Jonathan Townsend wrote:Kindly turn to page 72 of the article to the statistical analysis of the text itself.
Unfortunately, the article by Wiseman and Holmes represents more of a start of an investigation, than the end of one. The graph on p. 72 of the Feb 2011 issue is difficult to understand, and I don't think any real conclusions about authorship can be drawn from it. The authors assert that the data presented in their Fig 1 shows that Sec 3 and part of Sec 1 are more stylistically similar to each other than to the rest of the book, and thus are likely to have been written by the same author. But Fig 2 indicates that the same part of Sec 1 is more similar to a third author (Gerritt Evans) than it is to the Sec 3 of Erdnase (which is more similar to the writings of Edwin Sachs than to the other parts of Erdnase). So, is Sec 3 (the card tricks) written by the same author as the remainder of the book; or by Edwin Sachs; or by the person who wrote Sec 1a (who may or may not have been Evans, or copied from him)?
And how close together would Sec 1a/3 have to be to the rest of the book stylistically to indicate that the author was the same person? Wiseman and Holmes put quantitative values on their stylistic distance measurements, but don't tell us at what point the numbers get so big that they indicate different authors.
Where is their control analysis? What if you subjected a book with a single, known author, like one of Richard Kaufman's books (or perhaps someone with a more distinctive voice, like Racherbaumer) to this sort of analysis - would the results show similar variations in style? How big is the "spread" of a talented writer of a book of similar size and topic - more or less than that shown by the analysis of Erdnase?
Further, I would submit that the way they have divided the text into smaller groups, particularly within Sec 3, skews the results. Sec 3 has 3 distinct authorial voices:
a. Patter -- much more flowery language than elsewhere in the book. Within quote marks.
b. Straightforward expository text (such as that between the words "Card Tricks" on p 171 of most standard editions and "Patter and execution" on p 172).
c. Direct instruction -- commands within parentheses in the Card Tricks section. Many articles ("a", "an", "the") are omitted here, and that would corrupt any statistical analysis of the frequency of small words. (This omission is a stylistic preference exercised by the author; and not due to the fact that one author uses the words with a different frequency than another).
Note, for example, that the Patter of the card tricks is the only place in the book where the author uses the first person singular pronoun "I" everywhere else he uses the editorial "we". Again, this should be accounted for in any statistical analysis of small word frequencies
Wiseman and Holmes' failure to segregate the text by authorial voice in the Card Tricks section, as opposed to separating it into contiguous blocks of text, has perturbed their results.
And despite these critiques (which I've discussed with the authors when the article came out), I think this is an important analysis and needed to be done. EATCT
should be compared to itself and to other relevant works with as many tools as possible, and by as many investigators as possible.