GENERAL  ARTICLE

Pythagoras and Diophantus

Uwe Storch

T h e T h e o r e m o f P y th a g o ra s is e m p lo y e d a s a
sta rtin g p o in t fo r d isc u ssin g th e irr a tio n a lity o f
th e sq u a re r o o t o f 2 a n d its im p lic a tio n s in M a th -
e m a tic s a n d c u ltu re . It m o tiv a te s to c o n sid e r th e
e q u a tio n o f th e c irc le o v e r th e ¯ e ld o f ra tio n a l
n u m b e r s a n d o th e r ¯ e ld s w h ic h le a d s in a n a t-
u ra l w a y to th e u se o f m o d u la r a r ith m e tic fo r
Uwe Storch is Professor
D io p h a n tin e e q u a tio n s. (emeritus) of Mathematics
at Ruhr University
L o n g b efo re th e b eg in n in g o f th e C h ristia n era , in m a n y Bochum, Germany. He has
a n cien t cu ltu res o n e ca n ¯ n d h in ts a n d d o cu m en ts th a t been a visitor to TIFR,
th e so -ca lled P y th a g o rea n th eo rem w a s k n ow n . It w a s Mumbai, and he has given
u sed m a in ly in th e fo rm : T h e sq u a re o f th e h y p o ten u se at various times lectures at
the Department of
c o f a rig h t-a n g led tria n g le is eq u a l to th e su m o f th e
Mathematics, IISc,
sq u a res o n th e o th er tw o sid es a ;b, n a m ely, c 2 = a 2 + b 2 Bangalore, on subjects of
(F igu re 1 o n th e n ex t p a g e). his main interests:
Algebra, in particular
T h e w id ely trav eled P y th a g o ras (ca . 5 8 0 { 5 0 0 B C ) w a s Commutative Algebra, and
w ell aw a re o f th is th eo rem , b u t, certa in ly, h e w a s n o t Algebraic Geometry.
th e ¯ rst w h o k n ew it. A s a ju sti¯ ca tio n o f th is th eo rem ,
o n e ¯ n d s ¯ g u res in th e fo rm o f p u zzles, fro m w h ich th e This is the translation of an ar-
d esired eq u a tio n ca n b e v isib ly o b ta in ed (see F igu re 2 ticle with the same title (‘and’
o n th e n ex t p a g e). E x a m p les in C h in ese tex ts a re o f replaced by und) published
online in a slightly shortened
C h o u P ei S u a n C h in g fro m th e H a n era (ca . 2 0 0 B C { version by the press centre of
ca . 2 0 0 A D ), w h o h a d a ccess to h isto rica l d a ta o f th e Ruhr University, Bochum (Ger-
seco n d m illen iu m b efo re C h rist1 . many) on December 22, 2005
as part of a Christmas calendar
T h e sq u a re sta n d in g o n th e v ertex in th e la st p u zzle in under the epigraph Was die
F igu re 2 illu stra tes th e eq u a tio n c 2 = (a ¡ b)2 + 2 a b. Menschen vor 2000 Jahren
It ca n b e fo u n d in In d ia in a tex t o f B h ^a ska ra (1 1 1 4 { wussten (What men knew 2000
years ago). The notes were
1 1 8 5 ?) w ith th e rem a rk : L o ok a t th a t! E u clid (a ro u n d
added later by the author.
3 0 0 B C ) in h is E lem en ts2 a lso u sed in h is ¯ rst p ro o f
(I, x4 7 ) th e eq u a lity o f su ch a rea s. T h is is a ca u se o f 1-15
Please see section 'Notes'
co n cern , sin ce th e p ro o f is n o t co n v in cin g , a s it is b a sed at the end of the article (p.699
o n a petitio prin cipii: T h e ca lcu la tio n w ith a rea s, onwards) .

RESONANCE  July 2009 691

 GENERAL  ARTICLE

esp ecia lly th e fa ct th a t th e w h o le is th e su m o f its p a rts,

ca n b e ju sti¯ ed o n ly la b o rio u sly. H ilb ert (1 8 6 2 { 1 9 4 3 )
h a s d em o n stra ted th is in th e fo u rth ch a p ter o f h is b o o k
G ru n d lagen der G eom etrie in 1 8 9 9 , w h ich g iv es a m o d -
ern a n d co m p lete a cco u n t o f th e g eo m etric p a rts o f th e
E lem en ts o f E u clid 3 . T h erefo re, E u clid 's seco n d p ro o f o f
th e T h eo rem o f P y th a g o ra s, w h ich u ses th e ra tio s o f th e
sid es o f th e sim ila r tria n g les A B C , C B D , A D C in F ig-
c2 = a 2 + b 2 u re 1 (cf. E u clid 's E lem en ts, V I, x8 ), is m o re co n v in cin g :
a = c = p = a , b= c = q = b, c 2 = c(p + q ) = cp + cq = a 2 + b 2 .
Figure 1. H ow p ro b lem a tic th e n a iv e ca lcu la tio n w ith a rea s a n d
v o lu m es co u ld b e, is sh ow n v iv id ly b y th e so -ca lled p a ra -
d ox o f B a n a ch { T a rsk i: In 3 -d im en sio n a l sp a ce, if b a lls
V ;W o f d ia m eters 1 a n d a n y c > 0 (e.g . c = 1 0 1 0 o r
c = 1 0 ¡5 0 ) a re g iv en , th en V a n d W ca n b e b ro k en u p
in to ¯ n itely m a n y p a rts V 1 ;:::;V r a n d W 1 ;:::;W r , re-
sp ectiv ely, so th a t, fo r ea ch i = 1 ;:::;r , th e p a rts V i
a n d W i a re co n g ru en t4 .
Keywords In a rig h t-a n g led tria n g le w ith eq u a l sid es a = b (o r in
Theorem of Pythagoras, irratio-
a sq u a re w ith sid e a a n d d ia g o n a l c), b y th e T h eo rem
nality, Diophantine equations,
modular arithmetic.
o f P y th a g o ra s o n e h a s c 2 = 2 a 2 . F ro m th is it fo llo w s
th a t b o th a a n d c ca n n o t sim u lta n eo u sly b e `n u m b ers',
if b y `n u m b er' o n e m ea n s { a s th e a n cien ts (n o t o n ly th e
G reek s) d id { a ra tio n a l n u m b er, i.e., th e ra tio r = s o f
tw o p o sitiv e in teg ers. T o w h o m w e o rig in a lly ow e th is
in co m m en su ra b ility resu lt w h ich o n ly th e G reek s k n ew
a n d w h ich p u zzled th em trem en d o u sly, is still d isp u ted 5 .
W e k n ow o f th e ca su a l a rg u m en t o f A risto tle (3 8 4 {
3 2 2 B C ) th a t a p o sitiv e in teg er ca n n o t b e b o th ev en a n d
Figure 2. o d d sim u lta n eo u sly. P erh a p s, w h a t h e m ea n t w a s th a t

692 RESONANCE  July 2009

 GENERAL  ARTICLE

a n y p o sitiv e in teg er r ca n b e u n iq u ely rep resen ted a s th e Irrationality of 2 .

p ro d u ct r = 2 k ¢u o f a p ow er 2 k o f 2 a n d a n o d d n u m b er
u . (T h is is a sim p le fa ct. F o r G reek s, th e ev en n u m b ers
w ith k ¸ 1 w ere fem in in e a n d th e ex p o n en t k ca n b e
u sed to g ra d e th e fem in ity o f a n u m b er: 2 4 = 2 3 ¢3 b ein g
m o re fem in in e th a n 3 6 = 2 2 ¢9 a lso w h en co n sid erin g th e
m a scu lin e p a rts 3 a n d 9 , resp ectiv ely.) N ow , su p p o se
2 = (r = s )2 o r, m o re g en era lly, 2 = (r= s )n w ith n ¸ 2 ,
r = 2 k ¢ u a n d , sim ila rly, s = 2 ` ¢ v . T h en it fo llo w s
th a t 2 s n = 2 1 + n ` ¢ v n = r n = 2 n k ¢ u n . S in ce th e p ow ers
v n a n d u n a re p ro d u cts o f o d d n u m b ers th ey a re o d d
(th is is th e h ig h lig h t o f th e p ro o f: a n y p ro d u ct o f o d d
p o sitiv e in teg ers is a g a in o d d 6 ) a n d h en ce 1 + n ` = n k
o r 1 = n (k ¡p `), w h ich is im p o ssib le fo r n ¸ 2 . T h e
sq u a re ro o t 2 o f 2 a lso a rises in th e fo llow in g p ro b -
lem , w h ich a g a in m u st h av e ca u sed a lo t o f trep id a tio n
a m o n g th e m u sic lov in g P y th a g o rea n s: If o n e w a n ts to
sk ip a n o ctav e w ith th e h elp of a to n e o f freq u en cy x in
su ch a w ay th a t th e ju m p from a p rim e w ith freq u en cy
a to th e d esired to n e is id en tica l w ith th e ju m p fro m
th is to n e to th e o ctav e o f freq u en cy 2 a , it fo llow s th a t
x : a = 2 a : x o r px 2 = 2 a 2 . T h u s, o n e h a s to to u ch a
strin g o f len g th `= 2 , if th e va ca n t strin g h a s len g th `,
a n d th ispis n o t a ra tio n a l p a rt o f th e va ca n t strin g . T h e
in terva l 2 : 1 is th e trito n u s ( = th ree tim es th e w h o le
to n e), in cla ssica l m u sic it is ca lled th e `d ia b o lu s in m u -
sica ', b u t it is p resen t ev ery w h ere in m o d ern m u sic, fo r
in sta n ce, in ja zz it is o n e o f th e `b lu e' n o tes.
A fter th e in a d eq u a cy o f (ra tio n a l) n u m b ers to d escrib e Separation of
g eo m etric q u a n tities w a s reco g n ized , A rith m etic a n d G eo - arithmetic and
m etry p a rted w ay s. T ill th e m o d ern tim es, th ey co n sti- geometry.
tu ted tw o d iscip lin es a m o n g th e ca n o n o f sev en lib era l
a rts (a rtes lib era les)7 . E u d ox o s o f K n id o s (ca . 4 0 0 { Exhaustion
ca . 3 5 0 B C ) h a d , h ow ev er, sh ow n a w ay o u t o f th is im - method and
p a sse in th a t h e ch a ra cterized a n y len g th b y th e en tire Dedekind
set o f ra tio n a l p ro p o rtio n s lesser th a n th is len g th . E u clid sections, see
a n d , a b ov e a ll, A rch im ed es (ca . 2 8 7 { 2 1 2 B C ) u sed th is Note 8 on p.702.

RESONANCE  July 2009 693

 GENERAL  ARTICLE

Origin of set `ex h a u stio n m eth o d ', b u t it rem a in ed , fo r th e m o st p a rt,

theory, see note 9 w ith o u t m u ch im p a ct. T h e C o p ern ica n tu rn , to de¯ n e a
on p.702.
rea l n u m b er a s a n o n -em p ty a n d strict sectio n o f ra tio n a l
n u m b ers w ith o u t a g rea test elem en t co n ta in in g w ith a n y
elem en t a ll ra tio n a l n u m b ers lesser th a n th is elem en t,
m a d e a n u m b er in d ep en d en t of a ll p h y sica l a p p ea ra n ce 8 .
T h is id ea w a s u sed n o t b efo re th e seco n d h a lf o f th e 1 9 th
cen tu ry u n d er th e in ° u en ce of th e set th eo ry d ev elo p ed
b y C a n to r (1 8 4 5 { 1 9 1 8 ). A fter th e p relim in a ry w o rk o f
B o lza n o (1 7 8 1 { 1 8 4 8 ) a n d C a u ch y (1 7 8 9 { 1 8 5 7 ), su b -
sta n tia l co n trib u tio n s w ere m a d e b y C a n to r, D ed ek in d
(1 8 3 1 { 1 9 1 6 ), W eierstra ss (1 8 1 5 { 1 8 9 7 ) a n d o th ers.
The equation of C a n to r ca lled \ a n y co llectio n in to a w h o le o f d e¯ n ite,
the circle over the d istin ct o b jects (w h ich a re called th e `elem en ts' o f th e
field of rational set) o f o u r p ercep tio n o r o f o u r th o u g h t" a set a n d co m -
numbers (and b in es th is w ay in a sp len d id m a n n er tw o fu n d a m en ta l
other fields). h u m a n a b ilities: to d i® eren tia te a n d to u n ite 9 .
T h ere a re p len ty o f in sta n ces, w h ere th e ra tio s o f th e
sid es a ;b;c o f a rig h t-a n g led tria n g le a re ra tio n a l. If
w e restrict o u rselv es to th e ca se, w h ere c = 1 (w h ich
a rises, w h en w e g o fro m (a ;b;c) to (a = c;b= c;1 )), th en
w e h av e to lo o k fo r ration al so lu tio n s x ;y o f th e eq u a -
tio n o f th e circle x 2 + y 2 = 1 . E q u a tio n s w ith su ch
restrictio n s o n th eir so lu tio n s a re ca lled D io p h a n tin e
eq u a tio n s, in h o n o u r o f D io p h a n t (a lso D io p h a n tu s) o f
A lex a n d ria (p ro b a b ly a ro u n d 2 5 0 A D ), w h o co n sid ered
th em in h is m a in trea tise A rithm etika. In th e ca se u n -
d er d iscu ssio n , o n e ¯ n d s th e ra tio n a l so lu tio n s ea sily
Figure 3. Stereographic
w ith th e u se o f stereo g ra p h ic p ro jectio n , w h ereb y o n e
projection.
¯ n d s th e p o in t (t;0 ), w h ich is th e in tersectio n p o in t o f
th e x -a x is w ith th e lin e jo in in g th e N o rth P o le (0 ;1 )
a n d a p o ssib le so lu tio n (x ;y ) 6= (0 ;1 ) o f x 2 + y 2 = 1 .
(S ee F igu re 3 .) In th is w ay, (x ;y ) co rresp o n d s to th e
p o in t (x = (1 ¡ y );0 ) o n th e x -a x is a n d , co n v ersely, (t;0 )
g iv es th e p o in t (2 t= (t2 + 1 ) ;(t2 ¡ 1 )= (t2 + 1 )) o n th e
circle 1 0 . O n e n o tices th a t th e (red ) in tersectio n p o in ts
w ith th e d ia g o n a l ca n n o t b e o b ta in ed in th is w ay b y

694 RESONANCE  July 2009

 GENERAL  ARTICLE

a ra tio n a l t, b eca u se 2 is n o t th e sq u a re o f a ra tio - Indian formulae for

n a l n u m b er (b u t, o f co u rse, o f a rea l n u m b er). If w e Pythagorean
w rite t = r = s a s a fra ctio n a n d ex p a n d , w e g et (fo r
triples.
r > s > 0 ) th e sid es (a ;b;c) = (2 rs;r 2 ¡ s 2 ;r 2 + s 2 )
o f a rig h t-a n g led tria n g le w ith in teg er sid es, e.g . th e
fa m o u s P y th a g o rea n trip le (4 ;3 ;5 ) fo r r = 2 ;s = 1
(w h ere th e co n v erse o f th e P y th a g o rea n th eo rem is u sed ,
w h ich is fo rm u la ted ex p licitly b y E u clid in th e E lem en ts,
I, x4 8 : w h en ev er c 2 = a 2 + b 2 , th en th e tria n g le is
rig h t-a n g led ). T h ese a n d sim ila r su ch fo rm u la e a s w ell
a s sp ecia l ca ses ca n b e fo u n d ev en b efo re D io p h a n t in
th e cu n eifo rm w ritin g ta b lets o f o ld B a b y lo n ia n tim es
(ca . 2 0 0 0 { 1 0 0 B C ), in th e In d ia n S u lb a S u tra s (ca . 5 0 0
{ 2 0 0 B C ) w ith p recu rso rs (S a ta p a th a B ra h m a n a ) fro m
ca . 1 0 0 0 { 8 0 0 B C { th ese fo rm u la e a re o ften ca lled
th e `In d ia n fo rm u la e' { a n d in C h in ese tex ts o f th e H a n
era . F u rth erm o re, in th e co n stru ctio n s o f th e th ird m il-
len iu m B C o f th e m eg a lith cu ltu re in E n g la n d , o n e ¯ n d s
rig h t-a n g led tria n g les w ith in teg er ra tio s o f sid es, lik e
(4 ;3 ;5 );(1 2 ;5 ;1 3 ) a n d (1 2 ;3 5 ;3 7 ), a n d in a ru d im en -
ta ry m a n n er a lso in a n cien t E g y p t. T h e P y th a g o rea n
trip les w ere u sed fo r co n stru ctio n o f rig h t a n g les. F o r
ex a m p le, o n e d iv id es a ro p e in to 1 2 eq u a l p a rts a n d
u ses it to sp a n a rig h t-a n g led tria n g le w ith sid es 4 ;3 ;5 .
B a sed o n th e ov era ll p resen ce o f sim ila r co n stru ctio n s,
V a n d er W a erd en (1 9 0 3 { 1 9 9 6 ) co n jectu red th a t th e
k n ow led g e o f th e P y th a g o rea n th eo rem m u st h av e h a d a
sin g le o rig in a n d sp rea d o u t fro m th ere 1 1 . T h is is, h o w -
ev er, d isp u ted . A rch ea lo g ica l ¯ n d in g s a re, in g en era l,
n o t a s ea sy to in terp ret a s in th e w ell-k n ow n ch ild ren 's
rh y m e \ ... In B a b y lo n o n e h a s fo u n d a w ire / T h e a n -
cien t B a b y lo n ia n s, th ey / K n ew a b o u t teleg ra p h y / /
... in N in iv e o n e h a s fo u n d n o w ire / S o th e p eo p le o f
N in iv e k n ew / A lrea d y a b o u t w ireless teleg ra p h y."
A s in d ica ted in F igu re 3 , o n e g ets fro m o n e so lu tio n
(x ;y ) , in g en era l, sev en o th er so lu tio n s o f th e eq u a - Symmetries of a
tio n x 2 + y 2 = 1 b y u sin g th e sy m m etries o f a sq u a re, square.

RESONANCE  July 2009 695

 GENERAL  ARTICLE

Figure 4.

¯ rst, b y ro ta tin g th e p o in ts (¡ y ;x );(¡ x ;¡ y ); (y ;x ) a n d

Octagon in ancient
th en , b y re° ectin g a t th e co o rd in a te a x es a n d a t th e
cultures.
d ia g o n a ls resp ectiv ely, th e p o in ts (x ;¡ y );(¡ x ;y ) a n d
(y ;x );(¡ y ;¡ x )1 2 . T h e in tersectio n p o in ts o f th e circle
w ith th e co o rd in a te a x es a n d th e d ia g o n a ls (a s fa r a s
th ey ex ist) fo rm g ro u p s w ith o n ly fo u r so lu tio n s ea ch .
Figure 5.
T h ey co n stitu te a n o cta g o n m a d e o f tw o sq u a res ro ta ted
w ith resp ect to ea ch o th er. S in ce a n cien t tim es th is h a s
p lay ed a n im p o rta n t ro le in co n stru ctio n s a n d a rt, w h ich
is certa in ly d u e to th e fa ct th a t th e n u m b er `eig h t' w a s
a h o ly n u m b er (b y th e w ay, a p u rely fem in in e n u m b er)
n o t o n ly in a n cien t a n d O rien ta l cu ltu res, b u t a lso in th e
C h ristia n tra d itio n . Its rela tio n to th e sq u a re ro o t o f 2
m a k es it a ll th e m o re in terestin g .
M ay b e, th e g rea test p erfectio n h a s b een a tta in ed b y th e
o cta g o n o f th e D o m e o f th e R o ck in J eru sa lem (see F ig-
u re 4 ).
It is, h ow ev er, b a sed o n B y za n tin e d esig n a n d ca n b e
co m p a red , fo r in sta n ce, to th e ch u rch o f S a n V ita le in
R av en n a (see F igu re 5 ). O n e co u ld ¯ n d th e o cta g o n a lso
in th e G ro tto , ca lled th e H o ly C ry p t, in th e (d estroy ed )
C o n sta n tin ia n C h u rch o f th e N a tiv ity in B eth leh em . A
b ea u tifu l ex a m p le, w h ich is still v isib le, is th e b a p tism a l
b a sin in C iv id a le d el F riu li (F igu re 5 ) fro m th e L o m -
b a rd ic era (8 th cen tu ry ).
A p ro fa n e ex a m p le is C a stel d el M o n te in so u th ern Ita ly
(see F igu re 6 ).

696 RESONANCE  July 2009

 GENERAL  ARTICLE

Figure 6. Castel del Monte

(13th century).

T h e o cta g o n (d eriv ed fro m th e sq u a re) p lay s a lso a fu n -

d a m en ta l ro le in H in d u relig io n a n d a rt a s m ay b e illu s-
tra ted b y th e ex a m p les in F igu re 7 .
W e w a n t to d raw a m a th em a tica l co n clu sio n fro m th e
g iv en sy m m etries. (`S y m m etry ' is a m a g ic w o rd in m a th -
em a tics a n d p h y sics to d ay.) L et p b e a p rim e n u m b er
¸ 3 . W e co n sid er th e eq u a tio n x 2 + y 2 = 1 o n ly fo r
in teg ers a n d id en tify tw o in teg ers, a s d o n e a lrea d y b y
G a u ss (1 7 7 7 { 1 8 5 5 ), p rov id ed th ey a re co n g ru en t m o d -
u lo p , n a m ely, w h en th ey d i® er b y a m u ltip le o f p 1 3 .
F u rth er, a s w a s d o n e b y L eg en d re (1 7 5 2 { 1 8 3 3 ), fo r a n
in teg er a , w h ich is n o t d iv isib le b y p , w e set (a = p ) = 1
o r (a = p ) = ¡ 1

Figure 7.

RESONANCE  July 2009 697

 GENERAL  ARTICLE

o r (a = p ) = ¡ 1 d ep en d in g o n w h eth er a n in teg er t ex -
Diophantine
ists su ch th a t t2 a n d a a re co n g ru en t m o d u lo p o r n o t,
equations and
resp ectiv ely 1 4 . A g a in u sin g stereo g ra p h ic p ro jectio n 1 5
modular arithmetic.
w e ¯ n d th a t th e n u m b er o f so lu tio n s o f x 2 + y 2 = 1
m o d u lo p is p + 1 if (¡ 1 = p ) = ¡ 1 , a n d (p ¡ 2 ) + 1
if (¡ 1 = p ) = 1 . (D u e to th e d en o m in a to r t2 + 1 in th e
fo rm u la e fo r th e stereo g ra p h ic p ro jectio n , tw o p o in ts a re
ex clu d ed if (¡ 1 = p ) = 1 .) H en ce, in b o th ca ses w e ¯ n d
p ¡ (¡ 1 = p ) fo r th e n u m b er o f so lu tio n s. O n th e o th er
h a n d , th e a b ov e m en tio n ed sy m m etries g iv e fo r th e n u m -
b er o f so lu tio n s th e ex p ressio n 8 k + 4 + 2 ((2 = p ) + 1 ) ,
w h ere k is th e n u m b er o f g ro u p s o f 8 so lu tio n s ea ch .
(In ca se (2 = p ) = ¡ 1 th e 4 in tersectio n p o in ts o f th e
d ia g o n a ls y = § x w ith th e circle d o n o t a p p ea r!) T h e
resu ltin g eq u a tio n p ¡ (¡ 1 = p ) = 8 k + 4 + 2 ((2 = p )+ 1 ) im -
m ed ia tely g iv es tw o v ery im p o rta n t th eo rem s o f N u m b er
T h eo ry : It is (¡ 1 = p ) = 1 if a n d o n ly if p is co n g ru en t
to 1 m o d u lo 4 , a n d it is (2 = p ) = 1 if a n d o n ly if p is
co n g ru en t to 1 o r 7 m o d u lo 8 (B ox 1 ). If b o th resu lts
a re co m b in ed , o n e g ets (¡ 2 = p ) = (¡ 1 = p )(2 = p ) = 1 if
a n d o n ly if p is co n g ru en t to eith er 1 o r 3 m o d u lo 8 .
N ow a d ay s, su ch m eth o d s o f m o d u la r a rith m etic a re o f
su p rem e im p o rta n ce in th e stu d y o f D io p h a n tin e eq u a -
tio n s a n d h av e ev o lv ed in to a fa r rea ch in g th eo ry. F o r
ex a m p le, u sin g th e a b ov e resu lts o n e ca n o b ta in th e fo l-
low in g fa m o u s th eo rem s ra th er ea sily ( p is still a n o d d
p rim e n u m b er): T h e D io p h a n tin e eq u a tio n s x 2 + y 2 = p ,
x 2 ¡ 2 y 2 = p (o r 2 y 2 ¡ x 2 = p ) a n d x 2 + 2 y 2 = p h av e in -
teg er so lu tio n s if a n d o n ly if p is co n g ru en t to 1 m o d u lo
4 , o r co n g ru en t to 1 o r 7 m o d u lo 8 o r to 1 o r 3 m o d u lo
8 , resp ectiv ely. (F o r th e p ro o fs o f th ese resu lts, see B ox
2 ). S u ch w o n d erfu l resu lts fo r D io p h a n tin e eq u a tio n s
w ere o n ly o b ta in ed a fter th e ren a issa n ce o f m a th em a t-
ics a t th e tu rn o f th e 1 7 th cen tu ry. In N u m b er T h eo ry,
Complements of F erm a t (1 6 0 1 { 1 6 6 5 ), E u ler (1 7 0 7 { 1 7 8 3 ), L a g ra n g e
the quadratic (1 7 3 6 { 1 8 1 3 ), L eg en d re, G a u ss a n d m a n y o th ers w ere
reciprocity law. a n a ctiv e p a rt o f th is d ev elo p m en t.

698 RESONANCE  July 2009

 GENERAL  ARTICLE

B ox 1.

These two theorems are the complements to the law of quadratic reciprocity (p = q ) (q = p ) =
(¡ 1) (p ¡ 1 )(q ¡ 1 )= 4 , where p ;q are distinct odd prime numbers. Note that the result for
(¡ 1= p ) is a special case of Euler's formula1 4 . The given proof by counting points on the
unit circle over the ¯nite ¯eld IF p = ZZ = ZZ p goes back to V A Lebesgue (1791 { 1875) . By
the way, for a n y a 2 IF £p the number of points on the ` circle' f(x ;y ) 2 IF 2p : x 2 + y 2 = a g
is p ¡ (¡ 1= p ) . Namely, there is a point (x 0 ;y 0 ) 2 IF 2p with x 20 + y 02 = a ¡ 1 , because the
set fa ¡ 1 ¡ x 2 : x 2 IF p g µ IF p contains (p + 1) = 2 elements and hence at least one square.
Then x 2 + y 2 = a is equivalent with (x 0 x ¡ y 0 y ) 2 + (y 0 x + x 0 y ) 2 = (x 20 + y 02 ) (x 2 + y 2 ) = 1
which gives a bij ection of the given ` circle' with the unit circle over IF p . (The number of
solutions of x 2 + y 2 = 0 in IF 2p is p 2 ¡ (p ¡ 1) (p ¡ (¡ 1= p ) ) ; i. e. , (1+(-1/p) ) p-(-1/p) :)

Lebesgue's method can also be used to establish the main part of the quadratic reciprocity
law now counting points on the unit sphere S q ¡ 1 = S q ¡ 1 (IF p ) := f (x 1 ;:::;x q ) 2 IF qp : x 21 +
¢¢¢+ x 2q = 1g over IF p . Quite generally, one has the cardinality formula #S 2 n = p n (p n +
(¡ 1= p ) n ) for n 2 IN which one proves (perhaps) most easily by induction on n using
the recursion #S 0 = 2 , #S m + 2 = p m + 1 (p ¡ (¡ 1= p ) ) + #S m ¢(¡ 1= p ) p (one counts the
elements in the ¯bres of the proj ection S m + 2 ! IF mp + 1 , (x 1 ;:::;x m + 1 ;x m + 2 ;x m + 3 ) 7!
¡ ¢
(x 1 ;:::;x m + 1 ) ) . In particular, #S q ¡ 1 = p (q ¡ 1 )= 2 p (q ¡ 1 )= 2 + (¡ 1) (p ¡ 1 )(q ¡ 1 )= 4 . On the
other hand, from any solution (x 1 ;:::;x q ) 2 S q ¡ 1 one gets by reiterated cyclic per-
mutation q ¡ 1 other solutions (x 2 ;:::;x q ;x 1 ) ;:::;(x q ;x 1 ;:::;x q ¡ 1 ) 2 S q ¡ 1 except for
the case that the given tuple (x 1 ;:::;x q ) is a constant one (x ;:::;x ) . Then q x 2 = 1
or q = (x ¡ 1 ) 2 in IF p . Hence, the number of such constant (= diagonal) solutions is
1 + (q = p ) . Altogether, #S q ¡ 1 = k q + 1 + (q = p ) for some k 2 IN . Comparing both
expressions
¡ for #S q ¡ 1 and using 14
¢ again Euler's relation p
(q ¡ 1 )= 2
´ (p = q ) mod q yields
(p ¡ 1 )(q ¡ 1 )= 4
(p = q ) (p = q ) + (¡ 1) ´ 1 + (q = p ) mod q , which is the quadratic reciprocity
law. Cf. also x71, Exercise 40 in G Schej a and U Storch, L eh rbu ch d er A lgebra , Teil 2,
Stuttgart: Teubner 1988 (an English edition by D Patil is in preparation) , or F Lem-
mermeyer: Kreise und Quadrate modulo p , M a th . S em esterberich te Vol.47, pp.51-73,
2000.

N o te s fo r S u p e r sc r ip ts 1 -1 5 in T e x t
1
H ere a n d in th e fo llow in g th e h isto rica l rem a rk s a re
ra th er su p er¯ cia l. F o r m o re p recise in fo rm a tio n o n e
h a s to co n su lt b o o k s o n th e h isto ry o f m a th em a tics,
fo r ex a m p le C B B oy er, a n d U C M erzb a ch , A H is-
tory of M athem atics, W iley, N ew Y o rk , 1 9 8 2 . A lso v ery
in terestin g is J S tillw ell, M athem atics an d its H istory,
S p rin g er, N ew Y o rk , 2 0 0 2 .
2
F o r a n E n g lish tra n sla tio n o f E u clid 's E lem en ts see:
T L H ea th , T he T hirteen B ooks of E u clid's E lem en ts,
rep r. D ov er, N ew Y o rk , 1 9 5 6 .

RESONANCE  July 2009 699

 GENERAL  ARTICLE

B ox 2.
p p
To prove these results, consider the quadratic algebras ZZ [ D ] = Z © ZZ p D (µ CI ) for D =
¡ 1;2;¡p 2 , respectively,
p with their multiplicative norm functions x + y D p 7! x 2 ¡ D y 2 =
(x + y D ) (x ¡ y D ) = det(¸ x + y p D ) (where ¸ x + y p D : z 7! (x + y D ) z denotes
p p
multiplication with x + y D in ZZ [ D ] ) , x ;y 2 ZZ . All three algebras are Euclidean
integral domains with respect to the absolute norm function jx 2 ¡ y 2 D j. This one checks
quite easily. In particular, they arep principal idealp domains and factorial (i.e., UFDs) .
Furthermore, for an element x + y D 6=p 0 in Z [ p D ] the absolute p normp jx 2 ¡ D y 2 j
is the index of the principal ideal h x + y D i = Z [ D ] (x + y D ) in Z [ D ] . (More
generally: For an inj ective group homomorphism ' : ZZ m ! Z m given by an m £ m -
matrix a 2 M m (ZZ ) the absolute determinant jdet ' j = jdet a j > 0 is the index of the
image im ' in ZZ m , i. e. , the order of the cokernel coker ' = Z m = im ' :) It follows: The
Diophantine equation jx 2 ¡ D y 2 j = n 2 IN ¤ , also called Pell's equation, has p a solution
(x ;y ) 2 Z 2 if and only if there exists a principal ideal of index
p n in Z
Z [ D ] . Since
any ideal of index n contains n ,p the ideals of index n pin ZZ [ D ] correspond bij ectively
to the ideals of index n in ZZ [ D p] = hn i = (ZZ = Z n ) [ D ] »= (ZZ = ZZ n ) [ X ] = h X 2 ¡ D i .
As a consequence we obtain: If Z [ D ] is a principal ideal domain, then, for a prime
number p , Pell's equation jx 2 ¡ D y 2 j = p is solvable if and only if D is a square in
IF p = ZZ = ZZ p . This proves all the claims made in the text. For D = 2 one has to note
that the solvability of x 2 ¡ 2y 2 = n and of x 2 ¡ 2y 2 = ¡ n are equivalent conditions:
(x ;y ) solves x 2 ¡ 2y 2 2
p = n if and only if (u ;v ) : = (x + 2y ;x + y ) solves u ¡ 2v = ¡ n .
2

(The norm of 1 + 2 is ¡ 1:)

But, even more can be said: One easily calculates the Dedekind's p ³ -function ³ D (n ) : =
³ Z [p D ](n ) , i. e. the number of ideals of index n 2 IN ¤ in Z [ D ] . Since ³ D (p ®1 1 ¢¢¢p ®r r ) =
³ D (p ®1 1 ) ¢¢¢³ D (p ®r r ) for pairwise distinct prime numbers p 1 ;:::;p r , ³ D is completely
determined by the values ³ D (p ® ) = 1 if p = 2 or if p is prime with p jD , ³ D (p ® ) =
® + 1 if p is an odd prime number not dividing D with (D = p ) = 1 and ³ D (p ® ) =
((¡ 1) ® + 1) = 2 if p is an odd prime number not dividing D with (D = p ) = ¡p1. These
formulae are true for a ll square-free integers p D 6́ 1 modp 4 because then ZZ [ D ] is a
so-called Dedekind domain, i.e., ZZ (p ) [ D ] = Z (p ) © Z (p ) D is a principal ideal domain
for every prime number p , where ZZ (p ) µ QI denotes the discrete valuation ring of those
p
rational numbers whose (reduced) denominator is not divisible by p . Z (p ) [ D ] has
exactly one (respectively, two, one) (p rin cipa l) maximal ideal(s) , the indices of these
ideals are p (respectively, p each, p 2 ) if p = 2 or p jD (respectively, (D = p ) = 1 ,
(D = p ) = ¡ 1 ) . This gives the values of the ³ -function ³ D above.
p
Now, assume that even Z [ D ] is a principal ideal domain. Then for n 2 IN ¤ the number
2 2 ¤
of solutions of Pell's equation p jx ¡ D y j = n 2 IN is e D ³ D (n )p, where e D is the order
of the group of units in ZZ [ D ] , because two elements in ZZ [ D ] generate the same
ideal if and only if they di®er multiplicatively by a unit. For D > 0 ( D not a square)

Box 2. continued..

700 RESONANCE  July 2009

 GENERAL  ARTICLE

Box 2. continued..

one always has e D = 1 , i. e.p jx 2 ¡ D y 2 j = np has no solution or in¯nitely many p ones.

m
For example,
p the units in ZZ [ 2 ] are § (1 + 2) , m 2 Z . The units in Z [ ¡ 1 ] are
§ 1;§ ¡ 1 , i. e. e ¡ 1 = 4 , and for D < ¡ 1 the only units are § 1 , i. e. e D = 2 for
D < ¡ 1 . So we get for D = ¡ 1;¡ 2 and n 2 IN ¤ the following very precise results
( vp (n ) denotes the exponent of the highest p -power which divides n , p prime) :
Q
1) The number of solutions (x ;y ) 2 ZZ 2 of x 2 + y 2 = n is 4 ¢ p jn ; p ´ 1 (4 ) (vp (n ) + 1)
if vp (n ) is even for all prime numbers p ´ 3 (4) dividing n , and 0 else (T w o S qu a res
T h eo rem o f E u ler-F erm a t) .
Q
2) The number of solutions (x ;y ) 2 Z 2 of x 2 + 2y 2 = n is 2 ¢ p jn ; p ´ 1 ; 3 (8 ) (vp (n ) + 1)
if vp (n ) is even for all prime numbers p ´ 5; 7 (8) dividing n , and 0 else.
p
Forp D < ¡ 2 the algebra Z [ D ] is never a principal ideal domain because then
ZZ [ D ] contains
p (exactly) one ideal ofp index 2 (namely the ideal generated by 2
and D + D ) , but no element x + y D with norm x 2 + jD jy 2 = 2 . The ¯rst
non-trivial algebra
p of this kind (with D square-free and 6́ 1 mod 4) is the famous
example Z [ ¡ 5 ] of Dedekind. p For D > 0 the situation is di®erent. For exam-
ple, if 0 < D < 100 , then ZZ [ D ] is a principal ideal domain if and only if D 2
f 2;3;6;7;11;14;19;22;23;31;38;43;46;47;59;62;67;71;83;86;94 g. Cf. G Schej a, and
U Storch: L eh rbu ch d er A lgebra , Teil 2, Stuttgart: Teubner 1988, x59, Example 8 pfor a
more thorough but still comparatively elementary investigation of the algebras Z [ D ]
and related rings.

3
F o r a n E n g lish tra n sla tio n o f H ilb ert's G ru n d lagen
see: D H ilb ert, F ou n dation s of G eom etry, O p en C o u rt,
C h ica g o , 1 9 7 1 .
4
A d iscu ssio n o f th e B a n a ch { T a rsk i p a ra d ox ca n b e
fo u n d in L M W a p n er, T he P ea an d the S u n , A M athe-
m atical P aradox, A K P eters, W ellesley, 2 0 0 5 , fo r ex a m -
p le.
5
N ow a d ay s it seem s co m m o n ly a ccep ted th a t th e P y th a -
g o rea n H ip p a so s o f M eta p o n t (a ro u n d 4 5 0 B C ) w a s th e
¯ rst w h o d iscov ered th e p h en o m en o n o f in co m m en su -
ra
p b ility (irra tio n a lity ), n a m ely fo r th e ra tio © = (1 +
5 )= 2 o f th e `g o ld en sectio n ', cf. K v o n F ritz: T h e d is-
cov ery o f in co m m en su ra b ility b y H ip p a su s o f M eta p o n -
tu m , A n n . M ath., V o l.4 6 , p p .2 4 2 { 2 6 4 , 1 9 4 5 . A t th e
tim e o f P la to (4 2 7 { 3 4 7 B C ) it w a s w ell-k n ow n to th e
`ed u ca ted ' p eo p le (a t lea st, P la to h o p ed so ).

RESONANCE  July 2009 701

 GENERAL  ARTICLE

6
T h e rea d er w ill reco g n ize th e ch a ra cteristic (a n d fu n -
d a m en ta l) p ro p erty o f a p rim e n u m b er (h ere o f th e p rim e
n u m b er 2 ): If it d iv id es a p ro d u ct o f p o sitiv e in teg ers it
d iv id es a t lea st o n e o f th e fa cto rs. H en ce, a n a n a lo g o u s
p
a rg u m en t p rov es th e irra tio n a lity o f n p fo r ev ery p rim e
n u m b er p a n d a n y in teg er n ¸ 2 .
7
T h e lib era l a rts co m p rise th e triv iu m : g ra m m a r, rh e-
to ric, lo g ic, a n d th e q u a d riv iu m : a rith m etic, g eo m etry,
a stro n o m y, m u sic.
8
E u d ox o s o f K n id o s characterized a `n u m b er' a b y th e
set A a o f a ll ration al n u m b ers < a . In m o d ern la n -
g u a g e, su ch a cu t is a su b set A µ QI w ith th e fo llow in g
p ro p erties:
(1 ) ; 6= A 6= QI . (2 ) A d o es n o t co n ta in a g rea test
elem en t. (3 ) If x 2 A , th en y 2 A fo r a ll y 2 QI w ith
y < x . D ed ek in d (1 8 3 1 { 1 9 1 6) p ro p o sed to de¯ n e a rea l
n u m b er a s su ch a su b set A of QI . T h en th e set IR o f
a ll th ese su b sets is o rd ered (b y in clu sio n ) a n d h a s th e
ch a ra cteristic p ro p erty o f co m p leten ess, i.e., ev ery n o n -
em p ty a n d b o u n d ed fa m ily A i; i 2 I , o f elem en ts in IR
S
h a s a su p rem u m (n a m ely i2I A i ) . T h e d e¯ n itio n o f
th e a d d itio n in IR is a lso v ery ea sy. F o r th e d e¯ n itio n o f
th e m u ltip lica tio n o n e h a s to b e a little b it m o re ca refu l,
a t lea st, a s fa r a s n o n -p o sitiv e n u m b ers a re in v o lv ed .
9
C a n to r w ro te: \ U n ter ein er `M en g e' v ersteh en w ir
jed e Z u sa m m en fa ssu n g M v o n b estim m ten w o h lu n ter-
sch ied en en O b jek ten m u n srer A n sch a u u n g o d er u n -
seres D en k en s (w elch e d ie `E lem en te' v o n M g en a n n t
w erd en ) zu ein em G a n zen ." C f. G C a n to r, B eitrÄa g e zu r
B eg rÄu n d u n g d er tra n s¯ n iten M en g en leh re (E rster A r-
tik el), M ath. A n n . V o l.4 6 , p p .4 8 1 { 5 1 2 , 1 8 9 5 . U su -
a lly, C a n to r's p a p er \ UÄ b er ein e E ig en sch a ft d es In b e-
g ri® s a ller reellen a lg eb ra isch en Z a h len " in Jou rn al fÄu r
die rein e u n d an gew an dte M athem atik (C relle's J ou r-
n al) V o l.8 4 , p p .2 5 8 { 2 6 2 , 1 8 7 4, is co n sid ered to b e th e
¯ rst p u b lica tio n o n set th eo ry.

702 RESONANCE  July 2009

 GENERAL  ARTICLE

10 Acknowledgements
F o r an y ¯ eld F w ith ch a r F 6= 2 th e stereo g ra p h ic
The author thanks Renuka
p ro jectio n y ield s a b ijectiv e m a p fro m th e p u n ctu red Ravindran and D P Patil, De-
u n it circle S 1 (F ) n f (0 ;1 )g = f (x ;y ) 2 F 2 : x 2 + y 2 = partment of Mathematics, Indian
1 ;y 6= 1 g ov er F o n to th e set f t 2 F : t2 6= ¡ 1 g µ F . Institute of Science, Bangalore
for translating this article, and
11
C f. va n d er B L W a erd en , G eom etry an d A lgebra in the latter for his encouragement
A n cien t C ivilization s, S p rin g er, B erlin , 1 9 8 3 . to publish it here. The author
also thanks Dr H Wiebe, Fakultät
12 für Mathematik, Ruhr Universität
T h e tra n sfo rm a tio n g ro u p w h ich a p p ea rs h ere is th e
Bochum for producing pictures.
sq u a re g ro u p (o r th e d ih ed ra l g ro u p ) D 4 o f o rd er 8 .
13
T h e co n g ru en ce rela tio n s w ere in tro d u ced b y G a u ss
in h is ep o ch -m a k in g trea tise D isqu isition es arithm eticae,
L eip zig 1 8 0 1 ; G erm a n tra n sla tio n b y H M a ser, 1 8 8 9 ,
rep r. C h elsea , N ew Y o rk , 1 9 6 5 ; E n g lish tra n sla tio n b y
A A C la rk e, W C W a terh o u se, C G reith er, a n d A W
G ro o ten d o o rst, S p rin g er, N ew Y o rk , 1 9 8 6 : T w o in teg ers
b;c a re co n g ru en t m o d u lo a n in teg er a : b ´ c m o d a (o r
b ´ c (a ) ), if a n d o n ly if th e d i® eren ce c ¡ b is a m u ltip le
of a .
14
(a = p ) is th e so -ca lled L eg en d re sy m b o l. It is ea sy
to see th a t (a b= p ) = (a = p )(b= p ) fo r in teg ers a ;b w h ich
a re n o t d iv isib le b y p . F u rth erm o re, E u ler's rela tio n
(a = p ) ´ a (p ¡ 1 )= 2 m o d p o r (a = p ) = a (p ¡ 1 )= 2 in th e ¯ n ite
¯ eld IF p = ZZ = ZZ p fo r a ll a 2 IF £p = IF p n f 0 g is a lso ra th er
elem en ta ry : In th e cy clic g ro u p IF £p o f o rd er p ¡ 1 th ere
is a u n iq u e su b g ro u p H o f o rd er u := (p ¡ 1 )= 2 ; in
fa ct H = H (x 2 ) is th e cy clic su b g ro u p g en era ted b y x 2 ,
w h ere x is a g en era to r o f IF £p a n d so elem en ts o f H a re
p recisely th e sq u a res in IF £p . F u rth er, y u = § 1 fo r ev ery
y 2 IF £p , sin ce ¡ 1 is th e o n ly elem en t o f o rd er 2 in IF £p .
T h erefo re (a = p ) = 1 in IF p if a n d o n ly if a 2 H , i.e.
au ´ 1 m od p.
Address for Correspondence
15
W e sh a ll u se th e stereo g ra p h ic p ro jectio n fo r th e ¯ eld Uwe Storch
IF p = ZZ = ZZ p w ith p elem en ts, cf. n o te 1 0 . Fakultät für Mathematik
Ruhr Universität Bochum
D-44780 Bochum, Germany
Email:uwe.storch@rub.de

RESONANCE  July 2009 703