fauna of Europe. aspect of the fauna is indeed strong. This is made
COLBERT: PLEISTOCENE VERTEBRATES 427

especially so by the presence in the Tatrot of such im- Pinjor, as of Middle Pleistocene age, unless we are
portant and definitive types as Equus, Archidiskodon willing to assign a great thickness to the Lower Pleisto-
and Sivatherium. Therefore, it seems to me that, cene of India, and include the Boulder Conglomerate
rather than showing Astian affinities as Pilgrim holds, with the preceding horizons in the earlier part of the
all of the evidence would tend to link the Tatrot with Quaternary. This alternative is quite possible. Owing
the overlying Pinjor, making it of Lower Pleistocene to the lack of fossils in the Boulder Conglomerate itself,
age. Granting that the Tatrot is separate from the direct proof of the age of this horizon is not feasible,
Pinjor, and forms a fauna more or less intermediate although it may be remarked here that Teilhard and de
between the Pinjor and the underlying Dhok Pathan, Terra have found evidence supporting the idea that a
nevertheless the aspect of this linking fauna is definitely correlation may be established between the Mogok fis-
Pleistocene, and it may be regarded as rather closely sure deposits and boulder fans, these in turn being
related to the succeeding Pinjor assemblage. Conse- roughly equivalent to the Boulder Conglomerate of
quently, if the Tatrot is of Lower Pleistocene or Villa- India.
franchian relationships, the Pinjor must be viewed The Narbada fauna would appear to be later than the
either as a fauna of post-Villafranchian age, retaining Boulder Conglomerate, and definitely of Middle Pleisto-
many Villafranchian characters, or as a later and sepa- cene age. Perhaps the first terrace of the Irrawaddy
rate manifestation of the same broad Villafranchian River is equivalent in age to the Narbada.
complex. Von Koenigswald places the Tji Djoelang fauna of
If the Tatrot and Pinjor faunas are considered as of Java in the Upper Pliocene, emphasizing particularly
Lower Pleistocene age, it then seems logical to regard the presence of Merycopotamus and the first appear-
the Boulder Conglomerate horizon, overlying the ance of Archidiskodon as being indicative of the Upper

FIG. 99. Map of southeastern Asia, showing the directions of expansion of the Lower Pleistocene fauna from India (circles
and solid arrows), and of the Middle Pleistocene fauna from China (squares and broken arrows). These are the "Siva-Malayan"
and "Sino-Malayan" faunas, respectively, of von Koenigswald.
428 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Pliocene age of this horizon. But Merycopotamus is LYDEKKER, R. 1876. Notes on the Fossil Mammalian Faunae
not an especially characteristic Pliocene genus, although of India and Burma. Rec. Geol. Surv. India, IX, Pt. 3,
pp. 86-106.
many authors seem to have the idea that it is. It is --. 1878. Crania of Ruminants from the Indian Tertiaries.
just as typical of the Lower Pleistocene in the Oriental Pal. Indica, (X), I, Pt. 3, pp. 88-171, Pls. XI-XXVIII.
region as it is of the Pliocene. And Archidiskodon, as 1880a. A Sketch of the History of the Fossil Vertebrata
has been said, is here taken as a marker for the begin- of India. Jour. As. Soc. Bengal, XLIX, Pt. 2, pp. 8-40.
ning of the Pleistocene. 1880b. Siwalik and Narbada Proboscidea. Pal. Indica,
(X), I, Pt. 5, pp. 182-292, Pls. XXIX-XLVI.
Similarly, the Kali Glagah fauna has every appear- - . 1883. Synopsis of the Fossil Vertebrata of India. Rcc.
ance of being Pleistocene in its affinities, rather than Geol. Surv. India, XVI, pp. 61-93.
Upper Pliocene, even though it contains a few primitive .1884. Rodents and New Ruminants from the Siwaliks
holdovers from the Upper Tertiary. Stegodon, Hip- and Synopsis of Mammalia. Pal. Indica, (X), III, Pt. 3.
popotanmus,Sus, Cervius, Antilope, and Bubalis cer- pp. 105-147.
tainly make this a typical Pleistocene association. .1885. Catalogue of Siwalik Vertebrata in the Indian
Von Koenigswald regards the Trinil fauna as of Mid- Museum. Part I, Mammalia. Calcutta.
.1886a. Catalogue of Fossil Mammalia in the British
dle Pleistocene age, about equivalent to the Indian Museum. Part II, London.
Narbada. This view accords with the ideas set forth in .1886b. Catalogue of Fossil Mammalia in the British
preceding portions of the present work, and need not Museum. Part IV, London.
be elaborated at the present time. MATTHEW, W. D. 1929. Critical Observations upon Siwalik
Mammals. Bull. Amer. Mus. Nat. Hist., LVI, Art 7, pp.
437-560.
BIBLIOGRAPHY OF PART III
MATTHEW,W. D., and W. GRANGER. 1923. New Fossil Mam-
BIEN, M. N., and L. P. CHIA. 1938. Cave and Rock-Shelter mals from the Pliocene of Sze-Chuan, China. Bull. Amcr.
Deposits in Yunnan. Bull. Geol. Soc. China, XVIII, nos. Mus. Nat. Hist., XLVIII, Art. 17, pp. 563-598.
3-4, pp. 325-347. OSBORN,H. F. 1929. New Eurasiatic and American Probos-
CHAKRAVARTI, D. K. 1937. A New Stage in the Evolution cidea. Amer. Mus. Novitates, no. 393, pp. 1-22.
of Stegodons. Quar. Jour. Geol. Min. Met. Soc. India, IX, 1936. Proboscidea, a Monograph of the Discovery, Evo-
no. 2, pp. 33-37. lution, Migration and Extinction of the Mastodonts and
CHHIBBER,H. L. 1934. The Geology of Burma. MacMillan Elephants of the World. Volume I, New York.
and Co., London, pp. 224, 232-249, 250-258. - . 1942. Idem., volume II, New York.
CLIFT, W. 1828. On the Fossil Remains of Two New Species PILGRIM, G. E. 1906. Fossils of the Irrawaddy Series from
of Mastodon, and of Other Vertebrated Animals Found on Rangoon. Rec. Geol. Surv. India, XXXIII, Pt. 2, pp. 157-
the Left Bank of the Irawadi. Trans. Geol. Soc. London, 158.
(Second Ser.) II, Pt. 3, pp. 369-376, Pls. XXXVI-XLIV. .1910a. Notices of New Mammalian Genera and Species
COLBERT, E. H. 1935. Siwalik Mammals in the American from the Tertiaries of India. Rec. Geol. Surv. India, XL.
Museum of Natural History. Trans. Amer. Philos. Soc., pp. 63-71.
(N. S.) XXVI, pp. i-x, 1-401. .1910b. Preliminary Note on a Revised Classification of
1938. Fossil Mammals from Burma in the American the Tertiary Freshwater Deposits of India. Rec. Geol.
Museum of Natural History. Bull. Amer. Mus. Nat. Hist., Surv. India, XL, Pt. 3, pp. 185-205.
LXXIV, Art. VI, pp. 255-436. .1926. The Tertiary Formations of India, and the Inter-
DE TERRA, H. 1937. First Scientific Field Report of the relation of the Marine and Terrestrial Deposits. Proc.
American Southeast Asiatic Expedition for Cenozoic Geol- Pan-Pacific Sci. Congress (Australia), 1923, pp. 896-931.
ogy and Early Man. Mimeographed, December 25, 1937. (Issued, 1926.)
. 1938a. Second Scientific Field Report of the American .1937. Siwalik Antelopes and Oxen in the American
Southeast Asiatic Expedition for Cenozoic Geology and Museum of Natural History. Bull. Amer. Mus. Nat. Hist.,
Early Man. Mimeographed, January 30, 1938. LXXVII, Art. 7, pp. 729-824.
. 1938b. Third Scientific Field Report of the American . 1938. Are the Equidae reliable for the Correlation of
Southeast Asiatic Expedition for Cenozoic Geology and the Siwaliks with the Cenozoic Stages of North America?
Early Man. Mimeographed, March 25, 1938. Rec. Geol. Surv. India, LXXIII, Pt. 4, pp. 437-482.
-. 1939. The Quaternary Terrace System of Southern Asia STAMP, L. D. 1922. An Outline of the Tertiary Geology of
and the Age of Man. Geog. Review, XXIX, no. 1, pp. Burma. Geol. Mag., LIX, pp. 481-501.
101-118. TEILHARDDE CHARDIN, P. 1936-1937. Notes on Continental
DE TERRA,H., and P. TEILHARDDE CHARDIN. 1936. Observa- Geology. Bull. Geol. Soc. China, XVI, pp. 195-220.
tions on the Upper Siwalik Formation and Later Pleistocene .1937. The Post-Villafranchian Interval in North China.
Deposits in India. Proc. Amer. Philos. Soc., LXXVI, no. Bull. Geol. Soc. China, XVII, no. 1, pp. 169-176.
6, pp. 791-822. .1938. Le Villafranchien d'Asie et la Question du Vilia-
FALCONER, H. 1868. Palaeontological Memoirs I, pp. i-lvi, franchien. C. R. S. de la Soc. geol. de France, no. 17, pp.
1-590. (Fauna Antiqua Sivalensis, pp. 1-556.) 325-327.
FALCONER, H., and P. T. CAUTLEY. 1846-1849. Fauna Antiqua TEILHARDDE CHARDIN, P., and M. TRASSAERT. 1937. The
Sivalensis. Proboscideans of Southeastern Shansi. Pal. Sinica, Ser. C,
HoPwooD, A. T. 1935. Fossil Proboscidea from China. Pal. XIII, Fasc. 1.
Sinica, Ser. C, IX, Fasc. 3. VON KOENIGSWALD, G. H. R. 1934. Zur Stratigraphie des
LEWIS, G. E. 1937. A New Siwalik Correlation. Amer. javenischen Pleistocan. De Ingenieur in Nederlandsch-
Jour. Sci., XXXIII, pp. 191-204. Indie, no. 11, pp. 185-201.
COLBERT: PLEISTOCENE VERTEBRATES 429
. 1935. Die fossilen Saugetierfaunen Javas. Proc. Ko- WADIA,D. N. 1919. Geology of India. MacMillan and Co.,
ninklijke Akademie van Wetenschappen te Amsterdam. London.
XXXVIII, no. 2, pp. 188-198. WOODWARD, A. S. 1915. On the Skull of an Extinct Mammal
. 1938-1939. The Relationship between the Fossil Mam- Related to Aeluropus from a Cave in the Ruby Mines at
P1. Burma.
Mogok, Proc. Zool. Soc. London, 1915, pp. 425-
malian Faunae of Java and China, with Special Reference 428, 1
to Early Man. Peking Nat. Hist. Bull., XIII, Pt. 4, pp. YOUNG,C. C 1932. On Some Fossil Mammals from Yunnan.
293-298. Bull. Geol. Soc. China, XI, pp. 383-393.
1939. Das Pleistocan Javas. Quartar, Band 2, pp. 28- . 1937. New Vertebrate Horizons in China. Bull. Geol.
53, Pls. IX-XI. Soc. China, XVII, nos. 3-4, pp. 269-288.
PLATE XIX

M.C.Z.
6290
M.
C.Z.
246 67
FIG. 1. Stegolophodon latidens (Clift). ANSP No. 14621, left lower molar. Crown view.
FIG. 2. Stegolophodon latidens (Clift). MCZ No. 6267, upper molar. Crown view.
FIG. 3. Elephas hysudricus Falconer and Cautley. MCZ No. 6290, palate with upper molars.
All figures three-fifths natural size.
PLATE XX

A.N$.P14623

Stegolophodon latidens (Clift). ANSP No. 14623, right mandibular ramus with third molar. Crown view above, external lateral view below;
one-half natural size.
PLATE XXI

FIG. 1. Stegodon elephantoides (Clift). ANSP No. 14624, right mandibular ramus with DM3s4. Crown
view above, external lateral view below.
FIG. 2. Stegodon elephantoides (Clift). ANSP No. 14622, left M2. Lateral view above, crown view below.
All figures three-fifths natural size.
PLATE XXII

M.C.Z

FIG. 1. Stegodon elephantoides (Clift). MCZ No. 6251, left M2. Lateral view above, crown view below.
FIG. 2. Stegodon elephantoides (Clift). MCZ No. 6254, portion of right mandibular ramus with second molar. Internal lateral view above
All figures two-fifths natural size.
PLATE XXIII

Stegodoninsignis birmanicusOsborn. ANSP No. 14653, left mandibular ramus with second molar. Crown view above, external la
one-half natural size.
PLATE XXIV

Elephas hysudricus Falconer and Cautley. ANSP No. 14654, left mandibular ramus with third molar. Crown view above, internal lat
one-half natural size.
PLATE XXV

6291
M.C.Z.

Elephas hysudricus Falconer and Cautley. MCZ No. 6291, left mandibular ramus with molar. Crown view above, ex-
ternal lateral view below, one-half natural size.
PLATE XXVI

~~~~~~~~~~~~~~~~

FIG. 1, Elephas hysudricus Falconer and Cautley. MCZ No. 6268, upper molar. Lateral view.
FIG. 2. Elephas hysudricus Falconer and Cautley. MCZ No. 6257, left M3. Crown view above, external
lateral view below.
All figures three-fifths natuiral size.
PLATE XXVII

14635
A.N,S.R

6260

FIG. 1. Hemibos triquetricornis Riitimeyer, ex Falconer ms. ANSP No. 14635, cranium with right horn core. Dorsal view
above, lateral view in middle, occipital view below.
FIG. 2. Hemibos triquetricornis Riitimeyer, ex Falconer ms. MCZ No. 6260, portion of cranium. Dorsal view above,
PLATE XXVIII

FIG. 1. Stegodon orientalis Owen. ANSP No. 14626, left DM4. External lateral view on left, crown view on right.
FIG. 2. Stegodon orientalis Owen. MCZ No. 6252, right DM4. Crown view above, ventral view (to show tooth marks made by Hystrix)
below.
FIG. 3. Stegodon orientalis Owen. ANSP No. 14625, right DM4. Crown view above, external lateral view below.
All figures three-fourths natural size.
PLATE XXIX

A.

14627
2

FIG. 1. Stegodon (?) or Palaeoloxodon (?). ANSP No. 14628, tusk.


FIG. 2. Palaeoloxodon namadicus (Falconer and Cautley). ANSP No. 14627, right M2. External
lateral view above, crown view below.
All figures three-fifths natural size.
PLATE XXX

FIG. 1. Palaeoloxodon namadicus (Falconer and Cautley). MCZ No. 6256, left M2. Crown view above,
external lateral view below.
FIG. 2. Palaeoloxodon namadicus (Falconer and Cautley). MCZ No. 6255, left M2. Crown view above,
external lateral view below.
All figures three-fifths natural size.
PLATE XXXI

AN.SP.
14645

FIG. 1. Rhinoceros sp. ANSP No. 14645, right P3. Crown view, natural size.
FIG. 2. Rhinoceros sp. MCZ No. 6278, right lower cheek tooth. Crown view above,
external lateral view below, natural size.
FIG. 3. Cervus sp. ANSP No. 14647, antler fragments. Base of antler, left, two-thirds
natural size; portion of beam, right, showing tooth marks of Hystrix, approximately natural
size.
PLATE XXXII

A,N.S.R 14657 6296


.I

5 14658
A.N.

FIG. 1. Elephas hysudricus Falconer and Cautley. ANSP No. 14657, tooth fragment. Crown view above,
lateral view below.
FIG. 2. Elephas hysudricus Falconer and Cautley. MCZ No. 6296, tooth fragment. Lateral view.
FIG. 3. Elephas hysudricus Falconer and Cautley. ANSP No. 14658, tooth fragment. Lateral view.
All figures three-fourths natural size.
PART IV
FRESH-WATER SHELLS FROM CAVE DEPOSITS IN THE
SOUTHERN SHAN STATES, BURMA

BY J. BEQUAERT
Museum of Comparative Zoilogy, Cambridge, Mass.

Plate XXXIII

LOCALITIES that the indentification of Brotia variabilis was no doubt


correct. Two specimens of Brotia persculpta were
The fresh-water shells obtained by Dr. H. L. Movius,
also found by Dr. Movius in the Tin-Ain cave.
Jr., in the course of his archaeological excavations in 2. Mongta-wa-ku.-This is a very large cave about
the Southern Shan States, are not numerous. Yet
two and a half miles southwest of the town of Taunggyi,
their study has led to some interesting conclusions re-
and about 8 miles (as the crow flies) to the northeast
garding variation, which it seems warranted to pub- of the Tin-Ain caves. It is near the top of the escarp-
lish, particularly in order to justify the nomenclature ment overlooking the Yawnghwe valley, some 1200 ft.
adopted for the several species. above the alluvial plain and approximately 450 ft. above
The material is from three localities, in Yawnghwe
sea-level. Several test pits were dug, but the shells
State, within an area of about 40 square miles. The
sites are fully described elsewhere in this memoir (see were all in surface accumulations mixed with broken
modern pottery of the Buddhist period. Two species
pp. 389-391 of Dr. Movius' report); but a brief ac-
count will be helpful to the malacologist. of Tiaridae were recognized: Melanoides tuberculata
1. Cave of Tin-Amin.-Three miles east of the town subsp. grangcri (Wattebled) and Brotia variabilis
of Yawnghwe and about one mile north of the village (Benson), both living at present in the same region.
of Le-pin, there are two small caves at the base of the No previous examination of these deposits appears to
have been made.
escarpment, about 350 feet above the level of the old
alluvial plain of Lake Inle. Shells were collected in 3. Rock-shelter of Miilg Pawn.-About 28 miles
the south cave only, from a test pit made in a small due east of the town of Taunggyi, to the south of the
side chamber near the right-hand side of the entrance, road, there is a rock-shelter on the west slope of the
together with bones of wild animals, but no achaeologi- Nam Pawn valley. From a test pit about 1 m. deep
cal material. The shells belong to two species: Taia many shells were removed, together with a few broken
intermedia Annandale (fam. Viviparidae) and Brotia bones, some stones which appeared to have been broken
persculpta (Ehrmann) (fam. Tiaridae). The Tin- artificially and some charcoal fragments. There were
Ain caves were investigated by N. Annandale and F. H. no pieces of pottery nor implements. The shells are of
Gravely, in 1917; and again by N. Annandale and H. S. two species of Tiaridae: Brotia baccata (Gould), living
Rao (and their assistants), in March 1922. In the in the same region; and Sulcospira praemordica Tryon,
published accounts (Annandale, 1918, 1919 and 1924; with a subsp. laevitestacea J. Bequaert, the typical form
Annandale and Rao, 1925) they are referred to as two also now living in Burma.
small limestone caves some 3 miles east of the town of
Yawnghwe, on the slope of the valley of the Hsin- Annandale believed that the shells of the cave de-
Dawng, a stream running down the escarpment toward posits of Tin-Ain had been buried there by a natural
the alluvial plain of Yawnghwe. Annandale (1918: means, namely by floods, a view which no doubt influ-
143; 1919: 218) listed from these caves three sup- enced him to regard the Taia as representing three
posedly "fossil" (i.e. extinct) species of Taia: T. obesa, "extinct" species. Curiously enough he did not reach
T. cylindrica and T. conica, and one species of Tiaridae: the same conclusion for the two species of Tiaridae, al-
Brotia variabilis (a living species). He later found a though he admitted that they were contemporary of
few imperfect shells of a second species of Tiaridae, on the Taia. To account for the absence in the Tin-Ain
which he based the new form planicostata of his Ac- caves of the smaller fresh-water snails (Lymnaeidae and
rostoma elongatum (Brotia persculpta of the present Hydrobiidae), commonly associated with Taia and
paper). As shown in the sequel, Annandale's three Tiaridae in the superficial and lake deposits of the same
supposedly "fossil" Taia appear to be individual vari- region, Annandale (1918: 143) suggested that the
ants of the living Taia intermedia. His figure shows larger snails were able to float on the surface of a lake
431
432 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
or pool until they were stranded at the base of lime- Taia naticoides race intermediaAnnandale and Rao, 1925, Rec.
stone rocks, where they were buried; their unworn Indian Mus., XXVII: 121, fig. 8 (living in Lake Inle, S. Shan
condition proving that they were not carried by running States).
Paludina naticoides "Theobald" Hanley and Theobald, 1873,
water.' It seems much simpler to regard the ac- Conchol.Indica, Expl. of Plates: 32, P1. 76, fig. 4 only (Shan
cumulations of Taia and Tiaridae in the caves and rock States). Not of G. Pfeiffer, 1828, nor of Theobald, 1866.
shelters of the Southern Shan States, as the work of Vivipara naticoides "Theobald"Kobelt, 1907, Syst. Conch. Cab.,
Man, who discarded the shells after using the snails as I, Abt. 21a: 149, P1. 30, fig. 5 only (copy of Hanley and
Theobald'sfig. 4).
food. This explains satisfactorily (a) the occurrence
Taia naticoides "Theobald"Annandale, 1918, Rec. Indian Mus.,
of quantities of shells in shelters at levels far above XIV; 126, P1. 15, figs. 16-17, P1. 16, figs. 3-6, and P1. 18,
those of old lake or river alluvial deposits; (b) the un- figs. 1-3 (living at Yawnghwe and in the He-Ho plain, S.
worn, though often damaged, condition of the shells, Shan States). Ehrmann, 1922, Sitzungsber. Naturf. Ges.
together with the absence of opercula, the latter having Leipzig, XLV-XLVIII, (1918-1921): 23, PI., fig. 9 (living
in Loikaw River, S. Shan States).
been removed with the animals; and (c) the lack of
small fresh-water snails, which were not used as food. Vivipara shanensis "Theobald"Kobelt, 1909, Syst. Conch. Cab.,
I, Abt. 21a: 411, P1. 77, figs. 4-5 (living; Shan States).
That the ancient aborigines of Burma included fresh- Preston, 1915,Fauna Brit. India, Moll., Freshw. Gastr. Pelec.:
water snails in their diet is scarcely to be doubted, for 93. Not of Theobald, 1876.
even nowadays certain species are commonly eaten in Taia shanensis "Kobelt" Annandale, 1918, Rec. Indian Mus..
XIV: 129, P1. 15, figs. 14-15, P1. 16, fig. 10, and PI. 18, figs.
Burma, Siam and Indo-China, as well as elsewhere in 4-6 (living in marginal zone of Lake Inle, S. Shan States);
the Orient. J. de Morgan (1885, Bull. Soc. Zool.
1924, Proc. R. Soc., London, Ser. B, XCVI: 69, fig. 6D.
France, X: 420-421) mentions several species of Brotia Annandale and Rao, 1925, Rec. Indian Mus., XXVII: 122,
used as food in Perak. According to Tapparone- fig. 9a. Not of Theobald, 1876.
Canefri (1889, Annl. Mus. Civ. Genova, XXVII: 300- Taia shanensisform callosa Ehrmann,1922,Sitzungsber.Naturf.
Ges. Leipzig, XLV-XLVIII, (1918-1921): 26 (living in
303), L. Fea bought in the native market at Mandalay Loikaw S. Shan States).
four species of Tiaridae, including a form of Melanoides Taia obesaRiver,
Annandale, 1918, Rec. Indian Mus., XIV: 128, P1.
tuberculata. A. Billet (1898, Bull. Scientif. France 15,.fig. 19, and P1. 16, fig. 2 (dead in cave deposits near Hsin-
Belgique, XXVIII: 329) noted that the natives of Dawng stream, S. Shan States). Annandale and Rao, 1925,
northern Tonkin eat large quantities of a species of op. cit., XXVII: 126.
Viviparus. The Laos of Siam use as food Viviparlts Taia cylindrica Annandale, 1918, Rec. Indian Mus., XIV: 130,
P1. 15, fig. 9, and P1. 17, fig. 2 (dead in cave deposits near
bengalensis and Brotia variabilis (W. M. Daly, 1903, Hsin-Dawng stream, S. Shan States). Annandale and Rao,
Proc. Mal. Soc. London, V, pt. 5: 281-282). 1925, op. cit., XXVII: 126.
The finding by Dr. Movius of shells of Tiaridae as- Taia lacustris Annandale, 1918, Rec. Indian Mus., XIV: 131,
sociated with evident signs of human occupation, lends P1. 15, figs. 10-11, P1. 17, fig. 1, and P1. 18, figs. 7-9 (dead in
further support to the view that Man was responsible alluvium, He-Ho plain, S. Shan States).
for selecting and transporting the fresh-water snails now Taia naticoides race lacustris Annandale and Rao, 1925, Rec.
Indian Mus., XXVII: 121.
found buried in the limestone caves of the Southern
Taia analoga Annandale, 1918, Rec. Indian Mus., XIV: 132,
Shan States. The deposits of the three sites in which P1. 15, figs. 6, 7, 12, and P1. 17, figs. 3-4 (dead in alluvium,
shells were found are of recent age, Dr. Movius in- He-Ho plain, S. Shan States).
forms me. They are apparently post-Pleistocene and Taia conica Annandale, 1918, Rec. Indian Mus., XIV: 133, P1.
perhaps not over a few thousand years old. Hence it 15, fig. 8, and P1. 17, fig. 8 (dead in cave deposits near Hsin-
is extremely unlikely that the shells could belong to Dawng stream, S. Shan States). Annandale and Rao, 1925,
"extinct" species. On purely conchological evidence, Taia op. cit., XXVII: 126.
intha Annandale, 1918, Rec. Indian Mus., XIV: 135, P1.
I had, moreover, reached the conclusion that the three 15, figs. 1-3, P1. 17, fig. 7, and P1. 18, figs. 10-12 (living in
supposedly "extinct" species of Taia from the Tin-Ain central region of Lake InlI, S. Shan States); 1924,Proc. Roy.
caves were merely variants of a species now living in Soc., London, Ser. B, XCVI: 62, fig. 2D, 67, fig. 5, and 69,
Lake Inle. fig. 6F. Annandale and Rao,- 1925, Rec. Indian Mus.,
XXVII: 123, figs. 9b and 10.
DESCRIPTION OF SHELLS Taia elitoralis Annandale, 1918, Rec. Indian Mus., XIV: 134,
P1. 15, figs. 4-5, P1. 17, figs. 5-6, and P1. 18, figs. 13-14 (living
VIVIPARIDAE in intermediate zone of Lake Inle; pool near north end of
Lake InlI; canal at Yawnghwe; S. Shan States); 1924, Proc.
Taia intermedia Annandale R. Soc., London, Ser. B, XCVI: 69, fig. 6E. Annandaleand
Plate XXXIII, Figures 1-9 Rao, 1925, Rec. Indian Mus., XXVII: 123.
Taia intermediaAnnandale, 1918, Rec. Indian Mus., XIV: 128, Taia crassicallosaAnnandale and Rao, 1925, Rec. Indian Mus.,
P1. 15, fig. 13, and P1. 16, figs. 7-9 (dead in alluvium, He-Ho XXVII: 124; figs. 11-14 (living in Yawnghwe River near
S. Shan Rec. Geol. Tai-O, S. Shan States).
plain, States); 1919, Survey India, L, pt.
3: 239, P1. 31, fig. 10.
Some 20 shells of Taia were dug up in one of the
1 Elsewhere, however, the somewhat contradictory statement Tin-Ain caves. As shown by the figures, they vary in
is made that the shells had been washed into the Tin-Ain caves,
mainly through holes in the roof (Annandale, 1924: 71; Annan- size, shape, development of the scaly knobs, and thick-
dale and Rao, 1925: 102). ness of the umbilical callus. So many transitions con-
BEQUAERT: FRESH-WATER SHELLS 433

nect the extreme forms, that I am forced to refer them TIARIDAE (MELANIIDAE)
all to one species. The most common form (Figs. 1- Brotia (Antimelania) variabilis (Benson)
3) agrees with Annandale's figure of T. cylindrica.
Fig. 7 is apparently Annandale's T. conica, while Figs. Plate XXXIII, Figures 11-16
4-6 are his T. obesa. Moreover, I am unable to sepa- Melania variabilis Benson, 1836, Jl. Asiat. Soc. Bengal, V: 746
rate these Tin-Ain shells from the several other simi- (Gumti River at Jonpur; Tally's Nullah near Calcutta; also a
larly sculptured Taia described by Annandale from var. A from Hughli River, Calcutta; and vars. B, C and D
from Sylhet; all localities in northeastern India). Souleyet,
living and dead material in the same region, all of 1852, Voyage La Bonite, Zool., II: 545, P1. 31, figs. 12-15.
which are listed in the synonymy. The oldest valid
Annandale, 1918, Rec. Indian Mus., XIV: 114, P1. 12, fig. 8
name for this complex appears to be Taia intermedia, (dead in cave deposits near Hsin-Dawng, S. Shan States).
at first described from dead shells, but later found alive Melania species A, Benson, 1830, Gleanings in Science, I, P1. 7.
in Lake Inle. The extreme likeness of my Fig. 3 and Melania (Melanoides) variabilis Brot, 1875, Syst. Conch. Cab.,
I, Abt. 24: 85, P1. 10, figs. 1, 1 a-d.
of Annandale's figures of T. intermedia may be noted. Melania varicosa Troschel, 1837,Arch. f. Naturgesch.,III, pt. 1:
T. intermedia is characterized by the short or elon- 174 (Ganges River, India). Philippi, 1843, Abb. Beschr.
gate conical shape, the whorls little convex or more or Conch., I: 59, P1. 2 (Melania), figs. 2 (cotype received from
less flattened, and the basal callus completely closing Troschel) and 3.
Melania variabilis var. varicosa Hanley and Theobald, 1874,
the umbilicus and occasionally very heavy, broad and Conchol.Indica, Expl. of Plates: 44, P1. 109, fig. 2.
prominent (crassicallosa Annandale and Rao; see my Melanoidesindica "Eydoux" H. and A. Adams, 1854, Gen. Rec.
Figs. 8-9). The sculpture consists of a variable num- Moll., I: 296, and III, P1. 31, figs. 5 a-c. [There is no
ber of prominent spiral ribs bearing small or large Melania indica "Eydoux" or "Souleyet." The latter used
only the French vernacular name "Mdlanie indienne" on his
nodules, either rounded or more or less raised into P1. 31 for the species which he referred to Melania variabilis
scales or squamae, particularly on the peripheral ridge. in the text.]
On specimens from cave deposits the squamae are often Melania variabilis var. echinata "Benson"Hanley and Theobald,
broken or worn through handling by man. Moreover, 1874, Conchol. Indica, Expl. of Plates: 44, P1. 109, fig. 3
(Assam).
the nodules and scales vary much in living specimens Melania variabilis var. cincta "Benson" Hanley and Theobald,
taken from Lake Inle, as shown by Annandale's figures. 1874, Conchol. Indica, Expl. of Plates: 44, PI. 109, fig. 5
Much confusion has arisen from the authors' mis- (Assam). Not Melania cincta Is. and H. Lea, 1851.
Melania variabilis var. aspera "Benson" Hanley and Theobald,
taking this species for Theobald's Paludina naticoides, 1874, Conchol. Indica, Expl. of Plates: 44, P1. 109, fig. 6
the correct valid name of which is Taia shanensis (Hindustan). Not Melania aspera Lesson, 1831.
(Theobald). As originally described and figured Melania crotula "Rafinesque"Brot, 1875, Syst. Conch. Cab., I,
(1866, Jl. Asiat. Soc. Bengal, XXXIV, pt. 2, No. IV: Abt. 24: 85 (as a synonym of Melania variabilis).
Melania variabilis var. vittata Theobald, 1866, Jl. Asiat. Soc.
274, P1. 9, figs. 1-3; fig. 1 is var. fasciata; figs. 2-3 are
Bengal, XXXIV, pt. 2, No. IV: 273, P1. 9, fig. 4 (Shan
var. carinata; Shan States), this is less conical than T.
States). Not Melania vittata Anthony, 1854; nor of Brot,
intermedia, with more convex whorls, and weaker 1860.
umbilical callus. The surface is either almost smooth Melania variabilis subvar. fasciata G. Nevill, 1885, Hand List
or more or less spirally ribbed, but the ribs lack the Moll. Indian Mus., II, (1884): 256 (Bhootan, India). Not
Melania fasciata Menke, 1828.
nodules or scaly protuberances of T. intermedia. Melania variabilis var. glabra Theobald, 1866, Jl. Asiat. Soc.
Vivipara shanensis Theobald (1876) was proposed as Bengal, XXXIV, pt. 2, No. IV: 273 (tepid springs of Nam-
a synonym or substitute for naticoides Theobald moo, Shan States). Not Melania glabra Is. Lea, 1841.
Melania variabilis subvar.microstomaG. Nevill, 1885,Hand List
(1866); but Kobelt's (1907) Vivipara shanensis is Moll. Indian Mus., II, (1884) : 261 (Sylhet; with description).
T. intermedia. In fact Kobelt's figures (P1. 77, figs. Not Melania microstomaIs. and H. Lea, 1851.
4-5) agree well with some of the specimens from the Melania variabilis var. semilaevigata G. Nevill, 1885, Hand List
Tin-Ain caves. On the other hand, I refer Kobelt's Moll. Indian Mus., II, (1884): 254 (Sylhet; Daarbund; Sil-
curi, Cachar; all in northeasternIndia).
figures 4, and 6 to 11 of Plate 30 to true T. shanensis Melania variabilis subvar. subspinosaG. Nevill, 1885, Hand List
(Theobald). The following names are additional Moll. Indian Mus., II, (1884): 253 (Tank opposite Bengal
synonyms of T. shanensis: Vivipara naticoides var. Club, Calcutta; without description).
obsolescens E. von Martens, 1899; Vivipara theobaldi Melania variabilis subvar. subtuberculataG. Nevill, 1885, Hand
List Moll. Indian Mus., II, (1884): 252 (Calcutta; with
Kobelt, 1907; Vivipara noetlingi Kobelt, 1908; Taia
theobaldi var. leonardi Rao, 1928; Taia theobaldi form description).
Melania variabilis var. subvaricosa G. Nevill, 1885, Hand List
globosa Rao, 1928; and Taia theobaldi form niger Rao, Moll. Indian Mus., II, (1884): 253 (Arakan and Pegu?;
1928.2 Sibsagar; Namtsik River; Samaguting; Kusiyra River,
Sylhet; Sudiya near Darjeeling; all in northeastern India;
2Paludina naticoides Theobald (1866) is antedated by Palu- with description).
dina naticoides G. Pfeiffer (1828). Theobald's names fasciata Melania (Melanoides) tourannensis "Eydoux and Souleyet"
and carinataare also not available,having been used before 1866 Tapparone-Canefri,1889, Ann. Mus. Civ. Genova, XXVII:
in combination with Paludina. I am unable to find a "Palu- 302 (Bhamo, Upper Burma). Not of Souleyet, 1852, which
dina naticoides Ferussac," as quoted by G. Pfeiffer. is probablya distinct subspeciesof B. variabilis.
434 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Melania (Brotia) hungerfordiana "Nevill" E. v. Martens, 1899, ends of the vertical ribs stop abruptly and sharply
Arch. f. Naturgesch., LXV, pt. 1, Heft 1: 37, P1. 4, fig. 6 some distance below the suture, exactly as in v. Mar-
(near Meungyaw, east of Lashio; affluentof the Salween R.; tens' (1899) figure of "hungerfordiana." Annandale's
Nampai R.; all in N. Shan States). Not M. hungerfordiana
G. Nevill, 1885,which appearsto be relatedto Sulcopsiraprac- single shell from the Tin-Ain cave (1918, P1. 12, fig. 8)
mordica (Gould). was like some of the Mongta-wa-ku specimens.
The foregoing references are only to individual vari-
ants of the typical form of the species in Northeastern Brotia (Antimelania) persculpta (Ehrmann)
India, Assam and Burma. B. variabilis is so protean,
even in the same colony or stream, that none of the Plate XXXIII, Figures 17-18
names listed above seem to deserve a separate stand- Melania baccata subsp. elongata Annandale, 1918, Rec. Indian
ing, even as varieties or mutations. In addition, I re- Mus., XIV: 115, P1. 12, figs. 3-7 (living in Yawnghwe River;
gard the following names as based upon variants, eco- dead in alluvium of the He-Ho plain; S. Shan States). Not
logical forms or geographical races of this species: Melania elongata Is. Lea, 1831.
Melania menkiana Is. Lea, 1844 (= M. plicata Is. Lea, Acrostomnaelongatum Annandale and Rao, 1925, Rec. Indian
1838, not of Menke, 1828; India); Melania herculea Mus., XXVII: 117 (living at Tai-O, Yawnghwe State, and
in He-Ho plain; S. Shan States).
Gould, 1846 (Lower Burma); Melania tourannensis Melania persculpta Ehrmann, 1922, Sitzungsber. Naturf. Ges.
Souleyet, 1852 (Cochin China, Laos); Melania spinosa Leipzig, XLV-XLVIII, (1918-1921): 18, PI., fig. 8 (living
Hanley, 1856; Melania swinhoei H. Adams, 1870 in Loikaw River, S. Shan States; with form biserialis and
(Hainan); Melania herculea var. sowerbii Hanley and form triscrialis).
Theobald, 1874 (Tenasserim); Melania jugicostis Acrostomnaelongatumlform planicostata Annandale and Rao,
1925, Rec. Indian Mus., XXVII: 117 (living at He-Ho; dead
Hanley and Theobald, 1874 (Tenasserim); Melania in Hsin-Dawng Cave; S. Shan States).
julieni Brot, 1875 (Cambodia, Laos); Melania godwini
Brot, 1875 ( Melanoides hanleyi Godwin-Austen, Two very incomplete shells from the Tin-Ain cave
1872; not of Brot, 1860; North Cachar, India); clearly represent B. persculpta (Ehrmann). Both be-
Melania sumatrensis Brot, 1875 (Sumatra); Melania
long to Ehrmann's form biserialis, which was renamed,
boeana Brot, 1881 (Sumatra); Melania provisoria rather unnecessarily, planicostata by Annandale and
Brot, 1881 (Sumatra); Sermyla chaperi de Morgan, Rao.
1885 (Perak); Sermyla perakensis de Morgan, 1885
The true status of persculpta is far from clear, as it
(- S. kintanensis de Morgan, 1885; Perak); Melania seems to combine certain features of B. variabilis with
sumatrensis var. mitescens Schepman, 1886 (Sumatra);
other more characteristic of B. baccata. Provisionally,
Melania subplicata Schepman, 1886 (Sumatra); Me-
lania curvicosta v. Martens, 1897 (Sumatra); Melania however, I retain it as a distinct species. It seems diffi-
cult, moreover, to separate persculpta from Melania
papillosa v. Martens, 1897 (Sumatra); Melania stricti- variabilis var. binodulifera G. Nevill (1885; based in
costa v. Martens, 1897 (Sumatra); and Melania curvi-
costa var. prestoniana Bullen, 1906 (Sumatra). A few part on Hanley and Theobald, 1873, Conchol. Indica,
of these names, such as tourannensis, herculea and PI. 72, fig. 7, and P1. 75, figs. 5 and 7), or from
sumatrensis, apply perhaps to valid geographical races. Acrostoma baccata form bituberculata Rao (1928, Rec.
Some other Brotia, such as B. pontificalis (v. d. Busch) Indian Mus., XXX: 444, figs. 17 a, b, d, e; N. Shan
(Borneo), B. siamensis (Brot) (Siam, Indo-China), States).
B. reevei (Brot) (Burma), and B. episcopalis (Is. and
H. Lea) (Malacca, Siam, Indo-China), are very Brotia (Antimelania) baccata (Gould)
closely related to B. variabilis, but are perhaps best re-
garded as distinct species.3 Plate XXXIII, Figures 19-23
A series of 19 shells of B. variabilis, some of them
Melania baccata Gould, 1847, Proc. Boston Soc. Nat. Hist., II:
much broken, were dug up in the cave M6ngta-wa-ku. 219 (Thoung-Yin River, Burma); 1862, Otia Conchol.: 200.
Most of them are moderately large and belong to the Hanley and Theobald, 1873, Conchol. Indica, Expl. of Plates:
typical form of the species in which the vertical ribs 32, P1. 75, figs. 1 and 4. Brot, 1875, Syst. Conch. Cab., I,
are fairly numerous and regular in the upper half of Abt. 24: 81, P1. 9, fig. 6.
the body-whorl, while the spiral ribbing of the lower Melania (Brotia) baccata E. v. Martens, 1899, Arch. f. Natur-
half of the body-whorl is distinct. In a few the verti- gesch., LXV, pt. 1, Heft 1: 35 (Myitnge, an affluentof Irra-
wady, near Bagwyo and Thibaw; Lashio R.; Nampari R.; all
cal ribs are broken up into nodules, being more as in in N. Shan States).
tourannensis. There are no true tubercles or spines, Acrostoma baccataRao, 1928, Rec. Indian Mus., XXX: 442, fig.
however; although in some shells (Fig. 13) the upper 17c (Namtu River at Hsenwi and Hsipaw; near Tangyan;
3 Melania carolinae J. E. Gray (1834, in Griffith, Cuvier's Lashio; all in N. Shan States).
Animal Kingdom, XII, Moll: 598, P1. 13, Moll., fig. 3; no Melania variabilis var. baccifera Theobald, 1866, Jl. Asiat. Soc.
locality) appearsto have been based upon an abnormalspecimen Bengal, XXXIV, pt. 2, No. IV: 274, P1. 9, fig. 5 (Nammah
of Brotia dactylus (Is. and H. Lea), of the Philippines, not of Stream, S. Shan States).
B. variabilis. Acrostoma baccifera Rao, 1928, Rec. Indian Mus., XXX: 447.
BEQUAERT: FRESH-WATER SHELLS 435
Melania variabilis var. turrita Theobald, 1866, J1. Asiat. Soc. Sulcospira praemordica (Tryon)
Bengal, XXXIV, pt. 2, No. IV: 273, P1. 9, fig. 6 (Shan Plate XXXIII, Figures 24-30
States). Not Melania turrita Mousson, 1849.
Melania variabilis var. pyramidalis Theobald, 1866, Ji. Asiatic Melania praemordicaTryon, 1866, Amer. Jl. Conch., II, pt. 2:
Soc. Bengal, XXXIV, pt. 2, No. IV: 274, PI. 9, fig. 7 (Nam- 111, P1. 10, fig. 3 (Burma). Hanley and Theobald, 1876,
mah Stream, Shan States). Not Melania pyramidalis More- Conchol. Indica, Expl. of Plates: 61, P1. 153, fig. 2 (Pegu).
let, 1849. Tapparone-Canefri,1874, Zool. Viaggio della Magenta, Mala-
Melania baccata var. pyramidalisE. v. Martens, 1899, Arch. f. col.: 45, P1. 1, fig. 7 (young). Brot, 1875, Syst. Conch. Cab.,
Naturgesch., LXV, pt. 1, Heft 1: 36 (Nampai R., Lashio R. I, Abt. 24: 108, P1. 13, figs. 8 and 8a.
near Lashio; N. Shan States). Melania baccata var. praemordicaE. v. Martens, 1899, Arch. f.
Acrostoma baccata form pyramidalis Rao, 1928, Rec. Indian Naturgesch., LXV, pt. 1, Heft 1: 36 (Chindwin River, an
Mus., XXX: 444 (dead in alluvium near Namtu River, affluentof the Irrawady, near Natu, N. Shan States).
Hsenwi, and near Hsipaw, N. Shan States). Tiara (? Acrostoma) praemordica Preston, 1915, Fauna Brit.
Melania baccata var. fusiformis Hanley and Theobald, 1873, India, Moll., Freshwater Gastr. Pelec.: 31.
Conchol. Indica, Expl. of Plates: 32, P1. 75, fig. 2 (Shan In naming this species, I have had the advantage of
States). Not Melania fusiformis Is. Lea, 1841.
Melania episcopalis "Lea" Hanley and Theobald, 1873, Conchol. studying four paratypes (from the Wheatley Collec-
Indica, Expl. of Plates: 31 and 32, P1. 72, fig. 7, and P1. 75, tion), at the Museum of Comparative Zoology. I fig-
figs. 5 and 7 (Diyung River, N. Cachar, India). Not Me- ure one of these (Fig. 24) for comparison. It will be
lania episcopalis Is. and H. Lea, 1851.4 noted that Tryon's original figure is good, except that
M4elania baccata subvar. recta G. Nevill, 1885, Hand List Moll. the aperture is drawn too narrow. Hanley and Theo-
Indian Mus., II, (1884): 262 (Upper Salween River, Burma; bald's figure is even better in some respects; but Brot's
without description,but with a reference to Brot's P1. 9, fig.
6 of 1875). Not Melania recta Is. and H. Lea, 1851. P1. 13, fig. 8, is a poor and misleading copy. The spe-
Acrostoma baccata form recta Rao, 1928, Rec. Indian Mus., cies is characterized by the shape, the unusually large
XXX: 445. body-whorl, the narrow and somewhat grooved base
Melania (Melanoides) subasperata G. Nevill, 1885, Hand List of the aperture, and the peculiar low and broad spiral
Moll. Indian Mus., II, (1884): 262 (Shan States; with de- ribs, which are slightly wavy at regular intervals, but
scription and reference to Theobald'sP1. 9, fig. 5, of 1866). do not form knobs or nodules. The operculum is un-
Melania iravadica Blanford (1869, Proc. Zool. Soc. known. From shell characters alone, S. praetmordica
London: 445; Upper Irrawady R. in Upper Burma and seems to be related to S. hiigelii (Philippi), rather than
Yunnan) appears to be a fairly distinct subspecies of to the species placed in Brotia, and I refer it provision-
Brotia baccata, as first recognized by G. Nevill (1877, ally to Suilcospira. I suspect that Melania (Pachychilus)
Jl. Asiat. Soc. Bengal, XLVI: 33). hungerfordiana G. Nevill (1885), of Upper Burma and
Melania henriettae J. E. Gray (1834, in Griffith, Pegu, was based upon a large, smooth specimen of S.
Cuvier's Animal Kingdom, XII, Moll.: 598, P1. 13, praemordica, like those described below,6 while Melania
Moll., fig. 2; no locality), of southern China, is ap- liimborgiHanley (1879), of the Mule-it Range, Tenas-
parently also a subspecies of Brotia baccata. If fur- serim, was possibly the young of this smooth form.
ther study confirms this opinion, the species will have The rock-shelter of M6ng Pawn yielded about 20
to bear the name Brotia henriettae (J. E. Gray), with specimens of shells with the characteristic shape of
the Burmese race subordinated to it as subsp. baccata S. praezmordica,although none show the pronounced
(Gould). Melania reticulata Is. and H. Lea (1851) broad spiral ribs of the living paratypes. In two speci-
is a synonym of B. henriettae.5 mens these ribs are indicated (Figs. 25-26), although
I refer to B. baccata some 20 specimens from the they scarcely show in the photographs.
rock-shelter of M6ng Pawn, which have the typical All the other specimens are smooth, except for oc-
sculpture of conical or rounded nodules in 3 or 4 spiral casional, irregular, axial growth-striae. It may be
rows between the suture and the periphery. Below the useful to distinguish the smooth form by name, as
periphery the body-whorl bears spiral ribs, also some- laevitestacea (subsp. nov.), at least until the variation
what beaded occasionally. The shells vary from tur- of the species in life is better known. The holotype
reted to pyramidal. Some of them agree well with (Fig. 28) measures 39 mm. by 22 mm., the aperture
Hanley and Theobald's P1. 75, figs. 1 and 2 (1873). 20 mm. by 12 mm., one and a half whorls remaining.
4 The true M. episcopalis Is. and H. Lea, of Malacca, Siam
and Indo-China,appearsto be the continentalrepresentative
of Melanoides tuberculata subsp. grangeri
Brotia pontificalis (v. d. Busch), of Borneo. It is very closely (Wattebled)
related to Brotia variabilis (Benson).
5The two "fossil" species of Brotia described by Annandale Plate XXXIII, Figure10
(1924, Rec. Geol. Survey India, LV, pt. 2, (1923): 98-100), Melania grangeri Wattebled, 1884, Jl. de Conchyl.,XXXII: 127,
from the Oil Measures of the Dawna Hills, Tenasserim, appear P1. 6, figs. 3-3a (Long-Xuyen, Cochin China). Brot, 1886,
to be closely related to living forms. His Acrostoma inter- RecueilSuisseZool.,IV: 95, P1.6, fig. 2 (Siam).
medium is scarcely separable from Brotia baccata; while his
Acrostoma cotteri shows more affinity to Brotia variabilis. The 6The shell figured by E. v. Martens in 1899 as M. hunger-
age of the Dawna Oil Measures is uncertain,but Miocene at the fordiana was not Nevill's species, but a variant of Brotia vari-
very oldest and probably much younger. abilis (Benson).
436 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Melania paviei Morlet, 1884, J1. de Conchyl., XXXII: 397, P1. Melanoides tulberculata sulsp. pyramis (Benson,
12, figs. 5-5a (Kah Sutine, Cambodia). 1836) (--Melania adspersa Troschel, 1837; llIelania
Melania (Striatella) tuberculatavar. myadoungensisG. Nevill,
1885, Hand List Moll. Indian Mus., II, (1884): 245 (Mya- tigrina Hutton, 1849), of India proper, sometimes
doung, Upper Burma). reaches about the size of grangeri, but the later whorls
Melania (Striatella) tuberculata var. subplicifera G. Nevill, of adult shells (particularly the body-whorl) are
1885, Hand List Moll. Indian Mus., II, (1884): 245 (Mya- markedly and regularly convex and separated by deep
doung, Upper Burma).
Melania (Striatella) tuberculataG. Nevill, 1877, Jl. Asiat. Soc. sutures. It would seem that the areas of pyramis and
Bengal, XLVI, pt. 2: 32 (Irrawady River at Myadoung, grangeri scarcely overlap.
Burma). Tapparone-Canefri,1889, Ann. Mus. Civ. Genova, One specimen of only 2 preserved whorls, fromi the
XXVII: 302 (Mandalay, Burma). cave of Mong Ta-wa-ku, agrees well with some living
Melania (Melanoides) tuberculataE. v. Martens, 1899, Arch. f.
Naturgesch.,LXV, pt. 1, Heft 1: 37 (Myitnge near Thibauw; material of grangeri, from Cambodia and Tonkin. at
stream near Bangyo, N. Shan States; Mandalay). the Museum of Comparative Zoology. One of these.
Melania tuberculataAnnandale, 1918, Rec. Indian Mus., XIV:
114, P1. 12, figs. 1-2 (living in Inle Lake and Yawnghwe R.;
kindly sent by Dr. Mermod, was called gralngri in
dead in alluvium of He-Ho plain; S. Shan States). Brot's collection.
Melanoides tuberculatusAnnandale and Rao, 1925, Rec. Indian
Mus., XXVII: 118 (dead in clay-pit above He-Ho gorge, S.
Shan States). BIBLIOGRAPHY OF PART III
Melania citrinoides Brot (1886), of Siam, is prob- ANNANDALE,N. 1918. Aquatic mollusks of the Inle Lake and
ably also a synonym of grangeri. This race, seemingly connectedwaters. Rec. Indian Mus., vol. 14: 103-182, Pls.
common throughout Burma, Siam and Indo-China, ap- 10-19.
pears to be fairly well characterized by the large size . 1919. The gastropod fauna of old lake-beds in Upper
(as much as 35 to 40 mm. for adult shells with eroded Burma. Rec. Geol. Survey India, vol. 50: 209-240, Pls.
summit and 6 or 7 whorls preserved) and the later 31-33.
. 1924. The evolution of the shell-sculpturein freshwater
whorls being distinctly flattened along the sides, with snails of the family Viviparidae. Proc. Roy. Soc., LTondon,
the sutures very shallow. The sculpture is as variable (B), vol. 96: 60-76.
as usual in this species, but in full-grown shells the ANNANDALE, N., and H. S. RAO. 1925. Further observations
body-whorl is as a rule deeply spirally grooved and on the aquatic gastropodsof the Inli water shed. Rec. In-
without axial ribs. dian Mus., vol. 27: 101-127.
-? -
"o" a a C

m _, S *T

0DC O-'sro

-* RH o ca

(D < - .
" c?I

.S-
La^

I _

", - '

Ot

ta

o-S

W S
^ _^
c-
1^
PART V
PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA
BY HELLMUT DE TERRA

Plates XXXIV-XXXV

I. REVIEW OF THE DISCOVERIES more highly developed human of the Homo soloensis
Ever since Professor Eugene Dubois found the fa- had type. It seemed that the long-lost trail of Java Man
mous skullcap of Pithecanthropus erectus, the "Java finally been rediscovered.
While such speculations were running high, news
Ape-Man," in 1891, the quest for Fossil Man has never came from eastern Java to the effect that a small fossil
quite ceased on the island of Java. Dubois himself, human skull had been found near
convinced that more fossils could be found, undertook and Modjokerto (Fig. 100
P1. XXXV, Fig. 4), some 25 miles west of Soe-
additional excavations, but these did not add much to
The great age
our knowledge of Fossil Man, except for vertebrate re- rabaja (von Koenigswald, 1936-b).
(Lower Pleistocene) assigned to this fossil, known as
mains of extinct mammals of the Pleistocene period.
Homo modjokertensis, added to the prospects of further
Between his latest explorations and subsequent research discoveries in the Pleistocene beds of the island. Now
there was a lull of six years. In 1907 Frau Lenore that each of
the three divisions of this epoch had yielded
Selenka of Munich, widow of a distinguished anatomist, different human
remains, it was fairly safe to pre-
organized an expedition to find more of the Java Ape- dict that contained an unbroken record of man's
Man. These new excavations at Trinil in central Java Java
evolution.
(Fig. 100 and P1. XXXIV, Fig. 1) yielded a large Upon his return from the "Symposium of Early
collection of fossil material, but with the exception of
Man," held in March 1937 under the auspices of the
one human tooth found at Sonde, no additional evidence
Academy of Natural Sciences of Philadelphia, Dr.
on Pithecanthropus came to light. Nearly twenty-five von
Koenigswald found a jaw fragment of Pithecan-
years later, in 1931, the late Mr. C. ter Haar of the thropus, more perfect than Dubois' specimen from Ke-
Dutch Geological Survey ("Dienst van den Mijnbouw")
deng Brubus, among the fossils collected during his ab-
discovered near Ngandong (P1. XXXIV, Fig. 2), not sence. The new fossil came from
Sangiran, near Solo
far from Trinil, a richly fossiliferous site in a terrace
(P1 XXXIV, Fig. 3), although its exact location was
deposit, containing skeletal remains of an extinct human unknown. Dr. von Koenigswald (1937-a) pointed out
species (ter Haar, 1934). Subsequent excavations, car- that this jaw could not have come from the black clay
ried out here by W. F. F. Oppenoorth (1932; 1936; of Lower Pleistocene
age, which underlies the beds con-
1937), produced not less than eleven skulls with "ne- taining the Trinil fauna at Sangiran, but that it must
anderthaloid"affinities (Homo soloensis), in association have been washed out of the Middle Pleistocene de-
with bone tools and bones of Late Pleistocene mammals. posits. This region had
long been known for its wealth
It was this discovery which initiated a new phase of of fossils, and it was here that Dr. von
Koenigswald
organized study on the problem of Java Man and his had also found the first stone artifacts in 1934.
geological background. Two months later, in August 1937, a native collector
In 1936 Dr. von Koenigswald (1936-a), as palaeon- brought some fragments of a new
Pithecanthropus skull
tologist of the Geological Survey of the Netherlands to the geological laboratory at Bandoeng. Immediate
East Indies, announced the presence of primitive stone inspection of the site (P1. XXXIV, Fig.
4) showed
tools in strata containing the Trinil fauna at Sangiran, that they belonged to a perfect skull, found near Sangi-
near Solo (Soerakarta). Soon afterwards, he and Mr. ran along the bank of the Tjemoro River, but that the
M. W. F. Tweedie (von Koenigswald, 1936; 1936-a) natives had broken it up in their eagerness to sell each
found Palaeolithic implements in a region lying south piece at the price of a complete specimen! Luckily all
of Solo, near Patjitan in south central Java (PI. II, the fragments were recovered-twenty-nine in all-
Figs. 1 and 2). At once the question arose as to within a radius of 100 feet of the block of sandstone in
whether these tools had been made by Java Man, or which the complete skull had been found. In the labo-
whether they were to be regarded as the products of a ratory at Bandoeng the fossil was restored (von Koe-
1938; 1938-a), and it is now called Pithecan-
1This is not intendedto be a full reporton Java Man, but nigswald,
rathera scientificaccountof an excursionmadein the company thropus Skull II.
of Dr. P. Teilhardde ChardinandDr. H. L. Movius,Jr., under In July 1938, shortly after our visit to Java, a third
the guidanceof Dr. G. H. R. von Koenigswaldin April 1938. calvarium of Pithecanthropus was discovered in the
437
FIG. 100. Map of Java Showing the Location of Sites Mentioned in the Text.
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 439
cliffs near Sangiran (von Koenigswald and Weiden- found, which Dubois originally described under the
reich, 1938). In this specimen, which represents a name of "A4nthropopithecus"(Verslag van het Mijn-
juvenile individual, the parietal and occipital regions wezen, Batavia, No. 4, 1891, pp. 13-15; see also Dubois,
were largely intact. The matrix consisted of sandy tuff 1894). Both specimens came from the left bank expo-
mixed with lapilli. The new fossil has become known sure opposite Trinil where the Solo River makes a sharp
as Pithecanthropus Skull III. bend (P1. I, Fig. 1).
During the following month, Professor F. Weiden- 1892-At the same place, but 15 m. distant from
reich, Director of the Cenozoic Research Laboratory at the skullcap locality, Dubois excavated a left femur of
Peking, China, while looking over some new Sangiran "Anthropopithccus" in August (Verslag van het Mijn-
material at Bandoeng, recognized in a collection of mam- wezen, Batavia, 1893, pp. 10-14). In October the sec-
malian fossils a fourth Pithecanthropus skull (Skull ond molar was found only 3 m. from the original Trinil
IV). The occipital and large parts of the parietal and site, and between it and the resting place of the femur
temporal regions are present in this specimen, which is (Dubois, 1896, p. 4).
thought to be an adult male. Finally during January 1896-Further excavations at Trinil under Dubois'
1939 a fragment of the right body of the lower jaw and direction yielded a third molar (a second left lower
an upper jaw, also belonging to Skull IV, came to light premolar). The locality is somewhat uncertain, but
(von Koenigswald and Weidenreich, 1939). The latter since it also came from the "sandstone formation" im-
was large enough to permit a reconstruction of the facial mediately adjoining the site, Dubois argued for the indi-
region to be made (Weidenreich, 1940). vidual identity of origin of all the Pithecanthropus re-
Hence within two and a half years proof of the mains (Dubois, 1899, p. 273).
Pithecanthropus population of Java had accumulated so 1898-1901-Seasonal excavations proved unsuccess-
fast that there was only one other site in Asia which ful with regard to additional Pithecanthropus material.
surpassed this area in the number of fossil human re- 1907-1908-Frau Lenore Selenka excavated at Trinil
mains: the Sinanthropus site-Choukoutien-near Pe- with the aid of German and Dutch scientists, among
king. The data obtained from such an extraordinary whom was W. F. F. Oppenoorth, who later discovered
contribution to our knowledge will be discussed pres- Solo Man. A human tooth, the "Sonde tooth," was
ently. Here mention should be made of the fact that it recovered from the surface about 4 miles upstream from
was mainly Dr. von Koenigswald's personal initiative Trinil, and in addition a large palaeontological collection
and his ability to organize native collecting which have was made (Selenka and Blankenhorn, 1911).
produced the results. 1924-Professor Ales Hrdlicka (1925, pp. 57-60)
visited Trinil and received the impression that the
II. SUMMARY OF THE DATA BEARING ON Pithecanthropus remains had been redeposited by stream
THE INVESTIGATIONS IN CENTRAL JAVA action.
1926-Professor W. D. Matthew examined the site
The history of the investigations in the Solo Valley, under the
guidance of Dr. L. J. C. van Es.
central Java, from 1890 to 1935 may be briefly sum- 1930-1931-New excavations were initiated by Op-
marized as follows:
penoorth, but no new material was found. Dr. van Es
1890-Eugene Dubois started collecting vertebrate conducted special geological studies in the area.
fossils in the Kendeng Hills and at Trinil. These 1933-1935-A new geological survey of central Java
localities had been known both to native bone-collectors was undertaken
by J. Duyfjes under the direction of
and to earlier Dutch naturalists, notably Junghuhn. On
J. Zwierzycki. In conjunction with this work Dr. von
November 24 Dubois found a fragment of a human
Koenigswald collected new and additional vertebrate
lower jaw exhibiting the alveoli of the canine and the fossils.
first and second premolars, at a place called Kedeng From this historical record it appears that any
Brubus, 25 miles E.S.E. of Trinil (Verslag van het charges of neglect, or of disinterestedness on the part
Mijniiwezen,Batavia, 1891, pp. 14-15; see also Dubois, of Dutch scientists at the important fossil human site
1892, p. 95). This was the first fragment of Pithe- of Trinil, are completely unfounded. On the contrary
canthropus which was found among the fossils occur- it would seem that there has been a great deal of ac-
ring in the "Kendeng Deposits" (the "Kaboeh Beds" of tivity both at Trinil and elsewhere in the Solo Valley,
the latest geological nomenclature), according to Du-
yet it remains to be determined whether or not the ap-
bois' statement. It was said to have come from the
proach was in all instances dictated by broad geological
sandstone-like andesite tuff characteristic of the fossil-
reasoning.
bearing horizon.
1891-At Trinil, Dubois collected in September the III. THE BEGINNING OF THE PLEISTOCENE
molar of a "chimpanzee," which he called Anthropopi-
IN JAVA
thecus (Verslag van het MijnweZen, Batavia, No. 3,
1891, pp. 13-14). This is nowadays regarded as an Geologically speaking the island of Java is of such
orang. One month later the famous skullcap was recent origin that one may regard it as one of the
440 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
youngest offshoots of the mainland of southern Asia. brate fossils. This formation was evidently laid down
Fortunately the period of its development coincides in in an estuary along the coast, from which streams
the main with the ascent of man, hence the geological swept the leaves of tropical plants and trees together
formations laid down since the emergence of the island with the carcasses of land mammals. In the coastal
date from a critical phase in human evolution. In Java waters these remains were embedded along with marine
several phases of man's fossil record have been pre- mollusks. Oostingh (1935) states that these marine
served in a most unique manner as a result of the inter- invertebrates had been transported into the lagoon by
action of volcanic, climatic and mountain-making forces. tidal or longshore currents, and that only the fresh-
Volcanic rocks cover some 28 percent of the island, water mollusks could be regarded as truly in situ. Of
while approximately 40 percent of it is composed of 26 species of fresh-water mollusks, eleven or 42 per cent
Upper Tertiary formations, mostly laid down by the belong to recent types. Ter Haar (1935. p. 36) con-
sea. The balance includes certain complexes of Old siders that the vertebrate remains are, comparatively
Tertiary and Cretaceous rocks of marine and igneous speaking, of greater significance for purposes of estal-
origin, but more important are the estuarine and fresh- lishing the geological age of the beds. Indeed the
water Pleistocene beds in which the land organisms fauna, which includes some 22 species described by
are preserved. Naturally an island as large as Java Stehlin (1925), van der Maarel (1932), and von
(48,504 sq. miles) was not formed during one single Koenigswald (1933; 1934), makes correlations with
phase of the Cenozoic Epoch. Rather the process is to other faunas of the Asiatic mainland possible. The
be pictured as a very gradual but by no means continu- most characteristic forms are represented by Archi-
ous one. At the close of the Miocene period there diskodon planifrons, a primitive elephant; Stegodon
probably appeared a group of islands in the region now trigonocephalls, an ancient though highly specialized
occupied by the mountainous and hilly upland of lime- proboscidean; Leptobos, a primitive kind of cattle:
stone rocks and schists at present called the Zuider Hexaprotodon simplex and H. antiqltus, two extinct
Mountains (Fig. 100) and the uplands of western hippopotami; also antelopes, deer and Colossochclis
Java. To the north lay a marine strait, beyond which atlas, the giant land turtle first found in the Siwalik
a few isolated islands, forerunners of the two anticlines Hills of India. This group, together with the other
now known as the "Kendeng Hills" and the North forms found in the Glagah Beds, presumably came from
Java ridge, may have risen. between India and South China; it is unquestionably
The name, Kendeng Hills, is given to a folded ridge related to the post-Pontian (Upper Pliocene or Lower
350-450 m. high which extends from south of Semarang Pleistocene) assemblage known as the Upper Siwalik
to the vicinity of Soerabaja. This ridge is drained by fauna (see Colbert, p. 424). A similar though per-
the Solo River, which descends from the Zuider Moun- haps older fauna comes from Tji Djolang, northeast of
tains and flows around the Lawoe Volcano before it Rantjah in western Java (Fig. 1), while the earliest
cuts through the ridge. Its lower course is bordered trace of land mammals on the island-a tooth of Acera-
by a second hilly range to the north. Van Es (1931, theriumi-has been found at Tji Sande, south of
p. 126) mentions that in the southern portion of the Cheribon (von Koenigswald, 1935; 1937). The fossils
Kendeng Hills a limestone conglomerate containing from Tji Djolang are in a very fragmentary condition.
pebbles of Miocene marl often forms the base of the and only a few of them can actually be determined.
overlying Middle Pliocene beds. The latter being of Among them is Merycopotamus, an anthracotheriid,
marine origin indicates that during Pliocene times the Stegodon, and a primitive hippopotamus. On the basis
movements in the Kendeng area were largely sub- of this scanty list it is hazardous to venture any precise
marine, although the sandy nature of the beds suggests correlation with the mammal horizons of India. But it
that the sea was shallow, with sediment being swept into can be stated that here also we are dealing with an es-
it from an adjoining upland. Doubtless these move- sentially Upper Siwalik type of fauna. In my opinion
ments of the sea-bottom became accentuated during the there is no valid reason to regard the oldest known
Middle Pliocene when the incipient relief was enlarged. fauna of land vertebrates in Java as older than the latest
It is difficult to reconstruct the ancient geography of Pliocene. As to northwestern India there are several
that particular period, but to judge from the presence of good reasons for assigning the Upper Siwalik fauna to
a mammalian fauna in the Upper Pliocene beds of cen- the Lower Pleistocene, contemporaneous with the First
tral Java, and particularly from the character of this Glaciation and First Interglacial in the northwestern
fauna, it would seem that Java was then linked to the Himalaya, as shown first by Matthew and Colbert and
Asiatic mainland. The geological evidence for this as- geologically substantiated by de Terra and Paterson
sumption is found at Kali Glagah in western Java, in (1939). Be that as it may, a land-bridge existed at
the vicinity of Boemiajoe, south of Tegal (Fig. 100), the close of the Pliocene between Java and southern
where Lower Pliocene marine beds are disconformably Asia on which herds of ruminants migrated to the
overlain by deposits, known as the "Glagah Series" newly-born land around the Sunda Sea (Fig. 99).
(ter Haar, 1935). These contain marine invertebrates, During this period Java appears to have been but a
as well as the remains of a terrestrial flora and verte- narrow peninsula, possibly less than 55 miles in width.
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 441

which is the narrowest stretch of land encountered on IV. THE LOWER PLEISTOCENE SITE OF
the island today. MODJOKERTO IN EASTERN JAVA
It is a curious fact that the formation of land was not
In 1936 two reports concerning the find of a fossil
accompanied by volcanic action, as were the later up- human infant calvarium appeared in an obscure peri-
lifts and foldings of ridges in this area. Even in the
odical of a Dutch engineering society in Bandoeng,
strongly volcanic region of eastern Java, there is no
evidence of volcanic action. Duyfjes (1936; 1938) Java (Duyfjes, 1936; von Koenigswald, 1936-a). One
of these dealt with the geology of the easternmost part
has shown that in Upper Pliocene times eastern Java
was still submerged beneath the sea, since Globigerina of the Kendeng Hills in the vicinity of Modjokerto, a
marls and diatomaceous shales were being deposited to town some 25 miles west of Soerabaja (Fig. 100).
This region had attained economic importance on ac-
a thickness of 450 m. Evidently at that time large por-
count of petroleum explorations, which warranted a de-
tions of Java continued to form part of the geosyncline,
which extended south and southwest of the Sunda tailed study of eastern Java. The Dutch Geological
Shelf and from which the impressive chain of the East Survey sent Mr. Duyfjes to Soerabaja for a closer in-
Indian islands was to emerge during the course of the vestigation of the geological structure, and during the
Pleistocene period. course of his field work the human fossil was found.
But the narrow peninsula existing during Upper The great age ascribed to it, its affinity to the Pithlecan-
Pliocene times soon began to rise more vigorously out thropuls fossils, and the fact that it had been the first
of the ocean waters. All along the southern slope of human fossil collected in Old Pleistocene deposits since
the Kendeng Hills the formations of Tertiary age are Dubois' discovery of Pithecanthropuls, all made the
overlain by tuffaceous sandstones and shales of fluvial Modjokerto find one of the most important contribu-
tions to the field of human palaeontology in Asia.
origin containing locally marine fossil-bearing beds.
By the beginning of the Pleistocene period, Java had Duyfjes (1936, p. 138) reported about the discovery
entered its first major phase of eruptions, and volcanoes as follows (translated):
rose in a long chain of smoking mountains extending Before the completionof Sheet 110 (Modjokerto) of the
from southwest of Soerabaja in the east to the region of GeologicalMlapof Java, the JavanesecollectorAndojo was
Bandoeng in the west. From then on the geological sent out to search for vertebrate remains in the Kaboeh
history of Java was determined by the interplay of vol- Beds (Middle Pleistocene), and to secure the accurateposi-
canic action, stream erosion, and fluctuating sea levels. tion of each find. He was able to visit once more the re-
At first the ocean occupied most of what is now low- gion of the Poetjaung Beds (Lower Pleistocene), so that he
might collect new material of the Djetis fauna. In doing
land both in the interior and along the coastal plat- so he discoveredat a depth of one meter in an excavation
form. One of these marine straits extended from the pit, dug into the Poetjang Beds, a fossil skull which von
region of Modjokerto westward into central Java as Koenigswald determined as belonging to an immature
far as Soerakarta; another must have existed in the hominid . . . The exact location which the author visited
on May 31, 1936, lies 3 km. north of Perning, and about
area southwest of Bandoeng. To the north of these 300 m. east of the road to Soembertengahwithin the con-
straits lay the ancestral Kendeng Hills, large portions cession of the Bataafsche PetroleumMaatschappij.
of which were in process of rising gradually from the
ocean bottom. This gradual emergence is indicated by According to Dr. von Koenigswald, the discovery was
the conformable contact between Pliocene and Pleisto- made in February. Hence the site was visited three
cene beds, which is commonly met with. But discon- months after the find was made. Considering that the
formities have also been observed causing breaks in the skull was found at a depth of only one meter, i.e. well
sedimentary sequence, such as at Trinil along the Solo within range of normal soil-making processes, one
River. As a result of the initial folding, which divided might suspect that the specimen had been redeposited,
Java into a number of west-east striking synclines and and perhaps in this way mixed with fossils of a differ-
anticlines, estuarine conditions prevailed throughout ent age from that of the infant skull. It was this aspect
the Lower Pleistocene in the depressed region north of which made a personal inspection of the site desirable.
the Zuider Mountains. Here fresh- and brackish-water
faunas thrived and intermingled with marine organisms. 1. THE GEOLOGYOF THE MODJOKERTO
SITE
Thus the Lower Pleistocene of Java is characterized by
a rather quick change in facies, and in the southern Near Perning, a few miles northeast of Djetis, the
basin volcanic and marine sediments are intercalated. Kendeng Hills are much lower than farther west. The
Since man first appeared at this time, this period is of land is flat with well dissected, undulating topography,
special interest. The infant skull found near the town especially at the crest of the shallow anticline found
of Modjokerto, west of Soerabaja, was discovered in about one mile north of Perning, on the road to
Lower Pleistocene beds. On the basis of our present Soembertengah (Fig. 100). The site is about 300 m.
knowledge, this is the oldest datable human fossil in east of the road and about one kilometer south of the
Asia, and for this reason it occupies a special place in village, and it is marked on the geological map (Sheet
our discussion. 116, Sidoradjo, 1938) (PI. XXXV, Fig. 4).
442 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
As indicated on the section (Fig. 101), the skull can be seen to good advantage about one mile north of
was discovered in the upper portion of the so-called Perning village. The road cuts into the two mollusk-
Poetjang Beds, or more accurately in the volcanic bearing beds which dip 35? south at the point where I
facies which overlies the marine clays. This locality saw them in 1938. Once the top of the ridge is reached,
lies in the eastern part of the ancient marine strait there appears a thick series of stratified tuffaceous sand-
which at that time occupied the lower Solo Plain and stone and ash-layers. The excavation pit in which the
parts of central Java. This strait was gradually filled skull had been found looked as fresh as though it had
up with fine volcanic material by eolian as well as by been dug only a few months prior to our visit. In
fluviatile agencies. Duyfjes (1938) has demonstrated reality it was three years old and had during that time
how the volcanic facies and river-laid conglomerates been exposed to tropical rains and strong surface wash.
gradually extended farther and farther seaward until Under the circumstances, it seemed hardly credible that
the basin was completely filled with sediments. Dur- the pit had survived such weathering, and yet the ques-
ing this process, the sea-level and the supply of sedi- tion arose whether there had been much weathering at
ment underwent fluctuations which caused a shifting all. This region is one of the driest in Java. The pit
of the shore-line. Two mollusk horizons indicate two itself lies in a protected gully on a gentle slope where

Soembertengah
Klagenblandong H. modjokkertensis
_I I
'
I I I
I _ I
I N
,. / f Tn +
. ~tirl Tii In t
_ .N
_.1

^ ''i'*'- . - -
,.kMi.e/ ^*' ' .-^
-'.
."&. I

FIG. 101. Geological Section North of Perning, Near Modjokerto (after Duyfjes).
1. Lower Pleistocene Poetjang Beds Containing the Djetis Fauna.
la. Marine Deposits.
lb. Fresh-water Deposits; m-marine layers.
2. Middle Pleistocene Kaboeh Beds Containing the Trinil Fauna.

marine transgressions which are 10 m. and 20 m. thick not much rain-wash was in evidence. The northern
respectively, and which appear as distinct layers in the slope exposes two layers of tuffaceous sandstone di-
290 m. sequence of tuffaceous conglomeratic sandstones vided by ? 10 feet of fine tuff. Vertebrate fossils oc-
exposed in the section. Remains of land vertebrates cur in two specific horizons of which the upper is ex-
occur chiefly in the coarser layers. These were formed posed in the excavation pit. From the gravel Dr.
during periods of intense volcanic activity, and they Movius and I extracted a tooth of Cervus zwaani, the
were laid down by streams along the coast (Duyfjes, guide fossil of the Djetis fauna. Other bones and
1938, p. 34). No doubt the animals lived in the for- tooth fragments could be picked up in the adjoining
ested foothills of the volcanoes; in fact they may well fields, indicating that we were standing on the very bone
have been killed by eruptions-their carcasses after bed from which the Djetis fossils had been previously
having been carried downstream eventually were em- collected. The dip of the strata is northward, and the
bedded in the sand. Some of them actually were car- axis of the anticline is half a mile to the south. The
ried out to sea and were preserved together with sea geological structure is perfectly normal and apparently
clams and other marine forms (compare Cosijn, 1931). without any break; at least the exposures along the road
The human skull lay in conglomeratic sandstone in the show an unbroken sequence, and the upper bone bed
northern portion of an anticline. The sandstone itself can be accurately determined in relation to the entire
was laid down in shallow water near the then-existing structure. Hence there is no doubt as to the correct
shore as indicated by the mollusk-bearing horizons assignment of the fossil layer within the sequence.
shown in the section (Fig. 101). The skull was There still remains the possibility that the human
therefore presumably swept into the coastal region by a fossil was derived from an eluvial or colluvial soil. At
stream descending from the adjoining volcanic region. first glance one does not notice much of any soil in this
Duyfjes (1938, p. 34) calculated that the shoreline at neighborhood. There is no laterite as there is in other
that time lay some 10 km. west of this region. parts of Java, and there are no younger volcanic ashes.
The entire geological sequence is well exposed along The ground looked surprisingly barren, yet consider-
the road to Soembertengah. The anticlinal structure able crops are cultivated in the neighborhood, including
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 443

especially sugar cane, which demands a good loamy or Epimacheiroduis zwierzyckii v. K.


silty soil. Patches of silt were seen, but no real soil Hyaena siznensis Ow. (-H. brachygnathus Dub.)
Paradoxirus sp.
profiles. On the slope near the pit there is some an- Canis sp.
gular talus; however, there was no trace either of Lutra robusta
landslides or of any other formations which might con- Ursus cf. malayanus Raffl.
ceivably be responsible for the redeposition of the hu- Ursits cf. kokeni M. et G.
man fossil. I admit that my observations were very Manis palaeojavanica Dub.
limited, and that it is remotely possible that the cover- Dr. von Koenigswald (1939, p. 35) in commenting
ing sediments have been stripped from this surface. on the above assemblage says: "Very characteristic for
Yet if these ever had been present and if they had been this fauna is the presence of Leptobos, Nestoritherium,
of a later age, they surely would have left traces of Epimacheirodus and Hyaena, which are missing in the
younger fossils in the valleys. Now the investigators younger beds. None of the genera present is re-
claim that no younger mammalian fauna was found by stricted to the Tertiary." The two antelopes and deer
them in this area, and hence an overlap of younger are considered good guide fossils. As regards Hip-
fossiliferous beds is unlikely. popotamus antiqtus, von Koenigswald states that of its
What is of decisive importance for establishing the six lower incisors the I2 have already been pushed up-
age of the Modjokerto fossil is the fact that the matrix ward without having suffered reduction in size. This
adhering to the original skull can be matched with that means that the species occupies an intermediary evolu-
from the layer exposed in the pit. This at least was tionary rank between the primitive H. simplex of the
the result of my experiment. I had seen the infant older Glagah fauna and the more specialized H. nama-
skull in Bandoeng, and I studied the matrix, taking de- dicus of the Trinil fauna.
tailed notes of the mineralogical composition. Later Of great interest is the association of Early Man with
in the laboratory at Bandoeng I examined the material anthropoid apes in this Lower Pleistocene horizon, be-
from the pit and found it to be identical with the matrix. cause it indicates that man had already differentiated
This convinced me of the Early Pleistocene age of the from them at this period. Orang and gibbon still live
fossil, or at least of its contemporaneity with the Djetis in Borneo and Sumatra, but they are now extinct in
fauna in Java. Java. The human skull is that of an infant about three
This Djetis fauna was named after a near-by village to four years old (von Koenigswald, 1936-b). The
where the tuffaceous sandstone has yielded a consider- bones are thin and the fontanelles closed. But since
able number of vertebrates. Dr. von Koenigswald the facial part and the base of the skull are missing, its
(1935) introduced the term for the Lower Pleistocene true phylogenetic rank is unknown. Dr. Weidenreich
fauna of Java, which is earlier than the Pithecanthropus (with von Koenigswald, 1939) has drawn attention to
or Trinil fauna. Here again the majority of the fossils the similarity between the skulls of Pithecanthropus and
occurs in the upper half of the tuffaceous facies of the this infant fossil. This is especially marked with re-
Poetjang Beds, usually in the gravelly parts. The list gard to the postorbital narrowing and sharp bending
given by von Koenigswald (1939, p. 35) is as follows: of the occipital, from which Weidenreich has concluded
Homo modjokertensis v. K. (= Pithecantthropus sp. ?) (1940-a, p. 376) that the fossil is that of an infant of
Simia satyrus L. Java Man. From a paleontological point of view, few
Hylobates sp. will object to the long life-span for Pithecanthropus
Symlphalangus syndactylus Desm. implied by Dr. Weidenreich's analysis, since it is pos-
Macaca sp. sible that here we deal with another species of Pithe-
Presbytis sp.
Leptobos cosijni v. K. canthropus.
Bos (Bubalus) sp.
Cervus (Rusa) zwaalli v. K. V. THE MIDDLE PLEISTOCENE SITES OF
Ccrvus (Rusa) problematicus v. K.
MIuntiacus sp. PITHECANTHROPUS IN CENTRAL JAVA
Traguluscf. kanchil Raffl.
Antilope modjokertensis v. K. 1. THE SOLO PLAIN IN THE VICINITY OF SANGIRAN
Antilope saatensis v. K.
Antilope sp. From the folded hills of Modjokerto westward to
Sus brachygnathus Dub. central Java, the anticlinorium of the Kendeng Hills
Siis coerti v. K. exposes an almost unbroken geological record of the
Sits sp. Pleistocene. Thus far only the southern slopes have
Hippopotamusantiquuisv. K. been surveyed, but this has been done in such detail
Rhinoceroscf. sondaicusDesm.
Tapiruscf. indicus Desm. (- T. pandanlicusDub.) that every phase of the Pleistocene may be interpreted.
Nestoritherium sivalense Falc. The Lower Pleistocene Poetjagly Beds gradually lose
Stegodon trigonoceplialus praccuirsor v. K. their marine character, which is replaced by a volcanic
Elephas sp. facies and fresh-water deposits. As one approaches the
Felis cf. pardus v. K.
Fclis cf. tigris L. Solo Plain, near Soerakarta, black clays containing
444 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
CHART 1
COMPARATIVE STRATIGRAPHY ALONG THE SOUTHERN SLOPE OF THE KENDENG HILLS IN CENTRAL AND EASTERN JAVA

EASTERN JAVA GEOLOGICAL CENTRAL JAVA


PERIOD
Vicinity of Modjokerto AND BEDS Sangiran Anticline
(Kendeng Hills) (Solo Plain)

River gravel and sand, 14 m.


Sands and clays with black clay layers intercalated UPPER NGANDONG FAUNA-H. neanderthalensis soloensis
Tuffaceous sands PLEISTOCENE Volcanic breccia (lahar formation), 3 m.
Notopoero Beds River sand-tuffaceous, 4 m.

-ce- - - -- - - ---Disconformit X
Volcanic breccia (lahar formation) Agglomerate and river sand, 9 m.
River sands and clays, 4-240 m. MIDDLE River sand (cross-bedded)-tuffaceous, with clay bed at
Tuffaceous sandstone and clay, 60 m. PLEISTOCENE base, 27 m.
Agglomerate Kaboeh Beds TRINIL FAUNA-Pithecanthropus erectus
Cemented agglomerate, 4 m.

Volcanic facies, 4-260 m.: tuffaceous littoral sands Black lake clay, 8 m., with diatomaceous and tuffaceous
and gravels with two mollusk-bearing horizons LOWER layers and one marine clay bed (50 cm.) intercalated
DJETIS FAUNA-Homo modjokertensis PLEISTOCENE DJETIS FAUNA
Marine facies, 4-450 m.: argillaceous beds with foram- Poetjang Beds Volcanic breccia (lahar agglomerate), 30 m.
inifera clays Corbicula bed (lacustrine)-limnic and tuffaceous, 3 m.
Unconformity (?)

Upper Kalibeng Beds: Globigerina marl and sandstone UPPER Balanus limestone, 0.50-1.50 m.
Diatomaceous marls PLIOCENE Turritella bed-argillaceous sand, 30 m.
Globigerina marl and calcareous sandstone 4450 m. Kalibeng Beds Marine deposits-bluish-gray clay

layers with fresh-water mollusks become more frequent. ciated with Early Man and, more important still, lahar
In the domelike hills of Sangiran (Fig. 102), fresh- or deposits were instrumental in building up layer after
brackish-water beds dominate the Poetjang Beds. Thus layer of water-laid sediment of volcanic derivation. For
it is in the area west of the Solo River where we reach this reason it is essential to review briefly what is known
what was the coastal region of Lower Pleistocene times. of their origin. A very good concise description is
Here the great plain of Solo extends from the faulted given by van Es (1931, pp. 62-63), from which we
rim of the Zuider Mountains on the south to the slope quote the following passage:
of the Kendeng Hills on the north. This plain was filled
After an eruption, volcanic ashes, heaped around the cone
by a fresh-water lake in which black laminated clays and of the volcano, are swept by a big rain storm into the steeply
diatomaceous beds containing fresh-water mollusks were
descending valleys of the rivers. The mud-laden water,
deposited. Chart 1 gives the sequence of strata at having high specific gravity and great velocity, picks up
Sangiran, which is compared with that at Modjokerto. boulders from the river bed and undermines the steep slopes
There is a slight unconformity between the Corbicula of the valleys, thereby adding new material to the already
swollen rivers. The swiftly descending "lahar"-in a way
bed and the underlying marine Balanus limestone, and a
comparable to the "muhrgang" in Switzerland-with
in the southern part of the Sangiran dome the latter is height often attaining several dozens of meters, arrives in
missing altogether, since the fresh-water bed lies on an the lower plains, where it floods the banks of the rivers
older (Upper Pliocene) marine layer. Volcanic activity and spreads out, destroying everything it meets, whilst
was great at that time as indicated by a boulder breccia often big tracts of land are covered by it. The "lahar"
deposits are easily recognizable by being unstratified and
or "lahar" formation (volcanic mudflow agglomerate) unassorted. The material is mostly of volcanic origin and
which covers the lake beds up to a depth of 30 m. This consists of tuff in large quantities mixed with sand and
mudflow must have completely filled the lake, and when gravel and boulders of andesite of sometimes amazing di-
it was laid down it may have scooped up parts of the mensions. Where the lahar passes over sediments of dif-
ferent origin, part of these may be easily picked up, espe-
bottom of the basin because lumps of marine fossil-
cially as the eroding force is exceedingly great.
iferous clays are enclosed in beds containing clams.
But as van Es (1931, p. 62) states, "notwithstanding the It is this dual role of mass-movement and scooping
occurrence of marine fossils, the boulder breccia is not or erosive power which makes a lahar deposit a very
a marine deposit." These lahar deposits have played a puzzling geological feature. For one thing the capacity
significant geological role in the earliest periods of of scooping and picking up older sediments explains
human habitation on this island. In them we find why at Sangiran patches and lumps of older marine
buried remains of the fauna and flora which was asso- clays were found in the volcanic boulder breccia. Also
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 445
it would seem that such a formation might pick up and Paleolithic artifacts. The site can be reached on
bones or plant remains and transport them a consider- foot from a fairly good road which leads to Sangiran,
able distance, necessitating special care in the interpreta- a village located in the center of the domelike structure,
tion of geological sections. which rises 170 m. above sea-level and some 140 m.
At Sangiran the lahar formation was flooded once above the plain of Soerakarta, or Solo, north of this
more by the fresh-water lake in which black clay, diatom ancient native town (Fig. 100). The Kali Tjemoro, a
beds and tuff layers alternate. Van Es (1931, p. 64) small tributary of the Solo River which flows eastward
suggested, with some resrvation, an Upper Pliocene age from the foot of the great Merapi Volcano, has super-
for these fresh-water beds, chiefly because they contain a imposed its meandering course on the anticline, and in
greater percentage of extinct species of fresh-water mol- doing so has exposed to view all the formations from the
lusks than those found in the Trinil Beds which he con- marine Lower Pliocene through the Uppermost Pleis-
sidered Lower Pleistocene. However, more recent col- tocene. Before reaching Sangiran, one may ascend the
lecting by von Koenigswald has proved the presence of prominent rim of the slightly elliptical dome. Here one
the Djetis fauna in this stage, hence their Lower Pleis- notices how the volcanic boulder deposits form a dis-
tocene age has been established. Of special interest is tinct escarpment enclosing a maze of dissected hills
the find of a new large anthropoid ape in these beds.2
The Sangiran lake suffered a short invasion by the composed of less resistant members of the Late Ceno-
zoic marine-lacustrine series. The more important ex-
sea which left a thin layer of yellow marine clay, only
50 cm. thick, containing a few marine mollusks. This posures lie a few miles downstream, near the village
of Bapang, where the stream has cut down to the base
incident tends to show that the region was almost at
of the Lower Pleistocene. Fig. 102 shows the succes-
sea-level during the Lower Pleistocene, and that the
sion of strata at the Pithecanthropulssite of Bapang, near
ocean had easy access to the lake basin. But this defi-
nitely was the last submergence of central Java, because Sangiran, and P1. XXXIV, Fig. 3, illustrates the land-
the Middle and Upper Pleistocene formations are devoid scape in which this section is located.
of any marine beds. Layer 1 is a thin agglomerate, 4 m. thick, comiposed
of a cemented deposit of gravel with fresh-water gas-
OF THE SANGIRANSITE
2. THE GEOLOGY tropods, including Mclauiia. This marks a distinct hori-
zon that can be followed a great distance downstream.
Along the Tjemoro River, near the village of Bapang, Beneath lies a black, lacustrine clay, which marks the
there is an uninterrupted sequence of Pleistocene beds beginning of the Lower Pleistocene Poetjang Beds.
which furnishes a very complete record. It was here This-alternates with tuffaceous sand and contains a rich
that Pithecanthropuisremains were found in 1937, and vertebrate fuana of Djetis type. According to van Es
since then this region has become the most prolific in (1931, p. 63), the black clay contains fresh-water mol-
Java. It has yielded no less than three calvaria and two lusks. It originated from a fine mud deposited in the
mandibles, in addition to the countless vertebrate fossils center of a fresh-water lake.
2
Unpublished-information communicated through Professor Layer 2 is a gray to greenish-colored sand with fine
F. Weidenreich. pebble layers and cross-bedding. The constituents are
w S E
@0 00. a

NofopoeroBeds
.. .- .
4
*
aa 0flaa
" -':: . .
? .n.....-.'-
. .': ' - ... :.'. . ...
0
NGANDONG
FAUNA
(artifacts)
Kaboeh Beds
TRINILFAUNA
(36 m.)

Poetjang Beds
DJETIS FAUNA

FIG. 102. Geological Section on the Tjemoro River at Bapang, Near Sangiran. (Partly after Van Bemmelen.)
1. Cemented Basal Agglomerate, 4 m.
L. Landslide Material (?).
2. Cross-bedded Sands and Fine Gravel with Basal Clay Bed, 27 m.
3. Sand and Clayey Silt Containing Plant Remains, 9 m.
4. Volcanic Breccia (Lahar Formation) and River Sand, 7 m.
5. River Gravel and Sand, 14 m.
446 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
well waterworn and consist of volcanic rocks typical of accumulation of cross-bedded coarse debris. These
the Merapi region from which the river descends. This mark the uppermost portion of our Pleistocene se-
horizon is some 27 mn.thick including the basal clay bed, quence near Sangiran. Since this alluvium was also
and it represents a fluvial deposit laid down in a shifting spread over the plain, it is obvious that the folding of
stream channel. The river carried a great deal of vol- the Sangiran anticline is of later date.
canic material, ashes as well as lapilli; in addition it must The last coarse layer-Layer 5-of the sequence
have picked up a great many bones of land fauna which heralds an abrupt change of relief somewhere in the
populated the foothills of the great volcanic range in the upper reaches of the stream (P1. XXXV, Fig. 3). It
south. This would explain the wealth of such fossils, lies disconformably on the cross-bedded sands and
representing the Trinil fauna, in the tuffaceous sands gravels which make a compact and more resistant cap
at Bapang, and particularly the occurrence of human above the soft underlying beds. There is a notable
remains. It was in this layer that the second Pithe- wealth of various rock fragments in this upper gravel
canthropus skull as well as the lower jaw were found unknown in any other Pleistocene deposits in this re-
(von Koenigswald, 1937-a; 1938; 1938-a). Confirma- gion. In addition to the common volcanic rocks one
tion of their original stratigraphic position was received finds Tertiary cherty limestones, black chert and quartz,
when we visited the site in April 1938. At that time a clearly indicating that the stream descended from a
rockfall lay some 20 feet above the stream, blocking a western upland where Tertiary formations were already
small path which leads along the left bank of the river being denuded. In view of the fact that the drainage
(P1. XXXIV, Fig. 4). Ten feet above there was a underwent profound changes during this stage, it would
large scar in the cliff from which the rockfall had come. seem that the Upper Pleistocene ended rather abruptly
According to Dr. von Koenigswald, his native collector with the uplifting and erosion of the plain, which for so
had found the skull in perfect condition in a block of long had been the scene of heavy alluviation. This
sand amidst the rockfall. Now the block could still analysis is in complete agreement with the results ob-
have been reinserted in the scar, hence there is little tained from tectonic studies in central Java. These
doubt that the fossil came from layer 2. This means have shown that mountain-making was most active at
that originally the skull lay 35 to 40 feet above the for- the very close of the Pleistocene (van Bemmelen, 1937).
mational boundary between these and the Lower Pleis- This picture of a basin being filled by stream-action
tocene (Djetis) beds. is typical of the entire Late Pleistocene history of the
The geological age of the Sangiran skull becomes clear island. While volcanic activity developed on an ever-
from further analysis of the upper sequence. Layer 3 increasing scale, and large masses of eruptive rock
(9 m. thick) is a coarse, cross-bedded sand, which may debris came to be deposited, the streams carried masses
well represent a flood deposit, with clayey silt containing of loose volcanic material into the plains. Occasionally
plant remains above. The latter have thus far yielded a lahar deposit would advance rapidly into the lowland,
no recognizable forms, but I am confident that fossil burying channels and blocking rivers, thus causing a
plants could be collected in this horizon, which is appar- temporary ponding of water and the deposition of lami-
ently the same as the one at Trinil (Layer 5, see p. 447). nated clays. In these fossil leaves have been preserved,
Layer 4 is some 7 m. thick, and it makes a prominent revealing some of the ancient flora. But the eruptions
cliff. There is a slight dip to the north in these beds. were by no means local, since the Notopoero Beds con-
At the base lies a bouldery agglomerate, and in it occur tain the same type of lahar formations and boulder
blocks that measure up to 10 feet in diameter. The breccias between Sangiran and the vicinity of Mt. Pan-
matrix is silty to sandy; it contains small subangular dan, south of the Kendeng Hills, as well as in eastern
pieces of andesite and lateritized scoriaceous lava. The Java at Djombang (van Es, 1931, p. 122). No doubt
peculiar structure of this sediment, with its angular this increase in volcanic activity was connected with a
blocks lying in a fine-grained medium, cannot be ex- period of uplift in central Java, as Dr. van Bemmelen
plained otherwise than by catastrophic deposition, such (1934) has demonstrated in the vicinity of Bandoeng.
as results from volcanic mudstreams. Hence this layer Erosion was intensified, and while the axis of the
is a lahar formation similar to the one at Trinil (see Kendeng Hills was rising, the adjoining basins subsided
p. 447). It marks the beginning of a new kind of sedi- further and received masses of detritus from the rising
mentation brought about by a new eruptive phase of the land. Rivers now began to incise their course into the
neighboring Merapi Volcano. In the opinion of the slopes of the rising folds. In this manner the first
Dutch geologists at Bandoeng, especially Dr. von Bem- stream terraces came into existence, and it is with these
melen, this bouldery agglomerate marks the base of the that the later chapters of human evolution in Java are
Notopoero Beds of Upper Pleistocene age. It is suc- connected. A typical record of these events is found
ceeded by gravelly sands and silt beds, undoubtedly along the Solo River, in the Residency of Madioen, be-
representing a quiet-water deposition. At a later stage, tween the famous Pithecanthropus site at Trinil and the
however, torrential stream deposition set in, causing the village of Ngandong (see p. 454).
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 447
3. THE MADIOEN PLAIN IN THE VICINITY OF THE Pliocene marine regression. This uplift may not have
TRINIL SITE been sufficiently violent to have caused an unconformity,
but there is, according to all available information, a
The plain of Solo extends eastward into the plain of clear break in the sequence from marine to fluvio-
Madioen, which surrounds the northern rim of the great lacustrine sedimentation. Duyfjes (1936, p. 147)
Lawoe Volcano. The Solo River, which has incised places the lahar breccia at the very end of the Lower
a channel some 35 feet deep in the Pleistocene for- Pleistocene and considers that the Trinil bone-bed rep-
mations, meanders across it, its course being forced resents the lower portion of the Kaboeh Beds (Middle
against the Kendeng anticlinorium by the resistant rocks Pleistocene).
of many ancient lava flows which the Lawoe and Willis This period was one of great volcanic activity during
Volcanoes spread in this direction. As a result of its which streams were repeatedly ponded as a result of
position the river is passing repeatedly from one Ceno- lava and mudflows which advanced onto the Madioen
zoic formation into another, thereby exposing many Plain. Doubtless the river built its flood plain higher
good sections (compare van Es, 1929). Geologically and higher, and its southern tributaries swept down
speaking the Madioen Plain is part of the same ancient from the ever-rising slopes of the Lawoe Volcano,
marine strait which in Upper Pliocene times divided the bringing with them the remains of those animals and
Zuider Mountains in the south from the ancestral Ken- plants which thrived in the adjoining hills. It is this
deng Hills in the north. For this reason we find here combination of erosion on the volcanic upland and rapid
the same type of stratigraphy as at Sangiran-i.e. the accumulation in the river flat, together with catastrophic
Upper Pliocene or Kalibeng Beds of Duyfjes (1936) mudflows and ash falls, which accounts for the abun-
are conformably overlain by the tuffaceous sandstones dance of fossil remains in this region. But since the
and agglomerates of the Poetjang Beds which in turn depositional agencies lacked regularity and gentleness
are followed by the Middle Pleistocene Kaboeh and Up- of action, most of the fossils were embedded in frag-
per Pleistocene Notopoero Beds. As at Sangiran, the mentary condition.
marine strait was gradually filled up with alluvial de-
posits containing much volcanic detritus as the ancestral 4. THE GEOLOGYOF THE TRINIL SITE
Solo River heightened its floor. Carthaus (1911) and
van Es (1931) claim that the contact between the Trinil lies on the right bank of the Solo River in the
Pliocene, represented by coral limestone and Globigerina Residency of Madioen in central Java (Fig. 100 and
marl, and Pleistocene is uncomformiable,but Duyfjes P1. XXXIV, Fig. 1). The country in this vicinity is
(1936) has not upheld this view, and from his investi- flat except to the north, where there rises the low
gations it becomes increasingly clear that facies changes wooded range of the Kendeng Hills. Below Ngawi
are coimmnon in )bothformations, causing minor overlaps. the Solo River leaves the Madioen Plain and turns
Be that as it may, the Lower Pleistocene is here de- abruptly northeastward, flowing across the Kendeng
cidedly more influenced by volcanicity than at either Hills through a steep valley, presumably of antecedent
Modjokerto or Sangiran. It consists of some 100 m. of origin. On this stretch the stream exposes an anticlinal
tuffaceous sandstone and agglomerate, the latter pre- thrust-fault structure, in which marine Miocene and
sumably representing a lahar formation from the Lawoe Pliocene formations, as well as Pleistocene beds, are
Volcano, as Carthaus (1911) has suggested. His involved.
cross-section (1911, P1. VI) shows this layer between The most important cross-section through the Pleis-
two bands of bluish-black clay underlain by sandy clay tocene sequence at Trinil is the one already referred to,
which in turn overlies a thin conglomerate. The latter which is presented by Carthaus in the Selenka Expedi-
is presumably the same horizon as that at Bapang (Fig. tion Report (Selenka and Blankenhorn, 1911, P1. VI).
102), near Sangiran, hence we must assume a brief A modified version of this is illustrated in Fig. 103. As

FIG. 103. Geological Section at Trinil. (Partly after Carthaus.)


1. Black Clay with Globigerina Marl.
2. Sandy Clay.
3. Lahar Conglomerate (Lowermost Pleistocene).
4. Bone-Bed.
5. Gray Tuff-Sandstone with Plant-Beds.
6. Gray Sandstone and Silt with Bones.
7. Gray Sandy Tuff.
8. Terrace Sand and Loam.
448 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
a general characteristic of this sequence of strata, we rix of whitish-gray sandstone of which Layer 5 is
mention only three outstanding features: (1) the vol- mainly composed. From the literature (especially
canic origin of the sedimentary components, (2) the Carthaus, 1911), it appears that this is a volcanic tuff,
stratified, often cross-bedded structure, and (3) the but my observations at Trinil suggest that it is a river-
presence of fresh-water mollusks in most of the layers. laid sediment with volcanic components. These are
Layer 3,3 the basal deposit, can be studied only at low derived from an andesite rock, most commonly found in
water, but it was carefully analyzed by Dr. Carthaus of the foot-hills of the neighboring volcanoes. Duyfjes
the Selenka Expedition (1911). He interpreted it as (1936) mentions that Layer 5 contains vertebrate re-
a volcanic agglomerate or lahar conglomerate, formed mains of Trinil age and so does von Koenigswald (1934,
by a mudflow from the Lawoe Volcano, as mentioned p. 188). Fresh-water mollusks are quite common, in-
above. According to Duyfjes' (1936) recent regional cluding Unio and Melania; locally, as Duyfjes points
studies, this agglomerate is part of a formation of uni- out, this layer contains Globigerina, derived from the
form consistency in the Kendeng Hills, containing the underlying marine Pliocene marls.
remains of the Lower Pleistocene Djetis fauna. These Layer 6 is also a fine volcanic tuff redeposited in slug-
Poetjang Beds are characterized by a volcanic facies, gish water, as demonstrated by the frequency of clay
underlain by a thin marine layer, and by limnic, mollusk- layers.
bearing sandstones and tuffs. Hence it is evident, Layer 7 is a conglomeratic, partly cross-bedded, tuf-
Duyfjes states, that the Poetjang Beds represent re- faceous sandstone from 3 to 6 m. thick. This is clearly
washed volcanic material. Near Trinil they are about exposed on the right bank near Trinil. Bone fragments
100 m. thick and overlie conformably the Late Pliocene and entire skulls of Stegodon and crocodile occur in this
Upper Kalibeng Beds of marine origin (Layer 2). horizon, together with fossils of Trinil age. I investi-
For the history of Pithecanthropus it is rather signifi- gated a layer of lime pellets in the upper 30 cm. of
cant to note that at Trinil itself there are only 16 feet this stratum. These pellets are apparently derived from
of these lahar deposits between the Trinil Zone and the a fossil soil and their presence unquestionably marks a
coral formation of Pliocene age. time break in the sequence. This observation is sig-
Layer 4 in the section is the famous Trinil bone-bed. nificant in view of Duyfjes' (1936, p. 147) statement
It consists of tuffs, pumice and small lapilli, with occa- that the Middle Pleistocene Kaboeh Beds are discon-
sional, large rounded blocks of andesite and andesite formably overlain by terrace deposits.
lava, intercalated with few clay lenses. According to Layer 8 is a fine, fluvial tuffaceous sand, covered by
Carthaus (1911), the bone-bed is 40 cm. to 1 m. thick. dark clay soil (Duyfjes' Notopoero Beds). This layer
Since some of the blocks occurring in it weigh up to 100 lies on top of the fossil soil horizon, and apparently be-
pounds, it is apparent that they could not have been longs to a stage of terrace-making, traces of which are
transported by streams. Furthermore the bones show visible on the left bank opposite Trinil. In the dark
no signs of water transport. Still more convincing is soil-cap I also found limy pellets, but they are fewer
the absence of entire skeletons, which shows that the in number and smaller in size as compared with those
carcasses of mammals were broken up and their remains occurring in the fossil soil layer. This is the first hori-
widely scattered prior to their being embedded in the zon in the entire sequence which has not been affected
Trinil layer. Obviously such phenomena corroborate by folding, although there are indications that the ter-
Carthaus' statement concerning the catastrophic origin races at Ngandong and elsewhere have been tilted, as
of the bone-bed, and this point will be further discussed will be discussed below.
presently. The clay layers, on the other hand, suggest The Lower Pleistocene Djetis fauna (see p. 443) is
that the volcanic mudstream was reworked by river ac- well represented at the near-by locality of Guning
tion; this may have led to temporary ponding and depo- Boetak (Fig. 100), some twenty-five miles east of Trinil
sition of clays and silts. In the light of this interpreta- (von Koenigswald, 1934, p. 188). This horizon lies
tion, the fauna of Layer 4 must be considered as con- between two volcanic agglomerates, of which the higher
temporary with the deposition. is also represented in the Trinil section; it is the lahar
Layer 5 is partially represented by plant beds (60 to conglomerate (Layer 3 in the section). At the Trinil
80 cm. thick) composed of strata of thin and very bi- site the chief fossil-bearing localities are the bone-bed
tuminous clay, containing well-preserved leaves, roots, directly overlying the lahar formation and the plant bed
flower fragments and fruits. At certain places the above it (Layers 4 and 5). The fauna as listed by
leaves were seen to form a thick band of lacerated vege- von Koenigswald (1939, p. 38) is as follows:
table matter, such as results from flood deposition. The
Pithecainthropus erectus Dub.
palaeobotanical analysis of this material is given by J. Homo sp. (--Pith. erectus Dub.)
Schuster (1909) and in the Selenka Expedition Report Simia satyrus L. (-=Pith. erectus Dub.)
(1911). These plant-bearing clay beds occur in a mat- Symphalangus syndactylus Desm.
Hylobates cf. leuciscus
3 This is apparently the lowest stratum exposed to view at the Macaca div. sp.
site of the former excavation. Bos (Bibos) sondaicus palaeosondaicus Dub.
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 449
Bos (Bubalus) bubalis palaeokerabau Dub. but in order to visualize the cataclysmic nature of these
Bos (Bubalus) sp. events at Trinil, let us turn to a description of the
Cervus (Axis) lydckkeri Mart. (= C. liroceros Dub.)
Cervus (Rusa) hippelaphus Cuv. ravishing effects of a lahar inundation given by Jung-
Muntiacus mntuntjakkendengensis Str. huhn (1852-54, p. 707 ff.). Junghuhn described such
Dziboisia kroesenii Strem. an event most vividly from the vicinity of the great
Tragzlus kanchil Raffl. Kloet Volcano in Eastern Java (Fig. 100), as follows:
Sus brachygnathus Dub.
Sus macrognathus Dub. The great navigable stream, the Brantas which drains
Hippopotamus namladicus Falc. the Kloet Volcano on its semicircularcourse, became so
Rhinoceros sondaicus Desm. (- Rh. sivasondaicus Dub.) swollen overnight and carried such a tremendousload of
Tapirus cf. indicus Desm. uprootedtrees, dead buffalos,wild cattle (Bantengs), mon-
Stegodon t. trigonocephalts Mart. keys, turtles and crocodilesthat the great bridge at Kediri,
Elephas cf. namadicuis Falc. the largest on the island of Java, . . . soon gave way and
Cryptomastodon martini v. K. was completelydestroyed. The water was as black as ink
Fclis palaeojavanica Str. and consistedof 25 per cent of fine sediment,which when
;elis tigris L. (=F. trinilensis Dub.) evaporated,furnisheda very fine volcanic ash mixed with
Felis parduls L. pumice stone. In a delta branch,the Kali Gempol,on the
Felis bengalensis Kerr. (-F. microgale Dub.) border of Soerabaja and Pasuruan, floated . . . besides
Viverricula malacccnsis Gmel. other wild animals,a tiger and eight humancorpses.
Viverra sp.
Mececyonz trinilclsis Strem. From the same vicinity comes the strange story of a
Canis sp. Dutchman, owner of a coffee plantation, who was un-
Ursus cf. imalayanus Raffl.
Lepus sp. willingly carried away by the lahar and emerged alive
Hystrix div. sp. to tell the tale of how he floated in a warm muddy
Rhizomys cf. sumiatrensis Raffl. paste of sand without ever feeling any hard knocks
Ecliinosorcx sp. from floating boulders (Carthaus, 1911, p. 27). This
unusual experience might explain why the bone frag-
In this group the straight-tusked elephant (El. cf.
ments of the Trinil bone-bed show so little wear.
namnadicus), and the Stegodon, represent distinctly spe- This picture of a catastrophic burying near Trinil
cialized proboscideans not found in the Djetis fauna.
The former species is truly a guide fossil to the Middle gave rise to the conception that the fossil plant remains
found in the thin clay band above the lahar bed were
Pleistocene, being a form closely allied to El. antiqtius
of Europe. Of Stegodon, the Selenka expedition ex- swept down from a higher forest region (van Es, 1931,
cavated several whole skulls complete with tusks. An p. 14). But Schuster (1911, p. 245) in describing the
idea of the richness of the bone-bed may be obtained plant remains came to the conclusion that they grew
near Trinil, representing a type of flora closely akin to
from the original report of the Selenka Expedition
that now living some 3,000 feet above the Madioen
(1911): a total of 3.5 bones per square meter were Plain. Schuster, as well as Elbert (1907), interpreted
counted. The distribution of the bones is irregular,
but the relative percentages of the various genera rep- this as an indication of a climatic change, and in our
resented is rather uniform in the various excavation opinion they were justified in this assumption. The
made Dubois and van Es that the
pits. The most numerous (about 22 per cent) are suggestion by plants
were swept into the plain by the mudflow is neither
horns, especially those belonging to the small Axis
kroesenii, the only survivor borne out by the uniform character of the flora, nor by
lydekkeri and to Dlboisia
of the antelope group in Java. Second only to these the fact that the finest plant tissues are perfectly pre-
are teeth and skull fragments of pigs. Hundreds of served. Had this material been transported from higher
deer antlers have been found at Trinil and very few regions, it surely would have reflected the mixed nature
of them show any signs of water transport, although of plants growing at various altitudes. It has been
curiously enough there are no complete skeletons. claimed that in some cases the identifications of the
This led Carthaus (1911, p. 27) to the assumption that plant remains are unreliable, since they are mainly based
the bones were embedded in a lahar stream which had on leaves. However, in Schuster's report a good many
previously passed over a landscape that had been rav- fruits, flowers and stems are pictured, and their per-
ished by volcanic eruptions to which many of the forest centage in relation to the entire collection far exceeds
dwellers had fallen prey. The hills were burnt and what one is accustomed to in palaeobotanical work. If
covered by ash falls, and the rain-water carved gullies these identifications are questioned, one wonders how
into the soft burial ground, thereby digging up thou- many determinations of fossil plants would stand a
sands of skeletons which were carried away by the critical appraisal by botanists.
rivulets and scattered along their courses. Then came Of the 52 species of plants found in the clay layer,
the mudflow sweeping away the previous wreckage and only 21 occur on the island at present, and of these four
reburying the skeletal remains of the forest animals are still living near Trinil (Schuster, 1911; 1911-a).
along with the debris of burnt or broken trees and un- But there is no truly extinct form. Van Es (1931, p.
derbrush. We will refer later to the plant remains, 11) considers this as an argument against the great age
450 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

gists (compare Elbert, 1907; Dubois, 1908; Dietrich,


of the plant bed, but do the interglacial floras of Europe
behave any differently? Indeed they prove rather con- 1924). In reviewing these writings, two types of ap-
clusively that changes of flora during the Pleistoceneproach are apparent. According to one, the mammal
fauna is related to that of the Asiatic mainland. This
were more in the nature of vertical shifts of plant zones
than of horizontal migrations. Amongst the fossil method of determining the age has to date been em-
plants there are some which Schuster considered as ployed with varying results. The other approach was
climatic indicators, such as Reevesia wallichii, of the
attempted by van Es (1931), who tried to base the
family Sterculiaceae, and a citrus tree, Feronia ele- Pleistocene stratigraphy of Java on the ratio between
phantum, found in India from the Himalayan foothills extinct and living invertebrate species found in the
southward to Ceylon. The laurel, Altyngia excelsa, fresh-water deposits. This latter method, however.
still grows some 3,000 feet above the Madioen Plain. cannot lead to reliable results, because of the incom-
plete number of fresh-water fossils, and more so be-
There were five more species of laurels in the ancestral
forest, and amongst the trees there was one which at cause of the varying adaptability of fresh-water animals
present has a very limited altitude range. This is theto environmental changes.
Rasumela tree, Liquidambar excelsa, a tall member of Shortly after the discovery of Pithecanthroputs, Du-
the Javanese forest not found below 600 m. or above bois (1908) claimed that the mammal fauna of Trinil
1200 m. In the underbrush of the forest which shel- was similar to that found in the Upper Siwalik beds of
tered members of the Pithecanthropus family, thrived aIndia, which at that time were held to be Upper Plio-
species of snowball, Viburnum coriaceum, which in cene. The mammal fauna collected by the Selenka
Java rarely grows below 1200 m. Elbert (1908; 1911) Expedition, as described by Stremme (1911) and
pointed out that the plant layer contains some membersJanensch (1911), did not permit of drawing final con-
of the fig-tree family which today grow at the lower clusions as to age. This collection, although being
limit of the cool and at the upper limit of the temperate
very remarkable as far as well-preserved specimens
plant zone in Java. They required temperatures 6-8? were concerned, unfortunately did not contain many
C. lower than those prevailing near Trinil at present. type fossils, such as Elephas, which would have al-
This thermal difference corresponds to an altitudinal lowed a clear stratigraphic analysis. Stremme, how-
variation of about 1100 m. and a snowline of about 3000ever, pointed out that certain affinities existed between
to 3100 m. The latter computation has a distinct the Narbada fauna of central India and that of Trinil.
analogy to the snowline depression of the Second But Dr. von Koenigswald's recent collecting of verte-
brate remains at the many neighboring localities have
Glaciation in the Himalayas, as I have pointed out else-
where (de Terra, 1940). Schuster insists that the permitted a new and more definite analysis of the
Pleistocene assemblage at Trinil corresponds most problem, and he has presented evidence to the effect
closely to the "flora of laurel plants," such as now that Dubois' Trinil or Kendeng fauna consists in re-
characterizes the Khasi Hills of Assam, located in theality of not less than three different Pleistocene mam-
rainiest corner of the Asiatic monsoon tract. Another mal assemblages (von Koenigswald, 1933; 1934; 1935).
point in favor of Schuster's argument is the fact thatThis new classification of the Pleistocene is based on
he discovered another fossil flora from the tufa de- the fact that the beds with the Trinil fauna are under-
posits at Lasem (Residency of Rembang) in which he lain near Kedeng Brubus, Gunung Boetak and San-
giran by strata containing older mammals such as
identified eleven species out of a total of twelve, which
appear at Trinil. In this flora an oak, Qltcrcus laiel-
Leptobos (a primitive bovid), Epimiachairodus (the
losa, occurs at present only in Sikkim, at the extreme sabre-tooth tiger) and a new species of Hippopotanllus
north of Bengal, between 1500 and 2400 m. Since which is characterized by the small size and initial dif-
ferentiation of the incisors as compared with the Hip-
these fossil plants were found at an elevation of 100 m.
it is evident that Java must have experienced wide popotalmus simplex from Kali Glagah (see p. 426).
climatic fluctuations during the Pleistocene. It ap- As a guide fossil for this stage, Dr. von Koenigswald
pears that this conclusion must have seemed rather in-recognized Cervus swaani, a small deer with almost
significant thirty years ago when nothing was known vertical antlers (see p. 443). Eleplias is represented
by a few molars showing lower and fewer lamellae than
of the nature of the sediments, soils and fauna charac-
terizing the Pleistocene of southern Asia and adjoin- Elephas namadicus. Stegodon, also recorded by mo-
ing lands, and reflecting climatic changes. However, lars only, shows affinities with St. trigonocephalls
to-day a new aspect of the palaeobotanical problem at praecursor rather than with the Trinil form. On the
Trinil is apparent from the improved knowledge of Ice other hand, the Trinil fauna, as collected from Sangiran
Age climates in India and China. and Kedeng Brubus, contains Elephas namadicus and
Stegodon trigonloccphalus,two forms belonging to a dis-
5. THE AGE OF Pithecanthropus AND THE tinctly later stage of evolution (see p. 449). Von
GLACIALCYCLE
Koenigswald (1937) has called this fauna "Sino-
The geological age of the Pithecanthroputsbeds has Malayan" because of the presence of orang, rusa deer,
been discussed by many palaeontologists and geolo- Malayan bear, tapir, rhinoceros and Stegodon. This
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 451

assemblage contrasts remarkably with the Lower long before the initiation of the terrace-making cycle
Pleistocene group, where such types as Leptobos during the time of the formation of the Poetjang or
(primitive bovid), primitive elephant, primitive hip- Trinil Beds. These, as previously mentioned, repre-
popotamus, and Epimiachairoduts (sabre-tooth tiger) sent a period of rapid alluviation or filling of the basin
indicate close affinities with the Villafranchian fauna of north of the volcanic range. Although it is true that
Eurasia. Because of its relationship to the Upper during the Middle Pleistocene the sedimentary supply
Siwalik fauna of India, von Koenigswald called this was greatly augmented by volcanic action, it is like-
older mammal group a "Siva-Malayan" fauna. Thus, wise true that these beds could only have accumulated
while it is perfectly clear that along the Kendeng Hills as a result of stream action. In the basins proper all
both a Lower and a Middle Pleistocene fauna are rep- the Middle and Upper Pleistocene deposits are water-
resented, it still remains to be determined on a more laid, and at the time of the Trinil fauna the streams
precise basis how the Javanese succession may be cor- were rapidly aggrading their beds. This might well
related with the well-established sequence on the Asiatic have been the result of a higher ocean level-i.e., an
mainland. Interglacial period. However, since the bone and plant
\\e know that Asia experienced an Ice Age and beds appear at the very base of the Middle Pleistocene
that repeated glaciations in its southern mountain ram- Kaboeh Beds, it is obvious that stream aggradation
part gave rise to distinct cycles of erosion, clearly re- had not yet fully begun. Quite possibly erosion per-
corded in the foothills and adjoining plains of India, sisted for a short time, in spite of the fact that its
Burma and Central Asia (de Terra, 1937, 1940, 1941; action was interfered with by the advance of a lahar
de Terra and Paterson, 1939). The immediate effect stream and by the temporary ponding of the river.
of each glaciation on these piedmont lowlands was a In such an event both bone and plant beds would
general process of aggradation in the valleys of those fall into the late stage of a major glaciation, an as-
streams which descended from the highlands. They sumption which would also help to explain the peculiar
aggraded their courses chiefly owing to an overloading character of the Trinil flora. Schuster, as well as
of sediment supplied as a result of severe glacial denu- Elbert and Carthaus, seem to agree that this flora
dation in the mountains. As a rule such stages of existed in a pluvial climate, which Blanckenhorn (1911,
glacial advance correspond to stages of alluviation in p. 264) correlated with the First Glacial or First In-
the adjoining plains, which means that during the terglacial. But in view of the new stratigraphic stud-
Glacial periods the streams deposited gravel and sand, ies, it is evident that this particular Pluvial period
thereby heightening their floors. When, during an In- should be younger, since Lower Pleistocene beds are
terglacial period, the rivers incised their courses in present below the Trinil Zone. Accordingly, the lat-
these deposits, terraces were left above their newly- ter does not fall into the beginning of the Pleistocene,
formed valley floors. Hence in a general way the but rather into its middle portion, hence the Pluvial
glacial cycle made for rhythmic alternations of fill- period must correspond to the second major glaciation
ing and cutting, corresponding as a rule to glaciation on the Asiatic mainland. On the assumption that
and deglaciation. Ever since A. Penck (1882), R. A. Schuster's determinations are correct, the Trinil flora
Daly (1934), and others have speculated on the ef- represents a lowland vegetation of a slightly refriger-
fects of the glacial cycle upon ocean levels, the ques- ated equatorial belt. It so happens that the Second
tion of stream behavior in nonglaciated lands has Glaciation was the most effective as far as the Hilna-
been placed in a new light. On the assumption that layan region is concerned (de Terra and Paterson,
glaciation causes ocean level to drop, Molengraff and 1939, pp. 225-229), and we may therefore justly as-
Weber (1921) and Brouwer (1926) and Lehmann sume that its effect upon the warm, humid, tropical belt
(1936) have discussed the possible eustatic effects was particularly strong at that time. The ocean level
which such events might have on the geographical was still low, but may have begun to rise shortly after
configuration of the various islands which constitute the plant-bearing horizon was laid down, because the
the Malayan Archipelago. Valuable as these con- bulk of the Kaboeh Beds overlie the clays containing
siderations are, they seem to lack an objective approach the fossil flora. For this reason we are inclined to cor-
to the question of the origin of stream cycles in Java. relate the actual layer containing the Pithecalnthropits
Lehmann (1936, p. 107) discussed these relationships remains with the Boulder Conglomerate Stage of the
in a purely speculative manner and suggested that in Siwalik foothills, rather than with the somewhat more
the case of the Solo River the two terraces were laid recent Narbada Zone, as von Koenigswald has pro-
down during stages of high ocean level, representing posed (1939, p. 50). Whereas this correlation is to be
the last Interglacial and Early Postglacial periods re- regarded as a mere suggestion based on both fossil and
spectively. In the following discussion on the terrace physiographical records, there is a strong argument in
problem near Ngandong, reference will be made to this favor of linking the youngest Siwalik fauna of India
view (see p. 454). Here it is only essential to point with the Trinil fauna, as Dubois (1908) maintained.
out that such relationships might have been established Now the Boulder Conglomerate of India is in a large
452 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
measure homotaxial with the moraines of the Second ture perhaps is the other argument concerning the
Himalayan Glaciation, a fact which connects the Trinil presence of two human genera in the same layer. Both
Stage with the Asiatic glacial cycle in a manner which Weidenreich (1938) and von Koenigswald (1939) em-
no previous workers could have appreciated. phasize the presence of Homo at Trinil, and the latter
Almost from the very beginning, Dubois' age deter- even goes so far as to state, "So bleibt als einzig sicherer
mination of the Pithecanthropuls finds at Trinil were Pithecanthropus Rest von Trinil nur das Schadeldach
questioned. One group of students pointed out that iibrig" (1939, p. 40). On another occasion von Koe-
the fossils were insufficient to permit of any strati- nigswald (1938) pointed out that the presence of arti-
graphic correlations, while others claimed that the re- facts of specialized type in beds containing the Trinil
mains may actually have been found in a redeposited fauna at Sangiran suggests the contemporaneity of
condition. Hrdlicka (1925) emphasized the "second- Pithecanthroputs and true Homo. During our visit to
ary nature" of the deposits at Trinil and doubted this site in 1938 we did not feel altogether convinced
whether skull and femur of Pithecanthropus had really of this latter claim, not at least as far as the Middle
belonged to the same individual as Dubois had claimed. Pleistocene age of the artifacts are concerned. But it
Little can be added to these discussions except for one is peculiar that thus far no good fossil remains of Homo
point which concerns redeposition. have been recovered at this otherwise very prolific site
At the time of our visit it occurred to me that two near Sangiran. Furthermore the lack of any sure
low terraces could be recognized at Trinil. One can be traces of implements in the fossiliferous Middle Pleisto-
seen during high water extending a few hundred yards cene of Java leaves us with a certain shadow of doubt
downstream (compare PI. XXXIV, Fig. 1). It was regarding the above claims. To judge from our knowl-
present at the time of the Selenka excavations, as indi- edge of other sites of Early Man in other parts of the
cated by one of their expedition photographs (1911, Old World, it seems improbable that the two genera ex-
Fig. 17). Now, since Dubois as well as later investi- isted side by side. Hence the question of whether there
gators dug on that side of the stream-meander where has not been some redeposition in the Trinil horizon
deposition is still taking place, is it possible that their during Late Pleistocene or even post-Pleistocene times
pits were partially sunk into this younger terrace? To demands further field study.
all appearances, these deposits should be post-Pleisto-
cene, and if excavations had been carried out close to VI. THE UPPER PLEISTOCENE TERRACES
the river bank, they would have struck, in all probability, AND THE AGE OF SOLO MAN
the younger alluvial formation. It is therefore conceiv-
able that some -of the fossils excavated were derived Of the three divisions of the Pleistocene in Java the
from this horizon. Nevertheless, incomplete as the ini- upper is the least known. This is because the Upper
tial excavation records are, it would seem that most of Pleistocene formations have been largely denuded from
them, especially the Pithecanlitropus remains, were actu- the crests of the young anticlines, while at other places
ally discovered in situ in the bone-bed and not simply they are covered by recent alluvium and are not ex-
"found" as some of the text-books state. In fact Du- posed. Actually they are only known on the Solo River
bois was very emphatic about the exact location of the in connection with ancient stream terraces, at least the
finds, and the Selenka Expedition could also prove that fossiliferous division of them has thus far been found
most of their material was recovered in the bone-bed. only in these deposits. Hence our .description con-
A sceptic, however, might still argue that the Pithecan- cerns chiefly those localities which can reliably be re-
thropus fragments show varying degrees of weathering, garded as Upper Pleistocene.
and that the admixture of more modern human remains Along the slopes of the Kendeng Hills the Upper
with the older forlm argue for redeposition, as von Pleistocene deposits are known by the term of Noto-
Koenigswald (1939, p. 40) claims. Dubois himself poero Beds. Wherever I saw them in the field I was
says that the skullcap as well as the four femora de- impressed by the variety of rock constituents which they
scribed in 1932 and 1934 are intensely corroded,4 contain, by their very low dip, and by the prevalence
whereas the first femur and the jaw fragment from of rather coarse boulder-bearing layers in them. The
latter form the resistant escarpment at Sangiran (see
Kedeng Brubus are quite fresh-looking. However, such
variations in the degree of fossilization may well have p. 445 and Fig. 102), where they are derived from older
occurred prior to the final deposition of the material; eruptions of the Lawoe Volcano. Similar volcanic
one must remember that it was exposed for some time detritus is present along the Solo River downstream
from Ngawi, where fine, gray, tuffaceous layers and
prior to the formation of the bone-bed, and that it may tuffaceous sandstones reveal distinct cross-bedding.
well have been exposed on several subsequent occasions
While the three above-mentioned characteristics dis-
in the shifting sands on the flat river bottom during
the Middle Pleistocene. Of a more fundamental na- tinguish the Upper Pleistocene Notopocro Beds from
the two older divisions, it is evident that the formation
4 The femur fragment,
allegedly from Kedeng Brubus (Du- of terraces is the main event which introduces an en-
bois, 1935), is likewise stated to have been heavily corroded. tirely new chapter into the geological history of this
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 453

region. For from now on we are dealing with deposits Homo neandcrthalenlsis soloensis Opp.
associated with the Macaca sp.
closely present drainage lines-e.g. Bos (Bibos) sondaiculs palaeosondaicus Dub.
along the Solo River the Upper Pleistocene is docu- Bos (Bitbaluis) bubalis palaeokerabau Dub.
mented by terrace deposits only. Evidently there was Cervus (Axis) lydekkeri Mart.
an uplifting of the Kendeng Hills which caused the Cervus (Axis) axis javanicits v. K.
stream meanders to cut vigorously into the anticlinorium Ccrvius(Rusa) hippelaphusCuv.
until stream equilibrium was established. Under these Cervus oppenoorthi v. K.
Mulntiacus muntjak Zim.
conditions aggradation led to the deposition of the Sus terhaari v. K.
younger alluvium. Before we discuss the physiographi- Sus macrognatlhs Dub.
cal aspects of this new development, it is necessary to Sits ex. aff. vittatus Tem.
look into the geological and palaeontological evidence Sus vatuzalangensisv. K.
for the Upper Pleistocene age. Hippopotanmusnamadicus v. K.
Rhinoceros sondaicus Desm.
From the foregoing remarks about the Upper Pleis- Stegodon t. trigonocephalus Mart.
tocene, it seems evident that the terrace formation must Elephas cf. namadicus Falc.
be younger than the synclinical beds containing the Felis palaeojavanica Str.
Felis cf. pardus L.
Trinil fauna. In fact their superposition on the folded Felis tigris soloensis v. K.
Older Pleistocene was possible only after the uplift and Cuon crassidens v. K.
dissection of the anticline. This uplift is regional and Sits ex. aff. vittatus Tem.
very characteristic of the orogenic history of the Ken- It is an impoverished Trinil fauna, as von Koenigs-
deng anticlinorium. Duyfjes (1938, pp. 59 ff.), in de- wald has remarked, and was therefore
formerly re-
scribing the evidence for this Pleistocene folding in the garded as a facies of it. What exactly caused
special
Soerabaja area states (translated): impoverishment is not discussed, but we suggest two
After the Middle Pleistocene (Kaboeh Period) there possible reasons: (1) the isolation of Java due to the
again occurreda strong orogenic phase which made itself submergence of a land-bridge, or (2) the extermina-
felt all over eastern Java and to which the various anti- tion of certain forms by Stone Age man, or both. Of
clinal hill countries owe their present form. . . . The in- all these
genera Stegodon is the only truly extinct form;
tensity of the folding in the Kendeng anticlinoriumgrew the modern character of the assemblage is emphasized,
weaker toward the east . . . so that the various anticlines
one after anothersplit off from the foldedbelt and vanished however, by the absence of antelopes and of Crypto-
into the syncline to the south. ... In the north the anti- mtastodon,and by the presence of modern Indian deer.
clinoriumof northernJava was formed,the core of which Von Koenigswald himself admitted that it was the
alreadyexisted at the close of the Miocene. The intensity stratigraphical position of this fauna in terrace deposits
of the folding is here generally less as comparedto the which
Kendeng Hills.... The place of this youngest folding prompted this distinction from the Middle
phase in the geological time-scale . .. is indicatedby the Pleistocene. In my opinion, of equal if not of decisive
fact that it has influencedthe Middle Pleistocene Kaboeh importance is its association with a neanderthaloid type
Beds, while in the vicinity of Ngawi this stage is discon- of man who had a fairly specialized culture, which in-
formably overlain by terraces which are not folded and cludes tool implements made of bone and rock. While
which von Koeningswaldregards as Upper Pleistocene on it is
the basis of the vertebrates (Ngandong fauna) embedded impossible as yet to assess the technical achieve-
therein. ... In any case it is certain that this orogenic ments of these people in terms of those of other Upper
phase took place after the Middle Pleistocene while a still Pleistocene races in Asia, it is evident that the bone in-
younger age is not entirely excluded. dustry of Ngandong indicates a fairly high standard of
living as compared with that of such Middle Pleistocene
The last statement contradicts somewhat the former cultures as are found associated with Peking Man at
observation concerning the discordance between the Choukoutien. At
Karsono, on the Solo River, between
Middle and Upper Pleistocene, for obviously the folding
Ngawi and Ngandong, Dr. Movius and I found chert
must have taken place prior to the appearance of the and
jasper implements in 1938 in the gravel of the high-
Ngandong fauna. Hence, it would be more correct to est terrace-T, in Fig. 104-A. These implements
say that the folding took place solletime between the closely resemble the so-called Sangiran industry. The
deposition of the Kaboeh and the Notopoero Beds, and latter I believe to be of Notopoero age (see p. 455),
that it may have been revived during post-Pleistocene which would correspond with the
age of the Solo ter-
times. In fact we know that some of the stream ter- races. Archaeological sites will no doubt be found at
races were tilted, and that in central Java strong com- many new places in Java as soon as a systematic search
pression of the folded belt took place at a much later for them is begun. In any case it is my opinion that
period (van Bemmelen, 1937). an Upper Pleistocene dating of the Solo terraces can-
The palaeontological evidence for the Upper Pleisto- not be questioned, and on this basis it is safe to state
cene age of the fossiliferous terrace formations is based that the Upper Pleistocene in central Java forms a
on the following faunal list given by von Koenigswald distinct break with the previous development of the
(1939, p. 45): landscape. This break is marked by a disconformity.
454 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY

T2 :o.. ...o o'.-::6.oo


0/
oooo __ 0oo 00 0
000ooo0

A SoloR T3 Ti
(40- 0 m.)

T2

T3
06 .........
. 7f
(20 m.)

/o6
B RSo
l (4m.) /

FIG. 104.
A. Geological Section Through the Terrace Deposits of the Solo River, Near Karsono.
T2. Ngandong Terrace containing Artifacts.
T3. Silt Terrace.
B. Section Through the Solo Terraces at Padasmalang.
T1. High Terrace (40-50 m.).
T2. Low Terrace containing Vertebrates (20 m.).
T3. Silt Terrace.

From the beginning of the Upper Pleistocene on the presumably represent erosional features formed during
Solo River, the most important drainage system of the last great lowering of sea-level in the Pleistocene,
central Java has had to cope with the uplift of the i.e., the last glaciation. Fig. 5-B would indicate that
Kendeng Hills across which it flows. This uplift evi- there may be two substages in this Low Terrace; evi-
dently was not continuous, rather it was intermittent, dently it was greatly dissected as a result of a further
as in the case of most mountain uplifts, and the stream lowering of the base-level. Lehmann (1936, p. 99 ff.)
adjusted itself to the new conditions. These adjust- has shown that the Low Terrace (T2) was tilted and
ments are indicated in part by at least three terraces the same is true of the High Terrace (T1). It must be
which can be seen on both sides of the river between mentioned, however, that Lehmann assigns the High
Ngawi and Ngandong. At a village called Padas- Terrace to the "younger Old Quaternary" by which he
malang I made a sketch of these terrace profiles shown means a phase intermediate'between the Trinil Beds
in Fig. 104-B. The highest-T--is some 40-50 m. and the Young or Upper Pleistocene stage. The term
above the stream. Intensely dissected, it consists of "younger Old Quaternary" therefore is somewhat mis-
hills covered by rather coarse gravel and sand from leading and should rather read "older Upper Pleisto-
which fossil collectors of the Dutch Geological Survey cene." Thus it occupies an analogous stratigraphic
collected Stegodon and other remains of the so-called position between the Middle and Uppermost Pleisto-
Ngandong fauna. This level presumably is what Op- cene as in the case of T3-an erosion terrace-in India
penoorth (1932) called the "Ngandong" terrace after and Burma (see Table 1, p. 331). Now such an
the village downstream from Padasmalang where the analogy may seem puzzling in view of the radically
eleven skulls of Solo Man were found. The under- different origin of the several terrace systems. But,
lying Miocene bedrock outcrops along the slopes; its whereas in southern Asia streams degraded their
dip is 60? toward the west. The next lower terrace- courses during interglacial periods, the rivers of Java
T2-lies about 20 m. above the stream and is charac- and the coastal regions aggraded theirs, in consequence
terized by gravelly tuffaceous sandstone and fine tuff. of the higher ocean level. Hence, if such an analogy
Lehmann (1936, pp. 80 and 104) refers to T1 and T2 really exists, it appears doubtful if the Solo terraces
as the "High" and the "Low" terraces respectively. were really controlled by uplift rather than by changes
The former, T1 in my sequence, he assigns to the last of ocean level. Nevertheless Lehmann's excellent work
Interglacial. He pointed out that its average elevation must be followed by a more detailed field survey in the
of 50 m. above sea-level corresponds well with A. Solo Valley before we will have a satisfactory solution
Penck's (1933) requirement for a maximum rising of to the question of terrace correlations. It is very prob-
the ocean by 55 m. during complete deglaciation of the able that marine platforms exist on the coast which may
polar regions. While such a computation might ap- prove the key to correlations between high ocean levels
pear uncertain in the light of subrecent diastrophism and terrace deposits.
in Java, it fits rather well with our conception of Up- To complete the physiographical picture of the Solo
per Pleistocene stratigraphy in Java as shown by Chart terraces, it should be stated that a low bench (T.) ac-
2 (see p. 455). In such a case the slope below the companies the river over great distances (Fig. 104-A
Low Terrace (T2), as well as the Low Terrace itself, and B). Its height is 4-5 m. above stream level and
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 455

on it are found the "sawahs" or fields, and villages. when the new geological survey of this area was made
Lehmann (1936, p. 107) correlated this with a Post- by Mr. ter Haar in August 1931, a complete skull of an
glacial phase of marine sedimentation in the lower extinct buffalo was found by him near Ngandong (Fig.
course of the river. This phase of aggradation was 100). The next month W. F. F. Oppenoorth, then
followed in recent times by erosion, probably a result director of the Dutch Geological Survey, started an ex-
of a last lowering of the sea-level. cavation at this very place (Oppenoorth, 1932; 1937;
ter Haar, 1934). It yielded two human calvaria and
1. THE GEOLOGYOF THE NGANDONGSITE
three skull fragments; towards the end of the year two
The presence of river terraces near Trinil had al- excavation pits had produced not less than three thou-
ready been suspected by Elbert (1907, p. 658), and sand fossil animal bones. During the following year
CHART 2
PLEISTOCENE STRATIGRAPHY IN JAVA ALONG THE SOUTHERN SLOPES OF THE KENDENG HILLS AND SUGGESTED
CORRELATION WITH THE ASIATIC GLACIAL SUCCESSION
GEOLOGICAL

SUGGESTED
GEOLOGICAL DEPOSITS AND FAUNAL SUCCESSION GLACIAL
PERIOD CORRELATION

Silt terraces (Ts) and high flood plains


SUBRECENT SAMPOENG FAUNA-Proto-Australoid people POSTGLACIAL
Volcanism and uplift accompanied by tilting

Erosion-formation of T2 (20 m.)


Volcanism and earth movements 4th GLACIAL
V)
WADJAK MAN (?)

CP
0
Stream aggradation-T1 (40-50 m.)
0 NGANDONG FAUNA-Homo
Q, neanderthalensis soloensis Opp. 3rd INTERGLACIAL
ck
0F34 Volcanism
0(Z

Erosion and uplift


NGANDONG FAUNA 3rd GLACIAL
Volcanic lahar deposits

1h C fi ermfitvtv
C-)
:4
0
w
C,) Erosion and strong uplift of the hills
Aggradation in the synclines 2nd INTERGLACIAL

KABOEH BEDS (4175 m.)-fluvial


TRINIL FAUNA-Pithecanthropus erectusDub. 2nd GLACIAL
Volcanic lahar deposits

POETJANG BEDS (= 100 m.)-estuarine and fluvial


deposits containing volcanic material
DJETIS FAUNA-Pithecanthropulssp. (?) 1st INTERGLACIAL
Homo modjokertensis v.K.
0
I
UPPER KALIBENG BEDS-
Marine Globigerina and coral limestones 1st GLACIAL

UPPER LOWER KALIBENG BEDS-


PLIOCENE Globigerinamarl
456 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
two additional human skulls and one skull fragment by evergreen jungle, animals such as buffalo, deer,
were found in association with bone tools (Oppenoorth, bison, pigs and elephants came to drink or to ford the
1936), and as the excavation proceeded, six other tributary stream. In so doing some of them may have
calvaria were brought to light.5 This represents with- been caught in the quicksand during the period of
out doubt the richest haul of human fossils ever to have stream aggradation (T1). Others probably were
been made in an alluvial formation. preyed upon by jungle cats, and many may have been
The bones excavated by Oppenoorth were found trapped by man, for it is at such places that the trapping
some 20 m. above the river (PI. XXXIV, Fig. 2). of big game is facilitated by the boggy nature of the
Between the river and this site a narrow bench can ground. Very likely Stone Age man had a camp here.
be seen on which the small village of Ngandong is This is indicated by the presence of implements such as
built. This is Lehmann's Low Terrace-T2 in my harpoons, bone scrapers or knives, awls, and antler
sequence (see p. 454). Now, as stated above, the ex- picks. Oppenoorth remarks that many of the animal
cavation pits lie on the upper edge of the 20 m. plat- bones were split and artificially pointed (1936, p. 405).
form, a level which was formed during an erosion period At this site there is no doubt that the tools are con-
when the river incised its channel into the older T1 de- temporaneous with extinct man.
posits. Thus, since T, is an erosional terrace, the ma- The prevalence of buffalo and bison remains at
terial fourrdin it presumably dates from the preceding Ngandong has been interpreted as indicating a pre-
depositional cycle when T--the main or "Ngandong" historic sacrifice, or purposeful slaughtering of animals
terrace-was formed. At Ngandong itself the T2 level over long periods of time. In spite of the fact that the
slopes perceptibly towards the river, giving the impres- human skulls exhibit traces of decapitation, it seems
sion that some tilting has occurred. Twenty to thirty obvious from their stratigraphic position, described
meters above this level another flat surface appears, above, that they were swept together by the river rather
which is our first terrace (T,). than killed on the spot. Hence it is doubtful whether
The deposits on T, are composed of alternating layers any permanent settlement connected with the sacrificial
of loose tuffaceous sand (marly in places), and of silt killing of animals can be held responsible for the extra-
up to 3.50 m. thick. According to Oppenoorth (1932, ordinary accumulation of over 20,000 bones.
p. 51), the skulls and the majority of the other fossils
were recovered from a basal gravelly sand (40 cm. VII. THE STRATIGRAPHIC POSITION OF
thick) and the overlying limy tuffaceous sands (10-30 THE PALAEOLITHIC CULTURES
m. thick), which rest directly on Late Pliocene Glo- OF JAVA
bigerina marl. These layers are covered by 2.50 m. of
volcanic material of alluvial origin. This means that 1. THE SANGIRAN INDUSTRY
the human material was deposited at the beginning of a
In 1934 Dr. von Koenigswald found a series of primi-
stage of stream aggradation (see Chart 2, p. 455). tive artifacts of stone in what he took to be the Trinil
Probably the remains of Homo neanderthalensis soloen- horizon at Sangiran, near Soerakarta, and he speculated
sis were washed from the land surface into the river
on their possible association with "some human teeth
bed, and they may have suffered weathering prior to with marked Neanderthal affinities . . . found with a
their deposition. This contention is borne out by the
fauna typical of the Trinil horizon" (1936, p. 52).
thick incrustations found on most of the calvaria, which
While visiting Sangiran, we searched deliberately for
were covered by a hard, limy matrix similar in com-
these stone tools, first in the layers with a derived
position to the pellets observed in an exposure some Trinil fauna and later in the overlying Notopoero Beds
seventy meters distant from the skull site. Here small
(Layers 4 and 5 in Fig. 102). While the former were
pellets of lime form a thin layer in gray tuffaceous silt. found to be sterile as far as implements are concerned,
Animal bone fragments and teeth were similarly in-
the uppermost gravels (P1. XXXV, Fig. 3-Layer 5
crusted, and I have no doubt that this represents a
fossil soil at the base of the terrace, subsequently buried in the section, Fig. 102) proved to be rather rich in
under alluvium. precisely the type of flake tools which Dr. von Koenigs-
The two excavation pits at Ngandong lie south of a wald had ascribed to the Middle Pleistocene deposits.
In a later publication (1939, p. 42), he mentioned that
large river bend that follows the strike of the marine
Tertiary formations. Here a tributary joins the Solo the flake tools did not come from the same beds as those
River from the northwest, and it has assisted in widen- containing Pitlhecanthropius,but from layers somewhat
ing the loop by forcing the current towards the opposite higher in the sequence. While it is possible that im-
bank. This situation must have made for quick deposi- plements occur in the upper deposits, I doubt that these
tion in a quiet backwater of the Solo stream. In such beds belong to the Middle Pleistocene for two rea-
a protected corner, where mud flats were surrounded sons: (1) no artifacts were found by us below the
5 Professor W. J. Mijsberg very kindly allowed me to ex- upper gravel of the Notopoero Beds, and (2) the in-
amine the original Ngandong skulls in Batavia, which courtesy dustry found in the upper horizon is much too advanced
is gratefully acknowledged. to have been manufactured by so primitive a type as
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 457

Pitlecanthlropus. For these tools do not resemble in in a dry river bed, although a few were said to have
the least the earliest known Stone Age cultures in Asia, been extracted from a "boulder conglomerate" on the
such as the industry of Choukoutien; on the contrary, river bank. In a more recent article von Koenigswald
they can be compared only with the semi-precious stone (1939, p. 43) stated that this culture was known from
industry of central India, which is surely not older than surface sites only. The boulder bed was originally
the Upper Pleistocene. At Sangiran there are small described as having been folded, and this led the dis-
single- and double-ended scrapers, steep scrapers, cores, coverers to assume a Middle Pleistocene age for the ma-
blunt and pointed flakes, and burin-like tools. Similar terial, an assumption which at first appeared to be cor-
artifacts were collected by Dr. Movius and myself on roborated by the typology of the implements. These
the surface of a Solo River terrace (?T.) several miles were alleged to represent the "Chellean" of Western
upstream from Ngandong at Karsono (see p. 454). I Europe. The discovery was regarded as highly signifi-
presume that both belong to the same period, and that cant, because thus far no Palaeolithic hand-axe cultures
they represent an Upper Palaeolithic culture, which is had been found in Java or any other of the neighboring
definitely not to be associated with Pithecanthropus, islands. These new finds focused attention on the prob-
but which may possibly be connected with Homno n. lem of stone tool manufacture in the Pleistocene of
soloensis. This attribution is verified by their associa- southeastern Asia. The first impression evidently was
tion with the Ngandong fauna in the Solo terraces. that here a basic Palaeolithic culture had been found.
Another argument in favoring a Late Pleistocene age Now the Zuider Mountains are a karst region formed
for this industry is suggested by the fact that pebbles of by an uplifted and tilted block of Upper Miocene lime-
chalcedony and jasper (of which the artifacts are made) stone, which rests on a volcanic formation (Fig. 105).
are found in large quantities all along the Kendeng Hills While the latter is represented by shales and agglom-
in the Nlotopocro Beds, but not in earlier deposits. erates, the limestones form compact homogenous masses
of rock. Naturally both these formations differ con-
2. THE NGANDONG BONE INDUSTRY siderably in their resistance to stream erosion and to
weathering. The volcanic shales and tuffaceous beds
Oppenoorth (1936) has reported on a peculiar bone
might be considered the least resistant of the series,
industry found in the terrace deposits near Sidoredjo
except for the fact that they are made harder by layers
(near Ngawi) and Ngandong in the Solo Valley. Van of silicified tuff and fossil wood. These can be seen in
Stein-Callenfels (1933) also described some of these
tools and was inclined to regard them as Mesolithic or the bottom of the valleys, where the streams have cut
into this old volcanic formation, washing out sizeable
Early Neolithic. The former authority even spoke of
a harpoon of Magdalenian type, but this seems hardly pebbles or even boulders of silicified material. Whether
the structure of this karst landscape is normal, or
justified in view of the absence in Java of any known
dated cultures such as the Magdalenian of Western whether the limestone was thrust up on top of the vol-
canic series, was impossible to decide in the short time
Europe. Oppenoorth (1937, p. 359) argued that one
of the differences between the Solo and the Neanderthal at our disposal. I would even hesitate to state with any
Man was cultural, in as much as "the Neanderthal civili- degree of certainty that the limestone actually overlies
zation is characterized by stone implements, that of the volcanic series. But this question is of no great im-
Ngandong by implements of bone or stag horn." Of portance as far as the stratigraphic position of the
course this may be purely accidental, since stone balls Palaeolithic implements is concerned. What is signifi-
do occur and chert flakes have been found near by on the cant is the fact that the karst in this area is filled both
terraces. Naturally Solo Man made use of the enor- with red earth (terra rossa) and with a younger volcanic
mous supply of stag horn at the Ngandong locality, in ash deposit. The latter is the younger because it fills
fact this may well have been the sole supply of work- the dissected relief; the terra rossa on the other hand
able raw material at his disposal in this case. I believe is presumably of Pleistocene age, as it forms part of the
that the data are as yet far too incomplete to warrant fissure and sink-hole fillings of the Zuider karst. Dr.
any clear evaluation of correlations between typological von Koenigswald (1937, p. 29) stated that the fissure
stage of cultural development and geological age, in the fauna which he collected in this region corresponds to
case of the Solo Valley. Here lies a fruitful field for the Trinil fauna. Thus we seem to have in Java a close
research in a region which may possibly provide the analogy with the karst sequence of the Northern Shan
key to the problem of Stone Age civilization in southern Highlands (see p. 327). As in Burma, the Zuider
Asia. karst may be connected with a pluvial climate: very
3. THE EARLY PALAEOLITHICOF PATJITAN likely in the main with the second pluvial of Early
Middle Pleistocene age. Another analogy may be cited
An Early Palaeolithic hand-axe industry has been between Java and South China, where the fissure fauna
reported by von Koenigswald (1936; 1936-a) from the is also of Middle Pleistocene type.
vicinity of Patjitan in the Zuider Mountains (Fig. 100 Our investigations are not sufficiently thorough to
and PI. XXXV, Fig. 1). The implements were found permit us to discuss Lehmann's observations presented
458 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
in a special memoir dealing with the origin of the Pleistocene formations which may be called Middle and
Zuider karst (Lehmann, 1936). In general we can Upper Pleistocene. These probably correspond re-
confirm his impression that there has been a considerable spectively to the Trinil and Ngandong Stages of central
amount of tilting of the limestone massif, but whether Java. The two terraces mentioned above cannot as yet
this resulted in the karstification, as he claimed, seems 1)efitted with certainty into either of these divisions, but
to me doubtful. I presume that the formation of karst to judge from their relation to the Mt. Lawoe volcanic
here as elsewhere in southeastern Asia is chiefly a result series, it appears that the first is of Ngandong and the
of climate. Probably it began at the close of the Ter- second of post-Ngandong age.
tiary, when Java emerged as an island, and it may have As previously stated, Dr. von Koenigswald announced
received its greatest stimulus through the successive in 1936 that he and Mr. W. F. Tweedie of the Raffles
pluvial periods of the Pleistocene. The present drain- Museum, Singapore, had discovered Palaeolithic arti-
age of this region is directed northward, following the facts in situ in a tilted conglomerate in the vicinity of
dip of the tilted and uplifted limestone block. The Patjitan. We therefore expected to find implements in
karst streams originate near the coast, and they flow one of the terrace gravels. Now the bed of the Baksoka
through deep youthful valleys, some of which are up to River at the village of Poenoeng (P1. XXXV, Fig. 2)
300 m. deep. These deep valleys may date back to the is actually one of the main sources of the Palaeolithic
time when an incipient stream pattern had just started material. Most of the tools are made of silicified tuff;
to drain the karst depressions. On the basis of the fis- they are waterworn and apparently have been derived
sure fauna we know that the depressions existed during frolm an older alluvium. In the Baksoka Valley (Fig.
the Middle Pleistocene; presumably the karst was fully 105) there are two terraces: one lies 10 m. above
developed at that time. At a later stage (perhaps stream level and the other is 15-20 m. high. At the
Upper Pleistocene), the streams attained a more even base of the former is a thin layer of shingle, composed
gradient and the valleys were filled with alluvium: the chiefly of silicified tuff and fossil wood. The accumula-
present high-terrace deposits. Another dissection en- tion of silicified tuff and fossil wood in this lower ter-
sued. Again this was followed by a stage of alluvia- race greatly exceeds the present supply of these rocks
tion, leading to the formation of a lower terrace. Sub- in the recent stream beds. Hence it is quite certain that
sequent to the period of the first terrace, a great volcanic Palaeolithic man found a richer supply of suitable raw
eruption occurred in the neighborhood. The ashes and material during the time of his occupation of the area.
agglomerates resulting therefrom buried the karst and It was from the basal stratum of the 10-m. terrace
the existing relief. These volcanic deposits are pre- that we extracted a few rolled implements of "Patji-
sumably part of the younger Mt. Lawoe eruptive series, tanian" type, and I presume that this is the "boulder
and homotaxial with the Notopoero Beds of the central conglomerate" referred to by the above-mentioned
plain of Java. Inforlnation to that effect was furnished authors.
by Dr. van Bemmelen at Bandoeng, whose intimate A second level or terrace was seen some two miles
knowledge of the structural and volcanic history of the upstream from the village of Poenoeng. On the right
island was imparted to us on several occasions. Ac- bank there are dissected remnants of flat benches some
cording to his interpretation, the younger Mt. Lawoe 50 to 60 feet above the level of the stream. They are
volcanics can be easily recognized by their petrological composed of very coarse boulder gravel with 20 feet of
characteristics, hence there can be little doubt as to the red loam on top. This formation is free from volcanic
validity of the correlation indicated above. In other ash, indicating that it originated prior to the younger
words, in the vicinity of Patjitan there are at least two Mt. Lawoe eruptions. Quite possibly the implements

FIG. 105. Generalized Geological Section Near Poenoeng (Zuider Mountains).


1. Miocene Limestone-Karstified.
2. Red Earth (Terra Rossa).
3. Volcanic Ash.
4. Older Volcanics with Silicified Tuff and Fossil Wood.
T1. Upper Terrace.
T2. Lower Terrace.
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 459
were manufactured during this first terrace stage, but by a statement by von Koenigswald (1935) who main-
we did not actually find any in the boulder gravel or tained that "Sumatra . . . did not play any role in the
loam. Rainy weather prevented us from intensive connection with Asia since it has yielded no Tertiary
searching, as the ground was so moist and swampy that and Pleistocene mammal remains." His other argu-
it was difficult to find any clear exposures. My impres- ment is based on the fact that the so-called "Middle"
sion is that the question of the absolute age of the and "Upper Pliocene" faunas of Java are impoverished
Patjitanian industry can be determined only when the as compared to those of India. For instance the Equi-
terrace geology of the region is known. No doubt the dae and Giraffidaedo not appear in Java, and elephants
finds of promising fossiliferous fissure deposits in the and highly specialized pigs are also missing in these
Patjitan area would be greatly augmented if a physio- deposits. His conclusion is that Java cannot have been
graphical study of the entire karst drainage were done. intimately connected with Asia, not at least over long
Then it should be possible to trace the karst deposits periods. More recently, however, the same author
into the alluvium in the same manner as the fossiliferous (1939, pp. 48-49) voiced a different opinion. "The
karst formations of Burma were correlated with the growing Java," he said, "was populated first from India.
boulder fan stage of the depressions (see p. 324). This is indicated by Hippopotalmusand Merycopotanims,
One question which occurred to me was whether the both typical for the Siwalik fauna of India." He con-
distribution of the implements near Poenoeng could be tinued: "In the Djetis fauna on the other hand we find
traced to a special concentration of raw material. If it elements which are lacking in the Siwaliks: Simllia,
is true that fossil wood and silicified tuff are more com- Symiphalangus, Hylobates, Tapirus, Malayan bear etc.
mon in the older volcanic series than elsewhere, such a ..." This statement needs correction in as much as
supply could have been available only after the streams Simia is represented in the Upper Siwaliks (Colbert,
had cut down into this volcanic formation. This could 1935). Furthermore, how much do we know of the
hardly have taken place when the fossil karst fauna geographical range of the Indian Upper Siwalik fauna
lived in the region, because at that time karstification toward the southeast? Must the Siwalik elements in
was still in progress, and it was not seriously intercepted the Kali Glagah and earlier mammal faunas of Java
by rejuvenated slope drainage. In other words the necessarily have come from India? Indeed nor, be-
streams could not have reached the older volcanic series cause it is an established fact that the geographical range
(containing silicified tuff and fossil wood), before the tended southeastward into the Irrawaddy Valley of Up-
later part of the Middle Pleistocene. Therefore it is per Burma, and it may indeed have gone into what is
probable that the most primitive implements of the now Chinese Yunnan. Colbert (1940) has shown that
Patjitanian industry are late Middle Pleistocene or the Upper Siwalik type of horse in Burma, which I
possibly early Upper Pleistocene. collected in the Irrawaddy Valley in 1937, is identical
with Equils yunnanensis of southern China (see p.
VIII. JAVA AND THE QUESTION OF 3). Also there are close relationships between the
LAND-BRIDGES Middle Pleistocene fissure fauna of Upper Burma and
that of the Yangtze Valley. All this would indicate an
With the discovery of Late Tertiary and Pleistocene
extended range of the Siwalik fauna over southeastern
land mammals in Java the problem of their possible
Asia, pointing to the conclusion that the early land
migration route is connected. Up to the present zoo- mammals should have had a choice between two chief
geographers and geologists have been in rather close
agreement concerning the former land connection of migration routes: one leading via the Sunda Shelf, past
Sumatra or Borneo, and the other via Formosa, Luzon,
Sumatra, Java and Borneo as part of the so-called
"Sunda Shelf," which extended from the Malay Penin- and Borneo to Java. In this light it may be ques-
sula southeastward to the Straits of Timor and Macas- tioned whether von Koenigswald's differentiation be-
sar (Fig. 106). According to Rensch (1936), there can tween a "Siva-Malayan" and a "Sino-Malayan" fauna
be no doubt that mammals, as well as lower vertebrates, during the Pleistocene carries any significance with
dispersed from "Sundaland" to these various islands. regard to the past geography of the early mammal
Stresemann (1939) has recently shown that many migrations to Java. As far as our records permit us
birds also migrated on another route: via the Philip- to judge this matter, it would seem that the former
pines and Celebes. He suggests that some, such as the route was the most likely of the two.
weaver bird, reached Java during a dry period when During the Middle Pleistocene the island was popu-
grasslands extended over eastern Java, southern Borneo, lated, at least in the eastern lowlands, by men of the
the Lesser Sunda Islands, Luzon and eastern New Pithecanthropus type. The close phylogenetic rela-
Guinea. It is certain that the latter connection was of tionship between Pitlhecanthlropltsand Sinanthroplts
no importance as far as the majority of the land mam- would suggest a common origin for both-possibly
mals in Java is concerned, since it is much more likely somewhere in southern Asia. Presumably Java Man
that most of the migrations took place via Sumatra and migrated from there with the straight-tusked elephant,
Malacca. This generally excepted view was challenged with Cervuls lydetkkeriand Hippopotamus, an associa-
a 1Z0' \130 14
t110O 1120' ,130' 1
I

ORMOSA Pleistocene La
-40 Fathom
and Sahu
---- 100 Fathom
Maindrai
20Z .---. in Sunda r
==C = Migration
9 Fossils with Up
I.,.... 0 Fossil Man
(7.. A Palaeolithic cu
E Pleistocene cav
00
0 100 200 300 400 5
4r..- I L -wO

.:: -,iow.z i
I .:.

t...;
I:
10'
C',
L':. WA", 7
_, l e
C

N
..'..A

0 c
g rlc t.\
F"
:-Scl-I??.?:
r?r? ' =? r '5.
t =?'I:= -::::'=I ),.A"O",

t'
I["
Ir
O
a~oSU~';C

^ '- --

_z.. ,.-a . !) ^
*?.l" I
'
,/::
10l eV V ...:':.'".i
- .. -- -
M
-r
(I;e? ?
IA jf 0 C ,& 72

! A
h t
'100' l110o 1120?
--

FIG. 106. Pleistocene Land Connections in the East Indies During the Glacial Periods of Low Sea-level and the Suggested Migr
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 461
tion comlmon to both these extinct human races. But birds as the weaver in the Sunda area. Species and
the route of this interesting migration is yet to be estab- racial differentiations in their geographical range sug-
lished. It may have led via the Philippines, where re- gest that the area suffered at least two pluvial and two
mains of an Old Pleistocene type of fauna have been dry periods. Following a suggestion made by van
recovered in Luzon and Mindanao.6 On the other Steenis (1935, p. 398), Stresemann considers that these
hand, there are cultural relationships between Java and climatic changes were induced by the varying outlines
Burma which suggest the western route. A distinct of the Sahul Shelf (see Fig. 106), which during peri-
typological relationship exists between the Patjitanian ods of glaciation, or low ocean level, exerted a dessicat-
(Lower Palaeolithic) culture of Java, and the chop- ing effect upon the monsoon winds. But, in our opin-
ping-tool industry of Perak in northern Malaya. This ion, the effect of a general lowering of the temperature
in turn closely resembles the Early Anyathian culture during the Ice Age is the more plausible explanation;
of Burma, as pointed out by Dr. Movius (see pp. 376- and this in turn had a profound effect on the course
377). In any case, there was close correspondence both of the monsoon tracts. We have already mentioned
faunistic as well as cultural between the southern Asiatic Schuster's evidence for a lower temperature during the
mainland and Java during the Middle Pleistocene. The Trinil Stage, as well as the existing probabilities of cor-
Sunda Shelf must have had its maximum extension as relating this with one of the glaciations in the Hima-
a land-bridge during the Second Glaciation, which to layas (see p. 451). For the moment, however, it is
all appearances was the most effective in Eurasia. If not possible to decide what may have caused the grass-
the Fourth Glaciation caused a lowering of the ocean land avifauna to expand in the manner indicated by
level in the Sunda region by 72 m. (? 40 fathoms), as Stresemann. Nevertheless the question helps focus at-
Molengraff and Weber (1921) have pointed out, then tention on the necessity of investigating the amplitude
we may safely assume that during the Second Glaci- of climatic changes in the tropical belt of Ice Age Asia.
ation the drop was much greater. How much greater From the geological records, it appears that the tran-
is difficult to say, but the differences between the snow- sition from the Middle to the Upper Pleistocene was a
lines during the Second and Fourth Glacial Stages critical period in the early migration across Sundaland.
(1500 m. and 900 m.) in the Himalayan region make The combination of a lowering of the sea-level during
it seem probable that it may have been at least as much the Third Glaciation and the uplifting of eastern and
as 125 m. (? 70 fathoms, see Fig. 106). While this central Java suggests optimum conditions for land-
does not greatly alter the countours of the Sunda Shelf, making processes and faunal migrations. Indeed the
as reconstructed by Molengraff and Weber (1921, Fig. neanderthaloid race of Solo Man
probably dispersed at
1) for the last glaciation, it makes it virtually certain that time, since its remains are associated with the more
that during the earlier glaciation this shelf must have modern
Ngandong fauna. Its stratigraphic position in
stood as much as 100 m. or more above sea-level. No the oldest terrace
deposits at Ngandong (see p. 455)
doubt it was covered by jungle and dotted with marshy corroborates our contention that this fauna
appeared at
plains; in any case it must have been an attractive the beginning of the Upper Pleistocene. Of this group
habitat for the large ruminants with whom the early
namadicus and the Indian elephant sug-
human races were associated. These animals may Hippopotamus
gest the type of association one frequently encounters
have found it easier to move about in the alluvial tracts
when dealing with the bearers of Palaeolithic cultures
of this newly emerged land, traces of which have been
in Eurasia. Again the focus of dispersal can only be
detected by the accurate soundings of the Snellius Ex-
at, but it was probably not in Java in as much
pedition (Brouwer, 1926). The main drainage chan- guessed as this island is situated in too peripheral a position in
nels were apparently directed radially from a hilly up-
land around the islands of Billiton and Banka, with one the Sunda region. One feels that the present data
master stream flowing northward into a bay of the point toward Indo-China and South China.
China Sea, and the other master stream running south- As regards modern man, we shall base our discussion
eastward to an estuary due north of Madoera. It can of migration routes on the assumption that Wadjak
be readily seen how such drainage lines must have been Man represents a race ancestral in some way to the liv-
most primitive of modern peoples.
highly conducive to animal migrations to eastern Java ing Australoids,
and southeastern Borneo, since they afforded easy ac- This fossil Homo sapiens was discovered in 1888 at
cess to the easternmost terminus of the Sunda region, Wadjak in south central Java (Fig. 100) in what
at that time demarcated the Straits of Ma- seems to have been a cave or a fissure deposit. Du-
already by
cassar and Bali. bois (1920) described the two skulls, and he claimed
Another effect of the Ice Age on migration routes has that Wadjak Man might be an ancestral type to the
race. Pinkley (1936) on the other
lately been touched on by Stresemann ( 1939, pp. 412 ff.) living Australian
in his discussion of the distribution of such grassland hand denied this because of the "advanced character"
6 According to von Koenigswald (1939, p. 49), the University
of the Wadjak anatomy, which is characterized by a
of Manila's collection contains an as yet undescribed fauna con- general robustness of the skull and a slender body
sisting of Stegodon, Elcphas and Rhinoceros from Luzon. build: Recently Weidenreich (1939, p. 111) has sug-
462 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
gested a clear line of ascendence from Pithecanthropus mann's (1939, p. 414) recent hypothesis pertaining to
to Solo and Wadjak Man to the living Australoids. the introduction of fowl by prehistoric man in the
Unfortunately almost nothing is known of the geo- Celebes and adjoining regions merits our attention.
logical age of Wadjak Man. In any case he cannot be To judge from the present distribution of the jungle
much older than the last Glacial Period, since it is very fowl (Gallus gallus gallus), it appears that this sub-
unlikely that he was a contemporary of the neander- species is found by itself in the eastern Sunda Islands
thaloid people of Ngandong, who inhabited the fringe and in the Philippines, whereas in Java it is associated
of Sundaland during the Third Interglacial. Von with Gallus gallus bankiva. Stresemann considers
Koenigswald even went so far as to suggest that that this latter form was ecologically isolated from the
Wadjak Man might be contemporaneous with his post- other region as a result of having migrated to Java via
Pleistocene "Sampoeng fauna," an as yet largely un- Sumatra prior to the peopling of this area by modern
explored group from Sampoeng Cave in central Java man, who introduced the jungle fowl. In other words,
(Fig. 100). Mijsberg (1932) described skeletal re- the jungle fowl (Gallus gallus gallus) was introduced
mains with australoid affinities from this locality, and in the domesticated form via the Philippines and Su-
these appear to be associated with the Sampoeng fauna. matra to the Celebes and the Lesser Sunda Islands,
But, since he ascribed the manufacture of polished stone where in a few instances it reverted to wild life. While
axes and barbed arrowheads to this prehistoric race, such a method of dispersal does not necessarily imply a
we cannot see how this evidence throws much light on land connection across the Philippines, it seems to indi-
the problem of the dating of Wadjak Man, whom we cate that prehistoric races migrated this way to the
regard as an older type. Von Koenigswald, however, easternmost Sunda region. We suggest that it was the
considers it possible that his age does not exceed 5,000 Neolithic people who introduced the jungle fowl, be-
years, an assumption not supported by the ancient char- cause the distribution of polished celts, as shown by
acter of the Sampoeng fauna. In this assemblage we Heine-Geldern (1932), is almost identical with that of
have three mammals: Cervus eldi, Bubalus sp., Elephas these birds.
sp., which are now extinct on the island. The question
of whether or not Neolithic implements were really IX. SUMMARY OF PREHISTORIC
found associated with this fauna remains to be investi- MIGRATIONS TO JAVA
gated.
Taylor (1937, pp. 73 and 99) has discussed the mi- In conclusion let us summarize the salient points
gration routes of Early Man in the Sunda region with which emerge from the previous discussion regarding
reference to the first peopling of the Australian conti- the migrations across Sundaland (see Chart 2, p. 455,
nent. He envisioned a Negrito migration from Asia and Fig. 106):
across Sundaland during an "early glacial period" with- 1. During Upper Pliocene times Java emerged from
out stating what geological age he would assign to it. the sea and was linked to the southern Asiatic mainland.
He also assumed that a second migration of Australoids The newly formed peninsula became the habitat of
occurred "during an Ice Age when the Sunda and mammals which migrated from the southern portion of
Sahul areas were dry land." In our opinion this could Asia. The remains of this fauna occur in the fresh-
only have been at the time of the Fourth Glaciation, for water beds of western and central Java and indicate
Taylor proceeds to demonstrate how the primitive Tas- affinities with the Upper Siwalik fauna of India and
manians were obliged to migrate farther eastward to Burma. While this would suggest a migration route
Australia during the period when an enlarged Sahul via Malaya (Route 1 on Fig. 106), it does not exclude
Shelf still existed. Later on these aboriginals became the possibility of a more northern derivation, as indi-
stranded in northern Australia owing to the submer- cated by the geographical range of the Siwalik fauna
gence of Sahul-land "in a late Interglacial age." Again over large portions of what is now South China.
we interpret this to mean that the final separation of the 2. The Lower Pleistocene witnessed the continua-
Australian shelf from Sundaland was an Early Post- tion of free faunistic interchange with the mainland of
glacial event, rather than one referable to the Third Asia, and the island was populated by man (H.
Interglacial, as implied by Taylor's remark. Assum- modjokertensis), possibly a representative of the Pithe-
ing that Taylor's conception of migrations, based on canthropus race. In his company appeared other an-
the zonal distribution of races, is correct, one might say thropoids, including orang and gibbon. Large por-
that the Negrito and Australoid races invaded the tions of northern and eastern Java were submerged
Sunda region during the last major lowering of the under the sea at this time; volcanism became an impor-
ocean level, and that they occupied their recent habitat tant factor in sedimentation.
during Early Postglacial times. 3. In the Early Middle Pleistocene period, Sunda-
As to the later migration of peoples to Oceania, we land experienced its maximum emergence partly in
are entirely dependent on archaeological or ethnological consequence of a major maximum glaciation (Second
evidence, and these cannot be discussed without depart- Glacial Stage) and the lowering of the ocean level and
ing from the main thesis of this paper. But Strese- partly dtle to uplift. This made possible an influx of
DE TERRA: PLEISTOCENE GEOLOGY AND EARLY MAN IN JAVA 463
new faunistic elements from Malaya and South China DE TERRA, H. 1940. Geologic Dating of Human Evolution in
Asia. The Scientific Monthly, LI (1940) 112-124.
(Routes 1 and 2, Figs. 106), and probably led to a DE TERRA, H. 1941. Pleistocene Formations and Early Man in
wider dispersal of Pith ecanthroputs. The main part China. Institut de Geo-Biologie, Peking, No. 6 (1941)
of the Middle Pleistocene beds in Java, however, were 1-54.
laid down during the succeeding stage of high ocean DIETRICH, W. 0. 1924. Zur Altersbestimmung der Pithecan-
level (Second Interglacial), at the close of which thropus Schichten. Sitzungsber. Ges. Naturforsch. Freunde
Berlin. Jahrg. 1924, pp. 134-139.
strong mountain-making movements caused an uplift-
DUBOIS,E. 1892. Voorloopig bericht omtrent het oderzoek
ing of the Kendeng Hills. Volcanism was very active naar de pleistocene en tertiaire Vettebratenfauna van Su-
all over the island. matra en Java gedurende het jaar 1890. Natuurk. Tijd-
4. Following this phase of uplift, new drainage lines schr. voor Nederl.-Indie, LI (1892) 93-100.
were established in Early Upper Pleistocene times. A . 1894. Pithecanthropus crectus, eine menschenahnliche
Ubergangsform aus Java. Batavia, 1894. Also in: Jaar-
new type of Man-Solo Man-appeared with strong boek van het Mijnwezen, XXIV (1895) 5-77.
neanderthaloid affinities. His migration may be con- 1896. On Pithecanthropuls erectus: A Transitional Form
nected with the emergence of Sundaland during the between Man and the Apes. Sci. Trans. Roy. Dublin Soc.,
Third Glacial Stage. Stream terraces and alluvial VI (1896) 1-18.
. 1899. Remarks on the Brain-Cast of Pithecanthropus
formations contain records of prehistoric cultures. erectus. Jour. of Anat. & Physiol., XXXIII (1899) 273-
5. During the Late Upper Pleistocene, stream ag- 276.
gradation led to a mass burial of faunal remains in the 1908. Das geologische Alter der Kendengschichten
Solo Valley, possibly as ancient as the Third Inter- oder Trinilfauna. Tijdschr. Kon. Nederl. Aardr. Genoot-
schap, XXV (1908) 1235-1270.
glacial. The following phase of low ocean level .1920. De Proto-Australische fossiele Mensch van
(Fourth Glacial Stage) witnessed the appearance of Wadjak, Java. Proc. Kon. Akad. van Wetensch., Amster-
modern man (Honmo sapiens wadjakensis). Also the dam, XXIX (1920) 88-105; 866-887.
first migration of Negrito peoples may have occurred .1932. The Distinct Organization of PithecanthZropus of
Which the Femur Bears Evidence, Now Confirmed from
at this time, as well as a subsequent dispersal of Man Other Individuals of the Described Species. Proc. Kon.
to Australia. Akad. van Wetensch., Amsterdam, XXXV (1932) 716-722.
6. In the Postglacial Period no maximum lowering .1934. New Evidence of the Distinct Organization of
of ocean level has been recorded, although temporary Pithecanthropus. Proc. Kon. Akad. van Wetensch., Am-
sterdam, XXXVII (1934) 139-145.
land connections may have existed via Sumatra. Neo- 1935. The Sixth (Fifth New) Femur of Pithecan-
lithic peoples introduced the jungle fowl via the north- thropus erectus. Proc. Kon. Akad. van Wetensch., Amster-
ern route (Route 2 on Fig. 106) to the Celebes and the dam, XXXVIII (1935) 850-852.
Lesser Sunda Islands. DUYFJES, J. 1936. Zur Geologie und Stratigraphie des Ken-
denggebietes zwischen Trinil und Soerabaya (Java). De
Ingenieur in Nederlandsch-Indie. IV. De Mijningenieur,
BIBLIOGRAPHY OF PART V III (1936) 136-149.
.1938. Geological Description of Sheet 110 (Mod-
BLANCKENHORN, MI. 1911. Allgemeine Betrachtungeniiber die jokerto) : Geologische Kaart van Java, 1/100,000. Batavia,
wissenschaftlichen Ergebnisse der Selenka-Trinil Expedi- 1938.
tion. Report in Selenka and Blanckenhorn,1911. ELBERT, J. 1907. Uber das Alter der Kendenggeschichten mit
BROUWER, H. A. 1928. History of the Strand Line in the Pithecanthropus erectus. Neues Jahrb. f. Min., Geol., etc.,
Netherlands East Indies during Pleistocene and Post- Beil. XXV (1907) 648-662.
Pleistocene Times. Proceedings of the Third Pan-Pacific 1908. De nieuwste onderzoekingen over het Pithecan-
Science Congress, Tokyo, II (1926) 1807. thropusvraagstuk. Natuurk. Tijdschr. voor Nederl.-Indie,
CARTHAUS, E. 1911. Zur Geologie von Java. Report in Se- LXVII (1908) 125-142.
lenka and Blanckenhorn,1911. 1911. Die Selenka'sche Trinil-Expedition und ihr Werk.
COLBERT, E. H. 1935. Siwalik Mammals in the American Centralbl. f. Min., Jahrg. 1911, pp. 736-741.
Museum of Natural History. Trans. Amer. Philos. Soc., HEINE-GELDERN, R. 1936. Prehistoric Research in Indonesia.
XXVI (1935) 1-401. Ann. Bibliog. Ind. Arch., Leiden, IX (1936) 26-38.
1940. Pleistocene Mammals from the Ma Kai Valley of HRDLICKA,A. 1930. The Skeletal Remains of Early Man.
Northern Yunnan, China. Amer. Mus. Novitates, No. 1099 Smithsonian Misc. Coll., LXXXIII (1930) 1-379.
(1940) 1-10. 'JANENSCH, W. 1911. Die Proboscidier-Schadel der Trinil-
COSIJN, J. 1931. Voorloopige Mededeeling omtrent het voor- Expeditions-Sammlung. Report in Selenka and Blanken-
komen van fossile beenderenin het heuvelterreinten Noor- horn, 1911.
den van Djetis en Perning. Verhand. Geol. Mijnbouw JUNGHUHN, F. 1852-1854. Java. 2 vols., Leipzig, 1852; 1854.
Genootsch.Nederl. e. Kol., Geol. Serie, IX (1931) 113-119. LEHMANN, H. 1936. Morphologische Studien auf Java.
DALY,R. A. 1934. The Changing World of the Ice Age. Geogr. Abhandl. Dritte Reihe. Heft IX. Stuttgart, 1936.
New Haven, 1934. MIJSBERG, W. A. 1932. Recherches sur les restes humains
DE TERRA,H. 1937. The Siwaliks of India and Early Man. trouves dans les fouilles de l'abri sous-roche du Guwa Lawa
Article in "Early Man" (edited by G. G. MacCurdy). a Sampung et des sites prehistoriques a Bodjonegoro
Philadelphia, 1937, pp. 257-268. (Java). Hommage Serv. Archeol. Indes Neerl. Congr.
DE TERRA, H., and T. T. PATERSON. 1939. Studies on the Prehist. Extreme-Orient, Hanoi, Jan. 1932, pp. 39-54.
Ice Age in India and Associated Human Cultures. Car- MOLENGRAAFF,G. A. F., and M. \VEBER. 1921. On the Rela-
negie Institution of Washington, Pub. No. 493 (1939) 1- tion between the Pleistocene Glacial Period and the Origin
354. of the Sunda Sea, and its Influence on the Distribution of
464 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY
Coral Reefs and on the Land and Freshwater Fauna. Proc. . 1935. Geological Description of Sheet 58 (Boemiajoe):
Kon. Akad. van Wetensch., Amsterdam, XXIII (1921) Geologische Kaart van Java, 1/100,000, Batavia, 1935.
395-439. VAN BEMMELEN, R. W. 1934. Geological Description of
OOSTINGH, C. H. 1935. Die Mollusken des Pliozians von Sheet 36 (Bandoeng) : Geologische Kaart van Java, 1/100,-
Boemiajoe (Java). Wetensch. Mededeel. Dienst v. d. 000, Batavia, 1934.
Mijnbouw in Nederl.-Indie, No. 26 (1935) 1-247. 1937. Examples of Gravitational Tectogenesis from
OPPENOORTH, W. F. F. 1932. Holmo (Javanthroplus) soloensis, Central Java. De Ingenieur in Nederlandsch-Indie. IV.
een pleistoceene mensch van Java. Wetensch. Mededeel. De Mijningenieur, IV (1937) 55-65.
Dienst v. d. Mijnbouw in Nederl.-Indie, No. 20 (1932) VAN DER MAAREL, F. H. 1932. Contribution to the Knowledge
49-74. of the Fossil Mammalian Fauna of Java. Wetensch. Me-
.1936. Een prehistorisch Cultuur-Centrum langs de Solo dedeel. Dienst v. d. Mijnbouw in Nederl.-Indie, No. 15
River. Tijdschr. Kon. Nederl. Aardr. Genootschap, LIII (1932) 1-208.
(1936) 399-411. VAN Es, L. J. C. 1929. Excursion Guide E. 5: Trinil. Fourtl
. 1937. The Place of Honmo soloensis among Fossil Men. Pacific Science Congress, Java, 1929.
Article in "Early Man" (edited by G. G. MacCurdy). . 1931. The Age of Pithecanthropus. The Hague, 1931.
Philadelphia, 1937, pp. 349-360. VAN STEENIS, C. G. G. D. 1935. On the Origin of the Malay-
PENCK, A. 1882. Schwankungen des Meerespiegels. Jahrb. sian Mountain Flora. Bull. Jard. Botan. Buitenzorg, XIII
d. Geogr. Ges. Miinchen, VII (1882) 1-70.
(1935) 135-263; 289-417.
.1933. Eustatische Bewegungen des Meerespiegels wah- VAN STEIN-CAILENFELS, P. 1933. L'industrie osseuse (ie
rend der Eiszeit. Geog. Zeit., XXXIX (1933) 329-339.
Ngandong. L'Anthropologie, XLVI (1933) 359-362
PINKLEY, G. 1936. The Significance of Wadjak Man, a Fossil VON KOENIGSWALD, G. H. R. 1933. Beitrag zur Kenntnis der
Homo sapiens from Java. Peking Nat. Hist. Bull., X
fossilen Wirbeltiere Javas. Wetensch. Mededeel. Dienst
(1936) 183-200. v. d. Mijnbouw in Nederl.-Indie, No. 23 (1933) 1-185.
RENSCH, B. 1936. Die Geschichte des Sundabogens: eine 1934. Zur Stratigraphie des javanischen Pleistocan. De
tiergeographsiche Untersuchung. Berlin, 1936.
Ingenieur in Nederlandsch-Indie. IV. De Mijningenieur, I
SCHUSTER, J. 1909. Ein Beitrag zur Pithecanthropus Frage.
Sitz. Ber. K. Bayer. Akad. Wiss., Math.-Phy. KI., Minchen, (1934) 185-201.
-. 1935. Die fossile Saugetierfauna Javas. Proc. Kon.
Abh. XVII (1909) 1-30.
1911. Die Flora der Trinil-Schichten. Report in Se- Akad. van Wetensch., Amsterdam, XXXVIII (1935) 188-
198.
lenka and Blankenhorn, 1911.
1936. Early Palaeolithic Stone Implements from Java.
.1911-a. Monographie der fossilen Flora der Pithecan-
Bull. Raffles Museum, Singapore. I (1936) 52-60.
thropus-Schichten. K. Bayer. Akad. Wiss., Math.-Phys.
. 1936-a. Uber Altpalaeolithische Artefakte von Java.
Kl., Mfinchen, Abh. 17, XXV (1911) 1-70.
SELENKA, L., and M. BLANCKENHORN.1911. Die Pithecan-
Tijdschr. Kon. Nederl. Aardr. Genootschap, LIII (1936)
41-44.
thropus-Schichten auf Java. Geologische und Palaeontolo-
. 1936-b. Ein fossiler Hominide aus dem Altpleistozan
gische Ergebnisse der Trinil-Expedition (1907-1908).
von Java. De Ingenieur in Nederlandsch-Indie. IV. De
Leipzig, 1911. This includes the following reports:
CARTHAUS,E. Zur Geologie von Java. Mijningenieur, III (1936) 149-157.
. 1937. A Review of the Stratigraphy of Java and Its
DozY, C. Bemerkungen zur Stratigraphie der Sedimente in
der Triniler Gegend. Relations to Early Man. Article in "Early Man" (edited
MARTIN-ICKE,H. Die fossilen Gastropoden. by G. G. MacCurdy). Philadelphia, 1937, pp. 23-32.
MARTIN, K. Notizen iiber die Sfisswasserbivalven aus den . 1937-a. Ein Unterkieferfragment des Pithecalnthropus
Pithecanthrlopus-Schichten von Trinil. aus den Trinilschichten Mitteljavas. Proc. Kon. Akad. van
FELIX, J. Die fossilen Anthozoen aus der Umgegend von Wetensch., Amsterdam, XL (1937) 883-893.
Trinil. . 1938. Ein neuer Pithlecanthropiis-Schadel. Proc. Kon.
STREMME,H. Die Saugethiere mit Ausnahme der Probos- Akad. van Wetensch., Amsterdam, XLI (1938) 3-10.
cidier. . 1938-a. Nieuw Pithecanithropis vondsten uit Midden-
JANENSCH, W. Die Proboscidier-Schadel der Trinil-Expedi- Java. Natuurkundig Tijschr. voor Nederl.-Indie, XCVIII
tions-Sammlung. (1938) 195-207.
CARTHAUS, E. Spuren von moglicherweise menschlicher 1939. Das Pleistocin Javas. Quartar, II (1939) 28-53.
Tatigkeit in den Trinil-Schichten. VON KOENIGSWALD, G. H. R., and F. WEIDENREICH.1938.
SCHUSTER, J. Die Flora der Trinil-Schichten. Discovery of an Additional Pithlecanthropus Skull. Nature,
BLANCKENHORN, M. Allgemeine Betrachtungen fiber die 142 (1938) 715..
wissenschaftl. Ergebnisse der Selenka-Trinil Expedition. . 1939. The Relationship between Pithecanthroplus and
STEHLIN, G. 1925. Fossile Siugetiere aus der Gegend von Silnanthropus. Nature, 144 (1939) 926-929.
Simbangan (Java). Wetensch. Mededeel. Dienst v. d. WEIDENREICH, F. 1938. Tatsachen und Probleme der Mensch-
Mijnbouw in Nederl.-Indie, No. 3 (1925) 1-12. heitsentwicklung. Bio-Morphosis, I (1938) 5-29.
STREMME, H. 1911. Die Siugetiere mit Ausnahme der Pro- .1939. The Classification of Fossil Hominids and their
boscidier. Report in Selenka and Blanckenhorn, 1911. Relations to Each Other with Special Reference to Sinaun-
STRESEMANN, E. 1939. Die V6gel von Celebes. Jour. ffir throplus pekinensis. C. R. Cong. Int. des Sciences Anth.
Ornithologie,LXXXVII (1939) 299-425. et Eth. Deuxieme Session, Copenhague, 1938, pp. 107-112.
TAYLOR,G. 1937. Environment, Race, and Migration. Chi- (Published in 1939).
cago, 1937. .1940. Man or Ape? Natural History, XLV (1940)
TER HAAR, C. 1934. Homio soloensis. De Ingenieur in 32-37.
Nederlandsch-Indii. IV. De Mijningenieur, I (1934) 51- .1940-a. Some Problems Dealing with Ancient Man.
57. Amer. Anthropologist, XLII (1940) 375-383.
PLATE XXXIV

FIG. 1. The site of Trinil on the Solo River, central Java.


FIG. 2. The main terrace at Ngandong, showing the excavated area.
FIG. 3. Cliff section at Sangiran. In the foreground, black clay containing the Djetis fauna (Poetjang Beds). Above, cross-bedded sands and
fauna (Kaboeh Beds).
FIG. 4. The site of the discovery of Pithecanthropus skull II at Sangiran. Right, the Tjemoro River. Left, Middle Pleistocene Kaboeh Bed
block in which the skull was found.
PLATE XXXV

FIG. 1. The Baksoka Valley east of Poenoeng, near Patjitan, south central Java, showing terrace deposits and karst topography.
FIG. 2. The bed of the Baksoka River near Poenoeng. The gravel underlying the 30-foot terrace (T2) is exposed on the bank of the river.
gravel-strewn surfaces as the one shown in the foreground, the Palaeolithic implements are found.
FIG. 3. The uppermost part of the cliff section at Sangiran, showing the disconformity between layers 4 and 5. Stone implements of Uppe
upper gravel layer (5).
FIG. 4. The site of the discovery of HIomomodjokertensis. Dr. von Koenigswald is sitting on the edge of the small pit where the skull was fou

You might also like