A

HANDBOOK
OF

ANTIQUE PROSE -RHYTHM

A.

:% DE OROOT

j.

8,

HE HAGUE

ftf*

THE UNIVERSITY
OF ILLINOIS
LIBRARY

460
G83h
V,
/

\v

CMss^S

Return this book on or before the Latest Date stamped below.

University of Illinois Library

tfi

APR

p,

200B

w 25

m
1181

DEC 1 1
SEP 3

1515

N0V2Q1S&
DU: Mfrffi

OCT
'APR

19

L161 H41

A HANDBOOK
OF

ANTIQUE PROSE-RHYTHM

BY

DR A. W. DE GROOT
CONSERVATOR OF THE LIBRARY OF

THE UNIVERSITY OF GRONINGEN

HISTORY OF GREEK PROSE-METRE

DEMOSTHENES, PLATO, PHILO, PLUTARCH AND OTHERS

BIBLIOGRAPHY, CURVES, INDEX

J.

B.

WOLTERS - GRONINGEN, THE HAGUE.

1919

^0

PREFACE.
This work
tries

to

apply and to propagate a

It

new

method
and
will

of investigation of the clausula.

must therefore
method,

be considered as a simple introduction

is

to this

based upon much pioneer work, which, as such,

naturally

be incomplete, and open

to

criticism.

have

tried to confine myself to facts only

I

and

to give

as

few hypotheses as

could, as

am

fully

aware

that the

few hypotheses found here,

fications.

will require

important modi-

Of one thing
with this method,
one,
at

am

firmly convinced, viz. that only

either with

that is

mine or a similar

any

rate

with a method of comparison, reliable

I

results can

be obtained.

am

also convinced that Zielinski

quite
fest',

wrongly looks upon

and
that,
if

his statistical basis as 'felsenis

the science of antique prose-rhythm

this

to

lay

claim

to

name,

it

will

have

to

change the

usual

method
bring

of investigation altogether.
this

To
of
\y

clearly

to

light,

have often been

obliged to discuss
brevity
their
I

the opinions of others.

so,

For the sake

have only done

when
if

could not agree

with

theories
this,

and opinions:

had not confined

purpose

myself to

my work

would have
I

failed in its

of being a brief introduction.

hope

that the reader will

U
cp

IV
understand that
it

is

not

my

wish or intention
character.
I

to give to

my work
the

somewhat polemic

By choosing
more
attractive.

the form of lectures

is

have

tried to

make
little

matter which

naturrlly not very readable, a

One

of the first things in future will

be

to investigate
text.

an absolutely non-metrical and non-rhythmical

As

wanted such a
tence-metre
as such.
of

text only for a

comparison with the senI

Demosthenes,

have taken Thucydides

Further

take

the

liberty

of

pointing out that the

words
col.

of

Mr.

Ammon
I

(Berl. Philol.

Wochenschrift 1918,

495)

viz. that

should have considered Procopius of

author,
;

Caesarea as

metrical

must be ascribed

to a

natural misunderstanding

of his valuable

remarks as well
I

as

of

those

of

professor

D.

V.

Hesseling in Leyden

shall

be glad

to avail
I

myself later on.

of Clark's

Of course
valuable

have made a frequent use

of

Fontes prosae numerosae,

le style

Laurand's

clear

E'tude sur

des discours de Ciceron, and of

many

other works.

This part will be followed by a second for the Latin

prose

and the
will

later

accent-rhythm, in which the biblio-

graphy
one.
I

be completed with an extensive systematical

it

shall also discuss in

the

modern
of a

theories of

Zander

and others from the standpoint

comparative method.

V
In conclusion
that
I

want
their

to

express
to
it

my

gratitude to those
First of all to
for the
I

have offered
J.

services
if

me.

professor

van Wageningen:

had not been

in

constant interest which he

showed
in

my

work,

could

hardly have finished

it;

his

remarks both on the form and

it

the contents have influenced

father
in
J.

many

respects.

To my

whose

invaluable assistance has facilitated

my task

no small degree.

To
I

professor G. Heymans, Dr. P.

van Rhijn, and Dr. F. Zernike, by whose publications

oral information

and

have improved
Mr.
J.

my work
who

consideassisted

rably.

To my

friend

G. Holthuis,

me

in constructing the curves.

And

last

not least to Mr.

in

G. Dudok,

who

offered his time

and valuable services

supervising the translation in a most unselfish way.

Digitized by the Internet Archive

in

2013

http://archive.org/details/handbookofantiqu01groo

CONSTENTS. *)
FIRST LECTURE.
Methodological problems.

There

is

a definable prose-rhythm

different authors

show

different distribution of metrical forms; the synkri-

seis of Plutarch

show

the

same

peculiarities as the Lives

themselves

6;

Demosthenes

avoids

www, wwww,
of

^^^-w^;
by choice

a metrical effect can only be produced either

of

words or by arrangement
out the

words

Thumb

points

necessity of comparing clausula14.

rhythm with sentence-rhythm

SECOND LECTURE.
Methodological problems ; sentence -metre of Demosthenes.
It

is

necessary

to investigate

longer series of syllables

18;

Thucydides

in entirely or

almost entirely careless of

the arrangement of long and short syllables in the sentence

as

appears

from a comparison with other authors and

of increasing

from theoretical calculations 20; tendency

avoidance
of

www, ^^w,
is

etc.

in

Demosthenes, and

Reference

made

to the pages.

VIII

other

tendencies;
is

Norden's

scheme

for

the

clausula of

Demosthenes

wrong

33;

how

to

determine the length

36.

of the clausula in

Demosthenes and other authors

THIRD LECTURE.
Methodological problems; clausula of Plutarch
Philo Judaeus.

and

Mathematical problems 40.

avoidance of
,

Tendency
,

of increasing

etc.

in Plutarch 42;
(

Plutarch has only three 'clausulae':

^w^

w w,

43;

tendency

in

favour of forms in the whole

sentence 49; tendency in favour of forms in the clausula

only 49; importance for textual criticism 50; length of the

clausula of Plutarch 25; clausula of Philo 54.

FOURTH LECTURE.
Plato

Clausula of Plato 59
the

importance of the fourth paeon

sj/llable

quantity

of

the

final

of the clausula is not

indifferent

62; Bornecque on the negative clausula 65;

is

the clausula heroica of Cicero

66;

not an hexameter-ending

evolution of the clausula of Plato 68; clausula of the

IX
five chronological
its

groups

chronology of the Phaedrus 69

metre has influenced

current

statistics of

words 75; therefore

determine
79.

the

method

of

using the

statistics to
is

the chronology of these dialogues

quite

wrong

FIFTH LECTURE.
Chariton, Libanins, Herodes, Lesbonax, typology.

Consequences

of

the

application

of

non-comparing
for Libanius

methods by Heibges
83;

for Charito,

by Heitmann
in the
;

typology

is
;

probably absent

Greek

clausula
of

{Novotny)

90

Zander's

hypothesis 94
96.

quintessence

prose-rhythm and of the clausula

SIXTH LECTURE.
History of the method of research ; first stage of the clausula.

Merits
Kroll,

of

Litzica,

Jordan,

Maas,

Thumb, Novotny,

Laurand 98; value

105;

of antique theory 102; value of

statistics

origin of the clausula 107; three stages in

1

the evolution of the Greek clausula,

Hellenistic
stage,
3;

classical stage, 2

rhythmical

stage
of

109;

theory of

impoverishment 110; general features

Greek sentence-

metre

112.

X
SEVENTH LECTURE.
Development of the clausula; Wilamowitz.

Typology

preserves

the
is

falling

rhythm

119;

the

quantity of the final syllable

lutions

not indifferent 121; reso-

and by-forms
1

123;
;

relation

between Greek and

Latin prose-metre,
of

25

in Latin

prose

we

note two groups

2.

metrical

authors:
1

1.

Sallust
of

and Livy,

Cicero and
typical

the

others

25

clausula

Sallust
of

and Livy a

Latin

clausula
of

127;

clausula

Cicero

infuenced

by

Hegesias
the
in

Magnesia or others 128;

stage

characteristics of

Hellenistic

130; incorrect view of Wilamowitz

'Asianismus und Atticismus' and 'Kultur der Gegenwart'.

EIGHTH LECTURE.
Last stage ; rhythm and the Greek language.

Origin of the Greek rhythmical clausula

rences

32 the
;

diffe-

between the Greek and the Latin accent-clausula

correspond with those between the Greek and the Latin

quantitative clausula
1

33

the

Greek rhythmical clausula

1
;

is

an imitation of the Greek quantative clausula

of

34 influence

prose-rhythm upon order of words 138; upon choice of

141
;

words

importance

of

prose-rhythm

for

textual

criticism 143.

XI

NINTH LECTURE.
Statistical

and mathematical problems

14.7.

TABLES.
, ,

Frequency

of

etc. in different etc.

authors

167, 172, ditto of of

of

w, www,
of

172, 176; clausula

Thrasymachus
183;

182;

the

more
184;

rhetoric
of

parts

of

Thucydides

Thucydides

Demosthenes;

sentence-metre of Demosthenes as compared with that of

Thucydides 186; clausula

criteria

of Isocrates 188; of Plato 190;

for
of

the

chronological order of Plato's dialogues;

194;
of

clausula

Plutarch

Philo

196; generel tables

196197; typology

of the clausula 197.

BIBLIOGRAPHY,

p. 200.

CURVES.
Thucydides' clausula and sentence-metre 219; Demosthenes 220; Plato 221; Philo 222; Plutarch 223; Livy 224;

Cicero 226.

INDEX.

A.

W. DE GROOT, A

handbook

of antique prose-rhythm,

I.

CORRIGENDA.
Page

IV, line 16, D. C. Hesseling

instead of

D. V. Hesseling.

XI,
80, 80,

15, 13,

Thucydides

Thycydides.
0-00
o/
.

4-9%

102,

21, p. 54 and 59 18, 'Quatenus Cicero ipse

p.

00 and

00.

numerum suum

oratorium

in

certam formulam redigere

.
. .

potuerit', instead of:

'Quomodo

potuerit'.
results.

123,

56, assertion results instead of assertion,

24,
4,
x

124,
127,

avoided
exists

sought.
exist.

135,
136,

137, 162,

passim p. 00. ), line 24, and as an hexameter ending: instead of and. instead of caesura. 2, diaeresis

note

23,
24,
10,
9,

65%
50
0/

50%.
65
0/
.

162, 163,
166,

psychical

pscyhical.

read

reed.

192,

27,

6q$ws
line
1,

defrwg
instead of

196197

(table),

TABLE OF CLAUSULAE,

TABLE OF CLAUSULA.

196197 (table), line 34 sqq.:

(10)

Polemo.
of:

instead
(10)

~ *
-

--^
-

Polemo.

- ^

v_^_.
;=:.

197, 197,

line

23,and
12,

8,

- ^
24:

instead of

w -^^.
metrique
134,
etc.

^
should be
130, 134.

201,

23

La

prose
130,

omitted.

228,

Wilamowitz

103,

instead of

ANTIQUE PROSE-RHYTHM
FIRST LECTURE.
Is

there indeed
will

such a thing as a sharply definable

first

prose-rhythm?

be the

question that arises with

most
of the

of

you,

when

discussing this subject.

Are we not
units

same opinion as those who found themselves

in

able to use the concept rhythm

prose" because of
occurrence of
to
It

vagueness?

Are we not

to ascribe the

all

kinds of so-called curious and typical

phemomena
first of all.

chance?
ought
to

This question must be settled

be answered with a positive no.

investigation
of

systematic

statistic

shows

us,

that

passage from the works

Demosthenes has another

distribution of metrical

forms

than any passage from the Lives of Plutarch.

this,
I

To prove

took from different writers pieces each consisting

of

1000 syllables.

These

syllables

are of course either

long or short, and only in a very few cases

quite sure

we
Out

can't

be

whether

the}/ are

long or short.

Here we have
of

consequently a perfectly objective

datum.

each

group

of

000 syllables
is

we

take every two short syllables,

syllable

between which there

no third short

and see

how many

long ones there are between.

There may be

long syllables between,

e.g. in juera

ww
w w
^[
xcbv tzglqovicov

or or
or

Jiagovotv

2,

yvvcux&v naQOvxwv
yvvcuxwv

3,

'[

etc.

We now
ones.

write

down how
1

often in every 1000 syl-

lables there are

long,
first
life

long, 2

long between 2 short

For the

1000 syllables from Thucydides and

of

those from the

Pyrrhus of Plutarch

found the

following figures.

Thucydides
180

Plutarch

237
144

113

57

60

40
18

40
23

20
2

4
1

Certain

relations
is

are

noticeable here, for

we

see at

once, that there

an intimate agreement between the

two columns:
presented.

in
it

both the form

occurs

ww

is

most strongly
is

re-

So

more

frequently that there

no

long

syllable

between two consecutive short

is

syllables

than that there

one, or than that there are

two between

two short ones.

When
tain

comparing the two columns,

we

notice a cer-

difference.

w w
The
all

w w
this to

The forms ^^, w ^, ^ occur more frequently in Plutarch,

w, and
the form

occurs as frequently in both, whereas Thucy-

dides oftener uses

w and w
now be
put are
:

w.

question might

we not to ascribe
connection with

chance?

Will not the following 1000 syllables

give quite different figures?

Will not,

in

what you pointed out above,

obtained?
syllables of
180,
It

just the opposite results

be

is

unimaginable that the following

will yield again for

1000

Thucydides
exactly
240,

ww

exactly

for

w w

113,

for

w
70,

w
then

exactly 57.

Suppose we had
the

120

and

our previous

statement could not be maintained.

in

We

doubt whether
is

grouping of long and short syllables there

any
very

difference

between the

two

authors.

We

doubt

much whether you have

sal to

sufficient material at

your dispo-

prove your case.

this
I

To

give

you the following answer.

is

Indeed the material of 1000 syllables

for our purpose.

not sufficient

Therefore

my

investigations are not based

on these 1000
I

syllables only.
I

Of

practically

all

the authors

investigated,

did not take one group, but twelve groups

of

1000

syllables.

The

question
in the

is

now: do

the figures

of all these

groups point

same

direction?
it

From

more extensive

investigation

appears very

clearly that this

is

indeed the case.

As an example

take the form

w w.
of
this

The frequency
Thucydides
is

form

in the

twelve pieces of

the following:

113 70 89 98 113 107 93 105 99 91 109 85.

In Plutarch:

144 143 152 135 142 134 148 141

126 135 138 127.

From
of

this

we

infer that all these figures in the

in Plutarch.

works
Thu-

Thucydides are lower than those

In

cydides they vary between

70 and 113.
In Plutarch,

on the other hand, between

126 and 152.

The

highest

figure

in

Thucydides

is

considerably

lower than the lowest

I

in Plutarch.

hope

that

by means

of these figures
is

have con-

vinced you somewhat of the fact that there

of Part

no question
de grouper 1000

de grouper
chiffres.
is

les chiffres, but of la science

per les

The average
Thucydides

figure for

w w

syllables

for

97.7, for Plutarch 138.8.

Are
form
This

we
in
is

to ascribe this difference to a preference for that

Plutarch or to an avoidance of

it

in

Thucydides ?

a question

we

shall

have

to

answer again and again

in the

course of our investigations.

is

Facts prove that this

difference
tarch.

owing

to a preference for that

form

in

Plu-

This appears from a comparison with other authors.

In the
crates,

same manner

also

investigated

Xenophon,

Iso-

Demosthenes, Plato's Republic, Plato's Laws, and

the Consolatio ad Apollonium, which has got amidst the

Moralia of Plutarch.

As an average

figure for

w w on

1000 syllables they give:

Xenophon
Isocrates

97-2.

852.
111-5. 105-0.
80-5.

Demosthenes
Plato, Republic
Plato,

Laws

Consolatio ad Ap.

102-0.
97-7.

Thucydides
Plutarch

138-8.

Probably you
another
conclusive

will

be convinced now.

But

we have
form an

proof.

Not long
of

ago

Stiefenhofer

convinced

us

that

the

ovyxQioeig

Plutarch

essential part of his biographies.

He showed
life

us that the

language

of

these

ovyxQioeis,

in

which Plutarch makes

of a

a final comparison between the

of a

Greek and

that

Roman,
This

quite corresponds with the language of Plu-

tarch.

in itself is

an argument deserving our careful

attention.

But secondly he showed us that they form an

essential part of his biographies, a part

of composition cannot
If

which

for reasons

be dispensed with.

the ovyxQioeig are indeed

the

by

Plutarch,

they have

to

show

same

peculiarities as the Lives themselves.

this

Is this the

case?

To prove

we

give

the

following

figures for the frequency of

ww

on 1000 syllables:

w_w

131

119 134 143 132 155 134 139.

Average
It

137.1.

seems

to

me
The

that the proof could not

have been

more
of

conclusive.

figures entirely confirm the results

the

previous

investigation.

Attention

should also be

paid to the other forms.

Another example.

In the

same manner
in

as

we have
Whereas,

investigated the series of long syllables

Greek prose,
never been
last,

we
as

can investigate the series of short syllables.

far

as
of,

know, the

first

possibility has

thought

Marbe already pointed

out

the

which
investi-

Thumb
gations

applied on a small scale.

In other

words

may be made how many

i.

short syllables there are

between two long ones,

e.

the frequency of the forms

,

w
We
wing

ww www
,

etc.

may be

determined.
follo-

then get for Thucydides and Demosthenes the

statistics

for

the

first

1000 syllables of the Pelop-

onnesian
first

war,

and

for

the

first

1000

syllables

of

the

Philippic.

Thucydides
302
139

Demosthenes
308
128 129
7

66
26
13

5 6 7
The
that

3
1

greatest difference

between the two columns

or
dactyl

is,

the

choriamb

ww

ww

is

more

frequent in Demosthenes, but that, on the other hand, the

forms
Will

www wwww
,

etc.

rarely

occur there.

this

be confirmed by an extensive investigation?

this
I

To prove
66
51

shall first give the figures for the

choriamb.

ww
71

Thucydides
69 80 63 63
57

72

62

73

80.

Demosthenes
129

93

112

132

126

95 126

101

106

103

109

81.

About

this there

cannot be the least doubt.

Now

the

longer series.

Thucydides.

Demosthenes.
3
7

26
26
21

13
11

13 2 3 10
5

8
6
2

28
22

3 4
7
1

2 2

23

14

5 13

3
3 2
1

29
35

9 8 8
12

13

30
20
31

6
12
7

3
1

10
7

24
315

12

3
19

108

35

106

Average.

Thucydides

133-9

67-3
109-4

26-3
8-8

9-0
1-6

2-9
0-7

Demosthenes 147*9
In a similar

manner, but with much smaller material

Thumb
he
proof of

already pointed out that Blass

that

was

right

when
This
it

stated

Demosthenes avoids the

is

tribrach.

Thumb

very important.
it

Munscher thinks

very unimportant, because

had already been discovered

before.
statistics,

He

forgets,

however, that Blass published no

statement might just as well have
of

so that his

been

incorrect.

For so much
to

what Blass has

later

said on

rhythm has turned out

and

be incorrect

on.

Marbe
which

Thumb were
This
first

the

first

to point out to us in

way
draw
cretic

a scientific proof could be given.

of
all.

But secondly

we

are

now

able to

the following

conclusions
at all

from our material: the

w
same
of

has not

been sought by Demosthenes

or choriamb
like to

to the

extent as the dactyl

ww
I

Recent scholars such as Norden

preference
ancients

speak about the

Demosthenes

for

the cretic.

Already the

make mention
sa}7
it

of this.
is

But

this

assertion

need hardly

not based on
in

statistics.

And

to

anticipate our further

investigations:

the clausulae
in

as

such,

the

cretic

does not play any part

Demosthenes.

It

will

be understood that these are not the only

results of
this:

our investigation.

Of great importance
This difference

is

also

between the metre

there
fact
is

of the

Republic of Plato and his

is

Laws
to

a great difference.

owing

the

that Plato in his later 3/ears seeks for series

of short syllables.

Already Blass suggested

(

this,

but he only
).

speaks of a preference for the tribrach

www

This

one-sidedness can be accounted for: Blass did not

yet the method of arriving at a

know

more

certain result.

10

For Plato we get the following

Republic.

figures

Laws.

Average.

w WW
4
-5 6

Average
112-8 47-5 40-3
13-3
5-3

146-2

64-0
30-7

90
2-7
1-2

2-8
0-5

7
-8

00
0-2

00
0-3

9
There are two curves
In the

0-0

here,

which cross one another.

and

Republic the forms

in the

ww

occur

more frequently than

forms
it

Laws, whereas with the other

mainly the tribrach which

is just

the reverse.
it

We
is

see

that

is

not
the

preferred.

Though
in

forms
don't

^^^>w
occur
so
often

and
of

wwwww
quent in
the

themselves

course, they are, comparatively speaking

Laws
I

than the tribrach.

much more freThe preference

of a

shown

for such a

form can be expressed by means

quotient.

What

want

to

say

is

this:

suppose there are

30 forms in the Republic and 60

occur twice as

in the

Laws, then they

is

much in

the

Laws

the quotient

60 30
:

= 2.

11

The
is

quotient for the forms

higher than that for

wwww and w w w ww www We find for

.

Republic
(average).

Laws
(average).

Quotient.
1-3 1-5

wwwwwww -wwwww www www

You
will

30-7

40-3
13-3

90
2-7
1-2

53
2-8

2-0
2-3

now

understand

why

Plato in his later

works

so often uses expressions such as:

xaxd xb dvvaxov, juexQineQ,
xafidiiEQ,

EJiavacpEocov,

ajzoho/uevwv,

etc.

You
tQonov

will

also
?

understand

why

he so often says
xivd

xivd

w^w^
:

and not
can

xqotiov

w ww.

For

such
choice

metrical

effect

only

be produced

either by

of words
of

the
i.e.

author

may make
instead

use of xa&djieQ
of

instead
fie%Qi7iEQ
etc.,

coojzeq,

www
i.e.

w,

or

of

instead

of ecooneo,

www
:

instead of

w w,
use of

or by arrangement of words
xQonov instead of xqotzov
^S^'.
xivd,

he

may make

xtvd

i.e.

wwww
we
fix

instead of

*W

The importance
later

of

all

this,

to

shall refer to

it

on

should not be undervalued.

Of

late elaborate

investigations

have been made

by means

of this

choice of words the chronological order of Plato's dialogues.

12
It

has

been found

that

Plato

used

in

his

later

works

xaddjiEQ instead of

ojojieq,

/u%QuieQ

instead of ewotieq.

To

us

this

becomes
than
I

quite

clear

now.

There

is

no easier

means

to get a particular

combination of long and short

syllables

by

substituting one

synonym
that also

for another.
in

And when
sula
of

tell

you now already

is

the

clau-

Plato

there

a gradual development, you will

the chronology
of

agree with
the metre

me that for may become

Plato's

works

of the highest importance.

also

spoke about Plutarch and

This
is

his preference for

the form
his

w ^.

not the only metrical peculiarity

that Plutarch

prose

shows.

avoids series of long syllables.

The most important is The forms

-,

etc.

occur here less

vestigated

frequently

than in

all

other writers

in-

by

us.

That textual

criticism

be

it

only

on

moderate scale
you.

might

avail

itself of this

need not
in

tell

That

this

same tendency does not occur

all this

Pseudo-Plutarch's Consolatio ad Apollonium will not surprise you.

clearly

And
Typical

perhaps

may
is

lead to prove

more
of the

and

effectively
for
II

the

genuineness of
e.g.

many

Moralia.

Demosthenes
ed. Weil):

the following

sentence (Olynth.

[5],

13 To
/uev

ovv ejiloqxov xal dmoxov xaXsTv avev xov xd Ttejigayjueva

elvai
xig

detxvvvai,
3

XoidoQiav
TiojTiox*

av

(pijoeie
3

xevtjv dtxaicog

'

xd de

jidvft

boa

engage

diEg~iovxa,

lop

anaoi xovxoig eXey^eiv,

xal

xal

f$Qa%Eog

Xoyov

ov/upaivEi

delo&ai,
exeZvov,

dvoTv &%' tfyov/uai

xal dXrjfisg vndqiEi,

ovjLKpeoeiv

etofjo&ai,

xov

r'

ojieq

cpavXov cpaivEodm, xal xovg vjiQX7ZJiXr]y/uvovg (bg a/jLa%6v xiva

xov <PiXmjiov ISeiv

oxi

ndvxa

Sie^eXtjXvv^ev

olg
ijxst

jiqoxeqov
xr\v

naoa-

xoovofXEvog
jxody/uax'

fxkyag rjv^'d'i], xal jzgdg avxrjv

I

xeXevxtjv xd

avxcp, i.e.

Typical
(Life of

for

Plutarch
I):

are

e.g.

the following

words

Philopoimen
fjv

KXiavSgog

ev MavxtvEia ykvovg xe ttqojxov xal dvvijd-Elg

xolg judXioxa xa>v noXix(bv, i.e.

From
Mexu

Plato

we may

quote (Laws IX, 856b):

xfjg

dk xd jieqI $Eovg xd tieqI xaxaXvoiv

noXixdag,

i.e.

or (858#)

rsXoiav, cb

g~V,

Txooxx&EjUE^a xi]v atQEOiv, xal dxEyytbg

14
ojojieQ

xaxe%ofjivoig

vojuo&haig
rjdrj

o/jloiol

yiyvoifieff
i.e.
'.

av

vno

/bteyd^rjg xivog avdyxrjg

vojuolhxeiv etc.,

It

was

my
of

intention to give

you an idea

of the imtrust-

portance of these metrical investigations and of the

worthiness
himself

their

results.
first

with this for the

The man who occupied time, was the psychologist

applied this method

Marbe, on whose suggestion

to

Thumb

Greek prose.
Also

On Thumb's work we

can

in

general

make
sqq.)

the following remarks.

in his revision of

Brugmann's Grammatik

(p.

666

he reproaches the
not

modern
all

clausulae-investigators

with

having

first

of

investigated the

immanent
opinion,

rhythm
and

of the

Greek language.
I

He

is

even

of

in this respect

quite agree with him, that the present

investigation

of

the

Greek

quantitative

clausula

has in

general
I

not yet led to definite results.

To

his

remarks

add the following.

(1)

The avoidance
is

of

^w^

www^
differ

etc.
f

by Demosthenes

not

sufficiently

appreciated

by him.

He

says

that the figures indicating the frequency of the

in Plato

different

forms
in

and Xenophon do not

it

much

from those

Demosthenes, and that

lies in

the nature

15
of

ordinary speech that the rhythmical form

ww^>

rarely occurs.
(2)

In his opinion about the clausula of

Demosthenes
so far one-

he wrongly agrees with Norden.

(3)

His investigation of the rhythm

it

is

in

sided that

takes the series of short syllables into account,

but not the series of long syllables.

also this
e.g.

We

have seen that

may
the

be of importance as a criterion of genuineness,

Plutarchian

for

and pseudo-Plutarchian works.

not taken the series of long sylis

The reason why he has

lables
that of
into account, is

easy to understand: his method

The method-Marbe is intended for modern languages, where seldom two or more accents follow one another. For in these modern languages we count the number of unaccented syllables which lie between two
Marbe.
accented ones, and in this

v~

way we

often

arrive

at

the

most interesting

results, as

has been pointed out by Lipsky.

Why

do

in

English two accented syllables succeed one

another less frequently than two long syllables in Greek

or in Latin ?

The

principal reason
into account

is,

because the accents

in

which have
Latin
of

to

be taken

do not depend, as

and Greek on an objective

syllables
really
is,

factor, as the quantity

the

but

on

more

subjective

factor,

sentence-rhythm,
different.

or emphasis, or stress.

This

is

something quite
the

According
syllable

to

Marbe's scansion
accentuated or

same word,

the

same

may be

16
not,

as

sentence-rhythm requires.
is

In

Latin,

however, a

syllable

either long or short.

The

one-sidedness, there-

fore, of the

method
it

of

Thumb

can be explained by going

back

to its origin,

being invented for a Germanic, not

for a classical language.

(4)

He

did
for

not

show

us

how
with
think

to

compare
those
shall

his
clau-

own
to

results

sentence-rhythm

of

sula-investigations.

Therefore
his

we

we
it

be able

improve upon
he

method by making
attempt
be
is

more complete.
the help

Why
his

failed in this
it

obvious

With

of

method

would

an

impossibility to

compare them.
of
of

We

gladly acknowledge
the

Thumb's great merit

correctly,

having
having

formulated
solved
it.

problem

but

not

SECOND LECTURE.
In

my

last lecture

spoke

to

you about a preference

which
in

for series of short or long syllables

in different writers.

Greek occur
such
as

Some
Plato,

prefer

combinations
avoid

^w^^^^
etc.

etc.; others distinctly

e.g.

Some, as

show

a clear development in
of

their prose metre, a

development

which

have only

given

you a

single instance: peculiar usage of

in
this

words

in

his latter

works has
may,

way found an unexpected

this question
is this all?

explanation.

We
Is this

in the

meantime, put

preference only based on series of short and long

syllables, or is there

more behind

it?

Has Demosthenes
form

^ ^ w w ,orfor ^^ ^^ ^^ above above ^ ^^ 'w etc.? When we know that w

possibly

shown a preference
*

for the

is

sought by Demosthenes,
the combination
*-*

we do
is

not

know

yet whether

w ^ ^ as much sought as ^ w We know that Demosthenes has also sought the form ^^ was then quite immaterial him whether he used series of dactyls or choriambs, w^ ^^ ^^ or ^^ ^^ ^^ So
just
.

it

to

we
only:

shall

have

to

investigate

more.

The

question

is

how?

18

For these investigations

years ago, which proved to
I

made use
to

of a

method some

me

be the correct one.

investigated each combination of a fixed

8.

number
I

of syl-

lables, e.g. 6, 7, or

and added

the

On practical grounds took 7 reason why will be explained later on

left

an eighth, the quantity of which was

out of

account.
e.g.

We
etc.

then get 2

128 possible series.

1

These are

short,

6 short and

long, 5 short

and two long

syllables,

We

then get the following scheme:

^______^
w^wwww~ _^_^^^^^

^^w^~
w_-w--.w>~'w>-i,
etc.

11

126
127

w
investigate

128

~
how
often these combinati-

We

can

now

ons occur in the different writers.

19

You

will believe

me when

say that

in Plato's later

dialogues the forms

^^^- _~ w_wwwww;~:
wwwww^-r,
in his older
etc.

are

more frequent than

works.

We

cut,

so

to say, a text into pieces of 8 syllables, and register these

combinations.
It

will

be observed that the frequency

varies the
in

of the different

combinations

the

different

writers.

In

the

same manner
be
In

last

8 syllables of each sentence

also

may

investigated.

Here

differences

will

be found.

the

same manner

the middle of the sentence

may be
will

compared with the end.

be noted here.

Probably also a difference

The

clausulae of one writer

may be comwill

pared with those of another.

Also here differences

be found.
So,
is

in

Plutarch, the clausula exttyovoiv, or

w ^,
Now,

much

more
or

frequent

than
:

in

Thucydides.

another important question arises has Thucydides avoided

this form,

was
sight

it

used with undeniable preference by

Plutarch?

At

first

one

would be

inclined

to
is

answer:
not the

Of course Plutarch has sought

this form.

For

20
double trochee one of the most favourite clausulae, which
are frequently found in Greek as well as in Latin prose?
I

cannot,

however, admit the strength

contrary,

of this arguthat,

ment.

On

the

we have
i.e.

to
it

acknowledge
will

without the help of some other data,

to decide

be impossible
Plutarch
or

which prose

the

prose

of

that

of

Thucydides

we must

consider as the prose of

ordinary writing, and which, on the other hand, as metrically

artificial.
It

might appear
paid

later

on that Thucydides
to

as
of

well

as

Plutarch
short

attention
in

the arrangement
prose,

long and
in

syllables

their

the

latter,

however,

a higher degree.
all.

This ought, therefore,

to settle:

to

be

settled first of

We
is

have

what may we
frequency
of

consider as natural and artless writing?

of

What

the

different

forms

to

be found

in

Greek prose

ordinary speech?
I

venture to say that

it

is

Thucydides who appears

of the

to

be entirely or almost entirely careless

of long

arrangement

and short syllables

in the sentence,

and

for

the

following reasons.
I

think,

we
of

are
there

justified
is

to

suppose that

in artless

natural

writing

no system

whatsoever
This

in
is to

the

arrangement

long and short syllables.

style.

be

our ideal natural

take
it

Let

us, for the

sake of argument,
syllables are

for granted that in

Greek long

much
:

more frequent than

short ones in a proportion of 2

1.

21

^,

This proportion of
the
true

frequency of
2
,

Ions: syllables
,

frequency ot short syllables

it

=
If

f is not

one,

but

may

serve for example.

in

this

be

the case, a form

ww will
You
will

general occur
that

more

frequently

than

w^w.
to
j

see

the latter will

as
the

show

tendency

occur \ time as

much

former.

The
or

form
as

ww ww as
,

however,

will occur as often as

w w,
of
x,

they consist of the

same number

long

and

of short syllables.
If

we

call the

frequency of the form

to 2 x.

^ww
it

that

of

of

^^ ^
as

will
is

come very near

The frequency
;

greater than that of

latter.

^^
of

is

twice
will

as

great

the

The frequency
2

w
of
I

be

x,

that of

2 x.

Now,

we may
etc.

calculate
in

the

frequency

^^^,
syllable

^^-, ^ ^,
I

Thucydides.

As
of

said before,

have mostly neglected the quantity

each form;
if

the last

of

we

investigate the last four S3dlables of

viz.

each sentence, only eight forms can be found,

22
If

we now

could point out that the mutual frequency

of

each of these eight combinations was only determined

of long

by the respective number

which they are composed
for special metrical

and short

syllables of

there would remain no

room
this.

tendencies.
of

We

shall

calculate

Now,

the

ratio

the

frequency of long syllables

compared with
577

that of the short ones in

in

Greek seems

to

be nearly

Plato

and nearly the

this

same

in

Thuof

cydides. We shall call ^^-^ we shall call x.

ratio

p.

The frequency

The

frequency, then, of

____ = __._ = w = = w_ = = w

px px
p x px
p2X
p2 X
2

(==

423

-2 x)

The sum
This
If

total

is

x
of

total

amount

3 p x = 13*10 x. the cases investigated by us is 2000.

3 px

3 p 2x

in fact these

2000 forms are distributed

to call

in

what we
fre-

are

accustomed
of

a merely accidental way, the

quency
that of

^^^ must be nearly ^^ nearly px = 10-01

= t^t^
etc.

100

% = 7.63 %,

%,

23

As
those,

a matter of fact these numbers tally exactly with

in reality

which were

found by

us.

comparison

of the figures follows here.

Theoretical
calculation.

Clausula-

Sentencemetre.
5-8o/o
0/

metre.
8-5o/
8-8,

7-63
10-01

o/
o/

7-9,10-4

10-7,

9-3, 11-3o/

14-19%
19-38<y

13-6,14-2,18-5%
18-3o/o
of

14-5,1 4-4,13-90/

20-3
really

o/

The

correspondence

these

numbers

is

extraordinary.

The form
last

w represents the frequency of w

i.e.

:=:

at the close of the sentence,

the
to

double trochee, the

syllable

of

which

is

assumed

be anceps.
frequency

Accorshould
first

ding to our theoretical

calculation
It

its

have been 14*19 per cent.

1000
14-4

was, however, for the

14*0
14-2
for

clausulae

of

Thucydides
figure
is

%.

The average
was
29*6

%, %.
the

for

the

second

With
first

Plutarch,

on the other hand, the frequency

clausulae
29-1
o/
.

thousand

%,

for the

second 28*6 %, average

think, the

problem has been solved now.

Thucydides
it.

does not avoid the double trochee, nor does he seek

difference as
is

A
a

shown above between 14-2% and 14-19%to

must only be ascribed

chance.

But in

this

case

24
difference of 14-2
/

and

29-1
I

%
will

is

draw an

inference from, as

more tnan enough to show later on. Clearly

this form.

does Plutarch show a preference for

We
are
artless
st3/le,

are,

therefore,
to

now

justified to

suppose that

we
of

allowed

consider

Thucydides as an example
i.e.

and natural ordinary prose,

not with regard to his

his sentence construction, his syntax, but with

regard
short

to

his

metre,

i.e.

the

arrangement
for

of

long and

syllables.

Perhaps the difference

the form

^
of the

(14-19

clausula
to a to

% 18-5 ww

%), for which the high frequency

5=1

in

Thucydides
example.
If

is

responsible, points

divergence from
conclusion

this

afterwards

we come
is

the

by our

investigations

that this

his

single favourite clausula, this result should not astonish us,

as

prose rhythm in

its

first

stage

shows an increasing
difference as

number

of favourite forms, not

Slight,

an ^impoverishment", as has
is this

been assumed.

however,

com-

pared with the difference which

will

be found

when we

compare Thucydides' metre with that of more artificial writers.

In

order to
I

show more
parts,
I

clearly

how

far

our figures

are reliable,
into

divided the investigated text of Thucydides

two equal
should

each comprising 1000 cases.

calculated
to

For

each of these parts

form.
I

the

frequency of each
attention
of

like

draw your
exceeds

especially

to the fourth column, in of figures

which the difference

It

each couple

has been given.

0-8%

hardly four times!

25
THUCYDIDES' CLAUSULA.
I

II

III
B.

IV
Difference of the

Clausula form.

Group A.
Percentages.

Group

Average.

Percentages.

groups

and B.

^^

www

0-8

0-2

0-50
14-20
2-00 5-20
1-25

0-6

w w

14-0
2-4 5-0
1-1

14-4
1-6

0-4
0-8 0-4
0-3
2-5
1-2

w_^^^
^^-ww

5-4
1-4

^^^ ^^

^^^^ ^>w ^^

4-7
2-7
9-1

2-2
1-5

345
2-10 9-35

^ -WW

w w

w w w

9-6 2-4
3.5

0-5

2-7

2-55 3-70

0-3 0-4 0-6

0-5
1-3

3-9
2-8
2-1

ww

3-4 2-6

3-10

235
6-25
7-85

^^
_ww_w

5-6
8-1

6-9
7-6

0-5 2-9
0-5

19-7

16-8
6-3

18-25

WW
that,

5-8

6-05

Objections have been made, more particularly by Miinscher

without
as

sufficient

ground,
of

took Thucydides as a

norm,
his

an example
I

natural

prose with regard to

difficulties.
I

metre.

hope

have removed these

But Miinscher made other objections, because

wrongly

26
neglected the difference between rhetoric and non-rhetoric
parts.
of

This

is

merely theory. Munscher refers

this

to a

work

Rollmann.

From

work

derive the following figures.

Periclis laudatio funebris, clausulae creticae et choriam-

bicae 76*6

%;

material investigated 207 cases.

(lib.
I,

Periclis oratio

cap. 140

144), the

same forms
manner
in

72-3%; material
Leaving

investigated 217 cases.

alone

the

altogether
et

arbitrary

which the clausulae

creticae

choriambicae

have been

added together;

for this

group comprises among others:

clausulae creticae:

clausulae choriambicae

leaving

alone

the

arbitrary

manner

in

which everything
clausulae
for

has been cut to pieces and

made

to

the

sake of his theory,

e.g.

the sentence on page 31

27
xr\v

xe

ya.Q

Jioliv xoivrjv TiaQexojuev

xal ovx eoxiv ore J-evrjXaoiag andQyofxev xiva

fi

juafirjjuaxog

rj

dedjuaxog
jur)

xQvepdh av

tig

xa>v TioXe/uicov idcov

(bcpeXrjdeir),

Ttiotevovteg ov xaXg

naQaoxevaig xo nXeov

xal andxaig
tj

xq>

ay

f/jucbv

avxcbv eg xaqy

evxpvxco.

There
of such

are, rightly understood,

9 clausulae!
in getting

In spite

means he has not succeeded

(on 207 cases), and

more than

the following differences:

76*6% 72*3% 59-3%

IV, 1-5).
If

(on 217 cases) in rhetoric passages, whereas (on

143 cases) in non-rhetoric passages

(lib.

we now

calculate (in

what manner we

shall explain

later on) the

probable error of these percentages, then we find

for 76*6
for 72-3 for 59.3
It

%
/

a probable error of
a probable error of a probable error of

2-0,
2-1,
.

2*3.
is

should be understood that this

not the probably

it

largest error, but the probable error.

Hf..f.

Would
1-2,

not

seem

79-^

that these ratios

=^,

or 1*3 and =^0, or

in addition

28
lo

the

curious grouping and in addition to the arbitrary

interpunction prove just the reverse of

In
figures

what
a

is

aimed at?

conclusion
to

would draw
slight

our attention to some

show what

differences

comparison

with Xenophon's sentence-metre gives.

A
Xenophon
sentence-

B
Thucydides
sentence-

c
Thucydides
clausula

Differences between

metres

metres

A&B
0-3
1-6
0-1

A& C
0-3
1-4

&C
o-o

0-8

o/

0-5

0-5

12-8
1-7

14-4
1-8

14-2
2-0
5-2
1-3

0-2 0-2
1-6

0-3
1-2
0-1

4-0
1-4
3-1

3-6
1-1

0-4

0-3
0-4 0-9

0-2

2-7

3-5
2-1

0-4
0-1

0-8 0-8
0-6

2-0
7-8

2-9

8-8

9-4
2-6
3-7
3-1

10

1-6

2-6
5-1

30
5-1

04
0-0
0-1

o-o
1-4

0-4
1-4

2-4

2-5
1-3

0-7
1-4

0-6
1-1

10
9-1

2-4
6-3
7-9

0-3

9-4
6-5

0-3
0-9

2-8 2-3
7-5

3-1

5-6

1-4
2-1

25-8

20-4
2-6

18-3
6-1

54

37

2-4

3-5

29

The
in

differences

that

occur are so slight that

we

in-

voluntarily

pay

attention to the

form

^ only, which,
figure,
in

Xenophon, shows a somewhat higher

o/
,

namely
and not

25-8
Is

over against

204%
is

and

18-3%

Thucydides.

there indeed any question

of a real difference

of

difference

that

the consequence of working with

settled

limited material?

These questions can only be

by

more extensive
parts
of

researches.

The very
of,

slight difference for

the other forms, differences

for the greater part, only

percentage, exclude the presence of metrical

tendencies altogether.

Quite different

is

the case

of

the

clever prose of Plato and Philo Judaeus.

will

not annoy you any further by giving figures

and
of

statistics.

The

question, however, of a certain basis

to

investigation
I

that

could not
I

gations.

will

much importance refrain from making more minute investinow give you a summary of the results
appeared
of so of

me

yielded

by a comparison

the metre of

Demosthenes
is

with that of Thucydides.

terised
(1)

Demosthenian prose

charac-

by

the following tendencies:

all

Preference for dactyls and choriambs in

:

com-

binations

(hg a/uaxov

uva rov Qikmnov

Ideiv.

(2)

Tendency

of reiterations.
is

Reiteration of dactyls,
reiteration of

choriambs, and cretics

trochees
is

sought

iambs and

avoided.

30
(3)

Tendency
or

of

combinations dactyls and choriambs

:

tend to succeed one another in any combination;

e.g.

of

ww w w,
cretics is
(4)

w^ ^w ^ w avoided,
e.g.

any combination
,

or

ww
as

^.

Tendency

of increasing avoidance; series of long

six long or three

and short sjjlables containing more than

short ones,
of

are avoided
increases
:

more strongly
elQijvijg

the

number
nlovoioi
dk

syllables

ovoy\q

ix

nxwywv
/nerd

yeyovaoiv,
jceqi

w ww w|
tieqi

xd

deovg xd

xaxdlvoiv

xfjg nolixeiag,

wwwwwww
is

think

you

will

agree that our method

to

be

preferred to that of

Marbe and Thumb.

The

latter

does

not distinguish between iambic-trochaic metre on the one

hand, and cretic metre on the other.

texts
it

For the following

would give the same

figures:

ww w w w v w w w w ^
viz.:

(eighteen syllables),

(eighteen syllables),

frequency

5,

in either of in either of

them,

^
When we
of our

frequency
try,

6,

them.

however,
1

to analyse

them by means
get the follow-

method

of the

28

possibilities,

we
|

ing distribution:

^ ^ ^ w _w_w
|

86, 118, etc,

for the former;

and

for the latter:

__

^_

110. 92, etc.

31

The

cretic

rhythm

is

preserved now, and

is

it

will

be

understood that the same

true for dactyls

and choriambs.

By means
been able

of

Marbe's method

we

should not have

to state that peculiar preference of

Demosthenes

for a reiteration of cretics, as for instance in:

jiQcbrov jukv ovv

ovx

aftv/urjieov,

which

certainly

exists,

nor to show that he avoids the

reiteration of

iambs and trochees.

Investigators of

modern prose-rhythm have more than

once pointed out the importance of such researches. They

often try to determine the average
syllables
that,

number

of unaccented states

between two accented ones. Gropp already

texts,

with reference to the following

the

the

method

of

Marbe would give

I

same

results.

_r_

_r

j__

jr_

Quite the same holds good for

II

___

__

__

___
of in-

Nowhere, however, has the simple procedure

vestigating longer series of syllables been suggested.

The tendency

of increasing avoidance,

which

men-

32

tioned above,
logical

will
of

be quite clear
view.
If

to

you from a psychotries

point
series
of

writer

to

avoid

long

long

syllables,

and
to

can

hardly

be
to

avoided

the

same degree.
increases,
In
Plato,
this
e.g.

The

latter

will

seem
the

him
of

to

be even worse than

the former.

As

number
all

syllables

tendency

will

become

the

stronger.

we

are able to state that in the

?

Laws

the preference for

the

^^^ wwww
may be sign we

etc.

as

compared with

Republic

expressed by the following quotients.

(By a

shall indicate that there is a preference for a certain

it

form, or at any rate that

occurs more frequently there

than in the text compared

sign indicates the reverse.)

Frequency
in the

Frequency
in the

Republic
syllables,

Laws
syllables.

on 1000

on 1000

Average.

Average.
40-3
13-3
5-3 2-8

Quotient.
1-3 1-5

_^^ ^^^^ ^^^^^ ^^^^^^

You
sentence.

30-7
9-0
2-7
1-2

(+)
(+)

2-0
2-3

(+) (+)

see that in this

way

it

is

possible to

compare

the metres of the sentence with those of the end of the

If,

however,

in

either case

we

should get the

33

same percentage,
the

the

existence of any special

metre

at
i.e.

end

of

the

sentence would be highly doubtful,

clausula.

there

would be no metrical

Could

it

be demondiffers

strated,

however, that the clausula-metre greatly

of the

from that
to

whole sentence, the clausula would prove

of

be from a metrical point

view an independent, or
But so great
is

rather a separate part of the sentence.

the

power

of a priori

argumentation and of tradition that

of

the problems the solution

often entirely neglected in

which

is

attempted here are

modern

literature,

because their
only rarely

existence and the necessity of their solution

felt

is

and acknowledged.

As
is

to

Demosthenian metre
are made.

(too often the

term rhythm

used

in this connection) the

most curious and doubtful

Antike Kunstprosa Norden
of

assertions

In

his

gives

the

following

scheme

favourite

clausulae

for

Demosthenes:

34
If

this

scheme be
through

correct,

higher percentages must

necessarily be found for these

forms

at

the

end

of

the

sentence

than

the

whole

sentence.

Thumb,

Munscher, and
it

many

others agree with him

by accepting

without any proof.

We

are able to compare the figures.

Demosthenes.

Frequency Frequency
in the

at the

end
Quotient

sentence
o/

of the sentence

8-0

8-0

90
4-1

11-2
4-0 4-3 4-8
5-8
3-7
3-1

(+)
(-) (-)

5-9

4-6

(+)

4-1

7-2

(+) (-)

-w

1-7

(+)

12-4

18-9
18-7

(++)
(-)

190

The form
The
is

^
dicretic

occurs as often in the sentence.

less

'famous'

form occurs even

frequently

in the clausula.

For

ww

w hardly half the number

found there.

The double spondee, which was formerly

considered as metrically identical with the double trochee

(

and

w w)

reaches 19-0% in the sentence,

only 18*7%

in the clausula.

'AUd

del,

as Aristoteles says

35
in his Rhetorics, SrjXrjv elvm xr\v xsXevxrjv,
ju)]dk
jutj

Sid xov ygacpca,

did

xr\v

TiaQayQCKprjv,

alia

did

xov qv&jllov.

Only one
clausula.
?

or two forms, however, give a really favourite

They

are the

double trochee:
?

txteyovoiv,

w
1

and

^^
trochee
clausula.
I

ww

e.g.
1

wvtEQ

[tiov

dv&Qcbmvov.

The double
8*9
it

reaches

24

in

the

sentence,

in

the

draw your

attention to the curve;

gives the

frequency of the forms

5-8
912
1316 1720 2124 2528 2932
For the sake
1,

www w_ww ww ww w_ w
w
have been indicated by
as follows:

of brevity they

5,

9,

13,

17, 21, 25, 29.

The
(5)

results

may be summed up
of

Tendency
is

dependence.

The

clausula of De-

mosthenes
metre.
(6)

almost entirely dependent on his sentence-

Tendency

of

independence.

In

the clausula as

such only a few forms are preferred.

The

question remains to be answered, a question of

36
the highest importance which
is

therefore often discussed:

is

exMyovoiv

ww

to

be considered as the real length

of

the clausula?

Does

the length of the clausula here com-

prise only four syllables?

To answer this question we ought figures we should expect to find, if this

the case.
that

to find out

what

should really be

Let

it

be assumed for the sake of argument

at

Demosthenes does not prefer any metrical form

end
of

the

the

sentence,

except

w w.
form
is

Let

it

be

assumed
quotient
is

that this
2.

preference can be expressed by the

of

The frequency
-

the

in the sentence

really

12-4/

our supposition

correct,
If,

we

shall

find

24*8

%
is

at the
2,

end

of the sentence.

however, the

quotient
the form

this

will hold

good

for all subdivisions of

w w. We

may

find e.g.:

Sentencemetres.

Clausula-

metres.

Quotient.

Sum
will

total

ww- w_w _w_w_w w w_w w w w_w

we might
call

/,

2 (+)
2

(+)
(+)

4
5

8
10

2 (+)
2

12%

24

2 (+)

In

other words:

whatsoever precedes, the quotient

always be the same.

This an
'ideal'

distribution of metrical

37
forms.

Unnecessary

to

say that such a distribution will

Often

be found hardly anywhere.

to
-f-

we have

to

be content

find for all subdivisions of the

sign,

form a quotient with a

or for

all

of

them a quotient with a

of

sign.

Only when our material comprises thousands

investigated,
it

cases

is

really

surprising to see
'ideal'

how

often the

percentages

approach the

figures.

Let us

now

assume the reverse: the quotients

divisions vary altogether, e.g. in this
>entence-

for the different sub-

way:
Quotier

Clausula-

metres.

metres.
0-2

1-0%

5(-)
5 (-)
2 2

20 40
6-0

0-4

8-0

(+) (+)

12-0

What would
First of all
it

it

prove?
that the

would prove

forms

ww w
it

w w w, which, as you see, constitute the form w w w!, are not preferred, but avoided. Further wouldprovethattheforms(w)w w wand(w) w
and
do not
say:
really belong together.
is

On

the contrary,

we must
not four

www

avoided,

w^
www
it

is

preferred.
is

This means that the length of the clausula

syllables:

w w,
if

but five:

For the quantity

is

of the syllable
it

or w w. which precedes w w,
gives a bad clausula;

not indifferent:

is

short,

38
if
it

is

long,

it

gives a

good

clausula.
is

Whenever
indifferent,

it it

appears
does not

that the quantity of this syllable

belong any more to the

'clausula' in its technical sense.

We may
syllable

state:

the

length

of the clausula ends with the

of indifferent quantity.
is

To

find

out this indiffe-

rent syllable
of

theoretically very simple.

Where
the

a clausula
of the

n syllables forms a metrically separated part

all

sentence,

subdivisions of
the

it

will

show

same
is

quotient.

Of

course,

smaller the material that

investigated,

the less frequently the figures approach their 'ideal' value,

and the more one has

to

be content with an unbroken

signs.

row

of

signs, or of

felt
:

obliged to

make

this digression to

answer the

question
it

how
is

long

is

the clausula of

Demosthenes? Does

really

comprise only four syllables:

indeed the case.

w w?
if

This

Even

not merely one,

but two preceding syllables are taken into account, rather

regular proportions are found:

Demosthenes.
Sentencemetres.
Clausula-

metres.
2-6
4-6

Quotient,
1-0
2-1

2-5%
2.2 3-2
4-5

(+) (+) (+) (+)

4-9 6-8

1-5 1-5

39
If

we

take only the preceding syllable into account,

the agreement

becomes even more

Demosthenes.
Sentenceetres.

striking:

Clausula-

metres.
7-7

Quotient.
1.53 1-52

4-7
7-7

(+)

We

(+) have now discussed the most important and most

11-7

elementary methodological questions.

We
us.

have seen what

a preliminary investigation of Greek prose-metre not say prose-rhythm

do

has taught

Only one methoand then

dological question remains to be answered,

shall pass

we

on

to the applications of our results.

THIRD LECTURE.
In

my

last

lecture
'

discussed

new method
concluded

of

investigating

antique
of

prose-metre.
gives

We
a

that

the

method

Marbe

only

very bad idea of

in antique prose.

the metrical tendencies

which may occur

We
to

answered the question what prose we were allowed

consider
as

absolutely

unmetrical.

We

investigated

the sentence-metre of Demosthenes, and also his clausula.

We

stated

that
to

the

clausula

of

Demosthenes

had

in

most cases
part of the
five
it

be regarded as a metrically non-separate

that in
this

whole sentence, but

one case out

of

was

w w.

Indeed

figure

was higher
as something

than

we

could expect.

From

this

we
is

inferred that Defelt

mosthenes has one clausula which

separate, viz. the double trochee.

You
metre
I

will

have seen that

quoted
the

in dealing

with antique prose-

never

antique rhetoricians,

where

they are speaking of prose-metre, nor the orators themselves.

I

had a reason
it

for doing so.

From

a methodological
first

standpoint
tive facts

is

necessary to investigate
find,

what objec-

we
of

can
the

to

treat

afterwards the notions

If

and ideas
rule
to
is

ancients,

which bear on them.

tempted, as
e.g.

this

neglected, one

may be

Norden

compose a scheme without any ground.

Or one may

41

be tempted as

e.g.

Blass,

Zander,

and Bornecque,

to

explain the facts from an hypothesis, instead of deriving an

hypothesis from the

facts.

It is

necessary

first to

ascertain

what

hypothesis can be inferred from the facts, and only

it

then to see whether

corresponds with the opinions and

assertions of the ancients.

Before

go on treating the metre

authors,
I

of Plutarch

and

some other

must deal with another merely

theoretical question.
It

has been found that

in

Demosthenes the frequency

of

the double trochee in the sentence

14*1
/
,

was

12-4/o>

the

sentence of Thucydides
dides 14-2%.

in the clausula of

Thucy-

On
we
1

the other hand in the clausula of

1

Demosthenes

got
/o

8-9

%.
1

We

concluded that the difference between

too great to be ascribed to chance

2*4

and

8.9

was

only.

Were we justified in doing so? To answer this question we shall have

Suppose
that the
is

to consult the

theory of probability.

tendency which
quite the

causes the frequency of a certain form,

in

same

two

different prose-texts,

how

great

is

the probability

that

we

shall get a difference as e.g.

between 12

% and

18%?

We
tigated.

may
The

calculate the probable error of a percentage.

This probable error mainly depends on the material inveslarger the

number

of cases investigated, the

more

the figures will approach to the 'ideal' figures. For

42
the exact

meaning

of the

probable error

refer here to

the textbooks on the theory of probability.

The probable
form

error of our percentage

is

12*4 for the

w w m Demosthenes
it

0-6745

J/

mm

'-.

This means that

is

as probable that the real figure lies

between 12-41-0 and 124 -f 1-0, or between 11-4 and 1 3*4, as that it is lower than 114 or higher than 1 34.

The

greater the difference from these figures, the greater

difference
will occur.
If

the improbability that this

we

have therefore a difference such as

18-9,
is

e.g.

between 12*4 and

is

the probability that this difference

due
is

to

chance

very small.
it.

We cannot but think that there

probability,
it

something

behind

The

however, that no special cause,

about
is

no special tendency brought

impossible.
casting a
calculate

not altogether
that
in

As
die

certainty

as
:

we

can

calculate
3,

we
no

have

6 chance to get a
exists
in

we

can

the

probability

that

our case.

The

probability

that
in

real difference exists,

is

between
:

124%

and

8*9

our case

smaller than

5000.

We

shall

now

pass on to the metres of Plutarch.

As we saw

in

our

first lecture,

Plutarch avoids the

will

accumulation of long syllables.

You

understand that

here the law of increasing avoidance prevails.

avoids
will

Whoever
in

presumably avoid

43 an even higher degree, and even more

and so

on.
is

This
It

really the case with Plutarch.

is

possible, as

we saw

above, to express the degree

in

which the

different series of syllables are avoided

by

a quotient: a quotient found by dividing the percentage

for

certain

form

in

Thucydides by

that for the

same

form

in Plutarch/

So

the quotients are here:

Thucydides.
Plutarch.

Quotient.

(113120)

10-3
3-8
1-7
1-7

6-3
3-3

1-6(-)
1-1

(121124) (125126)
(127) (128)

0-8 0-8 0-8

2-4
2-1

( (
(-)

2-8

9-3

The form

128 comprises series of eight and

more

than eight long syllables.

An
is

even clearer insight

into of
his
I

the metre of Plutarch

given

by a comparison

sentence metre with

refer

his

clausula.

To

understand

this,

you

to the cor-

responding figures.

Clearly three preferred forms must

viz.:

be regarded as a separate part of the sentence,

44

These three forms comprise the

'clausula' of Plutarch.

The other forms do not belong

endings:

to his favourite sentence

they

are partly metrically indifferent to

him,

partly avoided.

The

'famous' double cretic, which

modern

scholars like to consider as a preferred form everywhere,

has no greater preference than other forms.

The form

w w

w w w in

is

often

combined with

and

one scheme

for
all

it

is

still

often thought that the cretic

is

the base of
eg

prose-metre.
is

This

form,

however:
in

ndoiv

ydQag

eX&elv,

probably not avoided

of

a higher degree than

the

other subdivisions

w w.
Autarch.
sentence-

The

quotients are:

Clausulametres.
Quotient.

metres.

8182 8384 8586 8788 8990 9192 9394 9596

1-50%
2-00
1-60

3-10
3-15 2-55

2-07

(+) 1-58(+)
1-59(+)

1-80
1-90

3-85

2.14 2-16

(+)

4-10
4-05

(+)

1-40

2-89(+)

2-20
1-30

4-25
4.05

1-93(+)
3-12

(+)

45

The
give

quotient

for
is

w
The
Is

ww
other
there

is

2*89,
it

that
is

for

ww
lower
the

3*12.

forms,
an}7

true,

quotients.

indeed

connection
cretic,

between the double trochee preceded by a

and

same form preceded by a molossus?

question.
of

Only a more

extensive investigation can give a satisfactory answer to

this

Suffice

it

here to have stated the method

by means

which

it

can be answered.
of investigating the clausula of the
It

The common method

that
of

is

comparing only the figures

metres)

second column
be clear to

(clausula

with

each

other.

will

you
it

that

such procedure has no

scientific base,

nor can

give any reliable results.

The double

cretic,

indeed,

can hardly be regarded

Its

as metrically indifferent to Plutarch.

frequency in the
its

sentence-endings does not reach the figure of

in the

frequency

whole sentence.

So

it

has been avoided.

Plutarch.

Sentencemetres.

Clausula-

metres.

Quotient.
2-09 2-89

_w_ w __ w w_ w w w
In

13-90%
1-40

29-10
4-05
2.35

(+)

(+)
form

4-00
I

1-70(-)
to

passing

drew

your
it

attention
felt

the

w
of

w w.
sentence,

Whether
is

is

as a separate part

the

not quite certain.

That

w w

is

*6
a good clausula can hardly be doubted.
length of the clausula
in

But whether the

such cases, where the double

cretic,
eg

trochee
four,

is

preceded by a

comprises more than

xetgag iX&eiv, is yet a

viz.

seven syllables: ndotv

problem.

Only

to

state

this

problem, to show the

I

way

to its solution,

and

to point out its importance,

undertook

as

my

task.

Some

years

ago Mr.

Henderson

in

the

American
whereas

Journal of Philology tried to trace the origin of the Greek

accent-clausula to the

Greek

quantitative clausula,
it

other scholars are inclined to think that

is

of Latin origin.

This later Greek clausula shows only a few favourite

forms,
viz.:

avdQcbjiovg naidevovoiv

oo

oo

oo

oo

.,

or

dv&QCDTiovg Xinelv

oo

oo

oo

oo

.,

or
or

av$QWJia)v xarahyovrcov
leyovoi jimdevovxcov

...oo
. . .

oo

oo

oo

oo

oo...,

oo

oo

oo

oo

oo

J>...

In general, the
is:

main
last

rule of this

Greek

accent-clausula

between the two

accented syllables of the sentence

or colon there are two or four unaccented ones.

Now,
certain

if

the

supposition

is

correct that there

is

connection between long syllables and accented

ones
in

in later

Greek, and that

it

this

connection shows

itself

the

clausula,

may be

possible to find an evolution

from

47

to
i.e.

Cfl

0>

C/

CO

. .

.j

from the classic or quantitative

accent-clausula
syllables.

to the accent-clausula.

This

does not distinguish between long

and short
sign
co

Therefore

it

is

desirable to use the

is
still

for

a syllable, the quantity of which

not

felt;

a syllable anceps, on the other hand,

we

shall

indicate

by w.

Now,

we

can

perhaps

explain

co

c/>

c/>

J)

from

w
is

-^.

In this

way

w
however,

w
it

or

regarded as the original form of w w w w w w w: for

ti&oiv

instance

eg

%eTQag eWeTv or xqijoijuwtoltov (palverai beIf,

comes
that

ndoL xaxaXineTv.

could be demonstrated

^ w
and

and

^
my

w ^,

if

anywhere,

occur

as favourite clausulae only in one or a few

this is really

Greek

authors,

opinion, this theory cannot

be upheld.

As
the

far

as this

we have

regarded the
of

'clausula' as

end

of

the sentence
short
syllables

consisting

a combination of
to

long

and

which
it

is

preferred

other

combinations.
exists,

We may

call

the positive clausula.

There

however, another

'clausula',

being the end of a

sentence, but consisting of a combination of long and short

syllables,

which combination
is

is

avoided.
it

This negative
is

clausula

known

to us
.

from Plutarch:

formed there

by

the series

48

You should
have seen
in the
,

rightly

understand what

mean.

We
,

preceding chapter that the series

in the

etc.

sentence are avoided by Plutarch.

in the sentence-endings.

This

is

probably also the case

it

When, however,
the clausula

is

said that the form

clausula,

should
to
i.e.

be regarded as a negative
is

we mean
is

say that
that the
in

avoided as such,
of the sentence

form
a

at the

end

avoided

higher degree than in the sentence.

This form, thereis

fore, at the

end
In

of the sentence in Plutarch,

metrically
is

separated.

ordinary
In

Greek
it

prose
is

its

frequency

nearly

20*4%.

Plutarch
it

11-4% throughout
%

the

sentence.
In

In his clausula

is

7-8

passing
is

we

involuntarily

approach a question,

which

intimately connected with these problems.

By modern

scholars

the

clausula

is

often

used

in

defence of their opinions about readings of manuscripts,

or about conjectures.
to

Many

conjectures have been objected

clausulae, others

as

they would

give bad

have been

defended as they would give good forms.

preference

Of course the

shown by
as an

a writer for a particular clausula

may be used
a conjecture.

argument

in

favour of a reading or

But how?

We

have seen that Plutarch has only three favourite

clausulae:

49

If,

however,

we compare

the frequency of the different

clausula-forms in Plutarch with that in Thucydides, other

and more numerous differences appear.

that
at in

It

is

seen

e.g.

the

form
of

w ^^w
the
:

xexzrjjuevco juetqiol in
is

Plutarch

the

end

sentence

much more
i.e.

frequent than

Thucydides
occur

3*85%

2-0%,

nearly
this is

two times as
due
to a pre-

many

in Plutarch.

Whether

ference which exists in the clausula only, or whether this

preference also exists

of

in the

whole sentence,
is

is

a matter

no importance here: there

preference
the
for

undoubtedly a tendency,

a
of

the

form

w.

w^^
criticism

at

the

end
just

sentence.
it

As
as

to

textual

we must
is

as

well regard
,

favourite

clausula as the forms

-^^^
at

w Wj w
that
of

There

only

this

difference

these forms are preferred only or mainly

the

the

end
is

sentence:

the

first

mentioned form,

however,

preferred everywhere.
therefore,

We
(1)

must,

sharply

distinghuish between

two tendencies:
a tendency in favour of forms in the whole sentence, or the reverse.
(2)

a tendency in favour of forms in the clausula only,

or the reverse.

50

The
dides.
in

first

mentioned tendency

may be measured by
that of
in

comparing Plutarch's sentence-metre with

In our

Thucy-

example the form

/
,

Thucydides
1 1

the sentence reaches 20*3

last

in Plutarch

only

*4

%.

The
those

mentioned tendency has been measured by us

figures of the clausula of Plutarch with
of

by comparing the
of

the

sentence
it

Plutarch:

we

got

11*4% and

7-8%, from which

clausula
results
exists.

appeared that also here a negative

of the
is

The combination
difference

two tendencies

into

the

which

obvious

when we

7-8o/

compare the clausula _18-3o/

.

of Plutarch with that of

Thucydides:

If,

therefore,
to

we want
textual

to

apply the results of clausula

investigations
result of the

criticism,

we have

to

take the
to

two tendencies

into account.

So we have

take into account the deviations of the clausula of Plutarch

from the clausula

of

Thucydides.
not
irrelevant.

These
inclined
to

facts

are

For we might be
,

consider the

forms

w
as
of
is

>^^^

^ ^,

and

^
known

as the only three

to you, did not

good

clausulae.

Zielinski,

even consider the possibility

the
clausula-figures

making a comparison

of

with

sentence-metres, nor with non-metrical prose for the sake

of
is

textual

criticism.

According

to

him a frequent form

a preferred form, a preferred form in textual criticism

is

more

likely

to

be the authentic form than the others.

51

All his notions on this point are

It

most confused and naive.

form
xexTijjuevq)
/uetQia,

is

clear
in

e.g.

that

the
at the

^^w,
just

Plutarch

end

of the sentence is

little

more frequent than

in the sentence:

3*85%

3-70 %

When we
is

know, however,

that this
in

form as a

sentence-ending
viz.
1

much

less

frequent

Thucydides,

-8

%,

it

is

obvious that another tendency exists which

3*85.

has

produced the figure

to take into

Only

this

tendenc}/

we

have

account in textual criticism.

This distinction becomes even more important when

we know
as a
6-5

that the

form

was

felt

by Plutarch

good

clausula: avdQwnovg xojui6vtcdv: in the sentence

%j at the end 9*8 % Yet in the clausula of Thucydides The sum, the this form is nearly as frequent: 9*5 %
result
of
is

the

two tendencies which have been spoken

very small.

of

above

therefore
(i.e.

Obviously the former

tendency
in
(i.e.

the one bearing on the frequency of forms

is

the

whole sentence)

negative in Plutarch, the latter

the
is

one bearing on the frequency of clausula forms

positive in Plutarch: they neutralize each other.
criticism,

only)

As

to

textual

therefore,

we may

regard the

clausula form

w
up,
for

as metrically indifferent.
criticism

To sum
the
result
of

textual

we have

to take

two tendencies

into consideration.

To

find

out

this result

with that of

we may compare the clausula of a Thucydides. If we do so, we cannot

writer

expect

52
to

get so simple results as

when we compare

Plutarch's

clausula
shall

with

his

sentence-metre.

On

the contrary,

we
The
of

often get a confused system of tendencies in

at first sight are to

which

no principal features
double
origin

be seen.

of the differences

between the clausula

Thucydides and the clausula

instance
the

of Plutarch necessitates for

juergia

form
for

xextrj/ueva)

to
in

be regarded as a
spite of the fact

good clausula

textual
it

criticism,

that, as a clausula,

does not form a metrically separated

part of the sentence.

It

will

be expected

that the clausulae of the SvyxQioeig

of

Plutarch

show

the

same

peculiarities as those of his

Lives.

Already a very small material (246 cases) gave

me

the following figures:

Lives.
HvyxQLoeig

Thucydides
Clausula-

Sentence-

Clausula-

Clausula-

metre

metre

metre
27.007,,

metre

ww
r
By
find an

___
w

13-90%

29-10%
9-95"/,, 9-95%

14-2%

6-10%
6-507 6-50%

9O0Y,

5-2%

9-80% 9-807,,

900%

9-4%

investigating

more material we

shall of course

even more striking agreement.

In

conclusion
,

we would
it

ask the following question.

of

w, the real length Are ww w w w, w not www or w the clausula?

***

Is

53
or

w?

The

first

is

indeed the

case.

We

shall try to

determine again the indifferent syllable.

Plutarch,

(the quantity of the final syllable is neglected.)

Sentence- ClausulaSentence- Clausula-

metre

metre

metre

metre

81

0-50

82

83

100 140

095 215
150 T65 155 100
1.75
2 10

2 3

w-

84 85
86
87

060 090 070 090

-

4
5

010 020 030 020

0'20

0-15

015 040
0-65

+ +
-f

0'40

6
7

w-

88
89

0*90

8 9
10

090
1.00

1-60

WW,

90
91

2.50

0'50

210
195 180

92 93 94 95
-

190
1-20 l'OO

0-70 0-60

245 220
1-85

020 030 050 070 090 060 060 090 090 040

065
0*35

4-

+
4-

030
0-95

155
0-85

444-

095
135
1'40
1-40

+
4-

44-

96

no

T50

Sentence- Clausula-

metre

metre

105 106
107 108

w^>_w
w_-

w ~ w
ww w
w

w
w w w

110
0-80
1-10

065
1'45
4-

1-20
1-45

H-

0-70

109

110
111

112

w w w w

w w
w

+
444-

0*40
0'80
l'OO

090
135 145
1

w060

35

54

From

these tables

we

can derive the following:

Plutarch.
SentenceClausula-

metre

metre

Quotient

8188 89-96 1_8

9

ww w

6-90

12 65

16

^^^^_ ^^
^ w
w

780 200 610

1645

105108 109112

w w

3-70

2-80

305 995 475 505

183 (+) 211 (+) 153 (+)

1

63 (+) 128 (+) 180 ( + )

The
of the

slight differences

between each couple

of quotients

(1-83 2-11, 1-531-63, 1-281-80) prove that the quantity

preceding syllable
is

nearly or entirely indifferent.

That, however, a long preceding syllable

to a short one,

may be

preferred

cannot be denied.

The
scholars.

clausula of Plutarch has been neglected

by recent

The same

is

true for that of Philo Judaeus.

And
It

yet

also

here exists a system of metrical tendencies.

peculiarities in

shows many

common

with Plutarch.
>

The

sum total of the favourite forms in Philo is 84-4/ the sum total of the same forms in Thucydides is only 52-1 % The number of exceptional cases in Philo is 15-6/o> the number of the same forms in Thucydides reaches 47*9 %

A
of

clear

view

of Philo's clausula is given

by our graphical

scheme.

In the

same way

shall give

you a comparison
have
are

the corresponding figures in Thucydides (see curves).

As

to

these curves you will

remember what

said before: perhaps the slight differences

which we

55
able to
state

between Thucydides' sentence-metre and

clausula-metre are not only and exclusively to be ascribed

to

chance.

Perhaps they are caused by some metrical

tendencies.

Perhaps these tendencies

in the

will

prove

to

be

even more obvious

more

rhetorical than in the less

rhetorical parts of his work, although Rollmann's investi-

gations
ever,

may have proved

the reverse.

In

no

case,

how-

are these

differences so great here as in Plutarch,

or as in Philo.
In

Philo
If

the curve.

we we

clearly see

some culminating

points in

try to investigate them,

we

get for the

clausula of Philo the following scheme:

Quotient.
2-0
1-6

(+)

HI

__

(+) (+)
1-8
1-9 1-2

1-6

(+)
(+)

(+) (+)

4-5

IV

1-4 1-2

(+)

V
The
first

(+)

M
form:
xzrj/uaz
eXeItiezo
is

(+)

entirely absent in
I

Demosthenes, Plutarch,

and,

as far as

investigated, in
it

Isocrates, as a favourite clausula.

Very frequently

occurs

56
later

on

in

Chariton (5*9 %), and also

in

Plato

it

is felt

as

a favourite form.

The form
nearly
all

-^^^^,
Greek
is

yevojuevwv

is
It

very

frequent in

metrical

authors.

is

the typical

Greek
it

clausula,

which
if

entirely absent in Latin.

In Philo

is

only sought,
to

a long syllable precedes.

For,

if

we

try

determine the length of the clausula,

we

get the follow-

ing figures.
Philo.

Sentence-

Clausula-

metre

metre
1

Quotient
3'0

wwwww ^w^^
If

xo yevojueva
\

x (bv

yevojuevwv

% 5*9 %
3'0
to

'0

9-6%
is

1*6(4-)

The preceding
the clausula
to

syllable,
is

therefore,

not indifferent.
it

be

^^^^ ^^w^:.
speak
third
of
it

be a good clausula,

ought

The form ^-ww the know, much recommended as

(

fourth

paeon,

is,

as

you

a clausula b}f Aristotle

we

shall

below.
fourth

The
exUyovoiv
later

and
in

form:
all

naoiv

av&owjioig,

and

occur

nearly

our authors:

only Plato's

works avoid them, even

Frequency
Thuc.

to

an ever increasing degree.

of the clausula
Plut.

Philo

:=:.

Demosth.

Plato Rep.

Plato

Laws

exleyovoiv

14*2

189
8*0

29*1

182
15'8

139

5*7

naOLV avftocbnoig

9'4

98

65

60

57
In this connection
to
I

feel

obliged to

draw your

attention

a fact of the highest importance which occurs again

I

also in Latin.

think,

we

are allowed to regard the forms

zd yevo/uev
jtaoi

avQO)noig

naQEypvoiv

naOL xaxeXemo/JLE^a

^ ^^^ -^ ^^^ ^^2^

^^-^
In Plutarch,

as by-forms of

^.

where only three

forms were preferred,

In

grouping the

clausula-forms

we were not allowed to do so. we are not allowed to

syllables
for

arbitrarily
syllable.

substitute

two short

one long
a bad clau-

By

so doing

we

should often

make

sula of a
for
all

good one. But here the preference which appears

these forms

makes
in

it

very probable that there

is

some connection between them.

appear
cipal
to

Now,

these

by-forms

be preferred

a higher degree than the prinis

form.
but,

Of
I

course, their absolute frequency

much

lower,

think,

you

will

not

regard

this absolute

frequency as a criterion of preference any more! Therefore,

we ought
denoting
is

not to
a

regard

this

notion 'principal form' as

the 'by-forms', as

more preferred form than

often

thought.

On

the

contraiy,

it

denotes only the

it

form

which comes

first in

our scheme as

contains the

lowest number of short syllables

nothing more.

We
will

might say: they are the most simple forms.

the

As

regards

connection between them

and the by-forms

we

suspend our judgment on

this point for

a moment.

58

The
dicretic

significance of the
clausula,
is

fifth

form naoi

njuw/uevoig, the
It

generally overestimated.
is
it

is

often

thought that the

cretic

the
is

basis of

all

prose-metre,

and from
cretic is

this

postulate

deduced

that the double

one

of the

most preferred forms.

The
is,

deductive

method

in investigating antique

prose-rhythm

however,

arbitrary
i.e.

and absolutely wrong.

and
in

Only

in a

few writers,

in Philo

Plato the clausula ^

^ occurs
,

as a favourite form.

As you
,

will
etc.

see,

Philo avoids the forms

In

this

way we

get

some

insight

into

the simple

means by which
rhythm.
entirely.

the ancients built their admirable proseits

The
Only

right feeling for

details

we

miss almost

indefatigable industry of investigation and

of

very long practice

reading

aloud

can give us back

something of the beauty of the ancients.

FOURTH LECTURE.
As
in

far

as this
I

have dealt only with those authors

able to state changes in the
is

whose metre
of
their

was not

course

activity.

Probably there

no change,
is

no evolution
with the

in their clausula.

Quite different

the case

artist

of

whom
you a

intend to speak to you now,

with Plato.
In order to give
of

superficial idea of the evolution

of

his

metre,

shall

remind you

our former results

which we obtained with the help

for the Republic,

of the

method-Marbe

and end

for the

Laws.

We may regard the

of his

Republic
the
of

as

the

of the first

group

works

Laws
a

not only as the end, but also as the culmination

Plato
is

second period of his evolution.

his later

years prefers series of short syllables as obviously as he

avoids

and

^^-.
important
than
his

Even

more

sentence-metre

is

that of his clausula.

Not only as
It is

to his clausula, the

Laws
comRe-

show a

certain culmination.

very interesting

to

pare the frequency

e.g. of the

form

^ ^

in the

public with that of the

Laws.

60

Frequency
(i

of the clausula

w w.
164
14-3

n percentages).
Phaedrus
140

Thucydides
(first

thousand)

Convivium

Thucydides
(second thousand)
14-4

Phaedo
Theaetetus

141

135 160

Demosthenes
Philo

189
18 2

Parmenides
Republic:

Chariton

213
1

Book

12 5

Lesbonax
Herodes
Libanius
Plutarch
(first

14
1

II

149
122
121
12 6

16

III

161

IV

V
296
28-6

thousand)

VI
VII
VIII

15-8

Plutarch

135
161
14 3

(second thousand)
Plato

IX
14
1

Apology
Crito

X
Philebus
Politicus

169

15 8

50
7-1

Protagoras

140 133

Charmides Laches
Lysis

Sophistes
Critias

97
13
152

130
14*7

Timaeus
Laws:

Euthyphro
Gorgias Hippias minor

127
12-2

Book

82
7-5 5'0

152

II

Euthydemus
Cratylus

158
161

III

IV

5-3

Meno
Menexenus

142
11-4

V
VI

45 33

61
VII
VIII
6-4

XII

46

67
5-7

Epinomis
(first

IX

hundred)

X
XI
In

56
4-5

Epistula 7 ma

40 90

the

Laws

the following clausulae form a separate

favourite part of the sentence:

ii

(-) (_)

HI

ww w WW

IV

(-)

ww
we
find again in Plutarch,

Only one
viz.

of these
It

forms

ww^

is

the

famous fourth paeon.

By way
of the

of exception

refer to a passage in Aristotle,

of the ancients the contents of
in a

one

few passages

which have

been confirmed by us
(Rhet.
Ill

most

striking

manner.

He

says

8.

1409 a
Jiaiavog
. .

2):

eoxi

de

dvo

eidr]

dvxixeljueva dXXrjXoig, cbv xo

jLiev

ev

ciQxfj

dojuoxxei
xoeig,
fj

exegog d^e$ evavxlag, ov /3oa%eiai

(^ww)

do%ovoi

de juaxod xelevxaia

(ww- w

).

/uexd

de ydv vdaxd x

dbxeavov y)cpdvioe

vvg~.

And
. . .

little q>

above:
exQcbvxo juev

nmdv,

and Ooaovjudxov dg^djuevoi absent

in

The
typical

fourth

paeon

is

Latin

it

is

the

Greek

clausula.

62

Most important
as a clausula.

is

the

occurrence of

)w
Quotient.

To

this clausula

may

belong:

7iaoiv iyyiyvetat

( )

)
*

'

'

(-(-)
(-)-)
(

jidvTcov exyiyvsxai
jiaoiv

eyyiyveodm
is

final

There
in
all

an important reason for joining these forms:

is

of

them there

a preference to use a long

I

syllable,

and not a short one.

explain are
felt

can find no other hypothat

,

thesis

to

these
to

facts

than

the

last

three

syllables

be a

cretic

or a molossus

substituted for a cretic.

To show
speak
In
to

this

clearly

must make a digression

to

you

for a

moment
I

of the so-called syllaba anceps.

my

investigation

myself neglected the quantity of the

last syllable,

except in Plato.

For

this investigation

would
should

have taken up much time: instead

of 128 cases,
It

we

have had

to

investigate

256 forms.

is

the

general

opinion that the quantity of this last syllable is indifferent.

Josephy, however, rightly doubted the correctness of

opinion,

this

which doubt he based on most convincing

is

statistics.

The

value of these statistics

questioned by Miinscher
000) on the authority of

(Bed. Philol. Wochenschr. 1915,

col.

ancient texts: a methodological mistake.

The passages

of

Cicero and the others bearing on the clausula have deceived

63
us even more than those of
better

many

recent scholars:

we had

put them aside.

Besides

we may

cite the text of

Aristotle in favour of the opposite opinion.

But

all

these

questions are very unimportant.

is

only important one

Statistics

this:

what do

statistics

The show us?

show

that in the
is

Laws

a metrical form with

a long final syllable

always more frequent than the

final syllable in

same form with a short

well as in the clausula.

the sentence as

In connection with the greater

in

frequency of long syllables

general

we

could not have

expected

otherwise.

Only one form, however, deviates

sentence as well as in the clausula,

from
viz.:

this rule, in the

www; ^^ But
.

this
this

form

occurs

more frequently than

us,

need not surprise

when we
that

re-

member
quency

that Plato seeks the tribrach,

and

he avoids
fre-

the choriamb.
of the

We

may draw

a curve denoting the

forms ending

in

^,

as

well as a curve

denoting the frequency of the same forms ending in

It

is

and

most interesting see how the forms w www www are more frequent than ^ w-w and ^^ For quite the other forms the
to
.

all

ratio is

the reverse.

In this

way

it

appears again and again that

even the most

subtile metrical tendencies are

shown by

means

of simple statistics.
all

And

it

is

quite interesting to

observe that

our figures bearing on Plato are derived

of

from a paper

Kaluscha

who made

a rather curious

64
use of his
statistics:

the idea,
to him.

the notion of percentage

seems

to

be unknown

So we see

that the last syllable in Plato

is

is

not anceps.

Now
rfjg

the

same
for

true for the form

w ^i

where

yQacpeioyg dlxrjg is preferred to xQijjudrwv xg/joi/ua.

The

quotient
the

the

first
is

form as a clausula compared with

larger.

sentence-metre

The same

holds good for

After this digression on the 'syllaba anceps'

to the clausula of the

we

return

Laws.
either

It

appears that

in the dicretic

form

cretic

can, so to say,

be

re-

placed

by a molossus.

double

molossus,

however,

would destroy the character

of the cretic clausula.

Besides to the positive clausula

attention to the negative one.
tant
in

we

should also pay

The
else.

latter is

more impor-

Plato than
is

for

instance,

The double trochee, sought by most authors. Now, only few

anywhere
to

forms are avoided by Plato

trochee:
later a

such a degree as the double

ixUyovoiv.

On

the whole

we

can say that the

You

w
this

work has been composed, the lower the percentage. will remember that the form naoiv dv&Qcojiotg, w, plays a part in Greek prose metre: even
is

form

avoided in Plato.

For the bad clausulae we

get the following scheme:

65
I II

Ill

_WWW ww w ww w -WW w
(

seems
).

to

be preferred

IV

(including

w
etc.

w,
,

as clausulae).

The avoidance
for this reason that
logical

of
it

the
is

clausula

heroica

is

important

symptom
in

of a general

psycho-

law,

which

is

here and there very obvious in

antique prose-rhythm.

Any metre
dans

any way connected

with poetry

is

avoided.

Bornecque

in his

most important
de

work

'La
is

prose

metrique

la

correspondance

Ciceron'

even inclined
is

to believe that the

whole prose-

metre of Cicero

to

be explained by

this factor.

acknowledges only the negative

others on p. 198:
'Je

clausula.

So he He says among

crois,

d'ailleurs,

com me M. Meyer, qu'on peut

donner de
flatter

la

prose m6trique une theorie simple: sans

trouvee, je

me

de

l'avoir

me
...

hasarde a exposer

celle

que

les faits

m'ont sugger^e

Le

seul prCcepte absolu

les autres

que nous donnent, non seulement Ciceron, mais

grammariens,
a une
fin

est d'6viter

que

la fin

d'une phrase ressemble

de vers etc'
is

This opinion
agree with

very one-sided.
the

Probably you

will

me

that

metre

of

Demosthenes, which

recognises only the double trochee as a clausula cannot

be explained

in this

way.

Also the curves

of the

metre

66
of Cicero

and

of

Livy

will

convince you of the existence

of the positive clausula.

The main
the form of the

opinion of Bornecque,

however,
is

is

this:

word preceding

the last one

is

determined

by
of

the form of this last one.

This

even his definition

metrical prose: 'On dit qu'un texte est 6crit en prose

la

meHrique lorsque

forme meHrique du dernier mot de

la

chaque phrase determine

precedent
le

forme m6trique des mots qui

p.
1).

mot

final'

(La prose m6trique,

If

We
word

can only partly agree with him.

the

Demosthenes favours
last

form

w ^,
like
If,

he

likes to
in

have before a

avxovg
jiotovoiv

a form
avzovg.

tioiovoiv,

order to get the clausula

last

however, the

word

is

ixMyovoiv,

we do
the

not understand

how

it

can have influenced the

preceding one; for

length
in

we

are able to accurately determine

of

the

clausula,

and

to

state

that
all

exMyovoiv
the pre-

forms

itself

beautiful clausula,

and that

ceding words are indifferent.

his

By means

of this

argument

theory

is,

as

far

as

see, refuted in a

most simple

and convincing manner.

In

the

meantime Bornecque's remarks should not be

regarded as unimportant.

They

clearly

show

the signi-

ficance of the negative clausula.

They
it

are illustrated for

instance

by the

clausula heroica
it,

of Cicero.
is

He seldom
of

uses

it,

but wherever he uses

seldom the end

an hexameter.

We

know

that

a good Latin hexameter

67
in

Cicero's

days did not end

in

a four- or five-syllabic

word

or word-group.

The

current forms are

vincla resovit,

and

lamentabile regnum.

Only
ridarum,
here.
If,

a few

Greek words are excepted,

e.g.

Tynda-

phenomenon which need not be explained

however,

heroicae of
endings.
Zielinski
I

we look more closely Cicero, we see that they are

some

at the clausulae

not hexameter-

quote here

clausulae heroicae cited

by

from the speeches

of Cicero (Clauselgesetz p. 751).

Quinctius cupit

commemorare

tota res transigeretur

paulo post

commemorabo

praetura Siciliensi

passuum conficerentur
videt discruciatur
cautius composuisses

pecuniam corripiendam
copiosus
est,

commemoravit

conscios flagitiorum.

Although Cicero avoids the real hexameter-endings,

he tolerates
the

use of such forms as caperentur and

perlegerentur to a certain degree.

So important a
in

part

does the avoidance

of poetical

metra play

prose metre.

The avoidance
in this

of

ww

in Plato

can also be explained

manner.

68
It

is

very interesting to note

frequencies of
life.

how
in

gradual the evolution

the different periods

is

in the

some forms
of

of

Plato's

Bij

means

hypothesis

we

can

dis-

tinguish five chronological groups:

I.

Protagoras, Crito, Apology.

II.

Charmides,
min.,

Laches,

Lysis,

Euthyphro,

Gorgias,

Hippias

Euthydemus, Cratylus, Meno, Menexenus,

Phaedrus, Symposium, Phaedo, Theaetetus, Parmenides.

III.

Republic.
Philebus,
Politicus,

IV.

Sophistes,

Critias,

Timaeus.

V.

Laws.
is

This distinction

made by Kaluscha
It

in

his

paper

on the chronology

of Plato's dialogues.

contains valuable
is

material for the clausula, a material which

intentionally
in a rather

only partly used,

curious

and so

far

as

it

is

used

way.

No comparison whatever

with sentence-

metre

is

made.

before, to be

The notion 'percentage' seems, as I said unknown to him. I will not, however, speak
evolution
to

about

this

matter here.

The gradual

which

alluded appears

most clearly from the following

figures.

69

Group

II

III

IV

3-50/0

50
7-6 76

6*2
8.1 81

7-3
9-1 91

26
11*6

( )w

6*9 69

61
7*7

5-7
7-1

55 70
139

52

3-7

32
91

13
5-70/

143

143

These
collected

figures

have been calculated from the material

to

by Kaluscha, a material which we are sorry

fully

say he did not

work

out.

Some

of the clausulae

show

a gradually increasing preference for them, others

a gradually increasing avoidance.

to ask:

And we
of these

are allowed
in

what about the frequency works

the

forms
is

those

of

Plato's

chronology of which

not quite

certain ?

The
the

history of the opinions of

is

modern scholars about

chronology of his dialogues

students

somewhat

as follows.

The

who
As

first

occupied themselves

with

it

all

agreed on
earliest

this

point that the

Phaedrus was one

of his

works.

there

were doubts about the correctness

were made by many
most
of

of this theory, careful investigations

scholars, but

even

after these investigations

them

adhered

to this theory.

Some

time

later,

however,

statistical

evidence seemed

70
to

show very

clearly,

that the
is

Phaedrus was written

in

a later period.
"... Meistens

Raeder

quite right in saying (p. 248):

dargestellt, als

wird aber das Verhaltnis so

Inhalt geschopften

ob die aus

dem

Argumente den sprach-

lichen Kriterien schroff gegenuberstanden".

It

is

easy to understand that people should try to

this
difficulty,
it

overcome

was even

a scientific require-

ment, and he
the

who

has any

faith in statistics
late.

had

to join
to

few who have placed the Phaedrus

Owing
of,

this there

were scholars who, on the ground

or rather

with the help of the criteria which were formerly known,

tried to assign a later place to the Phaedrus, in order to

make

their

ideas
In

correspond
this

to

the results of statistical

I

investigations.

connection

may

quote Pohlenz,

who

Von Arnim (Gotting. Gelehrte Anz. 1916, Es war also nicht Willktir oder Zufall, wenn p. 259): fast alle modernen Forscher den Lysis mit dem Symposion in engste Verbinding brachten, und wenn v. A. diese gewaltsam lost, so wird man hier wirklich den Gedanken
says about
nicht los,

dass er aus seiner Sprachstatistik die Ueber1

zeugung von der fruhen Abfassung des Lysis mitbrachte

Nearly the whole of modern
literature

'.

on

this

subject

moves

in

the

same

direction.

In this connection a par-

ticular preference is

shown

in the different in

works

for the

treatment

of

parallel passages

different

works,

the

chronology of which one would

like to determine.

Some

71

of

them,
of

such as
their

Von Arnim,

are of

opinion that the

result

investigation

'vollkommen ausreicht die

Prioritat des

Phaidros unwiderleglich zu beweisen'. Others,

E.
J.

such as
jg.

W.

Kuiper (Tijdschrift voor wijsbegeerte,

XI) are honest enough to admit that the treatment of

parallel

passages

will

often

convince
all

the

writer

only.

Some
ations,

philosophers are, in spite of

statistical investig-

convinced
late;

that

the

Phaedrus cannot have been

written

this
is

view
still

Schleiermacher
Ovink. wrongly.
it

was formerly held by held by Natorp, Gomperz, and

which
on
this subject, as

But most

of

them have an other opinion, and

other cases, have failed on account
error: incorrect

For

statistical investigations

happens so often
the well-known

in

of

statistical

and wrong

grouping.
that

Too

little

account has been taken of the fact

different

tendencies

can lead to the same results.

As
to

easily as Kaluscha, a disciple of

Von Arnim, has

tried

determine the chronological order of Plato's later works,

he might have proved that the Phaedrus has been written

in the earliest period.

For, as

you

will

remember, there

are at least five clausulae-forms in the frequency of which a

gradual development of Plato's metre

is

obvious.

If

we we

take the frequency of these very forms into account,

should be inclined to date the Phaedrus as early as

possible.

72
Plato.

Frequency
Phaedrus

of clausula-forms. Chronological group

I

11

III

IV

V
37
13

WS_^W

V4

61
7*7

57
71
143

5'5

52 32
91

____
_w
w;

94
164

70
139

14'3

57
126
11-6

(_)

www^ _^ w
As you
will

49
62

3*5

50
7'5-

6*2
8-1

73
91

6'8

notice:

even
are

the

forms

^^^,
in the

ww w,
drus.

and

w w,
clausulae

more frequent
(

Phae-

The very

^ww^ and )^
to

^^>
we
have

which are preferred

here.

in the later

period are less frequent

You

will

be inclined

suppose that

here the commencement of a gradual metrical evolution.

This,

however,

is

not the case: quite other laws are at

work.

Among
language,

the

works
It

of

Plato

the Phaedrus

is

a very

particular work.

is

written in an exceedingly poetical

in Plato
t

which

is

found perhaps nowhere

nor

anywhere
that
it

else in Greek.

Nor

will

it

surprise

ou

to see

has

a poetical metre which

shows a logaoedic

character.

73
Plato.

Sentence-metre.
Phaedrus
Chapt.
1st

XXVI

2nd
thousand
syllables

Republic

Laws

thousand

Average

syllables

174

167

17

_
3 _4 _ 5 6
-

os

146-2

112-8

85
12

74
21

79S
165
5*5

640
30*7

47-5 40*3
13-8

8 2

90
2-7
1-2

10

53
2'8

8 9
It

00 02

05

00 03

is

easily seen that the

Phaedrus shows an abun-

dance
can

of dactylic

and trochaic metra.

Of

this fact

everyone

convince himself by reading some chapters aloud.

this

Remarks on
in
It

phenomenon have already been made

especially on the finishing prayer.

some commentaries,
is

seen that the forms

frequent,
etc.

w and ww are more wwww that on the other hand www

y

are rather avoided.

Not seldom there are sentences

like this:

Ilaoa
(251
e
),

yaQ

fjv

xo

JidXai

meganr)

ww ww w

etc.

74

We
viz.:
(1)

may
the

say that there

are
all

three
the

criteria

which

distinguish

Phaedrus from

other dialogues,
(3)

poetical

choice of words,

(2) poetical metre,

avoidance of hiatus.
of

We

shall

now

investigate the influence

statistics.

each of these factors on the results of the

I

Therefore
Barwick,

especially

draw your

attention to a

work

of

De

Platonis

Phaedri temporibus,

of 1913.

Of

course he does not yet

know

the peculiar metre of the

Phaedrus, but he
cleverly

is

a scholar

who

is

not content with

in

found probabilities nor does he get absorbed

general speculations which are pleasant to read but which

do not prove anything.

He

gives

his opinion in a
statistics,

few

words, he gives simple and convincing

not go too far in his conclusions.
correct insight into
that

he does

And

that

he has a
fact

the problem

appears from the

is

he

gives

an opinion somewhere which he

for

not

quite

able to defend,

which he has no explanation

clear only

even, and which

investigations.
I

becomes

by means

of metrical

mean

the use of

cbg in

the Phaedrus, to

which

shall refer presently.

In a

most convincing manner Barwick shows

to the hiatus stands quite alone.

that the

Phaedrus as
of

The number

hiatus

in

Plato's

former works varies between 0*44

and 5*85 a page.

In the later
It

works
is

this

frequency varies

between 31*18 and 45.87.

rather surprising to see

75
that

the

Phaedrus has

23.90,

a figure that

lies

exactly

between the two other groups.

That
the
hiatus
is

avoided

in

the

later

dialogues
is

appears very clearly from these

far as
I

figures,
If,

nor

this,

as

know, doubted by anyone.

however,

we

think

that the avoidance of hiatus

to to

determine the
fix

may be used as an argument chronological order, we should be inclined

between both groups,
i.e.

the Phaedrus

before the
In

Critias, Philebus, Politicus, Sophistes,

this case,

Timaeus, Laws.

however, the Phaedrus would have been written

after

the Parmenides

and the Theaetetus, which

is

very

unlikely.
viz.

There remains only one possible hypothesis,

the

that

Phaedrus belongs

to

the former group, to

the time
poetical

when
style
of

Plato did not yet avoid the hiatus: but the

of
it,

this

work
of

involuntarily causes a certain

avoidance
that
to
it.

which,

course,

is

not so strong as
special
attention

of

the

time

when he

pays

We
offers

shall discuss other peculiarities of the

its

Phaedrus,

as the statistical treatment of

style

by recent scholars

many illustrations of wrong statistical methods. The frequent use of xl jur]v is indeed, as has been
of a later origin of this

remarked by the advocates

typical for the later period.

work,

But what

is

remarkable here?

That the Phaedrus, as well here as

in the treatment of

76
hiatus,

stands

quite

alone.

Though

it

is

by no means

the last dialogue,

we
the

get here
affirmative

a percentage of 68-8 on
rhetoric questions,
next.
total of affirmative
i.

sum total of 24-4% more than

the

e.

the figure

coming

Frequency
Phaedrus
Republic

of

tl fxr\v

on the sum

rhetoric questions.

68-8
23-3

Sophistes
Politicus

19-7

37-7

Theaetetus

46*4
11*8

Philebus

347
37-2

Parmenides
Besides,
the
alia

Laws
relatively
xal
jurjv,

Phaedrus has a
fjaqv,

very high
as

number

of

ye

fxrjv,

and

may

be

inferred from Lutoslawski's tables, in his

for statistical investigations

propaganda work

on

Plato,

The origin and growtli

of Plato's

logic (p.

184.).

No

other

hypothesis
suit

can

satisfy

us

than
of

that

these

expressions

the

poetical

rhetoric

the

Phaedrus:

also in the tragedy the expressions are very frequent.

Also
In

in

another peculiarity the Phaedrus stands alone.

the

former dialogues
it

we

often find dfjlov on; only in

dfjlov <bq.

the later ones

is

sometimes replaced by

The
on

same

is

true

for the Phaedrus.

is

Now, we know
e.g.

that

in the later

works

avoided because of the hiatus arising

in

between on and a following vowel

eoxlv.

on

wcpefojuog

The Phaedrus, however, does

not avoid any hiatus

after on, as evidently in that period Plato did not consider

77
the
hiatus

as

troublesome: on

is

used 35 times before

cog

a vowel, but only 16 times before a consonant; whith

it

is

quite

the reverse!

Therefore there must be some

other reason

why

dfjXov

on

is

avoided

in the

Phaedrus,

and replaced by

dfjlov cog.

Now,
3^ears.

in the

Phaedrus

cog is
is

very much more frequent

than on, although this too

It

a peculiarity of Plato's later

should be noticed that also here the Phaedrus

cog

stands quite alone: the figure for

for

compared whith

that

on

is

again

much higher than

this.

that of all the other

dialogues which were written before the Sophistes.

Baris

wick also noticed

reason

He

rightly thinks that this

the

why

dfjlov cog is
is

so frequent in the Phaedrus, and

he asks:
(p.

why

chg

so frequent here?

His answer

is

643i).

'Quod qui factum

alius aliter iudicabit;
ilia

sit,

utrum mero casu an

alia re,

me quidem

iudice artissime cohaeret

res

cum Phaedri
understand

peculiari sermonis habitu.'

Is this

the true reason?

To
metra.

the

true

reason

must draw your

It

attention again to the

metre of the Phaedrus.

seeks logaoedic

Now,
on,
i.e.

cog
If

consists of one long syllable,

on

of

two
and

short ones.

a writer has to choose between

dfjlov cog

and

dfjlov
?

he has

to

choose between

w
The

w^^
however,

between a

cretic

and a
is

tribrach.

tribrach,

is

avoided here.

This

the true reason.

78

We
where a

shall put this to the test.

There are many cases

Investigating

favourite

form can be replaced by a bad one,

d>g

and the reverse, by using

in

instead of on.
all

chapter 22 and the following ones

cog

the passages
I

where
on,

and on are used

d>g.

in the

same

sense,

got 21

and 27 times

Has

Plato's

choice indeed

been

influenced
is

by

his

metre? You understand that

this question

of

importance from a methodological point of view.

Its

solution

may prove whether

of

Plato's choice of

words

can partly be determined by metrical tendencies.

The
times,
of
cog

result

my

investigation

is

that cbg

is

used 27

and

that in the majority of cases, viz. 23, the use

tribrach,

on would have given a

gives a favourite form.
times, in 13 of

whereas the use

is

of

On

the other hand on

used

21

which
of the

cog

would have made a molossus

^^ In the 42 cases, where Plato could choose between w wwww and ^^ w and ww on the one ^w^^^ on the other, he chose 34 times one
from one

good forms

w
(

or

side,

etc.

of the
to

former combinations.
that the

From

this

we

are allowed
peculiaris

infer

remark

of

Barwick about the

correct.

sermo

of the

Phaedrus was quite

The word on
if it

can be used only between two long syllables,

to

is

tried

get

w
it.

or

ww

hence the low frequency.

But without knowing the metrical tendencies he could

not explain

79
In
this

manner

the

influence

of

sentence-metre on
of statistics.

the choice of

words has been proved by means

Any
(1)

definite

metre can be attained only by two means:

choice of
If

words
first

(2)
is

order of words.
chosen, these

the

way
and

synonyms such

as

(bg

and
first

on, cdotieq

and xadaneQ come

it

into consideration in

the

place,

is

these very

synonyms

that play

a predominant part in

statistics.

Also

some other
in

peculiarities

of

the Phaedrus

can

be explained

this

way. Kadaneo occurs only 4 times,

whereas
origin.

ojojzeq

27 times; this fact seems to indicate a late

it

Now,

is
e

an easy means to get an anapaest or

):

a dactyl, e.g. (249

the

6 eQ<bv x(bv xaXcbv eQaoxrjg xaiehat.

it

In

peculiar language of the Phaedrus

may

also

be a

favourite archaism.
Tcb
ovxi

and
fits

fj

ov are avoided because of the hiatus.

So one thing
the

with the other:

all

these facts point to

same

direction:

whoever writes hexameters cannot

use any cretic

of

word

hope

that

have convinced you

my

opinion that to the study of Plato also belongs

that of his prose-metre.

When we
its

acknowledge now

that the

Phaedrus with
it

peculiar

color poeticus

stands quite alone,

is

clear

that in order to determine the chronology

to

we

are allowed
this

use only those

criteria

which do not depend on

80
color
poeticus.

So

the

hiatus

must be excluded as a
rd>

criterion,

as

well as the avoidance of

it.

ovn and

fj

ov,

which depends on
as
far

So

the clausula must be excluded

of

as

it

depends on the peculiar poetical metre

the
of

whole sentence.

On
of

this

metre,
for

the

main features

which are the preference

the

and

avoidance

w and ww www wwww

t

etc.

depends the comparatively high frequency

of the clausulae

-WW
blic

with long

final syllable

(Phaedr. 10-1 %, Repu-

7-94%),

ww_w:

(Phaedr. 9-4%, Republic 7-0%),

ww
tion
of

(Phaedr.

16-4%, Republic 14-0%) on the one

hand, and on the other the low frequency of

(Phaedr. 0-00 %, Republic 6-20 %).
drjlov
(bg

wwww
will

The

peculiar substitu-

for

dfjlov

on

in

some passages

be

sufficiently clear

now.

Without any doubt the

ionic dativi,
ju/jv

the superlativi, archaisms such as xadajieo and

to the peculiar character of the style.

belong

Surama summarum, some

us

31

criteria

are

known
It

to

which

we

use

as

an argument for or against the

is

grouping of the Phaedrus among the later works.

thought (see Barwick
this
p.

00 and 00) that the Phaedrus

in

case comes near to the Theaetetus and the Republic.

31
criteria

Of even these
Theaetetus,
in

only

are

missing
in the

in

the

the

Republic only one,

11

Phaedrus,

however, 20!

Of the remaining

many

occur in the

Phaedrus only once; these

of course are quite insufficient

81

to

prove anything, for many

of

them are found

to

also in

some works which undoubtedly belong

group.

the

former

Those which

finally remain,

discussed before:

they depend on the logaoedic metre.

to

This

may

suffice

conclude

that,

as far as

any

faith

can be put in

statistical

evidence,

the

Phaedrus was presumably written before

the group to

which the Theaetetus and the Republic belong.

Thus
In

different causes

may have the same

on
after
is
it.

consequences.

the later period

dfjXov

avoided because of the

As,

hiatus

which may

arise
after

however,
to

in the

Phaedrus an hiatus
this
this

on does not appear

be

felt,

factor

is

wanting.

The avoidance
statistics of

of the tribrach in
dijlov on.

So
to
is

the

work is the method

wrong.
of

real cause of the

of

avoidance of

using

words and phrases

determine the chronological order of Plato's dialogues

quite
It

is

not always sufficiently asked

the different dialogues
is

why

the use
the

words

in

not quite
of

same.

On

this

point the time-taking

work

Von
is

Arnim, published
failed.

in the

Sitzungberichte of Vienna (1913),

He vehemently

defends

that

the

Phaedrus

posterior to the

Republic.

Elsewhere he says
to

that his
I

arguments are quite

shall not dwell

sufficient

prove

it

irrefutably.
:

on these questions any longer here

think

his

proof (Platons Jugenddialoge und die Entstehungszeit

p.

des Phaidros,

161)

is

more ingenious than probable,

more
It

logical than psychological.

should always be borne in mind what must be expect-

82 ed from the metrical tendencies

in

Plato's

later

works.

The

peculiar metre, as
(1
)

we saw

above, can be arrived at

(2)

only by means of

order of words

choice of words.

The

first

way

is

often

chosen by Plato.
less

The
by

frequent
of

use of
tqojzov

rivd
rivd,

xqotiov,

and the much

frequent one

the avoidance of the lengthening

striking
for

position,
of
it,

and many other phenomena are

and would give ample material

examples

separate treatment.

To
first

these

may be added
is

the frequency of

compounds

the

part of which

/tew

or dia

or
as
of
in

nam and
,

so

on,

the

frequency

of

participia

such

y.aradtxojuevog,

diafoyojuevog,

dedojuevog,

the

frequency

forms such as

xaraudftcooLv
etc.

and so

on.

Words ending

og,

ov,

av

are used by preference before a vowel. Some one may find time to show this fact by means of statistics. And what words are more likely to be chosen or rejected

according to metrical preference than synonyms such as

cog

and on

in

the Phaedrus, or than

in

ooojieq

and xa&aneQ,

juexQiJteQ

and ecoomQ

the

Laws?

We

ought even to

say

that,

knowing the metrical

its

peculiarities of Plato's later

years but not

consequences,

we

should expect some

preference in the use of these very synonyms.

And when

we have found indeed are we not entitled to

of the

obvious tendencies

in this direction,

regard the one as the consequence

other?

FIFTH LECTURE.
About the ancient
read
in

clausula curious theories

recent

literature.

Every scholar has

tries to

his

may be own

system of grouping, which he

of statistics.

defend by means
is

Hardly anywhere the question

of

put whether
to chance.

the

frequency

these

forms

may be due

Very
can

rarely are scholars convinced that only one system

be the true one; neither are they convinced that an

exact and reliable

of

method ought

to

be found by means

which
system

this true

system can be found.

moment ago
accepted.
is

we spoke
a

of

Norden's peculiar system for Demosthenes,

appears
in

which
nor

to

be

universally

Neither

here

Zander any comparison

made
if

with
detail.

sentence-rhythm, nor with non-rhythmical prose in

Mostly these schemes are put before

to

us, as

any

doubt as
read
in

their reliability
(p.

were impossible.
gravitas

Thus we
ubertasque

Heibges

7):

*Sed
dicendi

cum Demosthene mortuo

minuerentur,

paucae

illae

formae

clausularum

magis magisque adamabantur

ceptae sunt.
logo non notae.
1.

et

postea a Romanis sus-

Sunt autem hae quattuor, nemini iam philo(sic).

2.

w ^ = Ditrochaeus = A. ^ w Creticus auctus trochaeo

z= B.

84
3.

4.

^ ^^ = Creticus bis positus = ^ ^w Hypodochmius = D.'

,

C.

As you
it

will see, the

scheme
1,

is

quite arbitrary. Besides

is

wrong.

The forms
by
I

2,

and 3 are favoured by

some
is,

authors, avoided

others.

The
with

last

form

w ww
he deals
in
(p. 9):

as far as

know, not favoured by any Greek author;

true for

the

same

is

Chariton,

whom

particular.

He

says that Chariton uses numerous forms

'Attamen meras formas principales plurimum valere perspicitur.'

Now,

the form

^ ^^
it

is

one

of his

formae

principales:

and even as such

is

avoided.

In reading this
to

and other passages we ought always

neither

remember that mann's work on

here

in

Heibges, nor

in
is

Herany
It is

the

clausula

of

Libanius there

question about a comparison with sentence-rhythm.

Thumb's
Seldom

special merit of having pointed out

is

its

importance.

the frequency of the discussed forms in other

authors mentioned,
that

sometimes, however,

in

such a

way
it.

no conclusion whatsoever can be drawn from

it

can be clearly proved that the forms ^ Though w do not metrically belong together, on and generally favoured, the other the contrary, - ^
is
is

generally

avoided,

these

forms

are

often

grouped
to

together.
fact

Though
in

there

can

be

no doubt as
is

the

that

Chariton
avoided,

w
it

^^

favoured,

but

^^

^^

is

contended that a choriamb

can be substituted for a

cretic.

85
It

should always be borne in mind that the

is

word

'system'

used

in different senses.

It

may mean: system

of

of interpretation, as e.g.

where we are speaking

of

Norden's
it

system or scheme of the clausula

Demosthenes, or
system

may mean system

1

of investigation, as e.g. our

of

28 possible forms.

and apodictic one.

be some differences

The The

first is

mostly a quite subjective

is

latter

quite

objective,

as

it

has not any premise, except the premise that there

in the

may

arrangement

of long

and short

syllables in different authors.

As
to

far as

it

is
1

possible to reduce Heibges' figures to

our system of the

28 possible combinations,

it is

possible

compare Chariton's metre with

clear
insight,

that of other authors.

however,

into his clausula-forms cannot

be obtained before his

gated.

sentence-metre has been investi-

The favoured sentence-endings may be somewhat

as follows.
Positi ve clausala in Chariton.

Thucydides.
(iwv) hyojuevcov
jtaotv

Chariton
11-3

haQyeoiv

( )U^ww ^^^^_
s_.

5-2

21
142

5-9
21-3

exMyovoiv
Jidoiv avftgcDTZoig

:=1

92

17-6
7-2

ovfxnaoiv EJ;EQ%EXai

w
He

W^

26
)

Heibges ignores the form

of all

^_^

the most typical

Greek

clausulae.

regards the form

^^ ^^

86
as

a by-form of

w w

without any ground.

The

negative forms are nearly these.

Negative clausula
tovg dv&Qcbjiovg
decor egya xafohai
Statistical

in

Chariton.
Thucydides.
Chariton.
3*1

183

> w
seems
to

79

2*0

evidence

prove that Chariton

all

shows
the

the strongest metrical tendencies of

the authors

as yet investigated

by

us.

Therefore the investigation of

in the

frequency of the different forms

of urgent necessity.

sentence re-

mains a question

The same work

has

as

that

for

Chariton by Heibges,

been done

for

Libanius by Heitmann.

The same

formae principales have been distinguished here:

Positive
(?)

Clausula of Libanius.
Thucydides.
Libanius.

142

161

'principal'

ww
in this

94
2-6
3-7

76
5-8 5-5

w_w_
By comparing
forms

way

the frequency of these

in Libanius with that in

Thucydides, the

preference for these forms in the former appears to be

quite
that

hypothetical.

We

are

even

justified

in to

supposing
Libanius,

the

form

ww
w

was

indifferent

whereas

was avoided.

Of course Heitmann

87

combines

w ^
It

with

w.

So he

is

induced to

give a comparison of frequency-figures for different authors,

which comparison
the least sense.

is

given
is

in

such a form that

it

has not

rightly

doubted by Maas whether

all.

any metrical tendency

form which points
to

exists here at
it

Nearly the only

is

which shows a very

low

figure.

Negative

(?)

clausula of Libanius.
Thucydides.
Libanius.

79
20*4

77
116
statistical

in

Here as well as
investigation
of
his

Chariton only a minute

sentence-metre can give convincing

typical

evidence.

In

any case the

Greek form:
Libanius.

Thucydides.

www^
is

52

28

not favoured.

Some
certain
that of

3^ears

ago

pointed out that there exists a

of

agreement between the metre

Thucydides and

Herodes and Lesbonax.

Clausula of

Lesbonax
(236 cases)

Herodes
(297 cases)

Thucydides
(2000 cases)

114

161

142

106
20-8

84
17 5

93
183

30

20

26

88
Besides these figures correspond strikingly with those
of

Thucydides'

sentence-metre

and with those

of

our

theoretical calculations.
Lesb.

Her. Thuc. claus. Thuc. sent.-m. Theoret.

161

calc.

1T4

142
183

144

142

This
of

208

175

204
in

183

fact constitutes

an important argument
in

favour

the

hypothesis
if

that

Thucydides there

exist

only

slight,

any,

metrical

tendencies.

On

account of

my
5):

researches on this point Munscher says (Bursian Bd. 70, 1 91

'Bestatigt

sich die Ahnlichkeit

von Thukydides einer

so

und

Lesbonax-Herodes

andererseits,
in

ware das

ein wichtiger

Beweis

fiir

den auch

Rhythmus bewussten
altattische Literatur\
put,

Anschluss dieser Archaisten an die

The
the

question should,
of

however, be
tendencies

whether

in

absence

metrical

we

are allowed to

speak of a conscious imitation. Indeed, where a preference

exists for exactly the

same forms
two
facts.

in different authors,
is

we

are naturally justified in supposing that there

nection between the

a con-

Where, however, these

is

preferences do not

exist,

a conscious imitation as

meant

by Munscher, becomes
Before

quite hypothetical.

we

continue,

will discuss with

you another
it.

methodological question
In

probably

we

can solve
in

Greek prose we have investigated

different

89
authors

whether there

was a preference
that

for

arbitrary

combinations of long and short syllables.

In

doing so
double

we have

not

postulated

the
all

cretic

or the

trochee should be the basis of

prose-metre.

We have
is

not even postulated that in prose-metre there

thing

such a

we

are

used

to

call

metrical

feet.

We

have

only and exclusively asked this question: what combinations of short

in

and long syllables are used by the authors

often

question,

how
this

are they used, and finally: what

has

caused

usage?

These causes, we have

seen,

were twofold.
Firstly

the factor

which might be

called the depenof, i.e.

dence on the material which the author makes use

the

words

of
is

the
at

language.

Where
of
this

only this factor of

dependence
part.

work, chance plays a very important

the
side
factor,

Where, by

other factors

are at

work

as well,

we

have

tried to

determine and to

analyse them as accurately as possible.

in different

We
of

found that

Greek authors there was indeed a preference

series.

for

particular

With

the

help

some simple
in

mathematical formulae

we have
is

succeeded

determining

how

great the chance

that the ratios

found inductively

are a consequence of the dependence on the material only.

That gave us a twofold advantage.

Firstly the probability that our results should not
correct,

be

was reduced

to a

minimum.

90

A
a

second important result was

to

this

that

we have now
is

means

determine

how

great

the

chance

that

in different texts different metrical

tendencies are at work.

This

may be

of

importance as a criterion of authenticity.

the metre of

Of course an agreement between

two

texts

may be due

to imitation.

striking agreement,

however,

such as between the Anecdota of Procopius and his other

works cannot possibly be ascribed

the

to

this factor.

On
ad

other hand:
is

if

we

should have determined that the

that in the Consolatio

chance

one million times greater

Apollonium quite
than
in
all

different metrical tendencies are at

to

work
this

works which can be ascribed

it

Plutarch

with any certainty,

has not yet been proved that

of

work should not be

Plutarch.

But when

it

is

consi-

dered that these metrical tendencies, looked upon from a

psychological standpoint,
the
of

belong for the greater part

function,

to

domain

of

the

secondary

the
will

importance

metre as a criterion of authenticity

be readily

understood.
I

point out

emphatically that in our investigation of

Greek prose-metre we have not postulated anything.

have put a question, and

We
:

we have answered

it

as follows

there exists a preference for special metrical forms.

An

other question

is:

is

there within these metrical forms a

preference for a special divison of words, for a special

typology ?

91

An

investigation

made by me, has given only

sense that there

negative
is

results, or rather positive in that

every

reason to assume that the Greek clausula has probably

not

had a preference

for a special typology.

*)

That

in

Demosthenes the avoidance

bles

of long series of short sylla-

was more
ones

difficult

where a

single long

word

consists

of short
this

(ex xov jzagelrj^vdoTog %qovov) is clear in itself

symptom ma}/ perhaps been

classed under the idea

of typology.

But the most important result of this investiis

gation into the clausula-typology

in

this: that at

any

rate
feet,
if

most Greek authors metre

is

(I

do not say metrical

or verse-feet)
existing
at
all,

primary, typology, on the other hand,

at

any

rate secondary.
of
is

By means
for
feet

of this

investigation the
interesting
it

method
be,

Novotny, however objective and

may

condemned

Greek prose.
and especially
is

His assertion that quite wrongly metrical

the cretic
rect.

have been postulated as primary,

in

quite cor-

His assertion that

is

Latin and Greek the

words

are metrical units,

on the other hand

just as apodictic

and wrong as the assertion

of his adversaries.

And where
in
first

we
of

can show that

it

in
is

Greek prose typology remains

necessary to investigate metre
In a disguised

the back-ground,
all,

and typology

last of

all.

form

we

have here again the old method

of investigation of Bor-

See

table.

92
necque, and
I

hope

that

Novotny

will not diminish the

value of his results by applying an objectionable method

of research.

There
though
here
is
it

are, indeed,

some
in

differences in word-division,

should be borne

mind

that the material used

very small. The general correspondence, however,

figures,

between the
and
for for

as e.g. for

w w
it

in

Thucydides

in Plutarch, clearly proves, that

is

not the preference

one type
avxov

w w, not Xeyovoiv ^n, or

of
,

e.g.

the preference only

*>'

for exteyovntv,

^,
vp*

which caused the great difference between the

14*2

figure

%
29-1

for the

group

and

for the

w w as a whole same group w ^

it

in

Thuc3^dides,

as a whole in

Plutarch.

All 'types' contributed to

in

nearly the same,

perhaps quite the same degree.

the group
is
all,

In Plutarch, at all events,

^w
the

is

felt

as a metrical unity.
if

The metre
at

primary,
is

typology,
It

existing in
that

Greek prose

secondary.

may be

the

avoidance of a

final

monosyllable should be classed under the idea of 'typoIt

logy'.

may be

that in

some other Greek authors

there

exists

a preference for certain types.

Perhaps even our

figures, although derived

from a vety small material (not

such differences.

yet 200
errors,

cases),

and therefore exhibiting large probable

to

perhaps even these figures point

In general,

however, the

ratios
,

agree so strikingly, showing

a culminating point for

w w,

and a gradual decline

93

from

this

culminating point on both sides, that metrical

tendencies cannot but be regarded as the basis of prose-

rhythm here.
Besides
question?

how

then does

Novotny consider

this
is

He

thinks that the end-rhythm of the period

artist

formed by the
especially the

by means
the

of

changing the word-order,

order of the
to
last.

last

word
is

of the

phrase and
7

the

word next

This

only parti} correct.

then
is this

Suppose, however, that he

of

is right,

why

order

words changed?
syllables,

Only
as
is

to get a certain

order of long
cannot exist
Besides:
:

and short
here.

an

accent-clausula

But what else

this

than metrel

we
in

have seen
Plutarch
get
it

what the length

is

of the clausula really is

w w,
is,

that

means: the author

is

tries to

w w;
get a

whatever precedes,

indifferent
of the

to him.

The word-order
to

however, only one

means

certain

clausula:

words.

For and

vincla
is

vine la, for this

solvunt,
if

we may add to it: choice of resolvunt we cannot substitute resolvunt bad clausula. But we can write vincla
it

this is done,

is

done

to avoid the metrical

to

forms

^^ w
period,

and

w, and

get the metre

That the clausula should be compared with the

of

rest

the

has

already been
in

suggested by Thumb.

The

really

new element

Novotny's 'Neue Methode der

Klauselforschung' has no value, as he tears asunder groups

94

which metrically belong together, and as he does not

to

try
his

determine the length of the clausula.

Does not

opinion remind us of the words of the French scholar

who

said:
la

'On

dit

qu'un texte est 6crit en prose melrique

lorsque

forme
la

du

dernier

mot de chaque phrase

le

determine

forme metrique des mots qui precedent

mot

final'?

As

tried

to

explain,

our

investigations

were an
is

attempt to ascertain, whether

in ancient

authors there

indeed a preference for certain combinations of long and

short
syllables.

That such a preference

all

really exists,

is

generally recognised nowadays. Not

scholars, however,

agree to assume

it

as

a basis

of prose-rhythm.

Some
^l
is

think that a metrical form has only then rhythmical meaning,

if

it

is

repeated.

So,

according to them,
if

only then a metrical clausula,

in his

it

is

repeated.

Zander,

famous Eurhythmia

(I,

p. 207) says:

'Nam
membri
redditur

quantum

quoque

loco

a
aut

postrema
alteri

parte

aut

geminatur continenter,

clausulae
clausulae:
sit

congruenter,
sit

tantum

habendum

est

sed ut

rhythmus

aliquis, utique

necesse est

aliqua

terminatio'.
If

these assertions are true,

we

cannot see

why

there

should be a preference for certain metrical forms. Besides,

reiteration

of

certain

metrical

forms

must occur more

95
frequently

than

we

should expect,

i.e.

more
in

frequently

than non-rhythmical prose.

This has not

any respect

been proved by Zander.

texts

He

only gives an analysis of

which he regards as

metrical,
is

and marks the

of mistakes.

reiter-

ation of forms.

This analysis

full

Firstly,
syl-

he assumes
lables can

in

a most arbitrary

way

that

two short

be substituted for one long one.

Secondly, he

assumes

that
.

w w

is

to

be regarded as identical with

to

Each

of these

assumptions ought
in

be sepa-

rately
tition

proved.
of

Of course
those

every text a frequent repestated.

some forms can be

striking
in

This repetition

will

be most
frequent

authors

who make
are,

more

use
e.g.

of

certain

forms.
in

These preferred forms,

of

such

as

^w
is

Demosthenes
to

course,

often repeated.
repetitions

Zander ought

have proved that these

occur more frequently than

the
repetition
of

we

should expect.
forms'.

'Rhythm

metrical

But

where
which

is

the repetition of the form >

is

^ in Plutarch,
Are we
to

form
|

so

common

there?

scan

w? We can, however, prove that w ^ w ^. nearly as much favoured as no connection between the preference for w
w
is

^
there

Is

itself

and 'rhythm'?
It

is

only an hypothesis.
is
its

But a dangerous one.

is

For what
of

basis?

It

based upon the assertions

it,

the ancients themselves about

assertions

which are

96

supposed
tition.

to

teach us that rhythm

is

identical with repe-

Leaving alone whether the passages which seem

to

support Zander's view have been correctly interpreted

by him

this argumentation
of

is

methodologically incorrect.

The

texts

the

ancients discussing these questions are

mostly

vague,

contradictory,

wrong.

You need
by

only

remember what
cies

a bad insight they give into the tendento

light
statistical in-

which have been brought

vestigations.

Anyone who,
in

as Blass, Bornecque, Novotny,

and

others,

investigating

ancient

prose-rhythm bases

himself on a doubtful hypothesis, or on ancient texts, can

never arrive

at conclusive results:

he has

built

upon sand.

I
I

am

sure

now

that

3'ou

understand
I

me

quite well.

have never contended, nor

shall

ever contend that the

lies

quintessence of ancient prose-rhythm

sively
in

only and exclu-

metre.

There

is

a rhythm of thoughts, and a

rhythm
This

in the

agreement between thought and language.

consists
in

rhythm

harmony which
intuition,

is

generally

only recognisable by means of

and which cannot

is

be

grasped by

statistics.

think,

this

the

deepest

essence of prose-rhythm.

'There
separable,

is

rhythm

of thought distinguishable,

if

not

from the rhythm

purely

of language, controlling

and

supplementing the

phonetic

rhythm.

In

poetry,

phonetic rhythm often overrides thought rhythm. In prose,

97
phonetic rhythm
rhythm'.
is,

as

a whole, subordinate to thought

These words

quote from Lipsky's most interesting

characteristic of prose

work
style'

'Rhythm as a distinguishing
(p.

3 and

4).

If

this assertion is

not objected

to,

we

also
is

know

that

there can really exist a prose-metre

which

not solely based upon the reiteration of the

Further,
if

same

metres.

this

is

recognised,

we

understand
certain

that a peculiar

harmony may be reached by using

forms
form.

at the

end

of the kola, without repeating the

its

same
in

An

interpunction does not lose

of

importance for

the rhythm
the

the phrase, the

if

it

is

not repeated.

And
It

same way

use of a favourite clausula suggests

the idea of the end of a period, or of a kolon.

marks

the end of a thought, and in this

way

it

unites thought

and form.

This

is

the quintesssence of the clausula.

SIXTH LECTURE.
To
origin,
task.

write
its

the

history
its

of

antique
is

prose-rhythm,

its

evolution,

decline,

an extremely

difficult

Hardly any special point has yet been treated ex-

haustively or even satisfactorily, and most pioneer-work has

yet to be
into
this

done.
of

And
own

yet the

more we work ourselves

the

field

investigation,

more

sceptic

we

be-

come

as to our

conclusions.

Let us

first

give a short

survey of the history of the method of research.

Immediately
in
felt

after the discovery of rhythmical

cadences

antique prose, eminent scholars, such as Krumbacher,

the

necessity

of

putting the

statements to the

test.
felt

This meant that they tried to convince those

who

somewhat
different

sceptical

as to the existence of such cadences

by showing
sentence.
for

that there existed a great difference

between
of the

authors in their treatment of the

close

The

first

most praiseworthy step was taken

clausula.
It

the

Greek accentual
(1)

was
and

investigated in
(2)

great outlines

how many

'good'

how many
quite conto

'bad' clausulae occurred in different authors,

and then the

results

were compared.
and
in
this

These

results

were

vincing,

connection

Litzica

deserves

be

particularly mentioned.

The importance

of

these investigations

was

entirely

99
misunderstood by

W. Meyer

himself;
to

it

seems

that this

eminent scholar was not quite able

discovery in
detail.

work

out his

own

clausula

The second important step for the Greek accentual was taken much later. Instead of regarding the 'good' clausulae as one unity, Dewing investigates sepathe

rately

frequency

of

the

different

possible
it

forms

w w,

cJ>c/)J3,

wc/5coc/>, etc.

In this

way
in

is

possible to

compare the frequency

texts.
I

of the

same forms
and
that

non-rhythmical

am

sorry that these

valuable researches were

not extended to typology,

to

Dewing

did not try

determine the degree of preference which exists for

the different forms.

as

Consequently he regards some forms

at
all,

favoured

which are not favoured

been made
of the

and other

forms as avoided which are not avoided

at

all.

Very
examples
clausula,

little

use

has

of

these excellent
quantitative
Zielinski,

for

the investigation
for

Greek

and

the Latin

clausula.

Norden,

Zander,

Bornecque,

went on constructing

their

a-priori

schemes, seemingly based upon inductive material, but in

reality

working with the most objectionable deduction.

excellent exceptions can be mentioned.
of

Only a few
are
the

These

remarks

Jordan,

Maas,

Thumb, following
convinced

the

researches of

Thumb, Novotny. Marbe for modern

is

prose-rhythm, most acutely remarks that he

not yet

by the

results

of

the

clausula-investigations

100
before the rhythm of the whole sentence has been carefully

studied.

It

can hardly be understood

why

critics

did not see the fundamental importance of these remarks.

As
and

far

as

know, only Novotny appreciates

their value

sufficiently,

though his a-priori opinions about typology,

of

his

somewhat vague conception

prose-rhythm

in

general have prevented him from

of them.

making the

right use

In

modern

literature

on the Latin clausula we

may

note a controversy which has

of

drawn
of

the general attention

scholars.
Zielinski.

This

is

the controversy between Bornecque

this

and

The importance
If

controversy has

been greatly overestimated.

scholars had

been con-

vinced from the beginning that metrical texts ought to be

compared with non-metrical

texts,

and

that only in this

way

reliable results can

be obtained,
facts

this

polemic would

never have arisen.

collected

The simple
of

would have been

without prejudice, tendencies would have been

stated

and on the ground

such data, the facts could

have been explained.

Here
rhythmical

and there
prose

interesting

comparisons with non-

are those of
to

may be Jordan, who

noted.

The most

interesting

extended his comparison even

regret
that he did not

typology.

We
it

highly

accom-

plish the

whole work according

to this

method.

Finally,

should never been forgotten that the reviews

101
of

so

eminent

critics

as Kroll and

Laurand cannot but

have exercised a greet influence on recent reseaches.

In

modern

investigations

where students are

and

in a posi-

tion to study antique statements

to collect material

them-

selves,

we

notice that this collecting of material has often

to believe that the correct-

been neglected. Thus some seem

ness of the theory of Zander depends upon the correct

interpretation of
of

some

antique statements about the essence

of

prose-rhythm, instead

depending upon the

results

of methodologically correct statistical researches.

On

the
is

quantity

of

the

final

syllable

in

the clausula there

polemic which, curious enough, ignores Josephy's valuable

statistics altogether.

Leo

*)

explains the avoidance of long

final

words

in

the

hexameter,

an avoidance

occurring

already in Cicero,

by means

of

antique rhetoric theory

(which

is

found

in Quintilian)

about the prosaic clausula,

static-

without trying to control this theory by means of

tics.

Comparative

researches will

show you

later

on

that

these long final

words are not avoided

Statistical

in the clausula

by

Cicero himself.

researches are often regarded

as something inferior, at any rate negligible.

In
reality

there

are only two

ways

to inquire into

antique prose-rhythm.

The one

is to

accept antique theory

about

it.

The

other

is

to collect statistic material.

*)

De

Statii Silvis.

102

The
sula
is

value of antique passages bearing on the clauof

course very relative.

Metrical tendencies are

partly unconscious, partly conscious.

As
i.e.

far as

they are

unconscious, or rather subconscious,

belonging to the

secondary

function

of

the

mind, the statements bearing

far as they are conscious,
it

on them are unreliable.

is

As

doubtful whether the artist

was

able to formulate them

himself.

The

facts

show us

that this has only exceptionally

been the

case.

Especially

the

more

subtile

problems,

such as the question of the quantity of the

last syllable,

typology, length of the clausula have hardly been formulated by them.

What
long
facts.

has been said by Quintilian on the

avoidance

of

final
I

words

in

the clausula does not

illustrate a correcter

agree with the

think,

we may

insight into the value of the antique statements

by changing

the

title

of

the

work

of

Wuest: 'De clausula rhetorica

in orationibus secutus
sit'

quae praecepit Cicero quatenus

into

'Quomodo clausulam rhetoricam

ipse

construxit

Cicero,
of

quatenus

concipere potuerit'.

Aesthetic

criteria

prose-metre belong to the secondary function.

Everywhere, however, there

terial

is

a huge mass of manot

first investigate,

which can be investigated.

Why

and then correct the ancients with the help

thus obtained?
In

of the results

some

respects

antique

theory

may become

of

importance.

This importance

may be

illustrated as follows.

103

According
is

to

our imperfect sense for metre, which

hardly anything more than substituting accentual rhythm

for quantitative metre,

we

are inclined to regard the form

as a

more pure and

original

form than

Likewise

in this

respect

we

prefer

to

The
especially
that
in

study, however, of

Greek verse-technique, more

Wilamowitz,

the

researches
tragic

of

have proved
is
it

the

Greek
form

trimeter

'resolution'

not a

symptom
to

of decline or

even
of

of evolution, but that

belongs

the

original

the

verse.
to

Now,
a

this fact

may
in

be

mentioned here as analogous

phenomenon

Greek and Latin prose-metre.

are nearly always favoured
cipal'

In the clausulae 'resolutions'

more strongly than

the 'prin-

forms.

In Philo, e.g.,
that
fact

1*9(4-),

whereas

of

w^^ w the form w

hesitate
of

has a quotient

reaches

only

*6(-}-).

This

does not prove anything of course.

regarding

We

may now, however,

'principal'

w
I

as a real

form

^^^ w.

think, in

this respect, the investigations of Zielinski

and others have

104

given

wrong bend

to

modern
I

thought.

There

is

no

argument whatsoever, as

far as

see, to consider w

w
Let

as the original form of a clausula which might also

the form

assume

^^w w.
cited
in

On

the contrary, classical evidence

of

may be
suffice
to

favour
in
this

the

opposite view.

it

quote

connection Laurand

(Etudes p.

187):

'Cependant un examen plus

attentif
dit

mene,
que
le

je crois,

a une conclusion differente: Ciceron

6gal au cretique,
c'est

p6on

est

a dire que

le
il

peon
dit

et le cretique

ont tous

deux cinq temps, mais

mis a
la place

ne
et

jamais que
le

le

p6on

est

du cretique

pour

remplacer.
totalement

Bien plus,
etrangere,

cette
ainsi

idee de la dissolution qu'a tous

ils

lui est

les

autres

thoriciens anciens

de

la

clausule;

indiquent un certain

nombre de pieds
la

ou de combinaisons metriques que Porateur peut employer,

mais jamais
Pautre,
ils

ne disent que Pun tienne

place de

comme

le

tribraque tient la place de Piambe dans

un

trimetre'.

Of course
thing quite
(

this

holds
is

good even more

for the so-

called substitution,

which

very often regarded as some-

common

(Norden).

plus forte raison la substitution leur est inconnue

la

ils

ne disent jamais que

syllabe

breve puisse etre

remplacee par une longue.

Cette substitution est

meme

contraire au principe sur lequel est fondee toute la th^orie

105

de Ciceron:
caractere

la difference

des durees.
clausules

Ciceron trouve un
qu'elles

different

aux

suivant

sont

formees de spondees ou qu'elle contiennent des syllabes

breves: une serie de longues donne, d'apres
lui,

plus de

lenteur et de gravite a la phrase; aussi, dans sa pensee,

la

clausule

dispondai'que
(

^)
loin

est tout a fait diffe1'

rente

du dichor^e
finales

w w)
my

d'en etre

equivalent.

Ces deux

ne devraient jamais etre confondues dans

les statistiques'.
I

can formulate

opinion about the value of ancient

theory by repeating what Leo has said of the Alexandrine

authors on the theory of verse-metre (Neue Jahrb. IX,
1902. p. 161):
'.
.
.

dass

man

nicht

von ihren Doktrinen ausgehend

observieren,

sondern an der observation ihre Theorieen

messen

soil'.

A
method
method

few words ought

of

to

be said about the second

i.

investigating

ancient prose-rhythm,

e.

the

of using carefully collected statistical material.

Some modern
as
is

clausula-investigators

seem

to

be

of

opinion that figures should be excluded from their papers

much

as possible.

Firstly,

the collecting of statistics

a dry and
of

somewhat
statistics

inferior

work.

Secondly a vast

collection
literary

obscures the right sense for the

and aesthetic problems.

Whoever

'calculates' percriticises

forms a useful work, whoever reads aloud and

106
aesthetically

does a more superior work.

Aesthetic spe-

culations on general rhythmical problems are generally more

appreciated than the careful investigation of the

I

facts.

think,
is

the

first

of

these

objections

to

statistical

researches

based upon a serious misunderstanding of

the true nature of clausula-research.

The

science of proseIt

metre

is

or ought to be an exact science.

It

really be-

longs to psychology.

requires psychological methods

and the exactness

despises

of psychological researches.

Whoever

experimental

work should

not occupy himself

with psychological sciences, nor investigate the rhythm of

ancient prose.

As

to

the second

objection

we would

say

that,

if

figures really obscure the

to illustrate, this
sity of

problems which they are meant

in

does not

any way exclude the necesif

using them.

On
is

the contrary,

they really obscure

these problems, this

due either

to the writer

who

uses

them were they need not necessarily be used, or

reader

to the

who

has not yet learnt to read

to

statistics.

There seems

be a vague notion that

illustrate

we

ought

to

use some figures (not too many) to

our opinions.

Many

scholars
I

make
think

a kind of apology for using these

it

percentages.

better to irritate an uninterested

reader by using them than to offer interesting assertions

which

will

prove

to

be wrong because

statistics

have been

ignored.

107

The
tive

neglect of statistics and the neglect of comparaof research has

methods
the
first

caused the inexact formulating

of

discoveries

by the authors (Bouvy, Meyer,

it

Blass)

themselves.

Further

has caused the

frequent

doubt as to their correctness, so that

not accept their conclusions.
identification of
It

many

students could

has caused the

common
it

frequency and favouriteness.

of discussions

Then

has

produced a large number

and controversies

about problems which could have easily been settled by

means

of statistics exclusively.

Finally

it

has caused the

writing of dissertations such as the one on the metre of

Libanius,

where there

is

perhaps no metre

at

all.

In discussing the origin of the clausula, the difference

between prose-rhythm and prose-metre

clearly appears.
of

We

must sharply distinguish between the origin

Greek

prose-rhythm and that of Greek prose-metre. The question

regarding the origin of Greek prose-rhythm
with that regarding the origin of Greek prose.
different
is

identical

An

entirely

problem

is

that

of

the

origin

of

Greek prose-

metre,

i.e.

the introduction of metre into

of the

Greek prose.

The mania
chus
as the

Greeks

to ascribe the invention of

everything to a definite
inventor

person has stamped Thrasymaof

Greek prose-metre.
the
first

Perhaps

Thrasymachus was

really

to

use metre in his

108
prose.
entitle

The few fragments which have come down

us at
all

to

us

events to say that he tried to produce

rhythmical effects by means of effective metres. Aristotle

as

the main

authority on this point confirms

this.

All

passages relating to Thrasymachus agree on

Besides, the fourth paeon, the most typical
is

this point.

Greek

clausula,

mentioned as the characteristic form

of his sentence-

endings.

That

his

speeches

in

the

first

book

of

the

Republic of Plato are not metrically written proves just

as
little

as

the rhythmical

speeches

in

Ammianus Mar-

cellinus can
If

prove that they should have been rhythmical.

these speeches of

Thrasymachus should have occurred

If

in Virgil,

they would consist of hexameters.

in

they should

have occurred

Plato's

later

works,

e.g. in the

Laws,

they would no doubt have taken the form of Plato's later

metre.

On

such points an agreement between form and

matter

cannot

be

maintained.

Thrasymachus

tries

to

produce a correspondence between rhythmical unit and

thought
verse,
unit,

phenomenon obvious everywhere

he
lets the

in

Greek

more
and

especially
this
is

end

of both coincide.
tried

Besides,
to

really

more important, he has

support rhythm by means of metre.

The combination

of

both tendencies has created the clausula.

As soon
the

as

metre had been introduced into prose,

able to take different forms.
of

evolution

was

Firstly

one could make use

metre to imitate verses, as De-

109

mosthenes
Phaedrus)

imitates

the

hexameter,
Further

and

Plato

(in

his
to

logaoedic

metra.

one could try

avoid monotony and create a melodious and fluent rhythm.

Thus Demosthenes avoids

avoids
,

^^^,
,

wwww

etc.

Plutarch

etc.,

on the other

hand he

favours the fluent iambic-trochaic metres. Again one could

try to

mark

the close of a period or of a colon

by using

a conspicuous metrical clausula.

The

origin of this last

tendency coincides with the origin of the carefully and,

so to say, architectonically constructed sentence.
I

cannot

see

why

the clausula should be

more

intimately connected

with the commatic style than with the periodic: Isocrates

clearly

shows
its

favourite forms.

*)

From

very beginning

this last

tendenc} strongly
7

avoids the use of those clausulae which are

Greek verse
forms which

w^< w and w
Nor
is it

common in w^)> and favours many

in

in

verse are found hardly anywhere.

entirely

These three tendencies have developed

different

manners.

always possible

to distinguish

sharply between them.

In

the evolution of the

stages.

Greek clausula we can

first

dis-

tinguish three
cal

The
stage

we

shall call the classicall

stage.

The
the

second

we

shall

the

Hellenistic

stage.

In
it

third

metre has

disappeared:

we

shall call
*)

the rhythmical stage.

p. 32 sqq.

See Wilamowitz, Asianismus und Atticismus,

110
In

the

first

stage the individuality of the author

is

is

prominent.

There

hardly any agreement between the

of

metre

of

Thrasymachus,
Demosthenes.
metre.
In

Thucydides,
it

of
is

Plato,

of

Isocrates, of

Typical for
his
first

the developis

ment

of

Plato's

works there

no

metre whatsoever, but

later

on metrical tendencies

arise,

and become gradually stronger.

In

this

connection

we must
It

just

point

to

a widethat

spread

misunderstanding.
in
its

is

generally

believed

prose-metre
of forms.

first

stage

shows an impoverishment
p.

Thus Laurand says (Etudes

effet,

189):
les

'En
clausules
primitifs.

Phistoire

ne semble pas indiquer que

diverses

doivent leur origine a quelques types

contraire, autant

Tout au

que nous pouvons en

et

juger,

les

formes ont 6te d'abord multiples

progressivement.

se

sont

renstreintes

Les Grecs du temps de

Ciceron en employaient de moins variees que Demosthene,

a

Rome

on constate de Ciceron a

Symmaque une

serie

d'appauvrissements successes'.

This opinion

is

due

to the fact that Cicero has

of

always

been the central point

modern

clausula-research, and

indeed from Cicero downwards a gradual 'impoverishment'

is

noticeable.

It

has given birth to

is

many

misunderstandings.
senses.
of
It

The word

e.g.

used

in

different

serves

either to express a decrease in the

number

forms used

111

(indeed, the
small),

number
forms,

of

forms used by Sophronius

is

very

or

it

serves to express a decrease in the

thus
e.g.

number

of

favourite

in

Plato

and Philo many

more

favourite forms are found than in Plutarch.

of

Whatever may be the meaning

assertion
is

the

word, the

wrong.

The former use of many forms are used by an author does

In

The question is somewhat as follows. the word is altogether irrational. How

not concern us.

nearly

all

metrical
in

authors

nearly

all

forms

will

be

found.
of

Only

the

later

rhytmical

writers the
In

number

forms used becomes more limited.

however,
are
to

Procopius of
(8*8%),
in
all

Caesarea,
possibilities
all

among

his

exceptions

be found.

Therefore,

his

prose

possible forms occur, though the

is

number

of favourite

forms
is

very limited.

The only

thing that interests us

the question which forms have been preferred to others.

if

Now,

the

number

of these preferred

forms were really

decreasing in the course of centuries,

we

might speak of
true. It

an impoverishment. This
is

is,

however, only partly

the

not true,

for

instance,

that

Greek

in the

time of

Demosthenes used more

of
(

favourite forms than in the time

Cicero.
*-"
i=i),

Demosthenes
Philo,

has

only
If

one favourite form

however, many.

we

should be inclined

to use the

word impoverishment
obliged
his

in the

former sense,

we

should

be
to

to

acknowledge

that

Demosthenes,

owing

avoidance of series of short syllables, uses

112
in

fact

far

fewer forms than Philo.

Philo,

on the other

hand, shows
nesia.

more
is

favourite forms than Hegesias of

liberal
in

Mag-

Cicero

more

the

use of forms than

Hegesias.

Only

for Latin the rule holds good.

is,

But

why?

Because

in Latin there

with the exception of Livy and

not of evolution, but of decline.

Sallust, only

one single

line,

This
tors.

line consists

almost exclusively of Cicero and his imita-

Here, indeed,

we

can speak of a real impoverishment.

To understand
Greek prose-metre,
the conditions

fully
it

the

position

of
to

the

writers
in

of

will
at

be necessary

keep

mind

which

that time

surrounded them, and

of our age.

which were
intellectual

entirely different
life

from those

The

of that period
it

was not

influenced to such

an extent by reading as

is

nowadays.
it

There were no

printed articles then, from which

follows that the large

quantity of newspapers, periodicals, pamphlets and

more
at the

important
course,

works,

which
then.

we now
The

read as a matter of

were lacking

desire for

knowledge

time of the sophists could only be satisfied by the spoken

word. Hence their interest

the

in puns,

hence the important place

in ancient literature.

dramas and the speeches occupy

Hence
of

the delicate sense (quite incomprehensible for us) for

the quantity of the syllables.

the
rhetoricians.

Hence
the

the important position

Hence

very

important

place

which rhythm and metre occupy

in their literature.

Hero-

113
dotus

and others read

in

their

works

to

the

audience in

Athens and
to

Olympia. Their
it

literature is hardly intended

be read, but

is

certainly intended to

be heard.
is

The rhythm which pervades

and
to bring to the fore the

the

whole sentence

one of the most effective means to express one's thoughts

most important parts

is

of the

argumentation.

This

rhythm

of

course

only partly

dependent upon accent or quantity

of syllables: without of accents a beauti-

any metre or any special arrangement

ful

rhythm can be attained by other means, such as

With-

repetitions, contrasts, climaxes, periodic construction.

out
less

any metre the arrangement

of accents in a

more or

conscious and a more or less schematic

way may

give the impression of a beautiful rhythmical sentence.

Sentence-metre

is

only one of the means to produce the

rhythm

of the period.
of

One
is

the

most simple features

of sentence-metre

the tendency to bring variation in the arrangement of

the syllables.

This intention expresses

viz.

itself

by means

of

a negative tendency,
of long,

a tendency to avoid long series

and long

series of short syllables.

in the orators;

This tendency will be most striking

it

appears indeed
the

in the

speeches of Demosthenes.
etc.

Not

only
him,

forms

www, wwww,
,

are avoided
,

by
to

but also

etc.

In Philo

and Plutarch, however,

series of long syllables.

this

tendency extends only

114

But sentence-rhythm
of the sentence, or at the

is

heard particularly

at the

end

end

of a part of a sentence.

The form
is

w^^
%)
:

is

the typical
Aristotle.

Greek

clausula.
it

It

especially

mentioned by

Of course

is

avoided by Demosthenes as representing a combination

of short syllables (2*6

for the

same reason
In

its

frequency
metrical

culminates

in
it

the

Laws (18-0%).
where

Plato's

development

appears rather

late as a separate part of

the sentence: in the Republic,

w
it

and

are already avoided,

it

is

not yet favoured.

In the metre of Isocrates

it

does not play any part, because

fact

has a short penult, which

criterion of

seems

to

have been
It

to

him a
to

good and bad

it

clausulae.

is

curious

note that in Philo

is

favoured only

when preceded

by a long
ret

syllable: xwv yevofxevwv

forms a favourite cadence,

yevo/Lteva is

avoided.

In Plato a distinct preference for

a long final syllable in this form

is

may be

noted

yevojuevwv
is strictly in is

preferred to yevo/ueva

a preference which
w w.
all
is

accordance with the passage of Aristotle cited above.

Next

to

this

form the most important part

played

by the double trochee:

for
it

noticeable preference

exists

in

nearly

authors, except in Thucydides

and Libanius.
tendency
in

The

existence,

however,

of

any metrical

these authors

highly doubtful, which has

been
that

sufficiently

emphasised by Maas; Munscher remarks

review of the work of Heitmann does

Maas

in

his

115
not allege anything serious against
it

except, that no com-

parison has been

made with
is

non-metrical prose!
facts in the history of

One
by
Plato.

of

the

most remarkable

Greek prose-metre

the avoidance of the form

of
it

w
is

The normal frequency

it

the

form

about

14-2%;
in the
in the

in

the period of the Republic

is

occurs

still

13-9%;

preceding dialogues

just a little

more

frequent,

following rather less frequent (see above p. 60), in

5*7
it

the

Laws about
because

%.

Even

in

Isocrates

it

is

rather

frequent,

has a long penult.

Plato obviously

prefers

ww

to

w
etc.

which

is

a priori intelligible:
fact for

therefore

what Shipley accepts as an established

(Treatment
p.

Latin prose

145),

cannot be said for

Greek prose:
'Cicero
Quintilian

and

all

modern

writers

on

the

subject recognize that the last syllable in the rhythmical

member is As to
that
their

treated as long'.

the preceding

syllables

it

may be remarked
Here and there

quantity

is

mostly indifferent.

w w over against w w but a preference for a preceding cretic: 'w w w can hardly be stated anywhere:
a preference seems to exist for
?

if

it

exists

at

all,

this

preference

is

at

any rate not very

strong.

It

may be

superfluous to emphasize again the

fact that the length of the clausula

b}'

cannot be determined
less

citing

antique statements or

by giving more or

116

probable hypotheses, but only by applying a mere inductive

method

of

comparison.
is

Of almost equal importance

It

the form

w.

is

nearly everywhere

more frequent
in

as a clausula than
it

in the sentence.

Only

Demosthenes

is

not favoured,

and
here,

in Plato,
it

who

takes a rather isolated standpoint also

in the Republic,

is

avoided already
works.
It
is,

more strongly
in Philo

in the later

however, very frequent

and Chariton, two authors who show a curious metrical

agreement on many
points.

On
very
later

the other hand, the form

in

w
we
due
.

w^i

which

is

common
works,
in in

Latin

prose,

is

found only

in

Plato's

the

fragments of Hegesias of Magnesia,

In Plato

and

later

authors.

are able to note a

preference for

ww

or rather for
it

w
to

we
of

do not yet know whether

is

to a preference for

w
this

w
form
In

or for ww w ww in
,

The occurrence
more

Hegesias appears

have

greatly
Cicero.

influenced
the
it

Latin

prose-metre,

especially

peculiar prose-metre of Livy and Sallust

in

we

look for

vain.

Plato
as

prefers

w
A

^w

as

well

he prefers

w w
>

'

to

to

more remarkable form

is

^w ww

which

is

favoured by Plato and Philo as well as by Charito.

It is

117

avoided by Hegesias of Magnesia, and

reason

this

may be

the

why

it

is

avoided

in Latin
is

prose too.

The form
(especially in Philo

avoided nearly everywhere

it

and Chariton): so

cannot be regarded

as an equivalent of

The form
in
is

w w. ww w
is

avoided everywhere; only

it

Thucydides and Demosthenes

is

rather frequent.

It

strongly avoided in Philo and Charito, and even

Its

more
is

strongly in Plato's later years.

normal frequency
is

about 6*5
in the

%, Laws

in the
it

Republic

its

figure

nearly the same,

is

1'3%!

Why

these forms

w^^}

w w, w^

w
I

w,

w
in

wv~"i

and no other forms are the

favoured ones

Greek prose, we cannot exactly determine.

it.

should like to suggest the following hypothesis about

It

may be

that

mainly two factors have decided in

favour of these forms:

(1)

the avoidance of verse-metres the

(2)

tendency

to

use

such

forms

as

produce

either
falling

a sharply

marked interpunction or an agreeable

in

rhythm because these forms

feet.
I

themselves contain

a repetition of metrical

As

to the first point,

need only remind you

of the

avoidance

of

^^^

^^ w,

and

of

w^

hardly occurs as a clausula in

The form Greek verse. As

.

118
to

the

second,

it

is

curious to note that nearly

in

all

these

forms contain a repetition

themselves:

to

w w consists of w and ^^ ^^ ^w consists of ^^ and ^^ w w consists of w and w ^ there a preference for w In w w w may therefore consists of ^;
is
it

and

^-.

Of course these phenomena do not explain

exhaustively.

the fact

Perhaps the problem has been incorrectly

formulated: for
question

who
in

can give a

satisfactor}^

aswer

to the

why

Greek verse these current forms are

generally used, and not entirely different ones?

SEVENTH LECTURE.
Before
discussing
the
relation

between Greek and

few more special

Latin prose-metre,
questions,
viz.

we

shall deal with a

the typology of the clausula, the so-called

i.e.

'syllaba anceps',

the quantity of the final syllable, and

the resolutions and by-forms.

The problem
to

of

typology

has not yet been investiis

gated for Greek prose-metre, nor

my

material sufficient

draw

certain inferences from.

Neither

can

it

be treated without comparing nonit

metrical authors,
the typology of

nor can

be treated separately from

verse.
In ancient verse,

Greek and Latin

existing, nearly

typolog}

7
,

wherever

always tends to preserve

the verse.
In the Latin

the rising or falling

rhythm

of

hexameter

it

avoids word-divisions like these:

In the tragic trimeter

it

avoids

The same
Now,
obvious
in in

effect is

aimed

at

by

avoiding:

Greek

verse, typology (to use this term)

is

many

cases

where any

influence of word-accent

120

cannot be supposed with sufficient ground.

is

Therefore

it

not necessary to assume that typology of the clausula

has been influenced by the dynamic accent of the words

composing
to
it.

it.

Nevertheless

some well-known

facts point

The

preference, for instance, for

in

Procopius and his avoidance of

cannot be explained very well without assuming that a

secondary accent causes

it.

To
in

ascribe,

however, the avoidance

of

Cicero, and of
diaeresis
other.

w
>

^^
it

w
,

to a

mysterious fear of

the

would mean
it

substituting

one problem

for
>-:

an

Why

should occur in the forms

and

ww

an d

why

is

as strongly avoided in

w w,

is

somewhat

strange.

The

true expla-

nation seems to be that in the types

something

of

the

falling

rhythm

of the period has

been

preserved, which suits the end of a period. In Procopius

the form
C/3

V>

t/)

C/)

shows the same

peculiarity.

121

So
in

the

favoured clausulae show a tendency to pre-

serve the falling rhythm of the period, a tendency occurring

poetry

as

well.

The

artist

can try

to

produce the

same

effect

by using

w ^

instead of

w w w.

The avoided forms, however, if also occurring in poetry, show in Latin prose a tendency to avoid exactly those ,^ types which are common in verse. The form i=i,

for instance,

which

is

most common

in the

few hexameteris

endings of Cicero's prose, of Lactantius, and others,

hardly found in any sound Latin hexameter of the postCiceronian period.

in

Livy

this

Though ,^ form
end
of

it

cannot be denied that also

is

w ^
in

the

most common,

it

should be borne in mind that the longest words generally

occur
at

the

a sentence,

and

that verse-ending

and sentence-ending only

some cases

coincide.

In this

manner

the

clausula

ww ^
the
final

occurs in prose most

frequently.

Closely connected with the problem

that

of
I

typology

is

the

quantity

of

syllable.

have already

referred to this
It

more than
this

once.

has always seemed to

insight
into

me

that in order to get a

dis-

clear

problem we should sharply

groups
of

tinguish
(1)

between two

quite different

phenomena:

as to the metre, the quantity of syllable does not

matter.

Whether

trimeter

consists

of
it

three

times
tri-

w w

or of three times

remains a

122
meter.
either
It

is

not true that the

first of

these two feet

of

is

trochee,
In
this

or
case

spondee taking the place

be shortened

trochee.
that

a long syllable

we may

should be obliged to assume

in

some

special

cases.

We

can understand that a

but

final syllable is

lengthened
this

by

pronunciation,

we

cannot

see

how

long

syllable can
(2)

be shortened.
to

as

the

metre the quantity

e.g.
It

of a syllable

does

matter.

The

metre,

in

a tragic trimeter requires a

long

final syllable.

is

really unimaginable that a short

If

syllable could be tolerated here.

this trimeter is

we

cannot but think that the

i.e.

last short syllable takes the

it

place of a long one,

positione.
It

that

is

lengthened, so to say,

is

not true, however, that

it

was
I

quite indifferent

to

the

Greeks, whether in such cases,

(2),

mean

the cases

of

group

the syllable

was

really long or short.

On

the contrary,

where the metre requires a long

to put a
1

syllable,

there

is

often a tendency

long

final syllable (see

is

Havet, Cours elementaire

in the

44).

This tendency

found

Greek and Latin pentameter, and

in

also in the tragic

trimeters ending

a cretic

word
occur

or word-combination:

is

preferred to

w<^.
cases
in

Now,

analogous

the

clausula.

In

wwww;

the metre requires a long

final syllable.

This

123
is

made probable by

the passage of Aristotle,

is

who

speaks

only of
Plato

^^^
is

it

confirmed by
for

statistical

evidence:

shows a preference
quite

>^^^

to

^^^-^.

Kaluscha

wrong

in

saying that Plato generally

prefers a long final syllable to a short one: this his assertion,

results
s-z

from his non-comparative method

he obviously prefers
such
as

of research.

In
in

w
other

w w
^^,

to

forms,

problem

he

prefers

An
lutions

even more

difficult

is

that of the reso-

and by-forms.
have often stated
of
in

We
be said
different

Greek, and the same ma}/

that
in

Latin prose-metre,

the

same author

forms are favoured which show some mutual

This

resemblance.

resemblance

is

mainly based upon

what we are accustomed

I

to call dissolution,

and

substitution.

think

we have even

reason to believe that these

real

two words do not exactly render and express the

state of things.

On

the contrary, as
that

we have

seen before,
syllables

we are not allowed to assert may be somewhere substituted one short syllable may replace
either

two short

for a long one,

nor that
can only

a long one:

it

be said for instance that an author favours as a clausula

the form
:=:.

by

or the form

ww^,

followed
'by-

It

is

even questionable whether the word

124
form'
is

correctly used in this connection:

'principal'
It

it

might wrongly

suggest that the

form

is

more favoured, or
whereas the

more
for

important.

is

stated e.g. that in Philo the quotient

^^-w ^ quotient for w

as

a clausula
sk:

is

1-9 (+),
I

is

only

^^^ ^ ^^^ ^ w w belong together. Why, (quotient 1*6 (+)), and ^ should be the most imhowever, the form w
sufficient

1-6 (+).

think

we have

reason to assume that

portant of these forms,

cannot see.

same
are
cipal'

facts in the

prose of Cicero,

You will note the where some 'by-forms'

of

more strongly favoured than

forms,

the corresponding 'princourse,

which phenomenon,
Zielinski's
I

does not
antique

appear

from

curious
referred

statistics.

To

evidence on this point

you

before.

This incorrect view of

viction

'resolutions', including the con-

that

form containing a series

of

three short

syllables

cannot be a principal form,

of

may have

caused

the

wrong conception

the

form

^^^

and the

many different opinions about Aristotle's The question remains: how can it be
rent forms metrically belong together?
It

fourth paeon.*)

seen that

diffe-

is

more easy

to state that
in

two forms do not belong

Greek prose

together.
is

Nearly everywhere
s^
is

favoured,

sought.

So

it

cannot be asserted

<)

Cf. esp.

Norden, 'Kunstprosa' 2

p. 916/17.

125
that

these

forms metrically belong together, unless

to
cite

we

should
fact that
It

be inclined

in

favour of this opinion the

both forms have been avoided by Plato!

is

more

difficult to

formulate,

when we

are allowed

to

assume

that different forms

do belong together. Analogy

with
it

of

verse-metre
relations.

may be
In

applied

caution
often
If,

to

show
two

such

verse-metre

is

allowed to
indeed,
of

use in a certain foot either

clausula-forms
firstly

or

ww.

contain the

same number

morae,
in

secondly differ only in so far that where

the

one form,

stands in the other

(e.g.

^^ stands ^w^

and

^
the

^), and thirdly both forms are favoured

to

nearly

same
this
,

extent,

there

arises

some
for

probability

that these

forms belong together.

probability.

Other arguments
Plato,
in

may
in

strengthen

In

instance,

w
metre
is

^w

in

<->

^ and

^^>

the

forms with long

a short one.

final syllable

are preferred to those with

of verse-

Too
*)

often,

however, the analogy

cited as a sufficient proof to support

most question-

able hypotheses.

Whoever would
relation
first

like to

consider the problem of the

between Greek and Latin prose-metre, should

distinguish

sharply

between two

entirely

different

*) See Heibges, and his way of grouping together and neglecting the simple form

ww

different

forms

126

The one group seems to begin chronologically with Cicero. The deviations of his successors from prose-metre point all to the same direction: impoverishment of favourite forms. The number of favourite forms becomes smaller; besides the number of
groups of Latin authors.
cases of these forms

becomes

greater,

whereas the excep-

tions gradually disappear: tendencies

become
is

laws.

The

other

group

of
It

authors,

however,

quite independent

from Cicero.

seems

to

be mainly represented by Livy

has
hardly

and

Sallust.

Their metre

any features

in

common

with Cicero.

Whereas Cicero avoids

is

the heroic

clausula, the

same form
is

favoured by Livy and Sallust.

The same
are

true

for

^^^n.
and

Whereas Livy and

Sallust avoid

w ^
to

'*->

^,

these very forms

among

the

most favourite
be no

of Cicero's sentence-endings.
difficulty in finding out the

There seems

origin of Cicero's clausula.

Hegesias of Magnesia shows

all

the

same metre
of

in

nearly

details.

It

is

not altogether

devoid
all

interest

to note that Cicero

has imitated from

the smallest

Greek

authors the rhetorician

exceptions
to
us.
all

who shows
is

number

of

in the

few fragments which have

interesting

come down

Not

less

the fact that

in

Cicero imitates

forms which are favoured

to

Hegesias,

whereas he adds only a few

the

them.

That he omits
form did not

form

^^ww

only

proves that

this

suit the Latin

language.

127
In vain, however,
sible

would we
the

try to determine a pos-

Greek
is

origin

of of

clausula of the other group,

Sallust.
It

that

the

group

Livy and

is

true that

there exist

some correspondence with

are,
this

Plato's later works,

but

the

differences

however,
origin

much

greater.

We
This

cannot

but

regard

as

purely Latin.

purely Latin clausula

may have

arisen under the influence

of the highly refined feeling for the quantity of syllables,

after

the

verse.

Romans had become more familiar with Greek The true origin of this Latin clausula, however,
be found
in Latin literature itself.
It

seems

to

seems

to

be an imitation (perhaps not altogether conscious)

of the

hexameter verse.
half hexameters,

In

Demosthenes we
his

often

meet with

and

prose shows a dactylic character.

That no connection whatsoever with the national Greek

verse-form should exist here,
In the

seems highly improbable.

same way

the metre of Li\y has been influenced

I

by Latin hexameters,

think

more

especially

by

Virgil.

Everybody remembers the beginning

of the praefatio:
. .

'facturusne operae pretium sim

.'
.

The
his

clausula of Livy
of

may have been

but
it

stimulated

by

knowledge

Greek

literature,

really strikes us

that

ww
forms

dactylic

and w ^ ^^ w and

are avoided, whereas the

^-^^

are favoured.

Of course we can speak

of a conscious imitation.

of a certain influence only, not

This influence quite corresponds

128
with the influence of Virgilian words and syntax on his prose.

Therefore

it

seems probable

that

we have

found here a

purely Latin clausula.

Now,
by
In
his

Latin prose-metre has been directly influenced

of

Hegesias

Magnesia
the

or

other

'Asianic'

writers.

few

fragments

following

forms

appear as

favoured ones:

most favoured:

ww w
among

(often

preceded by a

cretic),

rather favoured:

and

ww^-w.

Besides

we

note

others:

^ w^w
is

and

A
pared
the

curious

agreement with

Cicero

found

in the

comparatively high frequency of

with

w^^

w w. An other w w. frequency of ^
It

as comis

striking

feature

can be seen that Hegesias paid minute attention

to

his

metre as perhaps no other Greek writer, so that

his

prose reminds us of the constant monotony of the

Cicero omits the form
its

later rhythmical cadences.

wwww
the forms

only,

which proves

that

absence

in

Latin has to be

explained by the character of the language.

of

To

Hegesias he adds a few others, such as

his

^^ ^.
is

Though
than

number

of favourite

forms

really greater

that

of

Hegesias,
total

their

total

frequency
of

does not

reach

the

sum

of

the

cases

'good'

forms

in

129
Hegesias.
of
It is

clear

from these
to a

facts that the exact

modelling
to a later

Hegesias according
of

few patterns points

has
lost
its

period

evolution,
of

which

freshness.

development

more than one hundred years precedes.

shows a
for

Cicero, however, through far from original,

fresh-

ness

which only by
imitation:

his later

successors degenerates into a

on,

ridiculous
lost,
it

when sense

metre

is

finds a substitute in rhythmical

monotony. Hegesias'

metre shows the S3/mptoms

of decline, Cicero's richness

precedes the gradually increasing impoverishment. Cicero's

influence

seems

to

have entirely ousted the more typical

Latin clausula.

Generally

speaking

we may
It

say that

Greek and
points, also

Latin prose-metre, though agreeing on

many

show many
is

differences.

may be

that the Latin clausula

here and there longer than the longest Greek forms:

e.g.

w www w
Of
course,
heroica,

is

in

vain sought in Greek. In Latin

less

the negative clausula

may be
of

prominent than
is

in

Greek,

where the avoidance

features.

one

of the typical

an exception should be
but though
is

made
avoided

for
in
.

the

clausula

Latin,

Cicero as well as Apulejus tolerate w

Typology plays an important part

Here,

in the clausula in Latin.

some
in

of

the

Greek forms do not appear, more

and

especially

wwww

ww ww.
lively

Nor do we

find

anywhere

Latin the

metrical feeling of Plato,

130

whose

metrical

development begins only

late

and then

gradually shows stronger tendencies.

AH

these questions,

however, have

to

be investigated more closely.

The
Hegesias.

Hellenistic

stage
there
is

seems

to

us

to

begin

with
his

Though

some

difference

between

forms and those of the

distinguishes

later authors, his

monotony sharply
Greek

him from the

appear

classic period.

The main
prose-metre
For, beside

characteristics,
for

however,
first

of

later

the

time

clearly

in Philo.

wwww __ ww ww, w
bably Philostratus.

and

><,

w w, he uses the clausulae ww. With and w

w.

him agree Charito, Flavius Josephus, Polemo, and pro-

Of these forms Plutarch favours only

and

wwww, w w,
In

w
thinks

two

brilliant

works VVilamowitz contended some

years ago that the Atticistic reaction put an end to the

Greek
Asianic

clausula. *)

He

(quoting

Marx)

that

the
into

clausulae

(Hegesias

etc.)

were introduced
dominated by
should

Rome
Latin

at the

times of the Gracchi, and that the whole of

literature

was

uninterruptedly
In

this

Hellenistic

tradition.

Greek,
of

this

have been

otherwise:
*)

by the
Lit.

imitation

the Attici the

manner
p.

of

Die griech.
1900, p. 37

Hermes

des Alt. in 'Kultur des Gegenwarf, (Asianismus u. Atticismus).

103104;

131

Demosthenes and Isocrates would have been

I

restored.

think, this

assumption cannot be maintained. Philo,

Plutarch
metrical
Flavii

and Philostratus cannot be regarded as nonauthors.

It

cannot be asserted that under the

not exist any more.

the

clausula

did

Nor can

it

be

said

that

w w, w
Greek,
form.

and

w
is

w
realty

are the clausulae of the later prose:

more
later

typical

and
In

wwww ww ww a more
others.
his

typical

Greek

fact,

Wilamowitz bases himself

priori

mainly

upon antique statements and upon the a

of

investigations

Norden and
in

But

it

would lead
and the

us

too

far

to

discuss

detail

assertions

inferences

drawn from them.

EIGHTH LECTURE.
I

intend to discuss rhythmical prose in extenso later

It

on.

may be
words the

of

importance, however, to formulate

in

a few

position of the problem.

in

Modern

scholars

have not yet succeeded

explaining this origin from

the other hand there seems

Greek metrical prose.

to

On

be a striking agreement between the Greek rhythmical

clausula and the Latin.

If

we

try

to

establish a certain connection

between

the
first

Greek metrical
have
(1)

clausula

and the rhythmical,

we

shall

to
it

answer three questions.

possible that there exists any relation besyllable in

Is

tween the long

the quantitative clausula and

the accented syllable in the rhythmical cadence? In Latin, the existence of this relation cannot be denied.
tion,

The

rela-

however, between quantity and accent

in Latin

is,

of course,

much more intimate Which are (2)

quantitative
clausula,

than

in

Greek.

the typical forms of the later

Greek

which those

of

the

first

ffyzantine

cadences?
(3)

What
of

is

the typology of the metrical endings,

what
it

that

the

Byzantine?
of

shall

show

later

on

that

is

the typology
the

the Latin clausula which clearly

illustrates

origin of the later prose-rhythm; therefore

133
I

highly regret that

Harmon

hardly investigated the typology

of

Ammianus
this

Marcellinus, an author especially important

on

point.

Such an
of a

investigation

could

only have

been made by means

In
I

comparative method.
hypothesis in a few words,

order to explain

my
of

should like to draw your attention to the fact that at

time
of

the

the

origin

the

rhythmical cadence,

the

Greek language had

for

lost its metrical character.

The sense
~

metre had gone.

Also the difference between

and

'

had disappeared.

The metre
tible

of

ancient verse could be

it

made percepan accentual

to

the
in

ear
the
to

only by changing
in

into

rhythm,

same manner
mutilate
do, in this
r

which we ourselves are

accustomed
recited, as

Homer.
way:
t

Homer,

of course,

was

we

and so

on.

This

means

that

some long

syllables are

accentuated, whereas mainly the short ones are not accentuated. In this

way modern
syllable,

sense for rhythm

thus accented,

is satisfied.

It is

said

that

the

bears the

'ictus'.

This treatment would, no doubt, have greatly astonished

the Greeks themselves.

Now,
of

in

the

schools of the rhetoricians,

the metre

the

ancient orators could not be neglected,

when

the

clausula

was

treated.

Their

way

of

reciting

the quan-

134
titative

clausula

cannot but have

for
c/>

been

like

ours,

i.e.:

substituting

rhythm
m
v>\
c/>

metre.

Thus
v'

treateQ as

co

en

in

s-^

as
c/>

c/)

(s>

ww ww
in
c/>
',

was
""

N-''

as

{/i

tn

in

s *

'

as

(/>

c/>

(/)(/>

c/)

Reading
for
this

classical authors in this

manner was

difficult
call

reason,

that

syllables

which had what we

the ictus,

would be
really

just as well accentuated as the syllables

which had
this this
(I

the grammatical accent.

to

If,

therefore,
other,

clausula

had

be imitated

in

some way or

could only be done by making the two kinds of accent

mean the 'ictus' and the word-accent) coincide. So the Greek accent-clausula has not developed from
Greek
There
quantitative
is

the
it.

clausula, but
line

it

is

an imitation of

no regular

of

development, neither a

gradual transition from the one to the other.

Wilamowitz,

however,

is

wrong

in stating that at the

time of the Flavii

the old clausulae are done with. Yet there

may have been a

when metre and

time

when
slightest

a certain stagnancy begins,

lacking.

rhythm are
the

Besides,

in the accent-clausula not

attention

was paid

to quantity in particular

places, neither

where

in the quantity-clausula the syllables

'ictus'.

were

long, nor

were they had the

All this

would
if

have been the case, as has been said often enough,

the

Greek rhythm had developed from

the

Greek metre.

To

ascribe,

the origin of a

as W. Meyer does (Fragmenta Burana), phenomenon which forms so organic and

135
so important a part in the whole of the Greek genius in
literature,

to

the invention of
in

some

orator or rhetorican,

would be more

accordance with ancient conceptions of

history than with ours.

On
from

the other hand, the arguments for the derivation

Latin

are

not very

convincing.

*)

First of all the

is

argument
Latin,
is

that the use of accent in the clausula

typically

rather weak.

That the Greeks used

a time

this

accent

in the clausula is quite natural in

when accent-hymns
is

arise independently

from Latin. The other argument

the

great agreement between the

Greek and the Latin

as

clausula.

This great agreement, however, does not exist

Firstly
writer,

in reality.

Himerius,

as

far

see the oldest accent-

has only a preference for the form

in

eo

.
.

while

Latin
.

c/>

v>

c/>

dn

...
is
J>

occurs by the side of

())viv)w ...
that in

Secondly there
J>

every reason to suppose

. .

Greek
of

the form

&
.

v>

v>

.,

where

it

occurs,
viz. in

is

by-form

*).

Thirdly in Greek,
>

Pro-

copius of Caesarea, a form

of

& &

c/>

occurs,

no prototype

which
for

is
co

to

be found
c/5

in Latin
in

rhythm. The same holds

in

good
the
one.

J)

c/3

v)

Sophronius: so

both cases

number

of syllables after the last accent is a definite

Fourthly the typology of the Greek accent-clausula

has quite a definite character.

!)

As
18

to the

typology of the

2
)

See Dewing, Origin, p. 00. See De Groot, Prokopius, p.p.

20.

136

form
in

a>

v>

v>

</>

</>

in

.,

it

does not point back, as

it

does form

Latin,

to

a
,

Latin

(or

Greek) typology

of

the

w
with
di

w
eo co

or
.
.

w
of

w Wj

but

to

a connection
is

eo

.,

which connection not a trace

to

be found
of

in Latin.

In

Ammianus

Marcellinus the typology

the

clausula

corresponds

clearly
( J>

with
u>
,

that
c#>

of

the

quantity-clausula; thus nossc confingit

^ w
. ,

has been
clearly

most favoured
points
to

of all types of

dn

<n

<n

which

the typology

^,
co co cA
.
.

w
.,

and

Neither of the form

&

nor of

w, w
dn

ww-.

w
The

or

w
found
in

w
it

special

typology has up to
v,

now been
force

Greek.

The form
shows
is

In

v>

is

lacking in Latin.
i

In Procopius
of this

the typology

<n

typology

to

preserve the

falling

rhythm

also in

the word-division.

So
t
C/3
>

r
C/3

Ui
r

and

C/3

C/3

C/3

C/3

are avoided, just as well as

is

avoided

in

the

tragic

trimeter and

still

more

rigidly

of course

and
and
*)

^^ ,w.

This tendency of avoiding final words of four syllables has nothing to do with the avoidance of final words of five syllables.

137

The

clausula-typology has to be explained not

by a

mysterious fear for the caesura, but by a tendency to

intensify the falling rhythm.

Now,

all this

proves that there

is

a greater agreement
quantita-

between the Greek accent-clausula and the Greek

tive clausula than

between the Greek accent-clausula and the

In
en

Latin accent-clausula.
is
en

Greek the form

. . . :

en

en

en

en

dn

a by-form of
en
c/i c/3

dn

en

en

the

metrical

prototype

of

en

en

does not occur, unless

we want
of

to con-

sider

as

such,

a form which,
dn en en
dn

course,

develops
the

much more

easily into

en

en

In

Greek
fore.

form

w
some
c/5

^ w
dn

does not come

to

the

Quite in agreement with this

as

we
u>

observe that Himerius,

after him,

well as
dn en

of those that
<s>
.
. .

come
is

does not

know
ch
en en

oj

<s>

<s>

en

lacking in Latin, also

(

c/3

vj

the prototype

of

which
is

^^ w^)

is

likewise absent in Latin,

in

and which
others
the
in

highly favoured

later

Greek,

among
In

Philo,

Chariton and

especially Philemon.

Greek

quantitative clausula

typology recedes into the background.

this

In

agreement with

the typology of the

Greek accent-clausula must be

explained independently from Latin and from independent

and very general rhythmical tendencies.

Where
Latin

the

Greek accent-clausula deviates from the

is

and

that

the case in

many

respects

these

deviations correspond with those between the

Greek and

138
Latin quantitative clausula.

So

the

Greek accent-clausula
very imita-

can

in

no case be derived from the Latin.

it

Thus
tion,

becomes
is

clear that,

owing

to this

which
is
In

no development

(in its

more

limited sense),

as
dr>

it

in

Latin,
c/>

only in later accent-writers the forms

c/j
c/>

c/5

w and

v>

v>

c/>

crop up,

forms which both

have

their prototype in the quantitative cadences.

After having given in this wa}

history
in

a brief survey of the

of

of the

Greek
far

clausula,

and

Greek prose-metre
knowledge
of
facts

general,

as

as
it,

our
I

present

enables us to see
the
to

shall

deal in a few

words with

means by which

clausulae and metre can be produced,

results of our

wind up with a few remarks on the

investigations for textual criticism.

The
one
metre
the
sula
of

existence of a certain influence of rhythm

upon
is

word-order

cannot
only

be

denied.

Firstly
to
it

this

influence

the

two

means

arrive
is

at

definite

or

rr^/thm.
to

Secondly

clearly

seen

from

tendency
at

put words especially suiting the clauof

the

end

the sentence:
c

thus in Cicero 'AntoPlato nvd tqojiov'

of
c

nius', in
etc.

Plutarch 'yevojuevwv',
frequently used
it

yev6jueva', in

are

at

the
to

close

the

period or

colon.

Thirdly

is

known

have influenced the order

139
of the

two

last

words

of

the

sentence,

e.g.
c

in

Cicero:
tqojiov
,

widetur esse' instead of esse videtur, in Plato nvd

but not tqotzov
tiva,

and so

on.

The
facts.

problems,

however,

concerning

this

influence

cannot be studied very well, because of two deplorable

Firstly because the
insufficiently

laws
to

of

Greek word-order are

and secondly because

only
the

known
to
us.

us,

laws
less

of

Greek prose-metre and prose-rhythm are

Therefore
it

even

known

may

suffice

to

illustrate its

importance by a few examples.

of

A
of

few years ago Pohlenz dealt with the group

manuscripts of Plutarch

containing a Byzantine recension

some works,

in

which many clausulae have been altered

to suit

Byzantine prose-rhythm.
clear

A
into

example

of this

tendency

is

the alteration of

xal (ptJovovg xal xaxorjfteiag xal /nixQoloyiag xal (pilodrjviag

xal /MXQoXoyiag xal (pdodrjvlag xal xaxoijrJeiag xal qptJovovg.

More
or

often

one

of

the favourite clausulae

w w, or w
I

w has been changed into avoided

an example of each
of these cases.

www

forms.
rfj

shall

give

oipei JzaoaToeyeiv

becomes

JiaQaTQE%Eiv

rfj oyjet,

(www
a>
oo

becomes the avoided form

u>

w w w

beco-

mes

the favourite form

ojiovdaoxixog
del

J>

w w

c/>).

(paiveofiai
del

xal aoxvog xal jigofivjuog becpaivzoftai

comes

OJiovdaoiLxbg

xal

aoxvog xal jiQo&Vfiog

140
(

ww
w
u)
c/>

becomes
u)
u>

the

avoided

form
dn

v>

f/i

becomes the
icbv

favourite form
djnijUf]iov

& & & w

v> )

ovdev
al'oxQwv

anokEiJiEi

aloyowv
(

becomes

xo)v

ovdev

ajuijurjTov

djtohiTiei
,

becomes the
fa-

avoided form
vourite form
In

www
it

^ w ^ ^ ^ becomes the

J>c/>c/>ce/wc/>)'

Plato

may be observed
is

that at the

end of the
zivd,

sentence there
in

a preference for uvd Xoyov to Xoyov

order to get the favourite form

In Plutarch

^^^
yEvofihcov

many

expressions tend to be used at the

close

of

the

sentence,

such

as:

XfyojuEva

(wwww^
?

yoacpojuevcov
((

(www
xciteXijie

jiQog

avrov

)w w),
).

(www

(www

),

),

IJeqikMovs

deo/uerovg

(www

)f

(^--ww^),

oiKOVOjulav

(^--^w

{^^-s^s
In

),

dvvdjUECog

(ww^

fXEyaXocpqoovvi]v

Livy the same phenomenon occurs:

statuit, faciunt,

numerum,
used

attulerint, contulerint, rediit, habuit, facere,

are

to get at the

end the favoured clausula

ww^.
legiones,

The same
erunt,

is

true for redierunt, adgrederentur, habu-

conficiebat,

consequerentur,

populorum,

in deditionem, in

order to get

ww ^.
Romam venisset,
w.
belli,

Likewise Romanis bellum, convenisse,

disceptarent, tollendos curavit, usu

introductus, decer-

nuntur, are used to get the 'good' clausula

In
etc.

Demosthenes,

of

course, Svvajuiv, noXEjuog, 6 nooog,

are preceded by a long syllable, and are followed by

141

consonant,

in

order to

produce

ww
end

of course

these

words are

often found at the

of a sentence or
syllable, in xov TtoXe/uov.

kolon, provided they are preceded

by a long
. . .

order to produce

^^
or

e.g.:

ovx

em

Further the influence of rhythm upon word-order

may

be
of

illustrated

by a negative tendency.
ignoring
neglecting

The
his

later copyists

Procopius,

rhythm, have
x
).

often

changed the word-order

in

of his sentence-endings
I,

Thus
ak'iov

Procopius

(ed.

Haury

p.

12 10

loyov nollov
In this

has
the
that

been modified

into loyov ak'iov nollov.

it

way
seen

rhythm

of

Procopius disappears, and

can be

the choice between different clausulae each of

to

which he was able

choose has been made

in

such a

way

as to suit the rhythm.

The
examples
ojg

imitation of
of

Herodotus by Procopius

offers

many
2
).

the influence of rhythm

upon word-order
xaypvg
(

zaxeog ei% exaoxog

fxev

becomes
ovxoi

(bg eiypv

k u>ib

c/>)j
/ulev

vdjuotoi

xoiovxoioi

%@ea)vxai
(.
.

becomes
.

roiovroig

lyoCbvio "EqovXol to naXaibv vojuois

d)de

o>
in

os
c/)

.)J

erdooovro de

becomes

xal

hd^avro

code

v>

v>

).

With
choice

this question that of the influence of

is

metre upon
influence
to arrive

of words

closely connected.
it

Also

this

exists undoubtedly, as
at

is

one

of the
its

few means
is

a favourite metre.
!)

Further

existence

proved by

2
)

See Cronert [on Haury]. See Braun, Die Nachahmung Herodots durch Prokop.

142

simple

statistics

which show the use

of

of

wg and Su

in

the

Phaedrus between syllables

these figures
is
it

different quantity.
ojg

From
or on

can be inferred that the use of

influenced

by the metre.

The

choice

made between

them has mostly been determined by the choice between

good and bad metre,

and other metres.

in

casu the choice between logaoedic

The importance
following examples.

of this influence

can be seen from the

As

the metre has influenced the use of

etc. in

nomina, forms of the verb,

the hexameter to a great

extent, so the use of the *-forms of the aorist in

is

Demosthenes
}

determined by his tendency

to

avoid

^ww wwww,
and

etc.

He
!)

prefers edcoxajuev to

edojuev, iScoxaze to edoze,

8,

so on.
30, 37

When
the

comparing Isaeus
fact

12 with

Demosthenes

2
)

same

can be seen:

> ' 'w

axoifieozazov tXeyyov Isaeus.

axQifieoxdnjv jiaocbv

Demosthenes.

^^^^
*

avviaxe

ydo

on has been omitted by De-

mosthenes.

'ww

nvkg

k'dog~av

ov xaXrjdrj juaQTVQfjoai.

nveg ov

xaXrjdi] juaQTVofjoai edog~av.

Further the use of fjovfoodm in Plato's later works

becomes more and more

i) 2)

limited

in

favour of

edifoiv'6 )

3
)

Fuhr, Rh. Mus. 57, 1902, p. 426 sqq. Fuhr, Berl. Philol. Woch. 1904, col. 1030 sq. This factor has been neglected by Rodiger, Glotta

8,

1917,

p.

16.

143
(esp.
fiedg
eftelei

^^^^
and so

),

for

ojotieq is

used xaddneQ,
I

for

hoojiEQ

/jL%qi71q,

on.

Other examples

gave

before.

In quotations conscious alterations

can be stated.
to illustrate this to

More examples might

tendency.
1

easily
that

be given
I

sincerely

hope

have been able

show

the desirability or rather the urgent necessity of studying

and investigating the

of

relation

between choice
in a

of

words, order
satisfactory
to

words,

and

prose-metre

much more
I

and exhaustive manner. few

remarks

In conclusion
of

want

make

on the importance

metre for textual

criticism.

As
whether

regards criticism of manuscripts,

different

we

can investigate
differ

manuscripts
text.

of

the

same author

from the authentic

far

These differences may

point, as

as

we

are interested here, into

two

directions: they

may be

rlrythmical deviations or non-rhythmical deviations.

With non-rhythmical

deviations

mean

deviations which
to the

have been introduced without any attention

or the rhythm of the author.

metre
I

With

rhrythmical deviations

mean

deviations which have been introduced in order to

produce a certain rhythm.

peculiar instance of non-rhythmical deviations

is offe-

red by some manuscripts of Procopius of Caesarea. Cronert

thinks that these alterations have been

made
like a

in

order to get

more

'open' rhythm,

i.e.

something

rhythm with a

144
considerable

number

of

unaccented syllables between the

two

last

accented ones.
is

The

clausula: xavxr\

wxrjvzat for

instance
d(jxyV,

changed
on.

into q>xt]vxcu tavtfj] ugxriv

I'xojv

into fyajv

and so

The

correctness

of
if

this
it

remark seems

to

be

liable

to serious doubt.

For

were

correct, there
viz.

would be

here a tendency hitherto unknown,

a tendency to get

many
the

(the

more

the better) unaccented syllables between

last

two accented ones.

Procopius

This tendency has not yet

been found anywhere.

of the

As, however, the principal feature

is

rhythm

of

a tendency to get few unac-

cented
(.
.

syllables
to
.

between
.

the
. .

two

last
.

accented

ones

co

.,

...(oa),cB09,

co co co to

.),

most deviations

from

this rule

may seem

to

produce a more 'open' rhythm.

is

A
Here

more
the
of

interesting case

offered

by
I

the Byzantine

manuscripts of Plutarch's Moralia of which

spoke before.
into the

metre of Plutarch himself

the
later

is

changed

rhythm
struggle

centuries.

Here we observe the

metre of the second
of a later period.

between

the

sense

for

century, and the sense for

rhythm
of

As

regards
of

the

criticism

single

passages,
(1) in

our

knowledge

prose-metre

may be used

order to
(2) in

choose between

readings of different manuscripts,

order to alter the text of our manuscripts.

Of course

the

principal

factors

which have

to

be

taken into account here are those which have nothing to

145

do with prose-metre
general
of the

itself.

These

factors

are e.g. the

value

of

our manuscripts, the general character

Greek language, the syntax of the author in question, and so on. These factors must form the basis of all They must play a predominant part further speculations.
especially

there

where a

certain

reading occurs in

all

manuscripts, or in the archetype of our manuscripts.

In order to determine the importance of prose-metre
for

textual

criticism,

it

is

necessary
it

to

ascertain

what

value

we would
to

ascribe to

in

such cases where

we

have
quite

decide between two equivalent readings of two

manuscripts.
will

equivalent

In

this

case,

i.e.

ceteris

omnibus paribus, one

of

be inclined
is

to decide in favour

the

better

metre.

What
that
it

'better'

metre?

have
i.e.

tried to

show before

is

not the 'best' clausula,

neither the most frequent form at the end of the sentence,

nor the form which

sentence,
tence.

is

most favoured

at the

end

of the

as

compared with the metre

contrary,
it

of the

whole senneed not

at

On

the

is

the form the frequency of

I

which most deviates from non-metrical prose.

dwell any longer upon the different factors which are

work

here:

(1)

the

dependence

of

the author

upon the

very limited choice of words of the language, and upon

syntactical
rules, (2) the
(3)

preference for metrical forms in

the preference for metrical forms

the whole sentence,

at

the end of a sentence only.

The

strongest arguments

146
exist there

where forms which occur

such as
,

rather frequently in

the ordinary language are not or are hardly ever found

in

a certain

text,

e.g.

ww>^ w^^w-j
f
,

etc. in

Demosthenes,

etc. in

Plutarch,
later

ww w
3/ears,

and

*-"

>=*

as
less

clausulae

in

Plato's

and so on.

Much

convincing

may be

the

argumentation where

we have

to decide

between two more

or less favoured forms.

As
have

to

such

questions

many premature
to

conclusions
to

been

made.
of

They can mostly be

rules

traced

an

application
sufficiently
in

metrical

textual criticism without

knowing the
form

rules themselves.

Thus Baehrens

his

treatment of the clausula of Apulejus entirely nethe

glects

so that he rejects rea-

dings such as: acrius contemplantes

The

general merits, however, of his investigations are, of course,

hardly diminished by these errors of less importance.

NINTH LECTURE.
With
to

the philological problems presenting themselves

statistical

us in a
others

treatment of antique prose-rhythm

are
closely

some

of

more mathematical nature

that in

connected.
cases of
in

Suppose

one author on 1 000 syllables 1 00

w
there

have been found, and, on the other hand,

1

an other author on
is

000 syllables 200 cases

to

of the

same

form,
that

any reason

infer

from

this difference

there
texts

exists

a real metrical

difference
to

between the

two
that

which cannot be ascribed

is

chance?

Suppose

there

indeed a
is
it

real

metrical difference between

these

two authors,

correct to measure this difference

by means
to

of a quotient

v^n

2?

Would

it

be possible
greater

find

a figure which might express

is is

how much

the

chance
it

that

real
to

difference

exists in this case

than that

merely due
I

chance?
I

So

far as

know, these problems, which

the most important of
all

think

we

may regard as we have dealt

scholars.

those with which

as yet, have been fully neglected

by recent
towards

The
3'et

consequences

of

this

attitude

methodological questions have occasionally been

shown by

me.

And

the

same or nearly
in other

the

same questions

have presented themselves

branches of science,

148

such

as

e.g.

in

astronomy,
in
all

in

biology,

in

differential

psychology.

Now,

these

cases

they

have been

settled in a rather satisfactory

way.

We

shall often start

from the results obtained there as a basis for the following researches.
1.

In

classifying
in

the different metrical forms which

I

may

occur
of

antique prose,

often

referred

you

to a

scheme

the

128

combinations

of eight syllables that

may be
the
fact

either long or short, the quantity of the final one,

It

however, being neglected.

that
is

is

superfluous to emphasize
of the
is

in

doing

so,

no length whatsoever

clausula

postulated.

To

investigate eight syllables

not

the

same
fix

as to contend that the clausula comprises

eight syllables.

On
real

the contrary, this

is

the only possible

way

to

the

length of the clausula.

Where we

are able to state in the sentence a frequency of

^^
hand
at

>=;,

and

of

/ 2 /

for

>

w,
>

2%

for

on the other

the end of the sentence a frequency of

w w w,

4%

for

and

of

5%

f r

lt

appears from
first of

these simple facts that the quantity of the

five syllables in the clausula is indifferent.

these

Sentencemetre.

Clausula-

metre.
4-0 o/ 5-0 o/

Quotient.

2-0% 2-5%

2(+) 2(+)

149

From
been
of
felt

this

we may

infer that the

itself,

form

:=:

has

as a metrical unity in

and

that the length

the

clausula here

comprises four syllables, although

five or

perhaps even more have been investigated.

this

In

way

it

becomes
group

clear in

what way the

diffe-

rent combinations have to be grouped.

At

first

sight

we

might be inclined

to

in this

way:

and so

on.

This

way
in

of

grouping has been followed by

Kaluscha.

Serious objections have to be

justified

made
more
more
it

against
distant

it.

We
will

are

assuming

that

the

syllable

from the end

of the sentence the

its

quantity

be metrically

indifferent.

Consequently

is

desirable

not to separate cases such as

w w w and
their
first

w
syllable,
is

which

differ

only

as

regards
for

or

^^^^^
On

and

^w^w,
of

which the same

true.

the contrary metrically they belong together.

There-

fore the best

way

grouping

is

that in

which the same

combinations of the

final syllables

are put together, viz.:

150

and so
2.

on.

The

reliability

of the results of our investigation

firstly

depends on the exactness with which the quantity

has been fixed.
It

of the syllables investigated

In doing so
is

errors

have undoubtedly been made.

practically

impossible to determine the quantity of hundreds of thousands

of

syllables

without

making

mistakes

now and
is

then.

Where, however, the material investigated

such errors tend to neutralize each other.
with another far more important
indifferent
difficulty.

so large,

But
It

we meet
is

rather

whether
of

in

comparing the prose-metre

of Plutarch
noifjoco

with

that

Thucydides

we
is

shall
it

take e.g.

as

or as

nor
r\v

very important whether

,

we
ded

shall

regard avxov

as

or as
in

w
is

Provi-

that

we
to

apply the same rules

both cases, they

can never have bad consequences.

There

no reason

whatsoever

suppose that any rule consistently applied

e.g.

may

cause a difference

in

favour of the frequency of

151

ww
registered
author,

in Plutarch,

On

the contrary, suppose

we have
in

a material of

1000 cases investigated

as well as in

(a

non-metrically writing author),

whom we may
liquida

call

B.

We

some other have assumed that

syllable

muta cum

nowhere lengthens a

which

contains a short vowel.

We

state that

nowhere

differen-

ces arise which point to a preference or an avoidance of

some form
applied,

in B.

Let the
but

same procedure
let

of

collecting
that

material

be

us

now assume

muta cum
will

liquida

lengthens a syllable eve^where.

What

happen now?
to the first

Many
in

syllables

which were short according

scansion

become long now. These

syllables are distributed

to this the

a most accidental way.

Owing

number

of

cases of each combination in both statistics

the

(i.e.

that for

author

and

that

for

the author

B)

will

become
which

larger in proportion as the

the combination contains,
instance,

number
larger,

of long syllables

is

and the reverse.

For

the combination

wwwwwwww
If

will lose the

in

greatest
(i.e.

number
were

of

cases.

the

syllables

question

those syllables containing a short vowel before muta

liquida)
all

cum

penults, the influence of this proce-

dure would be
the frequency of

felt

only in a special way.

For instance
at the
all

the

would increase
If

expense

of the frequency of
of

^^.

they were

antepenults

the

sentence,

frequenc}^ for instance of

152

woulds increase
the the

at the

expense

of the frequency of

^
on.

'double spondee' would

get a higher percentage; in

same way
expense

the frequency of

w would increase
-w>^ ^,
and so

at the

of the frequency of

In both authors, in

as well as in B, this

would be the

case to the

same

extent.

comparison
of

of the final figures

resulting from our latter

way

scanning would yield the

our former investigation,
exist

same
viz.

result

as the

figures

of

that

no metrical difference whatsoever would

*)

between both authors.

The

question

becomes

quite different,

when we apply
Suppose we

different rules of scansion to these

two

authors.

compare authors living in times quite distant from one another,

so that the pronunciation has considerably changed. Suppose
that
in

muta cum
If

liquida lengthens

by

position, but

not so in B.
into

in

scanning one takes these differences

in the
It

account,

corresponding differences will arise

this factor only.

statistics,

which are caused by

may

appear that the author

of
I

suddenly shows a great frequency

w w,
say

or other inexplainable tendencies.

this

Nor need
of

much on

point:

in

comparing the metres

two

different authors

one should apply exactly the same

the remarks of Munscher, who expresses his such researches, as it had not been clearly stated in what way dubious quantities had been regarded, are incorrect. Compare Bursian's Jahresbericht.
*)

That

is

why

doubt as

to the results of

153
rules
for

each

of

them.
to

Even

if

the pronunciation has

changed according
to

time or place, these changes have

It

be quite neglected and on purpose.

does not matter

how we regard either of these dubious cases, whether we suppose that muta cum liquida lengthens a syllable,
or not.

Suppose
e,

that

it

could be proved that in Plutarch

to

the vowels
this

and v had
to

be regarded as being long,

evolution

had

be neglected.

For

if

we

did not,

any comparison with Thucydides

for instance

would be

made
metre
author,

impossible.
of

In

comparing, however, the sentencewith the

it

one

author

clausulae

of

the

same

we

are allowed to take

this
it

into account.

From
interesting

appears

again,
of

how

methodologically
is.

the

investigation

antique prose-rhythm

When

once by means of these seemingly inexact

statistical

methods the tendencies have been ascertained, the true

quantity of the
the
syllables

can often be determined with

If

help
of

of

even these tendencies.

liquida

an

author uses

cases
the

muta cum

by preference there where

e.g.

metrical tendencies
elg

favour a short syllable,

etc.

ovx

djtoKQovei,

avxovg

ajioxXiveiv,

at

the end of a sento

tence, while
this

he generally prefers

^^ w ^
mind

w,

may become
liquida.
it

the clue to his metrical appreciation of

muta cum
Then,

must be borne

in

that in the figures

given for the frequency of the metrical forms in the whole

154
sentence, those for the forms occurring at the end of the

sentence will necessarily be partly included. Should

to exclude to

we
is

try

them from our

statistics,

we

should be obliged

assume a

certain length of the clausula,

and

it

this

very length which

paring,
for the

we

are trying to determine.

In

com-

however, the figures

for the sentence with those

end

of the sentence, the real differences will

never

disappear.
differences

This comparison

will

show

all

the metrical

between the clausula and the remaining part

If,

of the sentence.

for instance,

an author wholly avoids

^^^

in the sentence,

but seeks this

same form

as a

sentence-ending, our statistics for the whole sentence will

comprise some cases of

it.

Nevertheless

its

frequenc} in
7

our clausula-statistics will be

much

greater,

and

in this

manner
In

the difference will appear very clearly.

order to
scholars,
in

be able
it

to

compare

results obtained

by

different

is

desirable to

mention what has

Therefore
I

been chosen
give a
list

such

dubious cases.

shall

of the principal rules applied here.

alei

ovv-,

not vv-]

and

dei

Tioieiv

a short

vowel
before

is
/fy*,

not lengthened before muta

fiv,

cum

liquida, except

yju,

yv, dfi, dv\

nor

is

a long vowel shortened

before an
final

initial
is

vowel

of a following

word;

elision of a short
initial

vowel
of

accepted as a rule anywhere before an

vowel

the next

word

(except of course on before a

155

vowel

etc.)

a long final vowel

fjv

is

never elided, nor short-

ened aviov
3.

In the second

place the reliability of the results

of the material, that is the
If

depends upon the extent

numtwo

ber of cases investigated.

we have
draw any

investigated

clausulae of the author A, and

B,

two clausulae

of the author

we

are hardty allowed to

inference what-

soever from the figures thus obtained.

in

If

we

should find

two forms

^=:,

and

in

B two forms
are
are

^,

then this does not prove:

(1) that all

clausulae of clausulae of
is

A
B

w ^,
w,
clau-

nor nor
sulae of
If,

(2) that all (3) that

there

any difference between the

and those

of B.

however,

we

should have investigated 1000 clauat

sulae taken
to

from

random,
?

all

of

which

will

prove

be

of

the
all

form
of

w
will

and likewise from B 1000

prove
to to

clausulae,
:==:,

which

be
infer

of

the

form
great

we

should

be allowed

with

probability:
(1) that in

A
B

about

100%
100%

of the clausulae are of the

form

w ^,
(2) that in

about

of the clausulae are of the

form
(3) that

;=:,

there exists a difference between the clausulaein general,

forms of

and those

of

in general.

156

From
number
of

this

example

it

will

be clear

to

you

that the
relia-

cases investigated highly influences the

bility of the results.

I

shall take another

example.

If

we

should take from

to

A
of

three
the

clausulae,

one

of

which would prove

other,

be

form

w w,

the

however,

of the

form

w, we should

not be

able to infer that probably in

sulae
of

about

33%
we

of the clau-

would be
form

of

the

form
If,

w w,
however,

and about 61
should

the

w.

have

found on 1000 clausulae 330 of the form

w w,

and

670

of

the form
If

:=:,

this conclusion

would hardly

be premature.
yield

a second group of 1000 cases would

the

same

figures,

our probability would virtually

become a

certainty.

The
lowing:

question
is
it

which

will

occupy us next
in

is

the

fol-

possible to

determine

scientific
is

way
able

what degree
to give?
4.
It

of certainty a material of 1000 cases

cannot be

my

task to deal here with the purely

it

mathematical side of the subject, as

sible

would not be posproblems

for

me

to

give you

an independent judgment on
practical side of the

such

questions.
us.

Only the
found

can interest

An

empirically

percentage

represents

only

seldom quite exactly the true proportions.

As

a rule the

157

percentage
ratios.

forms

only
error
1

an
of

approximation

to

the

true

The mean

such a percentage can be

expressed by a formula

mean
where / represents
cases investigated.

error

=
this

/>(100-/) | 1/ ^

-,

the percentage, and

the

number

of

Can

formula be applied to our

philological material?

Now, we can
material into

investigate

whether
ex

in

dividing our
,

groups the formula

of

1/

where

percentage

single

group

minus the average

percentage of the five groups, and n

groups, agrees with the formula
e2

=
N

the

number
>

of

1/

n~~

where

percentage of a single group, and

z= number of

cases of the group.

In order to
for

answer

this question

divided

my

material

the

sentence-metre of Livy into five groups of 200

cases.

For each

of

these

groups

calculated

first

the

percentage of each of the sixteen possible combinations

of 5 syllables (the last

being taken as anceps), further of

each 5 corresponding percentages the average, then with

the help of these figures
of
bx

and

e2

Of course our material

suffice to

every time

five

cases
el

does not
e2
-

expect an

exact agreement of

and

158

TABLE.
Livy, sentence-metre, five groups of 200 cases.
Difference
IT

3P"

Five groups of

^
Average

Fl

8*

between the
last

Number
of the

200 cases each;

number
Form.

\
%
0-9%
15

SI

two

of cases

form.

for each group.

2
7

T
II

colums,
l

62

(0

07
I'l

+08

916
1724

^- ww_www>-.-

3 5

5
7

23
2-8

T4

+ 03
0-6
-0-2 4-03

6
13
7

06
14 14

12

25-32 3340

15 10
2
5

8 11
3
7

57

16
1*1

^w_ w__
w-

9
7

26 60 66
111
2-7

4148

15 12
13
13

19

26
11

17

+0-9

49-56 5764 65-72

7380

14 16 10

18

0-7

15 15 25 30 26

34
12
1*4

22
11

+ 1-2
+01
0-3

81-88 8996 97104

105112

www _ww w_w ^ ^^

9
15

3
12

8 13 15
9
7

6-3 4-3

17

10 15

2-4

14
21
1-5

+ 10
03 + 0.7 + 0.4 + 13
+3-2

25 19 17 15 19
i

95
4-2

18
2-2

16

12 26 21 20 20 18 20 18 18 13

99
8-7

113120
121-128

25 23
5-8

21

44 24 46 20 32

166
1

20 26

The
in
in

result is that in

1 1

of the
is
It

6 cases the figure found

an a posteriori
5 cases
*)
it

way

higher than the a priori one;

is

smaller.

appears from

this that

indeed

f)

5
200,

average %.

159

the

formula

1/

m~

>

within the limits of the exis

actness necessary for our purpose,

philological calculations.
4.

applicable to our

In

dealing with problems to be considered

now

we
for
is

shall

assume

that

the

formula

1/

irp-

the

mean

error of an empirically found percentage

applicable to our philological material.

The
Let
are
it

question must be considered

of

somewhat
and
at

as follows.

be granted for the sake

metrically
of
6.

argument

that the
that

Laws
real

homogeneous

unity,

the
of

percentage
sentence
follows.
is
If

the
If

double trochees
this

the

end

the

be the case,
of

we may
is

reason as

material

1000 clausulae
of

taken from
is

the

Laws,

and the number

shall

double trochees
a.

cal-

culated,

we
on.

get the figure

Another group

of
b,

1000 clausulae will give for the same form a figure

and
b
6,

so

The
These
the

errors
errors

which
are

are

made

are

6,

etc.

probably

distributed

in

accordance with the law of accidental errors.

Half the
distributed

number

of

figures

6,

6,

etc.,

if

ideally, will lie

between
and +0-6745

-0-6745

|/*!50p!>

|/^p^,

the other half lies outside these limits.

160

Only 454 on 10000 cases with

lie

outside the limits

_2
_3

|/Al00p/)
j/^100-/)
To
simplify

and

+2 j/>flMp4.
will
lie

Only 26 on 10000 cases

and

outside the limits

-/, +3 j/X100 N
calculations
I

such

give here a table,

ErblicJi-

taken from
keitslehre, 2
nd

W.

Johannsen, Elemente der exakten

ed.,

Kopenhagen

1913, p. 74.

161

TABLE
of of

the

number

of deviations

on 10000 which,

in the case

an ideal distribution,
,
. .

will lie

be between deviation
.

and some positive or some negative deviation

Number
d
(deviation)
e

d
Number

of

Number
d
(deviation)
e

deviations on 10000

of deviations on 10000

of

(deviation)
e

deviations on 10000

between and d.

between
and
d.

between and d.
4987 4990 4993 4995 4997 4998 4999 4999 4999 5000 5000

0.00 0.05 0.10 0.15 0.20 0.25

1.50

199

1.55

30
0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95
1.00
1

05

1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45

398 596 793 987 1179 1368 1554 1736 1915 2088 2258 2422 2580 2734 2881 3023 3159 3289 3413 3531 3643 3749 3849 3944 4032 4115 4192 4265

1.60 1.65 1.70 1.75 1.80 1.85

1.90 1.95

2.00 2.05 2.10 2.15 2.20 2.25 2.30 2.35 2.40 2.45 2.50 2.55 2.60 2.65 2.70 2.75 2.80 2.85 2.90 2.95

4332 4394 4452 4505 4554 4599 4641 4678 4713 4744 4773 4798 4821 4842 4861 4878 4893 4906 4918 4929 4938 4946 4953 4960 4965 4970 4970 4978 4981 4984

3.0
3.1

3.2

3.3 3.4
3.5

3.6
3.7

3.8 3.9

162
5.

So with
will

the help of the figures obtained in this manto

ner,
is

it

be possible

determine

how

great the chance

that the difference

is

between two percentages which are

the (exact or inexact does not matter)
difference,

empirically found

expression
could
figure

of

real

a true tendency.

If

it

be assumed as a certainty
1

for instance, that the

in

42

for the frequence of

w w
it

non-metrical

prose

is

correct,

and

if

we

should find a frequency of

could be calculated

29%

of the

same form

in Plutarch,
is

how much

greater the chance

that there exists a real

metrical difference between these

two

texts
exists.

on

this point,

than that no such metrical difference

Another question
of

arises.

Suppose
prose
is

that the frequenc}'

in Plutarch

ww

in non-metrical

14*2%,

29-0%, how
sight

can

this

difference be
to

measured?

At

first

one would be inclined

measure the tendency

of

which has caused the deviation

290%

from 14-2%

by

taking the difference

29-0

between the two

figures:

If

this

14-2 (0/ ) 14-8. (%) should be done, the conclusion must be accepted

that the

tendency causing a deviation of

1

16%

in metrical

prose from
as
that

in

non-metrical prose
deviation of

is

exactly as strong
metrical prose
it

causing

50%

in

from
see

65%
that,

in non-metrical prose.

And

yet

is

easy to

when

only
it

of

a certain form occurs in

difficult for

non-metrical prose,

is

much more

an

artist

163
to

raise

figure

from
to

(%)

to

6 (%) than to raise a

to this difficulty in

figure

from 50 (%)

65 (%).

Owing
it

using the difference between two percentages as a for-

mula

for

measuring the tendency,

appears to be more

simple to express the strength of the tendency by means

.

# of a quotient -j-

16

and

65
j.

Is this

way
I

of expressing the strength of the

tendency

quite

correct?

am

not able to answer this question.

it

Leaving alone the question, whether

to

is

possible at

all

measure pscyhical tendencies

in

a scientific way, supit

posing for a
to

moment
is

that theoretically
difficult to

would be possible
In

do

so,

it

extremely
of

give this answer.

favour,

however,
cited.

the

quotient

some undeniable

facts

may be
If
it

be assumed that

in non-metrical
1

prose the form

^ ^
the
is
1
:

has a frequency of

%, and

that in this prose

ratio in
3,

which short

syllables stand to long syllables

the frequency of

and

of

w_ w ^^

will
will

be

be

3%, 9%.
where
this

Now,
ones
for
is

if

there exists

some
but

metrical text,

ratio of the

number

of long syllables to that of the short

quite

the same,

where

exists

a preference
the frequency

one metrical form only,

viz.

^ ^,

164
of

which form

is

twice as great,

viz.

24%, we

shall get

the following scheme:

Non-metrical
prose.

Metrical
prose.
Quotient.

3% 9%
w w w

6%
18%
in

think,

we

are justified in assuming that the tendency

the metrical

which has caused the higher percentage

text
is

the

same

for

as for

w w.
it

If,

however,

this

tendency should be the same,

in quite the

ought

to

be expressed
if

same way

in

both cases. Now,

is

the

difference

between the percentages

taken as a

means
and 9

to designate the tendenc}',

we

get 3 for
is

w w
we

for

>-i.

If

the quotient

taken,

get

2 in both cases.
coefficient

Neither the difference, nor a correlationratios.

can express here the true

TABLES.
In investigating
of

Greek

texts according to the

all

method
material

Marbe
not

have neglected
to

interpunctions.

Those who

are

accustomed

deal

with

statistical

collected in order to determine tendencies, a material the

errors

of

which tend
in

to

neutralize

each

other,

as

for

instance

experimental psychology, ma}/ have doubts

about the

reliability of this

method.

165

Neglecting
this

all

interpunctions

enables us to exclude
editors into diffe-

subjective factor introduced

by the

rent texts.

The varying

interpunction
authors,

would make any

or even between

comparison between
different

different

passages of the same author

in different editions

impossible.
this

One

serious objection

may be made

against

procedure.

By applying

it,

longer series of long, or

longer series of short syllables are often introduced which

are not

meant as such by the author.

For

instance,

if

an author avoids
the

^w^
In

he need not avoid

it

end

of

a period, for there

may

www represent ^^

at

the last syllable of a sentence or of a part of a sentence

may
in

be 'anceps'.

the

mean

time this introduction of

these longer series of syllables will never be a difficulty

stating

the

general tendencies,
If
,

as
it

e.g.

avoidance of

longer series of long syllables.

for

influences the figures

some forms

as
all

w^^^w

and

others, the

same

holds good for

authors investigated.

When
mined, one

once the general tendencies have been deter-

may

investigate

whether the quantity

of the

last syllable is indifferent,

and whether seeming exceptions

will

have

to

be explained by neglecting the interpunctions.

for the scanning of the clausulae:
first

The same holds good

statistical-philological

researches must in the

place

exclude subjective factors, without caring in the least for

the reproach of having

made mechanical

calculations.

166
If

we

should be inclined to take the interpunction

into

account,

editors, or
first

we must accept the we must make our own

in

interpunction
interpunction.

of the

In the
diffe-

case different authors would be treatened in a

rent

way,

the

second case a

new

subjective

factor

would be introduced, and a

really

dangerous one.

For
to the

we
we

are inclined to

model our material according

If

results

we
on.

expect.

we

expect an avoidance of

wwww,

are inclined to reed

If

^^,^^
is

instead

of

^^^^
we
The

and so

interpunction

taken into account,

run

the risk of finding too

tion
is

much

in

our material;
little

if

interpuncfirst-

neglected

we may

find too

in

it.

mentioned procedure gives results which can hardly be

corrected later on; the results of the last-mentioned, on
the other hand,

can be corrected.

For

if

once the ten-

dencies have been found

other

by

errors tend to neutralize each

is

we
is

can investigate what part the interpunction

In

plays in the rhythm.

syllables

Demosthenes a

series of short

often broken
for

interpunction, in Plutarch the

same

true

series of long ones.

Nevertheless our
in

method
authors, reserve.

clearly

shows

the tendencies working

both

and
In

we

can accept the results without the least

a material, on the other hand, which does

not neglect interpunction, errors do not tend to neutralize

each other

on the contrary, they do

just reverse.

167

Frequency
in

of

simple metrical forms on 1000 syllables

different

authors.

The

figures

are

derived from a

material of every time 1000 syllables.

THUCYDIDES.
Book
180 113 57
161
I.

150

165

70

-2-3

40
18 12 2

^4-5-6-1^8-910
11

58 28 32
13
7

89 47 46 23
13 4 6

98 62 40
19

160 113 57

200
107

165

35
15 10 13 5
1

60 37
13 7
11

93 55 43
19 5

188 105

173

64 40
14

99 68 36
13

155 91 62
31 21 14

169 109

167

46 33
17

8 8
5
1 1

4 3
1

3 3 3 3

4
1

3 2 3

8 4 2
2 2 2
1

8 6
1

15

8
3

6
7

2 5
1

4
1

4
1

10 5 5 8 3
1

85 49 30 32 6
10

4 3 2
1

1
1
1

1 1

12

1 1

13 14 15

XENOPHON.
Cyropaedia, book
I.

_^

161

171

149

137

-2

3^
^4

78 59 35
14 15 10

77

96
61

44 43 24
6
11

37
16 18

95 48 39 20
16 13
1

136 92 45
31 21

185
101

47
51 17

179 110 55 36 22
14
7

172 106

231 103

142 114

58 38
17

45
41

25
13

40 40 20
14 5 5
1

187 109 50 37
17

162

85
51 31 18 15
11

56-7 -8

20
7 5

9
1

910
11

6 4 2 2

5 2 7 2

9 8 4
1
1

8
7

4
3
1

3
1 1

4 2

6 5
2
1

6 9
7

4
1 1

6
2

3 3

12 13 1415

1
1

1 1

16

168

ISOCRATES.
Panegyricus,
I

sqq.
182 152 133

w 2
34^
5

140

84
57 32 22
17

159 74 59

173 75

171

85
51

40
19
11

48 36 20
9
11

156 87 47
41
17

36 23
17

158 97 57 35

84
61

89
72 23

78 70
31

38
12

162 79 67 44
15

170
91

151

99
53 29
21

54 37
21
11

20
13

25
11

22
11

6,7w '8^

8 5
1

11

5 3
1

910
11

3
1

6 4
2

4 4 3 2
1

13 12 7
2

6
2

17 5

16

6
2
1
1

4 4
1
1 1

2
1

8 4 3 3
1

5 2 2
1

6 6 4
1

10

18 5

5 2

4
2

6
2

13 14 15
12

2
1

1
1

1
1 1

16 17 18-

DEMOSTHENES.
Philipp. A.

Olynth. A.

Olynth. B.

Olynth.

<_<

10^
^11^

^2 w3w^ w4 w5 ^6~ w7^ ^8^ w9-

147 103

134
111

118 109

158 130

74 28 24
12

73 36
21

68 48 30
6
7
1

69 46
17 2

144 108 69 39
21 14
7

114 123
61

165 106 70
41 21

133 102 75

40
21 14
7

43 23
10 7 2
1

8
3 4

8 5 3 2 3
1

9
5 5 5

4 3
2

3
1

2
1

2
1

69

[DEMOSTHENES].
77(00?
ri]v

emoxoXrjV rov <PdtJZ7iov.

-^
3 4

^
2

172

95 67 43
18 14 2 4
2 2
1

150 86 74 38 23
14 8 2 3
1

567

8910
11

PLATO.
Republic.

Book

B
102

r
46
42
17 13

'
93
42 32 25
15

E
83 67 28
15 2 3

3 _4
^-2

119 65 45
14

102 57

105 62

118

118

8-.9 10 11 12 13 14 15 16

-6w7

5-

29 24
11

46
15 16

48 30
21 10

65
31

9 6

6
5 2 2
1

4
2
1

6
2
1

6 4
2
1

4 6
3
1

8 4 3
1
1

28 9 4
3 3
1

2
1
1

1 1

17 18^
20

19-

170

PLATO.
Laws.

Book

I.

II.

in.

IV.

_2 ~_3 ^4 5

81

50 32
17

82 62 28 20
14 9

66 59 27 22
14

85 57
31

93 53 28
14 14

76 59 35
21 13 4 5 3

96 42
31

83 68 30
15
11

23
13 2 6 4 2
1

27
15

6
^-1^

^8^ ^9-10

9 6 6
1

4
5

9 8 4

9 6
1

9
3
1

3 5

6
3
1

3
2
1 1
1

1
1 1

2
1
1

12

11

13 ^14^

PLUTARCH.
Pyrrhus. Book
I.

_-w

^
2
6^

237
144

196 143
71

190 152

225
135 77 27
13
7

186 142

229 205 202

134 72

^3~

A^
-5 -7
8

60 40 23
4
1

36
17

80 35
12

66 37
21

34
14 8
1

148 76 36
15
7

141

163 126

171

68 32
17

80 43
16

135 92 29
15

184 138 65 32 25
10 2
1

197 127

68 32
13
11

6 4
1

4
2 2

6
2 2

8
1

6
5 3

5 3
1

2
1

6 2 3
1

2 3

4
1 1 1

9
toil 12
1

171

[PLUTARCH].
Consolatio ad Apollonium.

_^
v_

^2

3 _4 5
6-

160 102 75

34
17
11

-7w

_9

_8

5 4

4
1

-10-

PLUTARCH.
ZvyXQlOEiq.

Dem.
.,

a.

Ant.

Thes. a Rom. Solon a Popl. Arist.

. .

a.

Cato.

W W
^2 ^3 w4w ^5^ w6-7-9-

/w

187
131

74 40
19 8 2
1 1

188 119 74 36
19 14
1
1
1

171

134 79
41 14

6 4
1

178 143 73 30 20 9
2
1

201 132 72

204
155 70
31 14 8 2 2
1

185 134 70

172 139 78

33
15 7

45
15 5
7

38
16 7 2 2
1

a-

4 3

172

FREQUENCY OF DIFFERENT FORMS ON

Averages.
,5

1000

SYLLABLES.

>->

o C

u U o

o B

Q
167-7 97-2 50-3

<u

Q
1988 1600

o D

Plato

o e
3*

"3

Rep. Laws.
a-

171-1

23 _4 5^6 7 -8
_9-

977
57
1

1589 139-1 1610 852 1115 905

185 8

1050

580
35-2

699
40-

70.5

565
35-4

366
197
101
7 3

383
19-3

128

198 137
7-8

22-3
9.4

405 205
140

828 138-8 1020 137-1 750 73-8 563 72-9 368 303 344 340
199
12 9

199

42
3'1

86 45 28
19

63
2.8

4-7

32
2-2

19
1-3

50 30 25
1-5

76 55
3-5

64
5-4

16-8 6-9 2-5

1-8

170 110

16-5

50 40 40
10

8-0 3-0
1-4

2-4

26
19 ro

2-1

14

14-15
-16-

11 12 13
17 18-1920-

-10-

0-8

05 03
0-2

07 02
0-5

08 07
01

04
01
0-5

04 04 03
01

05 02 03
0.2

0-8
0*3

08
0*3
o-i

0'2
O'l

02 03
01

03 02
01
o-i

01

01
01

THUCYDIDES.
Book
I.

2 3
4 5 6 7 8 9

302 139 66 26
13

399
128
51

384 340
135
71 21

138

341 142

26
11

69 28
6
3 2

80
22
2 4 2

312 139 63 23
14
7

8
2
1

361 139 63 29 9
1

317
137 57

343
135 72

372
142 62

347 123 73
31 10
1

384
110

35
8 5
1

30
8
3
1

20
12

80 24
7

3
1
1

3
1

3
1

3
1

2
1

173

XENOPHON.
Cyropaedia, book
392 124 63 28
8 2
2
I.

411 128 56

344
141

379
147

402
149
51

26
7

64 30
9
1

44
19

339 124 69
31
11

358
154

303
144 53 29
7 7
1

347
111

313
144

29
3
2 2

60 30
11

66
32

65
18
11

341 127 77

28
14
3

9 4
1

20
6
1

2
1

3
1

4 2

2
1

391 128 65 23 8 3
3

2
1

ISOCRATES.
Panegyricus
406
135
I

sqq.

64 23
7
1
1

382 119 68 24
11
1

395
113

374 323
131 121

358
138

345
121

365
137

373
128 74 20 8
3

380
140

361 132

369
128 78
21

54 27
16
1

52 30
11

68 35
6

68
19 13 2
1

68 23
13
1

69
24
7
1

54 36 9
1
1

54 36 9
3
1

6
2
1

5
1

DEMOSTHENES.
Philipp.

A.

Olynth. A.

Olynth. B.

Olynth.

R
340
149
101

336

330
157

345
154 112
3

282
129 132
10
2

327
130 126
5
2

338
167

2 -3 4 5 -6 7
-8

W--

128 129
7

93
13 2
1
1

95
5 3

341 137 126

13 3
1

13
2

174

[DEMOSTHENES].
IIqoq
rijv

moToA.r]v tov

&diJi7iov.

W
2 3 45

337 134 73 28
9 4

356
140 72 22
10
1

PLATO.
Republic.

Book

B.

r.

A.

,,

2 3 4 5 6 7 8

149 110 74 75 35 29 12 9 1 4 2 3
1

140 162 53 62 39 25 5 6
2
1

151

165

56 29
11

64 27
11

7
1

PLATO.
Laws.

Book

I.

II.

III.

IV.

3 _ 4_
-5 6 -7

87 128 57 43 43 38
13 12
1

128 122
41

30
12

50 36
10
1

98 118 49 42 50 41
17

108 113 56 42 41 43
12

11

4 4

4
3 2

4
3
1

16 6 3

15

5 2
1

6
3

175

PLUTARCH.
Pyrrhus.
233
.

Book
238
138

I.

244
145 68
51 6 2

235
160 79

228
142

2 3 4 5
6-

140 58
61 7

60
51
11

44
5
2

251 165 51 51
7

236
161 61

229 281
150 65
51

255 277
177

282
153

178

65 50
6 5 2
1

48 9
4
1

54 45
5
1

50 49
5 2

6
3
1

157 53 55 4
1 1

49 48
8 4

4
2

3
2

7 8

9
10
1 1

PLUTARCH.
Consolatio ad Apollonium.

3 4 5 -6 7 -8 9

328 154 56 27
10

PLUTARCH.
HvyxQioeig.

Dem.
__

i.

Ant.

Thes.a .Rom.
253

Sol.

a.

Popl.

Arist. a.Cato.

2 3

266
165 52 45
11

285
152 52

142

4 5 6

40
13 3
1

43 50
2

259 168 52 54
2 3

249
148

236
153

267 165
61

259
171

49
59
7

3 2

2
1

56 55 4 4
2

45
10
1

57 40 9
2

176

FREQUENCY OF DIFFERENT FORMS ON

Averages.
,
,

1000

SYLLABLES.

t-,

T3

A
O
X!

5
o o S

5
CO

O 3
ji

o s
0J

Plato

tj

ti

Rep. Laws.

"5

9.

3 6 7 8 9
4 5

-2

3502 3600 3693 3299 3465 1339 133-8 1286 143-9 137-0 1462 673 61M 643 114 3 72.5 640 8-6 25-0 263 26-9 265 307
9-0
9-8

12 8

249-7 3280 259-3 155-0 1540 1580

47-5

594
50-3
6-6

403
13-3 5-3 2-8

570 270
100

97
1-5 1-3

1-5

29
1'5

33
1-5

0-6
0-1

95 25

9-0
.2-7

528 485 73

1-2

28 08
01 01

30
10
1-0

26
0-8

0-4

03 02

03
01
0-2

0-5
0-3

10 11

o-i

FREQUENCY OF EACH OF THE

IN

128

FORMS
For

DIFFERENT AUTHORS.

For Thucydides Hude's edition has been used.

his

sentence-metre the beginning of the

first

book has
in Classical

been scanned.
Quarterly 1915,

For
p.

his clausula see

my

paper

231

sqq.

For Isocrates the

1886, has been used.

edition of Benseler

and

Blass, Leipzig

For Xenophon the

has

'editio

minor'
of

of

A. Hug, 1905,

been used.

The beginning

the Cyropaedia has

been taken.

177

For Demosthenes the

edition of

Weil has been used.

The
the

figures for his sentence-metre

first

have been taken from

1000 syllables

of Phil. A, item of Olynth.

and

B, and the others from the beginning of Olynth. F.

The

clausulae of

Demosthenes (898
F,
FIeqi
rfjg

cases) are those of Phil. A,

Phil.

Olynth. A, B,

eiQrjvrjg,

B,

ITsqI

rcov

XeQQovv7]ocp, Phil. F, IIeqI ovvldt-ecog.

For Plutarch the edition

has been used;
for

of

Lindskog and Ziegler

For

the other lives that of Sintenis; for

his

the Moralia that of Hatsidakis.

sentence-metre

the beginning of the live of Pyrrhus has been investigated.

For

his clausula see Classical Quarterly 1915, p. 231

32.
The

For Plato the

figures
for

edition of Burnet has

been used.

his sentence-metre
first

have been taken from the

beginning of the

book

of the
of

Laws.

For Philo the

been made use
of.

edition

Cohn and Wendland has

of

The

figures

his

sentence-metre

have been taken from the beginning

(vol.

of IIeqI (pdav&Qcomag

V,

p.

279

etc.)

The

figures for his clausulae

have

been derived from

etc.,

vol. V, p.

279 sqq.

(IleQi

<pdav&Qcomas

500 cases).

178

<u

73

73.

S
a.
<u

o u
a;

<u

as 3
o S
7)

S-3

J8
c
a;
C/3

O 3

14
Si

X
Sum
1

c a
C/3 C/3

so

Ph

o*

c
<u

IX

7}

total:

1000 1000 1000 1000J1000I

(J

1000 1000

1000 1000 500

1

2
I

2
I

2 3 4 5 6
7

1 1

o
1

2 3

6
1

3
1

2 2
3 3 4

3 4 5

2 10 4 8

5
5

6
2 5
7

1 1

9
:

6
5

8 9 10
11

2
1

12

5
2

3
3
2
1

3
1

16

8
12 3

! 15 9
7

6 10
4

8 5 6
7

4 6
7 7

12

6
7
3

5
10 8

9
5

13 14 15 16
17

13

14 18 14
11

6
7
1

4
1

8
5

10 14
19

3
12

8 9
10
1

2
1

4
2

11

3 8 3
11

5 3

18 19

6
3

7
13
5

6
13 2 10 5
11

2
5

5
2 J

20
21

6
1

17

22 23 24 25

10

6
1

8 8

?|
5

6 8 10 15
5

6 3
7

4
9
10
()

26
27 28 29 30
31

11

11

5 8
10

11

8 3
7

3 3 7 3
12

4 6 9

6 6 8
5

4
3
1
1

12
11

8
7 8
4

32

8 15

5 8
5

4
3

9
12

179

x
C3

7)

U3
.

ti (h

t
H
c/3

MB
O
a;
CJ <u

D s 1-3

i Si
^
en

O
<u

5
as

S
aJ

=3

S SO Q Q
C/3

a o fl U
rt

<u

Si

Oh

U
v:

c/3

Sum
33 34 35 36 37 38 39

total:

1000 1000 1000 1000 1000 898

000 1000 1000 1000

4
2 2
3

500
2 3

338

4
3
7

1
1 1

6
5

3 8

6
7

4 6
7

9
3 3

6 8
2

2 3
1

2 14

9
14
5
7

9 6
7

9
14 5

8
8

12 2 3

4
5 2 4
1

4 4
10

6 18
11

3 4

14
12

8
7

8
7

40
41

9
11

13

6
10

4
5 5
3 16 7
12
7
11

4 4 3 8 6 10
7
11

42 43 44 45 46 47 48 49 50
51

4 4 4 4 6
13

6 2 6 2
2 7
11

5
11

6
1

5 10
11

5 5 2
1

7
11

6 8
7

2 5 5

12 3 2

5
2

9 16 6 2 14
15

5 8 3 5 5 2 4 2 3 4 3 3 2 2

4 4
11 7

6
10
11 4
1

6
11

5 5

9 5
10
7

4
1

6
3
2

8
7

4
5 10 2 9

5 5 13

1
1

2
1
1

4
3

6 13 6
11

4
3 5

6
2 6
1

4
1

52 53 54 55 56 57 58 59 60
61

4 5 10
11
11

4 8
7

8 4
7

6
16 7 14
7

6 8 9 8
3 3 12

5 8 4 5 9 6 4 6
11

4
11

11

4
8
1

6
7 5

15 8 8
12 13 2

17

5
7
11

9 8
15

10

4
4 6 7 6
7
1

8 10 10 8
12

13
1

16
5

6 8
10
11

62 63 64

6 8
11

16 14
11

4
17

26

13

26

9 6 8 9 9 16 9

9 8 9 4
12
11

4 8
5 3
7

6 4 4 6
2

4
12

14 13
2 3 7 10 15 13

14 10 5
7

5
7

4
2
7

5 17 7 3

4 2
7
7

8 8 9

10 4

6 4 2

20
8
11

16

8 6 9

180

4;
4^

ai

03 73

V
T3

rt

4
o o c B o c O
is
(a 03

if}

% 2 C

tS

QJ

s
2.

a;

S3 x:

of U
o
e
<u c/)

H
C/3

iSB
0-(

sS
C/3

<u

c/)

Sum
65 66
67 68

total

1000

000 1000 1000 1000 898 1000 000 1000 1000 1000 500 338
!

4 4
11

5 3 5
7

2 10

2 3

1 1
1

2 3

69 70
71

4
3
7 13 4

4
10 18 15 5
11

6 8 4
3

6 6
7

4
7

1 1

8
10

72 73

6 9 6
14

6 8
10
5

3 3
1

2 3 5
1

19
18
11

6
5

74
75 76 77 78 79

7 5 9 7
7

14 12 9 8
11 11

6
13
11

4 4
5

29
8
18

25
13 17 12 10 13 14 2 3

14 3 15 12 3 7

2 3 6 3 10 18 6 16 5

3 6 5

1
1

4
3

2 2

6 8
5 12 15 4
1

8
3 5 13
11

8 8 4
11

10
7

6 8
5
7

4 4 4
7

6
5 21 3 5
11

80
81

82 83 84 85

10 16 4 6 9
11

3 12 5

5 5 16 3

4 9 8
17

22
17 16 17

5 13

5 2

2 2 2
1

8
5 10 8

6
13 18

1
1 1
1

4
1

4
7

11

9
10

10
7 5

7
7

86
87 88 89

9 8 8
2

3 6 3 9

11

11

9
10 10

10 14

20
14 19 17
11

23
16 14 14

4 9
2 5 3
4

3 12 5

14

8 6
3 17 3 5 6

4
2

8
7

8 13
7

6
11

5
11

6 9 7 9 9 9 10 5 19 12 10
7

8 8
3

3 8

6
3 2
1

9
17

20
8 6 5 13 9
11

16 16 5

18 18

4
8 6
10

4
7

24
12

20
16

4
11

11

90
91

6
7

11

92 93 94 95 96

16 10 14 5 14

14 13 12

14 10

34
16 21

26
18
11

8
12

8
13 10 7 21

8
7

8 9 10
10 16

22
12

10 10

4
11

25
21 18 14

16

8 15

18 14 17

28 26 23

6 2
13 13

6
16
12

8
10
12

181

d'

<u
J-.

ah 3 6
TJ
in

7)

Ih aj as

S ^ 7 a. o a

CI

<u

<u

*->

q>

-G

"tJ
</)

c/)

3 3
a3

si
5
3

a)

a a V
CJ

72 ~C
Du,

d?
<U

o C
c

rp -C 0H

o"3 V) 3
*

<u

Q
c/3
c/3

<u

U
338

S
WD

<U

(U c/3

CO

Sum
97 98 96 100
101

total

1000 1000 1000 1000 1000 898 lOOOl 1000 1000 1000 lOOOl 500

6 6
11

3 6

9
10 14
11

6 8 9
11

2 5
7
1

1 1

2
1

6 6
12
2

6
13
7

5 3 10

10
7

102 103 104 105 106 107 108 109 110

111

2 5 13

6
11

6
13 10 12
7

14 19 5 14

15 12 14

14
17

3 3
9 9
12
1

20
7

5 12 9 10

9
16 5

22
19
15
7

22
12

9 8
12 10 12

15 7

8
10 12 16 16 3 10

6 9 13
7

7 6
11
7

6 6 5 9 12
12

10

6
13 13

3 6 8 6 2 3
3 5
11

9 4 4
17 12 16

6 8
15 3

9
11

8
7

8
13
17

112 113 114 115 116 117 118 119

10 19
11

21

14 12 15

5 15
12

4
8 10 6
4 14
7

15

9 9
14

3 16 10
11

4 16
7

13 10 14 17 14 2

8
13 12 13

6 8 8
12

8
17

9
14

8
17 15 13

8
11

2
7

8
11

19

11

4 3

10 13 8 15

11 11

4 9
7

15
5

30
14 12 18 15
1

16

9
12

8 10

10

4
8
12

3 5
14

4
1

7 3
7

8 9
10
7

3
1

U
5 10
1

6
12 16 5 10 8 16 12

20
19

14
21 12 12
11

16 10

120
121

20
10 12 13 18 16

15
1

9 7
5
7

8 4
7

2 8 10

9
5
7

6
2

122 123 124 125 126 127 128

17 12

5 17

13 13 14 17 12

16 13 19
11

18
16 15

8 13
1

4 3
5

12 21 18 14

9
10

4 4

6
17

8
11

9
7

28

23

22 27 49

10

15

9
3

4 4 6 4 4
3

8
13 12 16 21 31

6 8 6 8
11

2 2 2
1

13
7

14

182

THRASYMACHUS.
not enable us to
his

The few fragments wich have come down to us, do draw certain conclusions with regard to
Only the form

preference for any forms.

w
altogether

shows a high frequency which may not be due

to

chance.

On

12 sentence-endings

we

find 427 of this

of form.

These 12 sentence-endings are the

following.

wwww ww *w
^

Cfjtovotv exdxeQOL

avdyxt] de Xeyeiv.
v7iooyj]0i tag ahiag.

w^ w w
s-g:

jiQog alkrjlovs dupixEoftai.

eooj(p()ovovjuev.

ii

TOIOVTOV
^o/ft)

EOtCLl)

w w w
w w
Among
and
the kola
rcdAecog axoveiv

Ivovxa.

ovoa naoLv.
jivvfidveofiai.

(pdoviKovvxag nao%eiv.
0(DCpQOvi,ElV ElCO'&EV.

we may
>=-:.

note

rolg vecdteqoioi

w ^,

183

THUCYDIDES.
Clausula of

some
II,

of

the

more

rhetoric parts of his

work:

I,

140-144;
l

35-46.
33 34 35 36 37 38 39 40
41
1

2 3

4
5

1 1

6
7

65 66 67 68 69 70
71

97

98 99
1

100
101

5
2 5

1 1 1

8 9
10
11

2 2
1

12 13 14 15

16 17 18 19

2
1

42 43 44 45 46 47 48 49 50
51

72 73 74

75 76 77 78 79 80
81

2
1

3
2 5

102 103 104 105 106 107 108 109 110

111

1
1

3
1

1
1 1

20
21

22 23 24 25 26 27 28 29 30
31

1
1 1

52 53 54 55 56 57 58 59

3
2

82 83 84 85 86 87 88 89 90
91

2 2 2
1

112 113 114 115 116 117 118 119

2
1 1

2
1

2
2 2 2

2
1

120
121

60
1

61

32

62 63 64

92 93 94 95 96

2
1

2 2

122 123 124 125 126 127 128

3 2 4 5

184

THUCYDIDES.
Comparison
of his clausula with his sentence-metre.

M-l
i

oo
O m
the

Number
of

3
Form.
CO
3>

P
is

J3

"u
rheto

yj
part

the form.

Clausu

Sum

total

of cases in

1000 2000

130

vestigated

17-18 81-96 21-24 1-8 9-16 35-36 69-72 37-40

105-112

_______ v_ ^
V

0-5%
144 18

0-5

142

16*2

Ss

-^

^^s
v.

20 36
^-^
l'l isi
^

20 33
5'2 5-4

v_.^_.w
v^v_^v_^
VJN_,

13

S-s

53-56 41-48 97-100 57-60 61-64 57-64 73-80 113-120 121-128 113-128 101-104 91-92 51-52

v_^ v^^ri
^.^
,

>,

^ ^^ ^
^ ^
v-

27 29 88 30
51
2-5

35
21

94 26 37
31

92
18

_^V

^-^

s^=^

^ ^
-

40 54 94 65
10-3 10*0

29 34 63 79
8'5

15

x_^_, ^
s

^-; v.^

97
18-3

^^

vjxw v

^ ^ ^^

204

208
9-2
/

26 23
1*5

61
2'3

11

The
in

great majority of the forms which as regarded

in

by Rollman as being favoured

Thucydides are avoided

}

Greek prose,

e.g.

w ww
,

w^

185

ww
17

32,

ww 101 104.
of

etc.

They comprise

nearly: 33
in

The forms

^_r

64,
in
if

comprise

Thucy-

dides'

sentence-metre 23*8
the

%,

in his clausula 18*4

%,
So,

the

clausula

more

rhetoric parts 14-6

%.

any forms have been favoured by him

(

in the

speeches

wwww? ww w?
to

and

others), the statements of

Rollmann seem

be wholly

false.

DEMOSTHENES.
Comparison
Number
of the form.

of

his

clausula with his sentence-metre.

Sentencemetre.
Clausulametre.
0*2

Favoured
forms.

Form.

17-18 81-96 21-24 1-8

9_16 35-36 69-72 37-40

105-112 53-56

41-48 97-100 57-60 61-64 57-64 73-80 113-120

121-128 113-128 101-104 91-92 51-52

www wwv. w_ _w_www wwww www www w>. www _ ww ww w w w w _____ www w ww w ww ww _____ w
ww _ _ w ww wv

o-i%
12-4

31

189 31

w
-____-

05
12

07 06
8

04
11
4-4

12

80
4*1

36 80 40
4-8

6-7

01

0-8

www

39 46
8-5

29
4-8
7*7

127 109
9-0

11*7 7-7

112
18-7 3-7

199
7-2

18 T7

30
31

186

Comparison

of the sentence-metre of

Thucydides and

Demosthenes

in

order to determine metrical tendencies in

to

Demosthenes. Attention should be paid

the fact that

the results could not have been obtained the

by means

of

method

of

Marbe.
etc.

are avoided.

Thuc. Dem.

Form.

%
1

S-8

^3 +
(

00
0-2

65 33 97
17 81

01
0-2

00 00 00 00
O'O

~
j
~~
I

s^^v^^O'O
^^^^^^ __^^^
^^^^
2

00 00 03 00
08 00

03
0-4

49
113 9 73
41

00
11
0*1

00 00
01

0*4 0-6

00 00
0-1

105

11

25 89 57
121

04
0'7

00 00
0*2

53 05

11
0'9

01

66 98 68
114 100 72

04 06
11

01

04
VI

00
0.1 0-1

01
0-1

09
PI

P3

00 04

E)

33 05

187
Dactyls and choriambs.
122 116 104

Thuc.

%
05
14
0-6

Dem.

^^(i=i)
+

is

favoured.

16

30
T9 T7

80
19

16

74 37

06 07
0-6

07 29 06
0-9 2-2

73

155

26
77

06
0'7

Reiteration of iambs and trochees

Thuc.
118
0'7

is

avoided.

Dem.
0*2

(-

85 43

07 04

04 03

(-

Reiteration of cretics

is

favoured.

Thuc.
56
10

Dem.
13 14 P2
(+) (+) (+)

11

07
TO

55

Combinations of

cretics

and trochees, item

Dem.
10 10

of

iambs

ands trochees, are avoided.

Thuc.
91

94 92
112 107 108

07 04
T6 19
1-3

08
15

10

07 08

188

The form
course.

^^ w
73 74 75 76 77 78 79

in the

sentence

is

favoured of

Thuc.

Dem.

Quotient

04
0'7

00
2-9
4-1
1-6

(-)
(+) (+) (+) (+) (+) (+) (+)

05 09 07
0-7

08
T8

2-0
3-1

22
17 16 17

2-4
1-6
1-1

80

10 16
of

The

highest

quotients, *nts,
l

course,

are those for the

forms ^^ ^^_ quotient that for ^^

and

ww

^^^

the lowest

Tendency
:

of increasing

avoidance (see

p. 30).

Thuc. Dem.
16
0-7

Thuc. Dem.
I

Quotient.

02
1-7

_ -

80 48
112 32

16
1'3

5-5

45
21
17

V2 (+)

96 64 128
>

19 15 14

\\\
05 07 no
>

16 /

19

1-1

(+)

26
2.8

26
2 8

1-5

09

3'1

(-) (-)

For

'w
?

^--

'"

'**

page

176.

ISOCRATES.
Comparison of the frequency
that of Thucydides.

of the clausula of Isocrates with

The
is

short penult

seems

to

have been avoided

by

Isocrates.
,.
,

Favoured
v_x,

every form with long penult: ^-~_

w,

w w

wws^

^,

^,

except

189
Isocrates (Panegyricus). Quotient. Clausula. Clausula. (sum total 2000 (sum total 338

Thucydides.

cases).

cases).

1-8 9-16
17-18 19-20

3'3

v_.w~^=i s~w-~W v^v^=i

w

v_^v_^ _s_;^=:

52 05
14

06 45
o-o

5'5

(
(
(

12
CM
1*2

21-24 25-32 33-34 35-36 37-40 41-48 49-52 53-56 57-60 61-64 65-68 69-72 73-80 81-96 97-100 01-104 05-112
13-128

^ ^
^

v__i=i ^_-^WV_,

^^ -^-^ ^^
V_*

20 50
0-8

12 12

39 03
0-9 0-9

17 13
2-7
1-4

V^^-"
x_.^=;

wv_^

13 21 37 16

23
15
2-7

15
0*6

v_^^ ^i=:
v_^=i

v_^^^._ v_^v_
-^-v_
'

^
iS*

26 29 34 23 35
7-9

18 12

14
2-4

33 39 60
11-3

>=*

10 VI 17 14
12
VI

(
(
(
(

(
(

(
(

i=i

%_^N

142

170
3*3
5-1

(+) (+) (+) (+)

(+)

V_.V_
>

^
V^

31 61

94
183

125 185

V2 13 10

(
(+) (+)

Only a comparison with Isocrates' sentence-metre can give a certain clue to his clausula-metre.

190

PLATO. LAWS.
Comparison of the clausula with the sentence-metre. The figures
for the clausula

have been derived from Kaluscha.

t fc g

Form ^ orm

3 p
-

c?2 3 xj *H G

VIZ
>*
<U CJ fl

8 6

D B

4)

<u U)

c V

a 3 3 13 p
CTJ

Favoured
forms.

Avoided
forms.

Cfl

C/3

c/3

D
62
126
1-6 (+) 1-9 (+) 1-3

1-8 9-16 17-24 25-32 33-40 41-48 49-56 57-64 65-72 73-80 81-88 89-96 97-104
105-112 113-120 121-128

36

3.8% 68 39 70 42
3*7

30
3-7

1'9

(-) (-)
(+) (+)

5'7

1'4 (+)

5'1

4*2

11
2-1

54
7.9

116
13-0 8-7
1*3

60
6-5
5'2

1-6 (+) 1-5 (+)

4-0
1-8

3'9

70
6-4

22 35 44
6-0

(-) (-)

2-0
1-5

(
(-) (-) (-) (-)

88
100

7.5

1-0 1-5
1-6

90
133

60
8-2

191

PLATO. LAWS.
Investigation into the quantity of the last syllable.

Sentence-metre.
A.
B.

C
g
<u

a u
C.

su a-

me

re.

D.

Frequency of the form

with short
final syllable.

Frequency of the form

with long
final syllable.

o 3

Frequency of the form

with short
final syllable.

Frequency
of

the forn with long-

V o
3

final syllable

1-8

1-4

9-16
17-24 25-32 33-40 41-48 49-56 57-64 65-72 73-80 81-88 89-96 97-104 105-112 113-120 121-128

2*2

19 41
1-71

+ +

24
3'8

38
8-8
1-1

+ +

2-5 3-6

23
T8
2-7

31! 23
21

4-

1-9

2-7
2-7

1-0

+ + ++ ++ + + + + + + + +

18

30 24
8'2

41
2-6

17 15

22 24 33
2-7

51

29 40 38 30 28 46 42 40 54 78

+
?

34
4-2

8'8

+ + + + + + + +

30 06
10 14 15

57 07
1-2

21

29
37 41
5'2

23
19

30

PLATO.
Investigation into the
resulting

quantity ot the last syllable.

Quotients
E),

from a comparison of columns

and C (= table

denoting

ergo the preference shown for the clausulae with short

final syllable,
F),

and from a comparison of columns

B and D (=

table
It

denoting the

same

for

the

forms with long

in a

final syllable.

appears that some clausulae are favoured

ending
in a

higher degree

when

long than in a short final syllable.

192

1-8 9-16
17-24 25-32 33-40 41-48 49-56 57-64 65-72 73-80 81-88 89-96 97-104 105-112 113-120 121-128

wwwww .www wwww; www ww_ww w_ww ww ww ww__w www www w_w ww w

1 1 1 i l
1

(+) (+)

2*0 (+)
2-1
1*7

Preference for

over

w www

(+)

(-) (-)
(+)

(-) (-)

1-3 (+)
l'l (+) 2-8 (+) 2-2 (+)

l
i

j
!

3 (+) (+) 2 (+) 8 (( 5 (5

7

1*5 (+)

4-3 2.3 2-2

1*4

(-)

i
1 !

4
4

(((((--

( (

(-) 11 (-) 1-3 (-) 1-5 (-)

PLATO.
Some
it

of the criteria the

occurrence of which makes

probable that a dialogue belongs to the second chronogroup, which comprises the Republic, the Theae-

logical
tetus,

and perhaps the Phaedrus.

in

indicates the occur-

rence

the

Republic,

Ph

in

the

Phaedrus,

in the

Theaetetus.

See Barwick's

tables.

(Barwick, table on page 56/7).

1

dedax;, dgfioraxa, oQdoiara Myeig

and

ahj&EOTara.
2.

viziXapeg in

answers
(

3.

Jiavtdjiaoi juev ovv

v w
-

T Ph R T R T Ph R

193
4.
5.

dvayxaicog, dvayxaioxaxa, etc.

drjXov
ijuol

6.
7.

yovv doxei

R T Ph R R T

yaQ ovv in short

answer

8. 9.

ovxovv or dXXd XQV

xal Jtcbg xal Ticbg av;
XL
rj

10.

11.
12. 13.

fjLYjV;

T T T T T Ph

Jiobg;
fj

Ttfj;

Jtcbg;

14.

R R R R R R R R

(Bai rwick, table

15.
16.
jzcbg
Ticbg jzcog

on page

59).

yaQ;

ydg av;
d'ov fxellei;

17. 18.
19.

xal n(bg; xal Ticbg av;

/

20.

%i

o ov;

83

3/

21. 22.
23.
24.

xi
xi

yaQ ov;
yaQ;

aXXa, xi /ueXXei;
alio, xi fxr]v;

R R R R T R T Ph R R T Ph R R R

(Barwick, table on page 62/3).

25. 26. 27.

frequent
/UaXQCp Or

drjXov d>g
jLLVQLO)

('.

dfjXov oxi)

Ionic datives

T Ph R T R Ph R

194
28.
eigrjiai

fin zurtickverweisenden
Relativsatzen)

29. 30.

iQQJj&tj (item)

xa&djzeQ

('.

ojojieq)

31.

ovzcog (:rcp Svn)

T Ph R T T Ph R T Ph R

PLUTARCH.
Comparison
of his clausula with his sentence-metre.

h I S

Form.

in

Favoured
o 3
forms.

17-18 81-96 21-24 1_8 9-16 35-36 69-72 37-40 105-112 53-56 41-48 97-100 57-60 61-64 57-64 73-80 113-120 121-128 113-128 101-104 91-92 51-52

www ww*-* www w^__^

w w

1-10%
13-90 3-70 2-00

1-10

10
2-1
l'O 1-6

www; www.

_w___w

^ ww w
_n

610
2-10

2910 385 305 995

1-70

(+ (+ (+ ((((+ ((-

1-6 (+' 1-2 1-2

580
4-io 6*50

4-70

w_ww
w

260
9-80 2-35 3-30 2-40
2 05

1-6
1-5

400
7-20 2*50
3-6o

1-7

w__ w ww w
w w
s-:

2-2
1-8

TO (1-4

320 680
5-io

225
4-30 4-70

((-

16(11 (1-8 (1-2 (1'5(1-6 (2-9 (+'

630 510
11-40

360
4-20

780
2-45 4-05 0-85

w w_ ww ww

390
1-40

2-40

2 8

195

Syn crises.
(246 cases).

14 58 916
1718 1920 2124

% 2
2

Lives
13'9

%
(Clausula).

Syncrises

(sentence).
3 8
1

61

25-34 3536 3740 4148 4952 5356 5764 6568 6972 7380 8196 97100
101

65

2910 995 980

(clausula). 2-70 9*0

90

2 2 2

4
2
1

6
5

27
3 2

104
128

105112 113120
121

9 4 4

196

PHILO.
Comparison
0)

of his clausula with his sentence-metre.

<ti

c
Jt.
1

Form.
B

3 3

S-I

Favoured
forms.

Avoided
forms.

<U

>

w w w^ w_-_w^ w^^w ^-^~ ^w www w_ ^w w 105-112 w ^w 53-56 41-48 w _w w 97-100 www ww 57-60 w ww 61-64 ww 57-64 w 73-80 w 13-120 w 121-128 113-128 w 101-104 ww w 91-92 w ^w 51-52 w_
1

17-18 81-96 21-24 1-8 916 35-36 69-72 37-40

ww>

ww:=:

0'4% 20
12'7 2-3

50
14
1-2

180
2-8

(+) (+)

30
5-9 2-3 3.2
3.1

10
9'8

3-0
1*7

(+) (-) (+) (+) (+) (+) (+) (+) (-)

(+) (+) (+) (+) (-)

w s_^^
^^^w

v^,

wwwwW

06 62
6*2

3-8

19

20
16
1'2

98
4-0 6-4 2*8 2-8 4-8 7.6 6'4

160
4-6
1-4

46
1-6

46 36
5-4

w w wW www w
-

-^ ^r:
<

--

V^/

^~-

ww

13
11

9-0

12

'

6 5
6-4

20 30
T6

32
2-2

w-^

w_

120
4-2

46
16
3-0

4-0 2-8

(-) (-)

(-)
(-)

26
1-1

20
|*7

15 (+)
(+)

w w

_^ w

w>

18

METRE AS A DISTINGUISHING CHARACTERISTIC OF GREEK PROSE-WRITERS.

Sentence-metre (%)
Clausula
(0/
).

TABLE OF CLAUSULA WHICH HAVE BEEN FAVOURED (OR AVOIDED) BY DIFFERENT AUTHORS.
Favoured.
(1)

Avoided.

Thrasymachus.
Thucydides.

_.__^

(2)

_ _^.

^
wwic

Sum

total of
:

cases

in-

(3)

Plato (Laws).

^^^~

vestigated

1000 1000 1000 1000 100011000 2000 898 2000 3778 3781 500

338

866 1646

16

5-6

5-4

916 3740 8196

105112

36 29
144

4-0

20
2-8
7-8

81 61 41 124 109 139

1-7 1-2 4-4

106

68 19
75

8.9 5.9

31
12-7

26 13-0 9.2 18-8 108 5-2 06 100 6-2 12-6 98 6-2 21 36 2-6 142 18-9 29-1 13-9 5-7 181
8-5

50 129 45 09 5-9
17-0 21-3
3-3

14-3
(4)

(~
Isocrates.
11-6
(5)

etc.)

4-1

4-8

97100 6572 2124 1718 5356 6164 117120 3536 9192 123124 7380 113-128 121128 125128 127128 4148
101

88 25
2-7
1-8

80
0-1

2-4
3-1 1-7

3-8

6'5 2-5

9.8

9'4

8*0

9-8 61
2.4

6-0

16-0 12-5 17-6

161 162 76 100

Demosthenes.
Philo.

11

31
0-1
4-1

40 20
10 31

58 37
11

05 30 54 60
VI

08
2-5
5-1

28 32 23 04
4-0

31

08
T2 31

46
6-2 2-8

(6)

35
2-0

47 39
11

60
12
o-o
1'8

05
2-6

51
1-4 1:4 3-1

46 59
0-4

54 50
1-3

40 32
3-5

02 40
4-8 4-3

48
3-4

3'4
4'9

24 23 20
1-6

20 46
5-4

7-2

5-8

5-6

33

41

21
1-4

23 24 65
20-4

18 32
12-7 19'9

2-0

23 20

13

23
2-4 7'9

56 258
11-4 7'6

21 2'1 14 52 5 64 223 6-3 12-9

1

08 30 33
117

20 06
3*0

09
7-7

41
0-8

06
70
1-3
8-2

47 42 25
10
3'3

100 167 9-0

13-3

5-1

62
4-5

4-1

80
5-2

3-2
1-1

6-4 3-6
1-2 6'4
4-2

18 3 18-7 9'7 112 54 5-8

7-8 120 14-2

2-0 113 2-0 4-6 185 3-0

1.6

(7)

Flavius Josephus.

w>~,w:

62
8-5

0-1
(8)

1-0

Charito.

^^^^

2-0

30
3-7

32
4-8
3'7

0-6

104

51

51

67
7-2

3'7

7-2

42

1-4

1-5

32

5-5

2-6

26
1-5

5152

37 15

26
1-2

39
2-4

6-1
1.1

17

17

31

25 09

16 18

51

(9)
') ')

Plutarch.

Figures derived from Heibges, for Bell. Jud. and Antiquit. Figures derived from Heibges.

(10)

Polemo.

197

TYPOLOGY OF THE GREEK CLAUSULA.

Only
help
intend
of
to

few features
material

of typology in their

main outline with the

little

have been studied.

The

following tables

illustrate

my

hypothesis that in Greek prose

factors

metre

is

more important than those rhythmical

tomed
to

which we are accus-

include

under the idea typology.

The

results

seem

to

corroborate this hypothesis.

For

it

can be said that within the

metrical groups taken into consideration here, viz.

wwori, ww-we.g.:

:=n,

the proportions of the typological subdivisions,

are nearly the same.

Though

it

may be
first

that the appearing diffe-

rences point to some underlying tendencies, the main rule holds

good

that

it

is

desirable to study

the metrical structure of the

sentence,
really

assuming as an hypothesis that the metrical groups are

rhythmical unities in themselves, and after that to take the

I

typology into account.

intend to

come back

to these

problems.

FORM
Form
last

w ^
Philo.

of the

Thucydides. Demosthenes.

Plutarch.

word:
2 15

^ ^ ^ -^ ^^

%
2-6

%
3
1-6

%
O'O

%
6 10 47
15 10
0-0

198

43
10

566
131

53 117
16 3

276
60-9
8'3

12

200
38*3

122
57-3
18'3

23
21

350

79
100*0

76

192

16 1000

4 60

67
1000

82

122 1000

198

word

Thucydides. Demosthenes.

Philo.

Plutarch.

ends:

^
-

(,)

% W
~
i=i

%
3 53

%
00
12

%
10

jir;

w ^ ^ ^

Wj ,^

15

50 33 6
106

19 142 472 311

5-7

12

130 57
2

215 529 232

12
lOO'O

15 8

00 90
487 333

26 34 4
76

342 447

54 37
10
111

53
1000

90
1000

lOO'O

245

FORM -x
Form
last

w.
Philo.

of the

Thuc ydides. Demosthenes.

Plutarch.

word

w _^
S

%
00
1
1

%
42
o-o
3 18
1

%
4
5

%
00 00
4 25 4

4-2
16-7

00 89
111

>

^~^

4
17

125
75-0
4-2

121

w-^w,;^

^^

S~-s

708
8-3

w^w^
word

27 8
1

600
17-8

758
121

00
24
lOO'O

4-2

22
1000
33

00
100-0

24

1000

45

A
-~-

Thuc ydides. Demosthenes.

Philo.
/o

Plutarch.

ends:
;=

^ w w ^^^i: v^ v_^,-^^^: ^ ^^-^=:

ww%-^=: s~*w,v_*>=:

(,)

% 00
1

% 28
00 83
55-6 30-6
2-8

o-o

% 00
00
4 26
17
8-5

2-8
1

4
5 31 14
1

v_*nX=
:

14

20
10

57'1

20
11
1

73 91 564
25-4
1-9

28-6
o-o

55'3 36-2

00
47

35

1000

36

1000

55

1000

1000

FORM
Form
last

-_Philo.

of the

Thucydides. Demosthenes.

Plutarch.

word:

!ri

% 0*0
9
12

% 36
!

%
1
!

1-5

% O'O
6
3

__^

^ w

<-;

^ w w

20
2

209 279 465

4-7

30
25
18 7
1

35-7

298
21*4
8-3

17 14

29
5

25-8 21*2 43*9

146

27
5

73 659
12-0

12
100*0

76 00
1000

o
41

o-o

43

1000

84

66

100-0

199

word
(,)

Thucydides. Demosthenes.
p=
i=i

Philo.
1

Plutarch.

w w
_v_x

ends:

% 00
9
13
14*8

%
3
2*0

% VI
18-1

% 00
6 8

w
iri
i=l

21-3

,w
-^

26
13

-^

426 213

30 38 32 47
1

198
25'2 21-2
31-1
1

17 14

98
131

149
45-8 20-2 0*0
100-0

43
19

30
17

492 279

00
61

0-7

00
61

1000

151

1000

94

1000

CONTRIBUTIONS TO A BIBLIOGRAPHY OF ANTIQUE PROSEMETRE AND ANTIQUE PROSE-RHYTHM. *)

Adams, C. D. De periodorum formis et successionibus mosthenis oratione Chersonitica. Diss. Kiel 1891.
Ahlberg, A.
p.

in

De-

W.

Prolegomena

in

Sallustium.

Goteborg 1911,
distinxit C. U.

175, n.

1.

Ammianus
Clark
etc. vol.

Marcellinus.
1,

Recens.

rhythmiceque

1910. G. Bericht iiber die Literatur zu Ciceros rhetorischen Schriften aus den Jahren 19051909. (Bursian's) Jahresbericht. Bd. 143, 1910, p. 112175. [on Zielinski], Berl. Philol. Woch. 1918, col. 482 sqq. [on De Groot], Berl. Philol. Woch, 1918, col. 673 sqq. Arndt, E. De ridiculi doctrina rhetorica. Kirchhain 1904. Arnold, C. P. Caesarius von Arelate. Leipz. 1894, p, 85, 46890. Ausserer, A. De clausulis Minucianis et de Ciceronianis quae quidem inveniuntur in libello de senectute. (Commentationes Aenipontanae I). Ad Aenipontem 1906. Bachmann, A. Aiax et Ulixes declamationes. Diss. Minister 1911.

praefatio. Berlin

Ammon,

Baehrens,
praefatio

W.
1910.

A.

maior.
n. d.

Accedit

Plinii

Panegyricorum Latinorum Panegyricus exemplar

latini.

editionis
editionis.

novae
Diss.

Groningen

Panegyrici

Leipz.

1911, p.

XXIX.
Rhein.

Zu Florus. Wiener Studien 1912, p. 402 sqq. Zu den philosophischen Schriften des Apulejus.

Museum
Compare

N. F. 67, 1912 sqq., 264 sqq. _. Zu Curtius Rufus. Rhein. Kroll, Glotta 8, p. 326.
.

Museum

68, 1913, p.

429 sqq.

Erwiderung. Berliner Philol. Wochenschr. 1918, col. 311. Vermischte Beitrage. Berliner Philol. Wochenschr. 1918,

col.

501 sqq.

Bainvel, J. V. La prose metrique et la prose rhythmique. Etudes religieuses 59, 1893, p. 143155. Ballou, S. H. De clausulis a Flavio Vopisco Syracusio scriptore historiae Augustae adhibitis. Diss. Giessen. Weimar 1912. Bayard, L. Le latin de Saint Cyprien. These Paris 1902, p. 298, sqq. Becker, H. Hermogenis Tarsensis de rhythmo oratorio doctrina.
Diss. Miinster 1896.

) For additions nally been inserted.

and systematic index see

vol.

II.

Reviews have only exepti

201
Bellet, C. P.
Paris 1897.

L'ancienne vie de saint Martial

et la

prose rhythmee.

Les
copaux.

Nom.
.

ed.

origines de l'Eglise de France et les fastes epossuivie d'une etude sur le 'cursus' et la critique.

Paris 1898.

La

prose

rhythmee

et

la

critique

hagiographique.

Paris 1899.
.

L'age de la vie de saint Martial.

Revue des questions

historiques 58, 1900, p.

1908, p.

540, 3138.

Beltrami, A. II 'numerus' e Frontone. Rivista di Filologia 36, 545566. Bickel, E. Die Schrift des Martinus von Bracara, Formula vitae honestae. Rhein. Museum, N. F., 60, 1905, p. 516520. Der Prosarhythmus. In: Gercke-Norden, Einleitung in
.

die Altertumswissenschaft,

2.

Leipzig

etc.

1912, p. 612 sqq.

Bigelmair, A. Blacha, F. v.
clericorum' ein

Zeno von Verona. Miinster 1904, p. 131. Der Pseudocypriamische Traktat 'de singularitate
des Novatian. Diss. Breslau 1913.

Werk

Blass, F.

Zeitschr. fur das

Gymnasialw. 34, 1880, p. 264. Rhein. Museum 33, p. 493 sqq. Attische Beredsamkeit 2 Leipzig 1887-98. De numeris Isocrateis. Univ. Progr. Kiel 1891. La prose metrique et la dialogue des orateurs.
.

Revue

de

Phil. 23,

1899, p. 334.
.

Lycurgi oratio

in

Leocratem
Literaturz.

etc.

Leipzig 1899.
1865.

Compare
Berliner
p.

Fuhr, Deutsche
Philol.

1899,

Drerup.
Boficl 1899,

Wochenschr.

1900,449-52.

C m
i

a,

73-76.
.

Die Rhythmen der attischen Demosthenes, Platon. Leipzig 1901.

.

Kunstprosa.

Isokrates,
36,

Kritische

Bemerkungen zu Platons Phaidros. Hermes

... V.

1901, p. 580 sqq.

Album gratulatorium
Observationes
pertz 1902, p. 53.
.

Herwerden

1902, p. 22.

in Platonis

convivium. Festschrift

Gom-

Die

Reden.

Neue Jahrbiicher
.

Textiiberlieferung in 9, 1902, p. 708.

Demosthenes' olynthischen

Die Rhythmische Komposition des Hebraerbriefes. Kritiken 75, 1902, p. 420. (Barnabas') Brief an die Hebraer. Text mit Angabe der Rhythmen. Halle 1903. Die doppelte Form der 3. Phjlippika des Demosthenes. Neue Jahrb. 13, 1904, p. 486. Die Rhythmen der asianischen und romischen Kunstprosa. Leipzig 1905.
Theol. Studien
u.
.

202
.

On

attic

Zur Rhythmik

prose-rhythm. Hermathena 14, 1906, p. 18. in den neutestamentlichen Briefen. Theol.

Korintherbriefen.

Studien

u.

Kritiken 79, 1906, p. 304. . Textkritisches zu den

Beitrage zur

Forderung

christlicher Theologie 10, 1906, p. 31. Compare Woch. klass. Phil. 1906. col.

Draseke,

1259.

Bonwetsch, G. N.
Diognet.

Der Autor der Schlusskapitel des Briefes an

Nachrichten der kgl. Ges. der Wiss. zu Gottingen 1902, p. 62134. Quid de structura rhetorica praeceperint grammatici atque rhetores latini. Thesis. Parisiis 1898. La prose metrique et le Dialogue des orateurs. Revue de philol. 23, 1899, p. 334342. Les lois metriques de la prose oratoire latine d'apres Revue de philol. 24, 1900, p. 201 sqq. le panegyrique de Trajan. Les lois metriques de la prose oratoire latine d'apres le Brutus. Revue de Philol. 26, 1902, p. 105. La rhetorique a Herennius et les clausules metriques. Melanges Boissier, p. 7379, 1903. Les clausules metriques dans Minucius Felix. Musee Beige 7, 1903, p. 247-265. Les clausules metriques dans Florus. Musee Beige 7, 1903, p. 16-36. Wie soil man die metrischen Klauseln studieren? Rhein. Museum 58, 1903, p. 371 sqq. Revue de Les clausules metriques dans l'Orator.

Bornecque, H.
.

philol. 29,

1905, p. 40-50. Les clausules metriques latines. Lille 1907. Compare P i c h o n, Revue critique 1908, p. 38789.
.

Bourgery, A.
34, 1910, p. 558 sqq.

La prose metrique de Seneque.

tonique.

Revue de

Philol.

Bouvy, E.
.

Rhythme

These Paris 1886.

Poetes et melodes. Nimes 1886. Bresslau, H. Handbuch der Urkundenlehre fur Deutschland und Leipz. 1880, p. 588 sqq. Italien, I. Brinkmann, A. Klassische Reminiszenzen. Rhein. Museum 60,
1903, p. 631.

Brugnola, V.
di filol.
etc.

39,

Sulla clausula Ciceroniana 'esse videatur'. 1911, p. 558, and in 'Arpinum' 1, 1912, nr. 1.

Rivista

Burdach, K.
Sitz.-Ber.

der

Burn,

Ober den Satzrhythmus der deutschen Prosa. preuss. Akad. 1909, p. 52035. [On Latin prose-rhythm]. A. E. An introduction to the Creeds and to Te Deum.
p.

London 1899, Candel,

240-42.

J.

De

clausulis

Paschale opus adhibitis.

These Montpellier.

Compare Curcio,

a Sedulio in eis libris qui inscribuntur Toulouse 1904. Rivista di filol. e d'istruz. class. 34, 1906, p.

203
critique 1905, 2, p. 478 sqq., 1905, p. 331. Ceci, L. II ritmo delle orazioni di Cicerone. I. La prima Catalinaria. Testo con la scansione delle clausole metriche. Torino etc. 1906. Compare Curcio, Rivista di filol. e d'istruz. class. 34, 1906, p. 349 sqq. Christ, "W. von Qriechische Literaturgeschichte 6 I, Munchen

349 sqq.

Lejay, Revue
9,

Hubaux,

Musee Beige

1908, p. 562 A. 9. Cicero. Brutus, erkl.

.

notes by

J.

von O. Jahn-W. Kroll. Berlin 1908. Orator, with introd. essays and critical and explanatory E. Sandys. Cambridge 1885. De oratore, erkl. von K. W. P i d e r i t. Leipzig

1886-1890.

The correspondence
L.

of

C.

etc.

by
.

R. Y.

Tyr

a.

C.

Purser.

Dublin etc. 1890-1906. Vol. 112, p LXVII note. Ciceronis orationum scholiastae, Rec. T. Stangl.
,

Leipzig 1912.

Clark, A. C.
p.

Zielinski's

Clauselgesetz.

Class.

Review

19,

1905,

164 sqq.

Cluniacensis of Poggio.
.

Anecdota Oxoniensia. Class, series, 10. The vetus Oxford 1905. Fontes prosae numerosae. Oxford 1909. Compare Amnion, Berliner philol. Wochenschr. 1910, p. 1605 1607. S e r r u y s, Revue de philol. 34, 2, p. 306. The cursus in mediaeval and vulgar Latin. Oxford 1910. Compare Wessely, Wochenschr. fur klass. Philologie 1910, p. 95051. Roberts, Class. Rev. 25, p. 57. B e r m a, Museum 18, p. 247-49. Prose- rhythm in English. Oxford 1913. See Oxford Magazine April 24, 1913. Recent developments in textual criticism. Oxford
.

1914, p.

15 sqq.

Clark,

C.

U.

Ammiani

rerum

gestarum

quae supersunt.

I.

Berlin 1910, p. VII.

It. Le cursus ou rhythme prosaique dans la liturgie de l'Eglise latine du 3 me siecle a la renaissance. Revue des questions hist. 51, 1892, p. 253261. (= Compte rendu du congres scientifique international des catholiques, 5 me section, Paris 1891 p. 103109). Enseignement. Toulouse-Paris 191 1, p. 182-84, 189-213. Le rhythme des oraisons liturgiques. Musica sacra (de Toulouse) 15, 1892, p. 19-21.

Couture,

et la litterature

Encore

la question

du cursus.

Musica sacra (de Tou-

louse) 16,

1893, p. 9-14.

Cronert,
N.
F., 54,

W.

Zur griechischen Satzrhythmik. Rhein. Museum,

1899, p. 580 sqq. [on Haury], Gott. Gel. Anzeigen 1916, p. 391.

204
Ciceronis et Calvi reliquorumque Atticorum genere Acide prope Catinam 1899, p. 5969. Dabin, A. Nos oremus. Paroles et chant. Revue du chant gregorien (de Grenoble) III, 1894-95, p. 106-11, 123-27. Dechevrens, A. Composition musicale et composition litteraire. Paris 1910, p. 118-168.
dicendi quaestiones.

Curcio, C.

De

Groot, see Groot De. Jonge, see Jonge De. Dewing, H. B. The origin
Greek.

De De

of the accentual prose rhythm in American Journal of Philology 31, 1910, p. 312328. Compare Maas, Byzant Zeitschr. 20, p. 306. Accentual cursus in Byzantine Greek Prose with
.

reference to Procopius of Caesarea. Transactions Connecticut Acad, of Arts a. Sc. 14, 1910, p. 415 sqq. Hiatus in the accentual clausulae of Byzantine Greek American Journal of Philol. 32, 1911, p. 188 sqq. prose. Di Capua, P.*) [on Shipley and others], Bollett. die filol. class.
especial
.

18,

1911-12,

p.
.

244

sq.

e le varianti del
.

Le clausule metriche nelf Apologetico di Tertulliano Codex Fuldensis. Monza 1912. From Scuola Cattolica 40. De numero in vetustis Sacramentariis: quae instrumenta

adiumenta ad clausulas efficiendas in Sacramentariis adhibeantur. From Ephemerides Liturgicae 1912, p. 459-476, 526-535, 591-600.
vel

Osservazioni
Tertulliano.
i

critiche
19, in

Bollett. di
.

filol.

class.

sul testo dell' 1912, p. 59-61.

Apologetica

di

Le clausole
Bollett.

S.

due accenti.

di

filol.

class.

Agostino con tre 1912-13, 19,

sillabe atone fra

p.

12 sqq.

[On

Terzaghi, Ciceronis

Glotta 6, 1915, p. 379 sq e le clausole metriche da osservarsi nella riforma et nella compilazione degli oremus e delle prose liturgiche. Monza 1913. From Scuola cattolica 1912. Una glossa in Quintiliano, Inst. Orat. IX, 4, 105. Boll. di filol. class. 19, 1912-13, p. 207. Cassiodoro, De institutione divinarum litterarum C. XV.
II

De Compare Kroll,

re publ. IV, 11].

'cursus'

19, 1912, p. 89 sq. L'evoluzione della prose metrica latina nei prima tre Didaskaleion 2, 1913, secoli D. C. e la data dell' Ottavio di Minucio. p. 1-41. Minucio Felice, Octavius, 7, 4. Didaskaleion 2, 1913. [On Ballou, Laurand, Raspante], Bollett. di filol. class. 20, 1913, p. 129 sqq., 202 sqq., 277 sqq.

Boll, di

filol.

class.

*)

On Di Capua, compare Laurand, Musee

Beige,

Bullet,

bibl.

18,1914,

p. 10 sqq.

205
cursus nel De consolatione philos. e nei trattati Didaskaleion 3, 1915, p. 269303. Le numerus dans le Breviarium d'Eutrope. Bollett. di class., 23, 1916, p. 17-18. Dostler, G-. Das Klauselgesetz bei Curtius. Kempten 1907. Draheim, H. Lateinischer Prosarhythmus. Wochenschr. fur klass.
II
.

theologici di Boecio,
filol.

1910, col. 1294-1302, 1353-1358. Die Anfange der rhetorischen Kunstprosa. Jahrbiicher fur class. Philol. 27, 1903, p. 234 sqq. Untersuchungen zur alteren griechischen Prosaliteratur. Jahrbiicher suppl. 27, 1903, p. 219. Compare Schwartz, Bed. Philol. Woch. 1903, p. 97; Thiele, Wochenschr. klass. Phil. 1905, p. 284.
Philol. 27,

Drerup, E.
.

Droz, E.
sancon 1885.

De

Frontonis

institutione

oratoria.

These

Paris,

Be-

Duchesne.
des chartes 1889,

p.

Dupuis,

I.

Note sur l'origine du cursus. Bibliotheque de l'ficole 16163. Le nombre oratoire. I Contra la 'prose metrique' II

Theorie de Ciceron, Roanne 1907.

Engelbrecht, A. Stilfragen bei lateinischen Autoren in ihrer Nutzanwendung auf die Kirchenschriftsteller. Zeitschr. fur die osterr. Gymnas. 53, 1902, p. 1-20.
Ernst. De genere dicendi et compositione rhetorica in prioribus Ciceronis orationibus. Progr. Neu-Ruppin 1885. Eysert, L. Zum Rhythmus bei Demosthenes. Zeitschr. fur die
osterr.

Gymnas.

45,

1894, p. 720-21.
II

Ferretti,
riane.

P.

cursus metrico

il

ritmo della melodie grego-

K.

De errore profanarum religionum. Edid. Praefatio p. XXV. Compare Baehrens, Rhein. Museum 67, p. 128. Freund, J. De C. Suetoni Tranquilli usu atque genere dicendi. Diss. Breslau 1900, p. 39 sqq. Fritzsch, F. V. De numeris orationis solutae. Progr. Rostock 1875. Fuhr, K. Rhein. Museum 57, 1902. p. 426 sqq. Bed. Philol. Wochenschr. 24, 1904, col. 1030 sq. n. 5. G-abrielli, A. L'epistole di Cola di Renzo e l'epistolografia Archivio della R. Societa Romana di ivale. p. 381-379 (p. 391-395, 420-422 on the cursus). Gastoue, A. Cours theorique et pratique de plainchant romain gregorien. Paris 1904. p. 189-193. Les origines du chant Romain. Paris 1907, p. 18691. Gatard, A. 'Accent' in Dictionnaire d' archeol. chret. I. Paris

Roma-Parma 1913. Pirmicus Maternus.

1908.

Ziegler,

W.

1907, p. 220-40.

Gatscha, F.

Quaestionum Apuleiarum capita

tria.

Dissertationes

206
Vindobonenses VI, 1898, p. 139-190; see p. 159-176. J. van. De statistiek en de taalwetenschap. Nieuwe taalgids 9, 1915, p. 65-95. Giry, A. Manuel de diplomatique. Paris 1894, p. 454462.
philologicae

Ginniken,

Gladitsch, J. De clausulis Quintilianeis. Diss. Breslau 1909. Golz, G. Der rhythmische Satzschluss in den grosseren pseudoquintilianischen Deklamationen.
Diss. Kiel 1913.

Glotta 8, p. 326. de. Vorlaufige Mitteilung zu Zielinskis constructivem Rhythmus, S. 5 ff. Berliner Philol. Wochenschr. 1914, col. 1054-55. Notes on Procopius of Caesarea. Class. Quarterly 1915,
r
1

Compare K Groot, A.
,

1,

W.

p.

97 sq.

prose.

Methodological investigations into the rhythm of Greek Class. Quarterly 1915, p. 23144. ii n s c h e r, Bursians Jahresbericht 1915, Bericht fiber Compare die Lit. z. zw. Sophistik 1910-15, p. 148 sq. Prosarhythmus. Berliner Philol. Wochenschr. 1915, col.
,

1135 sq.

Eine neue Methode der Klauselforschung. Berliner Wochenschr. 1917, col. 1158 [on Novotny]. Untersuchungen zum byzantinischen Prosarhythmus (Prokopios von Casarea). Groningen (Holland) 1918. See Ammon. Grospellier, A. Le rhythme des oraisons. Revue du chant gregorien (de Grenoble) 5, 1897, p. 102-176; 6, 1898, p. 38-43. Hanssen, F. Ein musikalisches Accentgesetz in der quanti,

Philol.

tierenden

Poesie

der

Griechen.

Rhein.

Museum.
dicendi

N.

F.

38,

1883, p.
in

222-244.

Harendza, G.
epistolis

usus

sit.

De oratorio genere Oratislaviae, p. 5965.

Glotta 4,

quo Hieronymus

Harkness, A. G.
poetry.

The first monosyllable in Latin prose and Americ. Journ. of Philol. 31, 1910, p. 154174.
p. p.

Compare Skutsch,

429.

Harmon, A. M. The clausula Haven 1910. (= Transactions of the

sciences 16, 1910, p. 117-245).

in

Connecticut

Ammianus Marcellinus. New Academy of art and

Munster 1912.

Hartmann, M.
Havet,
J.

De

clausulis Libanianis, Diss.

Questions merovingiennes VII. Les actes des eveques du Mars. Bibl. de l'ecole des Chartes 54, 1893, p. 597-692 (esp. p. 639-645). Havet, L. Prose metrique. Paper in the Grande Encyclopedic Paris s.d. 27, p. 804-806. Byz. Zeitschr. 8, p. 535 sqq. [On W. Meyer, Der accentuierte Satzschluss], Revue critique 32, 1891, p. 207 sqq. La prose metrique de Symmache et les origines du
, ,

cursus.

Paris 1892.

207
Compare Bornecque, Revue de Philol. 18, Revue critique 53, 1893. W. Meyer, Gotting.
1893.

Lejay, 1894. Gel. Anzeig. 1,

17,

Cicero,
sqq., 141 sqq.
,

De

Oratore.

Revue de

Philol.

1893, p. 33

Many

editions.
,

Cours elementaire de metrique Paris 3 1893. 517 sqq. La prose metrique de Martial.

grecque

et latine etc.

Revue de

Philol. 27,

1903, p. 123-124.
,

La prose de Pomponius Mela, Revue de

Philol. 28,

1904, p. 57-59.

Manuel de critique verbale appliquee aux textes latins. Paris 1911, pag. 89-96. [Letter adressed to Stangl], Wochenschr. f. klass. Philol. 29, 1912. col. 1357-58. Heibges, S. De clausulis Charitoneis. Halis Sax. 1911, p. 28 sqq. Heisenberg, A. [on Haury's edition of Procopius], Berliner Philol. Wochenschr. 1909, 7 Aug. Heitmann, M. De clausulis Libanianis. Diss. Miinster 1912. Compare as, Wochenschrift fur klass. Philol. 30, 21, p. 576 sqq. [Valuable review]. Hendrickson, G. L. The commentariolum petilionis attributed Decennial Publications VI, Chicago. to Quintus Cicero. Compare Jahresbericht des philol. Vereins zu Berlin 31, p. 287. Accentual clausulae in Greek prose of the first and second centuries of our era. American Journal of Philol. 1908, p. 280 sqq. Hermann, G. Opuscula I, p. 121 sqq. Hoflinger, J. Bobiensia. Diss. Wiirzburg 1912. Hofacker, K. De clausulis C. Caecili Plini Secundi. Diss. Bonn 1903. Hoppe, H. Syntax und Stil des Tertullian, 1903, p. 154 sqq. Hugh, Pitz Th. Aristotle's theory of rhythm. Proceed, of the Amer. Philol. Ass. 44, 1914, p. 23-26. Immisch, O. Sprach- und stilgeschichtliche Parallelen zwichen Griechisch und Lateinisch. Neue Jahrb. fur das klass. Altert. 29, 1912, p. 31 spp. Neue Wege der Platonforschung. Neue Jahrb. 1915, p. 270. Johnson, C. L. The motion of the voice in connection with accent and accentual arsis and thesis. Studies in honour of Gildersleeve. Baltimore 1902. Jonge, E. de. Les theories recentes sur la prose metrique en latin. Musee Beige 6, 1902, p. 262-79. Les clausules de St. Cyprien. Musee Beige 6, 1902,
,

Ma

W.

p.

344-63.

208
.

Les

clausules
in

metriques
di

dans
e

St.
d'istr.

Cyprien,
class.

Louvain
34, 1906,

etc.

1905.

Compare
p.

Curcio

Rivista

filol.

349 sqq.

Jordan, H.

Rhythmische Prosa

in

der altchristlichen lateinischen

Literatur. Leipz. 1905.

Compare Lejay, Revue

.

critique 1905,
in

Gibt
Krit 79,

es

Rhythmik

II, p. 480 sq. den neutestamentlichen Briefen?

Theol. Stud.

1906, p. 634. der altchristlichen Literatur. Leipz. 1911, esp. p. 57, 68, 230, 372, 452, 502. Josephy. Der oratorische Numerus bei Isokrates und Demosthenes mit Beriicksichtigung der Lehren der alten Rhetoren. Diss. Zurich 1887.
u.

Geschichte

Kaibel, G.
p.

Stil

und Text de

Uohxda

'Adijvalcov. Berlin 1893.

89

sqq.,

94 sqq.

Kalinka. [On Zielinski], Algem. Literaturblatt 1909, p. 69. Kallenberg, H. 'On und cog bei Plato als Hilfsmittel zur Bestimmung der
Zeitfolge seiner Schriften. Rhein.

Kaluscha,

W.

Museum 1913, p. 465 sqq. Zur Chronologie der platonischen Dialoge. Wiener

Studien 26, 1904, 2 Heft. p. 190 sqq. Kauer, It. Studien zu Pacianus, II. Jahresber. des k.k. Staatsgymn. zu Wien XIII. Bez. 1902. Kawczinski, M. Essai comparatif sur l'histoire des rhythmes. Paris 1889. Kayser, S. Etude sur la lanque d'Hypercides. Musee Beige 4,
p.

95-104. 201-222.

Kirchhoff,
Lipsiae 1902.

A.

De

Apuleji

clausularum

compositione

et

arte.

Kirsten, C.
lungen.
fertur,
p.

Bd.

7,

Heft

Quaestiones Choricianae. Breslau 1894. 2.

Hilario oratoriae

Breslauer philol. Abhand-

Kling, H.
institutionis

De

Pictaviensi artis rhetoricae ipsiusque, ut Freiburg 1909, Quintilianeae studioso.

52-56.

Klotz, A.
65, N.F.,

1906, p.

Kroll,

W.
. .

die expositio totius mundi et gentium. Philologus 113 sq. Berliner Philol. Wochenschr. 23, 1903, col. 207. Berliner Philol. Wochenschr. 25, 1905, col. 1659 sqq. Moderne lateinische Syntax. Neue Jahrbiicher fur das

Uber

klass. Altert. 25,

.

1910, p. 318-326.

Die wissenschaftliche Syntax im lateinischen Unterricht.

Berlin 1917, p. 66 sqq. Literaturbericht, Glotta, vol. 5 sqq., 1914 etc. See Cicero. Krumbacher, K. Geschichte der byzantinischen Literatur
. .

2
.

Munchen

1897, p. 8, 29, 1137 sq.

209
Langen, P. Bemerkungen iiber die Beobachtung des Wortaccentes im alteren lateinischen Drama. Philologus 46, 1888, p. 409, n. 2. Laurand, L. Etudes sur le style des discours de Ciceron, avec une esquisse de l'histoire du 'cursus'. Paris 1907. De M. Tulli Ciceronis studiis rhetoricis. Paris 1907, p. 48-49, 57 n. 3, 101-2. Die Satzschliisse der Interpolationen in dem Traktat De Unitate Ecclesiae. Berliner Philol. Woch. 29, 1909, col. 1015-16.
.

De numero
Le
rythme

oratorio
oratoire
I

in

Ciceronis

orationibus.

Vox

Urbid

12,

1909, p. 18-19.
.

dans

Ciceron.

L'enseignement

Chretien 29, 1910, p. 17-20.

,

Le cursus du

au VII e

siecle.

L'enseignem Chretien,

29, 1910, p. 276-80.

Julien de Spire.

Bonaventure.

Le cursus dans la legende de saint Francois par Recherches de science religieuse I, 1910, p. 35158. Le cursus dans la legende de saint Francois par saint Rev. d'Hist. Eccles. 11, 1910, p. 257-62. Les fins d'hexametre dans les discours de Ciceron.
.

Revue de

philol. 35,
,

1911, p. 75 sqq.
les

Le cursus dans Ce qu'on

sait

hagiographes dominicains.
ce qu'on ignore du cursus.

Analecta

Bollandiana (1912?)
,

et

Musee

Beige

17,

1913, p. 91-107.
Classical

Compare Lindsay,
Jahresbericht

Review

1913,

Dec,

Luterbacher,

des

Philol.

Vereins

Thomas,
f.
,

Revue

klass. Philol.

La
[on

critique 1913, 1 1914, col 465. theorie du cursus dans saint Augustin.
4,

zu Berlin 1913, p. 278 sq., Nov. Strauss, Wochenschr.

Recherches
18,

de science religieuse
,

1913, 6, p. 56971. Di Capua], Musee Beige, Bullet, bibligr.

1914,
in

p.

10 sqq.

Lehmann Haupt,
Athen.

C. F.

Schatzmeister und Archontenwahl

Klio 6, 1906, p. 313 sqq. Lehnert, G-. Bericht iiber die rhetorische Literatur bis 1906. (Bursian's) Jahresbericht. Bd. 142, 1909, p. 226-340, esp. 237 sqq. Bericht iiber die Literatur zu Apuleius und zurHistoria Apollonii aus den Jahren 1897 1914. Bursian's Jahresbericht 44, p. 12 14. Lejay, P. [On Meyer, Blass, Jordan, Candel, De Jonge, Zielinski], Revue critique, N. S. 60, 1905, p. 478 sqq. [On Anonymi de rebus Bellicis liber, ed. Schneider 1908], Revue critique 1909, II, p. 289-91.
, ,

Lenchantin De Gubernatis, M.
Asinio Pollione.

II

ritmo in un frammento di

Ri vista di Philol. 37, 1909, p. 385-96.

210
Leo, F.
,

Die beiden metrischen Systeme des Altertums.

Hermes

24, 1889, p. 285 sqq.

De

Stati Silvis

commentatio.

Progr. acad. Gotting 26,

in

1892.

Lietzmann, H.
Bonn
style.

Funf Festpredigten Augustins

gereimter Prosa.

1905, p. 3.

Lipsky, A.
Diss.

Rhythm as
1907.

a distinghuishing characteristic of prose

New York

Litzica, C. Das Meyersche Satzschlussgesetz in der byzantinischen Prosa. Diss. Miinchen 1898. Compare Ha vet, Byzant. Zeitschr. 8, 1899, p. 535 sqq. Lofstedt, E. Zu Senecas Briefen. Eranos 14, 1914, p. 142 sqq. Arnoliana Lunds Univ. Arsskrift, N. F. I, p. 12, 1917. Compare Baehrens, Bed. Philol. Woch. 1917, col. 1291.
, ,

Erklarung.

Lorenz, Th. De Compare S k u t s c Luterbacher, P.

zu Berlin 31, 1905,

Berliner Philol. Wochenschr. 1918, col. 311. clausulis Arnobianis. Diss. Vratisl. 1910. h, Glotta 4, p. 425 sqq.

[On

Zielinski], Jahresbericht

des Philol. Vereins

p. 263.

Maas, P.
Manasses.
Berl. Philol.

Rhythmisches zu der Kunstprosa des Konstantinos Byz. Zeitschr. 11, 1900, p. 505. [On W. Fritz, Synesios, Abh. Bayer. Akad. 1905],

Woch. 26, 1906, col. 777. [On H. Usener, Der

heilige Tychon], Byz. Zeitschr. 17,

1908, p. 611.

[On Dewing], Byz. Zeitschr. 19, 1910, p. 592. Wochenschr. fur klass. Philol. 1911, col. 1254 sqq. Die Rhythmik der Satzschliisse bei dem Historiker
Prokopios.
Byz. Zeitschr. 21, 1912, p. 52 sqq. [On Dewing], Byz. Zeitschr. 21, 1912, p. 287 sq. [On Heitmann], Woch. fur klass. Phiiol. 1913, no. 21. Mace, A. Essai sur Suedone. Paris 1900, p. 379-400.
.

Mack, K.
Wien
1914.

Der numerus oratorius

bei

Demosthenes. Programm

Marbe, K. Ober den Rhythmus der Prosa, Progr. Giessen 1904. Martin, A. [on Blass], Revue critique 1903, I, p. 9. Marx, P. Ad Herennium libri IV. Leipzig 1894. Prolegomena
p.

99-102.
.

gr. u. lat. metrik.

May,

J.

neueren Forschungen auf dem Gebiete der fur das klass. Alt. 20, 1908, p. 233. Rhythmische Analyse der Rede Cicero's pro Sex. Roscio

Ueber

die

Neue Jahrbucher

Amerino, Leipzig 1905. Zur Kritik der Proomien des Demosthenes. Leipzig 1905.

211
uber die Literatur zu Ciceros Reden aus den (Bursian's) Jahresbericht 1907, II, p. 123 sqq. Rhythmen bei Demosthenes und Cicero. Wochenschr.
Bericht
Philol. u.

Jahren 190306.
f.

klass. Philol. 24,

.

45.

Vers, in

1907, p. 48-55. Verhandlungen deutsch. Bremen, p. 68.

.

Schulmanner auf der

Reden.
Progr.

Rhythmen

in

Ciceros

katilinarischen

Durlach 1909.

Rhythmische Formen, nachgewiesen durch Beispiele aus Cicero und Demosthenes, Leipzig Fock 1909. Compare A m m o n, Berliner Philol. Wochenschr. 1910, col. 156671, Bitschofsky, Zeitschr. f. osterr. Gymnasien 60, p. 752 sq. Rhythmen in Ciceros Reden. Wiss. Beilage zum Jahresbericht des Gymnasiums in Durlach 1912.
.

Compare
40-42.

Loschhorn,

Wochenschr.

fur klass. Philol.

1913, col.

Meader, C. L. Types of sentence structure of prose writers. Transact. Americ. Philol. Ass. 36. 1905. p. 32. Meister, L. Kritische Bemerkungen zu den Reden des Demosthenes nebst einem Anhang iiber Ciceros Rede in Pisonem. Progr. Durlach
Gymn.
1914.
.

Tullianae ad libros qui inscribuntur De Lipsiae [1912], p. 7273. Meyer, G. [On W. Meyer, Der accentuierte Satzschluss], Berliner Philol. Wochenschr. 1892, col. 182 sq. Meyer, Fragmenta Burana, Berlin 1901. Sonderabdruckaus der Festschrift zur Feier des 150-jahrigen Bestehens der kgl. Gesellschaft der Wiss. zu Gottingen 1901. p. 148 sqq. Abhandl. d. Gott. Ges. d. Wiss. 1901, p. 157 sqq. Das turiner Bruckstiick der altesten irischen Liturgie. Gelehrte Nachrichten, Phil. Classe 1903, p. 163-214, esp. 209. Gott. Gesammelte Abhandlungen zur mittellateinischen Rhythmik, Berlin 1905. Miller, C. E. The relation of the rhythm of poetry to that of the spoken language. Studies in honor of Basil L. Gildersleeve. Baltimore 1902, p. 497-511. Misset. [On N. Valois, Etude sur la rhythme des bulles pontificales], Lettres chretiennes [revue publiee a Lille], 5, 1882, p. 8990. Compare L a u r a n d, Etudes, p. 365, note. Mocquerreau. Le cursus et la psalmodie. Paleographie musicale 5, 1893, p. 27-30. Note sur l'influence de l'accent et du cursus toniques

Quaestiones

oratore pertinentes.

W.
.

W.

latins

dans

le

chant ambrosien.
J.

Ambrosiana, Milan 1897,

fasc. IX.

Moller,
Minister 1912.

De

clausulis a Q. Aurelio

Symmacho

adhibitis.

Diss.

212
Compare Forster, Class. Journal 8, 15, Tol ki eh n, Wochenschr. fur klass. Philol. 31, p. 10-11. Munch, V. De clausulis a Valerio Maximo adhibitis. Diss. Vratislav. 1909. Miinscher, K. (Bursian's) Jahresbericht, Bd. 70, 1905, p. 150-152.
.

Die Rhythmen

in

Isokrates Panegyrikos,
in

Gymn.

Progr.

Ratibor, 1908.

Der Abschnitt vom Rhythmus

XcLQizeg
Berlin 1911.
.

Ciceros Orator, in

[On

De

Groot],

(Bursian's)

Jahresbericht,

Bd.

170,

1915, p. 148 sqq.

Nassal, P., Aesthetisch-rhetorische Beziehungen zwischen Dionysius von Halicarnass und Cicero. Diss. Tubingen 1910.

Compare

Zeitschr.

f.

osterr.

Gymn.

64, 4.

Minucii Felicis aetate et genere dicendi. Vorlesungsverz. Univ. Greifswald 1897, p. 18 sqq. Gotting. Gel. Anzeigen 1904, p. 310. Die antike Kunstprosa 2 Leipz. etc. 1909. Esp. Nachtrage. Agnostos Theos. Leipzig etc. 1913. Novak, It. In panegyricos latinos studia grammatica et critica.
.

Norden, E.

De

Prag 1901.
filologicke 1914, p.
.

Z nove litartury o rhythmu anticke pr6sy. Listy 88-102, 192-212, 414-422. Eine neue Methode der Klauselforschung. Berl. Philol. Wochenschr. 1917, col. 217 sqq. Paleographie Musicale. Par les Benedictins de Solesmes. Solesmes 1894. Partzinger, P. Beitrage zur Kenntnis der Entwicklung des ciceronischen stils. Landshut 1910, p. 128-129. Peters, K. Festschrift fur Raspe. Parchim 1883, p. 8-19. Petersen, E. Rhythmus. Gott. Gel. Abhandl. 1917. Peterson, W. The Vatican codex of Ciceros Verrines. Amer. Journal of Philol. I9aa, p. 409-36. Pichon, R. Lactance. These Paris 1901, p. 324-34, 439-40. Etudes sur l'histoire de la litterature latine dans les Gaules. Les derniers ecrivains profanes. Paris 1906, p. 297 sqq. Le texte de Quinte Curce et la prose metrique. Revue de Philol. 30, 1906, p. 90-100. Pietsch, G. De Choricio Patrocli declamationis auctore. Breslauer Philol. Abh., ed. by Forster, Heft 42, 1910. Compare M a a s, Wochenschr. f. klass. Philol. 28, 1911, col. 1253. Pohlenz, M. Eine byzantinische Recenzion Plutarchischer SchrifGotting. ten. Gel. Nachrichten 1913, p. 338 sqq. Poppel, G. van. Over rhythmisch proza in literatuur en liturgie. De Katholiek 140, 1911, p. 384-99; 438-52.
r

Novotny, P.

213
Pothier, J. Les melodies gregoriennes. Tournay 1880, p. 237. Previtera, L. De nutnero sive clausula sive structura sive cursu.
Syracusis 1899.

Compare Wochenschr. II metodo

,

f.

klass. Philol.

1903.

statistico

nelle

nuove ricerche

della prosa

metrica.

Giarre 1903.

Compare Laurand, E'tudes, p. 189, n. 5, Kroll, Bed. Philol. Woch. 1904, col. 1550, De Jonge, Bull. bibl. du Mus. beige 8, 14. Przychocki. De Gregorii Naz. epistulis quaestiones selectae. Compare Maas, Byzant. Zeitschr. 22, p. 246, Jager, Deutsche
Lit. Zeit.

Quentin, H.
Melanges

1913, col. 1180. La plus ancienne vie de saint Sevrin de Bordeaux. Couture. Toulouse 1902, p. 1214.

Radermacher, L.
65, 1906, p. 149.

Griechischer

Sprachgebrauch.

Philologus
attributo
u.
z.

Raspante,
a Lattanzio.
fruhchristl.

S.

II

nurnerus nel

De mortibus persecutorum
Studien
z.

Adern6 1913. Reitzenstein, R. Poimandres.

Literatur. Leipzig 1904.

gr.

agypt.

Richter, E.

Zur Entwicklung der romanischen Wortstellung aus

Thukydides und die Sophistik.
Diss. Leipzig

der lateinischen. Halle 1903.

Rittelmeyer, F.
1915, p. 81 sqq.

Rollmann, B.
Sabbadini, R.
1897, p. 600-601.
,

De numeri

oratorii primordiis. Diss. Miinster 1910.

Del 'nurnerus' in Floro.

Rivista di Filologia 25,

inscribitur obviis.
,

in dialogo qui Vergilius orator an Rivista di filologia 1897, p. 600 sqq. [on inscriptions], Riv. di filol. 26, 1898, p. 126.

De numeris

poeta

'cursus' Civilta cattolica, serie 18, vol. 11,

II

Saintsbury, G-. Santi, A. de.

History of English Prose Rhythm.

nella storia letteraria e nella liturgia. 1903, p. 24-39; vol. 12, 1903, p. 38-

52, p. 288-307.

Schanz, M.
Diss.

Rom. Literaturgeschichte. I, 2, p. 206-7, 1909. Schlicher, J. J. The origin of rhythmical verse in late Latin. Chicago 1900. Appendix I, p. 83-87. Schmid, W., see Christ, W. von. Schmidt, A. Zur Lehre vom oratorischen Nurnerus. Progr.
J. Das rhythmische 1893, p. 209-247.
,

Mannheim 1858. Schmidt,

Studien
15,

Element

in

Ciceros Reden. Wiener

Zeitschr, fur die osterr.

Satzrhythmus und Neposfrage.

Gymn.

46.

1895, p. 97-109.

Schmiedeberg, P.
Sangallensibus.

De Asconi

codicibus et de Ciceronis scholiis

Diss. Bresl. 1905.

214
Schober, E.
Diss. Halle 1904.

De Apulei Metamorphoseon compositione numerosa.

lege agr.

J. K. Clauseltheorie und Textkritik. Zu Cic. De Wochenschr. f. klass. Philol. 1914, col. 1430 sq. Serruys, D. Les procedes toniques d'Himerius et les origines du Melanges Havet 1909. 'cursus' Byzantin. Sheily, J. [On Clark etc.], Rhytmical prose in Latin and English. The Church Quarterly Review 74, 1912, p. 81 sqq. Shipley, P. W. The treatment of dactylic words in the rhythmic
II,

Schonberger,
59.

prose

with special reference to the sense pauses. Transact, 1911, p. 139 sqq. The heroic clausula in Cicero and Quintilian, Class. Philology 6, 1911, 410 sqq. Compare Di Capua, Boll, di Filol. class. 18, 1911-12, p. 244 sqq. Preferred and avoided combinations of the enclitic 'que' Class. Philology 8, 1913, 23 sqq. in Cicero. Compare S c h m a 1 z, Wochensch. fur klass. Philologie 1914, col. 239. Sievers, E. Ein neues Hilfsmittel der philologischen Kritik. Verhandlungen der Vers. d. Philol. u. Schulm. 47, 1903, p. 33. Sinko, Th. Die Descriptio orbis terrae. Archif f. latein. Lexikogr. 13, 1904, p. 535 sqq. Skutsch, F. Zu Favonius Eulogius und Chalcidius. Philologus 61, 1902, p. 193 sqq. Rhein. Firmicus de errore profanarum religionum. Museum N. F. 60, 1905, p. 262-72. Die lateinische Sprache, in Die Kultur der Gegenwart,
of

Cicero,
,

of the

Amer.

philol. assoc. 41,

I,

8,

1905, p. 421-424, p. 431.

,

Sprache, Metrik Wechselbeziehungen. Zeitschr. f.

Literaturbericht, 1913, p. 422 sqq., 427.
.

d.

und Rhythmus der Romer in ihren Gymnasialwesen 1909. p. 67 sqq.

1,

Glotta

1909,

p.

399.

413 sq.;

4,

[On

Gaffiot,

Smedt, C. de.

Analecta bollandiana 16, Stange, C. De Arnobii oratione. Saargemiind 1893 II: De clausula Arnobiana, p. 1336. Stangl, Th. Bobiensia. Neue Beitrage zu den Bobienser Ciceroscholien. Rhein. Museum 65, 1910, p. 88, 249, 420 sqq. Zu Ciceros rhetorischen schriften. Wochenschr. f.
.

Pour le vrai latin], Glotta 3, 1912. p. 366 sqq. Le cursus dans les documents hagiographiques, 1897, p. 501-06; 17, 1898, p. 387-92.

klass. philol.

1914, col. 21 sqq.

Steeger, T. Die Klauseltechniek Leos des Grossen in seinen Sermonen. Haszfurt 1908. Stiefenhofer, A. Die Echtheitsfrage der biographischen Synkriseis Plutarchs. Tubingen 1915, p. 27, A. 60 (= Philologus 73, 1914-16, p. 462).

215
Strauss,
Progr.
Diss.
class.

C.

Die

Klauselrhythmen

der Bobienser Ciceroscholien.

Gymn. Landau

1910.

Sundermeyer, A.

De

re metrica et rhythmica Martiani Capellae.

Bollett. di filol.

Marpurgi Cattorum 1910.

Terzaghi, N. Ciceronis De re publica IV, 11. 191112, p. 55-58. Compare D i C a p u a, Le clausule in S. Agostino. Le clausole ritmiche negli Opuscoli di
.

Sinesio. Didas-

kaleion

1912, p. 205-225, 319-360. Compare Maas, Byz. Zeitschr.

1,
f.

22,

1913,

p.

207,

Draseke,
Woch.

Wochenschr.

klass. Philol. 30,

15, p.

411-12;

Berl. Philol.

1906, col. 777; Byz. Zeitschr. 18, p. 342. Thalheim, Th. Zu Lykurgos und Lysias. Progr. Hirschberg i. I. Der Rhythm us bei Lykurgos. Schl. 1900. Compare Drerup, Berl. Philol. Woch. 20, 1900, col. 1313-18, Thalheim, Berl. Philol. Woch. 1900, col. 1534. Thiele, G-. Ionisch-attische Studien. I Gorgias. Hermes 36, 1901, p. 218. Thomas, E. Petrone 3 Paris 1912, p. 193-200. A. Arch. f. Papyrusk, 3, 1906, p. 458.
,

Thumb,

Germanisch-roman. Monatschrift 1911, p. 4. In K. Brugmann's Griechische Grammatik, 4th ed. by A. Thumb, Munchen 1913, p. 666-672. Satzrhythmus und Satzmelodie in der altgriechischen
.

Sprache.

Fortschritte der Psychologie

1,

1913.
f.

Hubermann, Wochenschr. Munscher Berl. Philol. Woch. 1915, nr.

Compare

klass.

Philol.

1913;

15.

servir

Notices et extraits de divers manuscrits latins pour a l'histoire des doctrines gramm. du moyen age. Notices et extr. des mss. de la Bibl. imper. 22, 2e partie 1868, p. 48185. Tolkiehn, J. [On De Jonge], Berl. Philol. Wochenschr. 26, 1906,
col. 907.

Thurot, C.

Tyrrel, R. Y. On the mathena 12, 1903, p. 280 sqq.

.

third foot of the

in

Greek hexameter.

Her-

Metrical

prose

the correspondence of Cicero.

(On
Frei-

Bornecque

Hermathena 13, 1905, p. 289 sqq., esp. 302. Unser, H. uber den Rhythmus der deutschen Prosa. Diss.
etc.].

burg 1906.

Vacandard,
17377.
la liturgie.
.

E.

Saint

Victrice.

Paris

1903,

appendice

B,

p.

Le cursus, son origine, son histoire, son emploie dans Revue des quest hist. 78, 1904, p. 59-102, 699. Valentin, L. Saint Prosper d'Aquitaine. Toulouse, p. 86588. Valois, N. De arte scribendi epistulas apud Gallicos medii aevi
Paris 1880, p. 7080.

scriptores rhetoresve.

216
Bibl. de fitude sur le rhythme des bulles pontificales. l'Ecole des Chartes 42, 1881, p. 161-98, 257-72. Vendryes, J. Recherches sur l'histoire et les effets de l'intensite These Paris 1902, p. 71-72. initiale en latin. Verral, A. Longinus on the rhythmus of Demosthenes. Review 19 (1905), p. 254. Class. Compare Lehnert, (Bursian's) Jahresber. 1909, 380, p. 321.
.

W.

Volkmann, R. Die Rhetorik der Griechen und Romer. Vossius, G. J. Commentariorum rhetoricorum libri VI 5 Marpurgi 1781. "Waltzing, J. P. Octavius de Minucius Felix. Lille-Paris 1909. Partie du maitre p. 137139; partie de l'eleve p. 192-93. Watson, E. W. The style and language of St. Cyprian, chapter I 13, Rhythm. Studia bibliaca et ecclesiastica, 4, Oxford 1896, p. 217-21. Wehofer. Untersuchungen zur altchristlichen epistolographie.
,

Sitz. Ber.

Wien Weil, H.

143, 1901.

Les harangues

de

Demosthene.

Paris

1912,

p.

LI,

esp. note 2.

Weinstock.
Compare Bed.

De

Wendland, P.

Erotico Lysiaco. Diss. Munster 1912, p. 86. Quaestiones rhetoricae. Progr. acad. Gottingen 1914.

1915, col. 28. in the pronunciation of Latin. Compare Berl. Philol. Woch. 1914. Wichmann, C. De numeris quos adhibuit Demosthenes in oraDiss. Kiel 1892. tione Philippica prima. Wilamowitz Moellendorf, U. von. Lesefriichte Hermes 34, 1899, p. 214-18, laa. Hermes 35, (1900) p. 37, p. 536-42. In Die Kultur der Gegenwarts i, g, 1907, p. 65 sqq.,
Philol.

Woch.

Westaway, K. M.

Quantity and accent

p.

104, p.

153.
in

Compare Heibges, p. 109 sqq. Wilson, H. A. The metrical endings

Journ. of the theological studies
5,

1904, p.

386-95;

Winterfeld, P.
Bennonis.
Justini
Sitz. Ber.
.

Der Rhythmus der

the leonine sacramentary. 6, 1905, p. 381-91. Satzschliisse in der Vita

Berlin 1901, p. 163-68. Uber die Translatio sanctorum

presbyteri.

Neues Archiv. der Ges.

in

f.

alt.

Alexandri papae et deutsche Geschichtsk.

27, 1901, p.

751-54. Die 4 Pabstbriefe

.

gard.

Neues Archiv.
.

d.

G.

f.

alt

der Briefsammlung des H. Hildedeutsche Geschichtsk. 27, 1902, p.

237-44.

Zu Avianus.

Rhein.
bei

Museum

57, 1902, p. 167, esp. n.

1.

61,

1902, p.

Der Satzschluss 623-26.

.

Favonius Eulogius.

Philologus
Rhein. Mus.

Satzschlussstudien zur Historia Augusta.

57,

1902, p. 549-558.

217
Philologus 63, 1904, p. 31520. des Eugippus und der rhythmische Satzschluss. Rhein. Museum 1903, p. 363 sqq. Wissowa, G. Genethliacon Carl Robert uberr. von der Graeca Halensis 1910, p. 59 sqq. Wolff, J. De clausulis Ciceronianis. Jahrbucher f. Philol. Suppl. 26, 1901, p. 581 sqq. Wuest, G. De clausula rhetorica quae praecepit Cicero quatenus Diss. Argentor 5, 1881, p. 227328. in orationibus secutus sit. Zander, C. Eurhythmia vel compositio rhythmica prosae antiquae. Eurhythmia Demosthenis. II. Numeri latini aetas integra. III. EuI. rhythmia Ciceronis. Leipzig. Ziegler, K. Neue Firmicus-Lesungen. Rhein. Museum, N. F.,
Felix,

Zu Minucius
Die

Handschrift

60,

1905, p. 273-96.
,

vid. Firmicus Maternus. Zielinski, Th. Das Clauselgesetz in Ciceros Reden. Philologus, S. B. 9, 1904, p. 589 sqq. Das Ausleben des Clauselgesetzes in der rom. Kunstprosa (I. Die Panegyriker; II. Cyprian). Leipz. 1906 (= Philol. S. B 10, 1906, p. 429 sqq). Textkritik u. Rhythmusgesetze in Ciceros Reden. Philol 65, 1906, p. 604 sqq. Der Rhythmus der rom. Kunstprosa und seine psycho logischen Grundlagen. Archiv. f. d. gesammte Psychologie 7, 1906, p 128-42. [On May], Bed. Philol. Wochenschr. 26, 1906, col 777 sqq., 286 sqq.
.

CURVES.

I.

THUCYDIDES.

a n
It
/

clausula-

metre.

;
-V"

*/

**\

v^^A*I

sentence-

metre.
*~n*+&*toLkjdx4 .

*l
S

>

>

>
!

fe

SJJi!
j
J

9
|

18*
i

> 1

#
1

>

*
1

*
i

>

These curves show the

sentence-metre

striking

agreement between Thucydides'

and

his

clausula.

The few

deviations

(e.

g.

for

been discussed above

due
to chance.

(p.

18485,

p.

2028); they are probably

220
II.

DEMOSTHENES.

clausula-melre.

sentence-metre.

These

curves show that Demosthenes has only one really

viz.

preferred clausula,
p.

p.

:>=d

See Norden, Antike Kunstprosa 2

911 and 923; and above

3334.

221
III.

PLATO.

a
ft

LAWS.

frequency of the forms ending in a long syllable. frequency of these very forms ending in a short
. .
.

syllable.

?JLcJ:*

From
the
last
p.

these

curves

it

appears

clearly,
is

that

the

quantity
all;
is

of

syllable in Plato's later

191.

works
owing

not indifferent at

see

above

That, though in general a long final syllable

more

frequent than a short

one

to the greater

frequency of

long syllables in the

Greek language

this
->-

rule does not hold

good

for
y

the

clausula

wv_/w

where

*w

is

preferred to

- w
over a

apparently owing to a general preference for a tribrach

dactyl.

On

the other hand

it

appears that

ww^
,

is

pre-

ferred to

wwww,
only

which
of

fact agrees

with the passage in Aristotle

who

speaks

the
is

fourth

paean (^-^-^w

Likewise

not
;

.-~
same

).

preferred to

the

is

222

w and < and inferred that above pp. 191 192, and 6164.
true
for
-

w, from which

it

may be
See

belong together.

IV.

PHILO.

clausula-

metre.
sentence-

metre.

3
I

*
I

J
>

3
i

J
t
I

>
i

?
I

>

*
j

U
?

These

curves

clearly

w~_.^ ^ w ^
See above
p.

(e.g.

),
and

-
show
that

the

clausulae
,

- ^n,

-,

>=:

have been preferred by him.

196,

and pp. 5458.

223
V.

PLUTARCH.

clausula-

metre.
sentence-

metre.

These curves show

his preference for

and no other forms.

See above pp. 33

39,

and

p.

185.

See

also

my

remarks

in Berliner Philologische

Wochenschrift

1917, col. 1158.

224
VI.

LIVY.

clausula-

sentence-

metre. metre.

These curves have been inserted

idea of the
reliability
first

in order to give a superficial

details.

of
is

our results in

The

material on
the

which the

curve

based comprises

1.000

cases from

sentence-metre of book IV, compared with 1.000 clausulae from the

from book XLII.

The other is based upon the same number of cases The curves are nearly the same. The agreement would have been much more striking, if not the frequency of sixsame book.

225

72
pZ,

mi n%

m
ft

H
.
.

3l
Bl
fl

clausula-

sentence-

metre.
metre.

a n
41 it
:
1*

u\
'/

y/
/'

"

'1
"1 "I

/
/ / *

Aw

^
*

ft

>

>
>

\ \

1 1

T
*

in Jurn
S

a
*

a ?
;

a
1

*
J

*
1

i
1

>

a
'

;
s

i
'.

i I

3
1

1 1 t

)
3

i >

syllabic

forms,

but that

of

five-

or /<?wr-syllabic forms had been

represented.

The
pp.

question will be discussed in the second volume.

121,
126,

See above
Latino,

112,

and

my

paper:

De numero

oratorio
-

which
in.

will shortly appear.

Livy prefers

:=:,

-"

:=:,

and

226
VII.

CICERO.

clausula-

metre.
sentence-

metre.

mm
i

WW W M

ni

T
i

'

These curves show the preference

See

of Cicero for

w.
volume

my

paper:

De numero

oratorio Latino, and the second

of this work.

INDEX
Ammianus
Apulejus
Aristotle

Marcellinus

108, 129,

133 146

61, 114
70, 81

Arnim (Von)
Baehrens Barwick
Blass

Heitmann Henderson Herodes Herodotus

Himerius

86 46 87
141

135 133

146

Homer
Isaeus
Isocrates
115,

74 et passim
8 et passim
65, 91, 99

142 188

Bornecque

Bornecque Zielinski
Charito
83,

100

130

Johannsen Jordan
Kaluscha
63, 68,

160

99
123,

Cicero 67 et passim Consolatio ad Apollonium 12, 90

171,
17; I,

149

175,

176 143

Krumbacher
Kuiper (W.
Lactantius
E.
J.)

98
71

Cronert

141,

Demosthenes

8 sqq., 29 sqq.
185 sqq.
I

121
104,

st

passim;
135

Laurand

110
105

see contents.

Dewing
Flavius Josephus

Leo Lesbonax
Libanius

101,

87

86
97

130
142

Lipsky
Litzica

Fuhr

98
112,
121,

Gompertz Gropp

71
31

Livy Lutoslawski

126 sqq.

76
87, 99
6,

Harmon
Haury Havet
Heibges

133
141

122
83, 125

Maas Marbe Marx Meyer (W.) Munscher

14 et

passim
130
99,

65 25 sqq., 34, 62

134
152

88,

228
Natorp
7

Norden Novotny
Ovink
Philemon Philo Judaeus
Philostratus

9 et passim, 33, 99
91 sqq., 99

Rodiger Rollmann
Sallust
112,

142

26
126 sqq.
71

71

137

Schleiermacher Shipley Sophronius

Stiefenhofer

115
111,

135
5

54 sqq., 112
130
11

Plato

et passim, 172,

176

Thrasymachus

107 sqq., 182

6,

190 sqq.; see contents

Thumb
Thucydides
Virgil

14 et passim, 84, 99

Plutarch

passim; see contents;

see Stiefenhofer

20 et passim
127

see Consolatio ad Apollonium;

Pohlenz

70,

139 130

Polemo
Procopius
Quintilianus
90,

Wilamowitz

130,

134

111,

120, 135,

143

Xenophon
101

28 94 sqq., 99
67
99,

sqq.

Zander

Raeder
arrangement of words
choice of words
ictus
133, ssq.
11, 11,

70

Zielinski

124

138

141

degree of preference and avoidance expressed by a quotient

negative clausula
47

43

order of words; see arrangement of the Greek rhythmical clausula 132 sqq.
origin of the clausula 107 sqq.

resolutions

103

CORRIGENDA.

Page

7, line 4, 336

instead of

308.

62,

23, 463

000.

URBANA UNIVERSITY OF ILLINOIS

KSoBOOK OF A N TIQUEP ROE-RHYTH

GRONI

lilt
3 0112

023611285