Monist 65 (1983): 302-314.


J. J. C. Smart
The Australian National University


Are words like 'before' and 'after' fundamentally temporal predicates, true of ordered pairs of events, process and the like, or are they fundamentally (non-truthfunctional) sentential connectives? According to surface grammar, of course, it would appear that they can be both, since we have sentences like 'The battle was before the earthquake' and also sentences like 'Socrates drank the hemlock before he died'. But which is fundamental? Which most nearly shows the logical form?1 Most philosophers and scientists have assumed that the former alternative is correct. Wilfrid Sellars opts for the latter alternative in his second Carus lecture 'Naturalism and Process' (hereafter referred to as CL2)2 and in his much earlier 'Time and the World Order' (hereafter referred to as TWO),3 though he concedes that in some ideal future 'scientific image' the former alternative may be appropriate. In what follows I shall be concerned with Sellars's views on 'the manifest image' unless I indicate otherwise.4 Another important philosopher who has elucidated temporal discourse in terms of 'before', 'after', etc.. as sentential connectives is P. T. Geach.5

Contrary to Sellars I do not see the same unbridgeable gulf between our ordinary 'manifest image' talk and that of some ideal 'scientific image'. For example I would hold that a physicalist account of Sellars's well-known pink ice cube6 is possible. More germane to the present purpose, I do not see a wide gulf between our ordinary talk of objects, such as tables, men, cats, stones, clouds, and stars as substances -- the permanent in change -- and Minkowski type talk of these things as four-dimensional space-time solids. I want to say that just as a man is so many centimetres tall, broad and thick, so he is (say) seventy years long in time.7 (Furthermore the special theory of relativity shows how by settling on some value, say unity, for the velocity of light, we can express the man's height, breadth, thickness and endurance in time in the same units.) What then are events and processes? Before giving my answer to this question I shall adumbrate some 'ordinary language' considerations.

We talk of processes changing or becoming something other (as a battle can become fierce). It would be odd, however, to talk of such a change or becoming itself changing or becoming something or other. Many years ago I tried to make an analogy between the "event-process" distinction and Gilbert Ryle's "task-achievement" distinction.8 One can run hard and for a long time, but one can hardly win hard and for a long time. The traffic light can change from red to green, but can the changing from red to green change? It seems that often 'event' works rather like 'result'. This ties in with etymology. where 'event' = 'outcome'. In Sir Walter Scott's Ivanhoe we get a perhaps somewhat archaic use of the word in this sense. Thus in Chapter 16 we have "the event of the tournament" and "he was justified by the event,'' and even more clearly, Chapter 39, "I am . . . friend or foe, Rebecca, as the event of this interview shall make me." Now 'result' has a propositional feel about it. ('What was the result?' 'I passed'.) Such uses of 'event' tie in with Sellars's views.

It is tempting to say that the reason we might not be able to talk of events changing is that they are (or are associated with) mere instantaneous time-slices of four-dimensional objects. To say that a thing changes is to say that earlier and later time slices of it are different, but a single time-slice can not be different from itself. Are events, then, if they are not propositional, perhaps instantaneous time-slices? Consider the beginning of a war. Was there an instant at which the war began? (Indeed, in view of quantum-mechanical indeterminacy, is there a definite instant at which anything begins?) Also we talk of a battle sometimes as if it were an event ('The battle happened in 1915') and sometimes as if it were a process ('The battle grew fiercer'). I am now therefore inclined to suspect that the difference between events and processes is a contextual one, not an ontological one (not a category difference). We speak of a process as an event when we are not much concerned with its inner temporal structure, and as a process when we are so concerned. Events need not be instantaneous. (Nor, I think, despite the archaic use mentioned in the last paragraph, need they be propositional. But I shall give reasons for thinking this later.) The fact that the oddity of talking of events changing may be contextual rather than categorical is indeed brought out by the fact, pointed out by F. I. Dretske in his valuable paper, 'Can Events Move?'9 that we do sometimes talk of a picnic moving indoors, and things of that sort, and it seems artificial to say that 'picnic' is ambiguous, so that it here functions as a process word, not an event word, as when we say that the picnic occurred on a certain date.

If event and processes are not to be distinguished ontologically, nor need states. States of an object can be regarded as unchanging processes, i.e., temporal stages whose time-slices are similar to one another. Could we follow W. V. Quine and say, furthermore, that events and processes need not be distinguished from objects, as conceived four dimensionally? Physical objects, events and processes, alike consist, as Quine put it, of "the content, however heterogeneous, of some portion of space-time, however discontinuous and gerrymandered".10 This view runs up against semantic difficulties, but I want to play with it for a moment or two.

Consider a battle. We might consider this as a space-time object consisting of the mereological sum of all the space-time objects (temporal stages of men, machines, projectiles, etc.) that make it up. Thus we would identify processes with temporal stages of space-time objects. Sometimes such objects cannot be specified separately from their stages. Thus thunder and lightning are processes, but we can say 'It thunders' and 'It lightnings'. We can supply noun phrases for the 'It' by simply saying 'The thunder thunders' and 'The lightning lightnings'." Indeed we could treat a chair as a process that lasts all the time the chair lasts, and say 'The chair chairs'. According to this view the difference between substances and processes is an artefact of language and of contextual interests. Qntologically they are the same.

We now come up against a difficulty for this attractively simple view. Consider Davidson's example of a metal ball that is (a) heating up and (b) rotating.12 There is an inclination to say that we have two processes here. But if we suppose that the heating up begins and ends just when the rotating does, then on the Quinean view I have canvassed we must identify event (a) and event (b) with the same temporal stage of the ball. Hence the events must be the same. On reflection, however, this is not so paradoxical as our first intuitions may suggest. Possibly these intuitions may depend on confusing event types with event tokens, i.e., classes of events with particular concrete events. The particular heating up is in fact a rotating, even though the class of heatings up of the ball may be different from the class of rotatings of the ball. The identity of the rotation with the heating up may be made plausible if we consider the matter from the point of view of the kinetic theory of heat. (I think the point could be made also from the point of view of caloric theory, or of pre-scientific common sense, though perhaps less persuasively.) The heating up is an increasingly violent vibratory motion of the molecules of the ball. But the motion is not only vibratory but is also a revolving about an axis. Similarly the rotation is a mainly revolving motion of the molecules, with vibrations superimposed on it. So the heating up and the rotation are the same. When we are considering the situation from the point of view of the kinetic theory of heat we think of it as a case of the ball getting warmer, and when we are thinking of the situation from the point of view of solving a problem in the mechanics of a rigid body we think of it as a rotation. Forgetting the contextual differences we may come to think that we are concerned with two events not one.

However there does seem to be a difficulty for this view. It arises from adverbs. The ball may heat up steadily and rotate jerkily. So the heating up is steady and the rotation jerky, ft looks as though in ordinary non-canonical language we need an ontology of events distinct from the temporal stages of substances. The difficulty would not, however, arise for science, if this can be expressed in Quinean canonical notation. 'Heating up' would not be an event expression but would be comparable to 'length' and eliminible in favour of the predicate 'heats up in . . . calories per second' just as 'length' is eliminible in favour of 'has a length in centimetres of'. Similarly we need not refer to rotations but get by with the predicate 'rotates at . . . radians per second'.13 However for ordinary language, we do seem to need an ontology that postulates events as a separate category of objects.

One possibility would be to construe events as ordered pairs of temporal stages of objects on the one hand and predicates or properties on the other. This would be to go over to a theory with affinities to those of Jaegwon Kim14 and R. M. Martin.15 According to Kim an event is the exemplification of a property by a substance at a time. (Martin construes an event in somewhat similar fashion, but attempts to elucidate properties nominalistically.) If we do it with properties we have problems about the respectability of these entities, though I am aware that Sellars claims to give a nominalist account of these. If we do it with predicates we run up against the difficulty that synonymous-predicates will distinguish events as different which should not be different. Moreover whether we do it with predicates or with properties we will distinguish events that should not be so distinguished, for example (as Donald Davidson has argued) Brutus' stabbing Caesar and Brutus' killing Caesar.16 Perhaps events should be regarded as an ontological category irreducible to any others as is perhaps Davidson's view. But whether reducible or irreducible these theories hold that events exist and are something other than temporal stages of substances.17


In the previous section I have discussed Quine's account of events and processes as not ontologically different from substances. If this were correct it would follow trivially that events exist. Other theories give a nontrivial account of events. I have mentioned Kim's and Martin's and alluded to Davidson's. Davidson's theory is attractive because of the lucidity of its underlying semantics, which is essentially Tarski's. Davidson's semantic theory tries to give a recursive semantics for a certain class of adverbs by supposing that the logical form of sentences containing adverb is best given by transforming them into sentences quantifying over events and such that the adverbs get replaced by predicates of events. Davidson also applies this sort of idea to other sorts of sentences, such as causal ones. Now it is possible that a different approach to the semantics of ordinary language might be possible, which would not quantify over events, but which would yield a truth theory of the appropriately recursive sort. (That is, which will show how the truth conditions of any of the infinitely many sentences of a language depend recursively on the satisfaction conditions for a finite number of constituents.) But as Davidson remarks, it is extraordinary how smoothly things work if one is allowed to quantify over events.18

Sellars's theory might be called a "no event" theory. He says (CL2 § 29) that "there are no events in addition to changing things and persons." Of course if the Quinean suggestion that events are temporal stages of objects were acceptable there would be no disagreement with the above quotation from Sellars, but the real difference comes out in the next paragraph, CL2 §30, where Sellars asserts that "There are no temporal relations." Sellars rejects the four-dimensional view of substances and he eschews temporal relations (use of words like 'before', 'after' as predicates) in favour of using such words as sentential connectives. Or rather, if such words occur as grammatical predicates, as in 'Socrates' drinking the hemlock was before his death', the sentences in which they occur give surface structure only, and the deep structure is better given by 'Socrates drank the hemlock before he died'. Despite the difficulties raised by Davidson against eschewing quantification over events, Sellars's view has some initial plausibility. After all it seems highly plausible that children learn sentences of the form 'A did X before B did Y' before they learn expressions of the form 'A's doing of X'. Sentences such as 'Socrates ran before he dined' and 'Nero fiddled while Rome burned' sound much less strained than 'Socrates' running occurred before his dinner', 'Nero's fiddling was simultaneous with Rome's burning', and so on.

Similarly (CL2 §15) 'Socrates ran' is more natural than 'A running by Socrates took place', just as 'Snow is white' is more natural than ' "Snow is white" is exemplified by snow'. Sellars argues that expressions like 'occurs' and 'took place' are 'alethic predicates', that is metalinguistic predicates defined in terms of 'true of'. Sellars elucidates 'true of by means of substitutional quantification but we can ignore this issue of substitutional quantification for the moment. Just as 'Being white is exemplified by snow' is a way of saying ' "is white" is exemplified by snow', so according to Sellars 'A running by Socrates at t took place' is a roundabout way of saying ' "runs" was true of Socrates at t'. (Sellars operates with a conception of tensed truth.)

According to this account event talk is metalinguistic talk, which itself is parasitic on the more fundamental assertions in the object language that not mention events. The metalinguistic talk mentions predicates like 'runs', not nouns like 'runnings', and the more fundamental object language talk uses such predicates, and once more does not have noun phrases for events. In this object language talk 'before', for example, occurs as a sentential connective, not as a relational predicate.

One must of course not confuse a sentence with a name, and so one must not fall into the confusion of thinking that sentences like 'Socrates ran refer to events. On the surface 'Socrates ran' is about Socrates, not about an event of Socrates running. Of course Davidson has given arguments, such as from the implication of sentences like 'Socrates ran' by sentences like 'Socrates ran quickly', that the deep structure of 'Socrates ran' is that of a sentence containing the noun phrase 'Socrates' running'. Nevertheless according to Sellars's theory it is 'Socrates' running was quick' that needs transforming away into 'Socrates ran'. According to Sellars, if 'Socrates ran' is about Socrates, not about an event of Socrates' running, neither is 'A running by Socrates occurred' about an event. Sellars' account of event talk as metalinguistic and of 'occurs' as an alethic predicate leads him to deny an ontology of events, while not wishing to deny that in a sense there are events, i.e., that sentences such as 'Socrates ran' are true.

In Section I of this paper I noted that Sellars's views tie in with the perhaps archaic use of 'event' as like that of 'outcome' or 'result'. There is a naturalness also in treating 'Socrates ran' as more fundamental than 'There was a running by Socrates'. There seem, however, to be powerful Davidsonian arguments (from semantics) on the other side, the apparent naturalness of Sellars's views notwithstanding. The appeal of Sellars's theory may diminish if we note that if 'before' and the like are sentential connectives they are non-truth functional ones. Thus if I say 'Wittgenstein wrote the Tractatus before Gödel proved the incompleteness of arithmetic' I say something true. But if I substitute for 'Gödel proved the incompleteness of arithmetic' the true sentence 'Russell wrote the Principles of Mathematics' I get something false. But if such sentences are non-truthfunctional what is their semantics?19 Davidson's strategy, of course, would be to show that they are not really non-truthfunctional at all, that the apparent non-truthfunctionality comes from misconstruing their surface structure, in particular from misconstruing 'before' as a connective. What alternative strategy does Sellars propose? What are his semantics for his semantics for sentences containing 'before'? How does he deal with adverbs? One would like Sellars to say more about this, though he does refer to the interesting and well-written dissertation by Jack Norman.20 However one might be dissatisfied with the Fregean semantics used by Norman, since it makes use of intensions, and might wonder whether it is not open to objections raised by Davidson in his 'Theories of Meaning and Learnable Language' to Church's theory of sense and denotation.21 Also in an earlier paper Sellars speaks enthusiastically about Romane Clark's theory of predicate modifiers. (This is mentioned in Jack Norman's thesis.) Sellars says that Clark's analysis "provides the key to the problem of the identity of events .... "22 Clark's theory is very attractive in many ways, though I am unclear about the ontic commitments of his semantics. Clark says at the end of the paper23 to which Sellars refers that "It is that language which the theory formalizes, and not our semantical discourse about it, which houses our ontology." This seems too easy a disclaimer: if we take semantics seriously we must surely count it as part of total science, and hence its ontic commitments as part of those of total science. I should be interested in Sellars's views on these matters.

Even if such a semantics can satisfactorily be given, and if Sellars can elucidate event talk as parasitic on talk with 'before', 'after', etc., as sentential connectives, how much does this prove metaphysically? It is on the face of it indeed plausible that we initially learn words like 'before', 'after', etc., by hearing them used in sentences in which they function syntactically as sentential connectives, and if this indeed turned out to be the semantically fundamental use it would not be too surprising. (This would be so even with sentences like 'Lightning before thunder', since here 'Lightning' and 'Thunder' function as one word sentences, equivalent to 'It lightnings' and 'It thunders' or my preferred formulation 'The lightning lightnings' and 'The thunder thunders'.) Nevertheless the fact that we may initially use 'before', 'after', etc., as sentential connectives does not show that we can not come to use them independently as relational predicates of events. Consider as a parallel Quine's account in The Roots of Reference of how a child might first learn quantification as substitutional, and then by an analogy between relative clauses and general terms leap (perhaps by a beneficial confusion of thought) to using variables and quantification as objectual.24

The issue is complicated by Sellars's distinction between the manifest image and the scientific image. It is further complicated by Sellars's insistence on the primacy of tensed language in the manifest image. I shall say something about this latter complication first.


In CL2 §27, Sellars refers to his argument in TWO where he claims that tense, in a broad sense including not only temporal inflections of verbs but also indicator words such as 'now', is an irreducible feature of temporal discourse.25 However when he says "indicator words such as 'now' " does he mean specifically temporal indicator words such as 'now' or any indicator words whatever including 'this'? It seems to me that if one has the word 'this' one can say all that one normally wants to say with tenses, using only tenseless verbs.26 Thus instead of saying 'Jones will come' one can say 'Jones comes later than this utterance', and so on, (where the italics signify the tenseless present). There are of course complications with the difference between perfect and imperfect, and other complex tenses, but perhaps they can be accommodated. What can not be accommodated are such metaphysically misleading sentences as 'The battle was future, is present, and will be past'. That these can not be accommodated seems to me to be a virtue, not a vice.

Sellars might object to the phrase 'this utterance' in the above elimination of tenses, because it seems to take an ontology of events (utterances) as fundamental. Instead of 'Smith comes later than this utterance' we could say 'Smith comes later than I utter this', and if one wishes to eliminate the indexical 'I' also one could say 'Smith comes later than the person who utters this utters this'. Sellars might still worry about the reference of 'this', which seems to be an event, that is, a particular utterance. At any rate the problem would not arise with 'Smith writes this', where 'this' can refer to a physical object, say a chalk mark on a blackboard.

Assuming, then, that one who disbelieves in an ontology of events can still think of some suitable referent for 'this' in the context 'utters this', Sellars's insistence on the necessity for tenses seems a very dubious one. Apart from the minor worries that I have expressed at the end of the last paragraph, this insistence seems to me to be independent of his account of expressions like 'before' as sentential connectives and of his denial of an ontology of events. But perhaps Sellars is understanding tense in a still wider sense, so that any indexical language counts as tensed. (Though this interpretation seems to be inconsistent with his denial of spatial tenses in TWO p. 558.) If any indexical language were to count as tensed, then for some purposes (but not, I think, scientific or metaphysical ones) tensed language would be indispensable. The indispensability (in some contexts) of indexicals has been well argued by John Perry in his well-known paper on the problem of the essential indexical.27 For example Perry gives the example of a person lost in the wilderness who knows he is in one of two different localities, which he is unable to distinguish by any perceptually relevant descriptive expressions. The one he calls 'here' is one or other of them, and which it is may be highly relevant to the problem of the best way of getting himself home. In such a case the indexical sentence 'Such and such a place is here' is not only not translatable into a non-indexical one (which is always the case) but furthermore there is no non-indexical sentence that will meet the person's needs. Again, a professor who has to get to a meeting at noon may move off at 11.50: he needs to know that 11.50 is now. However I think that all cases of the essential indexical are of practical relevance only and do not arise in theoretical science or metaphysics. So I do not see that tenses, or even indexicals in general, are needed to state any ontological fact about the world that could not be stated in non-indexical language.


Consider the question, discussed by Sellars in TWO, whether tenses are needed in discourse about Minkowski space-time. More importantly for our present purposes (for I have suggested that the two questions are largely independent of one another) do we need to refer to events in the physics of Minkowski space-time? Does acceptance of Minkowski involve acceptance of an event ontology and of 'earlier', 'later', etc. as relational predicates of events? Sellars does not say much about this in his Carus Lectures. There is more about his views about space-time and events in scientific theory in 'Time and the World Order'.

In TWO Sellars gives an account of relativistic space-time physics whereby it is an adjunct to the ordinary discourse of spatial objects enduring in a separate nonspatial time which he holds to be characteristic of the manifest image. According to Sellars, what relativity teaches us is not that space and time taken separately are unreal with a unitary space-time the only reality, but is that 'the metricizing of a set of events into a three-dimensional spatial array and the metricizing of spatially related events into a one-dimensional temporal array are not independent operations' (TWO p. 571). And I take it that here he still holds that talk of events is not ontologically serious talk, but is to be elucidated in terms of statements about things changing. The Minkowski world, on Sellars's view, seems to be just a mathematical device to enable calculations to be made in the manifest world of separable space and time. If I have interpreted Sellars correctly, his attitude contrasts markedly with that of Minkowski, in his famous statement that 'Henceforth space by itself, and time by itself, are doomed to fade away into mere shadows, and only a kind of union of the two will preserve an independent reality.'28 Nevertheless I think that Sellars holds that Minkowski's statement would be correct in relation to the scientific image, but he seems to hold that this is something for the far future, not for relativity physics as it is used today.29 At present, he says, we have for this conception only 'a partially covered promissory note'. Only in this not yet realized scientific world view does he think that we would have a genuine ontology of events and of 'earlier' and 'later' as relational predicates of events.

Earlier in this paper I manifested some indecisiveness and unclarity about the notion an event, and of course we have to contend with Sellars's arguments that even talk is merely a notational variant on more fundamental discourse that does not refer to such entities. So it seems that either it is unclear what events are or else Sellars is right in his 'no event' theory. This might be taken as a difficulty for special relativity if this is taken as an ontologically serious modification of the manifest image. It has sometimes been thought that the topology of space-time has to be elucidated in terms of causal relations between events.

Sellars's solution, as I understand it, is to be quasi-instrumentalist about Minkowski space-time. I wish to avoid the worry about whether there are events by denying the legitimacy of a causal theory of space-time.30 Even if events exist, there are not events everywhere, and the construction of space-time in terms of causal relations requires contrary-to-fact conditionals. It seems odd that so transparent a business as space-time geometry should be held to depend on such tricky matters as counterfactuals and their attendent modal notions. We should avoid all this trouble by simply postulating space-time and its geometry. Space-time is a genuinely explanatory entity just as electrons are. The geometry of space-time figures holistically in all physical explanations, but the point I wish to make can be made most vividly by means of a kinematical example. Consider the Langevin twin who travels by a space ship to a distant star and then returns. On his return he is younger than his stay-at-home brother. The explanation of this is purely geometrical. If AB, BC, and AC are all time-like (each within some light cone), the geometry of Minkowski space-time is such that in the triangle ABC we have AB + BC < AC. In general relativity the explanatory power of space-time is even more striking, where gravitation is explained by means of geodesics in a space-time of variable curvature.

For space-time to be genuinely explanatory it must exist. And if it exists then genuine space-time entities must exist. Sellars in TWO seems to me to treat space-time as a purely mathematical and instrumental adjunct to the manifest image (as merely a means of facilitating deductions of manifest image propositions from other manifest image propositions). As I remarked earlier in this paper I do not myself think that there is the great gulf that Sellars sees between the scientific and the manifest image, and it is not clear to me that saying that a substance changes at time t is to say anything different from saying that parts of a four-dimensional thing later than t are different from parts of it earlier than t.31

I shall now argue that the question of whether or not there are events is of no interest for the philosophy of physics. Can we identify events and processes with temporal slices or stages of space-time processes? If the answer is "Yes" then events are related by the space-time relations that uncontroversially relate these slices or stages. We have seen semantic objections to this. However if the theory of predicate modifiers, commended by Sellars, is acceptable, then it could be used to revive the Quinean identification of events and processes with temporal slices or stages, without going all the way to Sellars's "no event" theory. If the answer is "No," or if events do not exist at all, then there is still no problem for science. In that case science does not need to talk of events and processes: slices and stages of space-time objects (or related things) do the job perfectly well. Suppose for example that one wants to talk about the space-time interval between (1) the collision and mutual annihilation of an electron and a positron and (2) the collision of an electron and a neutron. Here we appear to talk of events but we can do all that physics requires by talking of certain discontinuities in the direction of the world line of the electron in (1) (treating the positron in Feynman's way as a 'bent back' part of the electron) and in the direction of the world line of the electron in (2). We do not need to worry about whether these entities are events, because whatever they are, they are what physics needs to talk about.32


1. Here I am using 'logical form' in something like Davidson's sense. See for example Donald Davidson "The Logical Form of Action Sentences," in Donald Davidson, Essays on Actions and Events (Oxford: Clarendon Press, 1980). However if 'before' is a non-truthfunctional sentence connective, we will be not all the way to logical form until we have got down to something extensional.

2. Wilfrid Sellars, The Carus Lectures for 1977-78, published in The Monist, vol. 64, no. 1 (Januarv, 1981).

3. Wilfrid Sellars, "Time and the World Order" in Herbert Feigl and Grover Maxwell (eds.) Scientific Explanation, Space and Time, Minnesota Studies in the Philosophy of Science, vol. III, (Minneapolis: University of Minnesota Press, 1962).

4. On Sellars's distinction between the manifest image and the scientific image see Wilfrid Sellars, Science, Perception and Reality (London: Routledge and Kegan Paul, 1963), ch. I.

5. P. T. Geach, "Some Problems about Time," Proceedings of the British Academy, 51 (1965): 321-36. Reprinted in P. T. Geach, Logic Matters (Oxford: Basil Biackweii. 1972^.

6. For the problem of the pink ice cube see Wilfrid Sellars, CL3 §§17-35.

7. P. T. Geach in his "Some Problems about Time" objects to such locutions as talking of temporal stages of philosophers having beliefs or temporal stages of cats eating mice. But it seems to me that his opponent could simply deny that these locutions are illegitimate, while conceding that they perhaps lack literary elegance.

8. See my article "The River of Time," Mind 58 (1949): 483-94. Reprinted with slight revisions in A. G. N. Flew (ed.) Essays in Conceptual Analysis (London: Macmillan, 1956).

9. F. I. Dretske, "Can Events Move?," Mind 76 (1967): 479-92.

10. See W. V. Quine. Word and Object (Cambridge, MA: M. I. T. Press. 1960), p. 171.

11. This suggests my answer to Sellars when he discusses 'absolute processes'.

12. Donald Davidson, Essays on Actions and Events, p. 178, cited in nl, above.

13. See W. V. Quine, Word and Object (Cambridge, MA: M. I. T. Press, 1960), §50.

14. Jaegwon Kim, "Events and their Descriptions: Some Considerations," in Nicholas Rescher (ed.) Essays in Honor of Carl G. Hempel (Dordrecht: D. Reidel, 1969).

15. R. M. Martin, Belief, Existence and Meaning (New York: New York University Press, 1969), ch. 9.

16. Donald Davidson. "The Individuation of Events," in Essays on Actions and Events cited in nl, above.

17. Acceptance of a Kim-like theory of events need not be incompatible with physicalism. Let a be a substance, P a physical property, and Q a property belonging to common sense psychology. Then the physical event might be (a,P,t) and the 'mental' event (a,Q,t). However, if Q were a 'topic neutral' property (see my article "Sensations and Brain Processes," Philosophical Review 68 [1959]: 141-56), then (a,Q,t) would still count as physical. Still it would be awkward not to be able to identify a sensation with a brain process, or a flash of lightning with an electric discharge, as would not be prohibited by a Davidsonian theory of events.

18. See Donald Davidson, Essays on Actions and Events, p. 166.

19. Dr. Aubrey Townsend has pointed out to me that though non-truthfunctional such sentences are referentially transparent: substitution of coextensive predicates is possible salva veritate. Townsend thinks of this combination of non-truthfunctionality and referential transparency as making more difficulty for Sellars's view.

20. Jack Norman, Events and Semantic Theories (Pittsburgh: University of Pittsburgh, 1974).

21. See Donald Davidson "Theories of Meaning and Learnable Languages," Proceedings of the 1964 International Congress for Logic. Methodology and Philosophy of Science (Amsterdam: North-Holland), 1965, pp. 383-94.

22. Wilfrid Sellars, Essays in Philosophy and its History (Dordrecht and Boston: D. Reidel, 1974, p. 212.

23. Romane Clark, "Concerning the Logic of Predicate Modifiers." Nous 4 (1970), 311-35.

24. W. V. Quine, The Roots of Reference (La Salle, IL: Open Court, 1973). See p. 104.

25 And also in "Metaphysics and the Concept of a Person," in Karel Lambert, The Logical Way of Doing Things (New Haven: Yale University Press, 1969), reprinted as ch. 11 of Wilfrid Sellars, Essays in Philosophy and its History, (Dordrecht: D. Reidel, 1974).

26. I regard this observation as of metaphysical interest but as not important for semantics. The semantics for 'this' should be given by the method suggested by Donald Davidson in his "Truth and Meaning," Synthese 17 (1967): 304-23, and this method can be applied to all indicator expressions without any partiality to 'this'.

27. John Perry, "The Problem of the Essential Indexical," Nous 13 (1979): 3-21.

28. H. Minkowski, "Space and Time," in The Principle of Relativity, a collection of papers by Einstein and others translated by W. Perrett and G. B. Jeffery, with notes by A. Sommerfeld, (New York: Dover Publications, (1952). See the first paragraph.

29. See Sellars's "Time and the World Order" pp. 593.

30. See my article "Causal Theories of Time," Monist 53 (1969): 383-95, and Hugh M. Lacey, "The Causal Theory of Time: A Critique of Grünbaum's Version," Philosophy of Science 35 (1968): 332-54.

31. If I am right there is no separate entity S outside the rectangles of diagrams VII-IX on pp. 576-77 of TWO.

32. I have read earlier drafts of this paper to staff seminars at the two philosophy departments at the Australian National University and at the philosophy department of Monash University. I am grateful to all who attended these seminars for very helpful comments in discussion, and I should particularly like to thank Dr. Aubrey Townsend of Monash University who made detailed comments on the typescript of an earlier draft.