The illicit Procrustean logic of the Middleton attribution studies in these respects seriously undermines the studies’ validity. By including and excluding plays in and from the sample universe based on the plays’ internal mannerisms alone, the studies depend upon a series of dubious assumptions. The studies assume that an author’s style doesn’t change over time with respect to the selected norms. Thus they exclude Blurt from the sample although it would have been among Middleton’s earliest surviving works. They assume that the writer could not or would not vary his style in the selected norms depending upon the circumstances in which he was writing. And they assume that the appearance or non-appearance of the selected norms lay solely within the writer’s control.
All these are dangerous assumptions, especially here. Writing styles of professional writers do in fact change over time based on accumulated experience and evolving environmental cues. Indeed, parallel stylistic studies are often employed to help fix the chronologies of Elizabethan playwrights’ canons. Professional writers do in fact adapt their writing styles to the purposes at hand. Playwrights in particular. They put into the mouths of their characters not the words that they themselves would have spoken, but the words that would have been spoken by the characters. If I am wrong in these respects, can someone please show me a stylistic study proving that Titus Andronicus and The Tempest were written by the same man?
Any notion, lastly, that Elizabethan playwrights controlled their words’ ultimate expression is obviously incorrect. Many agents interceded between the playwrights and the expressions. In the collaborative world of Elizabethan play writing other dramatists may have revised or added to a play. The actors almost certainly added to, deleted, and otherwise changed the playwright’s words. The scribes who transcribed the playwright’s words surely changed the expressions according to their own predilections. And the compositors who set the playwright’s words into print brought to the task their printing house’s house styles. Many of those house styles would have affected expression of the Middletonian mannerisms so cherished by our studies’ authors.
The incorrectness of these assumptions doesn’t just vitiate the Middleton sample universe constructed by our studies’ authors. It also infects the studies’ application of Middleton mannerisms extracted from that universe to target plays. There the assumptions are repeated again. The studies assume that the styles of alternative possible authors did not change over time, that the alternative authors could not or would not have altered their styles in the target plays, and that the authors controlled their words’ ultimate expression. Thus, we are informed, Middletonian mannerisms appear more frequently in The Revenger’s Tragedy than in Tourneur’s undoubted Atheist’s Tragedy. The studies conceive that Tourneur was a playwright incapable of incorporating more characteristically Middletonian mannerisms in one play than another.
That conception is particularly treacherous in Tourneur’s case. We know almost nothing about Tourneur. He was a professional writer in a sense. He apparently spent most of his adult life as secretary to several of the realm’s nobles. Those positions almost certainly required him to draft documents in the voices of his employers. So he was accustomed to varying his personal style with the occasions at hand. He apparently turned to play writing only when he was between secretarial employments. Surely when he did he brought different styles and sensibilities to each individual project. And when he did take up a new play writing project he surely prepared for the project by reading and seeing then popular plays, such as those by Middleton.
We have, in Tourneur’s case, a sample universe of only one undisputed surviving play. We cannot rationally insist that he was bound to repeat the style of that one play in every other play he wrote. If he was, I’ll say again, the author of Titus Andronicus cannot have written The Tempest.
My second principal objection to the Middleton attribution studies concerns the methods by which the test mannerisms have been selected. Essentially all of the studies’ authors are Middleton partisans. They aim to show that Middleton wrote plays and parts of plays otherwise unattributed to him. They explain generally that they selected their test mannerisms because of those mannerisms’ frequency in Middleton plays. But they don’t say what other test mannerisms might have been used instead, and why they were not. The foundational flaw in this procedure should be obvious. No matter how well intentioned, our authors’ selection of mannerisms may have been influenced by their goal. If you were writing those studies, would you select mannerisms that tended to show your man didn’t write or participate in the target plays? Probably not.
In any event, frequency of occurrence in Middleton’s plays is not, standing alone, a valid basis for selection of test mannerisms. Its use will bias results in favor of Middleton’s authorship even apart from bias introduced by the studies’ authors. Simply as a matter of ordinary statistical variation, some plays and parts of plays not actually by Middleton will contain more of the frequent Middletonian mannerisms than the average number of those mannerisms in all plays. If you alight on those plays and parts of plays and say they must have been by Middleton, you will be wrong. If we set aside for the moment all other flaws in these studies, let’s indeed assume that more of the frequent Middletonian mannerisms appear in The Revenger’s Tragedy than in the average Elizabethan play. Does that mean the play was written by Middleton, or does it mean the play was one of those not by Middleton that by ordinary statistical variation must have exceeded the average? Absent the external evidence, we would have no basis to say.
A valid statistical study would not involve any selection of mannerisms. It would examine all mannerisms altogether—all that appear frequently in Middleton’s plays, all that appear infrequently, and, if they are genuine Elizabethan mannerisms, all that never appear. That information would be compared against corresponding information from the target plays. Then we would know the exact extent to which mannerisms adopted by Middleton, infrequently employed by him, and avoided by him altogether, appear and don’t appear, in the target plays. The statistical probability that a target play not by Middleton would in all these respects closely match the profiles of all the plays from a legitimate Middleton sample universe would be slight. As the studies now have it, we are very far from owning that information.