Any kind of inquiry, research or analysis has its inherent limits. As scholars who pursue knowledge through research, it is key to be aware of these limits. For example: in the case of stylometric analyses, we must know that we are leaving the comfort of certainties and entering the region of probability and likelihood. We can adequately label this the ‘epistemological limit’ of our project; however, it isn’t the only one.
In the case of The Dynamiter Project, other limits – historical and authorial – arise. Which are these exactly? Well, though Robert Louis Stevenson was a prolific author, Fanny Osbourne was not. Aside from co-authoring (allegedly) part of The Dynamiter, Fanny published no more than ten short stories in her lifetime and a handful of prefaces and introductions. Two memoirs have been published posthumously. In addition to this, an obscure theatre piece was also (again, allegedly) co-authored with Stevenson. This is our historical limit: Fanny did not write as much as Robert. Thus, the corpus of our analysis is mostly limited to the size of Fanny’s literary output.
What are the authorial limits? For starters, Fanny did not write novels. Other than her putative collaboration in The Dynamiter, all the fiction she wrote is in the form of short stories; and since it has been widely proven that generic differences outweigh authorial differences, our corpus should only include her short stories. What’s more, when we look upon The Dynamiter as a work in itself, there are further authorial limits: Fanny alleged to have written two chapters of the book. What does this mean? That when we use stylometry to analyse The Dynamiter in search of authorship, our irreducible textual unit must be the chapter. It follows that we can’t separate the novel in different samples (say, of 10,000 words each); nor can we combine or separate chapters in order to make up for larger portions of text that we want to look at; and, of course, given that stylometry is the use of computational methods based on statistical analyses, nor can we close-read chapters in order to try to distinguish which part of the chapter was written by whom. Simply put, the text samples would be too small, the evidence insufficient and it would lead to faulty conclusions.
There are still other limits, more intimate than the strictly historical and more subtle than the strictly authorial; for example, the invisible hand of editors and/or publishers – are they even traceable? And if we go further afield: How was the relationship of Fanny and RLS with regards to their writings? Did one edit the other? Who had the last word? Did they recommend changes to each other’s work? – And a long, ricocheting cetera.
A priori, these are the limits of our research. The success of our analysis greatly depends on how well we manage to move within this limited framework without ignoring its inherent constraints.