(We've been looking at Pulitzer winner Stephen Sears' evidence handling and citational habits this week: please scroll down to Monday to start reading if you are joining us late.)

The story so far: we have these documents carelessly referred to as McClellan's "letters" to his wife; we know that they are not letters at all, but notes McClellan made from a source never seen; we know that his daughter augmented her father's notes from the same source, but we don't know the form of that augmentation; we know that McClellan's literary executor, W.C. Prime, transcribed both sets of notes, father's and daughter's, and created composite "letters" to Mary Ellen McClellan but we don't know his method of doing this; and we know that Stephen Sears, styling himself "the editor of the McClellan papers" has done the same. Additionally, we have Sears' word that Prime "severely censored" both sources and that he has restored Prime's cuts; that Prime made transcription errors, which Sears has corrected; and that many missing dates in these records have now been supplied by Sears.

My intention was to look at a long "letter" to identify any value added by Sears. I think, instead, that it would be more useful to look at a random group of "letters", 10 in all, for the same purpose. The documents chosen for this drill appear between pages 230 and 256 of the 1992 Da Capo edition of The Civil War Papers of George B. McClellan. Prime's corresponding "letters" are derived from an Easton Press facsimile of the first edition of McClellan's Own Story.

The first thing one notices in reading Sears' Civil War Papers is that the forms of address that would open any letter ("Dear Nellie," "Dearest Wife," etc.) are missing and that there is a uniform "To Mary Ellen McClellan." This should be enough to alert even a careless reader, one who skipped the introduction, that something odd is afoot. So it is probably a good thing – as long as we are sure this was a letter sent to her. Since we cannot actually be sure of that, some kind of alternate notation is needed, perhaps a comment, an asterisk, or a different way of starting the document.

The second thing we notice is that brackets are deployed in the dateline. Sometimes we see something like: [City name] July 19th [1862]. Sears has supplied the bracketed information and the document presumably bears the unbracketed data. This is the inference we are allowed to draw. No explanation from our editor, though.

I have not, after a cursory look, found any "letters" shown this way: [July 19, 1862], which would indicate that Sears supplied the whole date. Although he claims in the introduction to have supplied dates, what he mainly does is give us the missing year, which does not add a lot of value because the context gives us the year. Another service he supplies is to insert dates into the body of the "letter" where the "letter" is said to have been written after midnight. If, in a document bearing a time of 1:00 a.m., McClellan says "yesterday," Sears brackets a date two days earlier to show what "yesterday" refers to. Again, this is small help compared to the claim of having supplied missing dates.

We really need to know more about the dates attached to these notes; did they appear on McClellan's original note? Did they appear on May McClellan's material (much of which is dateless)? Were they added by Prime or Sears based on some surmise? What surmise? Did Sears accept a Prime date that was construed from the material? There are no indications from the self-styled editor of McClellan's papers. One would have to bypass this editor and go to the papers directly for an answer.

Sears is good enough to mark each "letter" AL copy (which means notes copied by McClellan) or AL copy; copy (which means the "letter" is a blend of George B. and May McClellan's notes). However, there is nothing in the text to show which component was sourced to which note taker. This is an egregious, student-level mistake that could have been corrected with the use of brackets and notes.

On to the 10 "letters." Half of them are sole-sourced ("AL copy") to McClellan's own notes, and half are sourced to both father and daughter ("AL copy; copy"). None of them have an entire date supplied by Sears; three have years supplied by Sears, the rest have both a year supplied along with the place the letter might have been written from.

Of the five father/daughter notes (starting on pages 235, 241, 243, 249, and 256 of Papers), four include additions to Prime's text and one does not. Aside from the additions, there are no substantive changes to Prime's text by Sears.

On the matter of transcription errors, Sears appears to have made only one fix. In one document about siege work (starting on page 249 of Sears' book), Sears renders a word "boyau" where Prime rendered it as "tuyau." One meaning of the French word tuyau, aside from pipe or tip, is "un canal de communication." A boyau is a zig-zagging trench forming a path (channel) of communication. There is not a lot of value in this transcript "correction."

Sears has again done well to restore some of Prime's deletions. Here are some extreme examples of Prime's redactions from one document ostensibly dated April 19 [1862]:

* "I have a little over 100,000 men, including Franklin's Division"
* "- the more there are in Yorktown, the more decisive will the results will be"
* "and am quite confident that with God's blessing I shall utterly defeat them."
* "Wade" was restored to the comment "Never mind what such people as … say"
* "I telegraphed the Presdt. last night requesting that McDowell might *not* again be assigned to duty with me."

Looking at the five stitched-together father/daughter documents: the first shows no Prime deletions restored by Sears; the second shows one restoration ("and then woe betide the guilty ones."); the third shows three restorations; the fourth was analyzed above; and the fifth shows one deletion restored by Sears and one deletion made by Sears from Prime (where Prime explains a reference to "Kentuck" as being to a horse; Sears lets us wonder about Kentuck).

I leave it to the reader to decide how censorious Prime's omissions are.

We have a collection of "letters" with dates of mysterious provenance, abstracted by two note takers and then stitched together using methods unknown. Shall we use these "letters" to build and validate timelines? To convey McClellan's moods at any exact moment? Or should they be handled with caution until a bona fide editor of McClellan's papers can go through the materials and create a properly annotated record?

A few words in Sears' favor tomorrow.