University of Virginia Library

2. Practical and Conceptual Issues

The methodology adopted by Furbank and Owens is obviously vastly more conservative than those employed by earlier Defoe bibliographers, and their conservatism is a good thing. But, as some reviewers have noticed, there is a significant problem with Furbank and Owens's attributional policies, as well as a variety of inconsistencies in the execution of those policies. I want here briefly

to survey a few of these issues, because understanding them is important to making further progress in the realm of Defoe bibliography. In section 3, I will turn to what seems to be the most crucial point: the Critical Bibliography presents the canon shorn of its most obviously implausible or illogical attributions, but its firm demarcations obscure important gradations. What a tabular representation can do beyond that is to make clear the distinctions among the different kinds of evidence (and different degrees of probability) on which the reduced canon is based. The necessity of doing this is a point to which we will return.

First, the one serious reservation about Furbank and Owens's policy on evidence. What they have to say about internal evidence is essentially sound, but their notion of external evidence is disconcerting.10 Any attribution made before 1790 is deemed "contemporary"—a loose definition of contemporaneity. (By this standard, James Joyce [d. 1941] would have been a living memory as late as the year 2000.) That some posthumous references should be considered valid is clear; in the 1730s, and possibly in the 1740s, Defoe was still very much a "living memory." There is no inherently right cutoff date. But why not stop at (say) the list of writings included in the 1753 Shiels/Cibber "life" of Defoe? Not much is attributed for the first time to Defoe in the years between 1753 and the mid-1770s, by which point we are nearly half a century beyond his death. Why did Furbank and Owens extend their notion of "contemporary" to include the 1770s and 1780s? Probably this was not an arbitrary decision, but one having to do with the fact that the only external evidence we have for linking Defoe to most of his "novels" dates from the years between 1775 and 1787. Of the fiction only Robinson Crusoe, Col. Jacque, and A Journal of a Plague Year had been publicly attributed to him as of 1753. The situation remained essentially unchanged for another twenty years. Captain Singleton is advertised as Defoe's in 1767,11 and in the 1770s and 1780s, a rogue bookseller named Francis Noble assigned to Defoe mangled versions of Moll Flanders and Roxana, as well as Memoirs of a Cavalier (the last mostly faithful to the original). After someone else credited Defoe with A New Voyage in 1786, Noble promptly issued a three-volume edition of that work under Defoe's name. I have discussed Noble and his connection to Defoe's fiction at length elsewhere.12 At the moment what matters is that Noble is all the external evidence we have for attributing two thirds of "Defoe's" most important fiction to him.

Concerns about what constitutes contemporaneity in external evidence notwithstanding, we need to ask just how consistently Furbank and Owens apply the criteria they promulgated for attribution. The answer is less reliably than one might hope, though in fairness one must admit that the nature and scope of the

material make perfect consistency impossible. Furbank and Owens admit that external evidence is not necessarily proof of anything, which is why they retain some of the items attributed to Defoe by Abel Boyer and Robert Wodrow, for example, while excluding others.13 Edmund Curll, we are told in De-Attributions, is not a sufficiently reliable source. Wodrow is likewise untrustworthy, because his "methods of reasoning" are faulty: he lumps groups of works together and assigns them to Defoe for no reason except that each refers to the others.14 About the unreliability of those sources Furbank and Owens are entirely correct—but we might compare the treatment of De-Attributions no. 130 to the Critical Bibliography entries on Moll and Roxana. No. 130 is The State of the Excise after the Union, about which Furbank and Owens have this to say: "Given the date of Wodrow's ascriptions ... they must count as external evidence." They are, however, "inclined" to doubt the attribution. Of Moll and Roxana, they conclude, "The attribution by Noble is evidently questionable, but given its date it must count as external evidence" (CB, 200). In the one case, the unreliability of the source overturns an otherwise plausible attribution. In the other, the unreliability of the source is trumped by their sense of the plausibility (or perhaps the desirability?) of the attribution. The State of the Excise after the Union should not be re-attributed on current evidence, but by the standards of the Critical Bibliography it seems more aptly described as "probable." Alternatively (and more safely), unreliable "contemporary" attributors might simply be disregarded—meaning that the canon shrinks yet further, and we lose Moll and Roxana. The fact that Furbank and Owens find some "contemporary" sources more reliable than others is itself unobjectionable. What is unclear, however, is why Noble's testimony (for instance) is more inherently reliable than Wodrow's or Curll's.

The evaluation of internal evidence is likewise sometimes disquieting. Several items in the De-Attributions volume are removed from the canon because one piece of stylistic evidence is in itself inadequate. For example: "Defoe was unquestionably fond of this anecdote, but on its own ..." (no. 324a), they say, and no. 334 includes a remark that is "very close to a line" from The Mock Mourners, but "the allusion, if it is one, could admit of various explanations."15 That one suggestive internal tidbit does not constitute a sufficient basis for attribution seems a sound conclusion. But several works labeled "probable" in the Critical

Bibliography are included on the grounds that they have at least two suggestive internal tidbits, each of which could presumably "admit of various explanations." If one allusion is not enough and a single close quotation is not enough, then do we really want to accept a work because it has one of each? At some point an accumulation of suggestive tidbits might perhaps seem persuasive, but where we have two or three and no reliable corroborating outside source, just how solid is the attribution? Alternative explanations need to be considered, as in the case of A New Test of the Sence of the Nation (112(P))—regarded as "probable" because two of its passages echo one issue of the Review. But the Review did have readers, some of them also writers, who might well have been influenced by its argumentation and rhetoric. More broadly, the evidence for including a work as "probable" in the Critical Bibliography is occasionally similar to the evidence that is ruled inadequate in De-Attributions; an allusion or discussion or quotation might render a work "probable" or it might not. Furbank and Owens rightly argue that each case must be evaluated on its own merits, which means arriving at different conclusions for cases with differing mixes of evidence—but in each instance the calculus that yields the result needs to be laid out in full for the reader.

Part of Furbank and Owens's procedure is to ask whether a work for which we have external evidence of Defoe's is a plausible attribution. Broadly speaking, this is a sensible policy. We want to know whether something ascribed to Defoe by a contemporary could have been written by him. One obvious problem with trying to determine Defovianness is that the works we can be certain he wrote are relatively few in number, and many of them date from early in his career. A second problem is that we need to allow for fits of temper and changes of mind. A work that does not square with what Defoe said elsewhere is not necessarily an implausible attribution. To say that Defoe "sometimes wrote badly, but ... was not capable of certain kinds of badness, as of certain kinds of excellence" is problematic.16 Highly skilled and competent people have bad days. Given all the uncertainties surrounding Defoe, "assume nothing" is probably a good guideline for proceeding.

A core principle for Furbank and Owens is that one should never make a new attribution without explaining one's reasons. But if one is undertaking a drastic reconsideration of the old canon, then one ought to take the opportunity to provide reasons for the retention of every item. For most of the "novels," no such case is made. We are duly told about Gildon's attribution of Crusoe to Defoe. The sequels, Farther Adventures and Serious Reflections, are presumably in the Critical Bibliography because Defoe scholars seem always to have assumed that the author of the first part is necessarily the author of what follows.17 This is not the place to test that presumption, but the point is that Furbank and Owens did not explain the nature of the evidence that justified the preservation of the sequels in the "definite" category. Unfortunately, the same could be said of Captain Singleton, Moll Flanders, A Journal of the Plague Year, Col. Jacque, Roxana, and A New Voyage.

The only significant work of fiction (other than Crusoe) for which Furbank and Owens give anything beyond the shaky external evidence is Memoirs of a Cavalier,

first attributed to Defoe by Noble in 1784. The additional substantiation consists of three facts: on several occasions Defoe claimed "to have in his possession a manuscript by an English gentleman who served under Gustavus Adolphus"; a reference in Memoirs is "reminiscent" of a passage in the Review for 15 July 1704 (that is, sixteen years earlier); and in general the "overall plan of the work ... strikes one as very Defoean" (194). The case for inclusion is not exactly compelling, though it is more than what we are told about Moll, Roxana, and some others. We should also be aware of the fact that, though the evidence for Memoirs of a Cavalier, Moll Flanders, Roxana, and Captain Singleton is functionally identical, only the first is downgraded to "probable." That they shrink from attaching the dreaded "P" to two of the three principal novels is understandable—but if one is going to trust Noble, then one really has to trust him. We need to be applying the same criteria (at least where the source is the same) regardless of how canonical the item in question is.

At this point I want to move beyond the particularities of inconsistencies and oddities, and to consider the broader utility of the Critical Bibliography. What Furbank and Owens have accomplished in this and in their two previous books is phenomenal. They have effected a radical reduction of a much-cherished but much-swollen canon. What they did needed to be done, and if scholars do finally commit themselves to a smaller canon more scrupulously justified, that will be thanks largely to Furbank and Owens's pioneering efforts. The question is not whether their Bibliography has value—it is extremely valuable—but what we can (and cannot) safely use it for, and how it can be improved upon.

Let us remember the big picture. Almost all of the items Moore attributed to Defoe were published anonymously. In a huge number of cases, therefore, we have to depend upon something other than self-identification for the attribution. Furbank and Owens wisely insist upon external rather than internal evidence if a work is to be considered "certain," but extrinsic evidence varies wildly in quality, and most of it is proof of nothing. For that reason, they test contemporary attribution against content and style—also reasonable, though we need to remember that the apparent plausibility of subject and approach is likewise far from probative. In the Critical Bibliography, Furbank and Owens confer "certainty" and "probability" for a variety of reasons, arriving at their verdict by weighing different kinds of evidence, and then sort everything into only two categories. This division into "definite" and "likely" is potentially misleading. An item that is ascribed to Defoe by an enemy in his lifetime and that is stylistically plausible, Furbank and Owens label "certain"—but such an item is nothing like as certain as those of which he unambiguously acknowledged authorship.

What we need are more categories—that is point one. But we should remember that these categories cannot really reflect degrees of probability. What I am suggesting, in other words, is not that we re-categorize works into "definite," "highly probable," "somewhat probable," "slightly less probable," and so on. We need a layout and set of groupings that make clear the nature of the attributional evidence, but to assume that we can calculate the percentage of probability is foolish and dangerous. We can take the nature of the dubiety into account, but not the degree. For example, we can distinguish between an antagonistic or disreputable source and a sympathetic and reputable one. If the sainted Jacob Tonson

had published something as Defoe's, we might be inclined to treat the ascription as reliable external evidence; if the sleazy Edmund Curll made the same attribution, we would be much more skeptical. We cannot afford not to distinguish among degrees of probability—but neither can we sanely pretend that the differences are measurable. Tonson could be wrong and Curll could be right. A similar limitation applies to internal evidence. A work that has a dozen parallels to Defoe's known writings might not be his, and one with only a single parallel might be. How does one compare three weak pieces of stylistic "evidence" as against one more potent piece? We can say that a particular attribution is based on "x" and "y." What we cannot do is add "x" and "y" and arrive at a verdict of ten per cent probable or ninety per cent probable.

The problem we are dealing with here is twofold. First, we have no way of judging how likely most of the attributions are. Rogue booksellers and malicious rivals are not ideal sources, misinformation abounds, several items were attributed to Defoe but also to someone else, and Defoe's denials of authorship are surely sometimes but probably not always disingenuous. This is unfortunate, frustrating, and unfixable. The evidence is the evidence. The second problem is one we can address and it will be the subject of the next section.