IV

PRIMARY AND SECONDARY QUALITIES

18. Primary qualities and ‘body’

Locke inherited from Descartes, or borrowed from Newton and Boyle, a distinction between ‘primary’ and ‘secondary’ qualities. [1] His attempts to define it in general terms are unsatisfactory, and for now a pair of lists must suffice. A thing’s primary qualities are its ‘solidity, extension, figure, motion or rest, and number’; [2] and its secondary qualities are its colour, temperature, smell, taste and sound.

It is often thought that this distinction is a shaky one, that Locke certainly did not put it to good use, and that we owe these two insights to Berkeley. I shall argue that the distinction is well-grounded and interesting, that Locke had grasped an important truth about it, and that Berkeley’s treatment of this matter is impercipient and unhelpful. Berkeley assimilated the primary/secondary distinction to that monolithic ‘theory of material substance’ which he thought he detected in Locke’s writings; and I shall argue that that is the dominating fact about his failure to deal competently with the distinction between primary and secondary qualities.

Locke has two general, true things to say about the primary/secondary distinction. One of them is his thesis that primary qualities are

In most of Locke’s theorizing, a thing’s primary qualities are taken to consist in its being (say) spherical, two feet across, and

[p89]

falling rapidly; [4] but here they are thought of rather as a thing’s being shaped, of some size, mobile, etc. That is, in the thesis that primary qualities are ones which a body cannot lose, it is determinable qualities which are in question and not determinate ones. Locke’s example reinforces this reading: ‘Take a grain of wheat, divide it into two parts; each part has still solidity, extension, figure, and mobility: divide it again, and it retains still the same qualities.’ [5] It is not clear that ‘solidity’ is a determinable, either in its normal meaning or in Locke’s specialized sense in which ‘solid’ means ‘impenetrable’. In this respect, as in others which we shall meet later in this section, solidity is a special case. [6]

Locke has a good point here, but he ought not to express it as though it were a prediction about the outcome of an experiment, for really it is a point about the meaning of the word ‘body’, or about the concept of a body or a physical thing. He does in fact know this: ‘People [mean] by body something that is solid and extended, whose parts are separable and movable different ways.’ [7] Indeed the word ‘primary’ for Locke partly means that these are qualities a thing must have to count as a ‘body’. (His other account of what ‘primary’ means is indefensible. [8])

Locke’s discussions of the concept of body involve detailed points (e.g. against Descartes) which are of some interest but which lie beyond my present scope. His general thesis that the raw materials which constitute the concept of body are to be found within the realm of primary qualities, and that secondary qualities are conceptually inessential, seems safe enough. Yet Berkeley apparently denies it: ‘It is not in my power to frame an idea of a body extended and moved, but I must withal give it some colour or other sensible quality… . In short, extension, figure, and motion, abstracted from all other qualities, are inconceivable.’ [9] There is an uncertainty of interpretation here, which is also suspiciously present in the corresponding passage in the Dialogues. [10] The quoted passage is, as it stands, true: a thing’s being extended, or its taking up space, must involve some

[p90]

spatial region’s being occupied by something -- some quality must be manifested in that region other than mere extension. But the quality could be solidity, which is on Locke’s list of primary qualities. [11] If Berkeley really is saying only that ‘body’ could not be defined out of extension, figure and motion, without recourse to solidity, then his point is correct but it does not count against Locke, or help Berkeley with the larger claim for which he is arguing, or justify his claim to be discussing ‘figure, motion, and the rest of the primary or original qualities’ (my italics). It is, indeed, a point which Locke himself makes and insists upon, in criticism of Descartes’ account of the concept of body. [12]

But perhaps Berkeley is, as he is sometimes thought to be, making a stronger claim to the effect that secondary qualities are essential to the concept of body: he may be saying, a little carelessly, that nothing could count as experience of a world of bodies which had primary but not secondary qualities. What he says, on that reading of it, is certainly relevant to Locke’s thesis; but it is also a manifest falsehood which could be believed, I think, only by someone who had lapsed into thinking of perception too exclusively in terms of sight. Granted that we could not see things to have sizes and shapes without seeing them to have (not necessarily chromatic) colours, the crucial point is that we could perceive objects to have sizes and shapes without ever seeing them -- and, it can be added, without ever hearing or tasting or smelling them either.

That Berkeley discusses this view of Locke’s is not due to his having sharply separated it from Locke’s other claims about primary and secondary qualities. On the contrary: it is little more than an accident that we can find in Berkeley an argument which goes against this Lockean thesis in particular, and the passages containing it are jammed into discussions of entirely different matters (see §24 below). In contrast with that, Hume’s section ‘Of the Modern Philosophy’ is devoted almost exclusively to expounding and criticizing the view that the concept of a body can be adequately based upon primary qualities alone. One notable fact about Hume’s treatment is its emphasis upon the primary

[p91]

quality so conspicuously ignored by Berkeley -- namely solidity. (Berkeley’s passing remark that solidity is ‘plainly relative to our senses’, though it bears on a certain view he attributed to Locke, has nothing to do with the thesis we are now considering. [13])

Hume says explicitly that the primary qualities other than solidity are an inadequate foundation for the concept of body, and then argues separately that the addition of solidity still does not save the day for Locke’s thesis. The argument about solidity starts like this:

Hume then proceeds to argue that the ‘idea’ needed to supplement and give content to that of solidity cannot be a primary-quality one, for it has already been shown that all the other primary qualities need supplementation themselves; and so it must be a secondary-quality one, specifically it must be an idea of colour; and so Locke’s thesis is wrong. As Hume expresses it a little later:

Hume is surely right that the notion of impenetrability needs to be supplemented, in the same way and for the same reason as does the notion of extension. The difficulty about getting ‘occupant of region x’ to stand on its own feet is equally a difficulty about ‘occupant of x to the exclusion from x of everything else’. Yet Hume’s claim that the supplementation must involve colour is obviously wrong, since it implies that the congenitally blind cannot, without borrowing from the rest of us, have any workable, contentful concept of body or of occupant-of-space.

[p92]

To see where Hume has gone wrong, consider what the congenitally blind have at their disposal which he implicitly denies them -- namely, the sense of touch. A little more adequately: a blind person can build up an account of what bodies there are, what their shapes and sizes and positions are, and so on, by means which fundamentally consist in his knowing what impediments there are to specific movements of (parts of) his own body. This requires him to have independent knowledge of how his own body moves, but that is all right: he can know where his body is and how it moves because he moves it and does not merely observe its movements. [16] So, I contend, an adequate basis for a knowledge of bodies can be provided by sensory means in which the sense of touch plays a large part and which need not involve any perception of things’ secondary qualities.

Hume denies this, on the basis of a very peculiar argument. He seems to take it that what he has to deny is a view which we tend ‘naturally’ to accept, namely that ‘we feel the solidity of bodies, and need but touch any object in order to perceive this quality’. He denies this on the grounds that ‘tho’ bodies are felt by means of their solidity, yet the feeling is a quite different thing from the solidity; and. . . they have not the least resemblance to each other’. [17] This obscure utterance does not, as it stands, give us any help at all, simply because there is not ‘the least resemblance’ between any feeling and any quality. Still, one gets some idea of what Hume means by this denial from looking at his reason for it: ‘A man, who has the palsey in one hand, has as perfect an idea of impenetrability, when he observes that hand to be supported by the table, as when he feels the same table with the other hand. [18] Apparently Hume wants to deny that tactual feelings relate to solidity as visual sense-data do to colours. Not only do we see colours, but–Hume will say–the notion of colour is purely visual in its sensory basis; there is a biconditional relationship, a two-way flow, between facts about visual sense-data and facts about sense-based colour-judgements. By way of contrast, Hume is saying, judgements about solidity can have either a visual or a tactual basis, as witness the man who has palsy. So tactual feelings are ‘different from’ solidity; we do not ‘feel the solidity of

[p93]

bodies’; and so Locke’s thesis cannot be rescued by invoking the sense of touch.

This argument completely fails. From the premiss about the man with palsy, Hume can reach his lemma that tactual feelings are unlike solidity only if he construes ‘Tactual feelings resemble solidity’ as entailing ‘Tactual feelings are required as a sensory basis for solidity-judgements’. But the denial of this does not entail Hume’s conclusion, which is a denial of ‘Tactual feelings suffice as a sensory basis for solidity-judgements’. Hume has moved from the premiss that T is not necessary for S to the conclusion that T is not sufficient for S, through a cloudy lemma expressed in such words as that T ‘has not the least resemblance’ to S.

 

19. The Analytic Thesis

Locke’s other general claim about the distinction between primary and secondary qualities is more interesting, though also the source of more problems, than is the meaning-of-’body’ thesis.

Briefly, and in Locke’s words, it is the claim that secondary qualities ‘are nothing in the objects themselves but powers to produce various sensations in us’, [19] or that ‘when truly considered [they] are only powers, however we are apt to take them for positive qualities’. [20]

To say that x has a power to produce S in me is to say, among other things, that if x were related to me in a certain way then S would occur in me. If that were all it meant, then Locke would be saying just that any statement attributing a secondary quality to a thing is equivalent to a counterfactual conditional of the form:

For example, the claim would be that ‘x is green’ means roughly the same as ‘If x were sunlit and were in the line of vision of a normal open-eyed human, he would have a visual field of such and such a kind’ (Locke would describe the visual field as ‘green’, no doubt; but never mind that).

[p94]

But the notion of a ‘power to produce’, and thus Locke’s central claim about the primary/secondary distinction, have more content than that. Locke’s claim has in fact two components -- the Analytic Thesis sketched above, and what I shall call the Causal Thesis, which is a view about what causes us to have the secondary-quality ideas that we do have. Since these are philosophically linked in a certain way, and are so interwoven in Locke’s text that I cannot cite any passage expressing one but not the other, I present them as components of a single ‘central claim’ of Locke’s about primary and secondary qualities. Still, they need to be examined separately, and my present concern is with the Analytic Thesis.

Is the Analytic Thesis true? More precisely, does it express a truth about secondary qualities which is not equally a truth about primary qualities? The italicized clause is vital. Locke wanted to contrast the two sorts of quality, and Berkeley’s main criticism was that no contrast was effectively drawn -- that anything true that Locke said about secondary qualities is equally true of primary. There is some excuse for Berkeley, in that Locke’s arguments are all rather poor and some of his formulations are downright misleading. Still, there is a legitimate contrast between primary and secondary qualities, and I contend that it is one which Locke noticed and tried to formulate and defend.

In his arguments for the Analytic Thesis, Locke repeatedly stresses the fact that one’s perceptions of secondary qualities may vary greatly according to the state of one’s body and environment. [21] Berkeley replied that that is equally true of primary qualities. [22] On the face of it, Berkeley seems to be right; but let us suspend judgement until we have considered, more carefully than Berkeley did, what might be thought to follow from Locke’s perceptual-variation point. Never mind whether it yields a contrast between two sorts of qualities–what does it show about any sort of quality for which it does hold? As to that, I conjecture that Locke had half-grasped, and was moving towards expressing, something like the following point. We are all familiar with the way in which something which tastes sweet to most people may taste bitter to a sick person. Now, if we reflect on this phenomenon, and on similar ones involving other secondary-quality

[p95]

perceptions, we shall see how thoroughly contingent it is that we are in a position to say of anything that it is bitter or green or noisy or the like. The occasional failures of agreement bring home to us how dependent our public secondary-quality terminology is upon the fact that we usually do agree in our secondary-quality discriminations -- the failures help us to realize that our notion of two things’ having the same colour, say, is only as secure as our ability to muster an overwhelming majority who see them as having the same colour.

Perhaps Locke was nowhere near having such thoughts as these. Anyway, they yield an argument for his Analytic Thesis; they embody a truth about secondary qualities which does not hold equally for primary qualities, and so they provide a basis for the contrast that Locke thought he could establish between primary and secondary qualities. The section that follows is a defence of all this. [23]

 

20. In defence of a distinction

I want to contrast two kinds of sensory aberration: in one, someone (C) sees two things as having the same colour when in fact they haven’t, and in the other someone (S) sees and feels two things as having the same size when in fact they haven’t.

C is confronted by a red thing and a white thing, and satisfies us that he sees them as having exactly the same colour. He believes our claim that they have different colours; and, since they differ in no other way, we could if necessary prove to him that we can see a difference between them which he cannot. Also, C could discover that the two objects reflect light of different wave-lengths, and might know that wave-lengths usually correlate with seen colours. But if he ignores what others say about the two objects, and ignores esoteric facts of optics, he may never learn that he has a sensory defect. A failure in secondary-quality discrimination, in one who is otherwise sensorily normal, can and sometimes does persist unsuspected through any variations in distance or angle of view, light-conditions, and so on.

Contrast this with the case of S who, going by what he sees and feels, judges a certain jug to have the same size as a certain

[p96]

glass which is in fact shorter and narrower than the jug. (Grice has discussed the case where what he sees supports one judgement and what he feels supports another. [24]) In this case, we can place the glass inside the jug; or fill the jug with water, and then fill the glass from it and throw away the remaining water; or place both vessels on a table and draw S’s hand across the top of the glass until it is stopped by the jug; and so on. What are we to suppose happens when S is confronted by these manipulations of the two objects? There are just two relevant possibilities. (a) When we manipulate the glass and jug, S takes in what is happening and thus quickly realizes that he was wrong about their relative sizes. (b) Each time we contrive a happening with the glass and the jug, S mis-perceives it so that what he sees and feels still fits in smoothly with his original judgement about their sizes.

To adopt (a) is just to abandon the attempt to put size-’blindness’ on a level with colour-blindness. If the point of the latter were just that secondary-quality perceptions can err, then we could say the same of primary-quality ones. What gives relevance and bite to colour-blindness, and to its analogues for tastes etc., is the fact that any such abnormality can persist indefinitely without the victim’s getting any clue to it from his other, normal sensory responses. The tricks with the glass and jug could be performed by S himself; they involve ordinary commerce with familiar domestic objects; and they are in a very different case from C’s attention to wave-lengths or to what other people say about things’ colours.

To get an analogy between size-’blindness’ and colour-blindness, then, we must adopt supposition (b). This requires us to credit S with such inabilities as the following. He cannot see or feel that the glass is inside the jug (or that the jug has not stretched or the glass contracted); he cannot see or feel that the glass is full of water (or that water remains in the jug after the glass has been filled from it); he cannot see or feel that his hand is touching the rim of the glass (or that his hand is stopped by the side of the jug). It will not do to suppose that as each trick is performed S sees and feels nothing: the analogy with colour-blindness requires that he shall have no reason to suspect that

[p97]

there is anything wrong with him, and so his visual and tactual states throughout must present no challenge to his belief that he is handling an ordinary pair of vessels which are of the same size. This is bad enough, but there is worse to follow. S must not only fail to see or feel the water left in the jug after the glass has been filled from it, but he must also have compensating sensory aberrations when the water is used to douse a candle or to dissolve sugar, or when it is thrown in S’s face. Similarly with any of the other sensory aberrations with which we prop up the initial one: each requires further props which demand others in their turn, and so on indefinitely.

The analogy has collapsed again. C’s colour-blindness was not clued by his other sensory responses although these were normal; but to keep S in ignorance of his initial sensory failure we have had to surround it with ever-widening circles of further abnormalities.

Strictly speaking, it is not quite true that C’s single failure of colour-discrimination could remain unclued by his other, normal sensory responses. If he sees no difference between R which is red and W which is white, how does he see a third thing R* which is in fact red? If his only sensory failure concerns R and W, then we must suppose that he efficiently sees R and R* as having the same colour and sees a large difference of colour between W and R*. Since ex hypothesi he sees R and W as having the same colour, this is impossible; and to salvage the story we must suppose C to be blind to colour-differences between red things and white things generally. This still does not restore the analogy with size-’blindness’, however. C’s single red/white failure spreads only into other red/white failures; whereas S’s initial failure to discriminate sizes had to be backed by failures also of shape-discrimination, movement-detection, sensitivity to heat, etc., ramifying out endlessly into virtually all his perceptions of his environment.

We have lost our analogy beyond recall; and we are losing our grip on an initial datum of the size-’blindness’ case, namely that we can agree with S about the identity of a certain glass and jug, while silently noting his error about their relative sizes. For now we find that S disagrees with us about countless visible and tangible aspects of our environment, so that it is no longer clear that we share with him a sensory awareness of a single objective world.

[p98]

From the foregoing discussion there emerge two closely related contrasts between primary and secondary qualities.

(1) There are countless exoteric general facts about how a thing’s primary qualities connect with its ways of interacting with other things: of two rigid things, the smaller cannot contain the larger; one thing cannot block another’s fall without touching it; a cube cannot roll smoothly on a flat surface; a circular disc’s imprint on wax will be circular; and so on, indefinitely. It is true that a thing’s colour, say, may also connect with its behaviour in relation to other things: brown apples are usually more squash-able than green ones, blue flames boil water faster than yellow ones, a red surface reflects lightwaves of different lengths from those reflected by a blue surface, and so on. But for colours and other secondary qualities we cannot make, as we can for primary qualities, a long tally of obvious, familiar, inescapable connexions of the relevant kind.

(2) Just because a thing’s primary qualities correlate in so many obvious ways with its modes of interaction with other things, we cannot intelligibly suppose that these correlations might persistently fail. There could be no point in crediting something with a shape, say, which was belied by enough of its interactions with other things.

As against this, there could be a point in calling a thing red even if this were belied by the wave-lengths of the light reflected from it, or by its flavour, hardness or chemical composition. If something’s colour were in sunlight indistinguishable from that of things agreed to be red, this fact could reasonably be reported in the words ‘That thing is red’, even if we had to add riders such as ‘though its light-reflecting properties are atypical for red’, or ’… though its taste is atypical for red wine, or ‘...though its temperature is atypical for red iron’. Since wave-lengths of reflected light (within the range to which humans are sensitive) do correlate with the colours seen by most people in sunlight, we do not need to decide for or against defining colour words in terms of how things look and treating associated wave-lengths as mere empirical correlates of colours. But if we had to decide, we could choose to give our colour terminology a purely visual basis and still have it doing most of the work it does for us now. Analogous remarks apply to the other secondary qualities.

[p99]

Not so, however, for primary qualities. The inter-relations between things in respect of their primary qualities are many and varied and tightly interlocked, so that we cannot isolate a subset of them and suppose that just those might continue to hold while all the rest failed. A fortiori, we cannot describe a partial breakdown, the survivors of which would support a working vocabulary of primary qualities. The only kind of breakdown we could hope to describe without losing control would be one involving the collapse of all but one of the normal correlates of some primary quality: for example, a world in which ‘the size of x’ had to be defined solely in terms of the visual field presented by x to an observer ten yards away, with none of the other actual correlates of size continuing to hold. This supposition is clearly self-defeating; for it bases ‘size’ on ‘distance’ while making it impossible to measure distance. That special feature apart, however, it is clear that no supposition of this general sort can preserve a minimal sense of ‘size’ analogous to the purely visual sense of ‘colour’. Any such supposition, in cutting away so much of what ordinarily goes with size, leaves no basis for a language of physical objects. Offering us a minimal sense of ‘size’, it robs us of everything that could have a size.

 

21. Corollaries

We are now in a position to see that Locke’s Analytic Thesis does express a truth about secondary qualities which is decisively not true of primary qualities.

Part of the point is that the Analytic Thesis equates x’s being green, say, with the truth of a conditional stating that under certain circumstances a specific, characteristic kind of sensory state would occur. Now, it may be true that x will be deemed square, say, if and only if our sensory states in respect of it are such as to warrant its being described as square, but this fact cannot be expressed by picking on some specific kind of sensory state and saying that the occurrence of that in specified conditions is more or less definitive of a thing’s squareness. The reason why this is so is brought out, I believe, by my analysis in the preceding section. It is just that a thing’s being square, or having any other specific primary quality, consists in its relating to many other kinds of things in specific ways, and all of these are comprised

[p100]

in the notion of our sensory states’ being such as to warrant our describing the thing as square.

But my analysis is also relevant in a different way. The Analytic Thesis says that a thing’s having a given secondary quality is its having a certain power; and just this, prescinded from any question of what sort of power, is inapplicable to the primary qualities of things. We can identify a glass, say, while remaining ignorant of or in disagreement over its secondary qualities; and so we have the notion of the glass as an object which, among other facts about it, has certain ‘powers’ to affect us in ways which are our basis for crediting it with colour, taste etc. But we cannot identify a glass independently of all its primary qualities such as location, size, shape, etc.; and so we cannot have the notion of the glass as an object which, among other facts about it, affects us in ways which are our basis for crediting it with primary qualities. Granted that everything we say about the glass is based on sensory states it causes us to have, it is still misleading to speak of its power to make us perceive it as having a certain shape, size, etc.; for that way of speaking suggests that we have some notion of it -- some way of identifying and studying the glass–independently of, and as a preliminary to, discovering what its primary qualities are. My analysis shows why we cannot have this.

There is another point on which a little light can now be thrown. It is often thought that whatever significant differences there are between primary and secondary qualities are in some way due to an underlying difference expressed by this:

There is presumably a truth lurking behind this, but (A) as stated does not capture it; for one can know that the starter’s gun is making a noise because one sees the smoke, and discover that the apple is green by tasting it, and so on. No doubt (A) can be modified so as to cope with these cases, but just what modifications are needed in order to turn (A) into something precise and true, and how interesting it would then be, I do not know.

[p101]

My main point, however, is that neither (A) nor anything like it can explain the differences between primary and secondary qualities which came to light through my contrast between size-‘blindness’ and colour-blindness. For these differences would have emerged just as easily if the size-’blind’ man had been blind. Depriving him of sight altogether would not have helped us to entertain the hypothesis that he mistook the relative sizes of the jug and the glass and maintained that mistake through a series of down-to-earth transactions with the two objects in question. That hypothesis would, indeed, be blocked by every one of the sample obstacles presented in the original story. So the facts about primary qualities to which I was calling attention cannot be linked with any facts about vision in relation to primary qualities; whence it follows a fortiori that the contrast I was drawing between primary and secondary qualities does not arise from anything expressed by the thesis (A) or any modified version of it.

If one could explain the differences between primary and secondary qualities by adducing facts about their respective sensory bases or correlates, I suspect that the crux of the explanation would turn out to be the fact that the sense of touch -- or rather of touch-and-movement -- is involved in all the primary qualities in a way in which it isn’t with any of the secondary. But that is only a suspicion. Someone should write a book on the epistemology of the sense of touch.

 

22. The Causal Thesis

The Causal Thesis about primary and secondary qualities is this: in a perfected and completed science, all our secondary-quality perceptions would be causally explained in terms of the primary qualities of the things we perceive. For example, our colour-discriminations would be explained by a theory relating the colour aspects of visual sense-data to the sub-microscopic textures of seen surfaces. While admitting that we do not in fact have the theory, or set of theories, which would yield these explanations, Locke is immensely sure that this is only because of our ignorance. He does not doubt that some theory of this kind is true.

This confidence is prima facie puzzling. The state of seventeenth-century physiology surely did not warrant it. The success of

[p102]

Newtonian physics in other areas might, of course, have induced an optimism about its chances of eventually explaining everything. Consider, for example, this argument of Locke’s for his central claim about secondary qualities:

This argues that an almond’s colour and taste are mere upshots or symptoms of its primary-quality ‘texture’, since the latter is all that can be altered by pounding. It is invalid: the assumption that pounding can cause only primary-quality changes in the object pounded is false, as is shown by what happens when an almond is pounded with a pestle. Still, Locke’s using the argument does suggest that he was encouraged in maintaining the Causal Thesis by a general faith in the power and comprehensiveness of a purely primary-quality physics.

But that is not all. Locke’s advocacy of the Causal Thesis can be partly explained on the basis of his acceptance–for good, fumbled reasons–of the Analytic Thesis. There is a very natural intellectual route from one to the other, and Locke comes into sharper focus if we suppose that he followed it.

By way of introduction, I call attention to three features of the Analytic Thesis. (1) According to it, secondary qualities are dispositional: ‘x is green’ is equivalent to a counterfactual conditional. (2) It represents secondary qualities as relational: ‘x is green’ means something about items (people) other than x, and could become false just because of a monadic change in those other things. (3) It represents secondary qualities as involving something mental: ‘x is green’ means something about the occurrence of a certain kind of idea. Putting these three features in a nutshell: according to the Analytic Thesis a secondary quality of a thing is its power to induce in something else an idea. The three features are quite distinct. They generate eight possible kinds of property, which are exemplified by the following eight adjectives: green (1,2,3), poisonous (1,2), depressive (1,3), explosive (1), stinging (2,3), indebted (2), worried (3), square. One notes with pleasure that of the three features credited to the secondary

[p103]

quality greenness, none is credited to the primary quality squareness.

Each of these features of secondary qualities will help to explain certain aspects of Locke’s handling of the Analytic Thesis. I start with the first of them, i.e. with the fact that secondary qualities are dispositional, which is just to say that secondary-quality attributions are equivalent to counterfactual conditionals.

It is sometimes said that anything of this form:

means the same as, or should be analysed into, something of this form:

I believe that (2) does more than bring out the meaning of (1). Still, western science has for centuries proceeded on the assumption that wherever (1) is true (2) will be true also; and this assumption -- or regulative principle -- has clearly been instrumental in scientific advance, e.g. by implying that if something is soluble in water it has a chemical, structural property which explains its solubility. It would be astonishing if Locke, with his feeling for and understanding of the western scientific tradition, had not assumed that dispositions are always causally rooted in non-dispositional properties.

If he did, then he would think that whenever something of this form is true:

then the corresponding statement of this form will also be true:

This comes close enough to saying that x, by virtue of some non-dispositional property that it has, will cause a G idea in... etc. We have nearly got Locke as far as the Causal Thesis.

Not quite, though, for the Causal Thesis says that our

[p104]

secondary-quality perceptions are to be explained through the primary qualities of the perceived objects. To express this, we must replace ‘For some non-dispositional ...’ by the stronger ‘For some primary-quality ...’. But of course that strengthening will be acceptable to Locke: of the two sorts of qualities-of-objects at his disposal, one sort, secondary qualities, have already been declared by the Analytic Thesis to be dispositional; and so the only candidates he has for the role of non-dispositional properties are primary qualities. And so we have arrived at the full-strength Causal Thesis.

Notice that the only grounds I can give Locke for saying that our secondary-quality perceptions are to be explained in terms of things’ primary qualities is just the de facto absence of any other suitable candidate. So if my conjecture about the movement of his thought is right, he ought to concede that the true causal explanations might turn out to involve not primary qualities but qualities of some now unknown kind. And so he does: ‘Secondary qualities.., depend.. . upon the primary qualities of [objects’] minute and insensible parts; or, if not upon them, upon something yet more remote from our comprehension.’ [27]

I have remarked that Locke argues poorly for his view about primary and secondary qualities. This fact can be explained. The trouble is that Locke had a false belief about what sorts of considerations were needed to support his central claim. Kneale says: ‘The distinction between primary and secondary qualities was a philosophical discovery, and Locke was mistaken when he wrote of it as though it had been established by experiments unfamiliar to plain men.’ [28] Kneale presumably thinks of the distinction as ‘established’ by whatever supports the Analytic Thesis; and I agree with him that this is a philosophical thesis whose support involves no rechercha scientific information, no appeals to microscopy or the like, but only to the unexciting kind of empirical material to be found in §20 above -- reminders assembled for a purpose. The further move to the Causal Thesis does not need strenuous argument: it is really just a matter of combining the Analytic Thesis with a highly respectable regulative principle or scientific working assumption. But Locke, having failed to distinguish the Analytic from the Causal Thesis, sees the latter as

[p105]

needing argumentative support; and, since it has implications for science’s future, Locke naturally thinks that its support must come from science’s past and present.

 

23. The other versions

Locke has two other contrasts -- or ways of drawing the contrast -- between primary and secondary qualities. I shall argue, with regard to each of these, that it is intelligible only if regarded as a fumbled attempt to express the central claim with which I have already credited Locke.

The first of the two is fairly plain sailing. Here is a typical expression of it:

Since ideas cannot resemble either bodies or qualities of bodies, this must be either discarded or transformed. The only plausible transformation is into something like the following: in causally explaining ideas of primary qualities, one uses the same words in describing the causes as in describing the effects (shape-ideas etc. are caused by shapes etc.); whereas in causally explaining ideas of secondary qualities one must describe the causes in one vocabulary and the effects in another (colour-ideas etc. are caused by shapes etc.). If this is not what Locke’s ‘resemblance’ formulations of the primary/secondary contrast mean, then I can find no meaning in them.

Suppose that they do mean what I have suggested. Someone might challenge Locke: ‘Why should we not explain colour-ideas etc. in terms of the colours of the objects which are seen?’ He would have to reply: ‘We could, but that would not be to give the most basic kind of causal explanation of secondary-quality ideas.’ In saying this, he would be reiterating the Causal Thesis. And that makes my main point about these ‘resemblance’ formulations: if they present us with anything we can get our teeth into, it is the Causal Thesis. There seems to be no other reading of

[p106]

them, no other way of representing them as more than mere paragraphs, inert word-sequences with which we can do nothing.

The one remaining kind of thing Locke often says about the primary/secondary distinction is to the effect that secondary qualities are not ‘really in’ the bodies to which we thoughtlessly attribute them. His remarks in this vein have three separate sources, corresponding to the three things the Analytic Thesis says about secondary qualities -- that they are dispositional, relational, and mind-involving.

The first of these is not very important, but there is at least one example of it: ‘Yellowness is not actually in gold, but is a power in gold to produce that idea in us by our eyes, when placed in a due light...’ [30] This use of ‘idea’ is unsatisfactory: the phrase ‘that idea’ seems to refer back to yellowness, but yellowness is a quality and not an idea! Locke is in a difficulty here which I shall discuss shortly. My present concern is with the first dozen words in the quoted passage. If they don’t express a crude inconsistency (yellowness is not in gold but is in gold), then I think they must involve the actual/potential distinction. I suggest, that is, that Locke wants to say that yellowness is only a disposition or ‘power’ of the gold and not a non-dispositional or ‘actual’ property of it. On that reading, the remark primarily expresses one aspect of the Analytic Thesis, and does not introduce any new line of thought which needs separate consideration.

The second of the three features looms larger. That is, Locke’s tendency to speak of secondary qualities as ‘not in the object’ seems to be in good measure due to his preoccupation with the fact that according to the Analytic Thesis secondary qualities are relational. Consider this passage:

Again the use of ‘idea’ is sloppy. The vital phrase, however, is not really in the gold, considered barely in itself’: Locke’s point is that the gold’s being yellow depends not just upon it but also

[p107]

upon other things (people), so that yellowness, rather than being ‘in’ the gold, is in a manner of speaking between the gold and the class of normal, sighted humans. Locke, I suggest, goes too far here. The notion of ‘where’ a secondary quality is, is metaphorical and so perhaps harmless in itself; but Locke seems to be cashing this metaphor, and going too far, when he says that secondary qualities are called ‘qualities’ only ‘to comply with the common way of speaking’. [32]

I think he would be prepared to argue somewhat as follows: a given kind of substance might be bitter at one time and tasteless at another, simply because in the interim there had been a suitable change in the taste-buds of humans. In such an eventuality, the substance would have lost its bitterness without changing in itself; but to change is simply to undergo some turnover in the qualities that one has; and so it follows that the substance would have lost its bitterness without losing any of its qualities. So bitterness is not a quality–and similarly, mutatis mutandis, for other secondary ‘qualities’ and indeed for relational ‘qualities’ generally.

That argument is not valid. For in the situation envisaged, the substance would have undergone a change, namely a change in its taste. It is natural to protest that for a substance to lose its bitterness m that way is not for it to change in itself; but the force of that is just that bitterness is a relational property–which is the point from which we started.

The third source of Locke’s ‘not in the object’ remarks about secondary qualities is by far the most important. Here are two examples:

Notice that in these passages what Locke says is not that secondary qualities are ‘not in’ the objects, but rather that secondary-quality

[p108]

ideas are not in the objects. How then does this mark a contrast between secondary and primary? There are only two assumptions on which it can do so.

(a) Primary-quality ideas are, literally, in the objects to which primary qualities are attributed. If this were so, then the claim that secondary-quality ideas are not in objects would indeed distinguish primary- from secondary-quality ideas and to that extent mark a distinction between the two sorts of quality.

(b) There is no distinction between secondary qualities and secondary-quality ideas. If that were so, then the denial that secondary-quality ideas are in objects would imply a denial that secondary qualities are in objects; and, since primary qualities are really in the objects to which they are attributed, this would mark a distinction between primary and secondary qualities.

Of these two quite different possibilities, (a) seems to me the more obviously untenable as an interpretation of Locke’s thought. We have seen that his elliptic use of ‘idea’ to mean ‘quality’ does sometimes run away with him; and it has been pointed out to me that Locke sometimes uses ‘sensation’ instead of ‘idea’, but never when primary-quality ideas are in question–a fascinating fact which is prima facie evidence of his wanting to get primary-quality ‘ideas’ out of the mind and into the object. But I just do not see how to carry this interpretation through, and I am content to jettison it since there is a much more plausible alternative.

For (b) is a thoroughly plausible reading of the ‘not in the object’ remarks which I have just quoted. Sometimes, it seems, we must admit that Locke tends to identify secondary qualities with ideas of them, as when he says of porphyry that ‘it is plain it has no colour in the dark’. [35] Consider also this passage:

Having just remarked that ‘every one readily agrees’ that ‘these ideas of sickness and pain are not in the manna’, Locke now argues that we ought to say the same about taste and colour; and his

[p109]

wording forces us to construe him as identifying secondary qualities (‘sweetness and whiteness’) with secondary-quality ideas (‘the effects’).

In thus identifying secondary qualities with ideas of them, Locke sins both against his clearly announced distinction between qualities of both sorts and ideas, and against his central claim that secondary qualities are powers to cause ideas. How did he get himself into this self-contradictory situation? Why, when he was possessed of the truth that secondary qualities are powers to cause ideas in humans, should Locke give countenance to the conflicting and false thesis that secondary qualities are ideas in humans?

His drift from truth into error can be explained as arising from three distinct mis-handlings of the truth.

The first of them consists in moving from ‘not in the object’ to ‘in the mind’. Of course this move is invalid if secondary qualities’ being ‘not [actually] in the object’ is so understood that it follows from their being dispositional or from their being relational; but I suggest that Locke may have been inclined to make it, and that this helps to explain his sometimes implying that secondary qualities are ideas.

The second explanation can be approached through Locke’s comparison of secondary qualities with pains, sickness, etc. This wholeline of argument arose, I submit, because Locke saw that he had flushed out a problem but then proceeded to misidentify it. The problem arises thus: we can report a thing’s power to cause certain states in us by calling it ‘sweet’, ‘white’, ‘warm’, etc., while its powers to cause other states in us are not reportable in single adjectives except clumsy (and explicitly causal) compounds like ‘sick-making’ and ‘pain-causing’. Why do we draw that line between the two classes of states of ourselves, and thus between the two classes of powers? Locke assumes, far too boldly, that we cannot have a good reason for drawing the line just there; and his conjecture about our bad reason for the line’s location really will not bear scrutiny.37 But his real trouble is that he has misunderstood what sort of line it is. Seeing that his Analytic Thesis about secondary qualities raises a question about how secondary qualities relate to pains and sickness, he has completely misunderstood what question of this sort it raises. The Analytic

[p110]

Thesis implies that ‘green’ means ‘apt to cause G ideas’, which is structurally like ‘apt to cause sickness’. But that does not even prima facie threaten to put ‘green’ on a par with ‘sick’, or greenness with sickness. It puts ‘green’ on a par with ‘sick-making’, and what it puts on a par with ‘sick’ is ‘having a G idea’. The Analytic Thesis about secondary qualities, when properly stated, does not pose any problem of the form ‘Why do we treat greenness differently from pain in the respect that. . . ?‘, and so it does not raise the question to which Locke addressed himself, namely ‘Why do we regard greenness but not pain as in the object?’ Still, it is not bewildering that Locke should think that he is confronted by this question; for the Analytic Thesis does raise a genuine problem which, being of the form ‘Why do we draw the line just there?’ and also involving pains and secondary qualities, could fairly easily be mistaken for the pseudo-problem which Locke raises.

The third mistake which may contribute to Locke’s tendency to assimilate secondary qualities to ideas of them is as follows. His Causal Thesis implies that a secondary quality of x is a power which x has because of its primary qualities–as a thing may be soluble because of its chemical composition. Also, in seeing xas green one has an idea which occurs because of x’s primary qualities–as a thing may dissolve because of its chemical composition. These are different types of dependence; but the fact that they can be expressed in the same words may have helped Locke not to notice when he was slipping from one sort of dependent to the other.

Recall once more the three features which I stressed in §22: secondary qualities are dispositional, relational, and involved with something mental. In the present section I have argued that Locke’s remarks about secondary qualities as ‘not in the object’ are fed by each of these sources. Because a secondary quality is dispositional it is only a power, and is therefore not an actuality in the object, and is therefore ‘not actually in the object’. Because it is relational it is ‘not in the object considered barely in itself’. And (though here I compress the argument) because it is obliquely mental it is ‘not in the object’ because it is in the mind. These uses of the ‘not in the object’ language are all deplorable, but their inter-relations are part of the structure of Locke’s thought.

[p111]

24. Berkeley’s conflation

Berkeley’s attack on this part of Locke’s work was addressed to the worst of all Locke’s formulations. What Berkeley attacked was the thesis that, while primary qualities are in objects, secondary qualities are not because they are ideas and are therefore in the mind. In this two-part thesis, what he dissented from was the part about primary qualities: Locke’s mistake, according to Berkeley, lay in his not saying about primary qualities what he did say about secondary, namely that they are ideas in the mind.

Berkeley did not–as I once inexplicably alleged [38] -- endorse Locke’s arguments for the mental nature of secondary qualities; but he used them ad hominem against Locke, and thought that he had valid arguments for the same conclusion. Remarking that Locke’s arguments ‘may with equal force, be brought to prove the same thing of extension, figure, and motion’ as they purport to prove about secondary qualities, Berkeley adds a disclaimer:

This does not adequately describe the short-comings of Locke’s arguments in this vein–e.g. the argument that water could not feel warm to one hand and cold to the other if ‘those ideas’ were really in the water, and the argument about sweetness and whiteness in relation to sickness and pain. Berkeley, however, does not care to examine those arguments thoroughly because he so likes their conclusion: he agrees with what he thinks Locke is saying about secondary qualities, and complains only over Locke’s failure to say it about primary qualities also.

I should mention in passing Berkeley’s treatment of number. [40] This, though I have carefully neglected it, does occur on Locke’s list of primary qualities; and Berkeley deals with it through a

[p112]

special argument which does not follow the general pattern outlined above. We owe to Frege an important insight about the concept of number, namely that given any stretch of reality there is no number-expression which is absolutely the right one to apply to it. [41] It may, for example, be one book, or sixty pages, or millions of molecules. Berkeley had some understanding of this, but he mis-handled it. He said: ‘The same thing bears a different denomination of number, as the mind views it with different respects’, from which he inferred that ‘number is entirely the creature of the mind’. But this is wrong. The item objectively is one book, and objectively is sixty pages, and nobody’s mind has anything to do with it. If someone asks of some part of the world ‘What number ?‘, the question cannot be answered: it must be replaced by a question of the form ‘What number of Fs?’ But Berkeley implies that the original question can be answered in more than one way, and that we cannot know which answer to give until we know something about the questioner’s frame of mind. In short, he thinks that the question requires psychological rather than logical completion. Frege’s point disinclines one to classify number as a primary quality, but not for any reason that supports Berkeleian idealism.

Berkeley’s treatment of number, although special in its argumentative tactics, is in its conclusion just like his handling of the other members of Locke’s list of primary qualities. The point, Berkeley thinks, is that all of these qualities share the feature of secondary qualities stressed by Locke–namely, they are in the mind rather than in the object. How can we assess the adequacy, exegetical or philosophical, of this part of Berkeley’s work? Locke’s ‘not in the object’ remarks about secondary qualities are, as we have seen, flimsy and unstructured and inconsiderable; we cannot explain them except through his having somehow drifted from his central insight, the Analytic Thesis; and really that is all that can be said about them. If they include stray sentences which Berkeley can adopt as expressing (part of) a genuine philosophical view of his own, that is just a typographical accident. When one has seen what the provenance is of the ‘not in the object’ remarks within Locke’s thinking about primary and secondary qualities, one will not regard them as expressions of any solid philosophical position.

[p113]

In thinking otherwise, Berkeley erred. The error, however, is not in itself an interesting one. Is there nothing else we can say?

Well, there is a point of some importance, arising from this:

Philonous is here challenged to explain the mistake, supposedly made by some philosophers, of distinguishing secondary qualities from primary in respect of the possession of ‘real existence’ (as distinct from ‘ideal existence’, i.e. existence in the mind). He rightly assumes that to explain this one must point to some genuine differences between primary and secondary qualities–but then look at what he actually offers! As well as being far from the centre of the primary/secondary distinction, it is not even true. In support of it Berkeley might point out that ‘hot’ is usually more relevant to pain than ‘round’ is, but what about ‘sweet’ as compared with ‘sharp’ or ‘hard’ or ‘large’ or ‘rapid’?

Berkeley hints that he could say more (‘among other reasons’), but this dismally inadequate offering is all that he actually produces. Apparently he could not find in Locke’s work the makings of any legitimate contrast between primary and secondary qualities. So he can and should be criticized not just for over-dignifying the ‘not in the object’ remarks, but also for overlooking or mis-estimating the other, better parts of Locke’s discussions of the primary/secondary distinction.

There is a further point about Berkeley’s procedures which is essential for an understanding of Locke and of the truth, if not of Berkeley himself. When Berkeley approves Locke’s saying that secondary qualities are in the mind, and deplores his refusal to say the same of primary qualities, he is thinking about the conflict between his and Locke’s theories of reality. In Berkeley’s view, Locke’s doctrine on secondary qualities is the thin edge ofa wedge which can be used to dislodge the entire veil-of-perception doctrine. When Philonous has coaxed Hylas out of a pre-

[p114]

Lockean position and into agreeing that ‘all sensible qualities beside the primary.., are only so many sensations or ideas existing no where but in the mind’, he sketches the next part of his strategy as follows:

Berkeley, then, sees Locke as having a view about secondary qualities which restricts or partly retracts the veil-of-perception doctrine, and a thesis about primary qualities which affirms a restricted version of the veil-of-perception doctrine.

This is a gravely wrong picture of how Locke’s central claim about primary and secondary qualities relates to his veil-of-perception doctrine, as I shall try to show. ‘But Berkeley didn’t intend it as a picture of Locke’s central claim–only of his peripheral "not in the object" version’–that defence of Berkeley is an over-simplification, but even if it were perfectly valid my points would still need to be made. I contend that Locke’s central claim, which is made up of the Analytic and Causal Theses fused together, has nothing to do with the veil-of-perception doctrine: it does not embody a version of that doctrine, or a qualification of or rival to it. The doctrine is just Locke’s mis-handling of the sceptical question about whether the objective realm is in any way at all as it appears to be; and I contend that this question must be answered affirmatively, on the basis of routine trust in the senses, before one can begin to expound and defend Locke’s central claim about the primary/secondary distinction. These contentions are non-trivial, in that their falsity is implied in virtually every commentary on this part of Berkeley’s work. This can be so only because most commentators do at least partly grasp Locke’s central claim, and are not completely in thrall to the language of ‘not in the object, because in the mind’. I believe that they have been led astray because the central claim itself, even if not expressed in the ‘not in the object’ form, can seem to be connected with the veil-of-perception doctrine. This pseudo-connexion is my main present concern.

One last reminder: the Analytic Thesis represents secondary

[p115]

qualities as dispositional, as relational, and as logically connected with something mental. Of these three features, it is the third which now assumes a major explanatory role. Because of that third feature, the following picture is plausible: the veil-of-perception doctrine says that statements about objects are logically dissociated from statements about ideas, and the Analytic Thesis re-affirms this for primary-quality attributions but concedes that it does not hold for statements purporting to attribute secondary qualities to things. If this were right, it would fully justify Berkeley’s tactics against Locke; but it is in fact wholly wrong.

If it were right, the Analytic Thesis would be a bore. Obviously some predicates of objects are logically connected with mental predicates: no one would doubt that we imply something about states of minds when we say that castor oil is nasty or that warm baths are soothing. If the Analytic Thesis were significant only as conceding that some predicates of objects are logically connected with mental predicates, it would have no significance at all. What gives it interest is not its saying (a) ‘Some predicates of objects have some logical connexions with mental predicates’, but rather its saying (b) ‘Secondary-quality predicates have these logical connexions with mental predicates’. Now, (b) does not tend to support the view that all predicates of objects are logically connected with mental predicates: Locke’s view about the status of secondary qualities is no more a step towards idealism or phenomenalism than is the Nazi valuation of Aryans a step towards a belief in the worth and dignity of all men.

Just as Locke’s claim about secondary qualities is not a significant restriction, so his claim about primary qualities is not a restricted version, of the veil-of-perception doctrine. If to (a) ‘Some predicates of objects have some logical connexions with mental predicates’ we add the rider ‘but primary-quality predicates don’t’, the result is indeed a form of the veil-of-perception doctrine, and does contradict idealism and phenomenalism. But if to (b) ‘Secondary-quality predicates have these logical connexions with mental predicates’ we add the rider ‘but primary-quality predicates don’t’, the result says only that primary-quality predicates are not connected with mental predicates in the way secondary-quality ones are. This does not challenge Berkeley or any competent phenomenalist: my defence of it in §20 conceded nothing to the veil-of-perception doctrine.

[p116]

Even those who reject Berkeley’s idealism, and phenomenalism as well, agree that his principal service to philosophy lay in his criticisms of Locke’s account of reality–his insight into what goes wrong if the distinction between appearance and reality is mishandled in certain ways. So it is lamentable that he should have muddied these waters by stirring in materials drawn not just from the substratum theory but also from Locke’s views about primary qualities. Here, for example, all three ingredients are present in concentrated form:

The mix-up does lead to positive error, of which I cite just one example. It is common ground among all philosophers who attach value to the primary/secondary distinction that objects do have primary qualities; the veil-of-perception doctrine allows that, just conceivably, ‘real things’ may have none of the properties we attribute to them; and substrata as such cannot have any properties at all. So Berkeley is prepared to say: ‘By matter... we are to understand an inert, senseless substance, in which extension, figure, and motion, do actually subsist,’ [45] and also:

‘The matter philosophers contend for, is an in comprehensible somewhat which hath none of those particular quali ties, whereby the bodies falling under our senses are distinguished one from another.’ [46] Berkeley is not here pin-pointing an inconsistency in Locke, but rather committing an inconsistency of his own which is generated by his misunderstanding of Locke’s problems.

The commentaries do not yield the same rich harvest of thorough, glad commissions of this mistake as they do of the one discussed in Chapter III. [47] Most commentators (again Armstrong is an exception) merely take Berkeley’s word for it that the veil-of-perception doctrine is integrally connected with the primary-quality thesis, and slide quickly over the gap where the connexion is supposed to be.

[p117]

25. The conflation’s sources in Locke

I can find no evidence that Locke saw his theory, or theories, about primary and secondary qualities as on the same level as the veil-of-perception doctrine. Positive evidence the other way occurs when he interrupts his discussion of primary and secondary qualities in order to say:

In a later passage he denies that any ‘correspondence or connexion’ can be found between our secondary-quality ideas ‘and those primary qualities which (experience shows us) produce them in us’. [49] It is true that Locke tries to confute the sceptic by covert appeals to empirical evidence; but even he would see that in the context of the anti-sceptical debate -- the veil-of-perception doctrine -- open references to ‘physical inquiries’ and to what ‘experience shows us' would be merely grotesque.

Nor does Locke confound his primary/secondary views with the issue about substratum-substance. In his principal exposition of the former, the word ‘substance’ does not occur once. [50]

Still, some aspects of Locke’s writings serve to explain, if not to excuse, the conflation by Berkeley and others of the primary-quality theory with the theory of reality.

One of these is the part played in Locke’s discussions of primary and secondary qualities by the notion of an idea’s resembling an object or a quality of an object. He sometimes says that there is such a resemblance in the case of primary-quality ideas but not secondary-quality ones; and this could tempt Berkeley to see the primary/secondary thesis as asserting the veil-of-perception doctrine for primary qualities while denying it for secondary.

But ‘resemblance’ plays only a minor role in Locke’s formulations of the primary/secondary thesis; and his standard way of presenting the veil-of-perception doctrine, or the issue over

[p118]

scepticism, is not in terms of resemblance between ideas and real things. The arguments I have quoted in § iz, indeed, prove that Locke sees the sceptical question to be neutral as between primary and secondary qualities. The sceptical question asks: ‘Are all our ideas just hallucinations or dream-states or the like ?‘, and the Berkeleian exegesis would have Locke reply: ‘Not all our ideas–only the secondary-quality ones.’ But in two of Locke’s most considerable discussions of scepticism he defends the veridicality of our ideas against the sceptic in terms of just four examples -- the taste of wormwood, the smell of a rose, the heat of a fire, and the whiteness of a page! [51]

That alone makes Berkeley’s exegesis inexcusable. Still, perhaps we should hunt out some more of its possible sources in Locke’s text.

Locke’s unsatisfactory classification of ideas into ‘real’ and ‘fantastical’ might entice one into Berkeleian errors. [52] He says that ‘real’ ideas do, while ‘fantastical’ ones do not, ‘have a conformity with the real being and existence of things, or with their archetypes’. [53] This might lie within the area of the veil-of-perception doctrine, e.g. by being connected with Locke’s notion of ‘real knowledge’. [54] Yet it is not clear that that is right. The ideas Locke calls ‘fantastical’ are ones ‘made up of such collections of simple ideas as were really never united, never were found together in any substance: v.g. a rational creature, consisting of a horse’s head, joined to a body of human shape. [55] That is, they are ideas which fail in an ordinary everyday way to ‘conform’ to ‘reality’. Is there room here for even the prima facie possibility that all our ideas might be ‘fantastical’? The real/ fantastical distinction is in fact a rather muddled piece of philosophical logic, concerning ideas in their role as meanings rather than as sense-data, so that its relations with the veil-of-perception doctrine are bound to be cloudy.

Even if the two were tightly linked, though, the real/fantastical distinction would support Berkeley’s exegesis only if Locke also said that primary-quality ideas are ‘real’ while secondary-quality ones are ‘fantastical’. But this second step is explicitly ruled out:

[p119]

There is, then, no support for Berkeley in the distinction between real and fantastical ideas.

Locke’s theory about real essences could also lead to misunderstanding. While pouring scorn on ‘those who [use] the word essence for they know not what’, [57] Locke gives to the phrase ‘real essence’ a sense which he believes to be legitimate:

From now on I shall use ‘real essence’ to mean ‘real essence of the kind Locke approves’, i.e. the microphysical primary-quality constitution of a thing from which flow all its large-scale qualities, primary and secondary. The theory of real essences is, then, integrally bound up with Locke’s Causal Thesis about primary and secondary qualities.

Locke’s notion of real essence is one of the finest things in the Essay, and he handles it almost flawlessly. Yet a careless reader might assimilate it to the theory of reality, or to that of substance, because of Locke’s insistence on how little we know about real essences:

[p120]

Here he is preparing to argue that real essences are not in fact our basis for classification: they cannot be, since we know so little about them. [60] But the passage has a larger purpose which should also be mentioned. With characteristic intelligence, insight and humility, Locke took every possible chance–of which the quoted passage is one–to stress the gap between the intellectual control which we do impose on the world and the science-plus-conceptual-scheme which we might find appropriate if we ‘cured our ignorance’.

Sometimes, however, he gives the impression of thinking that real essences are not just mainly unknown but in principle unknowable. For example, he says that ‘we are so far from being admitted into the secrets of nature, that we scarce so much as ever approach the first entrance towards them’, and defends this in a fine passage which starts thus:

Against this common assumption, Locke maintains that individual things are continually propped up, so to speak, by their causal relations with other things:

He concludes in his next section that, as well as the difficulty of discovering the ‘size, figure and texture’ of the ‘minute and active parts’ of bodies, there is the problem of finding ‘the different motions and impulses made in and upon them by bodies from without’; but these motions and impulses are also part of the real essences of bodies, for upon them ‘depends.. . the greatest and

[p121]

most remarkable part of those qualities we observe’ in bodies; and so ‘this consideration alone is enough to put an end to all our hopes of ever having the ideas of [bodies’] real essences’. [63]

Is Locke here presenting real essences as necessarily unknowable–as a ‘something we know not what’ or a something beyond the veil? If so, that might link the notion of a real essence, and thus the primary/secondary theory containing it, with either substrata or the ‘real things’ of the veil-of-perception doctrine. The answer to the question is ‘No’. Even when Locke is at his most pessimistic about our prospects of discovering the real essences of things, his pessimism is reasonable, argued, and unabsolute. [64] What he says is not that real essences are in principle unknowable, but only that there are reasons for suspecting that full knowledge of them would require scientific inquiries of a depth and scope that lie beyond our capacities.

It would be in the spirit of Locke’s theory of essences to suggest that we might, as scientific knowledge increases, eventually give up all our present ways of describing and classifying physical things. This suggestion could, unhappily, be expressed by saying ‘Perhaps nothing in the physical world is in any respect what it seems to be’; but this sentence, when it carries that meaning, has nothing to do with the veil-of-perception doctrine. Berkeley notwithstanding, it does not express a ‘sceptical’ position, nor does it in any significant sense ‘depreciate our faculties, and make mankind appear ignorant and low’. [65] Berkeley’s inability to do justice to this aspect of Locke’s thought seems to have been a matter of temperament as much as of intellect–apparently he could not get himself into the frame of mind of someone who, without jeering at the ignorance of others or bewailing his own, thinks it likely that science has so far barely scratched the surface of the real.

Another aspect of the theory of real essences could, at a hasty glance, seem to connect it with the substratum doctrine in particular. What Locke stresses is the difficulty of discovering the real essences of substances; and someone might infer from this that real essences are supposed to be unknowable in the same way, and perhaps for the same reason, that the ‘nature’ of substratum-

[p122]

substance is unknowable. This too would be a misunderstanding. Locke is interested in the distinction between real and ‘nominal’ essences, and this distinction works only in application to substances–a fact which Locke records, not very satisfactorily, by saying that the nominal essence of a property or mode is also its real essence. [66] The basic point underlying this dark saying is, I think, as follows.

The nominal essence of a red thing (qua red) is just the meaning of ‘red’ or the idea we associate with that word. The real essence of a red thing (qua red) is that primary-quality texture of it which causes it to look as it does and thus qualify for the description ‘red’. To discover a real essence, then, we must start with a meaning or an ‘idea’, and then dig for certain empirical facts about the items–the substances–which correspond to it. Consider now the property of redness. Its nominal essence is just the same as that of any red thing (qua red), that is, it is determined wholly by the meaning of the word ‘red’. And that is all there is to be said about the property itself, as distinct from the things that have it. The only way of going beyond what we mean by ‘red’ or by ‘redness’ -- the only way of filling in relevant empirical facts–is by going from redness to red things, from properties to substances. (Locke’s own account of the matter is clouded by his speaking not of ‘properties’ but of ‘simple ideas and modes’, and by his implicitly contrasting ‘triangles [of the sort studied in geometry]’ with triangular things or substances. If I was right in § about why he chooses geometrical ‘triangles’ as paradigm modes, then we can take him to be contrasting the property of triangularity with actual triangular things or substances; and then his basic point is as I have expounded it.)

Locke’s point, on this reading of it, is perfectly correct; there is no mystery about it; and it has nothing to do with substrata. For example, when he refers to ‘the unknown essence of that substance’, [67] all he means is ‘the microphysical constitution of that lump of stuff’ or .... of that kind of stuff’. He is, quite harmlessly, using the language of his excellent theory of real essences.

[p123]

[end]