The Absolute Arithmetic and Geometric Continua Author(s): Philip Ehrlich Source: PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association, Vol. 1986, Volume Two: Symposia and Invited Papers, (1986), pp. 237-246 Published by: The University of Chicago Press on behalf of the Philosophy of Science Association Stable URL: http://www.jstor.org/stable/192803 Accessed: 17/07/2008 14:07 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ucpress. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org. http://www.jstor.org The Absolute Arithmetic and Geometric Continual PhilipEhrlich Brown University 1. Introduction With the appearanceof J.H. Conway'sOn Numbersand Games (1976), the mathematical andphilosophicalcommunitieshave much to celebrate. It is Conway'simportant discoverythatthe familiarDedekindcut andvon Neumannordinalconstructionsarepart of a more generalconstructionwhich leads to a properclass of numbersembracingthe reals and the ordinalsas well as many less familiarnumbersincluding -o, co/2,l/co, /o and co-n,where cois the least infinite ordinal. Conway furthershows that the arithmetic of the reals may be extendedto the entireclass yielding a real-closed orderedfield (1976, pp. 40-42); thatis, an orderedfield where every positive element is a square,and every polynomialof odd degree with coefficients in the field has a solutionin the field. From a logical standpoint,real-closedorderedfields areprecisely those orderedfields which are first-orderindistinguishablefrom the orderedfield of reals (Tarski1951), or to put this anotherway, they are precisely those orderedfields which satisfy the first-ordercontent of the DedekindCut Axiom as encapsulatedby the algebraicformulationof Tarski's elementarycontinuityschema. (1951, p. 49). Conway constructshis real-closedfield by means of the following conventionsand recursiveclauses. Construction.2 If L, R are any two sets of numbers,and no memberof L is > any memberof R, then there is a number (L IR}. All numbersare constructedin this way. Convention. If x = {L IR) we write xL for the typical memberof L, and xR for the typical member of R. For x itself we then write {xL IxR. x = (a,b,c ...I d,e,f...} means that x = {L I R}, where a,b,c,... are the typical membersof L, and d,e,f... the typical membersof R. Definition of x > y, x < y. x > y iff (no xR< y and x <no yL), and x < y iffy > x. PSA 1986, Volume 2, pp. 237-246 Copyright? 1987 by the Philosophy of Science Association 238 Definition of x = y, x > y, x < y. x = y iff(x y and y >x). x > y iff(x y and[not] y >x). x < y iff y > x. Definition of x + y. x + y = {xL + y, x + yL I xR + y, x + yR. Definition of -x. -x = {-xR I -xL}. Definition of xy. xy = (xLy + xyL- LyL,xRy + yR xRyRI I xLy + xyR - xLyR, xRy + xyL - xRyL}. (1976, pp. 4-5). This particularreal-closedfield, which Conwaycalls No, is remarkablyinclusive, Indeed,subjectto the proviso thatnumbersbe individuallydefinablein termsof sets of von Neumann-Bemays-Godelset theorywith Global Choice, henceforthNBG (cf. Mendelson, 1979, Ch. 4), it may be said to contain 'All NumbersGreatand Small' (Conway 1976, pp. 3-4) or allfinite, infiniteand infinitesimalnumbers!3The complex numbersandp-adic numbersare, of course,precludeddue to the ordering. On the other hand,if one follows the standardpracticeof identifyingthe cardinalN awith the initial numberoa, a practicethatis warrantedby the Axiom of Choice (cf. Kunen 1980, pp. 27-28), thenNo may be regardedas containingthe cardinals(althoughtheirarithmetic being defined 'naturally'differsfrom the familiararithmeticof Cantor). Whatis more importantfrom the standpointof the presentdiscussion, however,is thatin virtueof the classical Tarskianrelationthatexists betweenreal-closedorderedfields andmodels of complete(elementary)Euclidean,hyperbolicand ellipticgeometriccontinuaof dimensionsn > 1 (Tarski 1951, 1959; Szmielew 1959; Schwabhauser1965), the ndimensionalCartesian,Klein and elliptic Klein spaces definedover No (henceforth Cn(No),Kn(No)and EKn(No))4may be regardedas providing analyticrepresentations, relativeto theirrespectivegeometries,of the spatialrelationsdefinedover all possible types of positions near andfar (thatarerepresentablein termsof sets of NBG). It is importantto emphasizethatthe set-theoreticprovisos affixed to the above intuitivecharacterizationsof No and its geometriccounterpartsare crucial,since if one also permitsnumbersthatare individuallydefinablein termsof properclasses, or even the more multitudinouscollections warrantedin, say, Ackermann'sset theory (cf. Ackermann1937; Reinhardt1970; and Fraenkelet. at. 1973, pp. 148-153), then even richerarithmeticand analyticgeometricspectraare definable,albeitnot in NBG. Such structurescould, for example, be constructedin Ackermann'sset theory, and, of course, one could always constructmodels of such structuresin termsof sets of ZermeloFraenkelset theory(ZFC) supplementedwith the assumptionof the existence of sufficiently many inaccessible cardinals;however, these sets, unlike those of NBG, do not coincide with the sets of ZFC (i.e., the standardsets of the workingmathematician). Moreover,unlike NBG (or even Ackermann'sset theoryfor thatmatter),these set theoriescannot be shown to be equiconsistentwith ZFC, a theorywhich, unlike NBG, only has sets. For referencesas well as an excellent discussion of these mattersthe readeris referredto (Fraenkelet. al. 1973, pp. 119-153), where NBG = VBG + Global Choice. In a series of recent investigations(forthcominga, forthcomingb, and forthcomingc) the authorestablisheda varietyof resultsthatwere intendedto lend precisionto, and 239 justificationfor, the above intuitivecharacterizationsof No andits Euclidean,hyperbolic, andelliptic geometriccounterpartsandto clarifythe relationshipthatexists betweenthese elementarycontinuaon the one hand,andthe classical (i.e. real) arithmeticand geometric continua (henceforthR, Cn(R),Kn(R)and EKn(R))on the other. Included among these resultsare novel (categorical)axiomatizationsof the classical continuamakinguse of the Archimedeanaxiom in conjunctionwith hithertoforeunnoticedcontinuityaxiomsbased on concepts adaptedfrom the theoryof homogeneousuniversalmodels, a theorylargely concernedwith the studyof models which intuitivelyhave all possible types of elements. (Cf. Jonsson's (1956, 1960 and 1965) and Morley-Vaught's(1962).) We believe we are justified in referringto these axioms as continuityaxioms since in the context of these axiomatizationsthey areequivalentto any of the more familiarcontinuityconditions including, for example, those due to Cantor,Dedekindand Hilbert. Workingin NBG we furthershowed thatif the Archimedeanaxiom is deletedfrom the aforementioned axiomatizationsone obtains (categorical)axiomatizationsof No, Cn(No),Kn(No)and EKn(No),respectively. It is the author'scontentionthatthese results may be naturally interpretedas supportingthe thesis thatwhereasthe classical arithmeticand geometric continuashouldmerelybe regardedas arithmeticand geometriccontinuamodulothe Archimedeancondition(that is, modulothefamiliar conditionthatprecludesthe existence of non-zeronumbers(respectivelynon-degenerateline segments)that are either infiniteor infinitesimalrelativeto one another),No and its geometriccounterpartsmay be regarded as absolute arithmeticand geometriccontinua(moduloNBG). The above thesis, which may be referredto as The ContinuumThesis, rests in parton the following maximalitythesis implicitin our previousremarks;namely,thatdespite the ratherpicturesqueflavorof the above intuitivecharacterizations of No and its geometric counterparts,these structuresare essentiallywhat the picturesquecharacterizationssay they are;thatis, in appropriatesenses thatcan be madeprecise, No and its geometric counterpartsare (moduloNBG) the maximalstructuresof theirarithmeticand respective geometrickinds,or to put this anotherway, these structuresrespectivelyexhibitall possible typesof arithmeticand n-dimensionalEuclidean,hyperbolicand elliptic geometricgradations (moduloNBG). If this were not the case it is hardto see how we could justify the continuumthesis, for intuitively,an orderedfield F or a Cartesian,Klein or elliptic Klein space defined over F may be said to have 'no gaps' and therebysatisfy the fundamentalintuitionunderlyingall theoriesof continuaif, andonly if, they exhibit 'all possible types of gradations.' Of course, as our previous remarkssuggest, in the context of a formaltheory,any attemptto lend precisionto the idea of 'all possible types of gradations'is necessarilydoomedto failureand must be replacedby the more modest attemptto make precise the idea of 'all possible types of gradations(modulosome set of theoreticalconstraints).' In the classical theory,types of gradationsare theoretically significantjust in case they areconsistentwith the archimedeancondition,andin the context of our theorythey aretheoreticallysignificantjust in cast they are (roughly speaking)definable in termsof sets of NBG. Thus, whereasthe intuitiveimportof the maximalitythesis is thatthe purportedcontinuahave no holes thatare definablein terms of sets, the intuitiveimportof the continuumthesis is thatbeing gap-freein this respect justifies theircharacterizationas continua. It should be emphasizedthatthe continuumthesis and the maximalitythesis on which it depends must be regardedas philosophical theses, theses of the same type as Church's Thesis concerningeffectively calculablefunctions,the Cantor-DedekindThesis regarding the structureof the classical Euclideanlinearcontinuumor the thesis thatassociatesthe areaundera curve with the definiteintegral. In each of these cases a proposalis being put forthto the effect thata particularformalnotionprovidesa satisfactory characterizationof a vaguely conceived intuitiveidea. This being the case, a rigorous proof of the identificationis out of the question;at best, one can only marshalevidence to the effect thatthe formal notioncapturesthe intuitiveidea in questionand, moreover,it is unlikely thatany non-equivalentformalnotioncould capturethe intuitionto as greatan 240 extent, given the varioustheoreticalconstraintsplacedon it. Ultimately,the acceptability of any such identificationrests on the clarityand simplicityof the underlyingtheoretical notions, theirharmonywith the intuitiveideas, and the power andinstructivenessof the mathematicalandphilosophicalimplicationsderivedin termsof them. In the pages thatfollow we will providea brief overview of some of the aforementionedresultsin an attemptto lend initialplausibilityto the maximalityand of the purportedcontinuathatwill be continuumtheses. All of the characterizations discussed are concernedwith theirrespective structuresas models of first-ordertheories. We hastento emphasize,however, thatin additionto its rich structureas an orderedfield (or a real-closedorderedfield) No has a distinguishedset-theoreticstructurewhich emerges from the recursiveclauses in termsof which it is defined, a structurewhose interpretationwill play a fundamentalrole in the more completeformulationof our theory to be presentedin the essay referredto in note 1. Thus, in an importantsense, from the standpointof the presentessay the conceptsof absolutearithmeticand geometriccontinua andthe conceptsof absolutearithmeticand geometric(elementary)continuamay be used interchangeably. With the scope of our essay now having been made clear, we will introducethe definitionsrequiredfor the formulationof the aforementionedcharacterizationtheorems. Following theirformulationwe will offer a few remarksin supportof our two theses, leaving a more detaileddiscussionfor the expandedversion of this essay alludedto above. 2. DefinitionsandCharacterization Theorems A model A for a theoryT in a languageL will be said to be universallyextendingif for any models B and C of T in L where B is a substructureof A, C is an extension of B, and the universesof B and C are sets, there is a model C* of T in L thatis a substructure of A such thatC is isomorphicto C* where the isomorphismis an extension of the identitymap on B. We will also say thata model A for a theoryT in L is absolutely homogeneousuniversalif it is absolutelyuniversalwith respectto T, i.e. every model of T in L can be embeddedin A, and if it is absolutely homogeneouswith respecd to T, i.e., given any two substructuresof A thatare models of T in L (whose universes are sets) and an isomorphismbetween them, the isomorphismcan be extendedto an automorphismof A, i.e., an isomorphismfrom A onto A. If T is a first-ordertheorywith an infinitemodel, then the universeof any universally extendingor absolutelyhomogeneousuniversalmodel of T is necessarilya properclass. Since truthfor first-orderassertionsis not, in general,definablein class-structuresin NBG, talk of such models is not, in general,permissible. No improprietywill be committedin our discussion, however, since all of the first-ordertheorieswith infinite models what will concernus are VA-theoriesi.e., theories each of whose sentences is Tequivalentto a sentenceof the form Vx1.. .Vxn 1y1... ym0,where 0 is quantifier-free, and the notion of a class-model for such theoriesis definablein NBG, assumingthatL is a set. See the author's(forthcominga) for referencesand details. As the readerwill recall, being orderedfields, R andNo arerelationalstructureseach consisting of a universe(i.e., a collection of numbers)togetherwith a set of relations defined on it. Here the sets consist of an orderrelation,the operationsof additionand multiplicationtreatedas specialrelations,and two distinguishedelements (the additive andmultiplicativeidentitiesof the fields) likewise treatedas special relations. With this in mind the relationshipthatexists betweenR andNo can be expressedby means of the following two theorems. 241 Theorem1: (I) The following sets of axioms constitute(categorical)axiomatizationsof R; (II) by deleting the Archimedeanaxiom from the following axiomatizationsone obtains (categorical)axiomatizationsof No. Axioms for orderedfields (expressed in L = {+, , <, O, 1}) Archimedeanaxiom Axiom of ArithmeticExtendability or alternatively ContinuityAxioms Axiom of ArithmeticHomogeneousUniversality wherebythe lattertwo axioms we mean the following: Axiomof ArithmeticExtendability:The collection of numberstogetherwith the correspondingarithmeticandorderrelationsdefinedon it constitutesa universally extendingmodel of the above statedaxioms.5 Axiomof ArithmeticHomogeneousUniversality: The collection of numberstogether with the correspondingarithmeticand orderrelationsdefinedon it constitutesan absolutelyhomogeneousuniversalmodel of the above statedaxioms. Theorem2: Same as Theorem 1, except thatthe axioms for orderedfields are supplementedwith the additionalaxioms requiredfor the theoryof real-closedordered fields. To formulatethe geometriccounterpartsof Theorems1 and 2, we requirethe following additionalterminology. For n > 2, let En,Hn and ELnbe the axioms for complete (elementary)n-dimensional Euclidean,Hyperbolicand elliptic geometryfound in Tarski's(1959), Szmielew's (1959), and Schwabhauser's(1965), respectively. In these axiomatizationspoints are primitiveindividuals,andequidistanceand the orderof collinearpoints areprimitive notions;equidistanceis a 4-place relationand the orderof collinearpoints is a 3-place relationof betweennessin the Euclideanand hyperboliccases and a 4-place relationof cyclic order (i.e., separation)in the elliptic case. Also let En(respectivelyHn; respectivelyEL1)be En (respectivelyHn;respectivelyELn)less the corresponding (geometric)elementarycontinuityschema, and let Hn beH* supplementedwith the formulationof Hilbert'saxiom of hyperbolicparallels found in Szmielew's (1961, pp. 134-135). En, Hn and EL~are equivalentto the standardaxiom sets for classical Euclidean,hyperbolicand elliptic geometryless the correspondingDedekindCut Axiom and the models of E, Hn and ELnare precisely the models of EF,H' and ELnwhich rule out the full infinitevarietyof elementaryspatial anomaliesthatareprecludedby the axioms for classical real geometry. So, for example, in these geometricspaces all angles can be trisected,all lines joining interiorandexteriorpoints of a circle intersectthe given circle, and all lines joining points on the opposite sides of a line intersectthe given line. See, for example, Tarski's(1959, p. 26) for the Euclideancase. Theorem3: Let n 2 2. (I) The following sets of axioms constitute (categorical) axiomatizationsof Cn(R),Kn(R)and EKn(R),respectively; (II) by deleting the Archimedeanaxiom from the following axiomatizationsone obtains(categorical) axiomatizations of Cn(No),Kn(No) and EKn(No). 242 En (respectivelyHn;respectivelyEL*) ArchimedeanAxiom (Geometric) Axiom of GeometricExtendability or alternatively J ContinuityAxioms Axiom of GeometricHomogeneousUniversality wherebythe lattertwo axioms we mean the axioms thatdifferfrom theirarithmetic counterpartsonly insofaras the referencesto numbersand the arithmeticand order relationsdefined on them arereplacedby referencesto points (i.e., appropriaten-tuples of numbers)and the correspondinggeometricrelationsdefined on them (i.e., the analytic interpretationsof the respectiveaforementionedprimitivenotions). See the reference cited in note 4 for details. Theorem4: Same as Theorem3, except thatEn,Hn and ELnare employed in place of En, H' and ELJ,respectively. 3. Mathematico-Philosophical Remarks In virtueof Theorem 1, the basis of our confidencein the arithmeticportionof the maximalitythesis shouldnow be clear;for not only is No an absolutelyuniversalordered field, but with the sole exception of No (and of course isomorphiccopies of No) given an orderedfield F therealways exists an orderedfield B (whose universeis a set) which is a properextension of some orderedsubfieldA of F for which thereis not alreadya proper extension of A containedin F thatis isomorphicwith B, where the isomorphismleaves the elements of A fixed. Thus, unlikeNo one can always supplementF with types of arithmeticgradationsthatit does not alreadyexhibit. By appealingto Theorem3, the geometricportionof the maximalitythesis can be arguedfor along similarlines, and by appealingto Theorems2 and 4 respectively,supportfor regardingNo and its geometric counterpartsas uniquemaximalarithmeticandgeometricelementarycontinua(modulo NBG) can likewise be obtained. It is the 'fact' thatNo andits geometriccounterpartsmay be regardedas unique maximalelementarycontinuamodulothe set-theoreticproviso which providesmuch of the motivationfor regardingthem as absolutecontinua;being elementarycontinuathey have a plethoraof the intuitivepropertiesgenerallyattributedto continua,andbeing the uniquemaximalsuch structuresmodulo whatis at once the most liberaland fundamental of all theoreticalconstraintsof standardcontemporarymathematicslends a strongair of legitimacyto theircharacterizationas absolute. In additionto these considerationswe believe thatthe remarkablestructurealsimilaritythatexists betweenthese structureson the one handand the classical continuaon the otherlends additionalsupportfor the continuumthesis. One'sconfidence in the thesis would undoubtedlybe enhancedstill furtherif it were possible to find a simple axiom, generalizingone of our most basic intuitionsregardingcontinua,which would single out the purportedabsolutecontinua' from the wealth of otherelementarycontinua. As the authorshowed in his (forthcoming b), and as we will now explain, this is indeed possible. of the purportedabsolutearithmeticcontinuum,one To obtainsuch a characterization simply supplementsthe axioms for real-closedorderedfields with the following assertion (where X and Y are understoodto range over sets). 243 (*) VXVY (X < Y -- zz (X < (z) < Y)). Similarly,to obtain a categoricalaxiomatizationof Cn(No)(respectivelyKn(No))one merely has to supplementEn (respectivelyHn)with the following geometricanalogueof (*). (**) VXVY ({aVxVy (x e X & y e Y - Baxy & x ? y) -> 9bVxVy (x e X &y e Y -> Bxby & b ? x & b y)}, where Bxyz is read y lies between x and z (the case when y coincides with x or z not being excluded). Finally, to obtaina (categorical)axiomatizationof EKn(No)one simply has to supplementELnwith the elliptic form of (**). See the author's(forthcomingc) for details. (*) asserts thatthe collection of numbersis absolutelydense and, intuitively,(**) assertsthateach line in Cn(No) andKn(No)is absolutelydivisible in the sense thatgiven any two subsets of points X and Y at which the line can be divided (i.e., intersected),if X < Y thereis a point lying between X andY at which the line can be divided further. The elliptic form of (**) has an analogousinterpretation.These axioms may thereforebe regardedas axioms of absolute densityand divisibilityrespectively. Thus, in virtue of our earlierremarks,we see thatthepurportedabsolutearithmeticcontinuumis simplythe uniqueabsolutelydense orderedfield whichpossesses no elementaryalgebraic limitations,and (exceptfor isomorphicstructures)thepurportedabsoluteEuclidean, hyperbolicand ellipticgeometriccontinuaare simplythose Cartesian,Klein and elliptic Klein spaces which admitno elementaryspatial anomaliesand whose lines are absolutely divisible. The term 'arithmeticcontinuum'is used to denotethe system of real numbersbecauseit is held thatthis numbersystem is completelyadequatefor the analyticalrepresentationof what is known as continuousmagnitude. While we believe it would be rash to asserta priori thatphysical science will neverhave reasonto rejectthis thesis, therecertainly seems to be no reason at the presenttime to do so. The theoryof absolutecontinua sketchedabove is, therefore,not intendedto challengethe role played by the classical theoryin physical theorizing. Nor do we wish to suggest that,unlike the classical theory,it provides 'the correct'analysis of continuity. Rather,we merely wish to call attentionto the fact thatNo and its geometriccounterpartsadmitnaturalinterpretationsas absolutecontinuaand, moreover,thatthese interpretationslend insight into the structure of the classical continuaqua restrictedcontinua. Notes lThis essay is an expansionof a portionof the materialpresentedby the authorat the meetings in Pittsburgh.Many of the resultsreportedat thattime have been deleted and will be discussed in an expandedversion of the originalpaperbeing preparedfor a forthcomingissue of Syntheseconcernedwith theoriesof continua. Proofs of a few of the deleted results can be found in Alling and Ehrlich's(1986a and 1986b). 2Forthe purposeof this essay, we are ignoring certainaspects of Conway's constructionthat would be requiredfor a formaldevelopmentin NBG. See Conway's (1976, pp. 15-16, 65) for details. 244 numbersgreat and small are equated 3Strictlyspeaking,in these characterizations with numberswhich are elements of (commutative)orderedfields. This, of course, includes all the numbersthatare containedin orderedabeliangroupsas well as any of the otherorderedalgebraicsystems which can be embeddedin an orderedfield. 4Forthe definitions of Cn(F),Kn(F)and EKn(F)where F is an orderedfield see, for example, (Schwabhauseret. al. 1983, pp. 16, 208 and 213). 5Thecharacterizationof No as a universallyextendingorderedfield was first establishedby Conway (1976, pp. 42-43). An alternativeproof was providedin the author's(forthcomingb). 245 References Ackermann,W. (1937). "ZurAxiomatikder Mengenlehre."MathematischeAnnalen 131: 336-345. Ailing, N. and Ehrlich,P. (1986a). "AnAlternativeConstructionof Conway's Surreal Numbers." MathematicalReportsof the Academyof Science Canada, VIII: 241-246. ___-_-_. -_-___- __ (1986b). "AnAbstractCharacterizationof a Full Class of SurrealNumbers." MathematicalReportsof the Academyof Science Canada, VIII: 303-308. Conway, J.H. (1976). On Numbersand Games. London, England: Academic Press. Ehrlich,P. (forthcominga). "AbsolutelySaturatedModels." -___ - _ . (forthcomingb). "AnAlternativeConstructionof Conway'sOrderedField No." Algebra Universalis. -_--__ -. (forthcomingc). "UniversallyExtendingContinua." Fraenkel,A.A., Bar-Hillel, Y. and Levy, A. (1973). Foundations of Set Theory, Second Revised Edition. New York, N.Y: North-HollandPublishing Company. Jonsson, B. (1956). "UniversalRelationalSystems." MathematicaScandinavica 4: 193-208. __-_____. (1960). "HomogeneousUniversalRelationalStructures."Mathematica Scandinavica8: 137-142. -_-_ -____. (1965). "Extensionsof Relational Structures."In The Theoryof Models. Proceedingsof the 1963 InternationalSymposiumat Berkeley. Edited by J.W. Addison, Leon Henkin and Alfred Tarski. Amsterdam:North-Holland Publishing Company. Kunen, K. (1980). Set Theory. Amsterdam: North-HollandPublishing Company. Mendelson, E. (1979). Introductionto MathematicalLogic (Second Edition). New York: D. Van NostrandCompany. Morley, M. and Vaught, R. (1962). "HomogeneousUniversalModels." Mathematica Scandinavica 11: 37-57. Reinhardt,W.N. (1970). "Ackermann'sSet Theory Equals ZF." Annals of MathematicalLogic 2: 189-249. Schwabhauser,W. (1965). "OnModels of ElementaryElliptic Geometry." In The Theoryof Models. Edited by J.W. Addison, Leon Henkin and Alfred Tarski. Amsterdam: North-HollandPublishingCompany. ?___________, Szmielew, W. and Tarski,A. (1983). Metamathematische MethodenIn der Geometrie. Berlin: Springer-Verlag. 246 Szmielew, W. (1959). "SomeMetamathematicalProblemsConcerningElementary HyperbolicGeometry." In The AxiomaticMethod. Editedby Leon Henkin, PatrickSuppes and Alfred Tarski. Amsterdam: North-HollandPublishing Company. ________. (1961). "A New Approachto HyperbolicGeometry."Fundamenta Mathematicae 50: 129-158. Tarski,A. (1951). A Decision Methodfor ElementaryAlgebra and Geometry(Second Edition). Berkeley and Los Angeles. Universityof CaliforniaPress. _ _ _ _ __ (1959). "Whatis ElementaryGeometry?"In TheAxiomaticMethod. Editedby Leon Henkin,PatrickSuppesand AlfredTarski. Amsterdam: North-HollandPublishingCompany.