The Truthabout K6nigsberg

BrianHopkinsand RobinJ. Wilson

BrianHopkins (bhopkins@spc.edu) is an assistant

professorat St. Peter'sCollege,a Jesuit liberalarts college
in Jersey City,NewJersey.He receivedhis Ph.D.fromthe
Universityof Washingtonforworkon algebraic
combinatoricspertinentto the representationtheoryof Lie
algebras.Otherprofessionalinterestsincludegraphtheory
and matheducation.He also enjoyschoralsinging,poetry,
and New YorkCity.

Robin Wilson (r.j.wilson@open.ac.uk) is Headof the Pure

MathematicsDepartmentat the Open University, and a
Fellowof KebleCollege,OxfordUniversity. He has written
and editedabouttwo dozen bookson subjectsranging
fromgraphtheoryto the historyof mathematics,and has
recentlywrittena bookon the historyand proofof the
four-colorproblem.He is very interestedin music,and he
has justco-editeda bookon musicand mathematics.

Euler's 1736 paper on the bridges of Kdnigsberg is widely regarded as the earliest
contribution to graph theory-yet Euler's solution made no mention of graphs. In this
paper we place Euler's views on the Kinigsberg bridges problem in their historical
context, present his method of solution, and trace the development of the present-day
solution.

What Euler didn't do

A well-knownrecreationalpuzzle concernsthe bridgesof Kinigsberg.It is claimed
that in the early eighteenthcenturythe citizens of Kinigsberg used to spend their
Sundayafternoonswalking aroundtheir beautifulcity. The city itself consistedof
fourlandareasseparatedby branchesof the riverPregelover whichtherewere seven
bridges,as illustratedin Figure1. The problemthatthe citizensset themselveswas to

Figure 1. K6nigsberg

198 @ THEMATHEMATICAL OF AMERICA

ASSOCIATION
walk around the city, crossing each of the seven bridges exactly once and, if possible,
returningto their startingpoint.
If you look in some books on recreational mathematics, or listen to some graph-
theorists who should know better, you will 'learn' that Leonhard Euler investigated
the Kinigsberg bridges problem by drawing a graph of the city, as in Figure 2, with
a vertex representing each of the four land areas and an edge representing each of the
seven bridges. The problem is then to find a trail in this graph that passes along each
edge just once.

c cd d g
A e D

a b

Figure 2. The K6nigsberggraph

But Euler didn't draw the graph in Figure 2-graphs of this kind didn't make their
first appearance until the second half of the nineteenth century. So what exactly did
Euler do?

Ko N INGSBERGA

Figure 3. Seventeenth-century
Konigsberg

VOL.35, NO.3, MAY2004 THECOLLEGEMATHEMATICS

JOURNAL 199
The Konigsberg bridges problem
In 1254 the Teutonic knights founded the Prussian city of K6nigsberg (literally, king's
mountain). With its strategic position on the river Pregel, it became a trading center
and an important medieval city. The river flowed around the island of Kneiphof (lit-
erally, pub yard) and divided the city into four regions connected by seven bridges:
Blacksmith's bridge, Connecting bridge, High bridge, Green bridge, Honey bridge,
Merchant's bridge, and Wooden bridge: Figure 3 shows a seventeenth-centurymap of
the city. K6nigsberg later became the capital of East Prussia and more recently became
the Russian city of Kaliningrad, while the river Pregel was renamed Pregolya.
In 1727 Leonhard Euler began working at the Academy of Sciences in St Peters-
burg. He presented a paper to his colleagues on 26 August 1735 on the solution of 'a
problem relating to the geometry of position': this was the Kinigsberg bridges prob-
lem. He also addressed the generalized problem: given any division of a river into
branches and any arrangementof bridges, is there a general method for determining
whether such a route exists?
In 1736 Euler wrote up his solution in his celebrated paper in the Commentarii
Academiae Scientiarum Imperialis Petropolitanae under the title 'Solutio problema-
tis ad geometriam situs pertinentis' [2]; Euler's diagram of the K6nigsberg bridges
appears in Figure 4. Although dated 1736, Euler's paper was not actually published
until 1741, and was later reprintedin the new edition of the Commentarii (Novi Acta
Commentarii ...) which appearedin 1752.

B -k

Figure 4. DiagramfromEuler's1736 paper

A full English translation of this paper appears in several places-for example, in

[1] and [6]. The paper begins:

1. In additionto thatbranchof geometrywhichis concernedwith distances,andwhich

has alwaysreceivedthe greatestattention,thereis anotherbranch,hithertoalmostun-
known,whichLeibnizfirstmentioned,callingit the geometryof position[Geometriam
situs]. This branchis concernedonly with the determinationof position and its prop-
erties; it does not involve distances,nor calculationsmade with them. It has not yet
been satisfactorilydeterminedwhat kinds of problemare relevantto this geometryof
position,or whatmethodsshouldbe used in solvingthem.Hence,when a problemwas
recentlymentionedwhich seemedgeometricalbut was so constructedthatit did not re-
quirethe measurementof distances,nordid calculationhelp at all, I hadno doubtthatit

200 P THEMATHEMATICAL OF AMERICA

ASSOCIATION
 was concernedwith the geometryof position-especially as its solutioninvolvedonly
position, and no calculationwas of any use. I have thereforedecided to give here the
methodwhich I have foundfor solving this problem,as an exampleof the geometryof
position.
2. The problem,whichI am told is widely known,is as follows: in K6nigsberg...

This reference to Leibniz and the geometry of position dates back to 8 September
1679, when the mathematician and philosopher Gottfried Wilhelm Leibniz wrote to
ChristiaanHuygens as follows [5]:

I am not contentwith algebra,in thatit yields neitherthe shortestproofs nor the most
beautifulconstructionsof geometry.Consequently,in view of this, I considerthat we
needyet anotherkindof analysis,geometricor linear,whichdealsdirectlywithposition,
as algebra deals with magnitudes ...

Leibniz introduced the term analysis situs (or geometria situs), meaning the analysis
of situation or position, to introduce this new area of study. Although it is sometimes
claimed that Leibniz had vector analysis in mind when he coined this phrase (see, for
example, [8] and [11]), it was widely interpretedby his eighteenth-century followers as
referring to topics that we now consider 'topological'-that is, geometrical in nature,
but with no reference to metrical ideas such as distance, length or angle.

Euler's Konigsberg letters

It is not known how Euler became aware of the K6nigsberg bridges problem. However,
as we shall see, three letters from the Archive Collection of the Academy of Sciences
in St Petersburg [3] shed some light on his interest in the problem (see also [10]).
Carl Leonhard Gottlieb Ehler was the mayor of Danzig in Prussia (now Gdansk in
Poland), some 80 miles west of Kinigsberg. He corresponded with Euler from 1735 to
1742, acting as intermediary for Heinrich Kiihn, a local mathematics professor. Their
initial communication has not been recovered, but a letter of 9 March 1736 indicates
they had discussed the problem and its relation to the 'calculus of position':

Youwouldrenderto me andour friendKiihna most valuableservice,puttingus greatly

in yourdebt,most learnedSir,if you would send us the solution,which you know well,
to the problemof the seven Kinigsberg bridges,togetherwith a proof. It would prove
to be an outstandingexampleof the calculusof position [CalculiSitus],worthyof your
greatgenius.I have addeda sketchof the said bridges...

Euler replied to Ehler on 3 April 1736, outlining more clearly his own attitude to
the problem and its solution:

... Thus you see, most noble Sir, how this type of solutionbearslittle relationshipto
mathematics,and I do not understandwhy you expect a mathematicianto produceit,
ratherthananyoneelse, for the solutionis basedon reasonalone, andits discoverydoes
not depend on any mathematicalprinciple.Because of this, I do not know why even
questionswhich bear so little relationshipto mathematicsare solved more quickly by
mathematiciansthanby others.In the meantime,most noble Sir,you have assignedthis
questionto the geometryof position, but I am ignorantas to what this new discipline
involves,andas to whichtypes of problemLeibnizandWolffexpectedto see expressed
in this way ...

VOL.35, NO.3, MAY2004 THECOLLEGEMATHEMATICS

JOURNAL 201
 4
"

"• J_
.

I:11.

~ctl-z"

Figure 5. Ehler'sletterto Euler

Around the same time, on 13 March 1736, Euler wrote to Giovanni Marinoni, an Italian
mathematician and engineer who lived in Vienna and was Court Astronomer in the
court of Kaiser Leopold I. He introduced the problem as follows (see Figure 6):

A problemwas posed to me aboutan island in the city of Ktbnigsberg, surroundedby

a riverspannedby seven bridges,and I was askedwhethersomeonecould traversethe
separatebridgesin a connectedwalk in such a way that each bridge is crossed only
once. I was informedthathithertono-onehaddemonstrated the possibilityof doingthis,
or shownthat it is impossible.This questionis so banal,but seemed to me worthyof
attentionin that geometry,nor algebra,nor even the art of countingwas sufficientto
solve it. In view of this, it occurredto me to wonderwhetherit belongedto the geometry
of position [geometriamSitus], which Leibnizhad once so much longed for. And so,
aftersome deliberation,I obtaineda simple,yet completelyestablished,rulewithwhose
help one can immediatelydecide for all examples of this kind, with any numberof
bridgesin any arrangement, whethersuch a roundtripis possible,or not ...

202 : THEMATHEMATICAL
ASSOCIATION
OF AMERICA
 7~CA a
AftovO
"4'' .0/t~ " " G?
" CIjr
'
, .-mw"-'-"-44
1* wgo4D ?(4Z*
t."a-6i
r n
e
4 I
..-
)tc Mc 'Pee
,,F( VRls or14
,O..jR
Op . 00,; . .04
, * ..l " , .
•" " " ' " 0 • f"V. •
.9 . ."
Si • ,,en• I./u 3#4~4
eaA'ftftC_ t•I 9r 'av---- .'..•,n*,
'
4_•_
, o,4,)fo.4,,~

S, •,,& pw-.,
_._i,,
AN AV -

. ..
- , - [ - !•1
? .- . . !1,
v er
Ar,," A,,'
oflI JM0t-PMWCsw j
-

* >.3iuw
if pow
?'
AA/
Aor

d- v 7,
('V ?
,.. 4,.."-

Figure6. Euler'sletterto Marinoni

Euler's 1736 paper

Euler'spaperis dividedinto twenty-onenumberedparagraphs,of which the firstas-
cribes the problemto the geometryof position as we saw above, the next eight are
devotedto the solutionof the Kinigsbergbridgesproblemitself, and the remainder
areconcernedwith the generalproblem.Morespecifically,paragraphs 2-21 dealwith
the followingtopics (see also [12]):

Paragraph 2. Eulerdescribedthe problemof the K6nigsbergbridgesand its gen-

anddivisionof the riverinto branches,and
eralization:'whateverbe the arrangement

VOL.35, NO.3, MAY2004 THECOLLEGEMATHEMATICS

JOURNAL 203
howevermanybridgestherebe, can one findout whetheror not it is possibleto cross
eachbridgeexactlyonce?'

Paragraph 3. In principle,the originalproblemcould be solved exhaustivelyby

checkingall possiblepaths,butEulerdismissedthis as 'laborious'andimpossiblefor
configurationswith morebridges.

Paragraphs 4-7. Thefirstsimplificationis to recordpathsby thelandregionsrather

thanbridges.Using the notationin Figure4, going south from Kneiphofwould be
notatedAB whetherone used the GreenBridgeor the Blacksmith'sBridge.The final
pathnotationwill needto includean adjacentA andB twice;theparticularassignment
of bridgesa andb is irrelevant.A pathsignifiedby n letterscorrespondsto crossing
n - 1 bridges,so a solutionto the Kinigsbergproblemrequiresan eight-letterpath
with two adjacentA/B pairs,two adjacentA/C pairs,one adjacentA/D pair,etc.

Paragraph 8. Whatis the relationbetweenthe numberof bridgesconnectinga land

massandthe numberof timesthe corresponding letteroccursin thepath?Eulerdevel-
opedthe answerfroma simplerexample(see Figure7). If thereis an odd numberk of
bridges,thenthe lettermustappear(k + 1)/2 times.

Figure7. A simplecase

Paragraph 9. Thisis enoughto establishthe impossibilityof the desiredKbinigsberg

tour.Since Kneiphofis connectedby five bridges,the path must containthree As.
Similarly,theremustbe two Bs, two Cs, andtwo Ds. Inparagraph14, Eulerrecords
these datain a table.

region A B C D
bridges 5 3 3 3
frequency 3 2 2 2

Summingthe finalrow gives nine requiredletters,but a pathusing each of the seven

bridgesexactlyonce canhaveonly eightletters.Thustherecanbe no Kijnigsbergtour.

Paragraphs 10-12. Euler continued his analysis from paragraph 8: if there is an

evennumberk of bridgesconnectinga landmass,thenthecorresponding letterappears
k/2 + 1 timesif the pathbeginsin thatregion,andk/2 times otherwise.

Paragraphs 13-15. The general problem can now be addressed. To illustrate the
method Euler constructed an example with two islands, four rivers, and fifteen bridges
(see Figure8).

204 @ THEMATHEMATICAL OF AMERICA

ASSOCIATION
 Figure 8. A morecomplicatedexample

This system has the following table, where an asterisk indicates a region with an
even number of bridges.

region A* B* C* D E F*
bridges 8 4 4 3 5 6
frequency 4 2 2 2 3 3

The frequencies of the letters in a successful path are determined by the rules for even
and odd numbers of bridges, developed above. Since there can be only one initial
region, he records k/2 for the asterisked regions. If the frequency sum is one less
than the required number of letters, there is a path using each bridge exactly once that
begins in an asterisked region. If the frequency sum equals the required number of
letters, there is a path that begins in an unasterisked region. This latter possibility is
the case here: the frequency sum is 16, exactly the number of letters requiredfor a path
using 15 bridges. Euler exhibited a particularpath, including the bridges:

EaFbBcFdAeFfCgAhCiDkAmEnApBoElD.

Paragraph 16-19. Euler continued with a simpler technique, observing that:

... the numberof bridgeswrittennext to the lettersA, B, C, etc. togetheradd up to

twice the total numberof bridges.The reasonfor this is that,in the calculationwhere
every bridgeleadingto a given areais counted,each bridgeis countedtwice, once for
each of the two areaswhichit joins.

This is the earliest version known of what is now called the handshaking lemma. It
follows that in the bridge sum, there must be an even number of odd summands.

Paragraph 20. Euler stated his main conclusions:

If thereare morethantwo areasto which an odd numberof bridgeslead, then such a

journeyis impossible.

VOL.35, NO.3, MAY2004 THECOLLEGEMATHEMATICS

JOURNAL 205
 If, however,the numberof bridges is odd for exactly two areas,then the journey is
possible if it startsin eitherof these two areas.
If, finally,thereare no areasto which an odd numberof bridgeslead, thenthe required
journeycan be accomplishedstartingfromany area.

Paragraph 21. Eulerconcludedby saying:

When it has been determinedthatsuch a journeycan be made,one still has to findhow
it shouldbe arranged.For this I use the following rule:let those pairsof bridgeswhich
lead from one areato anotherbe mentallyremoved,therebyconsiderablyreducingthe
numberof bridges;it is then an easy task to constructthe requiredroute across the
remainingbridges,andthe bridgeswhichhavebeen removedwill not significantlyalter
the route found, as will become clear after a little thought.I do not thereforethinkit
worthwhileto give any furtherdetailsconcerningthe findingof the routes.

Note that this final paragraphdoes not prove the existence of a journey when one is
possible, apparently because Euler did not consider it necessary. So Euler provided a
rigorous proof only for the first of the three conclusions. The first satisfactory proof
of the other two results did not appear until 1871, in a posthumous paper by Carl
Hierholzer (see [1] and [4]).

The modern solution

The approach mentioned in the first section developed through diagram-tracingpuz-
zles discussed by Louis Poinsot [7] and others in the early-nineteenth century. The
object is to determine whether a figure can be drawn with a single stroke of the pen in
such a way that no edge is repeated. Considering the figure to be drawn as a graph, the
general conditions in Paragraph 20 take the following form:

If therearemorethantwo verticesof odd degree,thensuch a drawingis impossible.

If, however,exactlytwo verticeshaveodd degree,thenthe drawingis possibleif it starts
with eitherof these two vertices.
If, finally,thereare no verticesof odd degree,thenthe requireddrawingcan be accom-
plishedstartingfromany vertex.

So the 4-vertex graph shown in Figure 2, with one vertex of degree 5 and three
vertices of degree 3, cannot be drawn with a single stroke of the pen so that no edge
is repeated. In contemporary terminology, we say that this graph is not Eulerian. The
arrangement of bridges in Figure 8 can be similarly represented by the graph in Fig-
ure 9, with six vertices and fifteen edges. Exactly two vertices (E and D) have odd
degree, so there is a drawing that starts at E and ends at D, as we saw above. This is
sometimes called an Eulerian trail.
However, it was some time until the connection was made between Euler's work and
diagram-tracingpuzzles. The 'K•nigsberg graph' of Figure 2 made its first appearance
in W. W. Rouse Ball's Mathematical Recreations and Problems of Past and Present
Times [9] in 1892.

Background information,including English translationsof the papers of Euler [2] and Hierholzer [4], can be
found in [1]; an English translationof Euler'spaperalso appearsin [6].

206 @ THEMATHEMATICAL OF AMERICA

ASSOCIATION
 D1 E

C B
A
" d c

e b

Figure 9. The graphof the bridgesin Figure8

References
1. N. L. Biggs, E. K. Lloyd and R. J. Wilson, GraphTheory1736-1936, reissuedpaperbackedition, Clarendon
Press, Oxford, 1998.
2. L. Euler,Solutio problematisad geometriamsitus pertinentis,CommentariiAcademiaeScientiarumImperi-
alis Petropolitanae8 (1736) 128-140 = Opera Omnia(1) 7 (1911-56), 1-10.
3. - , Pis'ma k ucenym,Izd. AkademiiNauk SSSR, Moscow-Leningrad(1963), Fl, op.3, d.21, L.35-37
ob; El, op.3, d.22, L.33-41; F.136, op.2, d.3, L.74-75.
4. C. Hierholzer,Ueber die Mbglichkeit, einen Linienzug ohne Wiederholungund ohne Unterbrechnungzu
umfahren,Math.Ann. 6 (1873) 30-32.
5. G. W. Leibniz, MathematischeSchriften(1), Vol. 2, Berlin (1850), 18-19.
6. J. R. Newman (ed.), The Worldof Mathematics,Vol. 1, Simon and Schuster,New York, 1956.
7. L. Poinsot, Sur les polygones et les polyedres,J. Ecole Polytech.4 (cah. 10) (1810) 16-48.
8. J.-C. Pont, La TopologieAlgebriquedes Originesai Poincard,Bibl. De Philos. Contemp.,Presses Universi-
taires de France,Paris, 1974.
9. W. W. Rouse Ball, MathematicalRecreationsand Problemsof Past and Present Times,1st ed. (laterentitled
MathematicalRecreationsand Essays), Macmillan,London, 1892.
10. H. Sachs, M. Stiebitz and R. J. Wilson, An historicalnote: Euler's Kinigsberg letters, J. Graph Theory12
(1988) 133-139.
11. R. J. Wilson, Analysis situs, GraphTheoryand its Applicationsto Algorithmsand ComputerScience (ed. Y
Alavi et al.), John Wiley & Sons, New York(1985), 789-800.
12. - , An EuleriantrailthroughK6nigsberg,J. GraphTheory10 (1986) 265-275.

If I feel unhappy,I do mathematics to become happy. If I am happy, I do mathe-

matics to keep happy. Alfr6d Renyi
(Quoted in P. Turin, "The Work of Alfr6d R6nyi,"Matematikai Lapok 21 (1970)
199-210)

VOL.35, NO.3, MAY2004 THECOLLEGEMATHEMATICS

JOURNAL 207