1 Wittgenstein on Historical Propositions

Analytic philosophy of history is strangely moribund, at least as practiced by philosophers. The mid-twentieth century interest in the sub-discipline has almost completely dissipated, killed perhaps by a scientism that caused Anglophone philosophy to neglect historiography's status as a system of knowledge.Footnote 1 "Analytic philosophy of history" comprises the work of Dray (1957, 1964, 1966), Danto (1965), Gardiner (1952, 1974) and Walsh (1960). These writers applied the methods of analytic philosophy to historical explanation and knowledge, addressing such questions as whether explanations in history require general laws, and the place of causation in history. Their empiricist-inclined work contrasts with speculative philosophy of history, both Marxist and Idealist, the latter including Dilthey, Croce and Collingwood.

Wittgenstein's On Certainty was written in the period of analytic philosophy of history's flourishing, the late 1940s and early 1950s, though it was not published till 1969. Its author had little to say directly on philosophy of history, but On Certainty offers some pertinent remarks which have received little attention. Anscombe's short article on Hume and Julius Caesar briefly acknowledges its debt to these remarks—or at least to On Certainty's general stance. But responses to Anscombe have failed to recognise that her positive treatment develops ideas from it. For Anscombe, as for Wittgenstein, there are historical propositions that are Moorean propositions. In this article I look at the application of what I call Moorean propositions to history as a discipline. The result is a qualification of empiricism in philosophy of history.

On Certainty denies that there are foundations of knowledge, whether rationalist, empiricist, or Moorean propositions. For Wittgenstein, the latter form a heterogeneous class, distinct both from those non-empirical propositions considered certain by Descartes and his successors, and from the experiential certainties stressed by empiricists. Consider Wittgenstein's examples "I am called L.W.", and "This is N.N.":

Why is there no doubt that I am called L.W.?…One would not think that it is one of the indubitable truths (Wittgenstein 1969, §470, henceforth OC 470).


…what could make me doubt whether this person here is [my old friend] N.N., whom I have known for years? Here a doubt would seem to drag everything with it and plunge it into chaos…the foundation of all judging would be taken from me (OC 613–4).Footnote 2

There are many things that we seem justified in regarding as certain, but for which we find it difficult to adduce evidence. An example is that water boils at 100º C (at one atmosphere of pressure), a "very elementary [proposition] in our text-books":

What kind of grounds have I for trusting text-books of experimental physics? I have no grounds for not trusting them. And I trust them. I know how such books are produced—or rather, I believe I know. I have some evidence, but it does not go very far and is of a very scattered nature… (OC 600).

Such propositions are so fundamental to our understanding of the world that it seems impossible for us to deny them. Many of them once functioned as empirical truths, but now lie "apart from the route travelled by enquiry" (OC 88).

Moorean propositions seem to be empirical, Wittgenstein holds, yet turn out not to be. That is, they are empirical neither in the metaphysical sense of factual or contingent, nor in the epistemic sense of liable to be supported by evidence. Unlike ordinary empirical propositions—"The River Wear is in flood", "It was sunny yesterday in Durham", "The boss is off work because of illness", "There's no cheese left in the fridge"—they are not normally open to doubt. Rather, Moorean propositions function more like a kind of framework within which genuinely empirical propositions operate. Wittgenstein compares them to a riverbed, which must remain in place for our linguistic and epistemic practices to flow smoothly; he also likens them to the hinges of a door, which must remain fixed for language to function. For this reason Wittgenstein suggests that they make up what he calls a world-picture, a body of often unspoken and unanalysed beliefs that forms the basis of an individual's or society's belief-system; as he writes, "the inherited background against which I distinguish between true and false" (OC 94). One could perhaps talk of our common sense world-picture, or a Christian or scientific world-picture.

These Moorean propositions are certainties, perhaps, but not in the conventional philosophical sense; they are not "the indubitable truths", and indeed Wittgenstein regards them as non-epistemic, as neither known nor doubted. Echoing his view of mathematics, Wittgenstein treats them as normative or rule-like, and therefore empirical only in appearance:

"I cannot doubt this proposition without giving up all judgement." But what sort of proposition is that?…It is certainly no empirical proposition. It does not belong to psychology. It has rather the character of a rule (OC 494).

The rules of chess are one model. "The king moves one square" is an example; a game in which the king moves more than one square at a time is not chess. Similarly, Wittgenstein suggests, a context in which it is not accepted that I know that I have hands is not one in which questions about evidence, doubt, knowledge, belief, certainty, etc., can be asked or settled; it is not our normal "language-game". It is not that Moorean propositions ground the practice, as foundational propositions are normally regarded as doing. They are not "rules of grammar", more like "presuppositions of a language-game, practice or discipline". Wittgenstein's suggestion is that, unless they are accepted, the "game" of empirical enquiry is not being played (see Coliva 2010, p. 7). He challenges Moore's treatment of Moorean propositions as well-established empirical claims.

To reiterate, Wittgenstein controversially maintains, in his non-epistemic model of Moorean propositions, that Moorean propositions lie beyond the possibility of both knowledge and doubt. Clearly, this is an unusual kind of "certainty". Some writers treat them as bedrock or basic certainties, required for judgment. But there is a tension between regarding something both as a rule and as a "certainty", even a "non-epistemic" one, i.e. one that can neither be known nor doubted. In learning-contexts, even "non-epistemic certainty" seems the wrong term. "Certain proposition" masks the difference we want to stress—that "This is a hand" usually works as a rule, not as a factual assertion. So instead of "Moorean certainties", I call them "Moorean propositions".

A succinct statement of On Certainty's treatment of Moorean propositions is found at OC 308. There, Wittgenstein claims that the possibility of making judgments rests on accepting these propositions:

…we are interested in the fact that about certain empirical propositions no doubt can exist if making judgments is to be possible at all. Or again: I am inclined to believe that not everything that has the form of an empirical proposition is one.

What "we are interested in" summarises one of On Certainty's key insights: that propositions that appear to function empirically, are in fact framework propositions or rules. In his Remarks on the Foundations of Mathematics, Wittgenstein had considered an analogous analysis of mathematical calculation:

It is as if we had hardened empirical propositions into a rule. And now we have, not an hypothesis that gets tested by experience, but a paradigm with which experience is compared and judged… It is thus withdrawn from being checked by experience, but now serves as a paradigm for judging experience (Wittgenstein 1981, pp 324–325).

He is imagining that arithmetical propositions were once empirical, but had acquired a normative or rule-like status in which they determined the truth or falsity of genuinely empirical judgments. Note however that he writes it is "as if" this had happened. The process is also called "fossilisation" at OC 657.

In my Guidebook, I divided Moorean propositions into personal and impersonal categories. The personal category consists of (i) generally applicable propositions such as "I have two hands" about which most two-handed people are certain in their own case; and (ii) person-specific propositions, comprising truths that are certain for the speaker, such as "I have spent most of my life in the United Kingdom" and "My name is Andy Hamilton". I described the impersonal category as trans-historical propositions that are certain for anyone—such as "The Earth has existed for a long time", "The world contains lots of people", "I have ancestors". However, we will now see that this description is inaccurate—there are impersonal Moorean propositions that are historically specific. Indeed, there are Moorean propositions underlying all fields of enquiry; the present article focuses on historiography—the discipline of history—but also, in conclusion, looks at art criticism. "For many years, people have lived in what are now called the British Isles" is a Moorean proposition that underlies the discipline of history. "The earth existed a long time before my birth" is a Moorean proposition whose denial puts in question the disciplines of history and geology—someone who questioned this proposition "couldn't for example learn history" (OC 206).

For Wittgenstein, the boundary between well-established empirical truths and Moorean propositions is a vague one: "There are cases where doubt is unreasonable, but others where it seems logically impossible. And there seems to be no clear boundary between them" (OC 454). My Guidebook commented that the vague boundary between Moorean propositions and empirical truths is occupied by somewhat more specific historical claims—but again this is not quite correct, because historical claims such as "Napoleon Bonaparte existed" are unequivocally Moorean propositions. There are ambiguous cases, where the proposition can be interpreted either empirically, or as a Moorean proposition. "There was no revolution in Russia in October 1917" might be claiming that the October Revolution was a coup, not a revolution—still a matter for historical debate (see for instance Pipes 1990). But to deny that there was a violent uprising of any description in St Petersburg in October 1917, to insist that Kerensky carried on governing and that the Tsar and his family were not murdered, implies a massive conspiracy through forged documents, newsreels and so on—putting the fabric of modern historical knowledge in question. What other well-attested historical facts will someone question, if they question these? A parallel is with "My name is A.H.", which in one sense can be empirical but in another is a Moorean proposition, the doubting of which wreaks personal epistemic havoc.

"Lenin existed" is also a Moorean proposition; furthermore, at some point a well-attested historical fact becomes merely "bizarre to doubt", but not a Moorean proposition—on the model of "There's a sick man lying here". In Historic Doubts Relative to Napoleon Bonaparte (1819), Archbishop Whateley satirised hermeneutic interpretations of the Bible that questioned its historical truth, suggesting mischievously that these would also leave us doubting the existence of Napoleon. Although Christian theologians of more literalist inclination lost the battle against hermeneutics, Whateley's satire retains its force—if one doubts the existence of such a well-attested personage as Napoleon, what historical claim could not be questioned? Napoleon's existence cannot really be "attested" to; in Wittgenstein's metaphor, it lies outside the route of enquiry, and is part of our historical world-picture.

Whateley's view was echoed by Peirce, who anticipated some central ideas of On Certainty:

certain inferences which scientifically considered are undoubtedly hypotheses [are] practically…perfectly certain. [That] Napoleon Bonaparte really lived at about the beginning of this century [is] a hypothesis which we adopt for the purpose of explaining the concordant testimony of a hundred memoirs, the public records of history, tradition, and numberless monuments and relics. It would surely be downright insanity to entertain a doubt about Napoleon's existence… (Peirce 1998, "The First Rule of Logic", 54).

Wittgenstein would dispute that these "perfectly certain" propositions should be treated as "scientifically…hypotheses", but would agree with Peirce that it is insanity to entertain a doubt about Napoleon's existence.

As Anscombe recognises, the proposition "Napoleon existed" can be empirical in other cultures, or for novices. An historian writing when Napoleon was an obscure young officer, would not have heard of him; likewise in the year 2500, after a nuclear catastrophe in which much of human knowledge was destroyed. But by 1804, when he declared himself emperor, Napoleon was significant enough to be part of the historical framework. While "Kerensky dominated the Provisional Government" or "Kerensky became Minister of Justice in February 1917" are too specific to be Moorean propositions, "Lenin led the October 1917 revolution" is a plausible example of one. The same is true in the natural sciences: "The Earth is 4.5 billion years old" is too specific to count as a Moorean proposition, but "The Earth is billions of years old" would do. Whether a Moorean proposition is being challenged depends on how someone denies that Kerensky was the dominant figure, etc. If they said, "It wasn't Kerensky—it was his indistinguishable doppelganger", then a Moorean proposition is being challenged.

2 Anscombe on Hume, Wittgenstein and Historical Propositions

Wittgenstein's account is developed by Anscombe in a short and rather neglected article, and I now consider these developments. The general tenor of On Certainty, and perhaps the specific remarks just discussed, led her to develop a general account of historical propositions. Her article's most obvious focus is the inadequacy of Hume's treatment, but she offers a positive Wittgensteinian alternative. Concerning Hume's empiricist account, Anscombe argues that "Revision is incontrovertibly needed to secure coherence. The needed revision then reveals the position as incredible" (Anscombe 1973, p. 1). She expresses astonishment that Hume illustrates his account with this example:

we believe that CAESAR was killed in the senate-house on the ides of March; and that because this fact is established on the unanimous testimony of historians…Here are certain characters and letters…the signs of certain ideas…either in the minds of such as were immediately present at that action; or …deriv'd from…testimony…and that again from another testimony…'till we arrive at…eye witnesses and spectators of the event…this chain of argument…is at first founded on those characters or letters, which are seen or remember'd (Hume 2011, 1.3.4).

Wittgenstein and Anscombe have a dynamic concept of evidence—evidence for a proposition is person-relative, and depends on who is looking and when. They realise what Hume seems not to—that historical knowledge involves a dynamic picture. That is, the status of the proposition "Caesar was killed on the Ides of March" changes. Analogously, "Humans have not been to the moon" was originally a framework proposition, which became an empirical claim as rocket-powered flight became possible during WWII, then an empirical falsehood, and finally the negation of a contrary framework proposition.

According to Anscombe, Hume's comment that “‘Tis impossible for us to carry on our inferences in infinitum” means: the justification of the grounds of our inferences cannot go on in infinitum. As she writes:

Where we have chains of belief on grounds believed on grounds…we must come to belief which we do not base on grounds. The argument here is that there must be a starting point of the inference to the original cause, not that inference must terminate. Indeed, one reason why this passage of Hume's seems fairly ordinary and acceptable at first sight is, that he strikes one as just making this point, together with the one that the starting point must be perception.

However, she continues, Hume is arguing not merely that there must be a starting point, but that we must reach it in the justification of these inferences (Anscombe 1973, p. 2)—and this is incredible:

Belief in recorded history is on the whole a belief that there has been a chain of tradition of reports and records going back to contemporary knowledge; it is not a belief in the historical facts by an inference that passes through the links of such a chain. At most, that can very seldom be the case (Anscombe 1973, p. 4).

The second part of Anscombe's article develops a holistic account of historical framework propositions, derived from On Certainty, that contrasts with Hume's atomistic or molecularist picture.

She begins by citing a contrary account in middle-period Wittgenstein, "One of the rare pieces of stupidity in [his] writings":

That it is thinkable that we may yet find Caesar's body hangs directly together with the sense of a proposition about Caesar. But so too does the possibility of finding something written, from which it emerges that no such man ever lived, and that his existence was made up for particular ends (Wittgenstein 1991, IV 56).

This account is untenable, she believes:

What document or inscription could be evidence that Julius Caesar never existed? What would we think for example of an inscription saying "I, Augustus Caesar, invented the story of the divine Julius so that Caesars should be worshipped; but he never existed"? To ask a question Wittgenstein asked much later: What would get judged by what here? (Anscombe 1973, pp. 4–5).

Anscombe explains in a footnote that the question that Wittgenstein asked much later is found in On Certainty: "When he wrote On Certainty Wittgenstein would not have made such a suggestion. I am a good deal indebted to On Certainty in this article." A good example of "What would get judged by what" is the metre-rule in Paris. At one time, it helped to judge the length of other items; then when new methods of measurement appeared in the atomic age, the metre-rule was itself judged by them—and found to be shorter or longer than a metre.

There is, she holds, a "peculiar solidity" in the existence of Caesar, and key events of his life such as his assassination:

an expert on Julius Caesar [is] an expert on whether Caesar conducted such and such negotiations with Pompey or when he wrote his books…Not on whether Caesar existed. Contrast an expert on King Arthur (Anscombe 1973, p. 5).

She continues:

We know about Caesar from the testimony of ancient historians, we even have his own writings!…You were told it. And how did your teachers know? They were told it.


We know it from being taught; not just from explicit teaching, but by its being implicit in a lot else that we are taught explicitly.

However, as Anscombe comments, "it is very difficult to characterize the peculiar solidity involved, or its limits" (Anscombe 1973, p. 6).

That Caesar was assassinated has "a particular logical status of one kind of certainty…that that man, Caesar, existed and that his life terminated in assassination [could be called] in question only by indulging in Cartesian doubt". One cannot check that there was such a person as Julius Caesar. However:

suppose there were a schoolchild who first ran into Caesar through Shakespeare's play. Somehow he doesn't learn at once that this is not a purely fictitious story. He refers to Caesar as a fictitious character, and then someone tells him Caesar was a real historical character. He can check this; he can look into history books and find out that that's what Caesar is. But…I already know—I can at best remind myself of—the status of "Julius Caesar" as a name of an immensely famous man "in history". (To be sure, these things can change.) (Anscombe 1973, pp. 5–6).

So the status of the proposition "Caesar existed" is a dynamic one, that changes between ages and persons. However, if I frame the hypothesis that Caesar never existed, or was not assassinated, I see that "the 'doubt' [puts] me in a vacuum in which I could not produce reasons why such and such 'historical facts' are more or less doubtful" (p. 6). Anscombe agrees that there are vague boundaries, or gradations, between empirical and Moorean propositions in history:

there is the possibility of discovering that some obscure supposed historical figure is probably mythical, or is a conflation or the like. Things get corrected or amended because of inconsistencies.

But as she rightly argues, not everything can be put up for checking simultaneously (Anscombe 1973, p. 7).

Anscombe account of historiography is a development of the remarks in On Certainty, and is not made explicit there. This account interprets history as a chain of human activity, rather than evidence. She refers to a chain of tradition and reports, rather than of inference. As Wiseman and MacCumhaill write, "The 'historical past' is preserved jointly as witness testimony is recorded, preserved, repeated and passed down. To believe that Aeschylus wrote Agamemnon is to trust in continuous human endeavour that has preserved for us the testimony of those who bore witness" (MacCumhaill and Wiseman 2022, Ch. 1). The historian chronicles and explains the past through repeated human effort, copying, reproducing, retelling and reimagining. Anscombe treats the individual witness or observer as a member of a community of humans, that weaves together their common past in a mutual endeavour. Each contributes to the historical tapestry, keeping it alive by recording and preserving testimony through the generations. Historiography—the discipline of history—forms a holistic system. A doubt about a key fact, such as the occurrence of the Russian Revolution, puts the whole system in question.

An objection to the Wittgenstein/Anscombe account is this: Suppose that the people that we believe were Julius Caesar’s parents in fact never had a child, but that a child from some other family was substituted just a few weeks after birth and claimed to be theirs. That child then accomplished everything we attribute to Julius Caesar. Would they be Julius Caesar? Or is no child not born of Julius Caesar's supposed parents Julius Caesar?Footnote 3 My reply is: This is an unlikely development—that hundreds or thousands of years after a major figure lived, historical consensus concerning their parentage changes, based on the appearance of compelling new evidence. In contrast, there are figures surrounded by rumour from birth, such as the son of James II, later the Old Pretender, allegedly smuggled into Mary of Modena's chamber in a warming-pan.

One's response to this unlikely development in the Caesar case depends on whether one subscribes to a Fregean sense/reference account of "Julius Caesar", or a rigid designation view. If one subscribed to the Fregean view, one would say: Julius Caesar turned out not to have the parents we thought he had. Anscombe herself takes a nuanced position on this question, writing that:

In using proper names that we take to be the names of people we don't know, or people in the remote past, we implicitly depend on an "apostolical succession" of users of these names—or linguistic transforms of them—going back to original users…But a discovery that a name belonged originally to a period later than the life-time of the supposed bearer of the name at any rate reduces the status of the name: it becomes equivalent to some set of definite descriptions (Anscombe 1973, p. 4).

It is more possible that historical evidence appeared that there was someone who did almost all the things we attribute to Caesar, but that because of some local superstition about the time of year, the story became that he had died on the Ides of March, when in fact he died in the autumn. Similarly, no one thinks that it is necessary that Jesus was born on Christmas Day.Footnote 4

3 Generalising the Claim: a Wittgensteinian Account of the Artistic Canon

The claims made in this article can be generalised beyond the discipline of history, and I suspect that they apply to every discipline or rational practice. That is, every such practice rests on a river-bedrock distinction, I would argue. That is because every practice must involve both criteria for judging, and matters being judged. I will provide one further example, from aesthetics. Here we see the importance of framework propositions of the kind found in history—though here they are normative, while the propositions of history that we have been discussing are empirical. Wittgenstein did not write directly either on philosophy of history, or on aesthetics, but in aesthetics other writers have developed his ideas, for instance on a cluster or family resemblance account of "art". However, one idea that does not seem to have been developed is an interpretation of canonicity in the arts that is analogous to what On Certainty suggests about history.Footnote 5 As Hume argued in "Of the Standard of Taste", some great works themselves constitute a standard of taste or test of time; classic artworks become the basis, rather than object, of comparison. Thus Shakespeare's plays have become part of the bedrock in terms of which other potentially canonical works are assessed—a constitutive role that they probably acquired by the late eighteenth century. One cannot conceive of a canon of Western drama that excluded Shakespeare; his plays are entrenched in the canon, and help decide what will enter it. As Olsen writes, "to dislodge a play like Hamlet from the literary canon would leave no basis for critical discourse" (Olsen 1996, p. 79). Shakespeare sets criteria for what the English language—maybe any language—can do. His work expresses everything that language can express. Thus the canon illustrates Wittgenstein's argument that an empirical discovery could, over time, gain a special logical role in scientific investigation, as a hinge on which particular empirical judgments turn. In the aesthetic case, the process involves the evolution of an aesthetic or critical judgment—an evaluative proposition—into a logical or framework proposition.

Barbara Herrnstein Smith claims that a "great" book tends to "shape and create the culture in which its value is produced and transmitted and [so] perpetuate the conditions of its own flourishing" (Smith 1988, p. 50). Like other postmodern theorists, she finds this suspicious. The self-perpetuation claim is plausible, but it belongs to the process that Hume articulated and does not, as postmodernists like Smith claim, simply imply the hegemony of a dominant culture. Rather, "Shakespeare is in the canon of drama" or "Shakespeare is a great dramatist" are framework propositions in the sense of On Certainty. "Shakespeare might leave the canon" is as incomprehensible as "Napoleon did not exist". ("Shakespeare did not exist—Bacon wrote the plays" is more comprehensible, because we know so little of Shakespeare's life.) Though the surface grammar appears similar, to consider whether a lesser seventeenth century dramatist such as John Ford is in the canon is not the same as considering whether Shakespeare is. To say "Shakespeare is not part of the canon" is not about subtracting one person from it, but is rather to propose an alternative idea of canonicity, or indeed to reject it entirely. Debating John Ford's inclusion, in contrast, typically rests on a shared background of standards which, among other things, takes Shakespeare's works as a point of reference.

It is a sad commentary on contemporary philosophical attitudes, that philosophy of history has fallen into its current state of neglect. In this article, I have made some suggestions which I hope might help to revive interest in the sub-discipline, and results in a qualification of empiricism in philosophy of history. But these issues apply across all disciplines, and are of independent interest therefore.