You are on page 1of 7

Greek Astronomical Calendars: III.

The Calendar of Dionysios


Author(s): B. L. van der Waerden
Reviewed work(s):
Source: Archive for History of Exact Sciences, Vol. 29, No. 2 (1984), pp. 125-130
Published by: Springer
Stable URL: http://www.jstor.org/stable/41133707 .
Accessed: 08/06/2012 06:48
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Springer is collaborating with JSTOR to digitize, preserve and extend access to Archive for History of Exact
Sciences.

http://www.jstor.org

GreekAstronomical
Calendars
III. The CalendarofDionysios
B. L. VAN DER WaERDEN
1. Introduction
In Ptolemy's Almagestwe findeight observationsof the planets Mercury,
Mars and Jupiterdated in thecalendarof Dionysios. As an exampleI shall quote
the firstobservationof Mercury(Almagest IX 9, p. 264 Heiberg):
ofthe21stdayofHydron
In the23rdyearoftheera ofDionysios,in themorning
to the zodiacal signAquarius)Mercury
(i.e. in a partof the yearcorresponding
stood 3 lunardiametersnorthof the brilliantstar(<5)in thetail of Capricorn. . .
The datewas in theyear486 sinceNabonassar on EgyptianChoiak 17/18earlyin
themorning
(-261 February12).
havethedaynumber29, butthecodicesG and D have
Note.Somemanuscripts
Manitius
21.For goodreasonsLepsiusandBoeckhas wellas theGermantranslator
froman Arabicmanuthereading21. Gerhard of Cremona,who translated
prefer
script,also has 21.
In all cases Ptolemypresentsdouble dates: firsta date in theera of Dionysios,
and nexta date in the Egyptiancalendar.The latterdates can easily be reduced
to Juliandates.
In Chapter13 ofhis book "Sonnenkreise"1August Boeckh has reconstructed
the calendarof Dionysios, basinghimselfon the eightdouble dates of Ptolemy.
Accordingto Boeckh, the firstyear of the era of Dionysios began in the Julian
year -284 (= 285 B.C.) on June26, about the timeof the summersolstice.The
years with numbers4/7,An+ 1 and 4/2+2 had 365 days each, and the years
4/2+ 3 had 366 days. The yearwas dividedinto 12 partsnamedafterthezodiacal
signs,beginningwithCancer. The firstelevensignshad 30 days each, and the last
sign Gemini had 35 or 36 days.
1 A. Boeckh: Ueberdie vierjhrigen
Sonnenkreise
derAlten(Berlin1863),pp. 286340.

126

B. L. van der Waerden

as follows:
Otto Neugebauer2 criticizesBoeckh's reconstruction
to reconstruct
thecompleteDioMuch speculationhas beenspentin attempts
as longas we haveno hopeofchecking
nysiancalendar,a rathervaluelessenterprise
on additionalmaterial.
reconstructions
thesehighlyhypothetical
In myopinion,thisharshjudgmentis not justified.In whatfollowsit will be
is not based on "highlyhypotheticalreconshown that Boeckh's reconstruction
If
facts.
we
but
on
structions",
accept Ptolemy's double dates, we have no other
choice than to agree with Boeckh's conclusions.
2. Ptolemy'sDouble Dates
Following Boeckh, I shall divide the double dates recordedby Ptolemy into
3 groups:
A. Year numbersAn+ 1: years 13, 21 and 45.
C. Year numbersAn+ 3: year 23.
D. Year numbersAn: years 24 and 28.
In Boeckh's group B (years An+ 2) therehappen to be no dates.
The double dates in group A are:
Year 13: Capricorn25 = Athyr20/21 morning(Heiberg p. 352)
= -271 January18
Year 21: Scorpio 22 = Thoth 18/19morning(Heiberg p. 288)
= -264 November15
Year 21: Scorpio 26 = Thoth 22/23 morning(Heiberg p. 289)
= -264 November19
Year 45: Virgo 10 = Epiphi 17/18morning(Heiberg p. 386)
= -240 September4
In group C we have two entries
Year 23: Aquarius 21 = Choiak 17/18morning(Heiberg p. 264)
= -261 February12
Year 23 : Taurus 4 = Mechir 30/Phamenoth1 evening(Heiberg p. 265)
= -261 April25
In group D we have
Year 24: Leo 28 = Payni 30 evening(Heiberg p. 267)
= -261 August23
2 O. Neugebauer: A Historyof AncientMathematicalAstronomy(SpringerVerlag1975),Part Three,p. 1067.

GreekAstronomical
Calendars.Ill

127

Year 28: Gemini 7 = Pharmuthi5/6evening(Heiberg p. 266)


= -256 May 28
From these double dates we see that the years of Dionysios were divided
into 12 partsnamed afterthe zodiacal signs. In this respectDionysios followed
the traditionof Euctemon and Callippos, who used tropical years beginning
withthe summersolsticeon the day Cancer 1.
The firstyear of the "Era of Dionysios" began in the Julianyear -284. This
betweentheyearnumbersof Dionycan be verifiedby calculatingthe differences
sios and the Julianyear numbers.In the firsthalf of the Julianyears,beforethe
are
summersolstice,the differences
13 - (-271) = 284 in January
23 - (-261) = 284 in Februaryand April
28 - (-256) = 284 in May
but afterthe solsticetheyare
24 - (-261) = 285 in August
45 - (-240) = 285 in September
21 - (-264) = 285 in November.
Since the yearsof Dionysios began and ended withthe summersolstice,they
cannothave been Egyptianyearsof 365 days. It is easy to check thisconclusion
by comparingdates in years far apart. For instancewe may startwiththe last
double date of group A :
(1) Year 45: Virgo 10 morning= September4.
Adding 76 days and supposingthat the signsVirgo and Libra have 30 days
each, as in the parapegmataof Euctemon and Callippos, one obtains:
(2) Year 45: Scorpio 26 morning= November 19
witha possibleerrorof one of two days,because the signsVirgoand Libra might
have 29 or 31 days.
years earlierwe have the double date
Twenty-four
(3) Year 21 : Scorpio 26 morning= November 19.
The timebetweenthedates (2) and (3) is 24 Julianyears,or 24 Egyptianyearsand
6 days, but the Dionysian day numbersin (2) and (3) are exactlythe same. It
followsthat Dionysios did use years of 365-j days, and it also followsthatthe
signe Virgo and Libra had exactly30 days each.
betweenthe two nightsin the Egyptiancalendaris
In group C the difference
73 days, and the distancebetweenthe two Dionysian dates is 3 zodiacal signs
thatthesignsAquarius,Pisces and Aries
minus17 days. If we assume,tentatively,
have 30 days each, and if we assume (withBoeckh) that the Dionysian day did
not begin at midnight,but at sunriseor noon or sunset,the agreementbetween
73 days and 3 signs minus 17 days is perfect.

128

B. L. van der Waerden

Still more signscan be shown to have 30 days only. In group D we have the
equation
(4) Year 24: Leo 28 evening= August 23.
Adding 279 days and supposing that the signs Leo throughTaurus have
30 days each, one obtains
(5) Year 24: Gemini 7 evening= May 28.
Four years later one has fromthe text of Ptolemy
(6) Year 28 : Gemini 7 evening= May 28.
These two equations are in perfectaccordance,since 4 Julianyearsare equal
to 4 Dionysian years,as we have seen. It followsthatthe ten signsfromLeo to
Taurus have togetherexactly300 days.
This is a veryremarkableresult.In theparapegmaof Euctemon these10 signs
have together304 days, and in the parapegmaof Callippos, whichaccords well
withmoderntheory,theyhave together302 days. If 10 signshave together300
days, the remainingtwo signs Gemini and Cancer have together65 or 66 days:
an impossibilityin any reasonable theoryof the annual motion of the sun. As
Neugebauer (note 2, p. 629) rightlynotes, the Dionysian zodiacal dates, considered as solar longitudes,"deviatein an irregularfashionboth fromtrueand from
mean solar longitudes".
Thus we are bound to conclude(withBoeckh) thatDionysios used a division
of the year into 11 parts(Cancer throughTaurus) of 30 days each, and one part
(Gemini) of 35 of 36 days. As in the Egyptianand Alexandriancalendars the 5
or 6 "epagomenal days" were placed at the end of the year,whichis veryconvenientfor calendaric calculations.
If this is assumed, we can go back fromeach of the eightdouble dates and
calculatethe beginningof the Dionysian yearin each case. This has been done by
Boeckh, with the followingresult("Sonnenkreise",p. 317):
The yearsA, B, C, whoseDionysianyearnumbersare notdivisibleby 4, begin
on June26, and the years D begin on June27.
It followsthattheyearsA, B, D have 365 days each, and theyearsC 366 days.
If thisis assumed,all double dates are correctlyreducedto theEgyptiancalendar in Ptolemy's text,withthe exceptionof the first,whichshould read
Year 13: Capricorn 26 = Athyr20/21 morning= -271 January18.
The calendar of Dionysios is not an astronomicalcalendarin the strictsense,
because it is not based on a theoryabout the sun's course in the zodiac. It is a
convenientcompromisebetweenthe Egyptiancalendar with its 12 monthsof
30 days and 5 epagomenal days, and the more sophisticatedzodiacal calendars
of Euctemon and Callippos.

GreekAstronomical
Calendars.Ill

129

3. Greek Observationsin the Third CenturyB. C.


Many observationsof the sun, the moon, the planetsand the fixedstarswere
made in the firsthalf of the thirdcenturyB.C. in Greece and HellenisticEgypt:
A. Aristarchos of Samos observedthe summersolsticein -279 (Almagest
III, 1, p. 207 Heiberg).
B. TimocharisobservedVenus,the moon and the fixedstarsduringtheyears
-294 to -271 (AlmagestVII 3 and X 4).
C. The anonymousastronomerwho used the calendar of Dionysios observed
Mercury,Mars and Jupiterduringthe years -271 to -240.
D. Aristyllos observedthe declinationsof fixedstars (AlmagestVII 3).
Dennis Rawlins3 has analyzedtheobservationsof Aristyllos bytheMethod
of Least Squares. He found that the observationswere extremelyaccurate,
havinga standarderrorof about 4', and that theywere made, most probably,
later than -275. WhereasTimocharis had observedthe declinationsof 12 stars,
his successorAristyllos observed6 otherstars. Thus his programwas a continuationof that of Timocharis.
The latteralso observedoccultationsof fixedstars by the moon (Almagest
VII 3). I suppose thatthepurposeof thoseobservationswas, to studythe motion
of themoon withrespectto the fixedstars.For thispurpose,one would need not
onlythe declinations,but also the longitudesof the fixedstarsin question.So we
may suppose that Timocharis and Aristyllos observedlongitudesand declinationsoffixed stars.
These fixedstardata mightserveas a reference
systemforstudyingthemotion
of the moon and the planets. In fact, Timocharis himselfobserved distances
between Venus and fixed stars, and our anonymous observer(see under C)
observeddistancesof Mercury,Mars and Jupiterfromfixedstars.
of the calendarof
Thus one getsthe strongimpressionthatthe establishment
Dionysios and the observationsof our fourobserversall formedpart of a program,directedat the studyof the motionsof the moon and the planets.
In thehistoryof astronomy,we mayobservethatthe purposeof observations
of themoon and theplanetsalwaysis to determinethevalues of constantsoccurring in astronomicaltheories.This is true for Babylonian,Greek, Arabic and
westernastronomy.Astrologersneed panetarytables, and a necessaryprerequiof the constantsin a theory.
site forcomputingsuch tables is the determination
Thus it is not too bold to conjecturethat the observationsmade in Samos and
Egyptin the thirdcenturyB.C. had just this purpose.
methods
At that time,two typesof theoriesexisted.Linear, or arithmetical
wereused in the BabyloniantheoriesA and B and in Egyptianplanetarytables
writtenin the Roman age4. On the otherhand, geometricalmethods,based on
the assumptionof uniformcircularmotions,weredevelopedin Greece and Alex3 D. Rawlins: Aristyllos'
Date withVindication,
to be publishedin thisArchive.
4 See van der Waerden: Aegyptische
Centaurus16, pp. 65-91
Planetenrechnung,
(1972). See also G.Abraham: The Motion of Mars in EgyptianPlanetaryTables,
to be publishedin thisArchive.

130

B. L. van der Waerden

andria from the classical period to the time of Ptolemy. We now may ask:
What typeof theorydid Aristarchos, Timocharis,Aristyllos and our Anonymous aim at?
For Aristarchos the case is clear. He was the author of the heliocentric
theory,in whichthe planets(includingthe earth)weresupposedto rotatearound
the sun. If he wanted to determinethe constantsof this theory,he firsthad to
studythe apparentmotionof the sun. He could take the durationsof the astronomical seasons fromthe parapegma of Callippos, but he had to observe at
least one solstitiumor equinoctiumfor his own time,and this he did.
As forDionysios,his calendarfollowedthetraditionof theGreekparapegmatistsEuctemon and Callippos. His divisionof thezodiac into 12 signswas a tropical division: the sun was supposed to enterCancer at the timeof the summer
solstice,whereasin the Babylonian systemA the same solstice was located at
Cancer 10, and in systemB at 8. So the calendarof Dionysios belongedto the
Greek and not to the Babyloniantradition.
The unknownauthorof the observationsof Mercury,Mars and Jupiterlived
at the time of Aristarchos and used the calendar of Dionysios. He used, like
Dionysios, the Greek divisionof thezodiac, in whichthesignCancer beginswith
thesummersolstice.LaterEgyptianplanetarytextsall used theBabyloniansiderial
divisionof thezodiac, and a siderealyearlongerthan365-Jdays.For thesereasons,
it is not likelythat the planetaryobservationswere made in orderto determine
the constantsof a Babyloniantype theory.
What geometricalplanetarytheorieswere available between-271 and -240,
when these observationswere made?
The "primitiveepicycletheory"discussedin my paper5,whichwas probably
inventedby the Pythagoreans,cannot be used to explain the motion of Mars,
as Aaboe6 has shown. The same holds forthe homocentricspheresof Eudoxos
and Callippos, as Schiaparelli7 proved. Thus, if our observersin Alexandria
had a definitegeometricaltheoryin mind, the most probable candidate is the
heliocentrictheoryof Aristarchos of Samos.
Hence my hypothesis: The purpose of the observationof Aristarchos,
Timocharis, Aristyllos and our Anonymouswas to determinethe constantsin
the heliocentrictheoryof Aristarchos.
Wiesliacher 5
CH-8053 Zrich
(Received February2, 1983)

5 Van der Waerden: The Motion of


Venus, Mercuryand the Sun in Early Greek
Astronomy,Archive for History of Exact Sciences 26, pp. 99-113 (1982).
6 A. Aaboe: On a Greek
Qualitative Model of the Epicycle Variety,Centaurus 9,
pp. 1-10 (1963).
7 G. V. Schiaparelli: Scritti sulla stona della astronomia antica
II, pp. 3-112.
German translation:Abhandlungenzur Geschichteder Math. I (1877), Supplementto
Zeitschriftfr Math, und Phys. 22, pp. 101-198.

You might also like