A Widow and Her Soldier: The Case of the Pickett Letters

  1. 1. David I. Holmes

    Department of Mathematics and Statistics - The College of New Jersey

Work text
This plain text was ingested for the purpose of full-text search, not to preserve original formatting or readability. For the most complete copy, refer to the original conference program.


This research project concerns the authorship of the so-called "Pickett Letters" from the Civil War era. In 1913 LaSalle Corbell Pickett, the wife of Confederate General George Pickett, published The Heart of a Soldier, presented as Pickett’s letters to his fiancée, some of them dated from the field of Gettysburg. She told readers that these wartime missives from her famous long-deceased husband had "lain locked away from the world, the lines fading upon the yellowed pages, their every word enshrined in the heart of the noble woman to whom they were written". This book was re-printed in 1995. Numerous historians cited them, and excerpts appeared in anthologies and collections. Michael Shaara mined them for hie Pulitzer Prize winning novel The Killer Angels, and Ken Burns highlighted them in his 1990 television documentary "The Civil War".

The letters of this volume however, at least those dealing with Gettysburg, fall under considerable suspicion. The first published discussion of the letters was a two-page appendix in George Stewart’s book Pickett’s Charge (1959) in which Stewart cites several reasons for his suspicions concerning their authenticity, in particular the facts that the writer knows a great deal more than Pickett could possibly have known at the time, and that the style of writing is effusively sentimental. LaSalle Pickett’s strongest defender has been Glenn Tucker, author of High Tide at Gettysburg (1958) who finds it significant that LaSalle published the letters at a time when many of her husband’s veterans were still alive and could have pointed out glaring errors. More recently Gary Gallagher (Lee and his Generals in War and Memory, 1998) argues that LaSalle Pickett concocted the letters "relying on plagiarism and her own romantic imagination". Gallagher argues that LaSalle’s motivation for fabricating the letters was both financial and the wish to bolster with her husband’s own testimony the intensely romantic portrait of him which existed. When Gallagher compared eleven unpublished handwritten letters in the Arthur Crew Inman Papers at Brown University with the published ones, he felt he had found undeniable evidence of LaSalle’s subterfuge. The handwritten letters, apparently in George’s actual handwriting, did not match the published ones in content or tone. Gallagher’s article, it seemed, would end the debate.

Gallagher’s article was persuasive, but some lingering questions remained. Could it be that when George wrote personal letters to his wife, his tone altered? This did not seem too far fetched considering that Victorian couples like the Picketts often communicated with one another in a more intimate way than they would with other people’s lives. This research project offers a fresh perspective on the problem of the authenticity of the "Pickett Letters" through stylometry, the statistical analysis of literary style.

Sampling and Textual Preparation

In order to attribute authorship of the "Pickett Letters", a large number of control texts are necessary, preferably in the same genre. Samples of letters written home by other Confederate officers were gathered, also samples of letters written home by Privates. For George Pickett, the largest collection of his genuine writings is found in his military reports published in the Official Records. Textual samples were taken from such reports. In addition, prewar and postwar letters written by Pickett to friends and family, some in his own hand and some transcribed from a typescript collection in the Virginia Historical Society, were added to the mix. Pickett’s widow produced numerous books, articles and short stories after the General’s death in 1875. Textual samples were taken from these works, and additional samples were obtained from collections at various university libraries.

Letters from The Heart of a Soldier, the prime concern of this study, were put into machine-readable form and subdivided into seven samples of text. In addition, the eleven handwritten letters in the Arthur Crew Inman Papers were transcribed and prepared in machine-readable form before being subdivided into three textual samples. The target length for each sample was 3,000 words.

Stylometric Analysis

Using frequencies of large sets of non-contextual function words (a proven methodology in stylometry), and multivariate statistical techniques such as principal components analysis and cluster analysis, the methodology was first tested on the known material. The letters home from Generals and Privates separated out by author, intriguingly the variation between Generals and Privates being greater than the variation within these two categories. With George’s known works, the personal and the official material appeared broadly similar in their usage of function words, leading to the identification of a "George Pickett" style or wordprint. LaSalle Corbell Pickett, by contrast, displayed two distinct styles, one emerging in her published works and the other in her private letters. Together, three distinctive styles had emerged for our two candidates.

For the attributional phase, the statistical technique of discriminant analysis was used. The textual samples from George Pickett were recombined and then split into 12 smaller samples. Similarly the textual samples from LaSalle were recombined and then subdivided into new samples of approximately 1,000 words each. With 26 new samples drawn from three known and distinct styles, the number of variables was reduced to the 25 most frequently occurring function words in the combined corpus.

The textual samples comprising the eleven handwritten letters in the Arthur Crew Inman papers were placed by the discriminant analysis into the George Pickett style group. Six of the seven textual samples from The Heart of a Soldier were placed in the "LaSalle Published" style group. The remaining sample, consisting of letters supposedly penned early in the war, were classified into the "George" group. Perhaps these particular letters contain cores of authentic George correspondence, enough to have them attributed to George by the discriminant analysis. In conclusion, stylometry has shown a clear difference between the styles of the letters in The Heart of a Soldier and the eleven handwritten letters, a difference confirming Gallagher’s suspicions of LaSalle’s subterfuge.

If this content appears in violation of your intellectual property rights, or you see errors or omissions, please reach out to Scott B. Weingart to discuss removing or amending the materials.

Conference Info

In review


Hosted at New York University

New York, NY, United States

July 13, 2001 - July 16, 2001

94 works by 167 authors indexed

Series: ACH/ICCH (21), ALLC/EADH (28), ACH/ALLC (13)

Organizers: ACH, ALLC