...and counting (some musings)
Dad the Emeritus Historian of Graeco-Roman Egypt has frequently lectured me on my aversion to representative samples. When I research a project, I do my best to read everything on my bookshelves, in the state, in the next state, across the pond, and so forth--partly because I'm a fast reader, partly because I'm always afraid that the book left unread will be the book that causes my project to implode in a puff of foxed pages. Strictly speaking, such implosions are highly unlikely: if, say, fifty didactic novels adhere to one pattern with only minor deviations, and one didactic novel goes wildly off-message, then the critic just has one oddball to set against a clear norm. Now, it may be interesting that the oddity is, well, odd--however did that happen, the critic wonders--but that one lonesome novel's existence doesn't undermine any larger points about what didactic novelists typically say.
I started thinking about numbers while reading Pamela Regis' book on the romance novel. Regis, looking at how feminist critics survey the genre, complains that "[t]he practice of these critics reflects the widespread assumption that a very small sample of contemporary works can yield conclusions applicable to a whole range of romance novels" [1]. As Regis (and just about everyone else who writes on this genre) notes, there are literally thousands of romances out and about--so what constitutes a representative sample? I count thirty-seven novels in Regis' own bibliography, including a scant handful of texts published prior to the nineteenth century. This number is supposedly more representative in breadth as well as quantity; that is, Regis offers the reader samples of different kinds of romance (Western, SF, erotic, and so forth). And supposedly, we need "more" books in the set because romances are not strictly formulaic--meaning that romances cannot be simply "slotted" into the same historical spaces. If romances are really not formulaic, though, can you actually stop with thirty-seven novels? Does it matter that there are no African-American novelists in Regis' sample set, or gay novelists, or even male writers working under pseudonyms? (Regis does spend time talking about Richardson, Forster, and Trollope, but the men drop out once she begins analyzing twentieth-century bestsellers.)
The less formulaic the books in question, the more the critic has to read. Eventually, the critic stops when the books simply add quantity to the argument instead of nuance. At the same time, though, there's a longstanding tradition of doing literary history through close readings of exemplary texts--the case-study approach. Regis' own chapters on pre-twentieth-century romances actually fit that bill, despite her complaints about coverage elsewhere. That approach, for all its persistence, has actually been under fire for most of the twentieth century; it comes down to what the literary historian is looking for in his or her kaleidoscope. In my (just copyedited, hooray) article on Emily Sarah Holt, for example, I note that Andrew Sanders' The Victorian Historical Novel 1840-1880 (1979) is a fine survey, but that the book buries virtually everything that looks like a didactic novel. (Cardinals Wiseman and Newman manage to sneak in.) According to one twist of the kaleidoscope, omitting all the didactic novelists is just fine; according to another, the image suddenly fills with popular novelists of whom the twenty-first century public has never heard. It all depends on the priorities the critic brings to bear on thousands of texts...
[1] Pamela Regis, A Natural History of the Romance Novel (Philadelphia: University of Pennsylvania Press, 2003), 6.