Sie sind auf Seite 1von 8

Evolution of the Topological Concept of "Connected" Author(s): R. L. Wilder Source: The American Mathematical Monthly, Vol. 85, No.

9 (Nov., 1978), pp. 720-726 Published by: Mathematical Association of America Stable URL: http://www.jstor.org/stable/2321676 . Accessed: 12/08/2011 19:24
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.

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to The American Mathematical Monthly.

http://www.jstor.org

EVOLUTION OF THE TOPOLOGICAL

CONCEPI

OF "CONNECTED"

R. L. WILDER In memory andPaul M. Swingle ofEdwin W. Miller

basicconcepts ofone ofthemost Introduction. Thepurpose ofthis paperis to trace theevolution "simply connected"). Likemany other in Topology, viz.,thatof connected (notto be confused with it had onlyan intuitive mathematical concepts of a fundamental nature function), (e.g.,continuous ofAnalysis subtle demands meaning (suchas "connected figure" in geometry) until theincreasingly as one was notachieved, The latter and Topology forced formulation of a satisfactory definition. its within mathematisufficient might until a number of definitions had been proposed-each expect, becamemore general and abstract. cal context butquiteinsufficient as theconfigurations studied of the We tryto clearup, incidentally, theexisting regarding theactualauthorship confusion a "multiple." For several years, European definition ultimately adopted. Not surprisingly, we uncover because ofthedefinition, apparently topologists considered F. Hausdorff to be theprime originator from his notions was usually derived their knowledge of set theory and fundamental topological ofpublication by thetime of classic"Grundziige derMengenlehre" published in 1914[5].However, discovered had his 1944"Mengenlehre" of Hausdorff "Grundziige," which was a third edition the [61, in a noteat and calledattention thereto Lennes's earlier version of thesamedefinition (see below), theendofhisbook. of connected. definition" thedefinition was commonly calledthe"Lennes-Hausdorff Thereafter theterm "Hausdorff-Lennes Separation Manymodem textbooks on Topology seemto haveadopted in thedefinition. involved of separation Condition," or "Hausdorff-Lennes condition" forthetype Mathematical inhisAmerican Possibly this received stimulus from theuse oftheterm byS. Lefschetz speaksof the Society Colloquium volume entitled Algebraic Topology [10].On page 15, Lefschetz "so-called Hausdorff-Lennes separation condition." of thedefinition (p. 127)that In hisclassicwork on Topology [9],Kuratowski states in a footnote A connected "originates from" C. Jordan's CoursdAnalyse of 1893,and also citesLennes'swork. justification forKuratowski's statement is offered below. in the 1952English of his workon general the edition W. Sierpinski, topology [19],attributes Set Theory [13,p. 378],R. L. Moore definition to Hausdorff. However, in hisFoundations ofPoint ofManifolds attributes thedefinition to Lennes. In myownbook,Topology [20],I citedSchoenflies, although independently arrived and Hausdorff, theSchoenflies definition being thesame, Lennes, at, as thedefinition ofJordan which was citedbyKuratowski. further of literature, it seemsfairto concludethatlittle attention was paid to Without citing sinceLennes'definition was UnitedStatesjournalsduring theearlypartof thepresent century,
published in both the Bulletinof the AmericanMathematical Societyand the AmericanJournalof

Mathematics in 1906 and 1911,respectively. Perhaps, too, the same shouldbe said about the cited above(norin anyother, so far as I Hungarian journals, for nowhere in thetopological literature have observed) is thenameof F. Rieszmentioned with thedefinition of connected, in connection thesamedefinition byhimin 1906(in Hungarian) although as that given by Lennes was published and in 1907(in German).'

at Brown, ofTexas.He heldpositions hisPh.D.under Wilder received R. L. Mooreat theUniversity Professor He is now a fora longcareer downat Michigan up to his retirement. Texas,and Ohio Statebefore settling forAdvanced at theInstitute at Santa Barbara. He has heldvisiting Research Associate appointments Study, State. Southern ofTechnology, UCLA, and Florida California, California Institute Colorado, and he is a at theUniversity ofMichigan; He has beena Guggenheim Fellowand theHenry RusselLecturer oftheNational ofSciences. He has served as President ofboththeAMS and theMAAX member Academy His of mathematics. His maininterests are topology, and thecultural foundations history of mathematics, books includeLectures in Topology Volume,Topology of (editedwithW. L. Ayres);an AMS Colloquium to the Introduction Foundations and Evolution ofMathematical Concepts.-Editors Manifolds; ofMathematics; 720

1978]

EVOLUTION OF THE TOPOLOGICAL CONCEPT OF "CONNECTED"

721

The evolution.Unquestionablythe roots of the concept of connected lie in the notion of the continuous, but more specifically the linear continuum, which goes back as far as the Greeks,who struggled to clarifythe notion in the lightof Zeno's paradoxes. The history of this,so far as it is known,is alreadyadequatelycoveredin the literature. hold forthe contributions Similarremarks of the medievalmathematicians whose influence and philosophers, especiallyof the scholastictradition, on both Bolzano and Cantor have been widelycommented upon. Bolzano's contribution. Althoughthe theory of proportion givenby Eudoxus (and reproducedin Euclid's Elements)has been creditedby some as the equivalentof Dedekind's definition of the real it seems not to have figured continuum, in the analysis of the earlypart of the nineteenth century. Duringthelatter period,thegrowing the "locationtheorem"2 stressfora properbasis forestablishing of algebra, using only arithmetic (as opposed to geometric) means, led BernardBolzano to offer a proofof the theorem in 1817 [1]. A casual readingof Bolzano's worksconvincesone thathe had a as it is understoodtoday. remarkableintuitive knowledgeof the structure of the real continuum3 Along withthis,he evidently Considerthe following conceivedof thenotionof a generalcontinuum. definition is present (givenin his Paradoxien[2, p. 129]): "... a continuum when,and only when,we have an aggregateof simple entities(instances or points or substances) so arranged that each individualmemberof the aggregate has, at each individualand sufficiently small distancefrom itself, at least one othermemberof the aggregate When thisdoes not obtain,when so much fora neighbor. as a singlepoint of the aggregate is not so thickly surrounded as to have at least one at by neighbors each individualand sufficiently small distancefromit,thenwe call such a point isolated, and say for thisreason thatour aggregate does not forma continuum." Curiously,the motivationfor this definition, according to Bolzano's own testimony, lay in the paradoxes that plagued the philosophicaland mathematical conceptionsof time,space, and "substance." Bolzano reasoned that,by establishing of the abstractstructural a suitablecharacterization pattern commonto all theseconcepts,theparadoxes could be explained.The analogywiththeGreek dilemmaand the efforts to resolveit is striking. Now the Paradoxienwas written towardthe end of Bolzano's lifeand publishedposthumously in 1851,whiletheproofof the "location theorem," citedabove, was publishedsome 34 yearspreviously. But the motivationfor the latterwas strictly mathematicalin that it was to free analysis of its notoriousreliance on the geometric aspects of continuity. There can be littledoubt, however,that Bolzano's development of his intuition of the continuousin the latterworkwas contributory to his philosophicalconceptionof time,space, and substanceas continua.And it seems to represent thefirst attempt at a mathematical formulation of the topologicalnotionof connected.Since,as was to be the case forover a half-century thereafter, thedefinition of "connected"was tied to thatof "continuum," it would perhapsbe moreproperto termit the mathematical progenitor of the notionof continuum. However,thetimeforconsideration of pointsetshavingno compactness properties had not arrived in mathematics, and thereis littledoubt thatthe intuitive notionwhichBolzano (and after him Cantor) was trying to make precisewas equivalent, in its context, to thatwhichled laterto the "unrestricted" notionof topologicalconnectedness.

Cantor'scontribution. Cantor,who was familiar withBolzano's work,saw clearlythattheproperty used by Bolzano was insufficient to make precisethe intuitive notionof continuum. In a paper often called the Grundlagen [3, ? 10], he pointed out, forexample,thatsets consisting of severalseparated continua satisfyBolzano's condition. Moreover, he recognized intuitively that the compactness propertiesnow associated in topology with the notion of continuumhad not been required in Bolzano's definition, and pointedout thatthe complement of an "isolated" pointset in n-dimensional coordinatespace, En, n > 2, is a continuumaccordingto Bolzano.4 He also rejectedenlisting the concepts of time or space as aids in exploringthe mathematical notionof continuum, deemingthe relationship quite the reverse.

722

R. L. WILDER

[November

to Cantor, According a continuum in En must possess twobasicproperties, theproperty namely, ofbeing perfect and that In modem ifitis closed ofbeing connected. a point terms, setin En is perfect and dense-in-itself (i.e.,eachofitspoints is a limit In this ofit).5 he pointed outthe point connection insufficiency ofrequiring a point setto be onlyperfect in order that his itbe a continuum bygiving classicalexample of a totally disconnected perfect set-the "Cantor ternary set,"nowoften called simply the"Cantor set."(See footnote 11,p. 590,loc.cit.)He then defined connected as follows: A setT is connected point ifforevery twoofitspoints t and t',and arbitrary number given positive c, there always exists ofpoints thedistances afinitenumber tl,t2,... ., tnof T suchthat ttl,tlt2,.. , tnt'are all smaller thane (loc. cit.,575-576). Thenanyperfect and connected of En is a continuum, subset according to Cantor, whopointed outin a footnote (# 12,p. 590,loc.cit.)that no specialdimension was implied in thedefinition; a line,surface, solid,etc.,are all continua. forbounded Incidentally, subsets ofEl, this is equivalent to themodern definition ofa continuum. The mostimportant aspectof Cantor'sdefinition of continuum is his separation of the two concepts perfect and connected, timethe latter as an independent thusidentifying forthe first At thetime, property. however, Topology was virtually nonexistent as a field of study, and it could notbe expected that setshaving thesoleproperty ofconnectedness wouldreceive And anyattention. as alreadyimplied above,Cantor'sdefinition of connected was quiteadequateforthe study of continua. C. Jordan's contribution. The next oftheconcept ofconnected noteworthy is stepin theevolution found in C. Jordan's Cours d'Analyse.6 Apparently Jordan wasnotfamiliar with Cantor's definition of tenyears earlier, sincehe makes no mention ofit.Following a discussion ofclosedsets,7 establishing the notionof distance ("ecart")between them, and defining setsas separated whenthe distance between is greater them than zero, he gives a definition modern ofwhat he calls"unseultenant"-in terms "component"8-of a bounded, closedset,to wit:a boundedand closedset of points has a ifitcannot single component be decomposed intoseveral closedseparated sets."One seeseasily that thedistinctive character ofsucha setis thefollowing: 'For arbitrary between E,one can intercalate, anytwoof itspoints p, p', a chainof intermediate points of thesetsuchthatthedistance between consecutive pointsis less thane."' It is thisstatement thatJordan italicized, not the preceding definition. This coincides, of course, withCantor's and a necessary definition of connected and is proved sufficient condition fora bounded and closedsetto consist ofa single component (loc.cit., p. 26). It is then simple to prove forms a single (loc. cit., p. 27) that a subset ofthereallinewhich component and contains of twonumbers, between a and b, must contain number a and b. Thiscorollary every Bolzano'stheorem ofcomponent. seemsto havebeenthechief motive for Jordan's definition contribution. In 1904, A. Schoenflies thefirst ofhisfundamental researches Schoenflies' published intothetopological ofpoint of aspects settheory He was aware of Cantor's [18]. definition connected, which he cited(loc. cit.pp. 208-209), thateventhough theconcept of but wenton to comment distance formed a primitive for of geometric notion theaxiomatic basis hiswork, itwas preferable to givea purely set-theoretic ofconnected, definition he gives thefollowing: A perfect setis whereupon called connected if it is not decomposable subsets each of which is into[at leasttwononempty]9 perfect. Thisis,for bounded sets, theequivalent ofJordan's definition ofunseultenant which, according to the accompanying remarks, becameknown to Schoenflies onlyafter he had announced his own version. Stating thatJordan introduced the definition onlyto derive Cantor's formulation of the concept (with which he operated thereafter), Schoenflies observes that "connectedness is an important and fundamental property for Analysis Situs as a whole." Thisstatement represents an important step forward in theevolution of theconnectedness concept. Whereas Cantor onlyseparated thenotion from theother properties ofa continuum, Schoenflies nowelevated itto theposition ofa fundamental

1978]

EVOLUTION OF THE TOPOLOGICAL CONCEPT OF "CONNECTED"

723

property ofTopology, and went on to prove itsinvariance and to study theproperty especially in the context ofplanetopology. DespiteSchoenflies' recognition ofthefundamental ofconnectedness, hisviewwas still character limited, in that he expressed theopinion that, while thedefinition was formulated for perfect sets, it couldequally wellbe stated for(merely) closedsets;but"sincefor closedsetswhich arenotperfect, connectedness cannot comeintoquestion, it is sufficient to limit thedefinition sets"(loc. to perfect p. 173)!Thus, while making an important cit., stepforward, Schoenflies madeanother stepbackward. Youngvirtually coincides with that ofLennes and Riesztobe discussed itis interpolated below, here as a kindofcapstone to thework already described, as wellas ofintrinsic interest for itsadumbration oflater work in thetheory of connectedness. The classic book[22]ofW. H. and G. C. Youngl'introduces a definition ofconnected in terms of regions:""A setofpoints suchthat, describing a region in anymanner round and each eachpoint limiting point ofthesetas internal point, these regions always is saidtobe a generate a single region, connected setprovided it contains morethanone point. Henceif a setis connected theset gotby closing it is connected, and viceversa."12 Fromthisdefinition, theYoungsprove:A connected set cannot intoclosedcompobe divided nents(= subsets) without common points. Conversely a set which intoclosed cannotbe divided components without common points is,ifclosed, a connected set.We recall that this proposition was usedbyJordan and others to define connected in thecase where thesetin question is closed. The Lennes and Riesz definitions. It is remarkable that throughout the period discussed above-from thetime ofBolzanoto 1905, over haLf a century-the notion ofconnected wasconfined to closedsets; and thisin spiteof the factthatCantordivorced closurein his thenotionfrom definition of continuum. On the otherhand,it is not surprising, since attention was devoted to therealcontinuum exclusively either or to thesubsets of euclidean theplane), space(usually and theonly non-closed setsofimportance were ofa special suchas thesetofrationals oropen character, segments on therealline,and thecircular or triangular oftheplane. regions Of course, Jordan, and following himSchoenflies, had proposed of connected definitions which virtually begged forgeneralization to non-closed sets.And thisstepwas finally in 1905-06 taken by both N. J. Lennes and F. Riesz. Lennes gave his definition at a meeting of the American Mathematical in December Society 1905, anditwaspublished in theabstract ofhistalkthefollowing yearin theBulletin of thatSociety was presented to theHungarian [11].Riesz'sdefinition Academy ofSciences on January 22, 1906, andpublished later thesameyear [15].Herewasclearly a "multiple" -a case ofindependent invention bymorethanone investigator.
The workof W. H. and G. C. Young. Althoughchronologically the work of W. H. and G. C.

that it is meaningful in anyspacein which limit point is defined (although undoubtedly theauthor's thinking, likethat ofmosttopologists of thetime, was ofeuclidean spaces). Riesz'sdefinition has several remarkable features. In thefirst place,itis given in thecontext ofan essay [16] devotedto the relations between the "physical continuum" and the "mathematical continuum."'13 In defining thephysical continuum, Rieszusestherelation "unterscheidbar" between space points, not the topological notionof limit point.The definition proceeds as follows: "Das physikalische Kontinuum heisst zusamenhangend, wennes nicht in zweiTeilmengen zerlegt werden kann, dassjedes Element dereinenTeilmenge unterscheidbar sei vonjedemElemente deranderen Teilmenge." Noticethe striking resemblance to theJordan-Schoenflies definition, although Riesz makes no reference to thelatter. However, there is conclusive evidence in previous papers of Riesz' thathe was familiar with Jordan's Cours d'Analyse and hence, probably, with Jordan's definition.'4 In thesecondplace,thedefinition of connected fortopological spaces(he does notuse thelatter

Lennes' definition reads (loc. cit.): A set ofpointsis connected if in every pair of complementary subsets at leastone subsetcontains a limit pointofpointsin theother set. This is statedin such a fashion

724

R. L. WILDER

[November

forwhathe callsa "mathematical term) is given initially in modem is an continuum," which, terms, abstract ofsuchlater as were topological spacedefined axioms adumbrative byfour systems given by Hausdorff Kontinuum and Kuratowski. The definition "Das mathematische readsas follows: heisse die Komplemenzusammenhangende wennes nicht in zweioffene werden Teilmengen zerlegt kann, fureinander tarmengen two degrees sein."For subsets of such a space he thendistinguishes of connectedness: A setis calledconnected whoseclosures ifit cannot intotwosubsets be decomposed are disjoint; if forevery of it intotwo[nonempty] it is calledabsolutely connected decomposition oftheother. subsets, there exists at leastone element to one subset andis a limit which belongs point It is the secondof these, definition of course-i.e.,absolutely is themodem connected-that of connected. If Riesz had been familiar withthe modemdeviceof relativizing the topological notions of fora "closed"and "open,"he would, of connectedness presumably, thedefinitions haveidentified mathematical continuum and that of absolute connectedness. wasjoinedbya third, Actually, themultiple twodefinitions were which occurred when these given in hisbook of 1914(loc. cit.), viz.,F. Hausdorff's definition. Apparently whengiving his definition Hausdorff On theother did was unaware of either Lennes'or Riesz'definitions. hand,Hausdorff proceed, in thisbook,to study setsas topological in someof theproperties of connected concepts their ownright. until1921;we setswas notpublished thefirst to thestudy ofconnected However, paperdevoted of B. Knaster refer hereto theclassicpaperSurles ensembles and C. Kuratowski connexes [8].This in theevolution ofconnectedness because:(1) itestablished oftheconcept paperwas significant the factthatconnected have a variety of interesting spaceslacking compactness properties topological of foundations properties; (2) itgaveimpetus andin thelogical to a hostofstudies, bothin Topology settheory; the (3) it gavetheultimate to Schoenflies' emphasis statement, quotedabove,concerning fundamental character ofconnectedness."5 Froman evolutionary Concluding remarks. of the concept of pointof view,the development connectedness provesto be a revealing "case study." Its roots,as in the case of manyother mathematical concepts, areembedded in thecontemplation ofphysical time, space,and "substance." AtthehandsofCantor itfinally split off from philosophical and physical considerations to become a partof mathematical theory. But it was not easilydivorced from theconcept of continuum within whichit was first formulated-a consequence of its mathematical environment, whichconsisted chiefly of the studyof curvesand surfaces, examples of what Cantorcalled continuain the mathematical sense. This was a case of the operation of "environmental stress," in that the mathematical environment worked to confine thenotion within a restricted area. It failed to find itsproper placein mathematical theory until Schoenflies pointed outitsinvariance undertopological transformations, as wellas its independent status as a topological property. But although Schoenflies, who discovered essentially thesamedefinition thatJordan had given overa decadeearlier, madean important stepforward, Topology had still notgrown much beyond thestudy of configurations whosecompactness properties made theJordan-Schoenflies definition quiteadequate.Indeed, so much so, that when theYoungswrote their classic"The Theory ofSetsof Points" the decade between during Jordan and Schoenflies, theyseemto have deliberately phrased their definition ofregion (allowing a region toinclude boundary points freely) so that theCantor definition wouldbe preserved (see theremark aboveconcerning theYoungdefinition). Lennes'sgeneralization of thesedefinitions was apparently a resultof his consideration of non-closed sets.In thepaper[12]giving in detail theresults announced in the1906abstract (loc.cit.), he first defined connectedness for opensets(in euclidean space)byusing broken lines, an opensetU ifevery being connected pairofpoints a, b in U arejoinedbya broken linelying wholly in U (loc.

1978]

EVOLUTION OF THE TOPOLOGICAL CONCEPT OF "CONNECTED"

725

cit., p. 293). He observed thatby theCantor definition of connected, theunionof theinterior and exterior of a planar circle forms a connected theordinary set;moreover, that"iffrom continuum in spaceofanydimensions anysetwhatever which is nowhere denseis removed, theresidue would form a connected set"(loc. cit.,303,footnote). He thereupon gavetheform nowgenerally ofthedefinition accepted, remarking that it "applies in caseswhere theformer it renders doesnot."(In other words, setsconnected which our intuition tellsus shouldbe connected likethe and rulesout thosethat, complement of thecircle in theplane, should notbe termed connected.) One oftheremarkable features ofRiesz'definition, as we havealready noticed, is that itwasgiven in thecontext of an abstract topological space.Thisaspectof Riesz'work seemsalso to havebeen generally unnoticed forsometime, despite thefactthat calledattention Frechet [4,NoteB] to Riesz' abstract spaceaxioms as they werelater presented at theInternational Congress in Rome,1908[17]. In any event, his definition, although agreeing withLennes',achievesthereby its mostgeneral character, freed from all metric considerations. Although thelackofdiffusion from one country to another, which characterized earlier periods in mathematics, hadbegun to subside, theperiod during which thetopological concept ofconnectedness was developed still shows considerable lackofdiffusion. Lennes'and Riesz'work, bothpublished in reputable journals during thefirst decade of thecentury, was generally unknown until thejournal Fundamenta Mathematicae commenced publication in 1920.The occurrence of a three-member multiple during thefirst quarter ofthepresent century is quitenoteworthy. One further comment: One of thenoteworthy features of theKnaster and Kuratowski article referred to abovewas itspresentation ofparadoxical examples ofconnected setshaving no compactnessproperties. I have pointed out elsewhere [21] thecontribution thatparadoxcan maketo the development of mathematical concepts. The examples givenby Knaster and Kuratowski (loc. cit.) proved a great stimulation to thestudy ofconnectedness. Thereensued a sizableliterature devoted to theconcept, and in recent years this has engendered interesting questions in theFoundations of Set Theory.

Notes 1. My attention was first calledto Riesz'work in thisconnection by Professor C. E. Aull,to whom I am indebted forreferences thereto. 2. Thatis,ifa realpolynomial f(x) is negative for x = a and positive for x = b, then it is zeroat somevalue between a and b. (Bolzanostated, thatiff(x) and p(x) are continuous namely, realfunctions overan interval a < x < g, and f(a) <p(a), f(b) >T(b), then there exists a real number c such a <c Kb and f(c)= p(c).) 3. The notion of a countable densesubset (e.g.,therationals) escapedhim, to be sure.ButBolzanodid not encounter thesamesort ofproblem in realanalysis that forced Cantor's formulation ofthenotion. 4. It is curious thatCantor did notpointout thatthesetof rational points in El, therealline,is also a continuum byBolzano's definition. 5. Cantor's precise definition utilized thenotion ofderivative ofa point ifit set,a pointsetP being perfect coincides with itsfirst derivative P'. 6. We refer hereto thesecondedition, 1893, vol. 1,pp. 24-28[7].Thisis theedition in which theclassical concept of continuous curve was first defined; thelatter was to receive muchattention whenits "space-filling" character was discovered. 7. Jordan used theword"perfect" ("parfait") instead of "closed."We use thelatter term hereto avoid confusion with Cantor's use oftheterm "perfect." Modemterminology conforms to thelatter. 8. Noticethat Jordan did notuse theterm "connected." The assumption of "bounded" wasnecessitated by thedefining of "separated" in terms ofdistance. 9. Thebracketed condition is clearly implied in Schoenflies's statement, although notexplicitly stated byhim. 10. Thisbookwasintended as the"first attempt at a systematic exposition" ofCantor's ideason thetheory of sets.See Preface, loc.cit.

726

R. L. WILDER

[November

11. In modem terms, a region wasa connected opensetwith orwithout an arbitrary setofitsboundary points. TheYoungs defined itas generated bysuccessively overlapping triangles. Thedefinition is on p. 204, ofconnected 12. Ifthewords "andeachlimiting areomitted point" from this definition and "region" restricted to interiors oftheregions as defined then theabovedefinition bytheYoungs, to the"simple chain definition" of i$equivalent connected. See R. L. Wilder, loc. cit.,p. 34,Corollary 12.5. 13. "Ich suchenurden Weg,dervon den raumlichen Vorstellung zu dem[mathematischen] Raumbegriffe fiihrt" (loc. cit.). 14. See, forinstance, in thecollected works [14],paperA2 (1905),in which on he mentions Jordan's "ecart" thefirst page; paperA3 (1905),references to theCours sentence of d'Analyse; and paperA5 (1905),in thefirst which he defines "d'unseultenant," thesameterm that was usedbyJordan. 15. Of course, Schoenflies was notstrictly speaking of thetype of connectedness exploited by Knaster and Kuratowski, which was theLennes-Riesz definition nowgenerally accepted. References die ein entgegengedass zwischen je zweiWerten, Beweisdes Lehrsatzes, 1. B. Bolzano,Reinanalytischer Prag,1817. liege, ed. Fr.Prihonsky, einereele Wurzel derGleichung setzes Resultat gewihren, wenigstens and Kegan Paul, Routledge , Paradoxes tr.of theParadoxien by F. Prihonsky, 2. of theInfinite, London,1950. 5. Fortsetzung, Math.Ann.,21 (1883) linearePunktmannigfaltigkeiten, 3. G. Cantor, Ueber unendliche 545-591. Les espaces Paris, 1928. abstraits, Gauthier-Villars, 4. M. Fr&het, 1914. Leipzig, Grundzige derMengenlehre, vonWeit, 5. F. Hausdorff, , Mengenlehre, NewYork,1944. 6. Dover, Coursd'Analyse, 2nded., 1893, vol. 1. 7. C. Jordan, 2 (1921)206-255. Fund.Math., and C. Kuratowski, Surles ensembles connexes, 8. B. Knaster New York,1968. 9. C. Kuratowski, vol.2, Academic Press, Topology, NewYork,1942. Society, Topology, American Mathematical 10. S. Lefschetz, Algebraic Bull.Amer. Math.Soc., 12 (1905-06)284,abstract in non-metrical analysis situs, 11. N. J. Lennes, Curves ?10. Amer. J. , Curves in thecalculus of variations, applications analysis situswith 12. in non-metrical Math., 33 (1911)287-326. Providence, R.I., 1962. Mathematical Society, of PointSet Theory, American 13. R. L. Moore,Foundations 1960, vol. 1. Gauthier-Villars, Paris, 14. F. Riesz,Oeuvres Completes, (paper A6 in , A terfolgalom genesise, Math.u. Phys. Lapok,15(1906)97-122;16(1907)145-161 15. [14]). 24 (1907) 309-353 , Die Genesis Berichte aus Ungarn, des Raumbegriffes, Math.u. Naturwiss. 16. (paperA7 in [14]). , Stetigkeitsbegriff dei Matematici, Internazional Atti del IV Congresso undabstrakt Mengenlehre, 17. Romavol.2, p. 18. I, Math.Ann., 58 (1904)195-238. zurTheorie derPunktmengen, 18. A. Schoenflies, Beitrage Toronto, 1952. University ofToronto Press, Topology, trans. byC. C. Krieger, 19. W. SierpiAski, General R.I., 1949. American Society, Providence, of Manifolds, Mathematical 20. R. L. Wilder, Topology 1 (1974)29-46. , Hereditary Historia Math., stress as a cultural force in mathematics, 21. 1906. Cambridge, Cambridge Univ.Press, The theory of setsof points, 22. W. H. Youngand G. C. Young, CA 93106. DEPARTMENTOF MATHEMATICS,UNIVERSITY OF CALIFORNIA,SANTA BARBARA,
loc. cit.

Das könnte Ihnen auch gefallen