You are on page 1of 13

What is Algebra and What has it been in History?

Author(s): Hans Freudenthal


Source: Archive for History of Exact Sciences, Vol. 16, No. 3 (10.I.1977), pp. 189-200
Published by: Springer
Stable URL: http://www.jstor.org/stable/41133469 .
Accessed: 19/06/2014 21:18

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 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
WhatisAlgebra
and Whathas itbeeninHistory?
Hans Freudenthal
- What's in a name?
- Mit Wortenlässt sich trefflich
streiten.

0. WhoeverstartsreadingGreekmathematics is struckby largepartsthat


are overtlyalgebraicas well as other parts where algebra seemsto hideundera
geometricalcover.This fact
was a riddle for
many for as longas theproblems that
hadbesetGreekmathematicians andtheirsolutions werenotproperly understood.
The riddlewas solvedonlyafter19thcentury mathematics had facedthesame
problems and solvedthemina different way.
The viewthatlargepartsofGreekmathematics are algebraichas neverbeen
seriously EveniftherewerelittleleftofGreekmathematics,
challenged. onemight
concludefrom reportson integrals as calculatedbyArchimedes andfromreports
onnumerical astronomy thattheGreeksmusthavebeeninpossession ofpowerful
algebraictools.
S. Unguru* has recently challenged thisview.All whohavewritten about
Greekmathematics havebeenwrong, he claims.On whatgrounds?Has he dis-
coveredsensationalnewfacts?No, nothing!He has not eveninterpreted old
factsina newway.He simplysaystheyarewrong, and doesso withresounding
rhetoricalemphasis.Iftherhetoric is disregarded,theremainder consistsoflarge
extractsfromthe workof others,decoratedwithnumerousexclamation and
questionmarks, and a few, more concise statements, which can properly be
submitted to analysis.

1.Itismostdifficult
todiscover a generallineorargument
inUnguru'spaper.
The veryfewpropositions fromEuclid withwhichhe choseto makehispoint
aretakenfromcovertratherthanovertalgebra.In orderto realisethattheyare
covertalgebra,one has to knowmoreaboutGreekmathematics thanjustthese
fewpropositions; one shouldat leastknowhow thesepropositions havebeen
appliedinnumerable times.The propositions I mean,Elements 115and VI28,
containtermssuchas line,square,rectangle
andparallelogram,which, as happens
in Greekmathematics, maymeancertaingeometrical or algebraicideas
figures
likemagnitude, "square",product.The proofsthatfollowthepropositions are
frankly, not
though completely, geometrical. 115and VI28
Viewedgeometrically,
lookbadlymotivated anditishardtounderstand howsuchunattractivetheorems
* On theneedto rewritethe
historyofGreekmathematics.
ArchiveHist.Ex. Sci. 15,67-1 14(1975).

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
190 H. Freudenthal

couldhavebeencontrived bya geometrical mind.Mathematicians whoreadsuch


workare accustomed to ask foritsaim;theytryto findouthowthesetheorems
wereapplied.In thepresent casethisis easilydone:itappearsthattheseproposi-
tionswereusedas algebraictoolswithin Greekgeometry. Thetruthofthisasser-
tioncan easilybe checked : onehasonlytoascertain whetherinsomeapplication
thetermsline,square,rectangle, parallelogram have their
genuinegeometrical or
theiralgebraicmeaning. In almostall applications itturnsoutthatlines,squares,
rectangles,andparallelograms arenotinvolved butrathermagnitudes, "squares",
and products.Indeed,whenapplyingthetheoremone is allowedto disregard
theirgeometrical connotation, whichin factcan onlyhamperthe progressof
thought.This isa quitenormal ofmathematics.
feature TakeCalculusas anexamp-
le.ThewayinwhichtherecipesofCalculusareproveddiffers greatly fromthatin
whichtheyareapplied,andinordertoapplythemas mechanically as issometimes
required,youareoftenbetter advisedtoforget theprecisedefinitionsofderivative
and integral.As comparedwithhighbrow analysis,Calculusis lowbrowalgebra,
and thisis exactlythepositionof115and VI28 withrespectto theirbackground
theory.
2. It is hardto drawor to understand thisconclusion
as longas 115andVI28
are isolatedfromtheirhistorical context.
You haveto knowmoreaboutGreek
mathematics, and if Unguru findsfaultwithmathematicians and historians
before him,I thinktheironlyfaultis theirbroaderapproachto sources.
Unguru suggestsa quitedifferent originforthecommoninterpretation of
115and VI28: peoplediscovered thatyoucan notedownthesepropositions in
modernalgebraiclanguage,and thenconcludedthattheywerealgebra,geomet-
ricallydisguised.Thisbringsus tothequestionofhowGreekmathematics should
be edited.In fact,thereare various levelson whichthiscan be done.In a plain
translation, such as by T. L. Heath* the Greektext(VII) mayappearin the
version
as A is to £,so letC be to D,
and,as C is to D, so letE be to F.
I say,that,as A is to B, so is E to F.
whereasina comment or ina summary youmightfind:
Algebraically, ifa : b= c : d
and c:d = e:f
thena:b = e:f.

No doubt this is allowable but it would be absolutelyinadmissiblein the same


contextto replace propositionslike a:b = c:d by theirmore modernanalogues
ad = bc. It would not onlyspoil thecontextbut even make nonsenseofit.On the
otherhand,thereare manycases where,in commentariesor summaries,propor-
tions may be replaced by equivalentproductequations,and certainlyat points
whereancientmathematicianswould not have acted differently. Some delicacy
* T. L. Heath, The 13 books
ofEucliďs Elements.

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
What is and was Algebra? 191

is neededto knowin anyparticular case whichlanguageis mostsuitable.After


havingcheckedanewmanyinstances, I cannotbutaffirm thathistorians ofmathe-
maticspossessthisdelicacytoa highdegree*.Anyhow, itisunwarranted toquote
modernstylealgebraicalformulasfromhistoriansof mathematics without
identifying thelevelofpresentation to whichtheybelongand to insinuate that
conclusions aredrawnfromwrongtranslations.
3. Elements 115andVI28 havebeencharacterised as geometrically disguised
algebra.Severaltimes(e.g.p. 83) Unguru asks whygreatalgebraists suchas
Theaitetos,Archimedes, and Apollonios (he could have added more)should
haveinvented sucha disguiseandneversheditsyoke.Theanswerhasbeengiven
so manytimesinso manydetails, andeveninworkquotedbyUnguru(withparts
of theanswerincluded), thatit is hardto understand whyUnguru does not
knowit.Buttomakethings clear, I will give what I thinkare thethreemainreasons.
First,historical: It just happened that the Greek end ofthetorturous path
through foundations of mathematics, Eudoxos' theory, so excellent
was thatthe
Greeksdidnotaspiretoa better one.Indeed,eveninourowntimemathematicians
arestillfascinated byEudoxos' solution.
Second,philosophical: Thoughin dailyuse by laymenas well as mathe-
maticians, fractions weretaboo in highbrow mathematics, becausephilosophy
forbade thedivisionoftheunit.
Third,traditional: Once canonised,theElements weresacrosanct, liableto
additionsbut not to change.The mathematical community was small.To be
understood withinit,youhad to quoteEuclid and to speakhislanguage.
4. AfterthesegeneralremarksI shallnow startdicussingUnguru's more
concisestatements mentioned at theendoftheintroduction.
Quotations :
[a] ... ifthe mostfundamental difference between thealgebraic andgeometric
modeofreasoninglies,and I thinkit does,in thedistinction between
symbolic and extensive (i.e.,spatial)magnitude ... (pp.75-76)
[b] Thereis nottheslightest shredofgenuine historicalevidence thatEuclid
(or theother greatHellenistic mathematicians, letalone the Pythagoreans^
everusedequations intheirgeometrical works.Thesourcesdo notcontain
equations, (p.91)
[c] Do we find, then,anyalgebrain Euclid? (p. 113)
[d] I doubt it! Euclid's numbers aregivenline-segments, no abstract (p. 113)
[e] symbols, and Euclid's presentation is notsymbolic. It always(p. 113)
[f] dealswithdeterminate numbers of unitsof measurement whichare not
seenas representing specific illustrations, instances ofa (p. 113)
[g] concept ofgeneralmagnitude, (p. 113)**
* In this respectHeath is
particularlyadmirable.Unguru nevermentionsE. J.Dijksterhuis
(De elementen van Euclides I (1929), II (1930), Groningen,Noordhoff,and Archimedes, Copenhague
1956)whointroduceda symbolismadaptedto theGreekrhetoricalstyleofmathematics and to showing
itsalgebraicfeatureseven moreclearly.
** It is notclearwhetherthelastsentenceis a
paraphraseofa statementbysomebodyelse,though,
in any case, Unguru advances it as his own statement, and it summariseswell the main contentsof
thepaper.

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
192 H. Freudenthal

5. Inregardto [a] and[d] : Thephrase"symbolic andextensive magnitude" is


likesaying"written numerals and real numbers". Thereis indeeda widespread
opinionthatalgebra(andmathematics as a whole)is aboutsymbols, thatis,letters
and operationsigns.Accordingto a quotation(p. 113) from elsewhere (cf.the
to
footnote [g])
thesymbolic procedure [identifies]theobjectrepresented withthemeansof
itsrepresentation.
Thisis preciselywhatneverhappensin mathematics, be it algebraor geometry,
ancientor modern.It does not happen,becauseit mustnot happen.As every
mathematician knowsand as everyfreshman is warned,thisis thequickestway
tobecomeentangled inwildconfusion andcontradiction. Symbolsas understood
herearenottheobjectsofmathematics, as Unguru suggests, butrather theyare
partofthelanguageby whichmathematical objectsare represented.* Symbols
havebeenusedto do and to communicate mathematics, fromBabylonian times
downto our own.Symbolism has developedto evergreaterefficiency, as have
everyday languageand othercommunication techniques. Theefficiency ofa sym-
bolismis determined bytheease withwhichtheusercan movewithinit,bythe
algorithmic autonomy itprovides.Itisthevirtue ofsymbolism thatitallowsusfor
mostof thetime,ratherthanidentify object and symbol(whichis nonsense),
to forgetaboutwhatthesymbolmeans.It is indeedeasierto dealwith
(a + b)2= a2+ 2ab + b2
thanwith
thesquareoverAC equalsthesquaresoverAB andoverBC together withthe
doubleoftherectangle containedby AB and BC,
but evena modernmathematician, spoiltby our algebraiclanguage,can to a
certainextentlearnto speak,and to movein,theancientlanguage.Ofcoursehe
willneverattainArchimedes' skillindoingso,forthesimplefactthatheneednot.
So muchfor"symbolic"magnitude. Magnitudeis not symbolic, but it is
expressed bysymbols. Thismustbe so,becausetheonlyway to deal withmagni-
tudeis through symbols- exceptwhenwe manipulate them,using,forinstance,
a ruleror a balance.

6. Inregardto [d] and[e]. Numbers arelinesegments, andEuclid's presenta-


tionis notsymbolic. Is thistrue?Ofcoursenot.Almostneverin theElements or
anywhere elseinGreek mathematics doesAB mean a lineor a line It
segment. does
meana magnitude. It can evenhappen,and itoftendoes,thatthereareno points
A, B relatedto themagnitude, i.e.,AB is meaningful as a magnitude, whereas
A and B do not specify points.Similarly in Greekmathematics theexpression
"AB and CD takentogether" - however "addition"is described bya circumlocu-
-
tion can be nothingmorethanthesymbolicexpression forthe sum of two
magnitudes. At almostall placeswhereyouread"thesquareon AB"' youmay
forgetaboutdrawinga square,or imagining one,and youcan be surethatno
Greekmathematician everdid.Ifa passage abouttherectangle
is containedby
* To be sure,modernformalism mathematicsas a language.This,however,
allows us to interpret
fromwhatwe are used to callingmathematics.
is a tediousjob, and the resultis quite different

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
What is and was Algebra? 193

AB and CD, thenmostoftenno suchrectangle is to handand itis impossible


to
drawone,becauseABandCD arenotorthogonal toeachotheratonepoint(itmay
evenhappenthatone"side" is partoftheother).The Greeksavoidedaddingup
lengthand widthand area,as theBabylonianalgebraistsdid whenformulating
buttheysucceedednevertheless
theirproblems, in addingand multiplying the
factors
strangest whileonlyobeying certainconventionalrules.

7. In regardto [a]. Extensivemagnitude is opposedto intensive ratherthan


to symbolic magnitude, thelatter a
being logical monster. On the otherhand
extensive is
magnitude certainly not thedomain reserved to geometry and for-
biddento algebra.I am pretty surethatup to thebeginning ofwhathas been
called" arithmetisation
"
nobodyeverdoubtedthatalgebraas usedin geometry
as wellas inanalysiswasaboutextensive magnitude, andI canillustratethisview
byquotations from Descartes to Cauchy.
Ofcoursetherewereshadesof"extensivity". In H. Grassmann's Ausdehnungs-
lehre(theory ofextension)stresson extensionis particularlystrong,butnobody
woulddoubtthattheAusdehnungslehre is algebra,and thisnameis evenmade
explicitin vectorand tensoralgebra,thepresent versionofAusdehnungslehre.

8. In regardto [b] and [g]. Afterhavingfirmly deniedtheexistence ofany


algebrain Greekmathematics, Unguru weakenshis bold assertions to strong
doubt.As a matter offact,a fewpropositions fromtheElements mightprovetoo
smalla basisonwhichtoconcludemore.Euclid,nottomention Archimedes and
Apollonios,mighthavewritten more.Butas to Euclid's Elements thereis an
easywayto findoutwhether thereis anyalgebrain it.Unguru nevermentions
Elements V. Thisfact,together withhisstatement abouttheabsenceofalgebrain
Euclid, seemsto indicatethathe overlooked Elements V.
Indeed,Elements V is algebra,and nothing else,andI cannotimagine anyone's
challenging thisstatement. To be sure,itis notgeometric algebra-exorcisedby
Unguru- sincethereis no geometry at all in Elements V, wheretermslikeline,
square,rectangle are entirelyabsent.It is a generaltheory ofmagnitude - non-
existentin Greekmathematics -
accordingto Unguru independent of dimen-
sionor anycharacteristic ofspecificmagnitudes. In orderto distinguish thepure
algebraofV fromthematter ofII andVI, whichnecessarily dependson linesand
thelatterhas beenchristened
rectangles, geometric algebra.What'sin a name?
Withor without thatadjectivethesubjectmatter is stillalgebra.

9. Whatis algebra?Thereis no SupremeCourtto decidesuchquestions.


Nevertheless, "algebra"has a meaningineveryday languagejustas "chair"and
"table" have. For instance, at schoolalgebrais solvinglinearand quadratic
equations. Itis thekindofalgebratheBabylonians started
with.Wastheiralgebra
not algebra,becausetheirsymbolism was not smoothenough?Are "length"
and"width"muchworsethan"x" and"ynifyoucangiveclearrecipesforsolving
quadraticequationsin suchterms?Is it not algebraifthesumof thefirst10
squaresis laid downina numerical formula thatallowsone to extendtheresult
to anyn?Thisabilityto describerelations and solvingprocedures,and thetech-
niquesinvolvedin a generalway,is in myviewof algebrasuchan important

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
194 H. Freudenthal

featureofalgebraicthinkingthatI am willingto extendthename "algebra" to it,


as long as no othername is proposed,and as faras I know no othername has
been put forward.But whatis in a name?

10. In regardto [b]. IftheGreeksdid whattheBabylonianshad done before,


and what our school childrendo now, and called algebra solvingequations,is
thisthennot algebra?
The sources do not contain equations, Unguru says. "Equation" has three
meanings:
formalidentity,
conditionalequalityinvolvingunknownsto be made known,
conditionalequalityinvolvingvariables.
Whichmeaningsare absentin Greek mathematics?Is

(a + b)2= a2 + 2ab + b2

notan equation,ifitis formulated inwords?Are the"symptoms"forcircle,parab-


ola, ellipse,hyperbolanot equations,just because theyare writtenin thelanguage
ofrectanglesand squares? And finallywhatabout linearand quadraticequations?
Of course115is not thesolutionofa quadraticequation,butnobodyeverclaimed
itwas. VI28, however,is explicitlyformulated and theproblem
as problem-solving,
is a quadratic equation not fora numberbut fora magnitude.To be sure,the
Elementsis no textbook on solving equations. Such a text,however,exists:
Euclid's Data. I am prettysure if Unguru had been aware of its existence,he
would never have claimed there were no equations in Greek geometry.The
invariablestyleof theData is

Givencertainmagnitudesa, b,c and a relationF(a, b,c, x), thenx, too,is given,


and the proof of such a statementconsists in exhibitingeveryparticularstep
leadingfroma, b, and c to x.
For instance,ify is a linearfunctionofx, or in Euclid's language,
ifVis greaterthanx in ratio,
a shorthandphrasefor
ifv is up to a givenmagnitudegreaterthanx in givenratio,
thatis
y= ax + fc,
thenone shows how to findx fromgivena, b,y.

Or anotherexample:
iftwo segmentscontaina parallelogramundera givenangle,while the one
of themsquared is a givenarea greaterin ratio than the other,thenboth of
themmustbe given.

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
What is and was Algebra? 195

This is solved in a sophisticatedway,avoidingbiquadraticrelations.The method


ofsolutionis formalalgebra;no lines,squares,rectanglesin thesenseofgeometry
are involved.

11. ElementsV displaysEudoxos' marvelloustheoryof generalmagnitude,


and the transformations thatconstitutethe techniqueof algebraicoperationsin
Greek mathematics:alternation,inversion,composition,separation,conversion.
ElementsV is, as it were,a lineartheory,thatis, a theoryin whichno operation
changesthe genus of magnitude.ElementsVI is different. It deals withproducts
of
(rectangles) magnitudes, and forthisreason it is restricted
to geometricmagni-
tude, lines and rectangles. Elements VI takes up the equivalence of a:b = c:d
and ad = bc, whichis impossiblein ElementsV. It also deals with"compounding",
thatis, multiplying,theratios
a : b and c : d
into
a c : b d,
the most fertileamong the operations.It is unfortunatethat compoundingis
restrictedto threefactors,whereasin lengthyreasoningsone may need more.To
overcomethisdrawback,a sophisticatedtechniqueis applied.Indeed,compound-
ingis also feasiblewithintheframeofthepurealgebraofElementsV, for
a : b and b : c
maybe compoundedinto
a : c,
thecase called ex aequali in ElementsV. If
a : b and c : d
are to be compoundedusingthemeans of ElementsV, cumbersometransforma-
tionsare required.Unguru (p. 71,72) ridiculesvan der Waerden's characterisa-
tionof thismethodas cumbersome.Certainlyhe had not noticedthattheGreek
mathematicians,ratherthanthehistorians, feltthistechniqueto be a burden.This
is provedby a numberof cases in whichthe Greeks applied the easy methodof
ElementsVI, cases in whichit was not allowed,ratherthanthecumbersomeone
oftransformation accordingto ElementsV.

12. When I preparedthis paper, I looked forsome simple and convincing


examplesof algebraicprocedurein Greek mathematics.I opened Archimedesat
randomthrice,and twiceI was successful.
The firstexampleis fromArchimedes,Peri Hel 10:
Ifan arbitrarynumberoflinesexceedeach otherby thesame amount,and
iftheexcessequals thesmallest;ifmoreoverthesame numberoflinesare equal
to the largest;thenthe squares of the equal lines togetherwiththe square of
thelargestand therectanglecontainedby thesmallestand all thelinesexceed-
ingeach otherbythesame amountequals thricethesquaresofthelinesexceed-
ingeach otherby thesame amount.

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
196 H. Freudenthal

Heath's versionof thistext*reads as follows:


If Al, A2, ..., Anbe n lines formingan ascendingarithmeticalprogressionin
whichthecommondifference is equal to Ax, theleastterm,then

(n+l)^2 + ^1(A1+^2 + --.+ XJ-3(^ + ^ + ...+^2)^.


(The more compactwording has been achievedby two means: replacingvernacular
by artificiallanguage, and numbering variables by cardinals, a quite recent
mathematicaltool. I trustthateverybodywill agree thatHeath's versioncan be
in everycontextthatdoes not stresspurelylinguisticfeatures.)
said to be faithful
The proposition is proved for n= 8. Rather than translateArchimedes'
-
proofI reproduceHeath's version,whichI assure you is faithful therectangles
havingbeen replaced by products,the verbal numeralsby numericalones, the
"together"by "plus", "equal" by theequalitysign,the number8 by n.
A' A2 An- 3 An - 2 An- l

An An- I An -2 A3 A2 A'
Let the lines An, An_i, An_2,... Ax be placed in a row fromleftto right.
Produce An_1,An_2,... A^ until theyare each equal to An, so that the parts
producedare respectively equal to Ax, A2, ... An_1.
Taking each line we have
successively,
2An1= 2An'
(Ax+An_,)1= Ax1^A'^^2Ax-An_x,
(A2-'- An2) =A2 -'-A „_2 + 2A2>/in_2,

(An_,+Ai)2 = A2n_1+A12+ 2An_l-A1.


And,by addition,
(n+l)An2 = 2(A12+A22 + -+An2)
+ 2A1-All_i+2A2-AH_2+ - + 2AK_1-Al.
in orderto obtain therequiredresult,we have to provethat
Therefore,
2(ArAn_l+A2-A„_2+--- + An_l-Al)+ Aí(Al+A2+A3+---+An)
= A12+ A22+ ---+An2...(a).
Now 2A2- An_2= Al ■4An_2, because A2= 2Al,
2A3 ■An_3= Al ■6An 3, because A3= 3A1,

2An_l-Al=Al-2{n-')Al.
* T. L. Heath, The worksof Archimedes, chapters,Cam-
in modernnotation,withintroductory
bridge1897.
** The same formulais foundin cuneiformtexts.

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
What is and was Algebra? 197

It follows
that
2(A1-An_1+A2-An_2+ ---+ An_1-A1)+ A1(A1+A2 + ---+ An)
= A1{An+ 3An_1+5An_2+ -+(2n-lAl}. W

can be provedto be equal to


Andthislastexpression
V+V + -+4,2-
For An2= Ax{n-An)
= AMn + {n-')An}
= Al{An+ 2(An_l+An_2+ -+Ai)},
- =
because (n l)An An_1+Al
+ An_2 + A2
+
+ A1+An_1.
+ An_3
SimilarlyA2n_1=A1{An_1+2(An_2 + -+A1)},

A22= A1(A2+ 2A1),

byaddition,
whence,
= Al{An+ 3An_1+5An_2
Al2+ A22+ A32+ ---+An2 + .--+(2n-l)Al}.
that
Thustheequationmarked(a) aboveis true;and itfollows
(n+l)AH2+ A1{A1+A2+ A3+ .~ + AJ= 3(Al2+ A22+ ...+AH2).
Heath'sversionisnotmuchshorter thanArchimedes'. Heath wascareful enough
to writeA{ratherthani a. He usesthelatternotationina noteat theend,which
indeedshortenstheproofconsiderably:
(n+l)n2a2 + a(a + 2a+--- + na) = 3{a2+(2a)2 + {3a)2+ '-'+{na)2}.
Thereforea2+ (2a)2+ (3a)2+ ••• + (na)2

_ 2 n(n+l)(2n + l)
6
In a certainsensethefirstversionwas too careful.
Writing ia = Atis evenmore
to thehiddenspiritoftheproof,
faithful as we shallsee shortly.
In anycase,whether youreadthistextinone wayor another, youwillnotice
thatitis plainalgebrawithno geometricalconnotations.
Perhapsyouwillfeelthatyouhavelostthethreadofthestoryinthismountain
offormulas. You areright.Butthisis thewaymathematics hasbeeneditedfrom
ancienttimesdownto ourown.To be understood a mathematical papermustbe
readbackwards, andthisis thecase withthepresent text.Indeedthisprovesthat
Heath'sversionis faithful
to theformofArchimedes' text.

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
198 H. Freudenthal

Now I willedittheproofon thebasis ofwhatwe knowabout historyofmathe-


matics,usinga bit of psychology.The propositionis not about geometry,or ifit
is geometry,it is the discretegenus,cultivatedby the Pythagoreans, excluded
fromhighbrowmathematics,subsistingas an undercurrent, and surfacinga few
centurieslater,withNikomachos of Gerasa and Theon of Smyrna(2ndcentury),
namelythefiguratenumbers.
Let us firstintroducenotationsfora fewfiguratenumbers:
thenthgnomongn= 2 n- 1,
n
thenihtriangledn= £ '*
f=i

the nthsquare qn= n2,


the /byj rectangleri } = i /,
n

the nthsquared pyramid pn= Y¿ *2-


i

Of coursetheproblembehindthepropositionis calculatingthelast expression-


even thisis hiddenby Archimedes'formulation.
Note
<7i+;= <7I+ 4/+ 2rij,
n

i= l

Put the squares qx, . . . , qn upon each other such that the upper left-hand
cornersare on a verticalline:

/7th (n-l)th (n-2)ih (tt-3)th ...layer


-
qnis dividedinto a sequence of gnomons,such thata wall of height(n i+ 1) is
gnomon.Thus
erectedon the z'th
n

p»= I «f("-' + 1).


; _ i

Omit thetriangleon thediagonal:

P.-í/. = 2¿(i-l)(n-i+l)
1= 1

= 2n^j(n-j)
n-l
= Y 2r

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
What is and was Algebra? 199

n-l

= (n-l)^-2pnl
= («+l)^n-2pnq;
thus
= (n+l)^fl+ £/fl,
3pfl

which, indeed,is Archimedes'


proposition.
Ofcourseall notionsfromdiscretegeometry wereafterwardseliminated.
The
resultis formalalgebraicreasoning.
One is remindedofa methodnamedafter
Abel,thoughitis a generalfeatureof"discreteintegration".

13. For theotherexample,Perikon.kai sph.II, I takethetranslation


from
Dijksterhuis(Archimedes,
p. 124):
//therebe linesequal to one another,in any number, and to each of thesethere
be appliedan area witha quadraticexcess,and thesidesoftheexcessesexceedone
anotherby an equal amountwhichis equal to theleast,and iftherebe otherareas,
equal innumber and inquantity
tothefirst-mentioned each equal to thegreatest,these
willbe toall theotherareas ina lessratiothanthatofthesumofthesideofthegreat-
estexcessandoneof theequal linesto thesumof one thirdof thesideof thegreatest
excessandone-half ofoneoftheequal lines,and to theotherareas exceptthegreatest
it willbe in a greaterratiothanthissame ratio.
Theterminology ofapplicationofareasis usedto define whatwewouldcalla
quadraticfunction, just as theterminology "larger than in ratio"in theData
is usedto definethelinearfunction.Theproposition is usedto calculateintegrals
relatedto thehyperbola, and thenthefunction is relatedto thesymptom ofthe
conic.
Heath (p. 110)putstheproposition intotheformula
IfA1,A2... Anbeanynumber
ofareassuchthat

Al =ax + x2,
A2= a-2x + (2x)2,
A3= a-3x + (3x)2,

An= a- nx-'-(nx)2,

then n-An:(A1+A2+ '- + An)<(a + nx): k +


^h

and ri'An:(A1+A2+ --'+An_i)>(a + nx): U + ^H-

Theproofis bysumminga geometrical


series,usingtheproposition
of12,anda
of It is formal.
property proportions. entirely

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions
200 H. Freudenthal

14. Whatis algebra?I havetriedto answerthisquestionbypointing to some


vaguefeatures andtoa fewpreciseexampleswhich, inmyview,everybody should
recognise as algebra.Unguru mightat leastbecomeconvinced thatthegeneral
viewaboutalgebrain antiquity is notentirely
unfounded.
Or is itmorethanterminology? Arithmetical
puzzleshavebeena fashionofa
fewformorethanfour millenia. Solving them hasmeant usinglinearandquadratic
equations.WereGreekmathematicians notacquaintedwithsuchproblemsor
notable to solvethem?Or,iftheywere,did theynevernoticethelinkbetween
solvingequationsand Elements 115and VI28? Is thisUnguru'sopinion?Then,
ifhistory ofmathematics I wouldliketo haveitdonebysome-
is to be rewritten,
body who shows higheresteem for ancient
mathematicians.

Instituut Wiskunde
Ontwikheling Onderwijs
Tiberdreef
Utrecht/Overrecht
Holland

(ReceivedJune8, 1976)

This content downloaded from 91.229.229.96 on Thu, 19 Jun 2014 21:18:16 PM


All use subject to JSTOR Terms and Conditions

You might also like