Susan Stebbing (1885–1943), until very recently neglected by both historians of analytic philosophy and present-day philosophers, is beginning to be recognized as an important analytic philosopher in her own right (Chapman 2013; Beaney2016?; Janssen-Lauret 2017; Coliva 2021; West 2022), even as a “founding mother” of analytic philosophy (Janssen-Lauret 2024). Here I lay out Stebbing’s arguments against the assumption that reality is well-founded, which was commonly held in the early phase of analytic philosophy. Her arguments rest partly on her innovative metaphysical method, using her directional analysis, and partly on her philosophy of physics. I argue that Stebbing’s arguments indicate that she did not abandon directional analysis as is commonly held but continued to apply it in her philosophy of physics, and that her arguments still have much to teach us in the twenty-first century.

The canonical “founding fathers” of analytic philosophy, Moore, Russell, and Wittgenstein, all shared at one time or another an assumption that reality is well-founded, that analysis will terminate in simples. Stebbing revealed this assumption to be far shakier than previously supposed by making use of her sharp distinction between “same-level” analysis and “metaphysical” or “directional” analysis. She argued that while same-level analysis of language in terms of more language may often be analytic or a priori, metaphysical analysis never is. In doing metaphysical analysis, we are concerned with finding out what constituents of reality, in which arrangement, there are in the world if the sentence is true (Stebbing 1932a, 78–80). This can never be an analytic, logically true, or a priori matter because it makes a demand on reality. Reality may or may not be as described. Therefore, we cannot assume that analysis will terminate in simples; this is not a logical, analytic, or a priori truth but one which is beholden to reality being a certain way. It may be false.

Stebbing drew upon her philosophy of physics to argue that we cannot write off metaphysical analyses just because they sound “paradoxical” (Moore 1925, 54), analytically false, or counterintuitive, nor embrace them because they sound intuitive. Modern physics is full of counterintuitive analyses, such as “this solid table is, at the subatomic level, mostly empty space,” which we nevertheless have to accept as true. Philosophical analysis has to fit around the findings of physics, not dismiss them. In this way, Stebbing’s metaphysics made a key advance on Moore’s. She also argued that the new physics tends to suggest arguments against the kind of well-foundedness assumed by the logical atomism of Wittgenstein, whom she interpreted as falling into a form of methodological solipsism at odds with the methods of physics (Stebbing 1933a) and of Russell, who held that analysis terminates in simples with which we are acquainted, a physical impossibility for some subatomic particles (Stebbing 1932a, 72; Stebbing1937?).

Having explained Stebbing’s metaphysics and her proposal regarding the question of well-foundedness, I then compare it to the contemporary range of options on offer and conclude that Stebbing’s proposal fares quite well and remains a viable option for us today. Although Stebbing’s metaphysical analysis is sometimes compared to truth-making (Egerton 2021), I argue the resemblance is somewhat superficial because truth-making customarily involves some form of determination or necessity. Stebbing, by contrast, did not believe in bespoke metaphysical relations such as determination or metaphysical necessity.

Stebbing’s metaphysical analysis might alternatively be read as analysing a fact in terms of further facts, in terms of further facts, et cetera, and compared to grounding, with which it shares a chain-like structure. Yet Stebbing spoke out against several argumentative strategies and posits widely embraced by grounding theorists. She argued against the use of “What is it to be a so-and-so?” questions both in metaphysics and in philosophy of science (Stebbing 1932a, 74–75), against essentialist assumptions and modal discourse involving necessary truth or intensions (Stebbing 1930a, 28, 433), and against reliance on determinative relations in science (Stebbing 1930a, 392–394), in ways that could usefully inform present-day debates. She would also have rejected monism-based solutions to the problem of well-foundedness because she argued that the assumptions that there are multiple perceiving minds and multiple things perceived distinct from the observer are baked into both physics and philosophy from their inception (Stebbing 1929; Stebbing1937?). Although it is not a logical truth, ontological pluralism cannot in practice be intelligibly denied, according to Stebbing.

In Stebbing’s view, there is a chain-like structure to be found in metaphysical analysis, but it does not relate grounded facts to grounds. It relates higher levels of logical construction to lower levels. If we find out that there is a level of non-constructed, basic facts, then we will have discovered that metaphysical analysis is well-founded. But whether it is or not is neither a matter of logic, conceptual analysis, nor of metaphysics, but of empirical discovery.

1 Early Analytic Philosophy and the Assumption That Reality Is Well-Founded

Stebbing was in many respects an archetypal early analytic philosopher. She wrote in detail about the method of analysis (e.g., Stebbing 1932a). She tended to favour anti-idealism. She sought to build a philosophy fit for the twentieth century and beyond around the deliverances of modern physics and mathematics. Yet Stebbing differed from her colleagues Moore, Russell, and Wittgenstein in consistently questioning the well-foundedness of reality.

One key logical atomist argument for well-foundedness rests on the assumption that sense-data play an important role in analysis. Although Stebbing framed her objections to this style of argument as disagreement with Russell, I argue that they plainly also constitute both disagreement with Moore and progress compared to Moore, thus throwing into doubt the prevailing “Moorean” interpretation of Stebbing’s metaphysics.¹

Both Moore and Russell had held, from their very early works (Moore 1899; Russell 1903) that some sort of well-foundedness assumption was necessary in order to defeat idealism. They sought to falsify idealism by means of a theory of perception and cognition that sharply separated the mind from its object of judgement, external to that mind (MacBride 2018, 30–39). The two young philosophers aimed to refute both the epistemic idealism of Kant and the idiosyncratic ontological idealism of Bradley by maintaining that, contra Kant and Bradley, reality divides into discrete, individually cognisable constituents, and that we know this because our minds can grasp those constituents directly, and our words can name them directly. In subsequent years, Moore and Russell’s views evolved away from their early all-words-refer model and towards the more familiar logical atomist model on which a true sentence corresponds to a fact and a false sentence fails to do so. Yet both philosophers remained committed to the position that we can grasp and name at least some constituents of reality directly. They referred to these constituents as “sense-data.”

Moore was explicitly open-minded concerning the nature of sense-data. He considered the possibility that they might be the surfaces of objects (Moore 1925, 56), as well as the possibility that they might be mental items. Either way, Moore held, they were pivotal to analysis. Material objects, such as hands, he took to be analysable in terms of the sense-data directly presented to us (Moore 1925, 55–59).

Russell had similarly put forward a strong version of this type of view when he proposed that the process of analysis must terminate in a schedule of sense-data (and universals) to which we have direct cognitive access: “Every proposition which we can understand must be composed wholly of constituents with which we are acquainted” (Russell1910-11?). By holding on tightly to that view, Russell held, we are able to ward off the kind of idealism according to which “we never really, in knowledge, attain to the things we are supposed to be knowing about, but only to the ideas of those things” (Russell1910-11?), because sense-data are indubitable. We cannot be wrong about knowing them. If sense-data are the simples which we know directly and in which analysis terminates, we are on epistemological terra firma.

We are all familiar with the problem of epistemic access to unobservable posits of the natural sciences. The same problem occurs for epistemic access to abstract objects such as sets, numbers, logical properties, and functions for those who posit such entities. But even ordinary material objects, animals, plants, rocks, stars, planets, and artefacts, according to Moore, Russell, and Wittgenstein, are not presented to us directly but only via the medium of sense-data. Whether sense-data are themselves the surfaces of such objects or mental representations of them, belief in material objects such as human bodies or tables then appears to be the result of a risky inference. The classic logical atomist solution, which Stebbing called “Russell’s reformulation of Occam’s Razor” (Stebbing 1932a, 75; 1933a, 25), was that “logical constructions are to be substituted for inferred entities” (Stebbing 1933a, 25). We cannot observe the table directly. But, Stebbing wrote, “it is nonsense to talk of this table as an ‘inferred entity’; hence, they [Russell and his followers] conclude, it must be a logical construct of the given,” that is, of sense-data (Stebbing 1933a, 25).

Stebbing took Wittgenstein to share this Russellian assumption but to stray into even more unpalatable conclusions because, she wrote, “Wittgenstein simply takes it for granted that the given is, and could be, nothing but my own direct experience” (Stebbing 1933a, 27). As a result, “every genuine proposition says, and can say, only something about my present or my future experience” (Stebbing 1933a, 27). So Wittgenstein was, Stebbing concluded, forced into “methodological solipsism” (Stebbing 1933a, 27). This conclusion she took to be confirmed by Wittgenstein’s statement that “what solipsism means is quite correct, only it cannot be said” (Wittgenstein 1922, 5.62). Stebbing’s interpretation of Wittgenstein is controversial. Anscombe’s influential reading, for example, presents Wittgenstein as more realist than Stebbing does (Anscombe 1959). I will not assess Stebbing’s interpretation of Wittgenstein here; her arguments remain valuable even if she was incorrect to single out the early Wittgenstein as her foil.

While Russell denied solipsism, other minds were, on his view, also firmly in the realm of things which we cannot observe directly. He wrote,

If a person who knew Bismarck made a judgment about him […, w]hat this person was acquainted with were certain sense-data which he connected (rightly, we will suppose) with Bismarck’s body. His body as a physical object, and still more his mind, were only known as the body and the mind connected with these sense-data. (Russell1910-11?)

Stebbing, we shall see, took a different view.

2 Stebbing against Well-Foundedness: The Distinction between Same-Level and Metaphysical Analysis

Like her logical atomist colleagues, Stebbing took one of the main tasks of philosophy to be the clarification of our beliefs and our ordinary-language discourse through philosophical analysis. She further resembled them in positing facts with particulars and universals as constituents, although she also took Whitehead’s event ontology seriously as an alternative. But Stebbing’s theory of philosophical analysis differed in key respects from those of Russell, Wittgenstein, and Moore. She rejected the idea that analysis terminates in sense-data. She questioned whether we have good reason to believe that some ultimate level of basic facts will be uncovered. She held that what we analyse are sentences, not propositions or judgements. And she made an advance over all of the “founding fathers” by drawing her distinction between same-level and metaphysical analysis and applying it in her metaphysics and philosophy of physics.

Stebbing’s original views on analysis entail that we cannot expect reality to simply offer up a basic level of simples or simple facts in which analysis terminates. The existence of simples, or basic facts, is, she argued, not a logical or a priori truth. If it is a truth at all, it is one ascertained by inspection of the world. Unlike Russell, Moore, and Wittgenstein, Stebbing distinguished explicitly between metaphysical analysis, which she also called “directional analysis,” and same-level analysis.

Same-level analysis analyses a stretch of language in terms of more language. As a result, such analysis remains at the same level of logical construction: the high-level logical construct of language. Stebbing considered linguistic items such as words and sentences to be logical constructs out of tokens that are typographically or phonetically similar or conventionally associated with each other (Stebbing1935?). Analysis of language in terms of more language—such as conceptual analysis, definition, or analytic explication—may well be a priori, analytic, or based on synonymy. Metaphysical analysis, according to Stebbing, never is. Metaphysical analysis is concerned not with synonymy or analyticity but with uncovering what schedule of facts, what constituents in what arrangement, there is if a given sentence is true. It requires co-operation from reality. Reality may or may not contain the relevant schedule of facts. The claims “analysis terminates in simples,” “there are basic facts,” or “analysis terminates in sense-data” are by no means ungainsayable or logical truths. They can be coherently denied: “that there should be basic facts [is] not logically necessary” (Stebbing 1932a, 80).

Stebbing presented her theory of metaphysical analysis as disagreement with Russell. She framed it as an improvement on his “unfortunate reformulation of Occam’s Razor” (Stebbing 1932a, 75)—according to which we should replace inferred entities with logical constructions—and tentatively claimed some affinity with Moore, while noting that Moore did not use the term “metaphysical analysis” and might not agree with her views (Stebbing 1932a, 76, n.1). As a result, some Stebbing scholars have designated her view as “analysis practiced by Moore-Stebbing” (Milkov2003?) or as a “Moorean conception of analysis” (Beaney2016?). It can nevertheless be made apparent, I argue, that Stebbing’s metaphysical analysis is distinct from Moore’s analysis of propositions and that she was able to solve problems that Moore could not. Although Moore (and Russell) did not intend only to engage in same-level analysis but aimed to uncover “the nature of […] things” (Moore 1925, 55), Moore, unlike Stebbing, failed to keep metaphysical and same-level analysis sufficiently separate, landing himself in a muddle, which Stebbing managed to swerve.

Moore, in his “Defence of Common Sense,” had started off with a strongly anti-idealist message but found himself stymied in the final pages, unable to rule out an idealist analysis of “this is a hand.” He took for granted that any such analysis begins with “This is part of the surface of a human hand,” a statement that he took to be “undoubtedly a proposition about the sense-datum, which I am seeing” (Moore 1925, 55). Moore considered three analyses: one according to which the hand is a logical construction out of physical sense-data, the surfaces of objects; one according to which it is a construction out of sense-data conceived as mental representations of the material hand; and one according to which it is a construction out of “permanent possibilities of sensation” (Moore 1925, 57)—an idealist analysis à la Mill. Moore argued the first was unable to account for double vision, the second relied on the mysterious relation of “being an appearance of,” and then found himself in a kind of aporia, unable to rule out the intuitively unappealing idealist analysis, which he deplored as “paradoxical” (Moore 1925, 59). Although he never explained why he found it paradoxical, I hypothesise that he felt that our common-sense practice of calling hands “material things” (Moore 1925, 42) was at odds with the idealist analysis of hands, since items composed of permanent possibilities of sensation would appear to be mental rather than material things.²

Stebbing’s metaphysical analysis improved upon Moorean analysis by swiftly defanging the apparent paradox. On Stebbing’s view, an analysis that has a paradoxical sound to it is problematic only if the analysis in question is a same-level analysis, capturing some form of synonymy or analyticity. Metaphysical analyses may be perfectly viable even though they sound analytically false. A good example is the physical analysis that matter is, at the sub-atomic level, mostly empty space.

Although Stebbing did not press the above point against Moore, she could have done so because she made a comparable move in her rebuttals of idealism in the interpretation of physics. Stebbing’s expertise in the philosophy of science had made her well aware of paradoxical-sounding analyses in physics, such as “this solid table is, at the subatomic level, mostly empty space.” It would be fallacious to expect a macro-object, or a logical construct, to inherit all the properties of its micro-constituents, or vice versa (Stebbing1937?). Just as an acceptable analysis of a dependable physical table may take it to be mostly empty space at the subatomic level, and this analysis does not imply that the table itself is not solid, so could we theoretically conclude that a hand is, at the level of basic facts, made of permanent possibilities of sensation, and this analysis does not imply that the hand itself is not material. There may not be much that can positively be said in favour of the idealist analysis of hands or other material things, and this was indeed the line Stebbing took. But that is a separate issue, to be settled by an investigation of reality. The paradoxical appearance of the idealist analysis is not by itself sufficient reason to dismiss it as a metaphysical analysis.

Stebbing further differed from Moore as well as from Russell and Wittgenstein in dispensing with the central role allocated to sense-data in analysis. She explicitly rejected Russell’s claim that, as she put it, “a table is a class of appearances” (Stebbing 1933b, 503). We have seen that Moore, too, while not requiring that we view material objects as classes of sense-data, still felt that the process of analysis should have its roots in the type of claim that is “undoubtedly a proposition about the sense-datum” (Moore 1925, 55). But Stebbing’s metaphysical method was different: “We must not start from sense-data” (Stebbing 1932a, 72). According to her theory of perception, we are not simply directly presented with sense-data but thrown into a perceptual situation of determinate shades, sounds, smells, et cetera. Whenever we start to label individual surfaces or mental states, we have already started the process of generalisation and abstraction in some minor way. Thus, Stebbing wrote, “we must start from the perceptual judgment, made in a given determinate perceptual situation” (Stebbing 1932a, 72) when we engage in metaphysical analysis.

Stebbing was not averse to regarding certain posits as logical constructions, such as classes (Stebbing 1930a, 455), linguistic types (Stebbing1935?), and propositions (Stebbing 1933d, 78). She described going back and forth on the question whether tables and other apparently observable things are logical constructions (Stebbing 1933d, 2). Although she on balance felt that tables count as logical constructions, she objected to the view that they are logical constructions out of sense-data. We may say that tables are immediate referents of discourse about perceptual situations. We then enter into a process of abstraction to analyse what they are made of. They turn out to be made, ultimately, of subatomic particles: the basic facts referred to by sentences about tables are micro-physical ones (Stebbing1937?). Stebbing’s views on perception were more akin to those of Whitehead, who, in describing the perceptual situation as an event, sought to overthrow the “bifurcation of nature” into primary and secondary qualities and into mind and body, than to those of Moore.

Stebbing’s views on perception had grown out of her engagement with physics and its philosophy. In a relatively early paper, she defended a position she called “realism,” according to which both philosophy and the natural sciences start from “perceptual science” (Stebbing 1929, 147), comprising statements such as “I am perceiving a piece of paper,” “the piece of paper was here before I saw it,” and “others have seen this piece of paper, too.” Stebbing’s perceptual science explicitly encompasses physics as well as philosophy. It takes it as a given that perceptual objects have a duration and that other minds exist and can perceive the same objects. That other minds are not things to be known only by means of risky inference or logical construction, she appears to have taken as a clear and basic feature of the scientific method. Later, she was to criticise Eddington, for example, for thinking that “the inquiry concerning the nature of an other mind (called ‘Mr. X’) ‘must take place in the domain of my own consciousness’ (Eddington 1928, 268)”; Stebbing countered, “The difficulty is that Mr. X—indeed an army of Mr. X’s—must be assumed if physics is to be possible” (Stebbing1937?). Here we see her invoking her 1929 proposal of “perceptual science,” which has reliance on the observations of other observers running all the way through the scientific method like a stick of rock.

In the same vein, we may read Stebbing’s rebuttal of the methological solipsism she attributed to Wittgenstein as harkening back to her (1929) perceptual science. Stebbing’s reply to the methodological solipsist is brief: “I have the best of grounds for denying solipsism, namely, that I know it to be false. You, who are listening to me, and enable me to speak in the plural also know it to be false” (Stebbing 1933a, 27). Although at this point she included a footnote to Moore, who, unlike Russell and Wittgenstein, also took the assumption of other minds to be basic, this position was really original with Stebbing herself. And, I argue, it derived from her consideration of the scientific method rather than from a “Moorean” sense that the denial of other minds is paradoxical. In 1929, she had relied on the principle “Other people besides myself have seen that piece of blotting paper” (Stebbing 1929, 1) as part of perceptual science, a pragmatically necessary assumption for progress in both physics and philosophy.

3 Stebbing’s Metaphysics and the Question of Well-Foundedness: Metaphysical Analysis Post-1934

We have seen that in Stebbing’s estimation, the question whether reality is well-founded cannot be settled a priori because neither well-foundedness nor its negation is a logical or conceptual truth.³ Statements such as “there are simples” or “there are basic facts” can be coherently affirmed or denied. So can “analysis has no stopping point.” Same-level analysis, such as conceptual analysis, or analysis within a completely conventional or postulational system, will not help us answer the question of the well-foundedness of reality. How, then, did Stebbing propose to answer it? Although the passages in which Stebbing indicates what her answer would look like are compressed, I believe that they contain promising material to inform the present-day debate.

In brief, Stebbing’s alternative answer was that if there are simples, they have to be found in the world by means of the method of metaphysical analysis. Stebbing’s metaphysical method was somewhat naturalistic. She took physics and philosophy to share the same starting point of “perceptual science.” She looked towards physics, rather than sense-data, to find out what hands and tables were made of. Where Moore asserted that analysis begins with what is “undoubtedly a proposition about the sense-datum” (Moore 1925, 55), and Russell, though interested in the philosophy of physics, contended that a table is a set of sense-data, Stebbing’s examples of analysis tended towards the physical: water is made of hydrogen and oxygen (Stebbing 1932a, 67), matter is ultimately made of sub-atomic particles (Stebbing1937?). Instead of the analyses which she attributed to Russell and Wittgenstein, respectively—taking a table to be “a class of appearances” (Stebbing 1933b, 502) or “an experience of mine” (Stebbing 1933a, 28)—Stebbing held that tables ultimately consisted of quarks and leptons (Stebbing1937?): some quarks are arranged into protons, combine with electrons to form atoms, atoms of various sorts combine to form molecules, arranged into cells, arranged into cellulose fibres, arranged into planks, in turn arranged table-wise. The quarks’ proton-wise arrangements here play the role of basic facts. Evidently, their playing this role is not a conceptual truth but an empirical discovery. She also regularly drew upon examples of the analysis of socially constructed entities in terms of natural entities: “The action of the Council is a logical construction out of a set of facts each of which is a fact about one individual member” (Stebbing 1930a, 504).

Like Carnap and Russell, Stebbing felt that analytic philosophy had a role to play in spelling out the implications of the increasingly structural theories of the new Einsteinian physics, which yielded “a constructed system stated in terms of imperceptibles, the system being such that it permits, under certain conditions, of interpretation by reference to perceptual elements” (Stebbing 1933d, 9). Stebbing expressed sympathy for the tradition of Pearson, Mach, and Kirchoff, also embraced by Carnap, according to which “science does not explain but describe” (Stebbing 1930a, 392). Yet she sought to refine some of their rather crude pronouncements, since “a complete description of natural motions [as Kirchoff proposed] is impossible, and if it were not impossible it would be useless” (Stebbing 1930a, 393). Instead, a fruitful scientific theory takes the form of a “constructive description,” which provides fruitful abstractions and generalisations by “attending to certain properties of what there is in Nature, by constructing hypothetical entities (i.e. constructs) whose function is to aid in the correlation of what is sensibly observed, and by using whatever mathematical methods may serve to develop the correlation” (Stebbing1937?).

Historians usually say that Stebbing abandoned metaphysical analysis, either after (1933d) (Chapman 2013, 94) or at least by (1939) (Beaney2016?). On my interpretation, Stebbing did not abandon metaphysical analysis and, in fact, continued to apply it in her philosophy of physics.⁴ It is true that in (1939) and in (1942), she disavowed the exact account of metaphysical analysis that she had given in (1932a). But, read closely, these disavowals are of the Russell-Moore well-foundedness assumption that analysis terminates in sense-data, not of the distinction between metaphysical and same-level analysis. Stebbing wrote, “I was protesting against the view that there is any problem of justifying inferences from sense-data to perceptual objects” (Stebbing 1939, 73). What she had abandoned by then was her (1932a) hope that Moore’s project might coincide with hers: “I did not then clearly see that Moore’s discussion was also entangled with epistemological problems” (Stebbing 1939, 73), as Russell’s had been. In her (1942) retrospective on Moore, Stebbing again made clear that Moorean analysis was, in her view, stymied by the well-foundedness assumption: “Moore certainly has suggested that the analysis must terminate in sets of propositions about sense-data […] There seems to me to be no good reason for asserting that there are such” (Stebbing 1942, 527). Stebbing did, at times, express the worry whether metaphysical analysis without the well-foundedness assumption could be on sufficiently solid ground. But that did not stop her from apparently applying her method of metaphysical analysis in her philosophy of physics. Although she did not invoke it by name there, the analysis she deployed must be metaphysical analysis since it applies to statements that appear analytically false at the level of ordinary language.

Physicists such as Eddington had argued that the modern theory of subatomic particles meant that matter could no longer be viewed as solid. Stebbing argued that this inference relied on the fallacious assumption that predicates that apply to macro-physical objects also apply to their micro-physical components, so that if a macro-object is hard and solid, all its constituents are hard and solid. But such a principle, though it may have an analytically true sound to it—just like Moore’s assumption that if a hand is material, its constituent parts should be material—may be falsified in the case of physical and metaphysical analysis. Indeed, Stebbing wrote, it is so falsified because “no concepts drawn from the level of common-sense thinking are appropriate to sub-atomic, i.e. microphysical, phenomena” (Stebbing1937?). Instead, “it would be more appropriate to say that the modern physicist no longer believes that the table consists of solid atomic balls, than to say that ‘the table no longer possesses solid reality’” (Stebbing1937?).

Stebbing’s opposition to Russell’s contention that “every proposition which we can understand must be composed wholly of constituents with which we are acquainted” (Russell1910-11?) may also have derived from her understanding of physics. If physical simples are quarks or electrons—we know that physical objects can be at least as small as this—then some physical simples resist knowledge by acquaintance. As a matter of physics, we are unable to observe electrons directly. Electrons are smaller than the wavelength of visible light (Stebbing1937?). So Stebbing knew that there were some simples, arguably termini of metaphysical analysis as practiced in physics, with which we cannot possibly be acquainted. This fact further confirmed her position that analysis should not be expected to terminate in simples, which are the objects of acquaintance.

Though naturalistically oriented, Stebbing’s system was more metaphysical than that of Carnap, whose Aufbau Stebbing admired but critiqued for engaging only in same-level analysis (Stebbing 1933a, 1933d), or the later self-professed naturalism of Quine. Quine would have agreed with Stebbing’s dictum that “the metaphysician is not concerned to discover any new facts; he does not add to the sum-total of human knowledge in the way in which the natural scientist or the historian does” (Stebbing 1932a, 65). The Quinean naturalist philosopher famously builds her philosophy around the deliverances of the sciences instead of seeking to build a prior metaphysical foundation for them, and so, too, for the Stebbingesque naturalist. But Stebbing made clear that she regarded metaphysics as “a distinctive branch of philosophy” (Stebbing 1932a, 65) with its own methods, though not with its own bespoke metaphysical relations or facts.

Stebbing set apart her own chosen method, the method of metaphysical analysis, from other, more traditional methods in metaphysics, such as the deductive method of Spinoza and McTaggart, which rests upon axioms (Stebbing 1932a, 66–67), and Aristotle’s methods (Stebbing 1930a, 432–434). Her own method of metaphysical analysis implies a metaphysics of levels. Her other term for metaphysical analysis, “directional analysis,” indicates that the process tends towards ever greater simplicity. Its goal is to “determine the elements and the mode of combination of those elements to which reference is made when any given true assertion is made” (Stebbing 1932a, 79). Stebbing’s metaphysics of levels can be precisely characterised; the language of “levels” is not metaphorical. The lowest level is the level of simples, if there are any, combining into basic facts. Higher levels represent increasing amounts of logical constructedness. Same-level analysis, as we have seen, connects stretches of language—high-level logical constructions out of language-tokens—to more language, remaining at the same level. So, although same-level analysis may be said to have a “direction” in the sense that, for example, the right-hand side of a definition elucidates the left-hand side, it is not directional analysis in Stebbing’s precise, technical sense of descending down the levels in search of greater simplicity.

Stebbing’s range of ontological categories was also greater than Quine’s. She had long accepted the existence of particulars and universals as basic, though admitting that Whitehead’s event ontology was a worthy alternative against which she could not offer any strong arguments (1925, 315–316). Stebbing posited facts, with universals and particulars as constituents, writing, “A configuration of elements is what is usually called a fact. To the configuration is due the unity of the fact; to the elements it is due that there is something to be configured” (Stebbing 1932a, 80, her italics).

4 Analysis of Sentences vs. Analysis of Facts: Comparison with Recent Debates

Stebbing’s metaphysics bears some resemblance to truth-maker theory, to which it has been compared (Egerton 2021). One obvious respect of resemblance between truth-making and metaphysical analysis is that both Stebbing and most truth-maker theorists posit facts whose constituents are particulars and what Stebbing sometimes called “characteristics,” namely properties or relations (Stebbing 1933d). The other respect of resemblance is Stebbing’s contention that what philosophers analyse are not concepts or things, but sentences. Unlike many present-day truth-maker theorists, Stebbing also denied that we analyse propositions in metaphysical analysis. Whether the truth-bearers are sentences or propositions, the view that analysis has a truth-bearer as a point of departure and ends with facts remains a striking commonality between metaphysical analysis and truth-maker theory. Yet I will argue that the surface-level similarity is, to an extent, deceptive.

The proposal that we analyse sentences, as opposed to concepts or propositions, is one that Stebbing presented as a difference between her and Moore. She wrote,

I prefer to use somewhat different language from that used by Moore […] Where he speaks of “knowing what a proposition means, in the sense of being able to give a correct analysis of its meaning” I prefer to speak of “knowing the analysis of a sentence” […] I believe that what we analyse are expressions, of which sentences are one kind; and that when we analyse a sentence expressing a proposition what we obtain is not another proposition but another expression. (Stebbing 1933a, 9)

We saw above that Russell, too, had spoken of “propositions,” but he used “proposition” in a sense in which the contemporary debate uses “fact.” Stebbing, by contrast, wrote, “I think that a proposition is a logical construction out of a set of facts in which someone is using a sentence to express what he is truly or falsely judging.” As her propositions were not sui generis meaning-entities residing in an abstract third realm, her use of the term is not precisely Fregean, but it is clear that she did not use “proposition” in its Russellian sense.

Stebbing, then, bypassed propositions as meaning-entities in her metaphysical analysis. She viewed metaphysical analysis as crucially involving sentences and other linguistic expressions on the one hand and facts on the other, writing, “Metaphysics is a systematic study concerned to show what is the structure of the facts in the world to which reference is made, with varying degrees of indirectness, whenever a true statement is made” by means of a sentence (Stebbing 1932a, 65). Stebbing appears to have had in mind that metaphysicians analyse sentences as used on a given occasion rather than abstract sentence types. She noted frequently that sentences of the same type may have different meanings on different occasions of use (Stebbing 1930a, 149).

Although Stebbing’s system resembles modern truth-making theory in positing the ontological categories of fact, particular, and universal, there are also striking differences. One is Stebbing’s lack of reliance on modality or intensionality. In taking truth-bearers to be sentences rather than propositions, as well as in being devoid of any assumption of necessitation between sentence and fact, Stebbing’s position resembles the views of Quine, Tarski, Goodman, and other mid-analytic extensionalists more than present-day truth-making theory. Note that the Stebbing quotation in the previous paragraph is entirely non-modal. It contains no “must” or “ought” or “necessarily.” The same is true of her formulation of the aim of metaphysical analysis, to “determine the elements and the mode of combination of those elements to which reference is made when any given true assertion is made” (Stebbing 1932a, 79).

These extensional formulations were neither a coincidence nor the result of an oversight on Stebbing’s part. Apart from potential Humean scruples about the necessary connection between particular and universal in a fact (MacBride and Janssen-Lauret 2022, 83–84), Stebbing’s position was largely compatible with Quinean-Tarskian extensionalism.⁵ She consistently disavowed the ascription of metaphysical necessity to the world (Stebbing 1930a, 175–176, 265–266, 433). Stebbing countenanced “logical necessity” (Stebbing 1930b, 285) and the necessity of analytic truths (Stebbing 1933c, 193), but not, apparently, metaphysical necessity. She did not regard causation as necessitation, writing instead, in her chapter titled “Causality,” “The question of necessity does not arise for the practical agent and cannot arise for the scientific investigator until he has generalized from the particular instances so as to obtain the form whenever \(X\), then \(E\)” (Stebbing 1930a, 265).

The truth-maker theory of the late twentieth and early twenty-first century, in addition to positing propositions as truth-bearers and facts as truth-makers, also often posits a peculiarly metaphysical relation of truth-making. According to these accounts, truth-making is necessitation (Armstrong 2004, 5), the relation in virtue of which something is true (Armstrong 1989, 88), or a link between a fact and the essence of a truth-bearer (Lowe 2006, 203–210). Stebbing, by contrast, declined altogether to posit essences or bespoke metaphysical facts and relations: “metaphysics is not concerned with a distinctive region of fact” (Stebbing 1932a, 66). Stebbing’s thesis that metaphysics does not have its own distinctive subject matter in conjunction with her moderate naturalism yielded the view that a modern, scientifically informed philosophy develops in tandem with modern science to dispense with notions of determinative explanation, necessitation, and essence.⁶ Stebbing wrote, “Modern theories of organic evolution have combined with modern theories of mathematics to destroy the basis of the Aristotelian conception of essence” (Stebbing 1930a, 433).

Although an account of truth-making as entailment (cf. MacBride 2013, sec. 1.1), which Stebbing regarded as a primitive logical relation (Stebbing 1930a, 222), might in principle have been open to her, Stebbing did not take that path. Instead, she affirmed that the entailment relation runs both from truth-bearer to analysis and vice versa: “If \(\pi_{1}, \pi_{2}, \ldots \pi_{n}\) is the analysis of \(p\), then \(p\) entails and is entailed by \(\pi_{1}, \pi_{2}, \ldots \pi_{n}\)” (Stebbing 1932a, 85).

Yet Stebbing’s text pulls in two different directions concerning the question of what metaphysicians analyse. In addition to speaking of metaphysical analysis as a relation between a sentence and an array of basic facts, she at times also described it simply as an analysis of facts. For example, “At present I shall use this expression [‘proposition’], but later I shall inquire whether the analysandum may be regarded as a fact” (Stebbing 1932a, 77) and “I think that metaphysics is primarily concerned with the analysis of facts; it is not concerned with the analysis of things, though the special natural sciences may be so concerned” (Stebbing 1932b, 310).

The twenty-first-century reader might be tempted to draw a different comparison here: that Stebbing’s metaphysical analysis is not like truth-making but like grounding. Grounding is often taken to be a relation between a fact and another fact or facts, which is asymmetric and transitive, and which can form a chain-like structure with \(p\) being grounded in \(q\) being grounded in \(r\) and \(s\). Whether it is well-founded is a subject of debate (e.g., Dixon 2016). The truth-making debate is largely unconcerned with questions of well-foundedness, chain-like structures, and their logical properties, such as asymmetry or transitivity. Truth-making need not be a one-one relation, with one truth-making fact per proposition; logically molecular propositions might be made true by sets or collections of atomic facts instead of by logically molecular facts. Nevertheless, the truth-making relation is not generally thought to have a chain-like structure (Fine 2001, 25). Metaphysical analysis, by contrast, does, and Stebbing explicitly commented on it.

There are some passages in Stebbing’s work that, in isolation, appear to suggest that the question of well-foundedness might be fruitfully addressed by reading metaphysical analysis as a type of, or analogue of, grounding and bringing Stebbing’s answers under the grounding umbrella. But I will argue that an interpretation of Stebbing in grounding terms cannot be maintained. Stebbing consistently argued against many of the metaphysical tools in the grounding theorist’s arsenal: metaphysical determination, priority, essence, metaphysical “why”-questions in science, metaphysical ultimacy, and the metaphysical distinction between appearance and reality. What’s more, several prominent grounding theorists compare analysis unfavourably to grounding, but I will show that from Stebbing’s text we can extract a promising argument in favour of analysis over grounding.

At first, the case in favour of reading Stebbing’s metaphysical analysis as a grounding-analogue may seem strong based on certain passages. In the following, she appears to attribute a chain-like structure to an array of non-basic and basic facts:

A fact \(F\) is based upon a fact \(F'\) when \(F\) cannot be unless \(F'\) is. If \(F\) is based upon \(F'\), then \(F\) contains a configured element \(F'\). Since a simple fact contains no configured elements, it cannot itself be based upon any other fact. (Stebbing 1932a, 80)

The immediate reference of a proposition is never a basic fact, but it is in conformity with usage to say that a proposition asserts a fact, and if the proposition be true there is an ultimate reference to basic facts. We cannot tell by simple inspection whether a proposition is true or false, but we can determine the immediate reference of any proposition. A proposition is an assertion; an assertion entails a thinker, but the immediate reference of a true proposition does not depend upon its being asserted. Consequently, we must admit that there are non-basic facts. But non-basic facts are facts of a different kind from basic facts. (Stebbing 1932a, 81)

Elsewhere, she commented on the logical properties of the chain relation, calling it “asymmetrical and transitive,” properties that are also widely ascribed to the relation of grounding:

I am in the habit of describing the analysis involved in metaphysical inquiry directional in order to contrast it with other forms of analysis, which may be circular. To say that the analysis of \(F\) is directional is to say that if \(F\) be analysed into \(a\), \(b\), \(c\), then \(a\), \(b\), \(c\), are on a lower level than \(F\); and if \(a\) be analysed into \(a_{1}\), \(a_{2}\), then \(a_{1}\), \(a_{2}\) are on a lower level than \(a\). The relation of being on a lower level than is clearly asymmetrical and transitive. To say that \(a\) is on a lower level than \(F\) is to say that \(a\) is in some sense simpler than \(F\). (Stebbing 1932b, 311, n.4)

She added that while metaphysical analysis is assumed to be well-founded, assumed to terminate in simples, “there is a tendency to assume [i.e., without argument] that an ultimate element is an absolutely simple element” (Stebbing 1932a, 89), but this, again, is an assumption that can be false because it is not logically or conceptually necessary. Perhaps simples can be discovered or encountered in the world. Stebbing found it “plausible” that “an absolutely specific shade of colour, or taste, or sound, may be simple in the required sense” (Stebbing 1932a, 91). Nevertheless, she wrote, that suggestion remains logically contingent, beholden to what reality is actually like: “To assert that a basic fact is an absolutely specific fact is to make a significant assertion about the constitution of the world. It is not to make an assertion about synonymous expressions. It may be false” (Stebbing 1932a, 89). She therefore worried that it is uncertain whether, in circumstances where the field is dominated by the well-foundedness assumption, we can do metaphysics at all. “Metaphysics awaits its Galileo” (Stebbing 1932a, 93).

5 Analysis of Facts in What Sense? Stebbing against Grounding

A central claim of many versions of contemporary grounding theory is that we certainly can do metaphysics, appealing to bespoke metaphysical grounding or determination relations. Some are also explicit that this method is to be preferred to analysis (Fine 2001; Berker 2018). But Stebbing would not have awarded the founder of grounding the title of “metaphysical Galileo.” We have seen that she opposed positing specifically metaphysical relations. She supported her claim that “metaphysics is not concerned with a distinctive region of fact” (Stebbing 1932a, 66) with detailed arguments against the existence of distinctively metaphysical determination relations.

Canonical statements of the grounding project include “there is a primitive metaphysical concept of reality” (Fine 2001, 1). Such statements also explicitly trade on the contrast between appearance and reality (Fine 2012, 41). Stebbing objected that such metaphysical claims can neither be empirically supported nor are they generally supported by sound metaphysical argument:

The phrase “ultimate nature of reality” implies that reality has a nature that is not apparent. “Ultimate” cannot here be so interpreted as to signify that which could be discovered by analysis or by experimental observation. [Metaphysicians assume] the opposition of Reality to Appearance. It is important to ask what is the nature of this opposition.

Consider the opposition of a chemical compound to its constituents. […] Only a very muddled chemist could suppose that hydrogen is more ultimate than water in any sense other than “chemically more simple.” The case is quite otherwise, however, with the opposition of the ultimate nature of reality to its apparent nature. This distinction is not yielded by experimental observation; it is not yielded at all. On the contrary the philosopher who accepts the distinction starts from the ultimate. (Stebbing 1932a, 67)

Advocates of grounding would counter that they do have sound metaphysical argument to support their views, for example, via appeal to a “constitutive conception of essence” (Fine 2012, 71), a conception going back to the Aristotelian roots of essentialism, free from the modern disease of conflating an essence with a necessary property. Stebbing, who knew Aristotle’s text very well, cannot be accused of conflating essence and necessity. Though correctly describing Aristotelian essence in detail, Stebbing made clear that it did not meet her standards of intelligibility.

Aristotle’s notion of essence is difficult to understand. He nowhere clearly explains it, but seems to take “essence” as a technical term to be left undefined and by means of which he defines those predicables that are to be contrasted with it. (Stebbing 1930a, 429–430)

Stebbing’s standards for intelligibility in metaphysical explanation were exacting. This is evident, for example, from her reflections on the use of mereological composition terminology:

It makes sense to say that lemonade is composed of lemon-juice, water, and sugar. […] It makes sense to say that water is composed of oxygen and hydrogen, although this is a different usage of “composed of” from the usage in the statement about lemonade. […] But what meaning can be assigned to “the ocean is composed of water?” […] To this, it seems to me, the correct answer is that the question involves a misuse, or at best a wildly Pickwickian use, of “composed of.” (Stebbing1937?)

Her intelligibility constraints appear to rule out the “classical mereological relation” (Wilson 2014, 539; Berker 2018, 763) discussed in the modern grounding debate, which allows for statements such as “the ocean is composed of water.”

Along similar lines, Stebbing argued that the metaphysical use of the vocabulary of “priority,” “determination,” “ultimacy,” “appearance,” and the like goes beyond our ordinary-language use of cognate terms and strays into the realm of misuse. We have seen that she was willing to admit logical necessity, though not specifically metaphysical necessity. Stebbing was careful to ward off the intrusion of metaphysics into her logic. Concerning the proposal that axioms may be defined in terms of logical priority, she objected, “But logical priority is not absolute. The notion of logical priority is obscure. Its discussion has been encumbered with difficult and dubious metaphysical assumptions” (Stebbing 1930a, 175). She went on to argue that priority could perhaps be defined in terms of simplicity, but that as “simplicity is also a relative notion” (Stebbing 1930a, 175), not an absolute one, the problem is not thereby solved.

Stebbing was happy with the use of the word “determination” in its mental or epistemological sense, as, for example, in “the determination by experiment of those properties of phenomena that vary quantitatively” (Stebbing 1930a, 313). But she warned that in slipping into using “determines” in a metaphysical sense, we may unwittingly slide into running those two senses together: “The question of one-one [causal] determination belongs to the retrospective attitude; it concerns knowledge, not action. […] The statement of a causal uniformity is a generalization; consequently, it involves abstraction” (Stebbing 1930a, 264).

Stebbing’s arguments here found an echo in Carnap, who, one year later, noted that “prior,” etymologically speaking, means “before.” In its metaphysical use, by contrast, “it is not supposed to mean the temporally prior any more, but the prior in some other, specifically metaphysical, respect” (Carnap 1931, 225). Carnap, of course, drew much stronger, globally anti-metaphysical conclusions from this lack of intelligibility.⁷ Soon afterwards, Stebbing wrote,

The Logical Positivists, including Wittgenstein, agree in rejecting certain traditional, and still not uncommon views, concerning the nature of philosophy. […] With this rejection I also agree. The views rejected are those which hold that philosophy is concerned with the “ultimate nature of reality.” But in this phrase “ultimate” stands for nothing. (Stebbing 1933a, 5)

Stebbing, of course, did not derive an anti-metaphysical conclusion from her rejection of such traditional metaphysics. We have seen that she believed in a specifically metaphysical method and affirmed ontological commitments to particulars, universals, facts, and physical objects. Yet Stebbing’s reservations about the vocabulary of “priority,” “determination,” “ultimacy,” and “essence” were not of the nature of a merely sceptical challenge, holding out for this vocabulary to prove its usefulness, as modern critics of grounding often do (Wilson 2014; Koslicki 2020), nor even the stronger kind of scepticism that largely questions the intelligibility of “grounding” and its associated vocabulary (Daly 2012). She disavowed necessity, essence, and classical mereology. She took her arguments to license a fully-fledged “rejection” of metaphysical systems, which assumed that such vocabulary referred to bespoke metaphysical relations. Her anti-grounding conclusions were stronger than those of Wilson and Koslicki, who admit determination relations but no overarching grounding relation, or even Daly, who proposes that grounding claims might be “cases of restricted necessities” (Daly 2012, 98).⁸

Stebbing would also have opposed monism, both substance monism and priority monism, as a solution to the problem of well-foundedness. Substance monists maintain that there is (“ultimately”) only one thing. But in Stebbing’s view, “‘ultimate’ stands for nothing” (Stebbing 1933a, 5) in its metaphysical use, and the assumption that there is only one thing runs afoul of the method of perceptual science. To construct a physical theory requires a plurality of observers and theorists, and the same is true of philosophy (Stebbing 1929, 147; Stebbing1937?). Theorising is impossible without the assumption that multiple perceiving minds can perceive the same, mind-external objects. Pluralism is, of course, not a logical truth. We can, without contradiction, say that there is only one thing. But, Stebbing would have said, I cannot intelligibly maintain that there is only one thing when I do physics or philosophy of physics. I cannot do philosophy and coherently maintain that there is only one thing if I adhere to even a moderate philosophical naturalism. As she objected to Wittgenstein and the early Carnap, we (multiple persons) know that the assumption that there is only one thing is false. It is falsified when I draw upon physical knowledge or when I interact with other people. “Theoretical physics has developed by the continual modification of common-sense views through a stage of what might be called perceptual science […] unless perceptual science is true theoretical physics cannot be true” (Stebbing 1929, 148). Perceptual science includes the assumption of multiple minds.

Monism failed, as far as Stebbing was concerned, both because of monists’ denial of the naturalistically necessary assumption of the ultimate existence of multiple perceiving minds and because of their persistent reliance on “Aristotle’s notion of priority in nature” (Schaffer 2018, his italics) to argue that the whole is metaphysically prior to, or more ultimate than, the parts. Stebbing maintained that Aristotelian metaphysics of this sort was of dubious intelligibility and had, in any case, been obviated by modern science (Stebbing 1930a, 433).

Last, I will extract from Stebbing’s text a response she could, and likely would, have made against the charge that analysis is merely linguistic and therefore inferior to grounding. Fine, for example, writes that when we analyse “the couple Jack and Jill is married” as “Jill is married to Jack,” it is “the point of the reduction to show that couples are a ‘logical fiction’ and hence not really existent,” but objects that such “reduction is a semantical matter” (Fine 2001, 9, his italics). And Berker objects that in metaphysical disputes, proponents of a certain view “disagree with their opponents—and with each other—over substantive matters, not over linguistic or conceptual matters” (Berker 2018, 739, his italics).

Stebbing would have considered the Fine-Berker objection to analysis to rest on a clear mistake, the mistake of conflating same-level analysis with metaphysical analysis. Same-level analysis is linguistic or conceptual, explicating language in terms of more language. When we engage in metaphysical analysis, by contrast, we “determine the elements and the mode of combination of those elements to which reference is made when any given true assertion is made” (Stebbing 1932a, 79)—where “determine,” as usual with Stebbing, is used in the epistemic rather than the metaphysical sense of the word.

Stebbing’s metaphysical analysis is neither linguistic nor conceptual. It gives a full account of the basic facts, the constituents of the world, and their arrangement, which are there if the sentence is true. Stebbing, like Fine, disapproved of the implications of calling logical constructions “fictions” (Stebbing 1933b, 502) because couples, for example, are not fictitious. But, unlike Fine, Stebbing held that the actions and properties of couples can always be satisfactorily explained in terms that mention only the individual members of the couple. So it is appropriate, by her lights, to regard a couple as a construct. We do for couples as we do for councils:

We may say that a College, or the Council of a College, or a Committee, or a Nation, have acted in a certain way. Thus, for example, we may say, “The Council have elected A as chairman.” This statement says something about each member of the Council, but it does not say of each member that he elected A. But a set of statements could be found, jointly equivalent to the original statement, which would be each a statement about one individual member. The action of the Council is a logical construction. (Stebbing 1930a, 504)

Stebbing went further and argued that metaphysical analysis is, in fact, the only way to step outside a cycle of same-level definitions, a merely postulational system, and formulate metaphysical hypotheses about what demands our claims really make on reality. To start with a specification of a constitutive essence, a question of the form “What is it to be a so-and-so?” she argued, would be useless because it traps us in a postulational cycle. Metaphysicians who attempt to begin with the specification of constitutive natures have thereby gained no knowledge of whether anything exists that really has that nature. While they intended to investigate reality, in practice they remain stuck engaging in same-level analysis instead of getting down to the proper business of metaphysics, metaphysical analysis. Without metaphysical analysis, they have no way to get at reality.

The point I wish to emphasize is that it is a grave defect in metaphysical method to begin the investigation of problems by asking: What is it to be a so-and-so? For example: What is it to be a material thing? What is it to be a cause? The only possible form of answer to such a question is a definition, which leads us nowhere. We must begin with commonsense facts, such as I see this candle, or This blow on his head killed this man, or Her remarks made him angry. It is useless first to define “material thing” or “cause” and then to ask whether the terms so defined are exemplified in the world. (Stebbing 1932a, 74)

In summary, Stebbing’s trenchant objections to many of the pivotal arguments and machinery of grounding mean that a grounding interpretation of Stebbing cannot be upheld. How, then, are we to account for Stebbing’s passages quoted above stating that “the analysandum may be regarded as a fact” (Stebbing 1932a, 77), that “we must admit that there are non-basic facts […] non-basic facts are facts of a different kind from basic facts” (Stebbing 1932a, 81), and that “the relation of being on a lower level than is clearly asymmetrical and transitive” (Stebbing 1932b, 311, n.4)? These seem to present metaphysical analysis as a chain-relation that links a fact to another fact or facts, linked to another fact or facts, and so forth. I suggest that there are two possible readings that make sense of the Stebbing passages that I quoted at the end of section 4.

According to the first possible reading, Stebbing meant that there really are non-basic facts, involving properties and relations distinct from those involved in basic facts, but the properties and relations in question are physical (or biological, or mental), not metaphysical. In Stebbing’s claim that “a fact \(F\) is based upon a fact \(F'\) when \(F\) cannot be unless \(F'\) is” (Stebbing 1932a, 80), the “cannot” must be read as expressing not metaphysical necessity but either logical necessity or a restricted physical necessity. Textual evidence clearly revealed that Stebbing did not believe in higher-level facts formed from special metaphysical relations like necessitation, constitution, or classical mereology. Yet this does not exclude the possibility of some higher levels of facts consisting of lower-level ones standing in physical relations, known to us as the result of empirical discovery. Two up-quarks and a down-quark combine into a proton, which is orbited by an electron to form a hydrogen atom; the hydrogen atom’s electron combines with another atom’s electron into a cloud to create a chemical bond that holds together a molecule; macromolecules combine into DNA strands, et cetera, all composing physical or biological facts.

On this reading, it is unproblematically and literally true both that there are certain non-basic facts and that there are no distinctively metaphysical relations or distinctively metaphysical facts. Macro-facts are formed out of basic facts plus physical properties and relations, and perhaps specifically chemical, biological, physiological, or psychological properties and relations. While this interpretation makes ready sense of much of Stebbing’s text, it is not obvious that it fully accounts for the chain-like structure that Stebbing attributes to “being on a lower level.”

According to the second possible reading, all of the levels except that of basic facts are strictly speaking levels of logical construction, and some of Stebbing’s discourse about non-basic facts must be read as a mere façon de parler, a dispensable shorthand to be explicated in terms of incomplete symbol theory. On this interpretation, Stebbing’s claim that “we must admit that there are non-basic facts. But non-basic facts are facts of a different kind from basic facts” (Stebbing 1932a, 81) constitutes an oblique way of expressing that we must admit the “non-basic-fact” manner of speaking, even though reality only contains basic facts. As Stebbing put the point elsewhere, “To say that the table is a logical fiction (or construction) is not to say that the table is a fictitious, or an imaginary, object; it is rather to deny that, in any ordinary sense, it is an object at all” (Stebbing 1930a, 502). When Stebbing wrote, “The relation of being on a lower level than is clearly asymmetrical and transitive. To say that \(a\) is on a lower level than \(F\) is to say that \(a\) is in some sense simpler than \(F\)” (Stebbing 1932b, 311, n.4), she meant that the relational predicate “being on a lower level than” does not really stand for a relation because it is always flanked on at least one side by an incomplete symbol, which disappears upon analysis.

There is some textual evidence in favour of this second reading of Stebbing. She wrote, for example,

We may perhaps say that “\(\mathrm{S}\)” in a given usage is an incomplete symbol when “\(\mathrm{S}\)” occurs in an expression expressing a proposition and “\(\mathrm{S}\)” is neither a name nor a descriptive phrase referring to a particular which is a constituent of the proposition through some property belonging to a particular. (Stebbing 1930a, 156)

Stebbing made explicit that, concerning discourse about linguistic types, propositions, sets or classes, numbers, and mathematical points and lines, she took the line that these are logical constructs that disappear upon analysis. At times, she also suggested taking this line concerning macro-physical objects:

It does not make sense to say that a logical construction can be substituted for a persistent substantival object, although it is sense to say that a table is not a persistent substantival object, and that every statement about this table can be finally translated into a set of sentences in which the word “table” does not occur. (Stebbing 1933d, 23)

Another advantage of the latter interpretation is that Stebbing’s theory, on this interpretation, bridges the gap between truth-making, which relates a sentence to a fact, and the kind of metaphysical explanation that relates facts to a further fact or facts and forms a chain-like structure. The contemporary debate assumes that truth-making cannot form chains because sentences or propositions, on the one hand, and facts, on the other, are not sufficiently alike in kind. But if the chain-like structure is a feature specifically of the relation of “being on a higher level of logical construction,” then since sentences (or propositions) and non-basic facts are both logical constructs, they are sufficiently alike to feature in the same role in the chain-like structure.

6 Conclusion

Stebbing’s positive proposal concerning the question of well-foundedness is one that combines her own, sui generis kind of metaphysical analysis of perceptual and other ordinary facts with a certain kind of naturalism according to which questions about the structure of reality need to be approached by a divide-and-conquer method assigning different sub-questions to different branches of science and philosophy (Stebbing 1943). Analytic or a priori methods will not settle the question whether reality is well-founded. Modern physics proves incompatible, in different ways, with the atomisms of Russell, of Moore, and of Wittgenstein and leaves room for a non-wellfounded reality. The new physics requires us to believe in a plurality of objects and to accept analyses that appear very unintuitive; these analyses are metaphysical, at least in that they make demands concerning the size and arrangement of the components of reality. Stebbing’s naturalism is thus interestingly different from Quinean or Carnapian naturalism, being more metaphysical. Her system brings with it a robust metaphysical apparatus. Though one devoid of necessity, fundamentality, and determination, it includes facts, particulars, and universals (including relations). Stebbing’s system and metaphysical views remain defensible in the twenty-first century and deserve to be better known to us now.