Out of curiosity, I tried using Voyant to roughly test whether Ulysses complies with Benford’s Law. I did searches for “one” through “nine,” and then graphed how often they appear, as well as their deviation from what Benford’s Law would lead one to expect. I was hoping to discover that Joyce had used a particular number unusually often, but sadly no such outliers appeared. Still, it’s interesting to note that this weird statistical law had power even over someone as original as Joyce, especially since deliberately invented numbers often violate Benford’s law (hence its use in fraud detection).

This was my data set, taken from Voyant. The Error column was calculated as |Percentage-Benford|/100

Here are my graphs. The error is highest for one and two, perhaps in part because “one” has meanings beyond its usage as a number. In the second graph, the blue scatterplot is the frequency of each number out of the total data set, and the orange marks are the expected distribution based on Benford’s law.

One other interesting tidbit I turned up while tinkering with Voyant is that numerals (1 as opposed to “one”) rarely show up except near the very end of the book:

I’m kind of curious to know why this is.