DID DEFOE WRITE MOLL FLANDERS AND ROXANA?" asks Ashley Marshall, in an acute and well-reasoned essay in the present issue of Philological Quarterly. It is a good question, and--maybe rightly?--Marshall thinks the answer we give about these famous novels in our writings about the attribution of works to Defoe is not satisfactory. We present these works as having a secure place in the Defoe canon, even though the evidence we bring forward for his authorship of them is, by our own admission, somewhat problematical. In Marshall's view, since no good case has ever been put forward, they cannot be regarded as "definitively" by Defoe. Scholars, including ourselves, ought to face up to this problem, and should in future present these novels as only "dubiously attributed to Defoe, and preferably with special emphasis on the uncertainty."
Defoe attribution is notoriously an extremely knotty problem, not really paralleled in the case of any other classic English writer, and the forming of bibliographical principles in regard to it calls for a certain understanding of its strange history. Attribution of works to Defoe has gone on for three centuries, and indeed is still going on; and to complicate matters further, many new attributions come later to be dropped, for one reason or another. Again, our confidence in any given ascription may depend on our estimate of the character and skill of the person who first made it. (1)
That said, it follows that, with Defoe, there is a particular need for strict principles of author-attribution, and before coming to the particular case of the novels, we would like here to quote the general principles we put forward in our Critical Bibliography of Daniel Defoe:
1. In arguing for an ascription, one should not "forge chains" of attribution, that is to say base any part of one's argument on some other merely probable attribution, but should draw solely upon works indisputably by the author in question.
2. If making a new ascription, one should always explain one's reasons.
3. One should not regard the fact that an ascription seems plausible--i.e, compatible with the author's known style or interests, etc.--as in itself sufficient reason for making it; nor should one be tempted to include a work in the canon "provisionally," i.e. until some better candidate for authorship appears. (The reason, in both cases, is that inclusion of a work in the canon causes a qualitative change in its status, transforming the way in which later scholars are expected to regard it. It is infinitely harder to get a work out of the canon than to put it there in the first place.) (2)
In attempting to set out a new bibliography of Defoe's works following these principles, we thought it necessary to distinguish clearly between attributions to Defoe which could be regarded as "certain" and those which could only be regarded as "probable," and so we divided items accordingly, with "probable" attributions being identified by a "(P)" added to their running numbers. It is worth summarizing briefly the reasoning by which we came to identify works as "certainly" or "probably" by Defoe, highlighting in particular our understanding of the different roles of "external" and "internal" evidence.
Some of the reasons why we can regard a work as certainly being by Defoe are quite obvious. For example, we may have a manuscript of the work in his handwriting, he may have signed it with his own name or initials, or he may have included it in the True Collection of the Writings of the True Born English-man (1703) or in A Second Volume of the Writings of the Author of the True-Born Englishman (1705). In other cases the reasoning will involve inference, or will depend on a convergence of external and internal evidence. On the vexed and complicated question of the relative rights of external versus internal evidence, our view has been that external evidence has a kind of logical priority over internal evidence. …