Anthyphairetic Ratio and Eudoxan Proportion

Author(s): D. H. Fowler
Source: Archive for History of Exact Sciences, Vol. 24, No. 2 (1981), pp. 69-72
Published by: Springer
Stable URL: http://www.jstor.org/stable/41133607 .
Accessed: 01/08/2013 23:41

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

This content downloaded from 200.26.133.57 on Thu, 1 Aug 2013 23:41:01 PM

All use subject to JSTOR Terms and Conditions
 RatioandEudoxanProportion
Anthyphairetic
D. H. Fowler

by W. Hartner and B. L. van der Waerden

Communicated

In a recentarticlein this Archive,R. C. Riddell [5] gave a most exciting

speculativeexplorationof the astronomicaland mathematicaldevelopments
associated withthe name of Eudoxus, and set out for considerationnew ways
in whichtheymighthave been interrelated. He also discussedthe way in which
some of theseideas mighthave grownout of earliermathematical procedures,but
herecame to a negativeconclusionwithregardto proportiontheory.In particular,
when discussingthe idea of anthyphairesis (called antanairesisby Aristotle),
he concluded(p. 9) : 'To breakout of this[anthyphairetic] circle of ideas it seems
thatone would need the stimulusof some freshconstruction," and thenproposed
that his analysis of rotatingnestedspheresleads to thisrequirednew insight.I
wishto pointout herethattheEudoxan definition of proportioncan growdirectly
and naturallyout of theolderanthyphairetic procedures,withinthekindofcontext
that Riddell described,and thus,I hope, add furtherscope to his stimulating
proposals.
The problem,in generalterms,concernsthe meaningof a ratio a: è, wherea
and b are eithernumbers,homogeneousmagnitudes, or intervalsof time.Riddell
referred briefly to the proposal that,before the developmentof Eudoxan propor-
tion theory,the ratio a:b mighthave been describedby the anthyphairesis, the
application of the 'Euclidean' subtractionalgorithm, to a and b. for
Thus, example,
ifa = 16,b = 1 (we shall alwayssupposethata > b), we see that(16, 7) is trans-
formedbythealgorithmto (9, 7), thento (2, 7); theremainderis now less thanthe
originalsmallernumber,so we reversetheroles of thetwo numbersand proceed:
(7, 2) to (5, 2) to (3, 2) to (1,2); reversingagain: (2, 1) to (1, 1) to (0, 1); and here
the process terminates.Then thispattern:two subtractions;threesubtractions;
two subtractionscharacterisesthe ratio, and, for convenience,I shall writeit
hereas 16:7 = [2, 3, 2]. We do not have anywherein the survivingevidenceany
explicitand unambiguousreferenceto anthyphairetic ratios; indeed we do not
have any evidenceof how ratiosof magnitudeswerehandledbeforethe develop-
mentoftheEudoxanproportiontheoryin Book V ofEuclid's Elements,butI have

This content downloaded from 200.26.133.57 on Thu, 1 Aug 2013 23:41:01 PM

All use subject to JSTOR Terms and Conditions
70 D. H. Fowler

argued(in[2],[3],and[4])thatthereis goodcircumstantialevidenceforbelieving
thattheremightthenhave beena highlydevelopedand subtleanthyphairetic
ratiotheory.
Riddell described(pp. 7-9) howEudoxanproportion theoryis relatedto the
patternsetup bytheoccurrences oftwoperiodicevents;letme heredescribea
similarpatterngenerated by twolinesegments a and b (a> b), and laterrefer
to thecontext
briefly ofperiodic, events.On thesameline,
astronomical,
typically

A' A2 Ai A% As
0| 1 H- i 1- S i- h 1 r1 1 r1-! r-
Bx Bi ft ft ft ft Bi ft ft fto flu B'i Bu

set out OAlA2A3... witha = OAt = AXA2= A2A3= ..., and OBiB2B3...
with b = OBX = BiB2 = B2B3 = ...; then the resultingpatternof points (say
B^^^B+AiBsBeA^BsBçAt ...) willcharacterisetheratioa:b. In factwhat
is definedin Euclid's Elements,V, Definition5, is proportion,or equalityof
ratios: a, 6, a, and V are in proportionifa and 6, and a and V both definethe
same pattern.In otherwords,wheneverAkcomes beforeBh Äk willcome before
ifka < Ib,thenka < lb' etc.
B¡, etc.;or,equivalently,
Thus the two proceduresfor definingratios lead to two different kinds of
the of subtractions, and in the
patterns:in anthyphairesis, alternating sequences
Eudoxan theory,thepatternscreatedby the two sequences multiples. these
of Are
twopatternsrelated?Clearlytheymustbe; but Riddell did notpursuethispossi-
bility,and onlyremarked(p. 9) that"thischaracterisation [ofratio,byanthyphai-
resis]does not resemble V, Definition 5."
Take, then,the patterngivenabove and see how we can deduce fromit some
of thetermsof theanthyphairetic ratio,a:b = [n0,nl9n2,. . .]. It is clearfromthe
firstthreetermsBlB2Al that n0 = 2, since their patternclearly implies that
2. OBX <OA1<2>. OBU or 2b < a < 3b; we next deduce from the pattern
(BlB2Al)(B3BAA2)(B5B6A3)B1B8B9AA9which generates successive multiples
of the intervalB2AU to B*A2 then B6A3 then BS(B9)A^ that ^ = 3, since
3(0 - 2b)<b< 4(a - 2b). Beyond this thepatternas a wholeso farwillagain
repeatitself,and will thus measure the size of the thirdorderdifference A3B7
second order difference and so more terms will be needed to
against the B2AU
determinethe nexttermn3; and so on.
Descriptionsof thepatternsthatcan occurand how theyare relatedto anthy-
in the
phairesisweregivenbyE.B. Christoffel [1] and H.J.S. Smith[6] 1875-6;
readermightpreferto deduce theprocedurefordecoding patterns the forhimself
ratherthanattempting to understandthefollowingrecursivedescription,and he
would benefitgreatlyfromtheexercise.Let AndenoteA repeatedn times; and if
S and Tdenote anytwosequencesof¿'s and B% letSnTdenote5 repeatedn times
followedbyT. Thensuccessively longerinitialsegments ofthepatterncorresponding
to the anthyphairetic ratio a:b = [n0, nu n2,...] are definedby: So = Bn°A...;
=
St = SpB ...; S2 = SÏ>S0 ...; ...; Sk+l Sp+iS^ ..., where the A's and £'s
are subscripted in increasingorder.Conversely, onlypatternsthatare susceptibleof
such a descriptioncan arise fromratios.Periodicpatterns(whichmusttherefore
containcoincidentA's and 5's similarto theinitialO in thefigure)correspondto

This content downloaded from 200.26.133.57 on Thu, 1 Aug 2013 23:41:01 PM

All use subject to JSTOR Terms and Conditions
 Ratioand EudoxanProportion
Anthyphairetic 71

commensurableratios and terminatinganthyphaireses;while, conversely,the

termination of the anthyphairesis is to be interpreted as an instruction to repeat
thecorresponding patternindefinitely.
It mightbe objectedthatthesepatternsofA's and fi'sare hopelesslyunwieldly.
Indeed theyare. I have recentlyproposed,in [4], that the reason that epimoric
ratiosplayso importanta role in Pythagorean and Academicmusictheorymaybe
connectedwiththeirsimpleanthyphairetic expansions; and the same may also
have been true of early theoreticalastronomy.Simple ratios would give rise to
simplepatterns,whichcould easilybe understood.But thedifficulties inherentin
of
decodingpatterns greatcomplexity mightbe yetanotherpressure,in addition
to those I have indicatedin [2], forcingattentionaway fromthe anthyphairetic
information encoded in the patternand towardsthepatternitself,and thusto the
Eudoxan definition of proportion.
One of the contributions of Riddell's articleis to finda commoncontextof
Eudoxus' reportedinvestigations in astronomyand in mathematics, wherehe is
celebratedfor contributionsto proportiontheoryand geometry.I would like
to finishby arguingthat anthyphairetic ratio theoryalso fitswell withinthis
same astronomicalcontext.Consider,for example, any two idealised periodic
astronomicalevents.A recordofmerelythesequence,and notthetimings,of these
two eventswill give rise to a patternsuch as I have just describedand, if this
patternis maintained, it willeventually carry,in coded form,sufficient information
for an approximationof the ratio of the two periodsto any requiredaccuracy.
(The decoded anthyphairetic ratios can be furtherdecoded into approximating
ratios of numbers.)Moreoverthe patternis self-correcting; if an observationis
or
ambiguous missing, it can be
eventually determined or reconstructed fromlater
information. Of course, this applies only to preciselyperiodicphenomena,the
"uniformand orderedmotions"thatPlato was said, by Eudemus(see [5], p. 3),
to have consideredas fundamental, and anysecularvariationswillplayhavoc with
thepattern;but thisdoes seemto accordwellwiththeattitudeto astronomythat
we findin the Republic,528e-531d.

Note.Whilewaitingfortheproofsof thisarticle,I was shockedto hear of the

deathof Ronald Riddell, fromleukemia,on January12th,1981,just beforehis
forty-third
birthday.AlthoughI knewhimonlyfromthecontents of his threearticles
in recentvolumesof thisArchive and one letteron thesubjectof thisarticle,I feela
personalsenseof loss. I wouldliketo dedicatethisarticleto hismemory.

References

1. E. B. Christoffel, Observatio Arithmetica,Annali di Mathematica, 2nd series 6

(1875) 148-152.
2. D. H. Fowler, Ratio in Early Greek Mathematics,Bulletinof theAmericanMathe-
maticalSociety,New Series,vol. 1 (1979) 807-846.
3. D. H. Fowler, Book II of Euclid's Elementsand a pre-Eudoxan Theory of
Ratio,
Archivefor Historyof Exact Sciences,22 (1980) 5-36.

This content downloaded from 200.26.133.57 on Thu, 1 Aug 2013 23:41:01 PM

All use subject to JSTOR Terms and Conditions
72 D. H. Fowler

4. D. H. Fowler, A New Speculationon theDiscoveryof Incommensurability, preprint.

5. R. C. Riddell, Eudoxan Mathematicsand the Eudoxan Spheres,Archivefor History
of Exact Sciences,20 (1979) 1-19.
6. H. J. S. Smith,Note on Continued Fractions,Messengerof Mathematics,2nd series,
6 (1876) 1-14; reprintCollected Works,vol. 2, Chelsea, New York, 1965, 135-147.

MathematicsInstitute
Universityof Warwick
Coventry
(Received September5, 1980)

This content downloaded from 200.26.133.57 on Thu, 1 Aug 2013 23:41:01 PM

All use subject to JSTOR Terms and Conditions