Above are two images of the Hinman Collator currently residing in the Memorial Library at the University of Wisconsin, another optical collating device that uses visual comparisons to highlight minute differences between seemingly identical versions of the same text. Hinman used the device in his landmark survey of Shakespeare’s First Folio; it allows the user — here, paper conservator Theresa Smith of Harvard — to “see” differences between two items by merging them in a single image. Areas of difference appear as a kind of grey area — a more subtle effect, perhaps, than the hovering text that is produced by the Lindstrand Comparator discussed in the previous post. The device also allows you to toggle between the two editions you are looking at, making subtle differences standout immediately. Prior to creating this device, Hinman had worked in military intelligence comparing pre- and post-bombing aerial photographs: the collator is thus one of several adaptations of military skills and technologies for literary analysis.
Smith and her collaborator, Daniel Selcer (Duquesne), gave a fascinating paper at Wisconsin two weeks ago which dealt with differences among Facsimile editions of Copernicus’ De revolutionibus. Here Smith is examining two facsimiles of De revolutionibus with divergent diagrams of the center of Copernicus’ universe (the sun), which in one edition appears as a circular outline and in another as a solid sphere. The Hinman revealed the differences in the center of the diagram immediately, but it also revealed other areas of discoloration and streaking which suggested that the underlying manuscript (closely guarded in Poland) might not be adequately represented by the facsimile edition. One of the interesting points of Smith and Selcer’s paper was that you must treat facsimiles as artifacts in their own right; they are not always indexical transcriptions of an original, but in certain crucial ways iconic: the technologies used in producing the facsimile proper introduce artifactual effects that make the facsimile a likeness rather than an absolute trace-copy of the original.
This device and the one I wrote about in the previous post are technologies for the identification of local variants, which is to say, variants that occur in one place on the page. The search for such variants has been crucial in the history of textual scholarship. Hinman, for example, was able to deduce from variants among surviving First Folios the order in which the forms were printed and their various states of correction. This, in turn, led him to reconstruct an “ideal” Folio which he hypothesized contained the latest or most corrected state of the book. (No single Folio contained all of the corrections, since as Hinman argued, “every copy of the finished book shows a mixture of early and late states of the text that is peculiar to it alone.”) While presenting some findings from the work I have been doing with Docuscope at Loyola earlier this month, Peter Shillingsburg made a very important point in connection with this idea: when you are interested in broad patterns, it appears that it does not matter what edition you are working with. This may not seem like a big deal to some readers, but for Shakespeareans, it matters quite a lot which edition of the plays you are using to argue your case. There are substantive differences between different printed editions of the plays, and in some cases, individual words or phrases — for example, Hamlet’s “dram of eale” — can be emended to produce significantly different readings of a particular passage.
So, would the findings I come up with using Docuscope change significantly if I switched from the Moby Shakespeare (a nineteenth-century edition produced at Cambridge) to the Oxford or Norton? Yes and no. No in the sense that I am interested in a form of variation that is not accounted for in the kind of textual scholarship practiced by Hinman. In looking for genre at the level of the sentence, I am looking for diffused rather than local variation: a kind of patterned deviation from the mean that occurs across the entire body of the text rather than at one crucial intersection. So if Docuscope were able suddenly to read the word “evil” once an editor had emended “eale” in Hamlet’s speech, there would be a slight uptick in one of its counting categories (“Negative Values”). But that uptick would probably not fundamentally alter the patterns being discriminated across the entire corpus of Shakespeare’s works. There are some statistical procedures which could register slight upticks in categories that are not used frequently, however, correlating them with others that are exercised all the time. What if there is a correlation between even slight changes in Shakespeare’s use of “Negative Values” tokens and the much more common “Description” tokens we explored in the histories? What if, in other words, a “dash of x” matters sometimes?
I think it is important to recognize this category of the “dash” or “pinch” in looking for broader patterns of variation in large populations of texts, because it sits somewhere between the “crux” local variants like “eale” and the global variation we see in uses of “the” or concrete descriptive nouns. Because Docuscope is looking at things sub species aeternitatis, as is were, we cannot say that it matters when such dashwords are used. Time sensitivity in use, immediate context: these are all crucial features that help us understand local variants. (And we are quite attracted to local variants, as the history of literary criticism and close reading shows.) Dashwords are different: rare, like an eclipse, but nevertheless part of a globally diffused pattern.