I

In 1975 the Clarendon Press at Oxford published Peter Nidditch's edition of John Locke's An Essay concerning Human Understanding. In his Introduction Nidditch says that his edition “offers a text that is directly derived, without modernization, from the early published versions; it notes the provenance of all its adopted readings (some of which are new, correcting long-established errors); and it aims at recording all relevant differences between these versions”. As Nidditch goes on to acknowledge, the “relevant differences” were many, “requiring several thousand registrations both in the case of material variants (deletions, additions, or changes of wording) and in the case of formal variants (changes of punctuation, parentheses, italics, etc.)”. The textual history of Locke's Essay is extremely complicated. While there is no manuscript of the first edition of the book, there were four editions in Locke's lifetime, each new one containing extensive and significant revisions, as well as a posthumous edition published shortly after the author's death. There was a translation into French made with Locke's cooperation and published in 1700, and a Latin translation came out a year later. Nevertheless, Nidditch managed to record all the material variants in footnotes to the text, in a way that makes it fairly easy to track the changes that Locke made to successive editions of the book, and to locate points at which judgements had to be made as a critical text was established on the basis of the chosen copy text. Sometimes a critical edition succeeds in completely changing the way that a text is read. Peter Laslett's 1960 edition of Locke's Two Treatises of Government is a good example. Nidditch's edition of the Essay did not have that kind of very dramatic effect on Locke scholarship. Rather, it made it possible for those without direct access to all the early editions to engage in careful, historically sensitive studies of Locke's account of human understanding. The result was a slow revolution in Locke studies that continues to shed new light on even the most familiar aspects of the Lockean philosophy.

The text presented in the Clarendon edition of Hume's Treatise looks completely different from that of the Nidditch edition of Locke's Essay. The only footnotes to the text are Hume's own, and all editorial matter is contained in separate volume. I believe that this was a misjudgement, and for two reasons. First, compared with Locke's Essay, indeed compared with most of the central texts of the Western philosophical tradition, the textual history of Hume's Treatise is straightforward. There was only one edition in Hume's lifetime. There is in manuscript only a tiny portion of the text, the final section of Book 3. Hume did add an Appendix to Book 3 of the Treatise, which was published eighteen months after Books 1 and 2, detailing a number of substantial changes he wanted to be made to the text of Book 1, and any editor of the Treatise has to decide whether or not to make those changes. This is one major editorial crux. The other major crux concerns a small number of handwritten corrections or amendments made by Hume himself to three copies of the Treatise, the most significant of which are in a copy of Volume 3 now in the British Library. The editorial issues raised by these two cruxes, and other more minor ones, could easily have been raised and discussed in footnotes, without cluttering up the page unduly. The second reason why it was a misjudgement to present a “clean” text is that David Fate Norton and Mary Norton have made a number of controversial alterations to their copy text, alterations which should have been made much more obvious to the reader. As it is, they are documented only in “Register B” of a section of the editorial apparatus entitled “Editing the Texts” (644–62), and the unwary reader is given no other indication of the differences between this edition of the Treatise and all previous editions.

The Clarendon edition of the Treatise is designed to supersede the edition produced by L. A. Selby-Bigge in 1888 and revised by Peter Nidditch in 1978. Selby-Bigge's edition was, in turn, designed to replace the edition published by T. H. Green and T. H. Grose in 1874. Where Green and Grose prefaced their edition with hundreds of pages of introductory material designed to show the reader how Humean empiricism was inevitably overcome by Hegelian idealism, Selby-Bigge wrote no introduction at all. He provided a fairly reliable text and, crucially, a detailed and extremely useful index (for the length of which he apologized in a one-page Preface). Selby-Bigge's edition coincided with, and was no doubt partially responsible for, the birth of modern Hume scholarship, in the form of cool and measured journal articles and monographs intended to save Hume from the treatment he continued to receive at the hands of the likes of Green and Grose. Selby-Bigge printed the Appendix in full at the end of the text. One of Nidditch's innovations was to mark with asterisks the places in the text where Hume intended parts of the Appendix to be inserted. He also added thirteen pages of Textual Notes, recording variant readings, including the modifications made by Hume to the copy of Volume 3 in the British Library. Again, asterisks are placed in the text to indicate passages which were substantially modified by Hume in that copy. Nidditch was not able to change the pagination of the Selby-Bigge edition, so his revisions embody a compromise. They were the best that could be done in the circumstances. But they did succeed in giving all of Hume's readers the sense that the Treatise as published was very far from being a finished piece of philosophy. In the Appendix and in the amendments to the British Library copy we see Hume worrying whether he has expressed himself as well as he could have done, and also, more significantly, revising both his conclusions and the lines of argumentation that generated those conclusions. The Treatise was abandoned rather than completed. Its young author rushed it into print in order, so he imagined, to change completely the philosophy of his age. He later came deeply to regret his own precipitancy. The text presented by the Nortons gives the reader no sense of any of this, and that is a shame.

In this review I shall concentrate on changes made by the Nortons to their copy text that I believe to be unjustified. But it is only proper to begin by indicating the virtues of this edition of the Treatise (which includes the 1740 Abstract and the 1745 Letter from a Gentleman to his Friend in Edinburgh). Volume 2 contains a 150-page “Historical Account of A Treatise of Human Nature from Its Beginnings to the Time of Hume's Death”; no less than three hundred pages of “Editors' Annotations”; a fifty-page bibliography including both the “primary sources that are quoted or to which explicit reference is made, either by Hume or the editors” and “the secondary sources we have cited” (980); and two indexes, the first to the historical account and the account of the editing process, the second to “Hume's Texts and the Editors' Annotations”. There is much here that will be very useful to Hume scholars, for example the photographic reproductions of the sole surviving part of the manuscript (the “Conclusion” of Book 3), and the lists of works cited by Hume that were in the library used by students in Robert Steuart's natural philosophy class (in which Hume enrolled in 1724) and in the library of Hume's nephew (incorporating elements of Hume's own library). (It is regrettable, though, that there is no index of, simply, the works cited by Hume in the Treatise.) The “Historical Account” gives a comprehensive and reliable summary of all that we know for sure about Hume's life up to the time when the Treatise was published. It also gives full summaries of all known contemporary reviews and responses to the Treatise. The three hundred pages of “Editors' Annotations” constitute a genuinely extraordinary achievement of scholarship and detective work. Often extensive quotations are given from authors to whom Hume refers or alludes, or who simply form part of the various traditions of thought in which Hume immersed himself, in Scotland, England and France, as he worked on successive drafts of the Treatise. Of course, not every annotation will contain news for everyone, but taken as a whole they present what is probably the best available guide to the intellectual contexts of the development of the first phase of Hume's philosophical thought.

II

Many of the most distinctive features of the Clarendon edition of the Treatise will be familiar to students of Hume from the Oxford Philosophical Texts edition published in 2000, also edited by the Nortons. Like the OPT edition, this one gives numbers to paragraphs, facilitating more precise referencing than mere page numbers allow. Unlike the OPT edition, however, this one includes in its margins the pagination of the Selby-Bigge edition. Like the OPT edition, this one alters or adds to Hume's text in the light of the corrections Hume himself made to three copies of the Treatise. Like the OPT edition, this one corrects Hume's transcription of quotations “by substituting readings that reliably represent the apparent sense of quoted texts and that are found in editions of those texts available to Hume” (632). Also corrected are what are presumed to be misspellings, including the spellings of the names of other philosophers—“Shaftsbury”, “Hutchinson” (and “Hutchison”), “Malbranch”, and the like—that are such a disconcerting feature of previous editions of the Treatise. Most significantly of all, like the OPT edition, this one interpolates material from the Appendix into the main body of the text, with “App.” added in superscript at the start and end of the relevant passages.

The Nortons say very little about why they decided to deal with the Appendix in this way:

There seems to me to be a more substantive issue here than this admits. In their work towards the production of a critical text, the Nortons went to remarkable lengths (fully documented) to inspect virtually all known copies of the first and only edition of the Treatise, work which has yielded one important result, regarding the footnote concerning the imagination at 2.2.7.6 in previous editions (371 of Selby-Bigge), which the Nortons convincingly show was intended to be replaced by the note at the end of 1.3.9 (117 of Selby-Bigge). But there was some reason to preserve the difference between the edition that Hume published and the second edition that he wanted for a time, and then gave up on in favour of recasting the matter of the Treatise in the Enquiries and the Dissertation on the Passions. The Appendix is a document that records the result of some hard thinking on Hume's part about some of his central doctrines, and might be regarded as at least a stage on the way towards a wholesale reformulation of those doctrines (especially the theory of belief and the account of personal identity). Better, then, perhaps, to have followed Nidditch in merely indicating the places where Hume instructed the Appendix passages to be inserted, and to have printed the Appendix in full at the end of Book 3, where it is located in the copy text. (In fact, the Nortons do print the complete Appendix as well, in Editorial Appendix 2, at 674–84.)

The question of what to do with the Appendix passages is of course a difficult one. In his editorial Preface, Nidditch might be read as implying that he himself would have altered Hume's text accordingly had it not been incumbent upon him not to alter the pagination of Selby-Bigge's edition. Rather more controvertible is the Nortons' policy of altering Hume's text in line with the manuscript amendments to three copies of the Treatise, and of silently correcting his quotations and spelling. To take the issue of spelling first, in the absence of a manuscript it is impossible to know whether Hume himself wrote, for example, “Hutcheson” and not “Hutchison” or “Hutchinson”. The Nortons, in support of their amendment, cite the fact that Hume consistently used “Hutcheson” in his letters, and suggest that the compositor “used an English spelling of the name” (647). Well, quite possibly, but why need the editor decide either way? Why not record the spelling as it is in the copy text, with a note to explain what might seem a bizarre lapse on Hume's part? Hume did of course read the proofs of his first book, and we can be sure that he read them carefully. That he did not object to “Hutchinson” is an interesting fact which it would be a shame to lose sight of. There are of course many substantive errors in the text as printed—such as confusions of “these” and “those”, and of singular and plural forms of nouns—which may be such as to deserve to be corrected, as not being what Hume intended. But here again things are more difficult than the Nortons admit. Is it the editor's task to be true to the author's intentions—generally rather hard to be absolutely sure about—or simply to record what the author allowed to be printed? This question is even more pressing with respect to quotations which are corrected by the editors of the Clarendon Treatise. Suppose Hume were quoting from memory. Would it not then be interesting that that is how he remembered the passages in question? Of course, these may be simple errors of transcription, but how to decide the question? Why not in each case present the text as Hume saw it through the press, and record in a footnote the differences between it and the passage as printed in standard editions? The fact is that Hume and his contemporaries worked to different standards of quotation than we would consider appropriate today. It was normal to adapt quotations to one's own purpose, and no one bothered about exactly reproducing the spelling and punctuation of original texts.

The amendments made by Hume to one of the copies of Volume 3 of the Treatise now in the British Library are often substantive and interesting. Almost all of them pertain to the “artificial” virtues, and to how the value accorded to those virtues changes when, once society is established and stable, it is no longer perfectly obvious that they serve the self-interest of each and every one of us. They do not between them suggest anything like a complete revision of the doctrines of the Treatise as first published, but they see Hume thinking more about the distinction between the natural and the artificial, and about the best way of characterizing the role of sympathy in the development of a moral dimension to justice and, especially, to allegiance. For example, at the end of his first delineation of the question of whether justice is a natural or artificial virtue, Hume claimed in the published version of the Treatise that it follows from what he has said that “we have no real or universal motive for observing the laws of equity, but the very equity and merit of that observance” (3.2.1.17), making it a puzzle why we attribute virtue to acts of justice and vice to acts of injustice. Hume changes this to “we naturally have no real or universal motive”. When you know that this “naturally” was added later, it comes to seem as though Hume might have felt the need explicitly to deny that there really is such a puzzle, and that he wanted to make it plain that justice can become like other virtues and thus derive its virtue from a non-moral motive. Close to the end of the next section, in the all-too-brief treatment of why justice and injustice become distinctively moral issues, Hume's account of the role of sympathy in this connection is supplemented by the intriguing remark that “This latter principle of sympathy is too weak to controul our passions; but has sufficient force to influence our taste, and give us the sentiments of praise and blame” (3.2.2.24). Why did Hume feel the need to underline here the fact that morality as he understands it is very often more a matter of fine feeling than of difficult action? On the one hand, Hume draws attention to the impotence of sympathy considered as a spur to action; on the other hand, in several of the revisions he reiterates and expands upon his claims as to how sympathy takes us out of ourselves, as he puts it in one of them, “into the most distant countries and ages, and much beyond our own interest” (3.3.1.9). As I have already mentioned, the places of most (not all) of these amendments are noted by Nidditch in his revised edition of Selby-Bigge with an asterisk, and the amendments themselves (in their original form, with no spelling corrections) are detailed in the Textual Notes. The Nortons insert them into the main body of the text without giving any indication on the page that these passages were not in the Treatise as it was published.

There is of course good reason to think that had Hume had the opportunity to put out a second edition of the Treatise he would have added these passages to the text. The Nortons are confident that they have identified other changes Hume would have made, given the chance. Discussing the difference between “poetical enthusiasm” and “serious conviction” in 1.3.10, Hume draws attention to the phenomenological difference between tenuous belief in fictions and “the eternal established perswasions founded on memory and custom”. At least, that is what he does in the Selby-Bigge/Nidditch edition, because “eternal” is what is in all known copies of the published book. The Nortons, however, believe that Hume must have meant “external”, and so alter the text accordingly. They explain,

I find it remarkable that producers of a critical text for scholars felt able to allow themselves this kind of inconclusive speculation as the basis for a substantive change to the copy text. For one thing, while it is reasonably clear what the difference between an external and internal impression might be (roughly, the distinction between an impression caused by something outside us and one caused by something inside us), I have no idea what the distinction between an internal and external belief might be. For another thing, it is surely the editor's duty to hold back and leave for interpreters and critics the puzzle of what Hume might have meant by “eternal established perswasion”.

Here is another case. In 3.3.2.6 Hume discusses the way in which we come to have something like moral approval for the proud man, simply because of the way in which our capacity for sympathy prompts us to share the proud man's good opinion of himself. But Hume is tentative here:

The Nortons are convinced that Hume meant “places”, not “pleases”:

I do not see that talk of placing yourself in a conceit is any more (or less) intelligible than talk of pleasing yourself in one. In any case, this, again, does not seem to me to be the kind of alteration that an editor should be in the business of making. “Pleases” is not obviously nonsensical, and may, for all we can certainly tell, be what Hume wrote and meant. It is a cliché that there is a fine line between editing a text and giving an interpretation of it, but here the Nortons cross that line, and without needing to do so.

Another example of unnecessary interference with the text is provided by the decision to change Hume's reference at 1.4.6.6 to “the identity we ascribe to plants and vegetables” to “the identity we ascribe to plants and animals”, said by the Nortons, who follow a suggestion first made by Roland Hall, to be “supported by the context and by the discussion of Shaftesbury to which Hume refers” (652 n. 135). If “vegetables” is a slip, it is very unlikely to have been a slip on the compositor's part. The mistake was made by Hume himself—and should have been left intact, with a footnote explaining why it is surprising that Hume did not write “animals”. And here is a further example: the Nortons change the title of 3.3.5 from “Some farther reflections concerning the natural virtues” to “Some farther reflections concerning the natural abilities”. They say that this is “the only appropriate title”, and explain as follows:

“Virtues” in the title of 3.3.5 may have been a slip on Hume's part. But it may also have been an element of his attempt to undo the distinction between virtues and abilities. Again, “virtues” is not obviously a mistake, and therefore should not have been changed.

These alterations to the text that was published in 1739 and 1740 are of course not made silently. They are fully documented, and argued for. The regrettable thing is that the text itself bears no record of them. As already mentioned, the only footnotes to the text are Hume's own footnotes. This means that in order to keep abreast of what the editors have done to the text, one has to be continually moving back and forth between the text and Volume 2. I suspect that many users of this edition, as I have done, will end up making their own marks on the text to indicate the location and nature of the most significant textual alterations that have been made by the editors of this edition.

III

For these reasons, I am not convinced that, from the point of view of the scholar in search of an authoritative critical text of A Treatise of Human Nature, the Nortons' edition marks a decisive step forward from Selby-Bigge as modified by Nidditch. It was not a good idea to lay the text out in this way, with all editorial matter rigorously separated from the text. All the other volumes that have so far appeared in the Clarendon Hume edition—Tom L. Beauchamp's editions of the two Enquiries (1998 and 2000) and of the Dissertation on the Passions and the Natural History of Religion (2007)—effect the same separation of text from editorial matter. The questions that are raised by this edition of the Treatise pertain, then, to the overall character of Clarendon Hume project. The Clarendon edition is, presumably, meant for scholars, and scholars want to be able to see both where the textual cruxes are and all of the material relevant to resolving them. Hume, as Duncan Forbes remarked, is terrible campaign country. A clean text, almost totally devoid of editorial markings, makes the country even more treacherous than it need be. What the Hume scholar wants, surely, is to be given a clear sense of where the trouble spots are. He or she wants it to be made plain where there are questions about the accuracy of the compositors, where Hume might have made mistakes himself, and where there is evidence that Hume might have altered his text given the chance of a second edition. While this edition of the Treatise does tell the reader these things, and a great deal more besides, it does not make them anywhere near obvious enough.

For the moment, Oxford University Press is keeping the Selby-Bigge/Nidditch editions of both the Treatise and Enquiries in print. This is curious, but a good thing. Selby-Bigge/Nidditch has most of what most serious readers of Hume need. It is much smaller and lighter than the Clarendon edition, and considerably cheaper as well. It goes without saying that every proper library should buy the Clarendon edition of the Treatise. Generations of scholars will undoubtedly benefit from both the “Historical Account” and the “Editors' Annotations”. But the Clarendon Treatise, considered in its entirety, is a missed opportunity.