BRING BACK SUBSTANCES! Picture an everyday object such as a vase. It will have certain properties. Perhaps it is pear-shaped, blue, 2kg and smooth. There is an important difference between the vase and its properties. The vase is object-like. It is a thing in the strictest sense of “thing.” It is the kind of thing we naturally denote using a noun: we prefer to say, “there is a vase on top of the table” not “the top of the table is vase-ish.” The shape of the vase is un-object-like. It is a thing only in a permissive sense of “thing.” It is the kind of thing we naturally denote using an adjective: we prefer to say, “the vase is pear-shaped,” not “the vase has a pear-shaped-ness.” Countless other items belong in the same group as the vase: plants, animals, planets and particles for example. Likewise, countless other items belong in the same group as the properties of the vase: sizes, timbres, masses and charges for instance. The distinction between object-like and un-object-like things is ubiquitous. It appears to be a part of the basic structure of reality. So anyone interested in understanding the basic structure of reality should be interested in understanding this distinction. How should we understand the distinction between object-like and un-object-like things? This essay defends a simple, traditional, but unpopular answer. I propose that the difference between object-like and un-object-like things is the difference between things that can and cannot exist by themselves. More picturesquely, object-like things meet the minimum threshold for being. You cannot order less Château Margaux than a glass, and you cannot order less reality than an object-like thing such as a vase. I find the idea of something that could exist by itself as clear, simple and vivid as any idea in philosophy. For this reason, I consider it suitable for use in our most severely rigorous theorizing. Furthermore, I think the idea of something that could exist by itself plays an important part in the history of philosophy. For in my opinion, this just is Descartes’s idea of a substance, one that he takes, largely unaltered, from Aristotle. In urging that what distinguishes object-like things is their ability to exist by themselves, I take myself to be urging that we embrace a traditional Aristotelian-Cartesian concept of substance. I seem to be all but alone however, both in my judgement that the idea of something that can exist by itself is clear, simple and vivid, and in my conviction that this is what Descartes means by “substance.” In startling contrast to my view, the literature on this theme tends to present the idea of something that can exist by itself not as clear, simple and vivid, but as incoherent. And in part for this reason, and for no other good reason I can find, commentators tend to suppose that this cannot be what Descartes means by “substance.” I hope this essay will alleviate my loneliness in both respects. It should be clear that this essay has twin aims, one theoretical the other historical. I defend the idea of substances as things that can exist by themselves both as a valuable tool for present-day philosophy and as an interpretation of Descartes. It is expedient to pursue these aims together because the same objection to the idea of substances as things that can exist by themselves has prevented its acceptance in both contexts. Here, as often, philosophy and the history of philosophy are most efficiently undertaken at once. My primary objective is to defend the coherence of the idea of substances as things that can exist by themselves. I set upon that task in earnest in §§ 5–6. But it might be helpful to sketch right away why philosophers reject the coherence of the concept of substance that I defend, and why I think that they are mistaken. I discuss several examples of philosophers who reject the idea of substances as things that can exist by themselves in § 5. These include Gottfried Leibniz, “Critical Thoughts on the General Part of the Principles of Descartes,” in Gottfried Leibniz, Philosophical Papers and Letters, 2nd ed., ed. and trans. Leroy Loemker (Alphen aan den Rijn: Kluwer, 1989); Peter Markie, “Descartes’ Concept of Substance,” in Reason, Will and Sensation: Studies in Descartes, ed. John Cottingham (Oxford: Clarendon Press, 1994); Joshua Hoffman and Gary Rosenkrantz, Substance: Its Nature and Existence (London: Routledge, 1997); E. J. Lowe, The Possibility of Metaphysics: Substance, Identity and Time (Oxford: Clarendon Press, 1998); Matthew Stuart, “Descartes’s Extended Substances,” New Essays on the Rationalists, eds. Rocco Gennaro and Charles Huenemann (Oxford: Oxford University Press, 2003); Benjamin Sebastian Schneider, “A Certain Kind of Trinity: Dependence, Substance, Explanation,” Philosophical Studies 129, (2006): 393–419; Gonzalo Rodriguez-Pereyra, “Descartes’s Substance Dualism and His Independence Conception of Substance,” Journal of the History of Philosophy 46, no. 1 (2008): 69–90; Patrick Toner, “Independence Accounts of Substance and Substantial Parts,” Philosophical Studies 155 no. 1 (2011): 281–297; and Richard Swinburne, “Cartesian Substance Dualism,” in The Blackwell Companion to Substance Dualism, eds. Loose, Jonathan, Angus Menuge, and J. P. Moreland (Oxford: Blackwell, 2018). Their principal reason for doing so is that they suppose that to exist by itself something would have to exist without any property with which it is not identical, and they judge that this makes no sense: as a matter of conceptual necessity everything must have some properties with which it is not identical. To put this another way, no candidate substance could exist entirely by itself because everything depends, for its existence, on its own properties. For example, a vase cannot exist without a shape or a texture, and a vase is not identical to its shape or its texture, so a vase cannot exist by itself. Likewise for other candidate substances. Philosophers have contorted the idea of substances in outlandish ways to avoid this problem. This is one reason why many now prefer to do without the idea of substance entirely. But I believe that such contortions have been gratuitous. For it seems to me—and this is the crucial claim of this essay—that the belief that the idea of something existing by itself is incoherent results from a simple misunderstanding. I have no objection to the suggestion that everything must have properties with which it is not identical. What I object to is the assumption that for something to exist by itself is for it to exist without anything nonidentical to it. This, I urge, is dead wrong: it is a very odd way to understand the idea of something existing by itself, and one that exemplifies a regrettable habit philosophers have of interposing the identity relation where it is not wanted. I conclude, on this basis, that the charge of incoherence is a chimera. The position I defend is unusual. But I reach it from a wide plateau of common ground. For my opponents recognize that the idea of substance that I defend is prima facie appealing. And I agree with them that nothing could exist without properties with which it is not identical. Our only difference is that they see these points of agreement as in tension with one another, whereas I argue that they are in harmony. This should be welcome news. Section 2 explains why we need the idea of substances as things that can exist by themselves. Sections 3–4 sketch the origins of this idea in Aristotle and Descartes. Sections 5–6 address the charge of incoherence. Section 7 considers some further objections. If my arguments are successful then philosophers should treat the idea of substances as things that could exist by themselves much more seriously than they have done, both as an important concept in the history of philosophy and as one that might serve us well today. We dismiss this idea at the cost of misunderstanding the history of philosophy and of depriving ourselves of a valuable tool for understanding the structure of reality. II The case for substances. The difference between the vase and its shape, color or texture is the difference between a property-bearer and its properties. So I am using “object-like” and “un-object-like” to highlight a contrast between the vase and its properties and not between the vase and things that might be thought un-object-like in some other way, such as the way in which liquids are less object-like than solids or the way in which occurrents are less object-like than continuants. The difference I am interested in might not be exclusively a difference between property-bearers and properties, but that is the paradigm case. This section argues that we need the idea of substances as things that can exist by themselves because it best explains what makes object-like things object-like. I do not claim that substances, so characterized, also best answer the question “which entities are fundamental?” The project of explaining what makes object-like things object-like and the project of identifying which entities are fundamental are frequently confused in the literature on substances. They are helpfully distinguished by Kathrin Koslicki and Howard Robinson. Kathrin Koslicki, Form, Matter, Substance (Oxford: Oxford University Press, 2018), 164–7; Howard Robinson, “Substance” in The Stanford Encyclopedia of Philosophy ed. Edward Zalta https://plato.stanford.edu/archives/spr2020/entries/substance/. For a comprehensive overview of recent approaches to substances understood as the fundamental entities see Donnchadh O’Conaill, Substance (Cambridge: Cambridge University Press, forthcoming). You might doubt that we need to resurrect the concept of substance to explain what makes object-like things object-like. For it is sometimes supposed that this can be done with more fashionable concepts. The main candidates are particularity and concreteness, along with their complements, universality and abstractness. A particular is an unrepeatable entity, something that can feature in the world only once. A universal is a repeatable entity, something that can feature in the world multiple times. It seems obvious that the vase is a particular: the vase is unrepeatable, there may be others like it, but only this one is it. But many philosophers would say that the properties of the vase are universals. For example, if the vase is the same color as the sky, they would say that there is something, the color blue, that is literally one and the same in both cases. A concrete thing is a spatiotemporal, causally active entity. An abstract thing is a non-spatiotemporal, causally inert entity. It seems obvious that the vase is concrete: it occupies space and time; you can see it, touch it, throw it. But many philosophers would say that the properties of the vase are abstract. For example, they would say that the color blue does not exist in space or time; you cannot touch it or throw it; strictly speaking, you cannot even see it, though you can, of course, see blue things. (This is the predominant usage of “concrete” and “abstract”. I introduce a second usage later in this section.) It is tempting to conjecture that the difference between object-like and un-object-like things is the difference between particulars and universals and/or between the concreta and abstracta. But a little reflection suggests that this cannot be right. For there seem to be examples of concrete particulars that are un-object-like in precisely the way that properties are. If so, then particularity and concreteness are neither independently nor jointly sufficient for being object-like in the way that the vase is object-like. To see this, compare the vase not to its shape or its color but to the dent on its left shoulder or to the scratch beneath its rim. Dents and scratches exist in space and time. And dents and scratches are unrepeatable: two vases cannot share literally the same dent. And yet the vase’s dents and scratches strike us as un-object-like in much the way that its shape and color do. These are all thin, attenuated beings that belong to the vase in a way in which the vase belongs to nothing else. Dents and scratches seem to be un-object-like concrete particulars. So too do bulges, impressions, smiles, postures, glows, gleams, states, surfaces and countless other items. If this is correct then we need an account of the difference that obtains between the vase and the kind of un-object-like concrete particulars listed here. Schneider, “A Certain Kind of Trinity,” 396 makes roughly this point. In fact, even without these examples, I would doubt that the particular-universal distinction or the concrete-abstract distinction can capture the difference between object-like and un-object-like things. For I accept the traditional argument that universals must be abstract, since a concrete universal would have, per impossibile, to be wholly present in multiple locations. See David Armstrong’s defense of concrete universals, Universals and Scientific Realism, vol. 1 (Cambridge: Cambridge University Press, 1978), 108. And I follow E. J. Lowe in thinking that if we posit abstract properties these must complement and not replace concrete properties, since properties do causal work. Lowe, The Possibility of Metaphysics, 98–100. These arguments jointly entail that there exist concrete particular properties. Again, it follows that particularity and concreteness are neither independently nor jointly sufficient for being object-like. Furthermore, if we do posit concrete universals, it is hard to see how this will help us to explain the difference between object-like and un-object-like concreta. For the question arises: why should we identify concrete universals exclusively with un-object-like things? To do so seems oddly unprincipled. To see this, suppose that both I and Lord Bamford own silver Aston Martin DB5s. And suppose the two cars are qualitatively identical, although Bamford’s is in his garage at Daylesford whereas mine is aboard my yacht. Proponents of concrete universals ask us to accept that the shape, color and texture of my car are concrete components of it that are numerically identical with the corresponding components of Bamford’s car. (If shape, color or texture does not make it to your roll call of “sparse” properties choose your own examples.) On the other hand, they tell us that the two cars are not numerically identical: they merely resemble one another in virtue of sharing the same properties. What I want to know is, on what basis do proponents of concrete universals make this judgement? Why not say that Bamford and I own one and the same car? They appear to have no principled answer to this question. The obvious answer is that Bamford’s car is at Daylesford whereas mine is on my yacht, and so they cannot be identical. In his classic defense of concrete universals, David Armstrong implies that this is the answer he would give. Armstrong, Universals, vol. 1, pp. 118–24. But it is far from clear that proponents of concrete universals can give this answer. For if we judge that the cars are distinct particulars on the basis of their different locations, consistency seems to require that we say the same of their properties, since concrete properties exist at spatial locations too. And if we judge that the properties are recurrent universals on the basis of their qualitative identity, consistency seems to require that we say the same of the car(s). To be clear, I have only the faintest idea of what it would be for a car to be wholly present in two places. (An appearance of conceivability can be given to this idea by supposing that the car time-travels back to a moment when it already exists, à la Back to the Future, though the conceivability of time travel is itself suspect.) But I also have only the faintest idea what it would be for the very same concrete shape, texture or color to be wholly present in two places. The two cases seem parallel. I put off further discussion of these arguments for a more detailed treatise and presume hereon that there is indeed an important difference between object-like and un-object-like things that is neither the difference between particulars and universals, nor that between concreta and abstracta, nor yet a combination of the two. What is this difference? A convenient route to the answer I favor goes via a second usage of “concrete” and “abstract,” one that is also sometimes applied to property-bearers and their properties, though it differs importantly from the usage introduced above and is less prevalent today. As I have said, in their predominant usage, “concrete” applies to causally active spatiotemporal things and “abstract” applies to acausal non-spatiotemporal things that exist in some kind of metaphysical Elysium. But there is a second, less common usage, on which “concrete” applies to things that are in some sense whole or complete whereas “abstract” applies to things that are fragmentary or incomplete. “Abstract” things in this sense can be as spatiotemporal and causally vigorous as you like. It is regrettable that the same words have ended up with such different usages, both of which are highly relevant to this enquiry. Quine is at least partly to blame. Still, if we sharpen our concentration to a nanotip, we can hope to avert confusion. On the second usage, abstractness is “the trait of what is less than its including whole.” D. C. Williams, “On the Elements of Being I,” The Review of Metaphysics 7, no. 1 (September 1953): 15. It is usually added that we bring to mind things that are abstract in this sense by an act of abstraction, “a process of selection, of systematic setting aside.” Keith Campbell, Abstract Particulars (Oxford: Blackwell, 1990), 3; cf. Ledger Wood, “Abstract,” in Dictionary of Philosophy, ed. Dagobert Runes (London: Routledge, 1944), 2; David Lewis, On the Plurality of Worlds (Oxford: Blackwell, 1986), 85–6. You start with the whole vase and then you set aside elements of it until you are left with a dramatically different kind of entity. This second concrete-abstract distinction is also sometimes alleged to capture the difference between object-like and un-object-like things. And this proposal is superior to those considered above, in that it promises to class items such as scratches and dents as un-object-like things, along with shapes, textures and so on, even though they are, in the first sense of “concrete,” concrete particulars. For intuitively, these things are all fragmentary or incomplete elements of the whole vase, which we grasp by an act of abstraction. Though promising, this proposal needs work. For almost anything can be thought of as a fragment of some including whole. This is true of the vase’s scratches and dents. But the same is true of its handle and lid. Even the whole vase can be seen as a mere fragment of the Hall of Supreme Harmony or wherever it happens to be. This is a well-known problem for the second concrete-abstract distinction. The classic exposition of that distinction is D. C. Williams’ celebrated 1953 dilogy. Williams, “Elements of Being I,”; Williams “On the Elements of Being II,” The Review of Metaphysics 7, no. 2 (1953): 171–192. After defining abstracta as incomplete fragments of greater wholes, Williams observes that this might describe anything but the “World All.” Williams, “Elements of Being I,” 15. But having recognized this problem, he adopts a relaxed attitude to it. Williams simply recommends that we use “abstract” for things that have the “special sort of incompleteness” of “fine” or “diffuse” entities, and “concrete” for things that are comparatively “thick” or “gross.” Besides a few evocative adjectives, Williams does not specify the difference between the pedestrian kind of incompleteness possessed by everything but the “World All” and the exceptional kind that characterizes colors, dents and so on. Neither do most proponents of the second concrete-abstract distinction. As such, the second concrete-abstract distinction gives us a promising indication of what the difference between object-like and un-object-like things consists in, but one that is distressingly imprecise. If we want a precise account of the difference between object-like and un-object-like things we need to determine what makes the kind of incompleteness that characterizes un-object-like things “special.” Why is it that the term “incomplete” seems especially appropriate to shapes, colors and textures, compared to lids, handles or vases? It is here that historical discussions of substance seem to be of value. For a plausible hypothesis says that the special incompleteness of abstract things, in the second sense of “abstract,” consists in the fact that in bringing these things to mind we set aside things without which they could not exist. On this view, abstraction is the act of presticogitation whereby we separate in thought what cannot be separated in reality. The completeness of concrete things will then consist in the fact that they do not need anything further to exist. Equivalently, concrete things are substances in the sense that they can exist by themselves. Williams himself acknowledges the appeal of this hypothesis. Williams, “Elements of Being II,” 179. But he drops it without much discussion. Williams is too sanguine. We need some account of the difference between object-like and un-object-like things. The thesis that this is the difference between substances and non-substances is the most promising view on offer. The support for this account comes primarily from modal intuition. Intuitively, the lid and handle of the vase really could exist without the rest of the vase, and the vase without its surroundings, whereas the shape, texture, scratches and dents depend, for their existence, on their bearer. A secondary source of support is historical precedent. (Arguments from authority do not add much weight, but they do not add none.) For, as I explain in the next section, the idea that object-like things can, whereas un-object-like things cannot exist by themselves, plays a distinguished part in the history of philosophy. Modal intuitions and historical precedent can turn out to be wrong. But they merit our provisional credence. Especially when, as in the present case, they promise to make sense of something that we are otherwise at a loss to explain. The idea of substances as things that can exist by themselves promises to account for a distinction that embarrasses the standard resources of present-day philosophy. A more specific reason why we need this conception of substance is to make sense of the distinction between “substance dualism” and “property dualism” in the philosophy of mind. I discuss this briefly in Weir, “Can a Post-Galilean Science of Consciousness Avoid Substance Dualism?” Journal of Consciousness Studies (2021). III Substances: the tradition. I suggest that the difference between object-like things and un-object-like things is the difference between things that can and cannot exist by themselves. Furthermore, I see this distinction as a valuable heirloom of Western philosophy, one that can be traced to Aristotle’s Categories and which is first unambiguously expressed (so far as I know) by Descartes. This section sets out the history of the idea of substance that I am defending as I see it. The next section defends the reading of Descartes adopted here. Aristotle tells us that some things exist “in a subject” in the way that the color white is “in” an individual body. Aristotle, Categories and De Interpretatione, ed. trans. J. L. Ackrill. (Oxford: Clarendon, 1963), 1a20–b9. Other things are “said of a subject” in the way that “animal” or “two-footed” is said of an individual human. On a popular interpretation, accidental universals are both “said of” a subject and exist “in” a subject, essential universals are “said of” a subject but do not exist “in” a subject, and accidental particulars exist “in” a subject but are not “said of” a subject. See Paul Studtmann, “Aristotle's Categories,” The Stanford Encyclopedia of Philosophy, ed. Edward Zalta https://plato.stanford.edu/archives/fall2018/entries/aristotle-categories/, § 1. Things of yet a third kind are neither “in a subject” nor “said of a subject,” for example an individual human or an individual horse. Aristotle calls things in the third group “substances” (“ousiai,” a noun derived from the verb eimi, “I exist,” cognate with the English “am.” “Substance” derives from early Latin translations of “ousia” and has a very different etymology). Substances have a kind of pre-eminence in the Categories because they are the only things that are neither “said of” nor “in” anything else, and everything else is “said of” or “in” a substance. Aristotle adds that “substance” also applies in a secondary way to the species and genera of substances of the primary sort, but I will not discuss that usage here. The color white in an individual body is an un-object-like thing. Likewise the genus animal and the differentia two-footed, insofar as these are taken to be universals, as readers of Aristotle usually suppose. See Porphyry, Isagoge et in Aristoteles Categorias Commentarium, ed. Adolf Busse (Berlin: Georg Reimer, 1887), § 1; W. D. Ross, Aristotle, (London: Methuen, 1923); Studtmann, “Aristotle's Categories,” § 1. (Some theorists identify genera with the aggregates of their members, but Aristotle shows no sign of doing so.) By contrast, humans and horses are object-like property-bearers. So too are their parts, such as hands and heads, and Aristotle also classes these as substances. Aristotle, Categories, 8b15–19. Aristotle’s distinction between substances and non-substances in the Categories, then, resembles the intuitive distinction between object-like and un-object-like things. It is therefore reasonable to hope that the “said of” and “in” relations that serve as Aristotle’s definientia will provide an insight into what makes object-like things object-like. Aristotle does not say much about these relations. But he does explain that when he speaks of things that are “in” a subject he means “what is in something not as a part, and cannot exist separately from what it is in.” Ibid., 1a24–5. Furthermore, Aristotle says that: All the other things are either said of the primary substances as subjects or in them as subjects. So if the primary substances did not exist it would be impossible for any of the other things to exist. Ibid., 2b5–6. These passages imply that “said of” and “in” mark relations of existential dependence. That is, if a is “said of” or “in” b, in the sense that Aristotle has in mind, then a depends for its existence on b. And since substances are neither “said of” nor “in” anything, this suggests, though it does not entail, that substances are not existentially dependent on anything. The idea that Aristotle’s substances are characterized by a kind of existential independence is widespread. If it is correct then, in my judgement, Aristotle’s idea of substance in the Categories is essentially the idea defended in this essay: a substance is an object-like thing, understood as something capable of existing by itself. For two reasons, however, it would be incautious to claim Aristotle’s authority for the understanding of substance I defend. First, the thesis that the substances of the Categories are existentially independent is a conjecture: it depends on the assumption that substances do not depend on other things for their existence in some third way, that does not involve being “said of” or “in” them. Secondly, commentators dispute what the existential independence of substances consists in for Aristotle, supposing his substances are existentially independent in the first place. See Gail Fine, “Plato and Aristotle on Form and Substance,” Proceedings of the Cambridge Aristotle Society 209 (1983): 23–47; Gail Fine, “Separation,” Oxford Studies in Ancient Philosophy 2 (1984): 31–88; Phil Corkum, “Aristotle on Ontological Dependence,” Phronesis 53, no. 1 (2008): 65–92. It might be possible to clarify Aristotle’s position by drawing on other works. But the interpretive issues are too intricate to address here. And in any case, it is the Categories that has played the leading role in imparting the idea of substance to subsequent thinkers. The influence of the Categories is immense. By late antiquity it had acquired a unique pedagogical role, along with an introduction by Porphyry, as the first text standardly taught to philosophy students throughout the Roman world. Friedrich Solmsen, “Boethius and the History of the Organon,” American Journal of Philology 65 (1944): 69–74; Giorgio Pini, “Reading Aristotle's Categories as an Introduction to Logic: Later Medieval Discussions about its Place in the Aristotelian Corpus,” in Medieval Commentaries on Aristotle's Categories, ed. Lloyd Newton (Leiden: Brill, 2008), 148–51. Boethius’s translation transmitted this practice to the Latin Middle Ages, and the Categories remained the leading introduction to logic in particular and to philosophy in general until it was deposed by the 1662 Port Royal Logic. As Giorgio Pini says: This short treatise (only 15 pages in Bekker’s edition) was the only philosophical treatise that has been uninterruptedly read, studied and commented on since Antiquity. No other single work had a comparable influence in Western philosophy. Pini, “Reading Aristotle’s Categories,” 145. The idea of substance introduced in the Categories, then, appears on the first page of the first text of a philosophy syllabus that educated the Western world for at least 1200 years. As a result, it features in the works of more or less all of the leading thinkers running from antiquity to early modernity. Few ideas in philosophy enjoy such a shining career. The Categories suggests that un-object-like properties cannot exist by themselves. But it does not explicitly state that their object-like bearers can exist by themselves. For this reason the Categories does not fully enunciate the idea of substance that I defend. And I know of no work of antiquity or the Middle Ages that does. It is only in Descartes, to my knowledge and in my opinion, that the idea of substance that I defend fully crystallizes. Descartes has two definitions of substance. The first appears in the Second Replies and reads as follows: Substance. This term applies to every thing in which whatever we perceive immediately resides, as in a subject, or to every thing by means of which whatever we perceive exists. By “whatever we perceive” is meant any property, quality or attribute of which we have a real idea. Rene Descartes, Oeuvres de Descartes, ed. Charles Adam and Paul Tannery, revised edition, (Paris: Vrin/C.N.R.S., 1964–76), vol. 7, 161. Translations of Descartes are from Rene Descartes, The Philosophical Writings of Descartes, eds. and trans. John Cottingham, Robert Stoothoff, Dugald Murdoch, and Anthony Kenny (Cambridge: Cambridge University Press, 1984–1991) (‘CSMK’). Descartes does not say that substances do not “reside” in other things in this way, but I assume that this is implied. If so, the Second Replies definition appears to be the Categories definition with the “said of” relation omitted. This is unremarkable. For the ancient tradition that begins philosophy education with Aristotle’s logical works was the basis of the Jesuit curriculum Descartes studied at La Flèche. Society of Jesus, Ratio Atque Institutio Studiorem Societatis Iesu. Monumenta Paedagogica Societatis Iesu, ed. Ladislaus Lucas SJ (Rome: Institutum Historicum Societatis Iesu, 1986), vol. 5, 398. Even Descartes’s Latin reflects the idiom in which the Categories definition was expressed at this time (this had changed since Boethius). Compare Descartes’s “ut in subjecto... existet aliquid” to Pedro da Fonseca’s “existens in alio ut in subjecto” in one of two textbooks in Aristotelian logic recommended on the curriculum Descartes studied. Descartes, Oeuvres, vol. 7, 161; Pedro da Fonseca, Institutionum Dialecticarum (Lisbon: Libri Octo, 1564). Descartes’s omission of the “said of” relation also reflects Fonseca. So the Second Replies definition is essentially the Categories definition. And like the Categories definition, it tells us that nonsubstances are existentially dependent without telling us explicitly that substances are existentially independent. In the Second Replies, Descartes is no easier to interpret on this point than Aristotle. But the Second Replies does not give us Descartes’s only definition of substance. In the Meditations, Fourth Replies, the letter to “Hyperaspistes” and the Principles of Philosophy Descartes adopts an alternative approach which I will call the “independence definition.” Descartes, Oeuvres, vol. 7, 44, 226; vol. 3, 429; vol. 8a, 24. The Principles statement of the independence definition reads as follows: By substance we can understand nothing other than a thing which exists in such a way as to depend on no other thing for its existence. Ibid., vol. 8a, 24. Descartes adds that only God satisfies this definition in an unqualified way but that we can class as “created substances” things “that need only the concurrence of God in order to exist.” Ibid., vol. 8a, 25. In my judgement, the independence definition says of substances exactly what I suggested of object-like things in § 2: substances are existentially independent things in the sense that they are capable of existing by themselves (or for created substances, by themselves setting aside God). I assume this interpretation for now and defend it in § 4. It is sometimes thought odd that Descartes should adopt two definitions of substance, but I see no cause for puzzlement. In the Second Replies, Descartes endorses Aristotle’s Categories definition. Elsewhere, he disambiguates that definition in favor of the existential independence of substances. The Categories definition has not been abandoned, merely clarified. It is plausible that in doing this Descartes does not even take himself to be departing from Aristotle. As William Kneale says: In many text-books of metaphysics after Aristotle's time a substance was defined as a thing which required nothing else for its own existence. The persons who first used this phrase probably intended it to be a summary of the remarks I have quoted from Aristotle. William Kneale, “The Notion of a Substance,” Proceedings of the Aristotelian Society 40 (1939–1940): 105. The “remarks” Kneale has quoted are the Categories definition and Aristotle’s comments upon it. Kneale does not identify the inventor of the independence definition. But his hypothesis is only strengthened when we suppose this to be Descartes. For Descartes never acknowledges that he is working with a new idea of substance or with two competing ideas. On the contrary, in both the Second Replies and the Principles Descartes insists that we can have only one idea of substance. Descartes, Oeuvres, vol. 7, 161; vol. 8a, 24. But if it is philosophically unadventurous, Descartes’s independence definition is historically pivotal. For after 1662 the Categories (with the rest of the Organon, restored to the West in the High Middle Ages) yields its unrivalled status in philosophical education to an emphatically Cartesian substitute, the Port Royal Logic. And like the Categories, the Port Royal Logic does not focus solely on what we now call “logic.” It also covers a considerable amount of metaphysics, including the independence definition of substance. Antoine Arnauld and Pierre Nicole, Logic or the Art of Thinking, ed. Jill Vance Buroker (Cambridge: Cambridge University Press, 2012), 30. The influence of the Port Royal Logic—more properly, L'Art de penser—after 1662 is comparable to the influence of the Categories before that date: Between 1662 and 1887 it was printed at least 63 times in French, 13 times in Latin, and 16 times in English. L'Art de penser was the dominant logic text in the Western world for over 200 years… Cartesian notions on a myriad of subjects found their way into the working thought of Western philosophers as much through L'Art de penser as through Descartes’s own works. Richard Watson, “The Port-Royal Logic in the Twentieth Century,” Journal of the History of Philosophy 5, no. 1 (January 1967): 55. Editions of the Port Royal Logic and its “copy-and-paste” descendants Cf. Rick Kennedy, “The Alliance between Puritanism and Cartesian Logic at Harvard 1687–1735,” Journal of the History of Ideas 51, no. 4 (October–December 1990): 549–572. probably make up the weight of Kneale’s “text-books of metaphysics.” These quickly entered the university curricula of the eighteenth century and greatly contributed to the influence of Descartes’s ideas. In summary then, the independence definition can be seen as a disambiguation of the time-honored Categories definition in favor of the existential independence of substances. For twelve centuries the Categories definition was taught to practically every philosophy student in the Western world. For the following two centuries the same was true of the independence definition. This tradition is on the verge of extinction. But for the reasons given in § 2, I claim that it should be rescued. A central reason why substances are near extinction is the belief that the independence definition is incoherent. As I explain in § 5, the charge of incoherence can be traced to Leibniz and exercises a great influence in the recent literature. But there is a second reason for the recent unpopularity of substances which deserves a brief mention here. Aristotle’s and Descartes’s substances are object-like things: humans, horses, minds and bodies. But there exists a different understanding of substance whose locus classicus is the following passage of Locke’s Essay: Substance [is] nothing, but the supposed, but unknown support of those qualities, we find existing, which we imagine cannot subsist, sine re substante, without something to support them, we call that support substantia; which, according to the true import of the word, is in plain English, standing under, or upholding. John Locke, An Essay Concerning Human Understanding, ed. R. Woolhouse (London: Penguin, 1997), vol. 23, § 2. Locke’s substances are not object-like things but recondite metaphysical postulates that must be combined with properties to yield object-like things. (Note how misleading is Locke’s allusion to the etymology of “substantia.”) Lockean substances are very different from Aristotelian-Cartesian substances. Cf. Markie, “Descartes’s Concept of Substance,” 75–6; David Wiggins, “Substance,” in Philosophy: A Guide Through the Subject, ed. A. C. Grayling (Oxford: Oxford University Press, 1995), 227; Stuart, “Descartes’s Extended Substances,” 85–6; Justin Broackes, “Substance,” Proceedings of the Aristotelian Society 106, no. 1 (2006): 160. But Locke implies that he is explicating the standard usage of “substance.” In doing so he initiates a long tradition in which philosophers (mainly British empiricists) assume that “substance” usually has Locke’s sense, and criticize the concept on this basis. If the Leibnizian charge of incoherence takes half the blame for the endangered status of substances in present-day philosophy, the confusion initiated by Locke takes the other half. The Lockean side of the story has been thoroughly addressed by Justin Broackes. Broackes, “Substance”. IV Reading Descartes. In § 5, I focus in on the objection that the idea of things of can exist by themselves is incoherent. Partly because of that objection, there is a tendency for commentators to assume that this cannot be what Descartes means by “substance.” Since it is the purpose of this essay to defend the historical, as well as the theoretical importance of the idea of substances as things that can exist by themselves, it is worth pausing in this section to explain why, setting aside the charge of incoherence, Descartes’s independence definition really does seem to mean what I say it means. Commentators divide between a causal reading and a metaphysical reading of the independence definition. These say (or say something along the lines of) x is a substance if and only if: (i) x could exist without being caused by anything other than x (“causal independence”); and (ii) x could exist without being accompanied by anything other than x (“metaphysical independence”). The causal reading finds favour with Louis Loeb, From Descartes to Hume: Continental Metaphysics and the Development of Modern Philosophy (St, Ithaca, NY: Cornell University Press, 1981), 94–5; Markie, “Descartes’s Concept of Substance;” Jonathan Bennett, Learning from Six Philosophers: Descartes, Spinoza, Leibniz, Locke, Berkeley, Hume (Oxford: Clarendon Press, 2001), 134–5; and Stuart, “Descartes’s Extended Substances.” The metaphysical reading is advanced by Rodriguez-Pereyra, “Descartes’s Substance Dualism,” 69–90; and, with qualification, Marleen Rozemond, “Real Distinction, Separability, and Corporeal Substance in Descartes,” Midwest Studies in Philosophy, 35 (2011): 240–258. Anat Schetman also prefers the metaphysical reading, but eventually adopts a third alternative that deserves consideration elsewhere. See Anat Schechtman, “Substance and Independence in Descartes,” Philosophical Review 125, no. 2 (2016): 155–204. Of course, I favor the metaphysical reading. Though popular, the causal reading is not reputed to harmonize gracefully with Descartes’s statements about substances or with his wider thought. Proponents of the causal reading concede that it performs poorly in these respects. Markie, “Descartes’s Concept of Substance,” 68–72; Bennett, Learning, 134; Stuart, “Descartes’s Extended Substances,” 89. They adopt the causal reading anyway, in large part because of the alleged incoherence of the metaphysical reading. See, for example, Markie, “Descartes’s Concept of Substance,” 66; Stuart, “Descartes’s Extended Substances,” 89–92. For this reason, I limit myself to a brief exposition of three reasons why, setting aside the charge of incoherence, the metaphysical reading seems preferable. First, as I explained in § 3, the metaphysical reading can be regarded as a simple disambiguation of Aristotle’s Categories definition. Aristotle implies that things that are not substances are metaphysically dependent. To add that substances are metaphysically independent is very natural. As Kneale says, this might even be thought faithful to Aristotle. But to add that substances are causally independent seems very unnatural. If Descartes intended such a startling break with the Categories definition, we would expect him to prepare his readers for this, or at least to acknowledge the fact. Instead Descartes himself presents a version of the Categories definition in the Second Replies and behaves as though the independence definition comes to the same thing. This only makes sense if he has the metaphysical reading in mind. Secondly, Descartes’s language suggests that substances are capable of unaccompanied existence not uncaused existence. For example, Descartes says: In the case of created things, some are of such a nature that they cannot exist without other things, while some need only the ordinary concurrence of God in order to exist. We make this distinction by calling the latter “substances” and the former “qualities” or “attributes” of those substances. Descartes, Oeuvres, vol. 9b, 47. A created substance… is the kind of thing that can exist without any other created thing; and this is something that cannot be said about the modes of things, like shape or number. Ibid., vol. 3, 429. Descartes says that a substance is able to “exist without” (exister sans/esse absque) other things. This suggests unaccompanied existence. Cf. Stuart, “Descartes’s Extended Substances,” 90. (The first quotation is an addition to the French Principles, but these additions were made or approved by Descartes.) Witness Descartes’s use of the same language when he defines substances as “really distinct” when each could “exist without” (exister sans/existere absque) the other. Descartes, Oeuvres, vol. 7, 162; vol. 9a, 1245. Following this, Descartes appeals to the method of doubt to show that the mind is really distinct from the body. Ibid., vol. 7, 171–2. Descartes uses the method of doubt to show that we can imagine the mind existing unaccompanied by a body, not uncaused by a body, so this gives a clear indication of how he is using these phrases. Thirdly, the independence definition is meant to distinguish substances from properties (“attributes”, “qualities” and “modes”). On the metaphysical reading this means that properties cannot exist unaccompanied. It is easy to see why Descartes might think this. For this is just what the Categories definition implies about properties and it is, moreover, intuitively plausible. By contrast, it is unclear why anyone would think that substances, but not properties, can exist uncaused. This idea has no precedent in Aristotle, and it really does seem incoherent. For a substance must have some properties. And so if a substance exists uncaused, it seems that its properties must do so too (those it has to start out with, at any rate). Worse still, if Descartes distinguishes properties from substances by their causal dependence, he must distinguish created properties from created substances by their causal dependence on things other than God. To do so has the curious consequence that God is incapable of creating properties. And because substances need properties, the result seems to be a total collapse of divine potency. For example, if God wants to create a rock, he cannot do so. For a rock must have a size, texture and so on, and these, as Descartes observes, are properties (“modes”), not substances. Ibid., vol. 8a, 25. Markie responds to this last concern by adding a clause to the independence definition that explicitly excludes properties. The problem with Markie’s proposal is that the only distinction the resulting definition gives to substances is that they are not properties. The part about causal independence has become purely ornamental. See Markie, “Descartes’s Concept of Substance,” 69. For three reasons, then, the metaphysical reading seems greatly preferable to the causal reading. I know of only one argument for the causal reading, aside from the claim that the metaphysical reading is incoherent. According to this argument, the causal reading is encouraged by Descartes’s claim that created things cannot be substances in an unqualified sense because they depend on God. For the dependence of created things on God is causal. See Loeb, From Descartes to Hume, 94–5; Stuart, “Descartes’s Extended Substances,” 181–2; Broackes, “Substance,” 137–8; Vere Chappel, “Descartes on Substance,” in A Companion to Descartes, eds. Janet Broughton and John Carriero (Oxford: Blackwell, 2008), 261. Gonzalo Rodriguez-Pereyra points out that this argument falsely presupposes that the dependence of created things on God is, for Descartes, only causal and not also metaphysical. Rodriguez-Pereyra, “Descartes’s Substance Dualism,” 84–5; cf. also Schechtman, “Substance and Independence,” 182–3. I regard Rodriguez-Pereyra’s refutation as decisive and defer to his discussion for the detail. I conclude that, the charge of incoherence aside, the metaphysical reading of the independence definition is greatly preferable to the causal reading. If the metaphysical reading turns out to be incoherent, we might have to adopt the causal reading anyway. But we should favor the metaphysical reading if at all possible. I now turn to a question about how we should formulate the metaphysical reading. There are two ways of formulating the metaphysical reading. I name these the “weak” formulation and the “strong” formulation. Similar distinctions are drawn by Corkum, “Aristotle on Ontological Dependence,” 71–2; Rozemond, “Real Distinction,” 225; Koslicki, Form, Matter, Substance, 142–4. These say that x is a substance if and only if: (i) there is no y such that y is something other than x, and the existence of x necessitates the existence of y (“weak metaphysical independence”); and (ii) the existence of x does not necessitate that there exists some y such that y is something other than x (“strong metaphysical independence”). Using “Ex” for “x exists,” rough formalizations might say that x is weakly metaphysically independent if and only if ¬Ǝy((y ≠ x) ∧ □(Ex → Ey)); and x is strongly metaphysically independent if and only if ¬□(Ex → Ǝy((y ≠ x) ∧ Ey)). However, “≠” is not strictly appropriate. For as I mentioned in the introduction, and as I explain in § 6, I deny that the identity relation plays a role in the independence definition. In plain English, the weak formulation says that a substance can exist unaccompanied by any other thing in particular. The strong formulation says that a substance can exist unaccompanied by anything else whatsoever: a substance can exist in total isolation. Rodriguez-Pereyra does not discuss the choice between the strong and weak metaphysical readings. But as I explain in § 5, he implicitly adopts the weak formulation in responding to the charge of incoherence. And though they do not adopt the metaphysical reading themselves, Bennett and Schechtman recommend the same strategy for its proponents. I know no enquiry dedicated to Descartes’s concept of substance that adopts the strong metaphysical reading. Marleen Rozemond comes closest in being open to a strong metaphysical reading with certain qualifications. See Rozemond, “Real Distinction”. Curiously, works outside Descartes scholarship often appear to take the strong metaphysical reading for granted. I favor the strong formulation. I grant that the weak metaphysical reading is an improvement on the causal reading. It is coherent and compatible with Descartes’s wider thought. Nonetheless, the strong metaphysical reading is a more natural interpretation of Descartes’s language. Three examples will illustrate this. First, in the Principles, Descartes defines created substances as “things that need only the concurrence of God in order to exist.” Descartes, Oeuvres, vol. 8a, 24–5. This suggests the strong formulation. For claims of the form “a needs only b” ordinarily express the idea that a needs no other thing whatsoever, besides b, not that a needs no other thing in particular, besides b. To see this, consider Martine Nida-Rümelin’s (2006; 2010) theory of mind. Martine Nida-Rümelin, “Dualist Emergentism,” in Contemporary Debates in Philosophy of Mind, eds. Brian McLaughlin and Jonathan Cohen (Oxford: Blackwell, 2006); “An Argument from Transtemporal Identity for Subject Body Dualism,” in The Waning of Materialism, eds. George Bealer and Robert Koons (Oxford: Oxford University Press, 2010). Nida-Rümelin says that minds are immaterial things that can exist without their actual bodies. But unlike Descartes, she says that minds cannot exist disembodied: a mind needs some body or other to exist. Now ask yourself, would we describe Nida-Rümelin’s position as one on which minds “need only the concurrence of God in order to exist?” (Nida-Rümelin’s theory plus theism, that is.) I suggest that we would not. To do so would be misleading given that Nida-Rümelin’s minds also need bodies. For this very reason, Nida-Rümelin declines to call minds, on her view, ‘substances’. See Nida-Rümelin, “An Argument from Transtemporal Identity,” 191. Secondly, the French Principles (AT IX B, 125) adds that substances can “exist without other things.” Descartes, Oeuvres, vol. 9b, 125. And the letter to Hyperaspistes says that a created substance “can exist without any other created thing.” Ibid., vol. 3, 429. These passages also encourage the strong and discourage the weak metaphysical reading. To read these as passages expressions of the weak reading we must construe Descartes’s language in in an unusual way, so that for x to be able to exist “without any other created thing” it is sufficient that there is no other created thing in particular whose existence x necessitates. This would be like saying that Agamemnon could have besieged Troy “without any other assailant” on the grounds that no other assailant in particular was indispensable, or that humans can live on candy floss “without any other food” because no other food in particular is essential. Such claims outrage ordinary linguistic conventions. Thirdly, still more difficult for the weak formulation is the following passage from the Fourth replies, where Descartes explains the synonymy of “complete thing” and “substance”: By a “complete thing” I simply mean a substance… I am aware that certain substances are commonly called “incomplete”. But if the reason for calling them incomplete is that they are unable to exist on their own, then I confess I find it self-contradictory that they should be substances. Ibid., vol. 7, 222. Descartes says that to call things “substances” that are unable “to exist on their own” (per se solae esse) entails a contradiction. Slightly later, Descartes defines a substance as a thing that can “exist by itself” (per se existere). Ibid., vol. 7, 226. Clearly these claims prohibit the weak metaphysical formulation. For the fact that x can exist without any other thing in particular certainly does not entail that x can exist “on its own” or “by itself.” Indeed, those are just the sort of phrases I have used to specify the strong metaphysical independence definition that I favor. To accommodate Descartes’s statements in the Fourth Replies, proponents of the weak formulation must take issue with the CSMK translation. Perhaps this is possible. For the Latin “per se” has various uses. But the CSMK translation is very natural. And it is hard to imagine any translation that would naturally suggest the weak formulation. (Several contextual cues also support the CSMK, but I spare the reader these here.) Descartes’s statements, then, consistently suggest that he has the strong metaphysical independence definition in mind. We might have to revise this judgement if the idea of something that can exist by itself turns out to be incoherent. But setting aside the charge of incoherence, the strong metaphysical independence definition is clearly the best reading of Descartes. I now turn to the charge of incoherence itself. V The charge of incoherence. The strong metaphysical independence definition says something very simple: a substance is something that could exist by itself. Substancehood is the minimum threshold for being; it is the first entry in God’s catalogue of weights and measures. I have suggested that this conception of substance has an important place in the history of philosophy, due to Aristotle and Descartes, and that it promises to make up for an important shortcoming in the conceptual resources of present-day philosophy, by explaining what makes object-like things object-like. The literature on this subject paints a very different picture. For philosophers tend to regard the idea of substances as things that can exist by themselves as incoherent, and this has encouraged the view that the idea of substances, so characterized, will be useless to present-day philosophers and cannot charitably be attributed to Descartes. This section explains the charge of incoherence and illustrates its influence on the recent literature. Philosophers judge the idea of substances as things that could exist by themselves incoherent, more or less invariably, because they suppose that to exist by itself, a substance would have to exist without any property with which it is not identical, and they judge that this is impossible. I call this the “property-dependence objection”. For it says, in effect, that nothing can satisfy the strong metaphysical independence definition because everything depends metaphysically on its own properties. The property-dependence objection is sometimes accompanied by a parallel “part-dependence objection” concerning composite substances. The property-dependence objection is more important than the part-dependence objection. For everything has properties whereas it may be that only some things have parts. Except where I specify otherwise, however, my comments on the property-dependence objection may be taken to apply to the part-dependence objection as well. An early example of the property-dependence objection appears in Leibniz’s “Critical Thoughts” on Descartes’s Principles. His comment on the independence definition reads: I do not know whether the definition of substance as that which needs for its existence only the concurrence of God fits any created substance known to us... For not only do we need other substances; we need our own accidents even much more. Gottfried Leibniz, “Critical Thoughts,” 389. Leibniz does not explain the contention that “we need other substances.” But in claiming that “we need our own accidents” he expresses the idea that nothing could satisfy the independence definition because everything depends on its own properties. The property-dependence objection has been extremely influential in the recent literature. It has encouraged some commentators to favor the causal reading of the independence definition over the strong metaphysical reading. It has encouraged others to favor the weak metaphysical reading (or at least it seems to have done so). And it has convinced these and other theorists that the strong metaphysical independence definition will be useless to present-day philosophers. I give some examples of each group. Peter Markie and Matthew Stuart both present the property-dependence objection as the main reason to favor the causal reading of Descartes’s independence definition. After introducing the definition Markie says (alluding to some of Descartes’s examples of substances): Does not the rock also “need” and “depend on” its material parts?... Do not God, a rock, Descartes’s mind, and Descartes’s body all “depend” on their essential attributes since they could not exist without them, and on qualities in general since they could not exist without some? Markie, “Descartes’s Concept of Substance,” 66. Markie expresses here both the property-dependence objection and the part-dependence objection. He then proposes that to “enable us to move forward” we should “assume that Descartes is concerned, at least partially, with a form of causal independence.” Likewise Stuart, after acknowledging the merits of the metaphysical reading, describes following “fatal shortcoming”: Because the existence of any candidate for created substance implies the existence of some properties belonging to it, no candidate for created substance can meet the requirement of being metaphysically independent. Stuart, “Descartes’s Extended Substances,” 89. I have substituted ‘created’ and ‘metaphysical’ for Stuart’s ‘secondary’ and ‘logical’ to match my terminology Stuart concludes that we should adopt the causal reading. The literature on Descartes tends not to distinguish the strong and weak formulations of the metaphysical reading. For this reason, proponents of the weak formulation do not give any explicit reason for preferring it. Nonetheless it is clear that if proponents of the weak formulation were to give such a reason, they would point (perhaps inter alia) to the property-dependence objection. This is clear because proponents of the weak metaphysical reading standardly take the view that Descartes should respond to the property-dependence objection by arguing that a substance need only be capable of existing without any given property, not without anything else whatsoever. For example, Rodriguez-Pereyra says: That substances need their modes to exist is not a problem… For a substance does not need specific modes to exist, and so there is no single mode without which the substance cannot exist. Rodriquez-Pereyra, “Descartes’s Substance Dualism,” 81. The same response is recommended by Bennett, “Learning from Six Philosophers,” 134; and Schechtman, “Substance and Independence,” 183. This response implies that the weak metaphysical reading can withstand the property-dependence objection in a manner in which the strong metaphysical reading cannot. Incidentally, I do not find this response satisfactory. For it leaves unaddressed necessary properties like number or duration and the necessary parts of composite substances, both of which Descartes seems to posit. See Descartes, Oeuvres, vol. 8a, 26; vol. 5, 166. But I do not develop this objection here. The property-dependence objection, then, plays an important part in accounting for the widespread preference for the causal reading and probably also the weak metaphysical reading of Descartes’s independence definition, over the strong metaphysical reading. The property-dependence objection has also convinced these and other theorists that the strong metaphysical independence definition will be useless to present-day philosophers. An influential example that is not primarily concerned with Descartes scholarship is Joshua Hoffman and Gary Rosenkrantz’s 1997 book on the nature and existence of substances. Hoffman and Rosenkrantz set out to revive the idea of substance, taking the independence definition as their starting point. But in a passage that resembles that quoted from Markie above, they say: Suppose there is a substance, for instance, a rock (call this rock r). It can be argued that if r exists, then there must exist many other beings as well. These other beings include some substances, for example, entities which are parts of r, and some nonsubstances, for instance… properties of r. Hoffman and Rosenkrantz, Substance, 144. Hoffman and Rosenkrantz reject the independence definition in its original form but go onto defend a novel account of substance that is inspired by it. A more recent example is Richard Swinburne who says: Descartes [says] that a substance is “a thing which exists in such a way that it needs no other thing in order to exist.”… The definition is unsatisfactory, because a substance could not exist without having properties—a substance can only exist if it has properties that make it that substance. Swinburne, “Cartesian Substance Dualism,” 36. Other examples may be found in works by Lowe, Benjamin Schneider and Patrick Toner. Lowe, The Possibility of Metaphysics, 141; Schneider, “A Certain Kind of Trinity,” 397–400; Toner, “Independence Accounts of Substance,” 38. As these examples demonstrate, the property-dependence objection is very influential. I do not mean to suggest that there exist no other objections to the strong metaphysical independence definition. I consider several in the final section. I do find that the property-dependence objection appears with the greatest frequency and is most commonly treated as decisive. And yet, I believe that it is based on a misunderstanding, as I now explain. VI Responding to the charge of incoherence. I have already sketched my reply to the property-dependence objection in the introduction. The objection assumes that to exist by itself a substance would have to exist without anything nonidentical to it. I concede that if this assumption were correct, the objection would be fatal. But the assumption is false. It is very unnatural to interpret the claim that something can exist by itself as the claim that it can exist without anything nonidentical to it. This section sets out my reply in greater detail. From here on I assume the strong metaphysical formulation of the independence definition, referring to this simply as “the independence definition.” Something close to the response presented here (or part thereof) has been suggested by Terrance Irwin and Michael Gorman. But neither is defending the strong metaphysical independence definition that I am defending. I discuss one respect in which this compels me to depart from Gorman in § 7. See Terrance Irwin, Aristotle's First Principles (Oxford: Oxford University Press, 1988), 83, fn. 23; Michael Gorman, “Independence and Substance,” International Philosophical Quarterly 46 no. 2 (2006): 147–159. I begin by making two assumptions concerning the domain over which the independence definition quantifies. The definition says that a substance is something that could exist by itself. But as we have seen, Descartes adds a qualification. A created substance, he says, should be defined as something that could exist by itself, leaving aside God. Descartes, Oeuvres, vol. 8a, 25. Equivalently, a created substance is something that could exist by itself when we confine our attention to the domain of nondivine things. This qualification is sensible. For if God exists, he does so necessarily, and so nothing will be able to exist without God. But there will remain an important distinction between things that can and cannot exist without anything else apart from God. For this reason, I propose to exclude God, along with any other necessary beings that might exist, from the domain of the independence definition. I also propose to exclude abstracta from the domain of the independence definition (abstracta in the primary sense of non-spatiotemporal acausal things). Many theorists hold that abstracta, if they exist at all, do so necessarily. So this restriction might follow from the preceding one. However, some theorists view some abstracta as contingent. I exclude these too, in order to focus on what I see as the crucial difference between object-like concreta such as vases and un-object-like concreta such as scratches and dents, or shapes and colors construed as concrete properties. I consider objections to the exclusion of necessary beings and abstracta from the domain of the independence definition in § 7. For now, I suppose that these restrictions are acceptable and proceed with my defense of the independence definition. The definition says that a substance can exist without anything else. Its opponents assume that this means that a substance can exist without anything nonidentical to it. This assumption is made explicit by Hoffman and Rosenkrantz and Lowe. Hoffman and Rosenkrantz, Substance, 2; Lowe, The Possibility of Metaphysics, 138. It is implicit in all works that takes the property-dependence objection seriously. And it usually goes unquestioned. This is an important oversight. For phrases like “x without anything else” and “x by itself” do not usually mean “x without anything nonidentical to it.” To get a sense of how odd it is to interpret such phrases in terms of the identity relation, imagine that someone asks you what is on top of Nelson’s Column. You will probably reply, “there is a statue of Horatio Nelson, the hero of Trafalgar.” If your interrogator asks whether there is anything else, you will say that there is not: the statue is there by itself. There is a sense in which you are wrong. For the statue’s left leg is also on top of the column; and its right leg; and its shape, and its color, supposing these are concrete properties. None of these things is identical to the statue. Still, your claim is only incorrect where we misconstrue it in terms of the identity relation. In the ordinary sense of the phrase, it is no objection to the claim that Nelson’s statue is the only thing on top of Nelson’s column that there are also things nonidentical to the statue up there, such as its legs, shape and color. Likewise, when Democritus says that there is nothing in the world but atoms, it is no objection if there also exist left atom halves and right atom halves and atom shapes and sizes. And when the Jesus’s disciples see his burial cloth “wrapped together in a place by itself” at John 20.7 they do not see a burial cloth denuded of its properties and parts. Ordinarily, when we talk about some concrete thing, x, being present “on its own,” “by itself” or “without anything else” we do not mean that x is present without anything that is nonidentical to x. It might seem surprising, in light of this, that philosophers consistently interpret the independence definition in this way. This is less surprising when one observes that this is not the only case where philosophers have shown a tendency to press the identity relation into services for which it is unfit. One of the great lessons of the last half-century of philosophy is that numerical identity is frequently too strong a relation to express important metaphysical theses. Consider the example of physicalism. Physicalism is the claim that reality is exhausted by physical things. It was once thought that physicalism could be more precisely defined as the claim that everything is identical to some physical thing. This assumption has since been abandoned because, intuitively, “higher level” things might be no addition to physical things, without being numerically identical to them. Cf. John Haugeland, “Weak Supervenience,” American Philosophical Quarterly 19, no. 1 (1982): 93–103. Hence, philosophers have long preferred to define physicalism by weaker relations. The property-dependence objection presupposes that something that exists “by itself” must exist without anything nonidentical to it. In doing so, it makes an intuitively coherent concept appear incoherent, in a way that conflicts with the ordinary meaning of the terms used to express it. This looks like another example of the well documented tendency for philosophers to interpose the identity relation where it is not wanted. It might be supposed that for this response to the property-dependence objection to stand, it is necessary to specify how we should interpret the independence definition, if not in terms of identity. But I suggest that this is not so. For if its parts and properties do not prevent us from coherently describing Nelson’s statue as “by itself” on top on Nelson’s column then whatever the correct analysis of “by itself” in that description might be, it seems inevitable that the same interpretation will be available in the case of the independence definition. The property-dependence objection presupposes that terms like “by itself” have a meaning in the independence definition that they cannot possibly have in the example of Nelson’s statue or Jesus’s burial cloth. This presupposition is prima facie implausible. So the onus is on proponents of the objection to motivate it, not on us to refute it. The same point can be made by appeal to cases in which philosophers are heavily invested. For example, I have observed that the consensus in the literature on physicalism is that we can make sense of the claim that reality is exhausted by physical things, even though not everything is identical with some physical thing. If so, we should be able to make sense of the claim that reality is exhausted by an object-like thing such as a vase, even though not everything is identical to the vase. Again, the onus is on opponents of the independence definition to explain why we should view the two cases differently. It is not necessary, therefore, to commit to a specific replacement for the identity relation in our interpretation of the definition, before we reject the property-dependence objection as confused. But we might reinforce the case against the property-dependence objection by showing that a promising replacement exists. It is to this that I now turn. To this end, it is useful to imagine how we would answer someone who claimed that we spoke falsely when we said that Nelson’s statue is “by itself” on his column, given the presence of its shape, color and limbs. We would, I suggest, urge that these things are “no addition” to the statue, they are “included in” or “nothing over and above” it. We would say the same if someone sold us the statue and accused us of stealing its parts and properties, and there is not a jury in the world that would convict us. It seems, therefore, that we should replace the identity relation with the “nothing over and above” relation. Though common in philosophy, the phrase “nothing over and above” tends to be seen as undesirably vague. So we had better try to identify precisely what makes the parts and concrete properties of an object count as “nothing over and above” it. When philosophers began to distrust the adequacy of the identity definition of physicalism, many proposed that the supervenience relation would serve better. To say that one thing supervenes on a second thing is to say that the facts about the first thing determine the facts about the second. So, following their lead, we might hypothesize that parts and concrete properties are “nothing over and above” their bearers because they supervene upon them. As Armstrong once proposed, “what supervenes is no addition of being” and so “the terminology of ‘nothing over and above’ seems appropriate to the supervenient.” David Armstrong, A World of States of Affairs (Cambridge: Cambridge University Press, 1997), 12. But supervenience is not a promising proposal in the present context. For it is now generally agreed that one thing might supervene on another despite being “something over and above” it. See Terence Horgan, “From Supervenience to Superdupervenience: Meeting the Demands of a Materialist World,” Mind 102, no. 408 (1993): 555–586; Jessica Wilson, “Supervenience-based Formulations of Physicalism,” Nous 39, no. 3 (2005): 426–459. This is because supervenience is only a matter of covariance. One can coherently posit that two items covary in the requisite manner no matter how distinct they are. It follows that supervenient entities are not the ontological “free lunch” that Armstrong hoped they were. For this reason, present-day philosophers tend to define physicalism not in terms of supervenience, but in terms of grounding. To say that one thing is grounded in a second thing is to say that the first exists or obtains “in virtue of” the second, in a manner that is usually supposed to be explanatory, asymmetrical and primitive. See Kit Fine, “A Guide to Ground,” in Metaphysical Grounding, eds. Fabrice Correia and Benjamin Schneider (Cambridge: Cambridge University Press, 2012); Michael Raven, “Ground,” Philosophy Compass 10, no. 5 (2015): 322–333. So, a second possibility is that parts and concrete properties are “nothing over and above” their bearers because they are grounded in them. The problem with this proposal is that philosophers disagree about whether parts and properties are grounded in their bearers or vice versa. Neither can we evade this difficulty by hypothesizing that the converse of the grounding relation is also sufficient for being “nothing over and above.” For objects are usually thought to be grounded in things that greatly exceed them, ultimately in the entire cosmos. Only a tiny portion of this vast ground will intuitively be “nothing over and above” the object. The task of specifying that portion sends us back to the drawing board. Identity, supervenience and grounding all fail to capture the sense in which parts and concrete properties are “nothing over and above” in their bearers. There is, however, a more promising proposal. Parts and concrete properties are “nothing over and above” their bearers, I suggest, because they are subsumed by them. Here, my reply to the property-dependence objection aligns with Michael Gorman’s. For by “subsumption,” I mean the relation that I take Gorman to express when he suggests that a substance may remain “self-sufficient” so long as it only depends on things that are “so to speak, within the thing in question.” Michael Gorman, “Independence and Substance,” International Philosophical Quarterly 46, no. 2, (2006): 151. As I observed above, my account of substances also departs from Gorman’s in important respects. To put this more precisely, it is useful to exploit an idea from § 2. I mentioned there Keith Campbell’s characterization of abstraction as a process of “selective ignoring” or “systematic setting aside.” Campbell, Abstract Particulars, 2–3. To abstract, according to Campbell, is to bring something to mind, and then set aside one or more elements of it. A plausible definition says that one thing subsumes a second if and only if it is possible to start with the first and arrive at the second by a process of abstraction as characterized by Campbell. An advantage of this definition is that it explains why subsumption is sufficient for being “nothing over and above.” If something can be arrived at by a process of setting aside it must have already been there waiting to be discriminated. The parts and properties of a vase are no addition to it because one can arrive at them by starting with the vase and subtracting. I do not claim that subsumption is the only way of being “nothing over and above.” Trivially, there is at least one other: identity. And there might be a third. For there seem to be cases where we start with one thing and arrive at a numerically distinct thing without setting anything aside but without making any addition either. For example, it is possible to start with Nelson’s statue and arrive at the quantity of sandstone that constitutes it, simply by noticing that there is something here whose identity conditions, at least by convention, differ from those of the statue in certain respects. The statue does not subsume the sandstone. Rather, the two coincide. And the move from statue to sandstone is not one of abstraction, in Campbell’s sense. Instead, it seems to be a matter of reconceptualization: we stop thinking of this thing under the concept “statue” and think of it under the concept “quantity of sandstone” instead. I say that there might be a third way of being “nothing over and above” because it is controversial whether things that exactly coincide, like the statue and the sandstone, can really be numerically distinct. Perhaps the distinction is merely conventional. But we need not trouble ourselves about this at the moment. It suffices for present purposes that if x reachable by abstraction from y then x is subsumed by and therefore “nothing over and above” y. Having so far clarified the “nothing over and above” relation, we can state our interpretation of the independence definition of substance as follows: Substance = (def.) restricting our attention to the domain of contingent concreta, x is a substance if and only if, the existence of x does not necessitate that there exists some y such that y is something over and above x. It is plausible that the vase is a substance, according to this definition, because: (i) it is plausible that the vase could exist without anything else, apart from its own parts and concrete properties; and (ii) the parts and properties of the vase are subsumed by and hence nothing over and above it. It is implausible that the shape of the vase is a substance, according to this definition, because: (i) it is implausible that the shape could exist without the rest of the vase (or at any rate, something of which it is the shape); and (ii) the rest of the vase (or that something) is not subsumed by or in some other way “nothing over and above” the shape. You can bring the shape to mind by starting with the whole vase and setting aside elements of it. But you cannot bring the vase to mind by starting with the shape and setting aside elements of it—on the contrary, you have to add to them. That, then, is my preferred analysis of the independence definition. The fidelity of this analysis to the ordinary meaning of terms such as “by itself” makes it a significant improvement on interpretations in terms of identity. At the same time, this analysis makes it clear why the definition is immune to the property-dependence objection. Still, the proposed analysis only reinforces my reply to the property-dependence objection and is not essential to it. For the ordinary meaning of terms like “by itself” in similar contexts makes it a desideratum of any interpretation of the independence definition that it should not require that a substance can exist without its properties or parts. I have suggested that our tendency to miss this is due to our overreliance on the identity relation. If so, Leibniz is, perhaps, to blame. For it was Leibniz who established the centrality of numerical identity to philosophical theorizing when he introduced his laws of identity. Gottfried Leibniz, “Discourse on Metaphysics,” in Gottfried Leibniz, Philosophical Papers and Letters, 2nd ed., ed. and trans. Leroy Loemker (Alphen aan den Rijn: Kluwer, 1989), 308. It is therefore unsurprising that Leibniz also appears to be first proponent of the property-dependence objection. More recently, philosophers have learned to be cautious in employing the identity relation. When we transfer this lesson to the independence definition, it becomes clear that the property-dependence objection is a chimera. VII Further objections. There remains the business of responding to other objections. I begin with the worry that in excluding necessary beings and abstracta from the domain of the independence definition, I forfeit my claim to have defended idea of substance that I find in Descartes. Following this, I discuss the concern that my characterization of subsumption tacitly relies on the idea of something “by itself” that it was introduced to illuminate. Finally, I address ten objections to the extensional adequacy of the independence definition as I interpret it. Objection 1. I have suggested that we exclude necessary beings and abstracta from the domain over which the independence definition quantifies. It might be objected that this is incompatible with my claim to defend the idea of substance as I find it in Descartes. For although Descartes excludes God from the domain of the independence definition, as it applies to created substances, Descartes says nothing to indicate that this restriction extends to necessary beings other than God or to abstracta. In response I note first that the exclusion of abstracta is only needed to provide for the special scenario in which there exist things that are abstract, contingent, and whose existence is necessitated by concreta (for example, on some views singleton sets satisfy these conditions). For necessary abstracta will be excluded with other necessary beings, and abstracta whose existence is not necessitated by that of concreta will pose no threat to the existential independence of concreta. The next part of the response depends on whether Descartes thinks there are necessary beings other than God, or contingent abstracta whose existence is necessitated by concrete things. This is hard to judge. The best candidates in his system are the “eternal truths.” Cf. Martha Bolton, “Universals, Essences, and Abstract Entities,” in The Cambridge History of Seventeenth Century Philosophy, eds. Daniel Garber and Michael Ayers (Cambridge: Cambridge University Press, 2008), vol. 1, 178–211. But Descartes’s understanding of the eternal truths is notoriously difficult to interpret. For this reason, it is prudent to make our response conditional. If Descartes does posit such things, it is clear that he ought to exclude them from the domain of the independence definition. For otherwise minds and bodies will not qualify as substances. We should therefore take their exclusion as implicit. If Descartes does not posit such entities, then the independence definition will be defensible in the context of his system already. The exclusion of necessary beings and abstracta may then be viewed as a measure that renders Descartes’s definition serviceable beyond the confines of his system. Either way, Descartes’s definition will be defensible in the context of his system. In the second case, the definition requires further specification to accommodate the existence of entities that he does not posit. But the adjusted definition will have a good claim to retain the spirit of Descartes’s under conditions he does not consider. I respond to a further concern about the exclusion of necessary beings when I discuss objections to the extensional adequacy of the independence definition a few pages hence. But first, I consider an objection to my characterization of subsumption. Objection 2. I have suggested that one thing subsumes a second if and only if it is possible to start with the first and arrive at the second by a process of abstraction as characterized by Campbell. An instructive objection observes that “start with the first,” here, must mean “start with the first by itself.” I grateful to Olley Pearson for this objection. For otherwise, the process of abstraction could result in almost anything. For any x not included in the vase, if you start with the vase and x, and set aside the vase, you will be left with x. It follows that to understand the definition of subsumption you must already grasp the idea of something “by itself”. This helps clarify the dialectical situation. Someone with no prior grasp of the idea of something “by itself” could, on that basis, disavow any understanding of the definition of subsumption. The definition of subsumption cannot be used to compel such an opponent to acknowledge the coherence of the independence definition of substance. For two reasons, I do not see this as a serious problem. First, most of us do have an antecedent grasp of the idea of something “by itself.” This is why we are able to make sense of the claim that Nelson’s statue is “by itself” on his column and why we are not bewildered by Campbell’s description of abstraction in the first place. Secondly, as I have emphasized, the proposed analysis of the independence definition is not essential to the case against the property-dependence objection. Neither, therefore, is the definition of subsumption which I have used to elucidate that analysis. Nonetheless, I grant that my response to the property-dependence objection would be improved by a definition of subsumption that did not rely on the idea of something “by itself.” I am not sure that such a definition is possible, but I would welcome suggestions. In the meantime I move on to objections to the extensional adequacy of the independence definition. These come in two varieties: charges of excessive breadth and charges of excessive narrowness. I begin with three charges of excessive breadth. Objection 3. It might be objected that the independence definition, as I interpret it, does nothing to guarantee that substances are unitary beings, rather than mere collections. If, for example, a handful of dust could exist by itself, then it seems that the handful of dust is, according to the independence definition, a substance. This conflicts with the idea that substances should be unities. Koslicki presses this charge against Gorman. Koslicki, Form, Matter, Substance, 187. I do not see this as a problem because I do not accept that unity is necessary to make something object-like rather than property-like. A handful of dust, no less than a vase, is a “thick,” “dense” property-bearer. It is not a “thin,” “attenuated” being in the way that the shape or color of the vase are, or its scratches and dents. To that extent, it is right that a handful of dust should be classed as a substance. For the notion of substance has been introduced to explain what makes object-like things object-like. I grant that if I had started with the idea of substances as “the fundamental entities,” and were trying to fill out this notion, then the disunity of a handful of dust might make it a bad candidate for substancehood. But as I noted in § 2, that is not my aim. If there is any lingering concern about the unity of substances, I suggest that we have recourse to our independent mereological commitments. If we favor strict composition laws, then we may deny that a handful of dust is a substance on the basis that, strictly speaking, no such thing exists. For we will say that a handful of dust is not a genuine fusion. If we favor liberal composition laws, we might say that a handful of dust is a genuine fusion. But if so, we cannot deny it the status of a substance on the basis that it is a mere collection. Objection 4. A similar objection observes that the independence definition as I interpret it does nothing to guarantee that substances are durable entities of the sort called “continuants” as opposed to process-like or event-like entities of the sort called “occurrents.” If, for example, a splash could exist by itself, then it seems that the splash is, according to the definition, a substance. This conflicts with the idea that substances should be continuants. I do not see this as a problem because I do not accept that being a continuant is necessary to make something object-like either. Suppose that by a “splash” we mean, not just the movement that a liquid performs, but the rich dynamic entity that includes the liquid and its properties during the period when the movement occurs. A splash in this sense is object-like in a way that makes it akin to a vase. By contrast, the movement taken by itself is un-object-like in a way that makes it akin to the properties of the vase. It is plausible that this is because the splash could exist by itself, whereas the movement could not. Objection 5. A third charge of excessive breadth observes that, on certain views, the independence definition will fail to exclude the very items we usually regard as properties. For example, in his Treatise Hume acknowledges the coherence of the idea of substances as things that could exist by themselves, but claims that this definition is true of everything. David Hume, A Treatise of Human Nature, eds. David Norton and Mary Norton (Oxford: Oxford University Press, 2007), 1.5.5.5. Likewise, Williams and Campbell suggest that abstracta in the second sense of “abstract,” such as colors, textures, or timbres, might exist in total isolation. Williams, “Elements of Being II,”, 179; Campbell, Abstract Particulars, 3, 21. And Jonathan Schaffer argues it is only a contingent truth that properties “cluster.” Jonathan Schaffer, "The Problem of Free Mass: Must Properties Cluster?" Philosophy and Phenomenological Research 66, no. 1 (2003): 125–138. I concede that the inability of a property to exist by itself is a basic assumption of the account of what makes object-like things object-like proffered here. This assumption seems reasonable because the idea of an isolated property is, at face value, incoherent. As Broackes says: The trouble is that it is hard to make sense of the idea of a case of one property and no other. Surely nothing can have a shape and no size—nothing can be circular and nothing else (e.g. not of a particular diameter); nothing can be red and nothing else (not of any extent or shape). Broackes, “Substance,” 156. If this basic assumption is false, then the account of what makes object-like things object-like proffered here fails. I think the assumption is correct, but I grant that it deserves further scrutiny elsewhere. I turn finally to seven charges of excessive narrowness. Objection 6. It might be objected that object-like things will fail to satisfy the independence definition because, although they subsume their parts and concrete properties, they do not subsume the properties of their parts and properties. For example, the red spot on Jupiter has the property of being red all over. But Jupiter lacks the property of being red all over. And it might be supposed that Jupiter cannot subsume a property that it does not possess. I am grateful to Donnchadh O'Conaill for this objection. And yet, it is plausible that Jupiter could not exist without there being properties of its parts and properties that it does not itself share. If so Jupiter will fail to qualify as a substance. In fact, I suggest that such examples merely show that for x to subsume the property expressed by a predicate F it is not necessary that x satisfy F. This can be confirmed by the test of abstraction. Start with Jupiter and set aside everything but the red spot. Now set aside everything about the red spot, apart from the property of being red all over. By successive subtractions we have advanced from Jupiter to the property of being red all over. It follows that this property is subsumed by Jupiter. Objection 7. A similar objection concerns the ingredients in a mixture. A cake, it might be urged, could not exist without the eggs that were used to make it. But it is not clear that the cake subsumes the eggs. Indeed, the test of abstraction suggests that it does not. For we cannot start with the cake and arrive by abstraction at anything recognizable as an egg. Here, it is useful to distinguish between ingredients that are destroyed in and ingredients that survive the process of mixing. The raisons, for example, are straightforwardly subsumed by the cake, since they survive the mixing. If the eggs are not subsumed by the cake this is, I suggest, because they have been destroyed. If so, they do not threaten the existential independence of the cake. By contrast, whatever has survived of the eggs (e.g. their atoms) will be subsumed by the cake no less than the raisons. Objection 8. I come now to a further objection concerning the exclusion of necessary beings from the domain over which the independence definition quantifies. The objection points out that because of this restriction, the definition will only be satisfied by object-like things such the vase if these things exist contingently. So the claim that object-like things are substances, as characterized by the independence definition, presupposes the falsehood of necessitism—the thesis that all things exist necessarily. It might be possible to meet the threat of necessitism by replacing the exclusion of necessary beings with the specification that that, in the sense of “necessitate” relevant to the independence definition, it is not sufficient for x to “necessitate” the existence of y that at every possible world where x exists, y also exists. Rather, “necessitate” must mean something stronger, to the effect that the existence of y follows from the nature of x specifically. While I am sympathetic to this response, it requires development elsewhere. This response might be roughly equivalent to Rozemond’s suggestion that for Descartes, the separability of substances does not require that one can exist without the other where this is prohibited by something other than the “union” between the two. See Rozemond, “Real Distinction.” A simpler response points out that proponents of necessitism standardly say that while everything necessarily exists, it is not true that everything necessarily concretely exists. See, for example, Timothy Williamson, Modal Logic as Metaphysics, (Oxford: Oxford University Press, 2013). Rather, those concreta that we ordinarily think of as not existing do exist, but not concretely. So necessitism can also be accommodated by excluding things that necessarily concretely exist, rather than things that necessarily exist, from the domain of the independence definition. Objection 9. A fourth charge of excessive narrowness says that the independence definition will not extend to object-like things that comprise concrete universals. For it is not clear that things will subsume their concrete universal properties as they do their concrete particular properties, given that concrete universals are meant to exist in multiple locations. For example, the color of Nelson’s statue, construed as a concrete universal, is also present at Edinburgh Castle and at Holyrood Palace. And it is hard to see how something present at Edinburgh Castle and Holyrood Palace could be subsumed by a statue in Trafalgar Square. It might be thought that this objection can be evaded by excluding universals from the domain of the independence definition, as Gorman proposes. Gorman, “Independence and Substance,” 130. But although this solution may work for Gorman, since he is not defending the strong metaphysical independence definition that I am defending, it is not promising in the present context. This can be seen by considering such maimed entities as the vase minus its shape or the vase minus its size. The vase minus its shape is not an object-like thing in the way the vase is. Neither, I assume, can a vase minus its shape exist by itself. A shapeless vase is a metaphysically deficient thing. But if shape is a concrete universal, and if we exclude universals from the domain of the independence definition, then we will be compelled to say that the vase minus its shape is a substance. For it will be capable of existing without anything else, leaving aside universals. (I assume here that the vase qualifies as a substance in other respects.) We cannot, therefore, exclude concrete universals from the domain of the independence definition as we can necessary beings or abstracta. Instead I propose two alternative replies to the problem to which concrete universals give rise. First, we might argue that a concrete universal must in fact be capable of being subsumed by something in one location while also being present elsewhere. For concrete universals are meant to be wholly present wherever they are instantiated. And the claim that x is wholly present in location L entails that, whatever else might be true of x, x is subsumed by the things in L. This might sound incoherent. But there is not obviously any addition here, to the apparent incoherence of things being wholly present in multiple locations in the first place, something proponents of concrete universals are already committed to tolerating. Secondly, we might stipulate that the extension of “some y such that y is something over and above x” in the independence definition is to be determined, not at the actual world, but in the counterfactual world where x exists without anything without which x is capable of existing. For even if the color of Nelson’s statue, construed as a concrete universal, is not actually subsumed by the statue, it might counterfactually have been subsumed by the statue, in a world where Edinburgh Castle and Holyrood Palace do not exist. It seems likely that one or other response will succeed. But if the property-dependence objection turned out to pose a unique problem for proponents of concrete universals, it would be reasonable to suspect that the problem lies with concrete universals. For the idea of something capable of existing by itself is intuitively coherent, whereas the idea of something capable of being wholly present in multiple locations is independently dubious. Objection 10. The final three objections are drawn from Hoffman and Rosenkrantz. Hoffman and Rosenkrantz, Substance, 22, 44. First, Hoffman and Rosenkrantz say that the independence definition entails that anything with essential origins is not a substance. This is a problem because, as Kripke argues, it is plausible that many object-like things have their origins essentially. Saul Kripke, Naming and Necessity (Cambridge MA: Harvard University Press, 1980). Schechtman, “Substance and Independence,” 184–6 raises a similar objection. For example, it is plausible that a given human being could not exist unless her parents exist. And yet, the parents of a human being are something over and above her. It seems to follow that a human being cannot be a substance. This objection has been adequately addressed by Gorman and Schneider. Gorman, “Independence and Substance,” 152; Schneider, “A Certain Kind of Trinity,” 400. The objection presupposes that a substance must be able to exist by itself for the entirety of its existence. More precisely: the essential origins objection presupposes at least that a substance must be able to exist by itself for the entirety of its existence. It may even presuppose that a substance could have been the only thing that ever existed. For it is not clear that something with essential origins that are something over and above it need ever exist simultaneously with those origins. Perhaps, for example, for any given human being, that human could have come into existence by in vitro fertilization after their parents ceased to exist. However, the independence definition need only be taken to imply that a substance could exist by itself at some time. On the first construal, the definition rules out substances with essential origins that are something over and above them, but on the second it does not. (If necessary, we could specify this in the definition by adding a time variable. E.g. restricting our attention to the domain of contingent concreta, x is a substance if and only if the existence of x at some time t does not necessitate that there exist some y at t such that y is something over and above x at t.) Objection 11. Secondly, Hoffman and Rosenkrantz suggest that the independence definition excludes substances with essential external relations. For example, they propose that a wife cannot be a substance, because wives cannot exist without husbands. I suggest that, as stated, this objection equivocates de dicto and de re modality. Aristophanic octopods aside, any given wife might have existed although her actual spouse did not, even if she could not have been a wife had some spouse not existed. Cf. Schneider, “A Certain Kind of Trinity,” 397. Perhaps there are better examples. For instance, Donnchadh O'Conaill has suggested to me that a vase only counts as a vase if it stands in external relations to a society. This example is more effective, insofar as it is plausible that vases are essentially vases. But it is less effective insofar as it is not obvious that being a vase really involves external relations. For although it is plausible that a vase must start out standing in external relations to a society, it is natural to suppose that vases can survive when the society dies out. I am skeptical that there are intuitive examples of object-like things that have essential external relations. If there are, it might be possible to defend the thesis that object-like things are substances by taking the view that, for example, a vase only seems object-like to us because we confuse it with the vase-shaped ceramic substance that would survive if the society that makes it a vase died out. But at present I do not think that it need come to that. Objection 12. Finally, Hoffman and Rosenkrantz suggest that the independence definition excludes concreta generally, because no concrete thing can exist without a place and/or a time at which it exists. Alfred North Whitehead and Lowe make the same point concerning place. Schechtman finds a similar concern in Spinoza. See Alfred North Whitehead, Process and Reality (Toronto: Collier-Macmillan, 1969), 75; Lowe, The Possibility of Metaphysics, 141; Schechtman, “Substance and Independence,” 183, fn. 52. In response, I observe first that this objection only poses a threat to the thesis that object-like things are substances on those “substantivalist” views that represent space and/or time as something over and above their concrete occupants. For this reason, it seems probable that Descartes has the resources to meet the objection within his system already. For Descartes appears to reduce space and time to the properties of extension and duration. To ensure that the independence definition is serviceable beyond Descartes’s system, we could stipulate that, in restricting the domain of the definition to contingent concreta, we mean to include the occupants of space and time only, and not their substantival containers, if substantivalism is true. Again, this specification might be regarded as retaining the spirit of Descartes’s definition under conditions he does not consider. There is also another sort of response that we might give to this, and to some of the preceding objections. We might urge that on certain revisionary metaphysical hypotheses, the things we usually think of as object-like turn out to be property-like and so, insofar as the independence definition is meant to pick out object-like things, it should exclude them. For example, the thesis that concreta necessitate the existence of space and/or time, construed as something over and above them, might be judged to entail the “super-substantivalist” thesis that what we usually regard as object-like things are really features of space and time, or of “spacetime.” Cf. Dennis Lehmkuhl, “The Metaphysics of Super‐Substantivalism,” Nous 52, no. 1, (2018): 22–46. Likewise, if some of the items that we naturally think of as object-like parts turn out to be non-detachable, as Alex Moran has recently argued, then we might judge that these are really property-like features of their bearers. See Alex Moran, “The Paradox of Decrease and Dependent Parts,” Ratio 31, no. 3 (2018): 273–284. If so, we might say that while pretheoretically—in the “manifest image”—vases are object-like things and hence substances, in reality they are neither. The independence definition might, then, capture what makes object-like things object-like both in the manifest image and on revisionary metaphysical hypotheses, even when its extension varies between the two. Indeed, it is plausible that Descartes’s own writings contain some applications of “substance” that reflect natural pretheoretical metaphysical assumptions and other applications that reflect his own revisionary theses. But I put off discussion of this proposal for a future enquiry. I am very grateful to Arif Ahmed, Susan Capener, Tim Crane, John Cottingham, Richard Holton, Maxime Lepoutre, Donnchadh O'Conaill, Olley Pearson, Stephen Priest, Howard Robinson, Mikołaj Sławkowski-Rode, Alex Moran, David Weir, Dean Zimmerman and an anonymous reviewer for the Review of Metaphysics for helpful comments on this essay and/or discussions about the ideas it considers. 44