Leibniz Equivalence.

On Leibniz's (Bad) Inuence on the

Logical Empiricist Interpretation of General Relativity
Marco Giovanelli

Universität Tübingen. Liststr. 17 D-70180 Stuttgart

Abstract

Einstein's point-coincidence argument as a response to the hole argument is usually

considered as an expression of Leibniz equivalence, a restatement of indiscernibility

in the sense of Leibniz. Through a historical-critical analysis of Logical Empiricists'

interpretation of General Relativity, the paper attempts to show that this labeling is

misleading. Logical Empiricists tried explicitly to understand the point-coincidence ar-

gument as an indiscernibility argument of the Leibnizian kind, such as those formulated

in the 19th century debate about geometry, by authors such as Poincaré, Helmholtz

or Hausdor. However, they clearly failed to give a plausible account of General Rel-

ativity. Thus the point-coincidence/hole argument cannot be interpreted as Leibnizian

indiscernibility argument, but must be considered as an indiscernibility argument of a

new kind. Weyl's analysis of Leibniz's and Einstein's indiscernibility arguments is used

to support this claim.

Keywords: Gottfried Wilhelm Leibniz; Logical Empiricism; Philosophical

interpretations of General Relativity; Point-coincindence argument; Hole Argument;

Indiscernibility arguments; Felix Hausdor; Hermann Weyl

1. Introduction

It has become commonplace in the literature to argue that Einstein's celebrated ar-

gument for general covariance, the so called point-coincidence argument (well-known

from Einstein's review of his general theory of relativity; CPAE 6, Doc. 30; Die Grund-
lage der allgemeinen Relativitätstheorie, 1916), if considered as a response to the also

celebrated hole argument (known from Einstein's private correspondence with Michele

Besso, Paul Ehrenfest, Hendrik Lorentz and others; CPAE 8a, Doc. 173, 178, 180, 183;

see Norton 1984), would amount to a defense of what has been famously labeled Leibniz

equivalence (Earman and Norton, 1987).

According to this line of thought, Einstein's point-coincidence/hole argument would

clearly resemble Leibniz's indiscernibility arguments against Newtonian absolute space.

In particular the role played by what we now call dieomorphisms (transformations that

preserve only smoothness and the uniqueness of the coordinates) in Einstein's argument

would recall that of translations (which preserve also the distance between any pair

of points) in Leibniz's shift argument against Clarke. In both cases worlds which can

be transformed into each other by suitable transformations might be considered the

Preprint submitted to Studies in History and Philosophy of Science March 16, 2011
same world. John Earman and John D. Norton drew the parallel in detail in their

seminal contribution (Earman and Norton, 1987) and the protagonists of the huge debate

created by this paper, such as Tim Maudlin (1988), Jeremy Buttereld (1989), John

Stachel (1993), Robert Rynasiewicz (1994), Carl Hoefer (1996) and Simon Saunders

(2002), endorse the shift argument/hole argument analogy, or at least adopt it as a

polemical target. From this perspective the hole argument should be considered as a

New Leibnizian Argument (Bartels, 1994; Bartels, 1996).

In this article I will try to show that reference to Leibniz in this context is in many

respects confusing. It is not a question so much of historical accuracy, but rather of

obscuring the radical novelty of Einstein's indiscernibility arguments in comparison to

those of Leibniz. I will try to bring out this point through what I may dare to call an

historical-critical reconstruction of the role of Leibniz's indiscernibility arguments in the

Logical Empiricist interpretation of General Relativity. Logical Empiricists actually tried

explicitly to understand the point-coincidence argument as an indiscernibility argument

of the Leibnizian kind, encouraged by the use of such arguments in the 19th century

philosophical debate about geometry. However, they clearly failed to give a plausible

account of General Relativity (Friedman, 1983; Ryckman, 1992; Howard, 1999). An

analysis of the reasons for and the origins of this failure should show, in my opinion,

that the comparison between Leibniz's indiscernibility arguments and Einstein's point-

coincidence/hole argument is misleading.

I will proceed as follows: developing an idea of Hermann Weyl's, I will argue that

Leibniz's celebrated thought experiments on the impossibility of noticing a universal di-

lation of the whole universe or its mirroring by changing east into west and so on, can

be considered as the rst attempt to dene the modern concept of  automorphism or

symmetry transformation, a structure-preserving transformation of space into itself, a

way of mapping the object onto itself while preserving all of its structure. In my opinion

Weyl's suggestion, although surely questionable from a strict philological point of view,

is nevertheless helpful in the attempt to grasp the theoretical signicance of Leibniz's

indiscernibility arguments and most of all to understand the role that Leibnizian ar-

guments played in the 19th century debate about geometry. The ma jor protagonists

of that debate, such as Hermann von Helmholtz, Henri Poincaré, but also Felix Haus-

dor (to whom I will mostly refer), in considering automorphisms (or isomorphisms)

that preserve progressively weaker levels of geometrical structure, were able to generalize

Leibniz's thought experiments, showing that two worlds would be indistinguishable even

if they were mapped onto each other by any continuous deformation whatsoever.

Early Logical Empiricists (Moritz Schlick, Hans Reichenbach, Rudolf Carnap) explic-

itly interpreted General Relativity in the light of such Leibnizian kinds of arguments.

In particular they considered Einstein's  point-coincidence argument as an expression

of indiscernibility in the sense of Leibniz. I will suggest that the inadequacy of Logical

Empiricists' interpretation of General Relativity, which is now commonplace in the litera-

ture, depends exactly upon their misleading interpretation of Einstein's point-coincidence

argument as one of the many Leibniz-style indiscernibility arguments that appeared in

19th century debate about geometry.

The lesson that, in my opinion, we can draw from such a reconstruction is that, even

if the point-coincidence argument as a response to the hole argument can be consid-

ered a sort of indiscernibility argument, it cannot, however, be interpreted simply as a

restatement of indiscernibility in the sense of Leibniz. Interestingly enough, as we shall

2
see, it was precisely Weyl himself who already clearly recognized this point. In a rst

approximation, we can say that whereas Leibniz's indiscernibility is an expression of

apparent physical dierences that cannot nd expression in the mathematical apparatus

of the theory, what we may call Einstein's indiscernibility is the consequence of appar-

ent mathematical dierences that cannot nd any correspondence in physical reality. In

an appendix I will try to provide a formal elucidation of this point.

2. Leibniz's Indiscernibility Arguments and his Conception of Geometry

Hermann Weyl was perhaps the rst to emphasize the depth of the  philosophical

twist (Weyl, 1952, p. 127) that Leibniz gave to the simple geometrical notion of  simili-

tude or similarity between two gures:  Leibnitz [sic ] declared: two gures are similar

or equivalent if they cannot be distinguished from each other when each is considered by

itself, because they have every imaginable property of ob jective meaning in common, in

spite of being individually dierent (Weyl, 1934/2009, p. 21). Leibniz thus exhibited

the true general meaning of similitude (Weyl, 1939/1997, p. 15). According to Weyl

such a  `philosophical' denition of similar gures can be considered as the rst deni-

tion of the more general concept of  automorphism or symmetry transformation:  an

automorphism carries a gure into one that in Leibniz's words is `indiscernible from it

if each of the two gures is considered by itself '  (Weyl, 1952, p. 18). Automorphisms

are, namely, transformations of space into itself that leave all relevant geometrical struc-

ture intact, so that the result is indistinguishable from the original unless one refers to

something that does not participate in the transformation, which thus serves as a frame

of reference (Rosen, 2008; Kosso, 2000). The ction of a change that involves the entire

universe serves exactly to exclude in principle the possibility of such comparison. Two

worlds arising from each other by an  automorphic transformation, i.e. by a trans-

formation which preserves some geometrical structure, are to be considered the  same

world, since there is in principle nothing outside the universe with respect to which the

transformation can be referred. Thus, after the transformation, one cannot say if the

transformation has taken place or not, and it is impossible to establish whether one is

living in the original universe or in its transmogried copy.

In the next section I will try to show that Weyl's suggestion can nd supporting

evidence in Leibniz's texts. Of course no claim to provide an exhaustive and philologically

precise account of Leibniz's philosophy of geometry can be made in this context (for a

recent reconstruction see De Risi, 2007; still relevant for my exposition is Schneider, 1988;

see also Freudenthal, 1972; Münzenmayer, 1979; Wallwitz, 1991). The more humble aim

of this section is to show that considering Leibniz's indiscernibility arguments in the

light of his reections about geometry throws a dierent light on the meaning of similar

Leibniz-style arguments in the subsequent history of philosophy of space and spacetime.

This will be helpful further on, when we will consider the dierence between Leibniz

indiscernibility and Einstein indiscernibility.

2.1. Leibniz's Denition of Similarity and the Nocturnal Doubling Thought Experiment

Leibniz, as is well known, was unsatised by the traditional denition of  similarity

of gures (see De analysi situs, GM V, pp. 179-80; Leibniz 1976, p. 255). Every gure,

according to the period's mathematical parlance, includes besides quantity also quality or

3
form (GM V, p. 179; Leibniz 1976, p. 254). Similar gures were usually considered those

that have the same  form or  quality, but possibly dierent magnitude or  quantity : for

instance two equilateral triangles or two cubes may have the same  form but dierent

 size. However, such a denition appeared to Leibniz  fully as obscure as the thing

dened (GM V, p. 180; Leibniz 1976, p. 254). Indeed, according to Leibniz,  it is not

enough to designate ob jects as similar whose form is the same, unless a general concept

is further given of form (GM V, p. 180; Leibniz, 1976, p. 254). But the concept of form,

far from being obvious, is laden with cloudy metaphysical presuppositions.

To avoid a direct denition of similarity in terms of identity of quality, Leibniz proposed

therefore a sort of phenomenological denition of  similarity, considering similar those

gures  which cannot be distinguished when observed in isolation from each other (GM

V, p. 180; Leibniz 1976, p. 255). As far as we know, Leibniz provided such a denition for

the rst time in 1677 (Leibniz to Gallois; GM I, p. 180), and he later repeatedly insisted

that the Euclidean denition of similarity was just a particular case of his more general

phenomenological denition (De analysi situs, GM V, p. 179 and 181-82; Specimen

geometriae luciferae, GM VII, pp. 281-82). Here is one of the many passages that one

can refer to:

Things are similar in which when they are considered one by one, nothing by which
they can be dierentiated can be found as two spheres or circles (or two cubes or
perfect squares) A and B . For example if the eye alone without the rest of the body is
imagined now to be inside sphere A now in sphere B it will not be able to distinguish
them but it will be able if it considers both at once, or if it brings with it other organs
of the body, or another standard of measure which it applies now to one now to the
other. Therefore to distinguish similar things, they must either be present together,
or between them a third thing must be present to each successively. [Similia sunt
in quibus per se singulatim consideratis inveniri non potest quo discernantur, ut duo
sphaerae vel circuli vel duo cubi aut duo quadrata perfecta A et B. Ut si solus oculus
sine aliis membris ngantur, nunc esse intra sphaeram A nunc intra sphaeram B, non
poterit eas discernere, sed poterit si ambas simul spectet, vel si secum membra alia
corporis aliamve mensuram introrsum aerat, quam nunc uni nunc alteri applicet.
Itaque ad similia discernenda opus est vel compraesentia eorum inter se, vel tertii cum
singulis successive.] (GM VII, 30; tr. Cox 1978, p. 234)

It is possible, for example, to distinguish  an isosceles triangle from a scalene, even if we

do not see them together [ita triangulum isosceles facile discernitur a scaleno, etsi non

simul videantur] (GM V, p. 155). But if one wants to determine which is the larger of

two equilateral triangles, one  must compare the two triangles [collatione Triangulorum

cum aliis opus habeo] (GM V, p. 155). Then the quality of a gure is what  can be

known in a thing separately, while  quantity is what can be grasped only when the

gures  are actually present together (GM V, p. 180; Leibniz 1976, p. 254).

Leibniz was deeply convinced of the superiority of his own denition of similarity, for it

has not been deduced from the consideration of angles, which represents only one instance

of similarity, but from a deeper principle, that is from the principle of discerning  (LH

XXXV, I, 14, bl. 23-24; tr. De Risi 2005, p. 145). Euclid's denition of similarity

based on the congruence of angles is a special case which does not reveal the nature of

similarity in general (GM V, p. 181; Leibniz, 1976, p. 256). Leibniz explicitly claims

that it is possible to deduce the congruence of angles from his perceptual denition of

similarity, allowing congruence to be dened from similarity rather than the other way

4
around.

It is not possible here to discuss the geometrical implications of Leibniz's strategy,

adopted for instance in the Specimen geometriae luciferae (GM VII, pp. 281-82), to

provide a denition of similitude based on proportion of sides, rather than on angle

congruence (De Risi, 2007, p. 141.) However, in my opinion, following Weyl's interpre-

tative suggestion, we should be able to grasp the philosophical intuition behind Leibniz's

perceptual denition of similarity. According to Weyl, Leibniz seems to glimpse that

the important role of similitude in elementary geometry is related to the fact that trans-

formations which change the size of segments, but preserve both the ratio of lengths and

the size of angles, cannot change any geometric properties of gures. Geometry does

not have at its disposal any conceptual resources to establish the dierence between, say,

a smaller and a bigger circle; every objective statement about the one would hold about

the other, for the size of gures cannot be taken into consideration in geometrical theo-

rems: every theorem, construction, propriety, proportion or relation that can be found

in a circle, could be found also in the other [Omnia theoremata, omnes constructiones,

omnes proprietates, proportiones, respectus, qui in uno circulo notari possunt, poterunt

etiam in alio notari] (GM VII, 276). Thus similar gures (which have the same shape but

possibly dierent sizes) are clearly the same gure for the geometer. Their dierence

cannot be expressed conceptually, but emerges merely through an intuitive compari-

son [comparaison intuitive] (Couturat, 1961, p. 412), that is through what Leibniz calls

 comperceptio.

Leibniz's celebrated thought experiments serve exactly to show ideal cases, where the

possibility of such comparison is ctionally excluded. If, for instance, God were to dimin-

ish all appearances in and around us in a closed room (in aliquo cubiculo), preserving

the proportions (omnia . . . apparentia proportione eadem servata minuere), everything

would appear the same, and we would not be able to distinguish the state before from

that after transformation, without exiting our closed room and considering the things

that have not been diminished (nisi sphaera rerum proportionaliter imminutarum, cu-

biculo scilicet nostro, egrederemur) (GM V, p. 153s). The dierence would emerge

only through comparison between two situations: one that has been scaled, and one

that has not. Something must remain untransformed as a standard against which the

transformation is measured. If this comparison would be in principle impossible:

If God were to change everything conserving the proportion, we would lose all our
measures and it would not be possible to know how much the things have changed,
because it is impossible to determine a certain denition of measure or to conserve
it in the memory. From that I believe I could explain the dierence between size
and species, between quantity and quality. [At si quemadmodum alibi jam dixi Deus
omnia mutaret proportione eadem servata perisset nobis omnis mensura nec possemus
scire quantum res mutatae sint, quoniam mensura nulla certa denitione comprehendi
adeoque nec memoria retineri potest, sed opus est reali ejus conservatione. Ex quibus
omnibus discrimen inter magnitudinem et speciem inter quantitatem et qualitatem
elucere arbitror.] (GM VII, 276)

In the well-known terminology due to Leibniz, the original universe and the transformed

one would be indiscernible.

1 It would not even make sense to speak of a dierence. We

1 An attempt to give an account of the dierent forms (logical, metaphysical, empirical, etc.) that

5
cannot ascertain the change of a length in the course of time, but only the change of

its ratio to real standards of measure which are assumed to be unchanging (GP V,

134; Leibniz 1976, p. 147). Thus if all lengths were diminished or magnied by the

same factor, that is conserving the angles and the proportions among lengths, Euclidean

geometry would provide us with no tools for ascertaining the dierence. Any attempt

to establish through geometrical methods whether we are in the original or the scaled

universe would be in vain.

2.2. Leibniz's Denition of Congruence and the Static Shift Argument

Thus, Leibniz's nocturnal doubling-style thought experiment is simply the counter-

part of his denition of similarity. Leibniz himself explicitly establishes this connection

for the very rst time when he introduced this denition in his never sent letter to Jean
2
Gallois, to which we have already referred. Leibniz was so impressed by his own denition

that he repeatedly tried to provide an analogous phenomenological account of the notion

of congruence (see Schneider 1988 for list of passages). Two gures are congruent if

they can be distinguished not only through the simultaneous perception of both of them,

but also requiring the perception of a third ob ject: Congruent are those things that can

be distinguished only through the comperception with a third [Congrua sunt quae sola

comperceptione cum tertio discerni possunt] (LA VI.4a 565). Two similar gures that

dier in magnitude can be distinguished even if they are in the same place, for one can

be part of the other:

But if two things are not only similar, but also equal, i.e. they are congruent, they
cannot be distinguished even if they are perceived together, if not because of the
place, that is only when something it is assumed outside them and it is observed
that they have a dierent position respect to this third object. [Si vero duae res non
tantum sunt similes sed et aequales, id est si sint congruae, etiam simul perceptas non
discernere possum, nisi loco id est, nisi adhuc aliud assumant extra ipsas et observem
ipsas diversum habere situm ad tertium assumtum.] (GM V, 155)

Thus congruent gures, in Leibniz's scholastic parlance, are dierent  solo numero, only

by the reference to something external ( solo erga situ ad externa discernuntur; GM VII,

275), because one is more on the west or more on the east, more on the north or more on

the south, more above or more below, or because some another body is posited outside

them [unum alio orientalius aut occidentalius vel septentrionalius aut meridionalius vel

superius aut inferius esse vel alteri alicui corpori extra ipsa posito esse] (GM VII, 276).

Leibniz did not remain satised with such a denition of congruence (De Risi, 2007,

pp. 143.). However, it is plausible that the very famous arguments that Leibniz used

in his correspondence with Clarke can be considered the exact counterpart of such a

phenomenological denition of congruence, just as, as we have argued, the nocturnal

the principle of identity of indiscernibles assumes in Leibniz's work is impossible here and not only for
space constraints; on some of the forms that the principle assumes in Leibniz, see (Cherno, 1981).
2 Having thoroughly inquired, I have found that two things are perfectly similar when they cannot
be discerned other than by com-presence, for example, two unequal circles of the same matter could not
be discerned other than by seeing them together, for in this way we can well see that the one is bigger
than the other . . . In fact, if all the things of the world aecting us were diminished by one and the
same proportion, it is evident that nobody could make out the change. (GM I, p. 180; tr. De Risi
(2007, p. 58)).

6
doubling thought experiment is the counterpart of Leibniz's denition of similarity.

Two congruent gures A and B in the same plane may show many dierences, when

they are considered in their relations to some xed frame of reference. However, these

dierences are not geometrical dierences; every ob jective statement that geometry can

make for one gure will hold for the other. If therefore the reference to the third external

gure is eliminated, we would be left without any geometrical tools to establish if the

gure that we are considering is the gure A or the gure B.

Imagine that the whole universe has been displaced. Since the universe is everything,

in principle no external reference frames can be imposed upon it, so that the very concept

of spatial displacement would be inapplicable to the universe. Thus according to Leibniz

to suppose that the universe could have had at rst another position of time and place,

than that which it actually had . . . is an impossible ction (Leibniz, 1976, p. 667).

The details of Leibniz's static shift argument (following the nomenclature introduced

by Maudlin 1993) have been rehearsed so many times in the literature that I refrain

from doing so again here. The only aspect I would like to pinpoint is that it seems clear

that the goal of Leibniz's thought experiment is to make the simultaneous perception

of the original situation and the changed one in principle impossible - showing that

dierence of position is inessential to geometry, that it is geometrically no dierence at

all. Geometrically we cannot establish if we are living in the original or in the displaced

world.

Similarly, when in the Third Paper, Ÿ5, Leibniz so famously argued that if we inter-

change all matter east to west, or left to right, no dierence would emerge, he seems

to imply that the inner geometrical structure of Euclidean space does not allow one to

distinguish a left from a right-handed screw without reference to some third external

asymmetric object. The dierence between left and right is not a geometrical dierence,

as Leibniz seems to admit in this passage: But it is impossible to distinguish left and

right . . . if not for the fact itself or the perception, that the human being experiences

that a motion is more comfortable on one side that on the other [Sed dextrum a sin-

istro discerni non potest . . . nisi facto ipso, seu perceptione, dum ab uno latere motum

commodiorem quam ab alio homines experiuntur] (C Phil VII, D, II, 2, f. 30).

A passage of Leibniz's summarizes eectively the simple path that we have followed

through the rather messy conglomerate of his notes, drafts and manuscripts on geom-

etry: the quality can be observed in one thing, the quantity in two . . . the position in

three [qualitas est in uno observabilis, quantitas in duobus . . . positio in tribus] (A IV.1,

p. 393). If we resort to Weyl's interpretative key, this apparently trivial distinction of

the form of a gure from its position and magnitude seems to express the fundamental

dierence between conceptual denition and intuitive exhibition (Weyl, 1927b, p. 73;

tr. Weyl, 2009b, p. 11), the idea that is impossible to describe a position or x a unit

of length in a conceptual way, through geometric methods, and not by means of a

demonstrative this-here (Weyl, 1934/2009, p. 119). Dierences in position and magni-

tude are geometrically unascertainable and emerge only through an intuitive comparison.

Thus Leibniz's indiscernibility arguments seem to correspond exactly to such perceptual

denitions of congruence and similarity, in as much they serve to ctionally exclude the

possibility of such comparison.

However, precisely for this reason, it is easy to see that they do not succeed at all in

supporting the fully relational conception of space usually attributed to Leibniz, that

is to reduce space to the consequence of the  relations among bodies , as in the received

7
view of Leibniz's philosophy of space. On the contrary, the arguments arm something

very precise about the structure that space possesses independently from the  relations

among bodies. Leibniz's arguments work because it has been settled in advance that

the space, as is said in a recently published manuscript, is not only uniform, i.e. self-

congruent but also self-similar (LH XXXIV, I, p. 14, Bl. 23 retro; tr. De Risi 2005,

p. 140), that it is at or Euclidean. Space, as Leibniz explicitly points out, is not like

a spherical or a cylindrical surface, i.e. one of the surfaces that Leibniz dened as a

uniform locus, self-congruent, but not self-similar; but it is like a plane (LH XXXIV,

I, p. 14, Bl. 23 retro; tr. De Risi 2005, p. 145), which is everywhere internally similar

to itself  (GP VII, p. 22; tr. Leibniz 1976, p. 672).

2.3. A Glimpse into the Relationalism vs. Substantialism debate. Leibniz's Kinematic
Shift Argument
From this point of view it is not clear how Leibniz could credibly have believed that he

challenged Newton's conception of absolute space through his celebrated indiscernibility

argument. Newton's absolute space is obviously endowed precisely with the same Eu-

clidean symmetries of Leibniz's space. Leibniz's arguments seem to confuse the problem

of absolute position with that of absolute motion, that is the problem of the same posi-

tion in dierent times, which was Newton's concern (DiSalle, 2002b). Obviously, Leibniz

could extend his indiscernibility arguments to motion itself.

A ship may go forward, and yet a man, who is in the ship, may not perceive it (GP

VII p. 403; tr. Leibniz 1976, p. 705). As in the other indiscernibility arguments, also

in this Galilean thought experiment one can perceive the motion of the ship only by

reference to something outside the ship, that does not participate in its motion. Once

again the ction of the motion of the whole universe eliminates in principle the possibility

of such a comparison:
3 since there is nothing outside the universe, the motion of the whole

universe, per denition, cannot be observed, and when there is no change that can be

observed, there is no change at all (GP VII, p. 404; tr. Leibniz, 1976, p. 705).

Thus the static shift argument can be easily transformed into a kinematic shift

argument: To say that God can cause the whole universe to move forward in a right
line, or in any other line, without making otherwise any alteration in it, is another

chimerical supposition. These two states indiscernible from each other would be the

same state, it would be a change without a change (GP VII, p. 373; tr. Leibniz, 1976,

p. 705).

There was a time when the philosophical supremacy of Leibniz's relationalist account

of motion over Newton's alleged theological-metaphysical substantialism appeared un-

questionable. However, it has more recently become commonplace to argue that such

a generalized indistinguishability between motion and rest, the idea that there are no

3 In this sense Galilei's ship experiment is the prototype of every indiscernibility argument. According
to Galilei, uniform motion exists relatively to things that lack it, but for things that participate equally
in the motion, motion is as if it does not exist (come s'e' non fusse) (Galilei, 1632/2005, I, p. 205;
tr. Galilei 1967, p. 116). The indiscernibility emerges from the exclusion of the external reference to
other bodies lacking that motion. Similarly Leibniz's changes that involve the whole universe mean
 agendo, nihil agere (GP VII, p. 396), since the reference to something that has remained unchanged
is in principle impossible. There is therefore a connection between global symmetries, indiscernibility
and the possibility of subdividing the universe into isolated subsystems (see Brading and Brown, 2003,
p. 99-98).

8
exceptions to the general law of equivalence (Letter to Huygens, 12/22 June 1694, A

III.6, p. 131; tr. Leibniz, 1989, p. 308) is obviously incompatible with Galilean relativity.

Galilean relativity is the statement that motion in a right line is indistinguishable from

stasis, and not that also motion in any other line (however zigzagging) is indistinguish-

able from stasis. Leibniz's indiscernibility arguments, if applied in all their generality

to all motions, it is said, are unable to make sense of the privileged status of inertial

motion, the uniform motion in straight line, that one should be able to distinguish from

an accelerating motion. Thus Newton's bucket experiment correctly shows that circu-

lar motion is actually an exception to Leibniz's general law of equivalence. However,

Newton's alleged claim (Rynasiewicz, 2000) that this argument would provide evidence

for the existence of absolute space is commonly considered puzzling: Newton's laws of

motion presuppose absolute time, but not absolute space; absolute acceleration, but not

absolute velocity.

It has now become usual (Stein, 1967/1970; DiSalle, 2002a, 2006) to argue that the

Leibniz-Newton debate can be better understood if one assumes that it concerns not the

geometrical structure of space, but that of spacetime. Let me resort once more to Weyl's

account. Not only was Weyl probably the rst to make this point clear, but most of

all his position will reveal itself particularly signicant later on for the purposes of the

present paper. As Weyl famously pointed out,  the dynamic inequivalence of dierent

states of motion teaches us that the world bears a structure (Weyl, 1927a, p. 70 ; tr.

Weyl 2009b, p. 101). Since unaccelerated motion is also called inertial motion, one

also refers to such structure as the inertial structure. But in the concept of absolute

space this inertial structure is  evidently not sized up correctly; the dividing line does

not lie between rest and motion, but between uniform translation and accelerated mo-

tion. spacetime of classical mechanics has dened structure, since  straight lines can be

objectively distinguished from curves, but in the family of all straight lines one can single

out the vertical ones only by a convention based on individual exhibition (Weyl, 1927a,

p. 70 ; tr. Weyl 2009b, p. 101).

Classical mechanics appeals to the action of a background spacetime structure, one

may call it Galilean spacetime, in which the particles are immersed and against which

inertial and non-inertial motion can be distinguished (a body that is not moving in

a straight line is considered as being acted on by a force). With respect to Galilean

spacetime, it is usually argued,  Newtonian spacetime has, so to speak,  too much

structure , since Newton wanted to determine objectively what is a vertical straight

line (a body at absolute rest, the ob jective occurrence of two events in the same place).
4
Leibnizian spacetime, on the other hand, bears  too little structure : if Leibniz requires

the relativity of all motions, then he could not distinguish straight lines from curved lines

(inertial and non-inertial motions). Spacetime would be an amorphous mass of clay

(Weyl 1927b, p. 57; tr. Weyl 2009a, p. 41), without a real inertial structure. Then,

as Weyl points out, the concept of the relative motion of several bodies would have no

more foundation than the concept of absolute motion for a single body (Weyl 1927b, p.

57; tr. Weyl, 2009a, p. 105). In fact, if the automorphisms of spacetime do not preserve

4A much more sophisticated catalog of classical spacetimes is discussed in John Earman's denitive
study World Enough and spacetime (Earman, 1989). Leibnizian spacetime in particular is not completely
amorphous, because Leibniz seems to admit a foliation in hyperplanes of simultaneity. See Roberts 2003
for an overview and a criticism of the received account of Leibnizian spacetime.

9
any structure relational theories of motion would - as we shall see later - be dynamically

trivial (Lariviere, 1987).

Needless to say, Weyl's interpretation is dangerously close to that folk reading

(Huggett and Hoefer, 2009) of Leibniz's philosophy of space and time, which, as recent

literature has shown, is far from exhausting Leibniz's much more articulated conception

of the relationship between relationalism and absolutism about motion.

5 However, as we

shall see later, Weyl's suggestion is useful in order to restate once more what Leibniz's

indiscernibility arguments, in my opinion, implicitly presuppose. If the whole universe

were to move with constant velocity, we would certainly not notice the dierence from

a universe at rest: parallel world lines would be mapped into parallel world lines; the

ane structure of spacetime gives us no means of picking the vertical lines among the

others. All relevant geometrical structure would appear the same, in the universe at

rest and in the uniformly moving one. However, classical mechanics arms that if the

whole universe were rotating, parallel lines would be mapped into curved lines. The

original universe and the transformed situation would not be indistinguishable at all, for

the ane structure gives us the possibility of distinguishing between curved and straight

trajectories in spacetime.

This was of course hard to grasp without having the possibility of considering spacetime

as a single geometrical structure. However, exactly for this reason, we are in a better

position to recognize that it is the geometry, which has been settled in advance, that

decides the indiscernibility. Any proto-vericationist interpretation of Leibniz's famous

remark that motion depends upon being possible to be observed (GP VII, 403; tr.

Leibniz 1976, p. 705) (like Reichenbach's reading, as we shall see) obscures the fact that

it is the inner structure of Galilean spacetime that does not allow the establishment

of the dierence between rest and uniform motion, but does allow that between uniform

motion and acceleration. From today's standpoint, the fact, taken as obvious, that such

a mere geometrical structure is able to exert such an important inuence on physical

reality appears more than surprising. Not only does spacetime geometry make bodies

conspire to move in straight lines at uniform speeds, but it even opposes resistance when

one attempts to deviate bodies from such trajectories: It is probably fair to say that

anyone who is not amazed by this conspiracy has not understood it (Brown, 2005, p.

14).

3. Leibnizian Indiscernibility Arguments in the 19th Century Debate about

Geometry.

Weyl's intuition seems to oer a good insight into Leibnizian indiscernibility argu-

ments: in geometry two gures are considered the same gure if one can be carried

5 As is well known, in many passages Leibniz insists that a distinction between true motion and merely
relative motion can be drawn in terms of force or vis viva , that is mv2 . This claim is usually considered
inconsistent or at least circular (measurement of mv 2 depends on the denition a reference of frame);
this nevertheless makes clear that Leibniz's spacetime had probably a richer structure than what is now
called Leibnizian spacetime (Roberts, 2003, p. 553). In Roberts 2003, for instance, it is argued that
Leibniz would have even defended a form of absolutism about motion; for Arthur 1994 on the contrary
Leibniz was a full-blooded relativist. According to (Jauernig, 2008) it should be possible to reconcile
absolutism and relativism about motion by considering the dierent ontological levels, the dynamical
and the phenomenal levels, that characterize Leibniz's thought. See also (Slowik, 2006).

10
into the other by an automorphism (Weyl, 1927a, p. 79; tr. Weyl 2009b, p. 73), i.e. by a

structure preserving mapping of the space onto itself. That is now our interpretation of

Leibniz's denition of similar gures as gures that are indiscernible if each is considered

by itself (Weyl, 1927a, p. 79; tr. Weyl 2009b, p. 73). Thus, which worlds count as indis-

cernible depends on what kind of mathematical structure, we are concerned with. Once

this is known, we should be able to pinpoint what mappings preserve the structure, that

is the automorphism of this structure. In general transformations to which Leibniz refers

- the scaling of the entire world or, in his dispute with Clarke, the interchange east with

west or displacement of everything three feet east - are not arbitrary since they preserve

some  geometrical structure. The original and the copy are indistinguishable precisely

because all the relevant geometrical structure that was found before the transformation

will appear the same in the transformed situation. In other words, indiscernibility arises

because the geometrical structure we are considering does not allow the expression of

dierences that we might otherwise consider intuitively evident.

This point becomes particularly clear if one considers the story of indiscernibility argu-

ments in the 19th century debate on the foundations of geometry. The rapid development

that geometrical thought experienced from the 1830s opened to the protagonists of this

debate, such as Helmholtz, Poincaré and, as we shall see, Hausdor, the possibility of gen-

eralizing Leibniz's thought experiments (even if Leibniz is seldom explicitly mentioned):

two worlds will be indistinguishable not only if they are congruent or similar, but even

if they are mapped onto each other by any continuous deformation whatsoever, only re-

quiring that points that are close together before the transformation is applied also end

up close together. In other words, the 19th century debate seems to be dominated by

the tendency to consider transformations of space into itself that preserve progressively

weaker levels of geometrical structure. Two worlds arising from each other by a trans-

formation which preserves some geometrical structure are to be considered the  same

world.

3.1. Hausdor 's Geographical Maps

Helmholtz's and Poincaré's arguments have been abundantly discussed in the literature

(see for instance DiSalle 2006). I will therefore try to reconstruct summarily the 19th cen-

tury geometrical debate from a signicant, but less commonly considered, point of view.

The German mathematician Felix Hausdor, in his major philosophical work Das Chaos
in kosmischer Auslese
6 (Mongré 1898, now in Hausdor 2004, with the same pagination),

published in 1895 under the pseudonym of Paul Mongré (Paul to-my-liking), oered a

good overview of all the Leibniz-style indiscernibility arguments that were widespread in

the period's debate about geometry. Hausdor is surely better known for his fundamental

contributions to topology and set theory than for his early forays into philosophy of ge-

ometry, to which scholars' attention has been attracted only recently (Epple 2006, 2007).

Nevertheless Hausdor 's reections on the foundation of geometry provide a valuable

overview into the use of indiscernibility arguments in the 19th century philosophical de-

bate about geometry - a debate that, as is well known, profoundly contributed to shaping

the new-born philosophy of science at the turn of the century (Friedman, 1999).

6 Chaos in Cosmic Selection, as it might be translated in English

11
 Hausdor considered rst of all translations, rotations, mirroring and scaling of the

whole universe showing that a consciousness would necessarily remain unaware of them,

as long as all meter sticks used to ascertain any change, also underwent the same de-

formation (Mongré 1898/Hausdor 2004, pp. 84-94). But according to Hausdor, one

would consider two worlds as indistinguishable even if the objects of the universe were ar-

bitrarily distorted in arbitrary directions, by any deformation whatsoever, only requiring

that, in a rst approximation, it is free from discontinuities and singularities [Unste-

tigkeiten und Singularitäten] (Mongré 1898/Hausdor 2004, pp. 84-94): We would not

notice anything of certain space transformation: this is the refrain of my transcendental

dialectic [Von gewissen Transformationen des Raumes würden wir nichts bemerken: das

ist der Refrain meiner transcendentalen Dialektik] (NL FH Kapsel 49; Fasz. 1079; Bl.

26).

This was also the refrain of the whole period's philosophical-geometrical debate, which

was entirely dominated by similar thought experiments, showing in a popular illumi-

nating way [in populär einleuchtender Weise], how a space transformation can elude

our empirical perception [dass eine Raumtransformation sich der empirischen Wahr-

nehmung entzieht] (Mongré 1898/Hausdor 2004, p. 100). Hausdor is aware that the

Helmholtzian convex mirror [Helmholtzens Convexspiegelbild] (NL FH Kapsel 24; Fasz.

71; Bl. 33) already oered an example of such an approach, even if Helmholtz was more

interested in the possibility of visualizing non-euclidean relationships [die Anschaubar-

keit nicht euklidischer Verhältnisse] (Kapsel 24: Fasz. 71, Bl. 65) than of showing the

indeterminacy of the space structure. In an fragment of the Nachlass, unfortunately

undated, Hausdor confesses that he found similar reasoning also by others (Poincaré)

[auch bei Andern (Poincaré)] (NL FH Kapsel 49; Fasz. 1079; Bl. 4), who similarly argued

that our actual space does not dier from any space that one can derive from it by any

continuous deformation whatsoever.

However, Hausdor could consider such a type of thought experiment only as a special

case of what he called the principle of transposition, transformation principle, map-

ping principle; principle of substitutability [Übertragungsprincip, Transformationsprin-

cip, Abbildungsprincip; Princip der Ersetzbarkeit] (Kapsel 24: Fasz. 71, Bl. bl4 [7]).

Two spaces that are point-wise coordinated to one another, in such a manner that
their whole physical content participates in this point-correspondence, produce the
same mental image . . . Every space stands for a whole class of spaces, among which
no dierentiation, and thus also no choice, is possible. [Zwei Räume, die einander
punktweise zugeordnet sind, derart dass ihr gesamter physischer Inhalt an dieser Cor-
respondenz der Punkte betheiligt ist, erzeugen dasselbe Bewusstseinsbild. . . . jeder
Raum Repräsentant einer ganzen Klasse von Räumen, zwischen denen keine Unter-
scheidung, auch also keine Entscheidung möglich ist] (Kapsel 24: Fasz. 71, Bl. 65)

This approach is surely commonplace for someone who is familiar with the work of

Helmholtz or Poincaré. However, completely original in Chaos in kosmischer Ausles e is

Hausdor 's attempt to approach the issue using the modern concepts of Cantorian set

theory, then new-born (Cantor, 1874, 1878). This seems surprising at rst sight. Only

a few mathematicians at that time were interested in this eld and probably none at-

tempted to apply set theoretical considerations to philosophical and epistemological prob-

lems. However, precisely this rst unusual contact with Cantor's theory showed Hausdor

the potentialities of the  Mengenlehre, that he later dened as the groundwork of all

12
mathematics [das Fundament der gesamten Mathematik] Hausdor, 1914/2002, p. 1. In

1904 Hausdor began to work actively in this area, and in 1914, in his celebrated Grund-
züge der Mengenlehre , he developed the theory for which he is now rightly famous, the

theory of topological spaces, Hausdor 's labeling for sets endowed with an additional

structure in which distinct points have disjoint neighborhoods (actually what we now

call Hausdor spaces; see Hausdor, 1914/2002, p. 718f.; Hausdor (1927, 1935/2007,

p. 226f.)).

However, already in Hausdor 's geometrical-philosophical speculation, one can see

the tendency to consider space as point sets endowed with a structure. In particular,

Hausdor considers the empirical and the absolute space as two point sets. Between

these two, one can apply what has been newly called a transformation or a mapping of a

space onto the other [was man neuerdings eine Transformation oder Abbildung des einen

Raumes auf den anderen nennt] (Hausdor, 2004, p. 82 ), that is a transformation that

preserves some geometrical structure. That idea that dierence between the empirical

and absolute space would not fall in our consciousness [nicht in unser Bewuÿtsein fallen

würde] (Hausdor, 1903, p. 17) means now that they can be mapped onto one another

by such a structure-preserving mapping. Thus: geometry is not valid for a particular

space (the real one), but for all its univocal mappings [Also: Geometrie gilt nicht von

Einem bestimmten Raum (dem wirklichen), sondern von all seinen eindeutigen Bildern].

Once again, which worlds count as indistinguishable depends upon which geometrical

structure one wants to preserve. If all the relevant structure is preserved by the mapping

(one-to-one and onto, but Hausdor considered more general cases), then the original and

the deformed world would be regarded as the same and dierence between them would

escape any observation. The sense of the expression the same is here intended as mere

set-theoretical equivalence, a mere mapping, coordination, correspondence [Abbildung,

Zuordnung, Correspondenz] between sets:

Mapping in the sense of correspondence, coordination. The usual parlance assigns

to these pictures a certain similarity with the original. In this sense one usually
says that mind-processes are signs, not pictures of the external world, or that the
words are signs, and not picture of the concepts. Geographical charts provide already
a freer conception of the concept of mapping. [Abbildung im Sinne von Zuord-
nung, Correlation. Der gewöhnliche Sprachgebrauch schreibt dem  Bilde eine gewisse
Ähnlichkeit mit dem Original zu; in diesem Sinne sagt man, dass die Bewusstseinsvor-
gänge Zeichen, nicht Bilder der Aussenwelt oder die Worte Zeichen, nicht Bilder der
Begrie seien. Geographische Karten führen schon zu freierer Auassung des Begris
Abbildung. (NL FH Kapsel 49; Fasz. 1079; Bl. 8)

If the relationship between absolute and empirical space is conceived as an Abbildung,

a mapping in the set theoretical sense, it becomes clear that it does not make any sense

to demand any congruence or similarity from such a mapping [von dieser Abbildung

Congruenz oder Ähnlichkeit oder derlei zu verlangen] (Kapsel 49: Fasz. 1077, Bl. 4).

In 1903, in his inaugural lecture about  Das Raumproblem as an extraordinarius at

Leipzig University, in order to give an idea of this procedure, Hausdor eectively com-

pares the empirical space exactly to a geographical map of the absolute space: If this

conception is correct, then the original can undergo every transformation whatsoever,

without any change in the mapped copy: exactly as you cannot recognize in a geograph-

ical map, if it was drawn from the original or from another geographical map [Wenn

13
diese Auassung richtig ist, so muÿ man das Urbild einer beliebigen Transformation un-

terwerfen können, ohne daÿ das Abbild sich verändert: gerade so wie man einer Karte

nicht ansehen kann, ob sie nach dem Original oder nach einer anderen Karte gezeichnet

ist] (Hausdor, 1903). Thus in order to know the nature of the absolute space starting

from the empirical one, one has to know which kind of pro jection has been used, that is,

what kind of structure has been preserved by the mapping:

Our empirical space is like a three-dimensional geographical chart, a mapping of the

absolute space; but we lack the map legend, we do not know the mapping principle
and hence we do not know the original. Between both spaces there is an unknown, and
arbitrary relationship or correspondence, a completely arbitrary point-transformation
[. . . ] [T]he deformation however does not fall into our consciousness, because not
only the objects, but also we, and our measuring instruments are similarly deformed.
[Nun, unser empirischer Raum ist solch eine körperliche Karte, ein Abbild des absolu-
ten Raumes; aber es fehlt uns der Eckenvermerk, wir kennen das Projektionsverfahren
nicht und kennen folglich auch das Urbild nicht. Zwischen beiden Räumen besteht eine
unbekannte, willkürliche Beziehung oder Korrespondenz, eine völlig beliebige Punkt-
transformation . . . die Verzerrung fällt nicht in unser Bewuÿtsein, weil nicht nur die
Objekte, sondern auch wir selbst und unsere Meÿinstrumente davon gleichmäÿig be-
troen werden.] (Hausdor, 1903, p. 17)

For example, a map of the world shown in Mercator's pro jection accurately depicts only

the equatorial regions of the Earth's surface. As one moves nearer and nearer to the

polar regions, so the features of the map become progressively distorted. This distor-

tion is particularly pronounced for Greenland and Antarctica, which become drawn out

horizontally far in excess of their true proportions. The reason for this is well known,

of course, it being simply due to the fact that the surface of the Earth is spherical,

and it is not possible to represent a curved surface on a at map without distortion.

However, since all measuring instruments would be equally distorted, someone living

in the distorted situation would not notice the dierence. The only way to know the

structure of the absolute space from the empirical one, our geographical map, is to know

the Abbildungsverfahren , the kind of mapping, that has been used to draw the map

(Mercator, Stereographic, etc.). The empirical space is no faithful copy of the absolute

one, but only a mapping according to an arbitrary, indeterminable pro jection principle

[keine getreue Kopie des absoluten, sondern nur sein Abbild nach einem beliebigen, un-

bestimmbaren Pro jektionsverfahren] (RP 17). If we don't know which structure (angles,

areas, geodesics, etc.) has been preserved by the mapping, there is no way to infer the

real geometry of space.

4. The Point-Coincidence Argument as a Leibniz-Style Indiscernibility Ar-

gument: the Logical Empiricist Interpretation of General Relativity

Helmholtz and Poincaré, but also Hausdor/Mongré, are the authors to whom Moritz

Schlick refers in his celebrated Raum und Zeit in der gegenwärtigen Physik (Schlick,

1922/2006, tr. Schlick, 1978a), which achieved four editions from 1917 to 1922. If

Helmholtz's and Poincaré's inuence on Schlick is well known and well documented in

the literature (Coa, 1991; Pulte, 2006; Ryckman, 2005; Friedman, 1995), the relationship

between Schlick and Hausdor has only recently been brought to the attention of a larger

14
audience (Epple, 2006). For the aims of this paper, it is interesting in particular that,

after an exchange of letters in 1919-20,

7 Schlick, in the fourth edition of his work, felt

obliged to add a note to recognize the value of Hausdor 's philosophical reections.
8
This says a lot about Schlick's position regarding the connection between 19th century

Leibnizian indiscernibility arguments and the philosophical interpretation of the new-

born theory of General Relativity.

In particular, the third chapter of Schlick's booklet is entirely dominated by the re-

production of Leibniz-style thought experiments. Schlick starts by considering the case

in which the imaginary transformed world is geometrically similar to the original one,

then he imagines that the dimensions of all ob jects are lengthened or shortened in one

direction only, and nally he considers the case where the objects in the universe are

arbitrarily distorted in arbitrary directions (Schlick 1922/2006, p. 202; tr. Schlick

1978b, p. 227). Schlick's conclusion is invariably the same: as long as we suppose that

all measuring instruments, including our own bodies share the same deformation, the

whole transformation immediately becomes unascertainable (Schlick 1922/2006, p. 202;

tr. Schlick 1978b, p. 227):

In mathematical phraseology we can express this result by saying: two worlds, which
can be transformed into one another by a perfectly arbitrary (but continuous and
one-to-one) point-transformation, are, with respect to their physical reality, identical.
That is: if the universe is deformed in any way, so that the points of all physical
bodies are displaced to new positions, then [. . . ], no measurable, no  real change
has happened at all, if the co-ordinates of a physical point in the new position are
any arbitrary functions whatsoever of the co-ordinates of its old position. (Schlick
1922/2006, p. 204; tr. Schlick 1978b, p. 227)

Moving from the consideration of space to that of spacetime, it was easy for Schlick to

use this sort of argument in the context of General Relativity. The connection between

the 19th century debate and the new theory can be found in the, as John Stachel has

conveniently labeled,  point-coincidence argument ; an argument that Schlick could read

in Einstein's 1916 review article Die Grundlage der allgemeinen Relativitätstheorie . Ein-

stein, as is well known, maintained that the physical content of a theory is exhausted

by the catalog of the  spacetime coincidences or  verications of . . . meetings of the

material points of our measuring instruments with other material points (CPAE 6, Doc.,

30, p. 291f.). According to Schlick, this implies that all worlds that agree on such co-

incidences are equivalent, and that a choice among them is the result of an arbitrary

stipulation (an implication that Schlick called  the geometrical relativity of space ):

All world pictures which lead to the same laws for these point-coincidences are, from
the point of view of physics, in every way equivalent. We saw earlier that it signies
no observable, physically real, change at all, if we imagine the whole world deformed
in any arbitrary manner, provided that after the deformation the co-ordinates of every

7 The letters are preserved in Noord-Hollands Archief in Haarlem (NL): 102/Haus-2 Letter to Schlick
(23.2.1919, Greifswald); 102/Haus-2 letter to Schlick (17.7.1920, Greifswald).
8 Unfortunately, only after the publication of the second edition of this writing have I learned about
the most astute and fascinating book [Das Chaos in kosmischer Auslese]. The fth chapter of this
monograph gives a very perfect presentation of the considerations that follow in the text above. Not
only Poincare's reections, but also the extensions added above have been anticipated there (Schlick
1922/2006, p. 198, note 1, tr. Schlick 1920, p. 24, note 1).

15
 physical point are continuous, single-valued, but otherwise quite arbitrary, functions
of its co-ordinates before the deformation. Now, such a point-transformation actu-
ally leaves all spatial coincidences totally unaected; they are not changed by the
distortion, however much all distances and positions may be altered by them. For, if
two points A and B, which coincide before the deformation (i.e. are innitely near
one another), . . . as a result of the deformation . . . must be at the same point (or
innitely near) A. Consequently, all coincidences remain undisturbed by the deforma-
tion. (Schlick 1922/2006, p. 232; tr. Schlick 1978b, p. 227)

Schlick's interpretation of the point-coincidence is usually considered the expression of a

form of vericationism, as a commitment to a strong observability requirement. In this

sense the point-coincidence remark has been famously dened as the beginnings of the

empiricist and vericationist interpretation of science characteristic of later positivism

(Friedman, 1983, p. 24). It is indeed the case that Einstein's point-coincidence argu-

ment fascinated contemporary philosophers for its vericationist turn of phrase. But the

argument actually oered them much more. It allowed them to exorcise the novelty of

Einstein's newborn theory, by simply inserting it in the context of the by then already fa-

miliar 19th century debate about geometry. Einstein's point-coincidence argument is just

one of those many Leibniz-style philosophical arguments that one can nd in Helmholtz,

Poincaré or, as we have seen, Hausdor (Friedman, 1983, p. 47).

Schlick had by that time reached the philosophical stature necessary to transform

such an interpretative proposal into the received view. As is well known, he was a

trained physicist, having taken his doctorate under Max Planck, who notably singled

out Schlick as one of his best students, together with Max von Laue, later awarded the

Nobel Prize. But most importantly, Schlick and Einstein were in correspondence by late

1915; Schlick sent Einstein a copy of his paper on the philosophical signicance of the

theory of relativity (CPAE 8a, Doc. 296; 4 February 1917), which Einstein famously

praised for its unsurpassed clarity and perspicuousness [Übersichtlichkeit] (CPAE 8a,

Doc. 297, p. 389; 6 February 1917; see also Doc. 165; 14 December 1915). When

in 1922 Schlick was appointed to the chair in philosophy held earlier by Ernst Mach

and by Ludwig Boltzmann at the University of Vienna, he was already the recognized

philosophical authority on the sub ject of relativity.

Schlick's inuence on the philosophical debate on geometry and relativity was enor-

mous. The young Rudolf Carnap in his doctoral Dissertation Der Raum (Carnap, 1922)

drew mostly upon from Schlick's interpretation of Einstein's passage on the geometri-

cal interpretation of the point-coincidence argument - the idea that point-coincidences

are the only topological invariant, and are therefore unambiguous, whereas anything

else - pro jective, ane or metric structure is the result of a stipulation. Later Car-

nap remained essentially faithful to this approach, even after abandoning the Husser-

lian/Kantian framework in which it had originally been developed. And it was after hav-

ing corresponded with Schlick that Hans Reichenbach (1923/24), for whom Einstein later

created a chair in the philosophy of science in the physics department at Berlin, quickly

modied his early Neo-Kantian interpretation of the theory (Reichenbach, 1920/1977) in

the direction of the metric conventionalism for which he later became famous with his

classic Philosophie der Raum-Zeit-Lehre (Reichenbach 1928/1977 translated as Philoso-

phy of Space and Time in Reichenbach 1928/1958).

Reichenbach was perhaps the rst among Logical Empiricists to insist on the fact that

the indiscernibility arguments that ourished in 19th century philosophy of geometry

16
were nothing but variations of Leibniz's argument in the Leibniz-Clarke correspondence

(Reichenbach, 1924). Such arguments, in Reichenbach's vericationist reading, express

the idea that  it is meaningless to postulate dierences in objective existence if they

do not correspond to dierences in observable phenomena (Reichenbach 1928/1977, p.

495; tr. Reichenbach 1928/1958, p. 210). Reichenbach felt a profound philosophical

anity with Leibniz. According to Reichenbach, Leibniz went so far as to recognize the

relationship between causal order and time order (Reichenbach, 1949/1951, p. 300), and

most of all with his principle of the identity of indiscernible, discernible in connection

with the veriability theory of meaning, he laid the foundation of what Reichenbach

calls the  theory of equivalent descriptions (see for instance Lehrer 2004; Klein 2001)

where Leibniz's indiscernibility explicitly oods into the period's debate about geometry

Reichenbach's approach is well known. Only a set of dierent descriptions, rather

than a single description, can correctly describe the geometry of physical space, in as

much as these dierent geometries can be represented on one another by a one-to-one

correspondence (Reichenbach, 1949/1951, p. 298), so that all ob jects are assumed to be

deformed in such a way that the spatial relations of adjacent bodies remain unchanged:

In this context belongs the assumption that overnight all things enlarge to the same
extent, or that the size of transported objects is uniformly aected by their position.
Helmholtz's parable of the spherical mirror comparing the world outside and inside
the mirror is also of this kind; if our world were to be so distorted as to correspond
to the geometrical relations of the mirror images, we would not notice it, because all
coincidences would be preserved. (Reichenbach 1928/1977, p. 38-, tr. Reichenbach,
1928/1958, p. 27; my emphasis)

According to Reichenbach, Einstein's theory of relativity is the result of the recognition

of such a relativity of geometry, the recognition that a dierent choice of a coordinate

denition of rigid bodies or straight lines may yield dierent geometrical descriptions of

the world, that are however physically equivalent.

The connection of this strategy with Leibnizian indiscernibility arguments is made

particularly clear by Rudolf Carnap in a passage of his Philosophical Foundations of

Physics (Carnap, 1966), published in 1966:

Leibniz, the reader may recall, had earlier defended a similar point of view. If there is
in principle no way of deciding between two statements, Leibniz declared, we should
not say they have dierent meaning. If all bodies in the universe doubled in size
overnight, would the world seem strange to us next morning? Leibniz said it would
not. The size of our own bodies would double, so there would be no means by which we
could detect a change. Similarly, if the entire universe moved to one side by a distance
of ten miles, we could not detect it. To assert that such a change had occurred would,
therefore, be meaningless. Poincaré adopted this view of Leibniz's and applied it to
the geometrical structure of space. (Carnap, 1966, p. 148)

This passage clearly shows that the strategy adopted by Logical Empiricists to assimilate

General Relativity is disarmingly simple: Einstein's point-coincidence argument is simply

a Leibniz-style indiscernibility argument, the same kind of argument that dominated

the 19th century debate on the philosophical foundation of geometry. If Helmholtz,

Poincaré or Hausdor applied their arguments to space alone, Einstein simply extended

them to spacetime, that is to the intersection of world lines: a coincidence of two

17
world lines presupposes nothing concerning the metrical relations of space and time, so

metrical properties of spacetime are deemed less fundamental than topological ones.

The ob jective system of coincidences does not depend on an observer, it is therefore

independent of all arbitrariness and an ultimate fact of nature, whereas the metric

relations are frame-dependent and conventional. Point-coincidences represent a sort of

xed framework in which we can formulate an equivalence class of physically possible

geometries mapped onto each other by one-to-one continuous transformations (that is

probably what we would called dieomorphisms): Topological properties turn out to

be more constant then the metrical ones, so that the transition from the special theory to

the general one should be interpreted as a renunciation of metrical particularities while

the fundamental topological character of space and time remains the same (Reichenbach

1924, p. 115; tr. Reichenbach, 1924/1969, 195; see Ryckman 2007, 2008).

The vericationist avor of such a reading of the point-coincidence argument, the

idea (on which Howard 1999 famously insisted) that point-coincidences are taken to be

real because of their observability, and thus they qualify as invariant, seems however to

induce Logical Empiricists to underestimate the consequences of such an unrestrained use

of Leibnizian arguments. In such arguments, as we have seen, indiscernibility does not

arise at all from the impossibility of observing certain dierences physically, but from the

impossibility of expressing them geometrically. Just as Euclidean space does not allow

one to establish the dierence between left or right without a coordinative denition,

a bare topological space would not allow one to observe the distinction between

straight and curved lines. From such a conclusion, it is not hard to prognosticate that

Logical Empiricists' interpretation of General Relativity, despite its undisputed historical

relevance, was destined to failure from a theoretic point of view. Probably only the

implementation of a generalized Machian point of view in which the inertial force can

be interpreted . . . as a dynamic gravitational eect (Reichenbach 1928/1977, p. 247;

tr. Reichenbach, 1928/1958, p. 214), prevented them from seeing that the interpretation

they were suggesting would have made General Relativity dynamically empty.

4.1. The Failure of the Logical Empiricist Interpretation of General Relativity

If the interpretation of the point-coincidence argument as Leibniz-style indiscernibility

argument can be considered the core of the Logical Empiricist interpretation of General

Relativity, at the same time then it is the reason for its substantial inadequacy. As we

have tried to show, Leibniz's indiscernibility arguments express the global symmetries of

space or spacetime: two universes mapped by some kind of deformation would be indis-

tinguishable, if all relevant structure were preserved, since all that the theory considered

geometrically relevant would appear the same in both universes. If the point-coincidence

argument were an indiscernibility argument of this kind, its result would be therefore to

enlarge the spatio-temporal symmetry group of spacetime to the group of all one-one,

bi-continuous point transformations: only the topological features of events are preserved

by this group of transformations, that is, the notion of the  coincidence of two events

and the notion of two events being near one another in spacetime.

From a contemporary perspective, the mistake of such an account is easy to recognize:

the symmetry group of an arbitrary general relativistic spacetime is not the widest group

of all smooth coordinate transformations that preserve only point-coincidences, but the

narrowest one consisting of the identity alone. The use of the full group of admissible

transformations in General Relativity does not imply that we are working in the context

18
of a very weak, though xed, geometrical structure, but, rather, that we are working in

the context of a highly structured spacetime, endowed with a perfectly denite, although

variable, metric. In this context indiscernibility arguments in the sense of Leibniz do not

even make any sense. In an inhomogeneous space it really would make a dierence if one

would shift everything into a region of increasing spatial curvature, and the consequences

of doubling would depend on where the doubling were carried out (Nerlich, 1994, p. 152).

When Weyl (1924b), upon whose authority we wish once more to rely, reviewed Re-

ichenbach's Axiomatik der relativistischen Raum- Zeit-Lehre , (Reichenbach 1924, tr. Re-

ichenbach, 1924/1969) he found the book not very satisfactory, too laborious and too

obscure [wenig befriedigend zu umständlich und zu undurchsichtig](Weyl, 1924b). Be-

yond mathematical technicalities, what Weyl probably felt to be the most philosophically

extraneous was Reichenbach's identication of the philosophical achievement of Gen-

eral Relativity with the separation between the factual and conventional components,

those which are xed once and for all, and those which result from the stipulations of

rigid rods and ideal clocks. Weyl's reliance on the use of trajectories of force-free mass

points in the construction of the metric (Ehler, 1988)

9 is, on the contrary, the result of a

completely dierent philosophical attitude (Ryckman, 1995, 2005). As it is well known,

according to Weyl, the main novelty of Einstein's theory of gravitation was that it had

transformed the guiding inertial structure, that counts as standard for no-acceleration,

from a rigid geometric property of the world, xed once for all (Weyl, 1934/2009, p.

134) to a guiding eld (Führungsfeld , as Weyl famously called it) a physical reality

which is dependent on the state of matter (Weyl 1921/1968, p. 141 tr. Weyl 1921/2009,

p. 21): The distinction between guidance [Führung] and eld is preserved, but guidance
has become a eld (as the electromagnetic eld [An dem Dualismus von Führung und

Kraft wird also festgehalten; aber die Führung ist ein physikalisches Zustandsfeld (wie

das elektromagnetische)] (Weyl, 1924a, p. 198).

The philosophical meaning of General Relativity should be sought not in the distinc-

tion between arbitrary and non-arbitrary structures, but in that between dynamical and

non-dynamical ones. The dierence between Galilean spacetime and Special Relativis-

tic spacetime lies in the dierence between their inertial structures. Nevertheless, both

structures are non-dynamical in the sense that they are independent of their contents:

the unique ane connection (compatible with the spacetime metric), it is said, provides a

standard for absolute acceleration and rotation. The radical novelty of General Relativ-

ity does not consist in weakening such a xed background structure, but in transforming

the xed background into a dynamical one. Since all bodies are inuenced by gravity in

precisely the same way (equivalence principle), there are no physical phenomena inde-

pendent of gravitation that might serve to measure the background spacetime geometry;

on the contrary we can measure the acceleration of a particle in a magnetic eld relative

to the inertial tra jectory of a body that is not aected by magnetism. In other terms,

there is no unique decomposition of the ane structure into an inertial structure and the

deviation from this structure caused by gravitation.

9 So Weyl summarizes his position: the metrical structure of the world is already fully determined by
its inertial and causal structure, that therefore mensuration need not depend on clocks and rigid bodies
but that light signals and mass points moving under the inuence of inertia alone will suce (Weyl,
2009b, p. 103).

19
4.2. The Point-Coincidence Argument as a Response to the Hole Argument

Even if the distinction between dynamical and non-dynamical structures (or dynamical

and absolute objects, in James Anderson's terms; Anderson 1967) turned out to be not

suciently sharp (Giulini 2007; see Pitts 2006), nevertheless it is now a widely shared

opinion that it better expresses the philosophical spirit of General Relativity than the

Logical Empiricists' distinction between arbitrary and non-arbitrary structure. It is not

the case that the metric that was non-arbitrary in previous theories has become arbitrary,

but that the metric that was non-dynamical has become dynamical - it has become a

eld among others.

As we argued, the Logical Empiricists' failure to grasp this point seems connected

to their interpretation of the point-coincidence argument as a 19th century radicaliza-

tion of Leibnizian indiscernibility arguments. This mistake is of course at least partly

comprehensible. The echo of the widespread use of such arguments in one of the most

exciting philosophical debates of the turn of century was still vivid. Moreover Einstein's

Helmholtzian insistence on the importance of practically rigid rods and clocks (Ein-

stein, 1921, 1925) could be easily seen as the authoritative conrmation that this was

the correct interpretative context (Ryckman, 1996; see also Howard, 2005). Only re-

cent scholarship initiated by John Stachel (Stachel, 2002) discovered the key to a proper

understanding of the point-coincidence argument, with the help of Einstein's correspon-

dence from that period . The meaning of what we may call (following Rynasiewicz 1999)

the public point-coincidence argument can be understood only if one knows the private
point-coincidence argument as a response to the infamous hole argument. The details

of both versions of the argument have been rehearsed many times in recent literature,

so it does not seems necessary to repeat here the Hole Story, as Earman and Norton

ingeniously called it. I will attempt anyway a brief exposition.

(I) The public point-coincidence argument, as we said, appeared for the rst time in

Einstein's 1916 review article on General Relativity, and it is used to express the formal

request of general covariance, i.e. of coordinate-independent formulation of the laws of

nature: The laws of nature are only propositions about spatio-temporal coincidences;

therefore they nd their natural expression in generally covariant elds equations (CPAE

7, Doc. 4, 38; Prinzipielles zur Relativitätstheorie , 1918). In this context the argument

has nothing to do with indiscernibility, as the Logical Empiricists believed. If coincidences

are all that matters physically, then we ought to be able to use any coordinate system,

since all coordinate systems necessarily agree on such coincidences (Norton, 1995). The

young Wolfgang Pauli (at his third semester), in his celebrated  Enzyklöpedie article on

relativity (Pauli, 1921), sums up this line of reasoning clearly:

All physical measurements amount to determination of spacetime coincidences; noth-

ing apart from these coincidences is observable. If however two point events correspond
to the same coordinates in one Gaussian coordinate system, this must also be the case
in every other Gaussian coordinate system. We therefore have to extend the postulate
of relativity: The general physical laws have to be brought into such a form that they
read the same in every Gaussian coordinate system, i.e. they must be covariant under
arbitrary coordinate transformations (Pauli, 1958, p. 149).

In this form the argument is customarily regarded as physically vacuous, since such

a requirement can be satised by virtually all theories, independently of the content

of the laws. The unavoidable reference is to the now familiar argument of the young

20
mathematician Erich Kretchman, who as early as 1917 (Kretschmann, 1917) turned the

point-coincidence argument against Einstein: if the physical content of every spacetime

theory is exhausted by the catalog of spacetime coincidences, that is by topological

relations, then for this very reason all spacetime theories can be given in a generally

covariant formulation (Rynasiewicz, 1999).

(II) The private point-coincidence argument is a response to the hole argument, which

probably rst occurred to Einstein by November 1913 (Stachel, 1980/2002) at the latest.

In this form the argument appears actually as a sort of indiscernibility argument. As

it is well known, Einstein was worried that by means of a coordinate transformation,

two dierent solutions of generally covariant eld equations would arise at a single point

within a hole (a region devoid of matter and energy), whereas the sources outside the hole

have not changed. Einstein overcame this diculty when he came to realize that two sets

of eld lines that intersect in the same way, that is that agree on point-coincidences, dene

the same physical situation, since we do not have any means to separate a background

system of rigid grids from the eld lines. Hence, two metric elds whose geodesics

intersect one another in the same way are the same metric eld.

The nature of the point-coincidence argument as an indiscernibility argument is prob-

ably nowhere more in evidence than in Einstein's correspondence with Paul Ehrenfest in

late December 1915 and early January 1916 (CPAE 8a, Doc. 173, 26 December 1915, and

Doc. 180, 5 January 1916). Ehrenfest, in a letter that no longer exists, presumably asked

Einstein to consider a situation in which light from a distant star passes through one of

Einstein's holes and then strikes a screen with a pinhole in it that directs the light onto

a photographic plate. The question is whether the same point on the photographic plate

would have received the light after a coordinate transformation. In fact, the coordinate

transformation would change the metric in the hole, determining a dierent geodesic

trajectory of the light rays.

In his answer (CPAE 8a, Doc. 180), Einstein imagines representing the situation

described by Ehrenfest on completely deformable tracing paper [Pauspapier]. If one

deforms the tracing paper arbitrarily then one would obtain a solution that is math-

ematically a dierent one from before (CPAE 8a, Doc. 180, p. 238). Deforming the

paper means deforming the coordinate systems, and according to the well-known rules

of tensor calculus, one would obtain a dierent metric eld, and therefore a dierent

geodesic trajectory of light rays. Einstein's worries that this would jeopardize the uni-

vocality of the description of nature disappeared, when he realized that this is only a

mathematical dierence -  physically it is exactly the same. (CPAE 8a, Doc. 180, p.

239; my emphasis). In fact the background coordinate system (the orthogonal drawing

paper coordinate system) with respect to which the situation would have appeared de-

formed is only something imaginary [eingebildetes]: What is essential is this: As long

as the drawing paper, i.e., `space,' has no reality, the two gures do not dier at all. It is

only a matter of coincidences, e.g., whether or not the point on the plate is struck by

light. Thus, the dierence between your solutions A and B becomes a mere dierence of

representation, with physical agreement  (CPAE 8a, Doc. 180, p. 239; my emphasis; tr.

from Howard and Norton 1993).

4.3. Weyl on Leibniz's and Einstein's Indiscernibility Arguments.

Logical Empiricists' interpretation of the point-coincidence argument fails to grasp the

meaning of both versions of the argument (Ryckman, 1992; Howard, 1999). The public

21
point-coincidence argument is not an indiscernibility argument at all. The private in-

discernibility argument is an indiscernibility argument, but, as we shall see, not in the

Leibnizian sense. I argued that the reason for such a failure is not so much the conse-

quence of a rather clumsy attempt to nd eminent precursors of their vericationist

point of view. It is rather the fact that, given the philosophical context in which they

developed their interpretation, they could hardly resist the temptation to interpret the

argument more geometrico , as one of the many Leibniz-style argument that dominated

the 19th century debate on the foundations of geometry. Only recently has the knowledge

of Einstein's correspondence convincingly shown that these remarks of Einstein's link-

ing general covariance and point-coincidences should be understood against a completely

dierent background, one that the Logical Empiricists could not, if not partially, have

known (see Einstein's letter to Schlick, CPAE, Doc. 165; 14 December 1915). However,

it is signicant that it was precisely Hermann Weyl had already clearly seen the dierence

between Leibniz-style and Einstein-style indiscernibility arguments during roughly the

same years when the Logical Positivists were publishing their philosophical reections on

relativity.

Weyl provides on many occasions his own version of a indiscernibility argument á la

Leibniz applied to spacetime: Let us imagine the four-dimensional world as a mass

of plasticine traversed by individual bers, the world lines of material particles (Weyl

1927b, p. 73; tr. Weyl 2009b, p. 105). According to Weyl it is impossible to distinguish
conceptually between the system of lines and the system of curves resulting from them

by a spatial deformation (Weyl 1927b, p. 21; tr. Weyl 2009b, p. 24; my emphasis). In

fact only such relations have an objective signicance as are preserved under arbitrary
deformations of the plasticine. The intersection of two world lines is, for instance, of

this kind (Weyl 1927b, p. 73; tr. Weyl 2009a, p. 129; my emphasis). In this way,

however, the world would be an amorphous continuum without any structure (Weyl

1934/2009, p. 129) or better without any post-dierential structure: Only statements

concerning the distinctness or coincidence of points and the continuous connection of

point conguration can be made at this stage, but it would be impossible to distinguish

the straight lines from the curved ones (Weyl, 1932/2009, p. 41; my emphasis). Thus

every guiding structure, every standard for distinguishing inertial motions and deviation

from the inertial motion would have been lost. The consequence of the Leibnizian strategy

stubbornly pursued by Logical Empiricists is here eectively described by Weyl. Such

a strategy clearly destroys the possibility of dealing with the problem of the relativity

of motion, since no solution of the problem is possible as long as one disregards the

structure of the world (Weyl 1927b, p. 65; tr. Weyl 2009b, p. 105; see Coleman and

Korté 1984).

Einstein's indiscernibility argument, the point-coincidence argument as a response to

the hole argument, cannot therefore be confused with an indiscernibility argument in the

sense of Leibniz. The great achievement of General Relativity, as Weyl never tires of

telling, lies in the fact that the inertial structure of the world is not rigid, but exible,

and changes under material inuences (Weyl, 1934/2009, p. 133). The standard for

distinguishing between inertial and non-inertial motion has itself become dynamical, that

is, in Weyl's parlance, a guiding eld: acceleration means deviation from the trajectories

of particles sub ject only to gravitation, tra jectories that, however, depend in turn on the

contingent distribution of matter. In this context an indiscernibility argument, as Weyl

writes in a nice scientic-philosophical dialog, Massenträgheit und Kosmos , published in

22
1924, serves to avoid the mistake that Einstein committed in 1914 [den gleichen Fehler,

den Einstein 1914 machte] (Weyl, 1924a, p. 202).

The reference is clearly to the hole argument that was rst published in January 1914

(CPAE 4, Doc. 26; January 1914) in the  Bemerkungen added to the article version of

the Entwurf paper written with Grossman (rst published as separatum CPAE 4, Doc.

13; 1913) and repeated in two papers both dating from early 1914 (CPAE 4, Doc. 25,

January 1914; CPAE 6, Doc. 2, May 1914).

In fact Weyl rst observed that  if the matter disappears, the guiding eld must be-

come undetermined  [bei verschwindender Materie muss das Führungsfeld unbestimmt

werden] (Weyl, 1924a, p. 202; my emphasis). The very same physical state can be real-

ized in innite possible mathematical ways.

10 In fact, if the laws of nature are invariant

under arbitrary coordinate-transformations, then I will get from a solution [of the eld

equations] by means of a transformation, innitely many new ones [wenn die Naturgeset-

ze invariant sind gegenüber beliebigen Koordinatentransformationen, so erhalte ich aus

einer Lösung durch Transformation unendlich viele neue] (Weyl, 1924a, p. 203). Weyl's

reference to Einstein's mistake is particularly signicant, since in 1913, as Einstein was

working on the Entwurf theory, Weyl was his colleague at the ETH in Zürich (Weyl came

to Zürich in Fall 1913, whereas Einstein left Zürich for Berlin in Summer 1914).

In Massenträgheit und Kosmos Weyl reformulates an indiscernibility argument à la

Einstein in the following way: I divide the world in two parts through a three-dimensional

cut that separates both its edges [Teile ich die Welt durch einen dreidimensionalen

Querschnitt, welcher ihre beiden Säume, voneinander trennt, . . . in zwei Teile]. Weyl's

reformulation of the hole argument is similar to that of David Hilbert in 1917 (Renn and

Stachel, 2007), in using a open space-like hypersurface (a Cauchy surface) that separates

the future from the past, rather than a closed hypersuface as Einstein. Then, Weyl

continues, if I apply only those [coordinate] transformations that leave unchanged the

part `below' [verwende [Ich] nur solche Transformationen, welche die untere Hälfte

unberührt lassen], but change the metric eld in the part above, then all these solutions

[of the eld equations] will describe also in the underpart the same state evolution as of

the original ones [so stimmen alle diese Lösungen gleichwohl in der unteren Welthälfte

mit der ursprünglichen überein] (Weyl, 1924a, p. 203; emphasis mine). According to

Weyl, Einstein's mistake depends on the fact that he initially overlooked that there

was a dierence only if the four-dimensional world were a resting medium [daÿ ein

Unterschied nur bestünde, wenn die vierdimensionale Welt ein stehendes Medium wäre]

(Weyl, 1924a, p. 203). However, as Weyl immediately emphasizes such a resting Medium

... is completely repudiated by the theory of relativity [Ein solches stehendes Medium

wird aber . . . von der Relativitätstheorie durchaus geleugnet] (Weyl, 1924a, p. 203).

Einstein's kind of indiscernibility argument, as Weyl's exposition shows, does not imply

that the inertial structure has been dissolved in a Leibnizian/Machian way, but that it

has been, as Weyl wrote in an essay of 1925, so to speak, freed from space  [vom Raume

abgelöst]. It has become an existing eld within the remaining structureless space [sie

wird zu einem in dem zurückbliebenden strukturlosen Raume existierenden Feld] (Weyl,

1925/1988, p. 4). Dierences that would appear only with respect to such a structureless

space are not real dierences. On the other hand, Weyl shows that the strategy implied

10  So gibt es, doch unendlich viele Möglichkeiten, wie sich dieser Zustand im Weltcontinnum realisieren
kann (Weyl, 1924a, p. 202; my emphasis)

23
in Leibniz's indiscernibility arguments, to which the Logical Empiricists resorted, has

simply, as we may say, freed space from the minimum of structure that can play the role

of inertial guidance. In these two opposite manners of grasping Einstein's famous remark

that spacetime has lost its last vestige of physical reality, one can recognize most easily

the dierent consequences of Leibniz's and Einstein's indiscernibility arguments.

It may be worth mentioning, even if only in passing, that Weyl's insistence on the fact

that, in General Relativity, the metric eld has been freed from the manifold (Korté,

2006, pp. 193, 201) should probably be understood in the context of his attempt to dene

mathematically the concept of manifold set-theoretically, via the concept of neighbor-

hood (Umgebung , Weyl 1913). Such an attempt, initiated by Hilbert (1902) in the

same years, was, as Weyl recognizes, was pursued more systematically [Systematischer]

by Hausdor in his Grundzügen der Mengenlehre (1914) (Weyl 1925/1988, p. 4; the

reference is to Hausdor 1914/2002). By contrast Einstein's work is based on an older

approach, where the manifold was considered as a number manifold, as the manifold

of all possible values of x, y, z, t. Thus it could be argued, that Einstein was forced to

introduce his indiscernibility argument precisely in order to wash out this additional

structure (Norton, 1989, 1992, 1999).

5. On Learning from the Mistakes of Logical Empiricists. Some Lessons for

the Recent Debate

My attempted historical-critical analysis of the Logical Empiricists' misunderstanding

of the point-coincidence argument as an indiscernibility argument in the sense of Leibniz

is, in my opinion, instructive for the animated debate stirred by Earman and Norton's

fundamental paper (1987). Weyl's version of the hole argument is astonishingly similar

to that of Earman and Norton (much more so than Einstein's original version), and also

addresses the same substantivalist opponents; but at same time Weyl is also careful

to distinguish the hole/point-coincidence argument from a Leibnizian indiscernibility

argument.

The point-coincidence argument cannot be simply considered a stronger version of a

famous argument due to Leibniz himself against Newton's substantival ontology of space

(Janssen, 2005, p. 74). The analogy between Leibniz's and Einstein's arguments could

actually appear prima facie very plausible. Leibniz considered two material universes as

indistinguishable or as the  same universe. Einstein, roughly two centuries later, found

himself likewise considering two eld congurations as empirically indistinguishable and

thus physically identical. In both cases there are apparently dierent  possible worlds

allowed by the theory that actually correspond to the same physical reality. Leibniz

referred to alternative worlds that dier from the actual one only in position, orientation

or magnitude, but agree in the measure of the angles and proportions of lengths. Einstein

could imagine alternative worlds that agree exactly with the actual world outside the

hole, while diering within the hole. In both cases what seems at rst sight a dramatic

dierence reveals itself actually as being no dierence at all, since the dierence is declared

irrelevant.

However, the similarity between the two arguments is only apparent. The dierence

should strike the reader when they simply consider the dierent structures of the two

arguments. Leibniz-style arguments, as we have seen, always presuppose a transformation

that aects every physical entity without exception ; if this condition is violated, the two

24
situations could be easily distinguished. But Einstein's argument explicitly violated this

condition, allowing a transformation that leaves everything unchanged with the exception
of a region devoid of matter. Of course both are indiscernibility arguments. They both

aim to ctionally eliminate the reference to a xed standard against which the change

can be measured, so that the apparent change would reveal itself to be no change at

all. However, such a standard is clearly dierent in each case: Leibniz-style arguments

dissipate the illusion of a transformation that would appear only with respect to some

physical entity; Einstein's argument dissolves the appearance of a transformation that

would emerge, only while referring to the rigid geometry of the empty space.

The now usual exposition of physical theories in terms of models, where the models

are intended to represent the physically possible worlds that satisfy the laws of the

theory, can be useful to see this point. I will provide a slightly more formal presentation

in the appendix. Each model comprises a dierential manifold and various geometric

objects on it, such as metric and matter elds. Indiscernibility arguments in the sense of

Leibniz work well in a theory with global symmetries that presupposes that all relevant

geometrical structure of space appears  the same in all universes or models governed

by the theory, so that the theory does not have the conceptual resources to distinguish

among them. In General Relativity, on the contrary, there are no non-trivial symmetries,

except identity. Therefore there is no spacetime background that would look the  same

across all possible universes or models allowed by the theory, so that any reference to

such a background cannot be used to distinguish among them.

Thus in theories where spacetime is endowed with global symmetries, it makes perfect

sense to apply Leibniz's indiscernibility arguments. Such arguments after all simply

postulate a trivial identity (Stachel and Iftime, 2005) of all models of the theory: it is, so

to speak, the very same model or better - in Leibniz's terms - models that are dierent solo
numero . On the contrary, Einstein's indiscernibility arguments make sense in theories

without global symmetries, where one has to deal with a plurality of dierent models that

are declared non-trivially equivalent, although they show, to resort to Leibniz's parlance

again, a more than numerical dierence.

As we have seen, Leibniz's arguments serve to identify the physically relevant geo-

metrical structure of a theory, so that physical dierences that do not nd expression

in such a structure should not be considered dierences. Einstein's argument signals

on the contrary the presence of a surplus mathematical structure, so that dierences

with respect to such a structure cannot be considered physically meaningful. In the rst

case we have dierent physical situations expressed by the same mathematical model, in

the second case dierent mathematical models that express the same physical situation

(the very same inertio-gravitational eld). Thus in Leibniz-style thought experiments

worlds that at rst sight physically dierent turn out to be mathematically identical ;
in the hole argument apparently mathematically dierent worlds reveal themselves as

physically identical.
If one can speak in both cases of indiscernibility, it seems to me that we have to do

with dierent forms of indiscernibility: (1) an indiscernibility that arises because there

is too little structure to express some alleged physical dierences - dierences that might

otherwise be thought to have physical signicance are therefore declared mathematically

irrelevant ; (2) an indiscernibility that arises, because there is too much structure, from

which apparent mathematical dierences emerge, that are however declared physically
irrelevant, since they express the same physical situation in reality. In the rst case, one

25
might say, indiscernibility is the consequence of underdetermination, since the theory

does not have the tools to express at rst sight real dierences. In the second case, we

nd overdetermination, because the mathematical apparatus of the theory introduces

dierences that do not have any correspondence in reality. In my opinion simply re-

garding the point-coincidence argument as a restatement of  Leibniz's equivalence, as

it is ritually repeated in the literature, would miss the dierence between these two

sorts of indiscernibility arguments: dierent physical situations that are declared math-
ematically indiscernible and dierent mathematical ob jects that are declared physically
indiscernible.

6. Conclusion. Leibniz Equivalence vs. Einstein Equivalence

The historical reconstruction I have attempted should have shown that neither the

public version of the point-coincidence argument, expressing the requirement of general

covariance, nor its private version, a response to the hole argument, can be interpreted

simply as a restatement of indiscernibility in the sense of Leibniz, without committing the

mistakes of the Logical Empiricists. Leibnizian indiscernibility arguments misleadingly

induced the Logical Empiricists to declare generally relativistic spacetime metrically

amorphous, as Adolf Grünbaum famously put it. The point-coincidence argument,

as a response to the hole argument, shows on the contrary that the main feature of

General Relativity is best summarized by John Stachel's celebrated motto: no metric,

no spacetime.

The Logical Empiricists believed that General Relativity was the result of a sort of

Leibniz's equivalence stricto sensu : in the Logical Empiricist interpretation of the point-

coincidence argument dieomorphisms play exactly the same role that translations

or scaling play in Leibniz's arguments, that is they expresses a global symmetry of

space or spacetime. However, such an approach clearly failed to give a plausible account

of General Relativity. Einstein's equivalence (as we may call it) must therefore have

a dierent meaning and dieomorphism must play a completely dierent role, a role

that it is more similar to that which it is usually called gauge freedom, akin to that of

electrodynamics (Giulini and Straumann 2006, p. 151). Just as in electrodynamics where

the same physically measurable eld strengths can be expressed by several potentials,
11
similarly in General Relativity an entire equivalence class of dieomorphically-related

solutions to the eld equations should correspond to one inertio-gravitational eld.

11 The freedom to choose among gauge-equivalent potentials is not a physical degree of freedom: it
rather results from the fact that we have many distinct mathematical objects all of which represent the
same physical state of aairs  (Maudlin, 2002, p. 2, my emphasis; see also Belot, 1998). Another usual
example of the role of surplus structure in physical theories is to be found in modern particle physics.
→
−
Let a single, free non-relativistic particle be described by the wave function ψ(x) . Multiplying this wave
function by a complex number of unit modulus, a phase factor of the form eιθ , gives a wave function
−
→ →
− →
− →
−
ψ ′ (x) = eιθ ψ(x) : ψ(x) and ψ ′ (x) dier mathematically by an overall global phase. However, they
represent physically the same quantum state: the probability distribution for position and momentum
and the time evolution of probability distribution would be the same. As these well-known examples
show gauge freedom arises because the mathematical formalism introduces dierences that are physically
meaningless. At the contrary, global spacetime symmetries arise because certain empirical dierences
(of position, velocity etc.) are not allowed to appear in the mathematical expression of any physical law
(see also Brading and Brown, 2004).

26
 We cannot enter the discussion on the legitimacy of such an analogy between General

Relativity and gauge theory (Weinstein, 1999), on which Norton himself, has insisted

(Norton, 2003). However, the simple possibility of establishing such an analogy, shows

the dierence between Leibniz equivalence and Einstein equivalence. Leibniz equiv-

alence erases dierences: indiscernibility results from the acknowledgment of a lack of

mathematical structure that serves to express such dierences. On the contrary, Einstein

equivalence declares dierences redundant; indiscernibility follows from the acknowledg-

ment of an excess of mathematical structure. Leibniz indiscernibility arguments, we may

say, try to convince us that are no dierences at all. It is more a kind of Leibniz identity

than a Leibniz equivalence. The Einstein indiscernibility argument wants to persuade

us not to worry if the dierences appear to be too many.

Appendix. Leibniz Equivalence vs. Einstein Equivalence in Terms of Space-

time Models

The use of spacetime models oers a very simple way to grasp what I think it is

the theoretical core of the historical reconstruction I suggested: the dierence between

Leibniz and Einstein indiscernibility, or between Leibniz and Einstein equivalence. Even

if the symbolic apparatus is now more than commonplace in philosophical discussion, I

will provide at least a rapid overview. Following the current parlance stemming from

(Hawking and Ellis, 1974, ch. 3; see also Wald) models of a spacetime theory consist

of a manifold, a metric with Lorentz signature, and optionally one or more matter elds

(electromagnetic eld, neutrino eld etc.) that can be regarded as material content of

spacetime. Each of such elds is assumed to satisfy the eld equations. The models of a

theory are those that satisfy such partial dierential equations.

From this point of view, one can easily see the dierence between a theory like Special

Relativity (SR) and General Relativity (GR), and the reason why Leibniz indiscernibility

arguments apply only to the former and Einstein indiscernibility argument only to the

latter kind of theory. In pre-general-relativistic theories one always has an a priori

chrono-geometrical structure, that is one always knows what the geometry is, independent

of obtaining any solution to the equations of motion. In General Relativity, on the other

hand, the relevant geometrical structure has no a priori prescribed values, but rather

obeys the equations of motion.

A model of SR has the form ⟨M, ηµν, , F 1, , F 2, ..., F n, ⟩ where M, ηµν , and Fi , represent
the spacetime manifold, the metric eld and the other elds (gravitational, electromag-

netic . . . ) respectively. SR exhibits global symmetries, because of the invariance of the

ηµν : in all models allowed by the theory ηµν = (−1, 1, 1, 1). Thus there is a xed space-
time structure, the metric structure ηµν, , whose ane structure Γλ µν = 0 represents a
xed standard of non-acceleration. This structure will appear the same in all possible

worlds.

A model of GR is given by a triple M = ⟨M, gµν , Tµν ⟩ where M, gµν , and Tµν
represent the spacetime manifold, the metric eld, and the stress-energy eld respectively.

Such models are taken to represent the physically possible worlds of General Relativity

when they satisfy Einstein's eld equations Gµν = κTµν (where Gµν is the Einstein

tensor describing the curvature of spacetime, and κ the coupling constant, proportional

to Newton's universal constant of gravitation). In contrast to SR, the gµν are then

subjected to the equations of the theory: they have become a eld among others. In

27
GR the symmetry group of ⟨M, gµν ⟩ is the identity group, since there is in general no
transformation that leaves gµν invariant. Hence there will be in general dierent models
′ ′′
of the theory: ⟨M, gµν ⟩, ⟨M, g µν ⟩, ⟨M, gµν ⟩, ... The Christoel symbols Γ µν gure as
λ

the components of the gravitational-inertial eld (so they are in general ̸= 0), whereas the
ρ
Curvature tensor or Riemann-Christoel tensor (Bµστ ) represents the eld gradient, or
λ
tidal eld. The fact that Γ µν is not a tensor is the formal expression of the non-unique

decomposition of the ane connection into an inertial and a gravitational part.

The analogy of the Leibniz shift can be found only in pre-relativistic theories, such

as SR: it can be represented by a map h between worlds or models (spatial or temporal

displacements, rotations, or boosts up to the speed of light) that preserves all relevant
geometrical structure (in the case of SR the metric structure uniquely determining the

inertial structure), that is h ∗ ηµν = ηµν (an isometry). If such a structure would not be

preserved, that is if h ∗ ηµν ̸= ηµν , the two universes or models would not be indistin-

guishable. The indistinguishability arises because the transformation produces the very
same model (it is a trivial identity). The original and the transformed situations are

indiscernible simply because they are mathematically exactly the same. The ηµν are the

only quantities pertaining to spacetime structure which can appear in any physical law.

Thus dierences that cannot be encoded in such a spacetime structure are not dierences.

In order to introduce further distinctions, for instance of spatial orientation, one has to

allow further aspects of spacetime structure to appear in physical laws (Wald, 1984, p.

60).

Einstein's hole/point-coincidence argument makes sense only in GR

12 : it implies a

transformation h (dif f (M)) that does not preserve the relevant geometrical structure
(the metric structure determining the geodesics), that is h∗gµν ̸= gµν . An indiscernibility

argument in the sense of Einstein is needed because the theory introduces dierent models
that, however, are declared physically exactly the same : the same gravitational eld

corresponds to an equivalence class of ⟨M, gµν ⟩- it is so to speak, a non-trivial equivalence.

The Logical Empiricists clearly confused Leibniz and Einstein indiscernibility (or bet-

ter, had at their disposal only Leibniz indiscernibility): noticing that the group of trans-

formations of GR, does not preserve the metric gµν , (h ∗ gµν ̸= gµν ), they declared the
metric conventional and identied spacetime with the invariant M held across models.
Dierences that cannot be expressed in M, like in a Hegelian night, are not dierences:
dif f (M) with a global spacetime symmetry. Einstein's indiscernibil-
that is they identify

ity argument moving from h∗gµν ̸= gµν concludes that spacetime is the equivalence class
′
(up to dieomorphism) of all ⟨M, gµν ⟩,⟨M, gµν ⟩, ..., so that dierences with respect to

M are not dierences, they are a mere mathematical redundancy: dif f (M) resembles
more closely an expression of gauge freedom (a whole family of gauge-related solutions

of the eld equations represent the same physical situation) (Wald, 1984, p. 438).

References

A: Leibniz, G. W., 1923. Sämtliche Schriften und Briefe. O. Reichl, Darmstadt.

12 For sake of comparison I follow here John Stachel in arguing that the hole argument cannot be
taken over to the pregeneral-relativistic case (Stachel, 1993). This opinion is however controversial. I
cannot however address this issue here. My aim is rather to emphasize the opposite issue that Leibniz
arguments work only in pregeneral-relativistic theories.

28
Anderson, J. L., 1967. Principles of Relativity Physics. Acad. Pr., New York [et al.].
Arthur, R., 1994. Space and Relativity in Newton and Leibniz. The British Journal for the Philosophy
of Science 45 (1), pp. 219240.
Bartels, A., 1994. What is Spacetime, if not a Substance? Conclusions from the New Leibnizian Ar-
gument. In: Majer, U., Schmidt, H.-J. (Eds.), Semantical Aspects of Spacetime Theories. BI Wis-
senschaftsverlag (1994), Mannheim, pp. 4151.
Bartels, A., 1996. Modern Essentialism and the Problem of Individuation of Spacetime Points. Erkenntnis
45, 2543.
Belot, G., 1998. Understanding Electromagnetism. The British Journal for the Philosophy of Science
49 (4), pp. 531555.
Brading, K., Brown, H. R., 2003. Symmetries and Noether's theorems. In: Symmetries in Physics.
Philosophical Reections. Cambridge University Press, Cambridge, U.K.; New York, pp. 87109.
Brading, K., Brown, H. R., 2004. Are Gauge Symmetry Transformations Observable? British Journal
for the Philosophy of Science 55, 645665.
Brown, H. R., 2005. Physical relativity : space-time structure from a dynamical perspective, 1st Edition.
Clarendon, Oxford.
Buttereld, J., 1989. The Hole Truth. The British Journal for the Philosophy of Science 40 (1), 128.
Cantor, G., 1874. Ueber eine Eigenschaft des Inbegris aller reellen algebraischen Zahlen. Journal für
die reine und angewandte Mathematik (Crelles Journal) 1874 (77), 258262.
Cantor, G., 1878. Ein Beitrag zur Mannigfaltigkeitslehre. In: Zermelo, E. (Ed.), Gesammelte Abhand-
lungen mathematischen und philosophischen Inhalts. No. 119-133. Olms, Hildesheim.
Carnap, R., 1922. Der Raum. Ein Beitrag zur Wissenschaftslehre. Reuther & Reichard, Berlin.
Carnap, R., 1966. Philosophical Foundations of Physics: An Introduction to the Philosophy of Science.
Basic Books, inc., New York.
Cherno, F., 1981. Leibniz's Principle of the Identity of Indiscernibles. The Philosophical Quarterly
31 (123), 126138.
Coa, A. J., 1991. The Semantic Tradition from Kant to Carnap: To the Vienna Station. Cambridge
University Press, Cambridge.
Coleman, R. A., Korté, H., 1984. Constraints on the Nature of Inertial Motion Arising from the Uni-
versality of Free Fall and the Conformal Causal Structure of Space-Time. Journal of Mathematical
Physics 25, 35133527.
Couturat, L., 1961. La logique de Leibniz d'après des documents inédits. Olms, Hildesheim.
Cox, C. B., 1978. Leibniz's Philosophy of Space. Diss. - Columbia University, New York.
CPAE: Einstein, A., 1996-. The Collected Papers of Albert Einstein. Princeton University Press, Prince-
ton.
De Risi, V., 2005. Leibniz on Geometry: Two Unpublished Texts with Translation and Commentary.
The Leibniz Review 15, 127132.
De Risi, V., 2007. Geometry and Monadology: Leibniz's analysis situs and Philosophy of Space.
Birkhäuser, Basel; Boston.
DiSalle, R., 2002a. Conventionalism and Modern Physics. A Re-Assessment. Noûs 36 (2), 169200.
DiSalle, R., 2002b. Newton's Philosophical Analysis of Space and Time. In: Cohen, I. B. (Ed.), The
Cambridge Companion to Newton. Cambridge University Press, Cambridge, pp. 3354.
DiSalle, R., 2006. Understanding Space-Time: The Philosophical Development of Physics from Newton
to Einstein. Cambridge University Press, Cambridge, UK ; New York.
Earman, J., 1989. World Enough and Space-Time: Absolute vs. Relational Theories of Space and Time.
MIT Press, Cambridge, Mass.
Earman, J., Norton, J. D., 1987. What Price Substantivalism. The Hole Story. British Journal for the
Philosophy of Science 38, 515525.
Ehler, J., 1988. Hermann Weyl's Contributions to the General Theory of Relativity. In: Deppert, W. H.
(Ed.), Exact sciences and their philosophical foundations: Vorträge des Internationalen Hermann-
Weyl-Kongresses, Kiel 1985 = Exakte Wissenschaften und ihre philosophische Grundlegung. Lang,
Frankfurt am Main [et.al.].
Einstein, A., 1921. Geometrie und Erfahrung. Springer, Berlin.
Einstein, A., 1925. Nichteuklidische Geometrie und Physik. Die neue Rundschau 36, 1620.
Epple, M., 2006. Felix Hausdor 's Considered Empiricism. In: Ferreirós, J., Gray, J. (Eds.), The Archi-
tecture of Modern Mathematics: Essays in History and Philosophy. Oxford University Press, Oxford.
Epple, M., 2007. An Unusual Career between Cultural and Mathematical Modernism: Felix Hausdor,
1868-1942. In: Charpa, U., Deichmann, U. (Eds.), Jews and Sciences in German Contexts: Case
Studies from the 19th and 20th centuries. Mohr Siebeck, Tübingen, pp. 77100.

29
Freudenthal, H., 1972. Leibniz und die analysis situs. Studia Leibnitiana 4 (2), 6160.
Friedman, M., 1983. Foundations of Space-Time Theories: Relativistic Physics and Philosophy of Science.
Princeton University Press, Princeton.
Friedman, M., 1995. Poincaré's Conventionalism and the Logical Positivists. Foundations of Science 2,
299314, now in [Friedman 1999].
Friedman, M., 1999. Reconsidering Logical Positivism. Cambridge University Press, New York.
Galilei, G., 1632/2005. Discorso intorno a due nuove scienze. In: Brunetti, F. (Ed.), Opere. UTET,
Torino.
Galilei, G., 1967. Dialogue Concerning the Two Chief World Systems, Ptolemaic and Copernican. Uni-
versity of California Press, Berkeley.
Giulini, D., 2007. Remarks on the Notions of General Covariance and Background Independence. Lecture
Notes in Physics 721, 105120.
Giulini, D., Straumann, N., 2006. Einstein's Impact on the Physics of the Twentieth Century. Studies
In History and Philosophy of Science Part B: Studies In History and Philosophy of Modern Physics
37 (1), 115  173, 2005: The Centenary of Einstein's Annus Mirabilis.
GM: Leibniz, G. W., 1850. Leibnizens mathematische Schriften. Schmidt, Halle.
GP: Leibniz, G. W., 1875. Die philosophischen Schriften von Gottfried Wilhelm Leibniz. Weidmann,
Berlin.
Hausdor, F., 1903. Das Raumproblem. Annalen der Naturphilosophie 3, 123.
Hausdor, F., 1914/2002. Grundzüge der Mengenlehre. In: Brieskorn, E., Chatterji, S. D., Epple, M.
(Eds.), Gesammelte Werke. "Grundzüge der Mengenlehre". Vol. II. Springer, Berlin, Heidelberg etc.
Hausdor, F., 1927, 1935/2007. Mengenlehre. In: Brieskorn, E., Hirzebruch, F., Remmert, R., Purkert,
W., Scholz, E. (Eds.), Gesammelte Werke: Deskriptive Mengenlehre und Topologie. Vol. III. Springer,
Berlin, Heidelberg etc.
Hausdor, F., 2004. Das Chaos in kosmischer Auslese. Ein erkenntnisskritischer Versuch. In: Stegmaier,
W. (Ed.), Gesammelte Werke: Philosophisches Werk. Vol. VII. Springer, Berlin.
Hausdor, F., 2010. Gesammelte Werke: Geometrie, Raum und Zeit. Vol. VI. Springer, Berlin, Heidel-
berg etc.
Hawking, S. W., Ellis, G. F. R., 1974. The large scale structure of space-time. Cambridge monographs
on mathematical physics ; 1. Cambridge Univ. Press, Cambridge.
Hilbert, D., 1902. Über die Grundlagen der Geometrie. Mathematische Annalen 56, 381422.
Hoefer, C., 1996. The Metaphysics of Space-Time Substantivalism. The Journal of Philosophy 93 (1),
pp. 527.
Howard, D., 1999. Point Coincidences and Pointer Coincidences. Einstein on the Invariant Content of
Space·Time Theories. In: Goenner, H. (Ed.), The Expanding Worlds of General Relativity. Birkhäuser,
Basel, pp. 463500.
Howard, D., 2005. 'No Crude Surfeit': A Critical Appreciation of The Reign of Relativity"., delivered
at 2005 Pacic meeting of the American Philosophical Assocation, manuscript.
Howard, D., Norton, J. D., 1993. Out of the Labyrinth. Einstein, Hertz, and the Gottingen Answer to
the Hole Argument. In: The attraction of gravitation : new studies in the history of general relativity.
Birkhäuser, Basel, pp. 3062.
Huggett, N., Hoefer, C., 2009. Absolute and Relational Theories of Space and Motion. In: Zalta, E. N.
(Ed.), The Stanford Encyclopedia of Philosophy, fall 2009 Edition.
Janssen, M., 2005. Of pots and holes: Einstein's bumpy road to general relativity. Annalen der Physik
14, 5885.
Jauernig, A., 2008. Leibniz on Motion and the Equivalence of Hypotheses. The Leibniz Review 18, 140.
Klein, C., 2001. Conventionalism and Realism in Hans Reichenbach's Philosophy of Geometry. Interna-
tional Studies in the Philosophy of Science 15 (3), 243251.
Korté, H., 2006. Einstein's Hole Argument and Weyl's Field-body Relationalism. In: Demopoulos, W.,
Pitowsky, I. (Eds.), Physical Theory and its Interpretation. Vol. 72 of The Western Ontario Series in
Philosophy of Science. Springer Netherlands, pp. 183212.
Kosso, P., 2000. The Empirical Status of Symmetries in Physics. The British Journal for the Philosophy
of Science 51 (1), 8198.
Kretschmann, E., 1917. Über den physikalischen Sinn der Relativitätspostulate. A. Einsteins neue und
seine ursprüngliche Relativitätstheorie. Annalen der Physik 53, 575614.
Lariviere, B., 1987. Leibnizian Relationalism and the Problem of Inertia. Canadian Journal of Philosophy
17 (2), 437448.
Lehrer, K., 2004. Reichenbach on Convention. Synthese 34/2, 237248.
Leibniz, G. W., 1976. Philosophical Papers and Letters: A Selection, 2nd Edition. Synthese historical

30
 library ; 2. Riedel, Dordrecht.
Leibniz, G. W., 1989. Philosophical Essays. Hackett Pub. Co., Indianapolis, IN.
Maudlin, T., 1988. The Essence of Space-Time. SA: Proceedings of the Biennial Meeting of the Philos-
ophy of Science Association 2, 291.
Maudlin, T., 1993. Buckets of Water and Waves of Space. Why Spacetime Is Probably a Substance.
Philosophy of Science 60 (2), 183203.
Maudlin, T., 2002. Thoroughly Muddled McTaggart. Or, How to Abuse Gauge Freedom to Create
Metaphysical Monostrosities. Philosophers' Imprint 2 (4), 119.
Münzenmayer, H. P., 1979. Der Calculus Situs und die Grundlagen der Geometrie bei Leibniz. Studia
Leibnitiana 11, 274300.
Mongré, P., 1898. Das Chaos in kosmischer Auslese. Ein erkenntnisskritischer Versuch. Naumann,
Leipzig, now in Hausdor [2004].
Nerlich, G., 1994. What Spacetime Explains: Metaphysical Essays on Space and Time. Cambridge Univ.
Press, Cambridge [et al.].
NL FH: Hausdor, F., -. Nachlass, the Nachlass is located at the Main Library of
the University of Bonn (D). An online catalogue can be found at http://www.aic.uni-
wuppertal.de/fb7/hausdor/ndbuch.asp. The papers on philosophy of geometry will be published
in Hausdor [2010].
Norton, J. D., 1984. How Einstein found his Field Equations: 1912-1915. Historical Studies in the
Physical Sciences 14, 253316.
Norton, J. D., 1989. Coordinates and Covariance: Einstein's View of Spacetime and the Modern View.
Foundations of Physics 19, 12151263.
Norton, J. D., 1992. The Physical Content of General Covariance. In: Studies in the History of General
Relativity: Einstein Studies. Birkhauser, Boston, pp. 281315.
Norton, J. D., 1995. Did Einstein Stumble? The Debate over General Covariance. Erkenntnis 42 (2),
223245.
Norton, J. D., 1999. Geometries in Collision: Einstein, Klein and Riemann. In: Gray, J. (Ed.), The
Symbolic Universe. Oxford University Press, Oxford; New York, pp. 128144.
Norton, J. D., 2003. General covariance, gauge theories, and the Kretschmann objection. In: Brading,
K., Castellani, E. (Eds.), Symmetries in Physics. Philosophical Reections. Cambridge University
Press, Cambridge, pp. 110123.
Pauli, W., 1921. Relativitätstheorie. In: Enzyklopädie der mathematischen Wissenschaften. Vol. 5. Teub-
ner, pp. 539775.
Pauli, W., 1958. Theory of Relativity. Pergamon Press, London.
Pitts, J. B., 2006. Absolute Objects and Counterexamples: Jones-Geroch dust, Torretti Constant Curva-
ture, Tetrad-Spinor, and Scalar Density. Studies In History and Philosophy of Science Part B: Studies
In History and Philosophy of Modern Physics 37 (2), 347  371.
Pulte, H., 2006. The Space between Helmholtz and Einstein: Moritz Schlick on Spatial Intuition and the
Foundations of Geometry. In: Hendricks, V. F., Jørgensen, K. F., Lützen, J., Pedersen, S. A. (Eds.),
Interactions Mathematics, Physics and Philosophy, 1860-1930. Springer, Dordrecht.
Reichenbach, H., 1920/1977. Relativitätstheorie und Erkenntnis a priori. In: Kamlah, A., Reichenbach,
M. (Eds.), Gesammelte Werke in 9 Bänden. Vol. 3. Braunschweig; Wiesbaden, Vieweg.
Reichenbach, H., 1924. Axiomatik der relativistischen Raum-Zeit-Lehre. Braunschweig, Vieweg.
Reichenbach, H., 1924/1969. Axiomatization of the Theory of Relativity. University of California Press,
Berkeley,.
Reichenbach, H., 1928/1958. The Philosophy of Space and Time. Dover Publ., New York.
Reichenbach, H., 1928/1977. Die Philosophie der Raum-Zeit-Lehre. In: Kamlah, A., Reichenbach, M.
(Eds.), Gesammelte Werke in 9 Bänden. Vol. 2. Vieweg, Braunschweig; Wiesbaden.
Reichenbach, H., 1949/1951. The Philosophical Signicance of the Theory of Relativity. In: Schilpp,
P. A. (Ed.), Albert Einstein. Philosopher-Scientist. Tudor, New York, pp. 289311.
Renn, J., Stachel, J., 2007. Hilbert's Foundation of Physics: From a Theory of Everything to a Con-
stituent of General Relativity. In: Renn, J., Janssen, M. (Eds.), The Genesis of General Relativity.
Springer Netherlands, pp. 857972.
Roberts, J. T., 2003. Leibniz on Force and Absolute Motion. Philosophy of Science 70 (3), pp. 553573.
Rosen, J., 2008. Symmetry Rules: How Science and Nature Are Founded on Symmetry. Springer, Berlin;
Heidelberg.
Ryckman, T., 1992. (P)oint-(C)oincidence Thinking. The Ironical Attachment of Logical Empiricism to
General Relativity. Studies in History and Philosophy of Modern Physics 23 (3), 471497.
Ryckman, T., 1995. Weyl, Reichenbach and the Epistemology of Geometry. Studies in History and

31
 Philosophy of Science 25 (6), 831870.
Ryckman, T., 1996. Einstein Agonists: Weyl and Reichenbach on Geometrv and the General Theory
of Relativity. In: Giere, R. N. (Ed.), Origins of Logical Empiricism. University of Minnesota Press,
Minneapolis, pp. 165209.
Ryckman, T., 2005. The Reign of Relativity. Philosophy in Physics 1915-1925. Oxford University Press,
Oxford, New York.
Ryckman, T., 2007. Logical Empiricism and Philosophy of Physics. In: Richardson, A. W., Uebel, T.
(Eds.), The Cambridge Companion to Logical Empiricism, 1st Edition. Cambridge University Press,
Cambridge [et al.], pp. XI, 430 S.
Ryckman, T., 2008. Early Philosophical Interpretations of General Relativity. In: Zalta, E. N. (Ed.), The
Stanford Encyclopedia of Philosophy: Fall 2008 Edition. plato.stanford.edu/entries/genrel-early/.
Rynasiewicz, R., 1994. The Lessons of the Hole Argument. The British Journal for the Philosophy of
Science 45 (2), 407436.
Rynasiewicz, R., 1999. Kretschmann's Analysis of Covariance and Relativity Principles. In: Goenner,
H. (Ed.), The Expanding Worlds of General Relativity. Birkhäuser, Basel, pp. 431462.
Rynasiewicz, R., 2000. On the Distinction between Absolute and Relative Motion. Philosophy of Science
67 (1), 7093.
Saunders, S., 2002. Indiscernibles, General Covariance, and other Symmetries. In: Ashtekar, A. V., Renn,
J., Howard, D. (Eds.), Revisiting the Foundations of Relativistic Physics: Festschrift in Honour of
John Stachel. Kluwer, Dordrecht/Boston/London, pp. 151173.
Schlick, M., 1920. Space and Time in Contemporary Physics, an Introduction to the Theory of Relativity
and Gravitation. Oxford University Press, Oxford/New York.
Schlick, M., 1922/2006. Raum und Zeit in der gegenwärtigen Physik. In: Neuber, M., Engler, O. (Eds.),
Gesamtausgabe. Vol. I,2. Springer, New York/Berlin.
Schlick, M., 1978a. Philosophical Papers. Vienna Circle collection ; 11. Reidel, Dordrecht.
Schlick, M., 1978b. Space and Time in Contemporary Physics. In: Mulder, H. L. (Ed.), Philosophical
Papers. Vol. 1 of Vienna Circle collection ; 11. Reidel, Dordrecht.
Schneider, M., 1988. Funktion und Grundlegung der Mathesis Universalis. Studia Leibnitiana Sonderheft
15, 162182, (Leibniz: Questions de Logique).
Slowik, E., 2006. The "dynamics" of Leibnizian relationism: Reference frames and force in Leibniz's
plenum. Studies In History and Philosophy of Science Part B: Studies In History and Philosophy of
Modern Physics 37 (4), 617  634.
Stachel, J., 1980/2002. Einstein's Search for General Covariance, 1912-1915. In: Einstein from "B" to
"Z". Birkhäuser, Boston, pp. 301337, the paper was rst paper read at the Ninth International
Conference on General Relativity and Gravitation, Jena in 1980.
Stachel, J., 1993. The Meaning of General Covariance: The Hole Story. In: Earman, J. (Ed.), Philo-
sophical problems of the internal and external worlds : essays on the philosophy of Adolf Grünbaum.
Pittsburgh-Konstanz series in the philosophy and history of science ; [1]. Univ. of Pittsburgh Pr.,
Pittsburgh, Pa., festschrift Adolf Grünbaum.
Stachel, J., Iftime, M., 2005. Fibered Manifolds, Geometric Objects, Structured Sets, G-Sets and all
that: The Hole Story from Space Time to Elementary Particles. arxiv.org.
Stachel, J. J., 2002. Einstein from B to Z. Einstein studies ; 9. Birkhäuser, Boston.
Stein, H., 1967/1970. Newtonian Space-Time. In: Palter, R. (Ed.), The Annus mirabilis of Sir Isaac
Newton, 1666-1966. The MIT Press, Cambridge (Ma.)/London, pp. 254284.
Wald, R. M., 1984. General relativity. Univ. of Chicago Pr., Chicago [et al.].
Wallwitz, G. G., 1991. Strukturelle Probleme in Leibniz Analysis Situs. Studia Leibnitiana 23, 111118.
Weinstein, S., 1999. Gravity and Gauge Theory. Philosophy of Science 66, 146155.
Weyl, H., 1913. Die Idee der Riemannschen Fläche, 2nd Edition. Mathematische Vorlesungen an der
Universität Göttingen ; 5. Teubner, Leipzig [et al.].
Weyl, H., 1921/1968. Elektrizität und Gravitation. In: Chandrasekharan, K. (Ed.), Gesammelte Ab-
handlungen. Vol. 2. Springer, Berlin, pp. 141142.
Weyl, H., 1921/2009. Electricity and Gravitation. In: Pesic, P. (Ed.), Mind and Nature: Selected Writings
on Philosophy, Mathematics, and Physics. Princeton Univerity Press, Princeton.
Weyl, H., 1924a. Massenträgheit und Kosmos. Ein Dialog. Naturwissenschaften 12, 11.
Weyl, H., 1924b. Rezension von: H. Reichenbach: Axiomatik der relativistischen Raum- Zeit Lehre.
Deutsche Literaturzeitung. 45, 21222128.
Weyl, H., 1925/1988. Riemanns geometrische Ideen, ihre Auswirkung und ihre Verknüpfung mit der
Gruppentheorie. Springer, New York; Berlin; Heidelberg.
Weyl, H., 1927a. Philosophie der Mathematik und Naturwissenschaft. In: Bäumler, A., Schröter, M.

32
 (Eds.), Handbuch der Philosophie. Olde, München/Berlin.
Weyl, H., 1927b. Zeitverhältnisse im Kosmos, Eigenzeit, gelebte Zeit und metaphysische Zeit. In: Pro-
ceedings of the Sixth International Congress of Philosophy. Harvard University, Cambridge, Mas-
sachusetts, United States of America, September 13-17, 1926. Longmans, Green and Co., Cambridge,
Ma., pp. 5458.
Weyl, H., 1932/2009. The Open World. In: Pesic, P. (Ed.), Mind and Nature: Selected Writings on
Philosophy, Mathematics, and Physics. Princeton Univerity Press, Princeton.
Weyl, H., 1934/2009. Mind and Nature. In: Pesic, P. (Ed.), Mind and Nature: Selected Writings on
Philosophy, Mathematics, and Physics. Princeton University Press, Princeton, pp. 83150.
Weyl, H., 1939/1997. The Classical Groups: Their Invariants and Representations. Princeton University
Press, Princeton.
Weyl, H., 1952. Symmetry. Princeton Univ. Pr., Princeton, NJ.
Weyl, H., 2009a. Mind and Nature: Selected Writings on Philosophy, Mathematics, and Physics. Prince-
ton Univerity Press, Princeton.
Weyl, H., 2009b. Philosophy of Mathematics and Natural Science. Princeton University Press, Princeton.

33