You are on page 1of 35
LAWRENCE SKLAR? Facts, Conventions, and Assumptions in the Theory of Space Time 1. Introduction Given a physical theory, there ate always those who will try to for- ralize i. At a minimum this entails an attempt to present the theory {informally in terms ofa number of basi propositions about the world from ‘which all the consequences of the theory in question can be derived informally. At a maximum such an wndertaking involves an exact and rigorous presentation of these fundamental propositions in some formal language (frst order quantieation theory or set theory, say), with the ‘implicit clam that all the consequences of the orginal theory could, if they themselves were suitably “formalized,” be derived formally from these rigorous fundamental propositions by the mere application of the formally specied rules ofthe logic in question Clearly such “axiomatizaton” can he a useful (or even crucial) compo- nent of angaing science or the philosophy af science. Axiomatization need not be merely a more or ess carefl display of some degree of technical ingenuity applied to no apparent purpose, even if that Is what it fe ‘quently turns out to be, But just how axiomatization can be applied to scientific or philosophic purpose i not always completely clear from the sxiomatizations themselves, The seape and limits of formalization as an aid to science, and especially to philosophy, ae the subject ofthis paper. But I shall treat these general questions within the limited context of some formalizations of the theory of space-time, since these particular cises provide highly laminating examples ofthe general issues ore 1am pl othe Jeb Sion Guerin Menor insti fr th oe Salorart Sage pese wich pe Son Sieh on eld hn ont ee 2m FACTS, CONVENTIONS, AND ASSUMPTIONS HL Formalizations and Their Differences All standard axiomatizations have some fii components, Fist of all there is the accepted logical framework, which may he restricted to *pely logical” tems or may encompass a vocabulary of mathematics that Js in general taken for granted. Then there are the primitice descriptive terms ofthe theory ia question. They are just “givens,” and although they ‘may be “explicated” informally outside of the formaltation, nothing within the formal presentation pretends to “define them or otherwise specify their “meaning.” There are the axioms, statements accepted as true and containing in thete vorabulary only the “logical” and primitive descriptive terms—st least iitilly Next there are the definitions, and over their status muck philosophical Abate can rage. In the definitions we have, standardly, new descriptive terms introduced on one side ofan equation oF equivalence, and phrases Involving only the primitive vocabulary or previously defined terms on the other side, Not all equations can serve as definitions. They must meet the conditions of eliminabilty and nonereatiity. If dention is lgiti- imate, one must be able to replace the defined term in whatever contest it appears in further on in the formalization by some expression using prim= tive terms alone. The theory that results when the definition fs included rust not contain any consequences phrasable in terms of the primitive vocabulary alone that would not be consequences were the definition iminated and the defined term replaced everywhere by its defining expression." Let us note some important ways in wh From one another: (2) Two formalizations may have all the same consequences, but difer ‘n that one and the same term appears 1s a primitive in one formalization and as defined term inthe other. Naturally this wil force the formaliza- ‘ions to differ also in which eonsequences they count as definitions and which as postulates. What we count as a primitive term and what as defined are in some sere not “internal” tothe formalization ut imposed wpon it by us. That 1s this diferential attribution of status to terms is not forced pon ws in any way by the consequences of the formalization. Let us call formal tias that differ oaly i regi to which terms are considered primitive and which defined, formalizations difering mevely in the statue attrib edt ter ich formalizations may differ 20 Lawrence SMar (2) Two formalzations may agree a to which terms ae primitive and which defined. And they may agrec in thatthe consequences ofeach Fanible in primitive ters lane are the same, Yet they may til der least one consequence which i phrased sing a defined term. Let us Say that such formalization der mercy in definitional comaguence a) Suppose two formalization difer in what they both agree to be a ‘consequence ramable in primitive tems. Then a we shall see, al par ties are Hikely to agree thatthe Formalizations are based on dierent theories. But suppose two formalization fer in some consequence. And suppose one formalization so categorizes terms as pzlive and dened tht # declares the differences tbe only in defitienal consequence, Suppose further that the other formalization declares sme of he dif ferential consequences tobe framable in primitive terme only, ive its duscation of terms as priv of defied. Let ws ell fomaliations ‘that ifr n his way, formalzations dering ony in ome acy definitional omsequence (4) Two loemalbations may agre ast which terms are primitive and which defined; and they may hive eae dhe same consequences. But ‘one formaliration may dedare, "in the margin,” some of these conse- ‘quences tobe definitions thatthe other cll postulates or axioms, and vice versa, Let ut say that two such formalztions difer only In the fatiibuton of defintional stats to consequence. ‘Therelevance and importance ofthese distinctions wil hcome clear ax sve proceed. Ishll be concerned primarily with fo fundamental ques (Gd) Given to formalizations that der in one of he ways nated shove, what god scentifc andor philosophical grounds could there be fr pre Serra one such formalization to another? '2) What are the scientific andor philosophical consequences of opting far one sch formalization over an alternative that ders from tin one a the ways noted above? 11, Formaliations and Minkowski Space-time A. Robbis Formaliation of Minkowski Space stead of continuing the dscassion at this abstract Deve, lt us send toa concrete ete. shall tart wth Babli’ ingenious, cht atl terapprecated aionatization of Minkowski spate Mont a wt ‘would view this at formalization of sored ett at sinc Hah preferred to contrat his theory with hat of Einstein, lett hep the ore entra form above, In Robs formalism one takes we priitve clas of ‘events andl one prnitiveeelation on them: is alter b, Intuitively Is after and only possible to propagate casa signal ro b toa, ad a and b are ditnet events. Robs imposes twenty-one axiomatic ‘conditions on their telatian, scent to guaratee tha the sratare of causal conneetblity among events fe tht among events in four ‘ioenstonal Minkowski space-time Robb then shows that tis possible to Inte deistons, in teens of tho ‘alte relationship, that are sulicient to capt such wellknown features of Minkowsht space-tane as tnelike connectivity, nl eonnectn~ ty, spacelike connectivity, ec. More surprisingly, he shows that one an ler deficitions of ‘sa timelike ier path,” sam intr com fvent to a given inter’ even for spacelike intra, stich that the Suce-tme structures so defined wil gan corespond inal hei geomet. He features to the ural sich strctres of Minkowshispuce-tine ‘The temporal asymmetry ofthe ‘after relationship sens ireevant to his main task, and indeed i canbe shown that ome starts with the symmetric reaionship corresponding to eithera is ater barb ialtra then, following outa Rob-ike contraction, one can const al howe Features of Miakowaki space-time that are thensches independent of a shoe for the “dcetion of tine,”* ‘Now Robb, being concemed ony to construct «formalized theory of space-tie, his tle tos about dhe relationship between the observable Dhavior of material objets like clack, rods nd fee parties andthe space-time structures s cnstmcted. He does tll us, however, that tis sn empirical fac that light ay travel nll straight Ins, a he constrcts ‘hem, and I assume he wonld take ith Sn cur fll physics we woul simply add more aioms to telus that atomic docks tease Hime-tike Intervals relative toa reference frame, and rg ode spacelike intervals relative to a reference frame and that fee partes travel tinelike stright lines. These are totally nondefinitonal on his view. We shall Slag thi in more deta hort, losophil Thesis Drawn fom Robb's Formation Let us ok atone aspect of an answer to our second question above What inyrtant pilosa cancun can be drawn fom apting for 1 shen frniation of athens? Jolin Winnie as aged lows Lawrence SHar Grau and others have asserted thatthe metic structure of space time is conventional sine, even given the topology of space-ime, no Particular st of cic seaions between the events Is sngled ot at prefered, Therefore the cheice of metric, i some sense, arity Hence the metric of spuce-tine is conventional." ‘While Rob dd not deus the problem offormalzing the topoony of specetine, it clear that for Mahon space tine, «frmalation is ‘salable that takes ony the ltr’ relationship between evens as pie tive, and in which we can fll define all the topological features of the ‘space-time, This certainly ten the eli sense ven the “afer” felationship, ad the Robian some governing, we can deine what it ‘ean for an event to be in the interior of the Forward (respectively thckward light cone of the eent. Then wean take asthe topology af the space-tine the Alexandr topology, i, the cause topology in which {Ml such interior of ight cones are open. This wil agree tn all its stru- ture wth the ws manifold topology on Minkowskt space-time Bat, Winnie continues, Robb hae shown thatthe mete congruence stevie ofthe space-time fe efnable by the afer’ relationship sne “The same basic relation defines siutineosly the tpolgia and metre features ofthe space-time, Thus we ctnnat ely Bet the tpology and stil ve fee choice at tothe met, hence the metre 5 not eeally Comentional Ifwe ot fr Rabb’ formaliztion we se tht, at esti the Context of Minkowski space-tine, the metic af the spacetime is nt realy a matter of “convention.” “A fall critique ofthis argument ill reveal much that of genera piilosphic importance. But raher dan attack this thesis det, 1 Sl ake more iets route, hope the reader wil bear with ne throu 2 lent digreson. , General Hlatvstic Considerations In the Snt edition of Raum, Zeit, Materie Weyl suggested that we could esuming spacetime fo bea fdinensinal pedo Remannian Imanld, completely map ot the etre ofthe space-time i which we live by using ight rays aloe. The fc that the light cone stractre of he space-time ithe trate aft contectity sogests that ‘ale isa fificient primitive on ich to found the definitions af all the mete concepts we tse to characterize the space tine. Buta Lorent piped 20 ‘ut Wey, and as Wesl noted ia atereditnsof hs bak, this hess is wrong, For any two noniometric space-tines that are conformally “equivalent (Le, such dat there He «one-one ange-peserving tan Teemation fom one tothe other wil have the sine ight cone strctre ‘Weal then painted oot that f we added to our bay of observation ata, soar consisting ofthe paths oat as, dhe paths ee partic, and if we assumed these fee partis traveled tne geadeses nthe Sce-time, then we ould indeed filly determine the mete: For there ‘sa one-to-one napping fom one space-tine to another hat preserves the pas of materal particles (he, the timelike geese) then the space ties are isometries up ta constant factor" Robs ess onthe definblty ofthe metre by casa conmectiilty fm Minkowsls space-time, a hn tht tll ws hat the mete most ‘etany ot uniquely xed by the ul strstre in Riemannian space times, are af eonse, completely compatible with ech ater, Suppose we iow that spacetime fe Minkowekian, Rob has shown ws that we can termine the interval separation of any pr of pints (lative to ven ‘separation taken at uni) yea of ght rays alone, light rays being {aken to demarate the bowaary of eanallyconnctible sets of events. But ite know only that the spacetime f Reman, then the rests ‘noted above tll stat we eld not even determine that these wt fat (Minkowskian} using Hat rays alone, much Tess messine elative Inter separation ally between al pat of pots, ‘To clay this point farther, we might reflect far & soment on an Important result of Zeta, le sbi hat s'space tne Minkowski, ‘ny automorphis ofthe spacetione onto tl that preserves, in bath Aiwtons, all the casa eect relations id the ine-ordering a sents that ate casaly connect (is er # a Robb sense, and 715 the mapping) after fia) and id alter ar P(e) bamember ofthe group, where ithe group genersted bythe ‘othochronss Lorente grasp, tanltions, and dans of the space time In ther word, dhe pete is Minkowekia, “sual ples the Lorente group"* ut Zea mt certainly dos not show us that any mapping between ‘70 psd Remannian spacetimes which «one-to-one mapping and ‘which preserver ‘alter in oth directions ix such an enended Lorentz "wonton ink his pall othe at tht ste ol shows Laverence SMar ‘statin Minkowsk space-time, the ltr’ relation defines ert lines ‘nd both spatial and temporal congruence, he has certainly nt shown ns that we ean tel which pseud-Rlemannian word we are Se byexing the at on ‘iter relations aloe. Fr, oace azn, evens Habba or hater ar satisfied, the space-time need ot even he Minkowski, Now if pace-tine s Reemanaian, one of to possiblities ld: eth the casa connect (or er) stracture aheys Robh’s axioms oF les not tn general it does wo fat, unless the space-time i confor ‘nally ht, the aus structure ofthe spacetime will ot even be lcally like that of Minkowski space-time a each point” And even the space time i conformally at, Le. lly conformal to Minknwski space-time, ebb’ s axioms sil ned not hold, for they ar of lal nate; they wt old only the space-time ts conforial to Minkowski space-time ina ‘lal way. Its, fn it, dhe global nature of Rob's aon tit allows him to define the metre structure erm ofthe cas hen hi ans sre saifed, For, at Wesl showed, the local canal structure iy in- Suficent of the metric events the Minkownkian xs. Now then, Rois aioms may nat hold. We allo his a poss, Delesng as we doin general relativity and therefore inthe posshlty that space-tine és wot loally conformal to Minkowk! space-time, I eed, when we lok tthe also inevtbly singular ate of onnlog cal solutions to the general relative Bld equations, Ht seems most probable that Rob's axons do wot hal And # Robb'saoms all, x0 loos his program a defining the congruence ofthe metric i terms of sa comet. ‘Bt suppore we did belive Robb's axioms hold. Should we sceept hisdelinitons of the metric congruence terms ofthe eal onnecth ty relation? Two posites are: 1) While we believe Rob's aions hol, we belive special elitiity fm genera dace not. From the general relathiste pint of view this corresponds to a theory af space-tine which, wale globally conor to Minkows space-time, i ot Hl Empty thsi revealed t us by the ft tht fe putes do nat tewel timelike straight lines a catsally defined in Rob re ne dt rmewsire Robbin spi intervals, and alomie clocks do ot mene Robbian timelike snterale (2) Not omy da we bee H's son hold, but we believe that the speci relate pretons ut fe aries, ros, and locks ol wll IN, Some Alternative Formallzations of Robbian ‘but non-Minkowskian Spacetime A. The Masinal Robbian Formation bls aims hold but From the general eave point of view, the soce-time ls believed to be non-Minkowskian, What altemative formal tons could we choose? ‘Option A The masa! Rebbian choice, rw we Keep a lose to the Rabbian analy ax posible His post lates fo ‘ter remain the same. And we hold to the sane definitions OF ‘course we must now drop such assumptions as tit fe particles evel tele stright lines, that rid vd tease interval et ov we hve some warrant fr taking this approach in the history ofthe ‘theory afterall In Newton space-time for example, we never assume that putes acting oly wnder the inflence of gravtationfallowed spce-time geodesics. Iastead we tok these gees t be straight lines tnd said that the trajectories of the parties were sorted fom the ess bythe graitational free. Now ofcourse, given the sumption that nad rods measure general relative spacelike interval and hat clocks measured general relative timelike intel, we must allow these “frees”, eter, potent) toinluence the bchavir of atrial bjt in the spacetine in other ways as well The gravitational free must become also «gravitational “tetchingsbrinking” Bel Tn thi ight iis interesting to ee ow Re himself as responded to the intradution of gener elatty. He wories about how hie space- time theory sgt to aren the Hight of general ela nan append to his The Absolute Relation of Tne an Space. is remas there ae Dri and enigmatic, but Think we ca see that something ike what we ate propising here as Option A, bt wth diferences, i going on Tabb sis that although there might be some reason fr sped of partes as tveling gases none of Eittes “complited geome ties." s a facon de pate,“ dows not inply any ‘curvature af pace, 1s meaning le space-time stil Hat, and imelike geodesic ae sl straight Hines Wile fice patees can be viewed as traveling srlesis” ina fiiows Riemannian spacetime, they donot really inet wee a al real in ner eltity even ht rays ay Fl Te geese on eet aia ann, eve his reply at Laserence Sklar ‘shard to comprehend. He seas to ange that there maybe aber caus inflences ha wiley his axons even fight pathe donot, In pti lar be thinks the propagation of gravitational afluence might take the place of ight marking ou the boundries fst of cnsaly connect fens The, hess, we coud stl adopt is spacetie theory, aking t ‘san empiri fact that gravitation selets the extremely easly connec tile events, and allowing, ike fre parties, toa infoenced by the smtter ldo fal to flow the"eal” godess f the spacetime. But ofcourse ths wll not do. Wf general relativity is coer, there can ven be space tines in which extremely eaaly conectible event sim iy do not moet Reis postulates onsmch as grovitatona cei ‘cao propazates with the velocity ofliht whether unr the inlence ‘of rea matter” or fn an empty now-Minkowskin spuce-tine. Robb Should ave fared the fact hat the ight of general relativity, even is ‘om for asl conectiity miht rea down, and with this whole Program fora theory of spacetime But our modest Robbianism is sll worth examining, Once again, & stale fat lt the Rabin axioms for eal connectility bald p— tee the spacetime is atleast saul coaforma to Mikowsh space tine cam seta formalization af ou space-time theory that pre server Hobbs aioms and defstions, even for such things as timelike ‘stoves inert lines) and spat and temporal congruences, and make bur theory ft the fats by changing the postulates that cancet the be avr of fce pate, rgd os, and lacks with the space-time str ture That i, we can intodice ray ata "poteat” superimposed on the spacetie. 1. A Formalzation Dierng in Defiationa Consequence Only ‘Option B: Choose formalization tat gies with Rob's on the sti of terms and on those consequences framable im only the (mutlly ‘ered primitive tems bat tht ile from Hob on the ditions ‘fered, acon definitional consequence generally. What would dis ak ike? Fire ofall the formalization would ake ax primitive only hoe ers ‘designed ws primitive by Rabb, And any ropition fovelvng only Primitive tonne would be a consequence af the new frit il ‘nly f were a enaeqence of Ras loristion thn would ler frm Habib’, fr expe y de ss pth ofa ine material patie Al whe fice particles would not travel tele geodesics, sg rods wold ot sete spiclite intervals, and clocks would not measure tele Inter, In this formalization the weal seatticasmptions about the relttonship between these material objects and these sacetine structures might hold But the Robbians and the proponents of hs new spacetime theory Init very well are that tel 9 formations were formalztions of the same theory, althovgh Ido not know i abl himself wok have cepted thi line How wou they ange? Perhaps a flows, “We agree ‘bout the primitive consequences oon formaizations. And the primitive Comequences af formulation conta the tly of factual inport. ‘We clams o define some nonprimtve terms in different ways, but hat {sa frechate in any cae subject only tothe orl coon of proper dltnabilty. Having chosen these dilleringdfintons, sadly supe ing thit we extract from our axioms apperenty compatible conse- ‘qences in eases which ey ae proportions combining defined ters, But this apparent neompatibiey i only apparent. To thik that here teal incompatty ol prey to alles of equivocation ede by fare to pay prope ltenton to the dilerngdeion we afer af he !nonprimitive terms. Ou ormalations, despite thee diving inguin appearance, really ave just diflerent ways af expressing one and the same theo. Weroted shove tht what we call the masinal Rebian formalization tad the “advantage” that inthe cae which Rb axioms were sad Tt én which not all of spel elatty ws covet, thi Formation ‘made grty play a eole sma n sme waa tt role a a Src fn [Newtonian mechanics and in spel slaty. The advocate ofthis new Foemalztion might an ike his about the ear ply “Ii ewe that in pre-generl clans phsis we trated gravitation a Free superimposed on the spacetime. In my formation, actualy ado servationally euisalent to Rabb’ though it, we all ei gravity realy ‘were curvale of pacetime. Thi i more ature way of allig Han Tubb’, Tor my way maintains the connection between righ rods at sta separations, and betwen acs and tempor separations, wich fesse the pre-reatstetheoy and which fe preserve bah i the ‘eal Vrms of cl att adn thet Formalzations of semen rltiy 1s oy formalin ww speak of particles ato aly by Loverence Sklar sevity as flowing curved tinelke geadesles, and we speak ef ened sace-time, even though we dd not talk this way i the pregeneral elas theory. But we rst realize that that older way’ of aking wa nly a way of talking. For just as we have a choice between the Robbin: Minkowski space-time ps gravitational potential els, and the curved space-time Formallzaton, to formalize oor present theory of ice-tine, which—in general elatiiatic terms lolly conformal 0 Minkowsl spacetime but not Rat, we bad the same options in pre general elastic phsis. For example, as Trautman has shown, we ean formalize Newtonian physics fom the standpoint of curved spacetime jst the maximal Rabbian shows swe can formalize or present cory from the tandpent of Hat space-time. Ia the curved space-time formal zation of Newtonian physic, the sine natal association af igi ode and docks with the appropiate Intervals fe maintained. But we now talk of feve particlesthose acted upon only by gravty—as ‘ollowing the timelike geese of curved spacetime." But we mst note that n both thee cases isola mater af how we talk, not what ee belce. For in bo cases the observational (etal) ontequencesof the altemative formalizations are the sae. Ths follows from the fact that the formalzatins agree onthe primitive and dined slats of terms and propositions and on primitive consequences, dilfering tml on definitions! consequence ‘Such a thinker might continne to defend a preference for his way of Fonmaization as fllow; "While both the maximal Robbin and my fr: imalization of thi space-time theory are hated om one and the same theory, mines superior to hein a numberof ways, even hough dilers fiom his merely n the definitions adopted forthe (agreed defined terms The oily advantage ofthe Robbian's formations hatin is theory asin ‘he unl Newtonian theory. parties acted upon by gravity do nt travel timelike geodesics, but instead ae acted pen by frees. AST have Just potted at, this “continuity” with pregenerl relativistic phys f nota ‘tea pint, for we could, sf we wished, even reformulate Newtonian Dhysls to meet the general relativistic way of talking which pts tted upon only by gravity alow Uli geodesics In «curved spa “My way of speaking his other continuities, both with preencral relative physicy and with general veatty, which the wail Tig ban's formulation lacks While: Newtonian ysex,spevial reali senor relativity and. my formulation all speak of rods and clocks as orectly messing space-time intervals, the maximal Robin drops this ‘common way of speaking forthe sole reason that this the only way he ‘an kep his Robbian definitions of congruence, ete “ay view further shown tobe preferable by the following reason: I believe Robs ania for nut contecibiity had, Far they had both in ‘the maximal Robian formalization and in mine. Bu kow about general relativity, s0 realize my present theory may be wrong, so wrong tha the obian ions really donot hold Should 1 Ascover this, 1 shall have change my tory: Now the main Robbian wil ot only have ge tp Belin the tt of his causal eonnectiblity axioms, he wil ave to ‘opal his definition of consrience mele genes te, as well for the’ very formulation of these definitions depends upon the Rabbian soma holding “Toon the other hand shall, ofcourse, have to drop my Robbian axioms of cust connectblty, But I shal he able to preserve my’ definitions of tine geodesics and congruence intact, for they do not depend, ny Fovnazation, on the Robbianasioms, and they corespond exactly the ‘ul general relativistic dfitions of thee space-time tris, definitions tehich ate adequate even when the spacetime i not lolly conformal MinkowsKt space-time. My formaluation then, while admit dif ‘ng only definitional from the maximal oan’, x more continous 1s way of talking—both with pre-general relativistic physics and with ener relativity a their usual Formultionsthan is the maximal Rb Than’ way of speaking. Thus one should opt or my formulation.” Here the reader will undoubtedly recall the familar thesis of Eddington, Slick, and Reichenbach as long as two Formalizationshaveall the same ‘sbserational consequence,” they ar the foaaliztions of exe and the ‘me theory, ering only in Ings oman. 'A this point several crucial Sues abot definition in formulations tuve Sally surfaced, an we must now make an nia ass pon the. The reader should ante that our tentative prabes here wl arly be the list moves we shall make om these ites, for we are now about to slag fakonental problems which wil pear wpa and again a8 we ‘mine partie Fornaliations of particular theories, thle snares on deences, the ras Lagi adopting tem, a he om sper of din Laeronce SHlar tions were ase on the sane teary, Fundamental ra Asstmeds entity priv consequences constitates identity of ebser “ational and of faa consequences Bat why should anyone assume this? [thik the answer is lr both fralitions the choice of primitive for of defined sats fra term, and the resulting clasifcaton of conse quences ofthe formalization nto pemitve and definitional, were de ‘Heme to relegate the totaly of abnervatonal and factual consequences ‘the Foonltons—these bo dase eingasiumed tose coextensive to the clas of primitive consequences, rat ofall the argument snsumes that 3 consequence of theory cannot be fc les i, atleast in pnp, abserahly ether the CGae arnt the case. We accept thi «theory cannot ental ay putative “fhe” unless they ave putative observable face for, inthe best ver ications veln, we acept that «Tata diference with no cbservable consequence only an apparent etal diferenc. Second, the angoment Seemes thatthe chseation of terms and propositions into primitive tnd defined not an aitary elasication, but one designed to mark ost inthe margins of the formalization thowe parts of coments that are ccs templet and dose tat are open to "airy" lng We choir ‘What shoudl count as primitive term in formalization? The frequent anor that primitive terme should be nly those whose ppliebity oF omapolicabity to situation can be determined witoet in any’ way presuppning the tath athe theory a queston. Atel, the fundamen tao ofthe formation ane to he presented as ramed inthe prin tive terms ony, And 6 the emp est ofthe coretness of these sions tha wil ll us wheter or 0! the theory's formalized is comes Buti we cod not determine the appli of «primitive term to « situation without presipposng the cnrectes af he theory in question, aw cold empire tsting ever set ver way? ‘One version a this these would have us reset the primitive terms those whove appleby nongpplbilty in some sense mate immediate sensory awareness.” Pritve terms are then the basi o> sernation terms af some rally empistic philsophies of wwe “This is cetanly Rois pontion with rept oiler print, For, hh say the rounds for choosing ler priv ae that we cs be drety conscious of snp of ese nt—th a igh "eons {cana hs sine sng moc nals tht my ‘heoretial presupposition or inference, el that one ofthese evens inthe Pairs alter th oer ‘The reer acqusinted with some history of philosophy who rade Robs introductory sections the stats of afr and on the “ality of simultanity.” and who encounters Robb’ bel tha snatinety e ‘event spatial separate i nonsense, wil recognize a dove sina Detween Roblis positon and that expressed in the think chapter of Bertson's reply to special relativity, Duration and Simultancty Berguon's work har been, 1 think, unjosty neglected, This lage bis cn fal, for many’ readers have Been dsatisled with hi acount a felivity beease of the ition rampant throvghout Bergsons ‘work, and because of the later parte of Duration and Silane which onsttte a confised attempt to refit of the socalled clock pardon in ec reltity. "| “This rationale for choieo' primitive lead immediately to philesop hi cal trouble: it lok a though our pice is gong to be fue on 4 Solpsistie, private, subjective observation base Reichenbach, in his wellknown aetmatiation of special relativity, was well aware ofthis problem, eventhough ebewhere ini work be tlk ore ike Rab and Seems to tke the primitive atthe “diel apprehendable by a cone ‘ousnes.” Ia his Book on spel relativity Hichenbach vss, fo xa le, coincidence of events aa primitive. But, he ys by ths Ten Physical coincidence, not coincidence of “sppearance” in some sje tive consciousness. Reichenbach als hose propaitions that are ramable ‘tly in term of his piv, clomentery fects, What makes thom sementary ts not immediate apprehendabiity by consiusness, but rather the fact tha hey’ "remain tavarane wth respect toa gest vey “Cipterpretations” For example, we perfor the Michelson Morley e- periment. We might have many dileent phys interpretations ofthe Fesuts But all ich theoretical accounts will agree a to wheter the ysl iteference lies dior dna remain coincident with desig ‘ued marks when the tnterferoteer wa ated.” ‘As stands, this definition of elementarity Ives something tbe sited, For sinely Tul have considered theory im which explained {vay the apparent wl esl the Mihelon-Morely experiment, not ‘wa done by mauming the ml rolls were x psc rity und by ming ate the allevative acomny fr the (Varentsian ether relativity for example), at by explain the resale Laverence Sklar ways aschectv ao somehow tbe aco foro te basso they wich po oe at ever propo and which Tcl even now tegintoconanct. TT prinitvens is tobe Wend wth lementarty in Richest sense then eran ith dec sprees Babi 1 ik Tice dfn interme laine varine with expect toa fret atl interpreting” mest be spent 0 ed rele arance wh spc to he retary of teretatons which me fawimmin ar lie hee snot ofthe experiment re tonr"The ear ay now bein se he pint of he ie of tere: sopesng tnt ven fo hepato arteritis on ihe thal chow sir t sree theory frmation ve 0 ‘Tet itnavor of tme erate, aced ante pear emp tone about he rng of feos we re ong cone. ‘May psp rears wl thie pan al he prenieldo tine the "thay dons oa trenching the sean ‘hich rerentyake t mrko the png ec need, Sins oh ind re wt have id: But Fal eee my moe {rer plop aguments shot is te dt Yo tine oe Spe te pene pln behind he pried diinton infomation, wich end ode tontine oeamitn oF the concrete pte we re eumebel ifcarpcte then re maton a iho rm ae talent pte, we sel prop sme hse athe term be theo hate platy omental en phy ston iindgpenfent a the scetance it theory. Natural the dened terms are raposed to hve jst th opposte ts he aos of he crave cect ten crn ye an at. We can he trav ths in enn ofthe pint she. Bt my be mae conve tent inrdae new sig term slr t soe ofthese phys {oes or latins owe ntoce dfn nd dd erm Toner ein the ms pestis mae nyt pe ustooabalay wnt sippsd ohne te ante ht wes he {Si conoctness or tcvrecnes wot prepping he hey torre the dation sve opal tone the ope ve ‘eingamenshie ral ya tx Sie edd ee scl y the tion snd spt tod oly the tei Station canard the ra he thee ‘question of esting the correctness ofa definition, although tis, of eurse, testable question whether the definition i "abn." sine the ext fence and uniqueness propositions necessary to egiimaie its «dfn tion are themselves expres in the pie tem alone, And since the primitive vocabulary exhausts those terms whose applicator rnonapplicabiity is desidable independent of aceptng the theory. the totality of empeal tests of the theory resies in drawing from i ts primitive consequences and testing ther, We are now in deep philsophil waters decd, and this wil hay be the lst we have to say about these matters. But we have eno, think to understnd at lst why oor mxmal Rabbi and his opponent ‘agree about a number of sues They aves as to which terme sre Drimitve and which defined inthe fmuzations, thet female Zations are ai In dhe stator they attbute to terms. They age thit their formulations are based onthe sume theory. They dlsgree only with regard to defitinal consequences oftheir theories, but they fagiee that these differences ate simply the result of having choyen ierng definitions for some terms, and that these diferences atin 0 way a mark oF empirical diagrement ©. A Formalization Diferng Only inthe Attibution a! Defsitional ‘States to Consequences Letus now return oor concrete case. We ate considering the options salable to win the sation a which, from the general elaine pont ‘of view, we believe that space-time i glbaly conformal to Minkowski space-time but not fat Physially this means we believe that wheess Robi aioms for causal eonnectility hol pes selativity does nt, Iban it fils om the “ater” sie. Free pats, i os and lcs ‘ho no, we Believe, comespnd in thee havior tothe geometry cone structed using haht (or eal connect or alter a Rob's manner as {hey woul if spec rlatsty were core. Here is another option we could eke informal our bel Option C: We adopt a frmalation tit ares with the minal Reb- bam about the print or dined stator ofthe tes Halo has eae the same consequenes, primitive and defo, as dhe main obs Fan formalization. Bit it allocates dfntioal and postationl sats to some af these comeguence nw ailleent ay. In this oration ve woul ge, fw exam, with the nl ta Lawrence Sklar Robin tat fice parties do not travel inert lines, tat rgd ads do ot measire spat intervals, and that cocks do not mearare timelike fotervals. But the formalization wiht declare tat wht defines an inertial line i stl Bxed by that consequence that relates the behavior af free paticles to inertia Hines, And the formalization would then decae that ‘what dhe maxinal Robin takes tobe «easel dfintion of inertial lines ‘postulate about the rltion of inertial ine structure to causal structure ‘As one might imagine, the proponent ofthis formalization wll usualy say he is most certainly ormalizing the same theory that the maximal Robbin formalising. He seven using the same words, and he believes the same terms to have thee eritoris of appli independent of a ceptng the theory. But he say, since his definitions are the maximal obbisn's postulates, and vie vera, he diagreing withthe maximal bb about the meaning of at leat some ofthe defined tems ‘Why would anyone prefer sich «formalization to the maximal Rab- bin’? tn thi case, I dou that anyone would, ut later we shall see 2 case in which just such a preference i plausible. The proponent of tht Formation might argue that hiss preferable othe maaitnal Robbans simply fsciise the meanings he gives othe defined terms ae closer to ‘what they ordinary mean than ae the meanings the maximal Rbbisn ves to them He might argue like this: “Our ordinary meaning of inertial lin’ that 1s something fed by the motion of fee putes. And our ordinary mcanings of spatial and temporal ‘intervals that they ate given by eods tnd cocks, The maximal Robbianfostead defines these quantities a ‘sl way. Now do net sa that he cenna so deine these terms. Nor do lain tha there is anything physially wrong with the theory he for rmalizes in the peclar way tha results when one gives sich novel mean- fogs to the ters. am only saying that my definitions, my meanings are dover to what the terms ‘mean all hn “This argument isnot very persuasive here. For while nay be “giving a novel meaning” to ‘inertial Ine’ to define causally as the maxinal Fobbian does, the meaning given this expression by the proponent of formalization © is peculiar as well For according to bit, i agroement with the suaimal Rabbian and in oppastion to Option B discsed abave ‘nd Option D discussed below, Fe parties donot travel neta les Now inertial ine” had an ondinary mang tw, T sppose, ah atic” Se the proponent of Option C5 allen Aiton of neta lin’ jst as peculiar int owe way 2 the ravine Robbins, As T note, we shall ater sees casein which just sucha move from the Robian cus formalization tan alternative seems more plas ie ‘Another possiblity i this: someone might agree with the maximal Robbian that ony causal connectbalty shouldbe taken a primitive. He ‘night, however, be reictant to assign “defiitional” or "postulational” Haus among the sentences of his formulation which svelved eems ter than the primitive tern, Ina sense be woud be uneling to assign ‘propositions in formalising the oe uniquely, ether of Suing meanings or ‘stating ats. Those who eschew the analytrynthte distinction would trgue inthis way, I believe "Nonetheless sch a thinker might stil agree with the maxinal Rabbian| that both formations have al the same “fica” consequences assuming, that, that they have the une praitve consequences. The ‘leu here wonld be of» formalization that, ike that of Option B. was “merely another way of formalzing the same theory the maximal Robbian formaliod,” superior in that by refusing to alloeate definitional vs. ps {ulation status to the propositions ofthe theory which iaolved vocab Tar other than the prinitive one, this formalization permite ws « more Aexble way of talking about theory change and 2 more realistic way of faking about meaning. Bascal this prion would lok ike this: “Afterall the real impor of| 4 theory is, indeed, ss abservational consequences. My formalization, like the saimal Robbins, hall the right observational consequences, snd these are cxptured in the primitive consequences that my formalza tion shares withthe maximal Robbin’. He, however, has taken the Further—and gratuitous and misleadngstep of alloatng “anal” and yeti" stats to the propositions of his theory which invalve noe Prive terms. Since thisar no empirical cansequence, he should’ t do (On this account, while important to demarcate the obseratinal inport of «theory, Further discrimination among its consequences by Imangna notes about thelr “analyte” or “synthe” status are pallens ‘fom the point of view of an adequate account of meaning, mislead Hut U shall not burden the rear with one more attack on the analicsynttic distinction, for U think we have more relevant Bs to fy, Laverence Sklar . A Formalzation Diving Only in One-Way DeBnitional Conse Suppose someone oere formalization of theory to acount fe the sace-tine ofthis wold tn which he, the maximal Robban, and the Droponent af Option B above all greed that this now frmaleaton fered from the earlier two i a primitive consequence. Then all would gre there wasn sense in ayng tal ths was ate formalization the ‘theory fomalzeddteratively by the maxinal Rabbian and the peo pment of Option B Instead they woul gre that his wa the formalin thon ofa new theory incompatible wth tht earlier foaled ‘Bat suppose the following pear station aor: «new formalization ‘propo. Let ell Option D. Option D: Here we are presente wth a formation, some of whose consequences ile fom those ofthe maximal Rabbian. But hi formal ‘tion al ders rom the maximal Hoa ia the status atribates to some terms in partculr, at ear one consequence of this new formalize ion i incompatible wth thove tat alow from the asia Rabbian's formalization, and it has the following interesting Feature: whereas the ‘masial Robbin delres this consequence to contain a defied term — etn, a et, ns venation the new oration, ih new tain ateibtion to terme, declares the consequence t9 be primi. Porter, lets suppote the warinal Rebian declares no comsequencet of ‘he now formation printive but incompatible wth the consequences afi owt, Let ar lnk a the station fit from the point of view of the mama oblian, aod then from the pint of view of the exponent a the new formalization. The minimal Robian sys: "Thi i jst ike the sitton Aseussed in Option B above. Tis new formalization difers rom mine ‘only in dfnitional consequence, for the ony consequences of "incom tile” with consequences drewsble from my formalization are the ‘ofuning whit take tae defined terms, So this ally anew forma ‘atom othe same theory at mine, excep ht this formulation ones Caress the theory dn a diffrent way. Hf Uhre are grounds tall for ‘homing song thse Frlizationy, they ar eet ie the grounds fo ‘hoosing between my maatmal Robbian formalization and Option shove. They are chokes of expression,” indent, gan, {hat Robb himself would have tolerated this mn) now let wee what the proponent of his ew Kraaiation ha sy: My formalization has consequence incompatible withthe dawn from the masa Reba formalization, Furthermore, 1 say dat these consequences, fame in what ake toe primitive temsalne, serve as {round for tevin the theory. They cota no terms how pict ‘or nonaplcbilty equres presapposing the tt ofthe theory. Cea ‘wht am proporagis nota new farmaiation tht merely 'eprne the theory formalized bythe matin Rosa proposing 4 new theory to acount forthe data compatible withthe theory formalized by the ‘nana Robbian ‘Why wou anyone delare that term taken by the maximal Robbian to be defined, realy «primitive term? This sub sue than ‘pets tobe st tight. Ove reson he might have for disgrecing with the masimal Robbian on this ive might be thi: he mit be king Primitive terms toe thowe wove applebity or nonapplcblty i Altrminable by “immediate cbneration,” and then saving there ate ‘more features ofthe world aces to immediate, totaly theoretically ‘mediated, tet than the maximal Rabbit allows, He might are, fr ‘example, thatthe maxinal Robbian is wrong in thinking that only Fs shout asa eoneceity are det accessible epistemiely. He wiht tert, for example, thal he san immediately tell wheter oF wot tw emeoincdent bt nearby rgd ade ae concn, jt a Ro asserts that we an imaediately tell hes oe of ts events nour consciousness sale the other, I that his line of reasoning, we are in for one oF ‘hone endless debate bout ust what faces to mmedlte sera tin, and about whether, wenn upon immediate aprehenstty a4 teri of primitienens, we ever cn constmt a selene that rea he vil of perception” and becomes tore hans lavike ptematizain of solipsitie conscoses ‘On theater hand, be might age Iie this: "Tdi tke private’ to snark ot determinable by mined inspection and 0 total independ ‘of thericing But fo have notion of primitivenesswbih inp tant in that # captures Tacs’ inthe following wey: propositions Framed etitely npr terns tat fat. They are toa oven. ‘hal Of eos tn sce sense convention that the words mena wa ‘hey mca, bt wh say that tere na frmalzation senitve, Vy indigo ton fable In hese ters ayer fae espst ft wyn iermsSn ean ‘ny rom the ee hy lay te Hey Laurence Sklar Whichever poston he takes about piv ters, he wil continue like this "While the masial Rbbian, for example, says tht fee pat sles do aot travel tneliegeodesls, 1s they do. While the proponent tf Option B als says they do, he sy he dsgreing with he maria Rabbian only as tothe meaning of timelike geodesic’ a erm whose meaning, fr beth of them. 1s fixed by the dieing deinitions in their ‘especie formalization. But Tsay we should take "nelle geodesic a 8 Pettiv tern in ur formalization: I disagree with the maximal Robbian ho fat ofthe word wheter Fee partes do o do nt realy travel timelike geudescs). Furthermore, wile the proponent of Option B ‘thought he was dsagresing with the maximal Robin only bot mean Ing forms he oo spropsina theory incompatible wth the msi Robbion’s The incompatihy, however, was dgused by his maken ttn of defined sats fo tine geodesic ™ ‘As we know, some would even deny the usefulness or inelligbity of ny sbvervationalnonabeervtional ot “Tata eanventona” distne ton llogether think we shall be able to comer them, witht ls | fontent a advcsting that al the terms of oration of «theory hoa propery be taken as primitive in or sense. E, Variations on Option D ‘The exponent of Option D Is most understandable when he takes the first linea reasoning we hve proposed for him; he beloves in "pare staervation languages”, he believer tn terms whose applicability or ‘nappy tou situation totaly hor adependent. He dnagrees ‘vith the masimal Robian and with the proponent f Option B oly abot ‘which terms are realy nthe “pre observation langue “The exponent who takes the other option lesly has his work et out for hi. Jost what his theory of mening and ust how do words get er meaning according tot? How does he rationalne fil ditinton Intween primitive and dened terms na formalization ihe ca pk ont the former by their “pure observational states?” Pehape le as sme Ik of theory in mind which, ike tha impli in the Resenbachian notion of "elementary fat noted above, reste upon a view ofthe role | Sntcedent theories nd theories i ctestining tot believing in fing the meaning ofthe tre in the theory we ae a mater of ct ‘roping shall return to these question, fot anewer them Bat bt Ie nts here afew ofthe less that rie when one reno th White shall nt try to develop the tory of meaning cee by h sition, shal try to say something aout another problem fice by the India who tes soc ln, "This individual proposes that his fszatin i hated on a theory incompatible with tat of the maximal Rabin, Bt he aces tat bath is formation andthe maximal Rabbia’s ive rset all the sme onsequenees about cal connetbity. Now he might consider that bhi owen wrong im hiking tat the fats abot cal conmecity were somehow “immediatly scene to obseration ina totaly theory. Independent way.” We sal, ed, eaplore this iste later in some deta Bat forthe moment suppose he grants that atleast thse prop ‘stios are open to immediate empl determination. Sil he sa, 1 take onrtheores abe incompatible cae they confit on other pro ‘tons which, wile | adit them toe wt direlly testable empl ‘ith prespposing the corecness af the theory Isl take then, Unlike the mana Robbin, tobe facial and not in any Sense matter of conventional choice about the meaning of words But, one inmedately aks, ow could one posbly decide which theory was correct? For example, how could one decide whether or not free panicle fellow tele goad, sven that this individ pro nesta this is afartual question notin ay way deftinal oe and et tel we cmnot dees correctness by ny “ented era” ‘news? Of couec ovr mazimal Rein bas wo problems ere for be sy hse are realy jst diferent formations aoe athe Se theo. Hot oar india contin deny this ‘One thinge ould do sot for septic: I dot relly belive fee pastes travel timelike geodesics, vale the mal Rabie, who says they do not IIT wanted to do wae demonstrate oy that here were tser theories hesder maximal Robbins that were te ell pose ‘shteraional tf mata Robbins) and yt incmpatle with Inavinal Riis, Actually I believe you haven the fitest rete gan over the other of these theses abot fre parties, OF cunye only ane af these theses canbe tive, but you simply can ever Kw which.” Skeptic. “Sather tig he could dos ptf Pineréan conventional, what. ser thi fy He sonld ay "Nowe weave tomate thea with i hse pumsle one wala onsen Nema et could {dine pede Gr ar eth intentionally for oo Lawrence Sklar choose which toe, We might even 2050 frat sy “Truth just 2 Imate of convention.” Nate, however, that Lam ot aig at oar Choice is merely a matter a chosing how we talk That what Ed- tlingon, Selick, and Reichenbach belive. Lm claiming ie choice tuvong comport, but expisilyindseviminale, bli.” ther postion he could opt rsa version pros Inchon which theory to adopt we must aay se, above and beyond the empi- {al evidence in voor agaist a theory, a nt «pion plasty of the theory. Without such elevates Intense tothe thoory in qe tionand ndependontof the empinal evidence for r agaist, theoretical ‘decslon-making could never get ander way. So even if to incompatible "cet theares have ll the sae observational consequences, thete ray stile od ean, onan a prio bass, to adopt one rather than the ster. ‘Or he might aset thatthe choice of theory is abways to some degree notte “continuity wth previous theory” ad that eonsiderations ‘this Kind could motte bis choice le ght "Suppose Thave two incompatible space tne theories that ae obvewatonally ndings ‘le: One ofthese theories ay sil be more “continuous withthe older theory replaces Tat, nas lee change nour bel to drop the old theory for this new alternative than Sor the other ternative. Tn {his crcmstance ite more remonable to elove the nee hypothes, shih is snimaly diferent rom our older rejected theory, than i i to believe the alternative more Tada theory Teh been alleged, for exanple, that even fone reat realy incompatle wit the Reichechlsn altemative of fat space-time and tives force, and even these two alteratve are obserationaly| Inditngsshable, we should sil opt for general relat since ‘more contiouoss wih" or “a more conservative change from” the oer theories of gravitation. Since this losphi chim hs een made with some frequency, s might be worthwhile to comment on it here “Thote are, think, wo probs with this approach (0) Fint fal cach ofthe new alternatives cab and ally wil he mote conservative” eth respect to the antecedent Hy in same spect or other. How can we decide which ay of eine conserve Should take precedence otherwise "weight the al of beng nae tive in some partelarrespeet? For example general relat no Camservative than the universal Force hyp btha the farmer case, clocks and rods at dierent spuce-tine lotions are congrt if they ae congruent when brought into colneence. And ths the way things were in Newtonian and spel relatisi physics. But in Newtonian psc, ad perhaps i what would have Been @ “natural” theory of gravitation from the special relatsti poi of wew, particles acted upon by grit dona rave ine geodesien. They are fot “Tree” but “forced.” This remaing true fn the univers force Iypotiesis, be fal to be tue in general ett. Crdly, the univer salforce hypothe more conservative than general reathity tat like Newtonian mechanis and special relativity, Is flat space-time Mhory, whereas general relatifs not ‘Which way of being conservative with reapect tothe older theory it more eaportant? Why? (@) A secon difily withthe approach i this: we ane told to he comenvative as possible with respec to antecedent theory Bat suppose, ‘when we have considered the options availabe tou fo he new theory. ‘veo back oer the older theory ani lize that to could have taken nother form abseratinaly ntngshale fom the fem it had. We Imig end up with this view one option for our new Uhary it mst conservative with resect to antecedent dhory att was. But the other ‘option inmost conservative with respect to anaes theory we could have Tid carer, which ebservationalyIndstingnishable from the older theory we did have Suppose for example, we decide that he universal force theory s most ‘conservative with rexpest to Newtonian mechanics because tke New tonian mechanics, Ht At spacetine theory. But then we discover Telleting on general slaty, something ike Trautman’ curved space tine version of Newtonian mechanic. And uppose general rela 5 ‘ort conservative with eect to His now ese antecedent theory! ‘The indvidal whoxe “conservative” doctrine we ae now examining cowl avoid this secon problem only by becoming more conserve. He ‘would hive to sy" aow rie that all long there was an ateraative ‘cory that was observational indistingishabl from my older theory Hot I bold to the allowing conservative maxims you believe Inpatients, there sno gal reason to drop your ele jst beense om nce tha there aor hype which just od a, ba 90 Inter than then yon Lane ts trope oan ay taking ther ory, ih atl hal ae eerie ty Loverence Sklar to thse nowy discord stomatins tot anda being the comet Statin pit comeatie mies theo: Tut anther problem ae is onion inthe pres content. We sayy ht Tat ad cone ln the gent ey, tol coset wo! hve bee ote nm droning He New Cin ad epg with era toy. Bath, en that we seep gene lay, ight i the tion ar, eee tre genral cnt, omsert thatthe her tha sl ve been ieidn the nner cnt ws Tetman coed spce-tine New toi mck and et Newt at scene hoy Te ng tobe omer th eet pyr a tw, but le ante © te-comeratve wl expec yt which, we me se, wold hare Seon art iad ne ben Tes comider ast option ond ho tl that tees any be compat ever tonal tinge, ad who Santo teller whch fick sterathe tre we shold at So Dose he es eth "Tbtee ht Rabb so ir as cme Bi hl: Dt Ta wll are that ht be wong sot he: Now Bel that fe ures donot tel Rtn inlets fr tel hat the ater camo pc ltd ot a Soper tl nt tas tay tomy reset ven, Rabon ald. What woud 1 82 The oy hey tune sale to apy that ce {erally te gene ely Habs Fegan ft Sly bol athe wo ut fe pres dost evel nie eon there neste try. Terre king tad he Fonte nor now emis ad conterng the pete Five nny repeat dsr the wr age lis hee treo computa thcone these scatman taal nels the nmin! Rat, onic pre dant vel nel fates te tere verso fener ety ste ts worl ‘Schall to Minow sae meat whch a Ins helt ie pts do tel tine eas Wha shold hater Tay {shold believe the later. Why? Reo contrary toy present eli, Rai' tn do ot ol 1 ‘Srl che move erly denne then fl eaten wd ot bly contralto Mink Sprectime tn hort tanto cl a th a 1 discoves could still belive, for example, that free particles trvel timelike ees. IT adopted the maximal Robin theory, however and then iscoveved tht Rob's aoms do not hold, I woul need to change ‘heoretil belies more radially, "Tam assuming the world is one ofthe possible general relativistic wold, Which one 21 lieve i oe a which Rabb’ asoms hd. But I might he wrong aon thi. So, given that theliee tat they do hol, I should elev that theory which sane of spectra of posible theories Ihave im mind, one of which wll hold even some of my preseat assumptions fa sh fo believe a theory in which Robb’ astoms old. Bt which of the many incompatible theories in which they held shou | belive sven that I think tere are many that ae observational indstingulh. thle? I should believe that theory which (1) ea speslzaton of general ‘lass of theoties, oe of whone members I would adopt HT no longer believed that Rob's axioms do hol and) selected because i the te momber of this cas im which Rob's aims do ol and which compatible withthe ebuervationl date ow the sof theres have in ids the clas of general elativis ticmodel ofthe work. The theory in this las fm which Rabb’ aims do hod one in which free patel travel tineie genesis, dre measure spatial intervals, and docks measire timelike fates; ce sven our data, the theory the genera telatitc lel f world tlabaly conformal to Minkowski space-time. Now I believe the world tot flat. $0 1 believe the fee parties all not travel the timelike sees ofthe macinal Robban theory. Therefore {should ehoote the theory that says fee particles do travel timelike gees, an which the maximal Rabbian's timelike geodesics are not really timelike sees at all T should not coo the maxoal Rebs’ tery, in which is timelhe geodesic are really neice geodesics bot fee part es jst dont travel them. "make this choice no onthe bass of my abserstion, fr the il "abian’s theory and nine pet he sane obseratinal eonsejuenees ater, I make this choice brcase my theory fa spec instance of a sso tar hear, ome a hich Hold adapt even if found ut, tsintay toy pestle ta ne bls hypteses abn exe sonnet x ag Hh manna Rls theory ds ot ma rally at eae hon sta ha so alae oe 2H Laseronce Sklar even if were to ind ou that sme of my present assumptions sbout the srl ace wrong “This inva s saying: theories can be genney ncompuible even 1 abservtionaly ndstingushabe. Bu Lay stl have grounds or pe fering one ove the ter The grands are not a pio plausibly, nor ven conformity with antecedent theory. They are rather the natural or ‘tonal role which the theory plays the Light my assumptions out the other possible theories the world which might hold—although Ido tt Believe they do~and which [wold adopt fund out that some of ny asemplions about the world re false ‘Once ashn, Ht asserted that theoreti deeision-making depends ‘pon some assumptions we make about the spectrum of possible theories tree likely to consider, and at jst upon the “ebservatonal fics.” We Shall ea to this pont in some dtl when we consider frmalizations ‘general relay ater this paper. F. A Fonnalization Dilfering Only i dhe Status Attributed to Terms 1 noted ears four ways in whi Femalztions can fer fom one sother, but 1 ave dsessed only three alternatives tthe masinal Fabs formalization, Uhave ved the fart, Option E, lls, prima- rly beeae itis the ane option which, tomy knowledge, no one has yet sed. Option B would be espoosed by Eddington, Sehlick, and Feichenbach, I bulieve, on the grounds tht Bs “deserptvely simph tha the mania Robbin formalization. Option © would be espoused by Someone who thinks the consequences of maximal Robbianis perfectly {MVrightasthey stand, but who as diferent predilection for attrbuting ‘letitionl or postulational status Yo them. & Quine, for example, who ‘vshes to eschew such narginal status attbuton altogether woul apt 2 Nerion of Option C- Some version of Option D would probably be ‘spouse by someone who, looking atthe dats iy the hyht of gener Felatvsticpnsiites, and disagreing with Rob's insistence tat only fausalconneeiity should count a «primitive nation, wou interpret the date to mean thatthe space-time af the word, although eohaly conformal to Minkowski space-time, as realy not at tall. Tere is Option E, Choose a formation exactly ike the sna Robbin, except that dillevent terns are designe pinative” ana i ferent terms are designated definitional Thon Iw dein tw wap ther (me ih take cna sonmectity a feito iste primitive; oF (2) one might tke some additonal terms aver and above “eaaly comnecble’ ae primitive. Since the latter is the more likely move, let me discs what might motivate ‘Someone making this move might argue: "obi wrong to take causal conoectblity at the only observational (oF, peshaps, factual feature ofthe world captured by a theory af space-time. For example, cong fence for spatial interval at «distance i also observational actu) a 6 ‘beingan inertia ine. But the maximal Rabbis e correct in thinking that the space-time of the word is Hat, that fee particles donot travel neta lines, ete What the maximal Roian says about the spacetime of the world and its connection wth material objects is correct: hei only wrong In thinking that some ofthese eansequences of is formalztion are mat ters af dfntion when they are really matters of ebservaton (or al 11 is no surprise that no one has ever alfred an approach like this Who, fced wth empiri facts that would normally lead general rl wrt o say tat spacetone was nonfat but glablly conformal to Min ows! space-time, woud everasert thatthe spuce-tine ws al and not ‘va matter of"deinition” or convention”. nt heause we coal ‘we wished, reformulate the general relativistic theory as a lat spacetime theory, but because we coukl observe that this was relly the cate? Or, ‘ven if we couldnt decide this beratonally, becuse there wer ober eauons for thinking tis Robbin account hath tly incompatible with ‘he general relativistic acount and factually corte? 1 Shall forego consideration ofthe other approach, with its dena of primitive stats to ented connectiblty, antl later, when we shall ‘examine that approach in another contest. 6. Summary of Part 1V ‘Let me summarize what hasbeen ging on inthe lst few sections. have been exploring the desiabilty or undestabilty ofthe Robbian Aetniton of such notions as inertial ine (neice geodestes) and spatial and tempor congruence, by subjecting Rabb's theory to sess. Mate inal stress consists sayin suppose Robh's anim for causal connect ‘ty do not even hold, Under tht tes his method of definition cleay Tweaks dw at then Ive Het ue sbjct Robhie thenry to less, fil toxone sites, Lt wsatnee we below that his axis of ae ssomestaity hk, but tht dhe postulates tht would supplement his Ahoy peel eltity were cortet back dna That, fet ame Laserence SHor we below that fee pte do ot trvel Robbian inertia ines, rg Fo do not measire Rebibiun spat congrience, and clocks donot meas- tre Hoban temporal congreace ‘lying hisses to bh theory, we ave von tha the allowing tions canbe delineated we could hod toa maximal Robbin formal ton io which Rebb's axons for causal connect are retuned itu, svar his deiitons. We woul then have to abandon the postulates Uthich woul supplement Hobbs orga theory and tell ws how free utiles, rods ad lacks behave relative othe spacetime structs as ‘etned by Rodd, ‘We cold adopt an erative formation in which the usual onnec- tion between fee parce, rods, and clocks and pace-tine stroctres fetuned at these tine retaining Robb’ axons or eas connect ity in oping his debits oferta ie and of patil nd temporal ongrmence, We sould muna tht this wae merely «dierent malic ‘ton ofthe same theory a the formation fist scribed, but that i fexpressed this theory in dierent, and preferable, way of speaking ‘We could adopt formalization in which all the comeqvences of the nasal Robbianformaltion were retained tact and in which the ‘me stite—primiive or defined ae atte t the terms the ‘mainal Robben formalization, but in which the consequences desig ‘ated "fnition” and “postulates” dlfleed from hore a estanated by the maximal Robben formalization. We coud manta tht this was formalztion of he sme theory stat fora by the minal Rab jan formation, but argu that frmlioed the theory in sway that fame “more untal meaning” to some ofthe dened terms. Or we could ‘opt a formation excl ike the rasa Robin's exept that init tre refine altogether tosh “in the mangos” deinitonal r portly ats to any af the consequences ‘Anatherallemative would be to adopt a Frmalzation ke the ove eseribed two paragraphs above and dec hat this was the formal tion os they diferent from thot Bint described, For, we cou a (ai, Rolo was ron in thinking that his "tints" of seh things | inet Kine and comarca itera ell were definitions they con he wed ax erp testable consengece of the theory Sa fash checein, ‘We could adept the second Kin of frmaliation dices a en, mana hat Grail a rt try tha thal Krai ‘by the masinal Robbian. We could ada thi there wat no servation sean of telling wheter the maximal Robbin or thialternative theory ‘wa correct, bat oon to declare tht there could be geniney incor {ble theories that were obserntinaly ndstingishable, We woul sp plement this approach by offering. one hopes, an accunt of how a term {auld have a meaningin a theory witha having “rly servations Ccteria of applicability, and an acnt of ow one ould rationally 20 ‘tout deciding just which of eo incompable batobseretionally inde ‘imieabe theories was comet. In any cae we woud either provide this last supplementary account or admit ounces skeptics or conven ‘nalts with regard to theo. Finally, we coud ake the view tha al the consequences of the man- mal Robbian formalization were correct, but thit the mai Roba Sormalization wae mieading in tht by incorrectly estrting rive status to casa connect, mitepresented some of ts tr obser tonal or fetal eonaequences as having mere dfntonal tat, Y. Alternative Formalizations of Minkowski Space-time A. The Prt We have subjected Rob's theory to vans stresses, and we have seen how it alls up. When theses so seat that his pastes or an connect break down, all would agree, I believe, that his mete ations must go by the board 8 well: But, a we have seen, 2 weaker stress, inrlcen to contd any of iam fr cael connects, but rficent to place stern the addtional portlstes connecting the Ihavior of mater ejects tothe space-time cure, c ent many options. Some ofthese ar masial Robbin, rtalning his metic defini tons intact, but others lead us to reject the Robbian formaliation of « space-time theory to greater or lesser degre. Even when the Ren theory i aubjet tno aervational stress whatever, there may stil be dispute a to whether the Roba formalization of space-time i core, sd, fore” whether itis est. Hobbs theory & now unaly thought of as one way of formalin spect atiity, even f Hob would not qe have viewed that wi {tus spe we belive special rlatety to be correct. Then we be- Ive that bb's hase aoa feats conectiiity had, and atthe ter states we wun ray ae a nalent Ra's theory Lawrence Sklar hold as wel Le, rg ods comecty measure spatial congruences a deine by Robb, clocks corel determine Rabbian trporl cong ‘ences and fre pts fallow Robbin inertial lies. If we do belive ths, are we then precided from adopting any of the options dss the sections above? We shall se tht lof these option a sil waa to us, each option diving from the obhian aiomatztion st before But some af the epson wll no Longer have the “plasty” they had boot, inthe sese thatthe ground fr adapting ther wll longer be svalable: On te othe band, some options, peevioly unmotivated and Unlikely ever to be espovse wll now become plane alternatives 'As before, Option Ais the Rabin, The only primitive term ix “e- sally connectibl,” and the aiomsfor tare Robs. The metic notion and {he notion of inertial line are defined asin Robb, For our complete formalization Robb’ theory i supplemented with mater ptltes that talus that rigid rods, clocks, and fre partes have the ws spec relativistic conection withthe underping space-time Featres. In Option B we would til fake casa connect sour oly print tive notion, andthe clas fcoseences of the formaliation we take a= Primitive would remain the same, But we would fe diferent definitions forthe metric nations ad fr the ation of inertial ine. What woul this mount to? Jat as oor maximal Robbian open amounted ear to ‘opting «ft space-tie formation of theory which in general felatistc tem, we would tke tobe «theory aa noni spacetine, ‘Mpg Option 8 would amount to proporing curved spcetne fne ‘mallation what we would tke, ram general elite pout ae, to ea theory of «Hat spacetine Eater wes thal one ght shoot Option B onthe grounds that it was “descriptively sinpler” than the masinal Rebbian formalization, Siting in th the way we wo lk othe general relativistic context, ‘he marinalHabbin formation did ot Btn thse ss ear Hat the Moblin correlations af the behavior of mater with the space-time structure ar, from the point of view of gener rely, the ih ere Tatons; and thats ither formalization is deveiptively simpler isthe Rabon ‘Option C remsine quite viable, Here we adopt frazation th sees with Rabb’ in counting only ssa same as pit haa the same consequences sv Ros oration. Ht 1 ei) allocates dfintonal al pont 20 ‘na manner deen fom Robs, of (eschews the marginal annotation sf pentuatioal and defnitinal state to proportions containing nom Deinitve tera ligt. 1 proponent of allerative () might arm like thi: “While all the consequences ofthe Rebbian formalization ar tac, is allction of ‘lefetiona and postlationa stats to them untae to the manning of the deine terms Is te that ee pales travel inertial ines Bat that swhat we mean by Inert line esas tr that neta ie ave the asocation with cus connectiilty which the Rabb friction ports. But thats afar aboot he wort, not proposition ve by the ‘meaning of nertal line” So Robh’s consequences ar all tv, but he {ake some ob tue ar a mater of fat that re realy tre by definition, td ein cores” “The exponent of alternate a) las a notion of meaning—or at least ths he dacs tha allows hin oll wha irellwdefiition and what ‘really postulate. The exponent of steoative (dink tht Rob and the exponent af alternative (a are equally confused, For, he alleges thre 1s simply no goo reason to label consequences postalational or dein ‘inal tall. Remember that hei til allowing that we cam determine which consequences ofa formation are “observatnal"and he sake Ing eiference among these consequence inde clet divergence of theory But beyond that he believes ite ftuos to pretend tht he ropostions af the theory even nomprimiive terms cn ay impor tant way be characterized ax defntonal or portation. nso a he onprimiive terms arden” bythe theory, he sys, the fining i spread throughout the theory a «whole. ‘Option D is again one that noone very key to capo, Like Option 1 iotee formalist that ders frm Robb’ in sone ofthe oop tive onsequences. Ths, ike Option B, i amounts to adopting a curved spcetine formalization for what we wold ordinary take tobe a theory ‘ala space tine word. Now whereas Option B tock his tbe ater ‘merely “rewordng” Rob's theory. Option D takes the neve formal thn the that theory sacompatibe with Rob's Choosing Option D smnts to dedi thi space-tine relly x curved In one alterative (Opts D cai that thi tes ebnervationally determinable et, in smth, ha, wh fat thatthe pace ved rather han at ‘Sime toaleratina daconery, testi tl andy mo mee ‘kta ater Loverence Sklar Eater we sa that this option waa matral one to take inthe ase in sshich spac-tne wa, from the gener relative polt of ie, euved Tat lotally confor to Minowskt space-time. We saw that consid rations ofthe gener thre contest of general relataty might very tell ead one to say thit the masimal Roan way of frmalzing the {Bory of space-time was not st milading with respet Lo meanings, tut simply wrong. And ifthe mail Robbin would argue da the to formalzations dd not realy fer in primitive consequences, the expo- rent of Option D arged tht they di for he argued that the axial Tobin's version of what was primitive wat to mare But nom, given that we belive thatthe word from the general relativistic point of ew, at, who would be likey to are for carved {pacetime a theory incompatible with the Ha pact tory, Bat forret? IF ane were to blive that Option D and the Robian ormaira- tion really re formalization of incompatible theres, he would be more than likely to declare the Robbin alternative correct, don the tame rounds on which the proposent of Option D argued eater for the frrectnes of his theory s opposed o tht ofthe mail Rian. "The proponent of Option E argues fora frmalzaton that difers fom the Robbin in no consequences: Bute dagrees with Robb about the ‘cope of primitive concep: Let us suppose, as we did eae, tht he Ailes fram Rob only in wanting more concepts as primitive, a let us ‘en postpone arguments about the primivenes of casa conectib {y. The proponent of Option E would then argue that, whereas Robs formalization bse on 3 corre hoor of space-time, ts misleading in that fils decaes some of consequence to be nonobsewvational ot nofctoal. He would say: “I, Ike Rabb, prefer the consequences of Option A to thove of Option B. Bt wherst the Robbin and the pro- ‘pment of Option B agree tat that two frmalizations are Based on the “ime theory, Fake I that they formalize incompatible theories. So the ‘hoe of the Robbian formliation over that of Option 8 8 oe of fac, not of mere expression. Like the proponent of Option D 1 consider it aca ater whether spacetine or ipa lt. But whereas he opts for sn implaasble curve spacetime pietr, opt for the Hat.” B, Can Definitions Be “Right” oF “Wrong”? Letus bok more closely a theft lleratve verso of Option Can at Option E. Both of therm Find the Haba fat deertve The lyst version of Option C finds Robb’ formalization ujusto the mean ‘ofword, Option finds Robb'sformalrtion vn tothe quent {5 “actual” a opposed t "detonal Gansder the following argument: “In the Robbian formalization the notion of congruence for spatial and temporal Interels defined by ‘ean of causal eonnectily. So the connection between the notion of congruence nd the cna notion used to define can hardy be sd to be ‘one orincoret. These, ikea, definitions, inorder toe acceptable, rust mect certain formal roqurements of existence and wniqueness of Towaertnd the EPS formation we shal rst have tlk ome oF the mathemati aspects of general teltistiespacethnes, Te fill teudo-Remannian metic tanild of general relatity i» compen Structure. From swe can abate sme spl tract. rt ofall the mand opal space Second ile Wil manifold. Next, these spacetimes Inve conformal struct Iofsitesinally this means the pss fcostructing the hyperplanes tthoginl to a sven Hine element. Sach sconatrtion resus in the intemal nall-cone stractre, 4. im the distinctions among timelike, rll and pclke connected events or phystlly, i thecal cone fly strc onthe space-time “The spacetimes also have a projectce sete. tnitesimaly this means that there «well defined notion of pale anspor ti ‘own diction, vector rot tangent apace to tangent space. Cloaly this Tessin geodesic structure onthe space ime. For the geodesics ne fot those coves whove tangent vectors ren fngent vectors when they ace pale tinported thelr ow diets “The spacetimes have an ane structure. nfitesinally this means tt there navel deine ton of parle transporting, inany dretion {vector from tangent cet tangent pace. Cblly ths ress the ‘Anability of an eine parmeter slong the geodesic. Equidstant ‘int longa geoesi ith respect to the affine parametee) are those ‘oints whose "connection vectors” ae pall lly thee i the fall ete strctre, which allow us to compare the interval teen abitrary par fpf pot events EPS asumes the mins of vent, ight ray a pte. Panties are nea, intuitively, to be ath of “ree” partcs. Axioms ae intduced onceraing the stones ofthe partis and of menage sent between them by ht ras An nn intrduced which allows to eoordntize the space by particles and ight ry, and armed that any to sue ‘ornate are eno elated to each other, This lets die & ‘ferent! topgy forthe mand of events, There are sme problems Heres but I shal discus them later. Next, ome om the Bight ay ate Intvatced that ae scent to allow ws to deine the sua ight cone Structure an the spacetime, ke, to defies conformal tte. Then {oma he particles are introduced that ate sficent o let ws define the projective statue of dhe space tine ‘Ba we sil dot have fall aie stractre fr the spacetine, mich lest ill metic stactare, EPS shows that ony oe aditinal on i ‘neces tonic the existence ofthe ill ine staetare, Tis an ism af compat Between the conformal and projective structures tes that the terior the ight cones, dial sv led withthe patil. That i, we ean bythe eo

You might also like