I spent the last evening / morning digitizing, analyzing, and graphing the data re: mutinies provided by A. Loez in the downloadable appendix to his breathtakingly comprehensive monograph Les refus de la guerre (Paris: Gallimard, 2010). Loez's book is an excellent example of French historical writing at--what I find to be, at least--its best. The monograph provides sensitive, critical appraisals of primary sources, many of which have either been overlooked by other historians or have recently become accessible because of the "boom" in published Great War témoins over the past decade. But more importantly, Loez's arguments--which I hope to synthesize in a later entry...--are underwritten by painstaking quantitative research. The "annexe" linked above provides detailed information about who the "mutineers" were, in terms of origin, class, occupation, and age.

I have three points to make after looking over and analyzing Loez's data:

Loez is much more fastidious than Rolland in his quantification of the number of men involved in any given mutiny. Rolland provides numbers for about 30% of the mutinies, while Loez provides numbers for about 87% of the mutinies. When I used Loez's data and calculated the 95% confidence interval for the total number of mutineers, I received a range of [21769, 33566] (you can see the spreadsheet here). Loez's data produce a range that is less than half of Rolland's. But Rolland and Loez list different mutinies as "mutinies," and their timelines of indiscipline--especially in the month of May--differ. The discrepancy between the two different ranges could be the result of different inputs during early May, or different size estimates in late May-early June. I plan to do a day-by-day comparison of their figures in order to find the roots of the difference.
Loez's work benefitted tremendously from Rolland's, to be sure. But more importantly, Loez's work benefitted from the explosion in publication and increased accessibility of first person narratives of the First World War in France between 2001-2009. For example, he relies on the témoins of René Clergeau (published 2001), Maurice Digo (published 2008), and Lucien Laby (published 2001), none of which appear in Rolland's bibliography. This is important because virtually all of the material we historians have relating to the mutinies comes from officer's reports, French military intelligence digests, and the incomparable Colonel Zopff, the French officer who organized, collated, and analyzed all the telegrams relating to indiscipline in 1917. While the archival materials produced by these officers, 2è bureau agents, and Zopff (especially Zopff... without him, there would be no real archival record of the mutinies at all) are invaluable, they nevertheless demonstrate a clear "observer bias" and view mutineers through the overlapping and distorting lenses of military rank and social class. Indeed, Mariot analyzed this problem with the archival materials in his essay "Pour compter les mutins...", and Loez's work made significant progress in correcting this methodological problem through his use of newly available materials.
Dynamic graphs produced from Rolland's and Loez's data show different geographical and temporal patterns in the spread of indiscipline. Yesterday, I posted a graph of Rolland's data. Compare that with a similar graph of Loez's data, which you can see here.

I hope to do a line-by-line comparison of Rolland and Loez in the next few days. I will post the data as soon as it is compiled.