ACD, the unreliable narrator

[mightymads]

Just finished reading Dangerous Work: Diary of an Arctic Adventure, a reprint of ACD’s journal he kept during his voyage on a whaler. When he wrote about this journey first in an article for the Strand Magazine in 1897, and later in his memoirs called Memories and Adventures in 1924, he put in the following way:

“I went in the capacity of surgeon, but as I was only twenty years of age when I started, and as my knowledge of medicine was that of an average third year’s student, I have often thought that it was as well that there was no very serious call upon my services.”

In contrast, here’s an excerpt from his actual diary of that period:

“April 11th [1880] Sunday. A dark day in the ship’s cruise. Poor Andrew was very cheery and very much better in the morning, but he took some plum duff at dinner, and was taken worse. I went down at once, and he died within ten minutes in my arms literally. Poor old man. They were very kind to him forwards during his illness, and certainly I did my best for him.”

From the editor’s note :

“According to Dr. Robert Katz: “By this point, Milne probably suffered irreversible damage (infarction) of part of the intestine with perforation of the bowel and leakage of bowel contents into the peritoneal cavity (peritonitis). Most conditions like this require surgical intervention, clearly impossible for Conan Doyle to provide on a whaler in the middle of the ocean. Regardless of diagnosis, there was not much that he could have done for this poor fellow.”

So there actually was a serious case with a fatal outcome, and out at sea ACD was powerless to help the man in any way. It’s one of the many instances when the real events are quite different from what ACD recounts later. In turn, this proves that Watson could do (and did) exactly the same.

Comments

[senmut]

Good point!

I find myself wondering if, faced with something he could not have beaten, if he unconsciously mitigated it in his recall.

[mightymads]

Could be! Likewise, he glossed over his father's alcoholism, for example. He wrote about his family living from hand to mouth, "I grew up in a bracing atmosphere of poverty" or something like that. I wonder if he truly believed there were no atrocities towards the Boers by the British when he wrote about the Boer war (which brought him knighthood) or was it deliberate lying through his teeth.

[cimorene]

I was googling about the origin of the name "Sherlock" the other day and instead stumbled on a website of a guy who studied ACD's ouvre with stylography and believes that most of the first half of Holmes canon was written by ACD's first wife Louise and most of the second half by his second wife. It looked to me like he didn't have a null hypothesis or a control, which seems to make it impossible to actually draw any conclusions anyway even apart from questions of how plausible that hypothesis is; but I didn't want to do a deep dive into how stylography is meant to be done right away so I'm fuzzy on the details.

Anyway, however unlikely these conclusions, one of his pieces of evidence was that ACD was apparently a known fabulist, and he gathered a bunch of examples of that here, but without citations (presumably you have to buy his e-book for those).

[mightymads]

That’s a very unusual theory. Thanks for sharing! :)

[graycardinal]

That is a fascinating Web site, and there is more than enough in the downloadable PDF files under the Authorship tab to get at the meat of the theories (save that he promises a chapter I can't find on the collaborative aspects of A Study in Scarlet).

My two preliminary conclusions are these: (1) he's thorough enough in documenting his methodology -- and honest enough to admit the limitations of the data -- that I generally trust his analytical instincts, and (2) that while I'd want to look more closely at the straight Doyle biographies we have to evaluate some of his non-statistical comments, his data certainly supports (and explains) the long-held observation among Sherlockians that there are distinct and sometimes sharp differences between pre-FINA and post-Reichenbach Holmes.

Also, I would be interested in sanguinity's take on the mathematically complicated chapter of the stylometrics paper (one of the free downloadable PDFs on the Web site).

[cimorene]

Oh, really? I skimmed the Stylometrics article, and it seemed like the methodology involved dividing the writings into packages and then the analysis simply compares each story he analyses to the different packages to see which ones it's most like, without any mechanism to account for a null hypothesis (ie that the variation among the different packages is within normal parameters of things written by the same person at different stages of life). I would have expected some information about comparing the within-package to between-package variation, or data about expected levels of variation within an author's lifetime, but I didn't see anything like that. I thought perhaps that was part of the background of stylometry, so I tried reading about it in Wikipedia, but I couldn't find anything that seemed to point to where I could find that particular information easily.

[graycardinal]

From what I gather, the underlying premise of stylometrics is that there isn't a null hypothesis in the sense you suggest - rather, the usage of a given core word-set is nearly as unique as a fingerprint and doesn't tend to vary significantly over time. More specifically, a stylometrics analysis is not a study of vocabulary; most or all of the words under study are "function words" rather than "content words".

Between the discussion here and some of the comments in the discussion on The Boscombe Valley Mystery, I found myself sufficiently intrigued to buy Shadow Woman, the full ebook outlining Allen's case for Louise as creator of Sherlock Holmes (and later, for Jean Conan Doyle as her successor as Holmes' primary chronicler). I have gone very rapidly through a good deal of it, and will shortly post a preliminary review to this community. Meanwhile, I will stand by my comments above: I think the stylometric analysis is fascinating (and broadly credible; the parallel case involving the Federalist Papers is instructive, I think), and there are aspects of Allen's arguments that align disturbingly well with fandom's long-standing debates about the differences between pre- and post-Reichenbach Holmes.

[cimorene]

I thought of the before/after Return difference right away when I was skimming the paper!

If there's no null hypothesis involved that's very interesting. I did notice the bit about function words, but of course we all know that a writer's style can vary consciously depending on what they're writing, and I haven't seen the foundational, I presume, data on whether function words are independent of that. If it's essentially as you say then of course that explains it. I also had difficulty, not being a math person, skimming over the method section - I'm used to understanding the basic statistics kind of numbers you get in scientific papers and this is... very much not that. If I remember right, the numbers are about weighting each test case towards one of two bodies of work, so the numbers are just against each other. The explanation I took in didn't give me a quick way to evaluate the amount of divergence ie how different a 'Louise' or 'Jean' result is in some... visualizable way... from the main body of other ACD works.

I suppose I'd have to read the book too, huh?

[graycardinal]

I admit to skimming pretty fast through the more complex math myself - I am very much hoping that we can talk sanguinity, who is in fact a certified math geek, into giving us a simplified explanation of what's involved.

Per my latest comment in the main post further along - yes, I really recommend reading the book, because the stylometrics are not (as Allen readily acknowledges, and I count this a point in his favor) in themselves capable of attaching a name to an unknown individual metric result. For that one wants more traditional literary and forensic clues, and the book does a much better job of developing those than the "Five Best Arguments" condensation does.

[earthspirits]

I'm reading Shadow Woman, and there's a lot there about ACD as an unreliable narrator and outright fabricator. Many examples are given and it certainly demonstrates that this was a life-long pattern of his. More on this after I've finished reading the book.

[mightymads]

I’m definitely interested in a review of that book. ACD was a storyteller, and he told various stories about his life, presenting them as was most convenient to him. Glossing over his father’s alcoholism, embellishing his adventures in the Arctic, weaving an unlikely legend about a ten-year platonic courtship with Jean. It was his personal matter until he whitewashed the British in the Boer War and was knighted for it.