Historical mysteries · The Bixby letter
In November 1864 a Boston widow received a letter of condolence of 139 words signed by Abraham Lincoln. It became one of the most often reprinted of his writings. The manuscript has never been found, and for more than a century it has been said that his young secretary John Hay wrote it. This report reads the whole printed record, reruns the one statistical study of the wording, and ends with a narrow lean to Lincoln.
In September 1864 a widow named Lydia Bixby came to the Boston office of William Schouler, the Adjutant General of Massachusetts, the officer who kept the state's military records, and showed him five letters. Each was from an army company commander, and each reported the death of one of her sons. Schouler told the story to the governor, John A. Andrew, who sent it on to the War Department in Washington with the request that the President himself write to her. On 21 November a letter of 139 words, dated from the Executive Mansion (as the White House was then called) and signed "A. Lincoln", was sent to Boston. Schouler put it into Mrs Bixby's hands on the morning of Friday 25 November, and the Boston Evening Transcript, an evening newspaper, printed it the same day.
The letter is short enough to quote in full, and section 1 does. It tells her that the President has been shown "a statement of the Adjutant General of Massachusetts, that you are the mother of five sons who have died gloriously on the field of battle", admits "how weak and fruitless must be any words of mine" against such a loss, and offers "the thanks of the Republic they died to save" and "the solemn pride that must be yours, to have laid so costly a sacrifice upon the altar of Freedom". Other newspapers reprinted it. It has been in anthologies of American prose ever since, and in 1998 it was read aloud in the film Saving Private Ryan. Its facts were wrong: two of the sons had been killed, one had deserted, one had been honourably discharged, and one had been captured and either deserted to the enemy or died in prison. The error was Schouler's, and it has no bearing on who wrote the sentences.
Who composed those sentences is the mystery. Lincoln signed the letter, but the manuscript disappeared. No draft, copy or letter-book entry has ever been found in his papers or anyone else's, the newspaper text is the only authority for the wording, and the "facsimiles" of a manuscript that dealers sold from 1891 onward are forgeries. Lincoln had two private secretaries, John G. Nicolay and John Hay, and Hay, then 26, drafted many of the letters that went out over the President's signature. Hay went on to a long public life (he was Secretary of State from 1898 until his death in 1905). During his last years a story went round Washington that he had written the Bixby letter, and after he died several men said that he had told them so in confidence.
None of those men wrote the story down while Hay lived. The earliest record of any of the claims is a letter of 1922, seventeen years after Hay's death and fifty-eight after the letter, and every one of them reached print through at least two intermediaries. Hay himself left the opposite on paper. In January 1904 he wrote to William E. Chandler, a former senator from New Hampshire who had asked him about a suspect facsimile, that "the letter of Mr. Lincoln to Mrs. Bixby is genuine", and around 1902 he told Robert Lincoln, the President's eldest son, that he knew nothing of where the printed text had come from and had never seen a draft or copy. He never mentioned the letter in his diary, in his letters or to his family.
Both answers fit the record, which is why the argument has lasted. For Hay: drafting letters for Lincoln's signature was his job; "beguile", the letter's most unusual word, is one he used often and Lincoln never; and a clipping of the letter is pasted into two of his scrapbooks. For Lincoln: he wrote his own letters of condolence (the two that survive from his presidency are in his hand); the request came to him from a governor through the Secretary of War, Edwin M. Stanton, and was the kind of letter he kept for himself; and Hay's own statements point away from Hay. Historians have divided along these lines since the 1940s. F. Lauriston Bullard, a Boston editor whose 1946 book Abraham Lincoln and the Widow Bixby is still the fullest study, argued for Lincoln, as did Joe Nickell, an investigator of questioned documents, in 1989, and Jason Emerson, a historian of the Lincoln family, in 2006 and 2008. Michael Burlingame, the author of the fullest modern life of Lincoln, has argued for Hay since 1995. Roy P. Basler, the editor of the standard edition of Lincoln's writings (1953), printed the letter as Lincoln's and called the claim for Hay "unproved".
In 2019 the question seemed to close. Jack Grieve, a linguist at the University of Birmingham, and six colleagues applied a technique called n-gram tracing to the letter. The technique counts how many of the short strings of letters and words in a questioned text (its "n-grams") also occur somewhere in each candidate's known writing, and gives the text to the candidate who shares more of them. Tested on more than 1,600 known texts by the two men, most of them shorter than 200 words, the technique almost never erred, and on the letter it chose Hay at every one of seventeen string lengths. The authors concluded that Hay "was almost certainly its primary author", and newspapers reported the mystery as solved.
Two kinds of evidence exist, and this report works through both. The first is documentary: everything Hay is reported to have said, everything he wrote himself, what he told Robert Lincoln, the paper trail of the request from Boston to Washington, and the record of what Lincoln and Hay each wrote in the week the letter was sent. All of it is in print, in Bullard's book, in Burlingame's and Emerson's articles, in Hay's published letters and diary, and in Basler's edition, which records for every Lincoln document whose handwriting the surviving copy is in. We read these sources in full rather than through summaries, dated every reported claim and traced who told whom, and counted from Basler's notes what came from the President's desk that week and in whose hand.
The second kind is stylometry, the statistical comparison of writing styles, and here the report does three things. It rebuilds the 2019 technique from the paper's description and checks that it gives the answers the paper reported, on the letter and on the known texts used to validate it (it does). It then runs three tests the paper did not run, each asking whether the technique still works when the known Lincoln texts resemble the letter in date, in kind of writing (its register, in the linguists' term) and in the form in which they survive (manuscript or print, which we call the edition). A second implementation of the same tests, written from the paper by a different hand without sight of ours, was run as a check on the first. Last, it tries the technique and three other standard methods on the evidence that matters most: known letters that Hay wrote for Lincoln's signature, to see whether any method can tell Hay writing as Lincoln from Lincoln.
The result is a lean. The 2019 attribution reproduces, and its validation reproduces, but under the added tests it proves sensitive to the design of the comparison: the same technique sends between roughly a quarter and a half of Lincoln's own elevated wartime prose to Hay (28 to 48 percent, depending on the pieces and the pool), the Gettysburg Address among it, and no method we ran detects Hay writing in Lincoln's name. We therefore treat the stylometry as neutral. What remains is documentary and close to a balance. On Hay's side are the reported claims and the scrapbook clippings; on Lincoln's, the office's practice and Hay's own written statements. Weighing those items as explicit numbers (section 5 shows each one and what happens when it changes) gives Lincoln about 53 percent, Hay about 42, and another hand in the office about 5. Changing any one weight moves Lincoln between 26 and 75 percent, and taking the 2019 study at face value puts him at 11.
The one piece of evidence that had seemed to settle the question, the 2019 n-gram study, reproduces under its own design and passes the validation its authors ran. Under the added tests, with Lincoln texts closer to the letter in date, register and edition, its attribution proves sensitive to all three, and no method we ran tells the two men apart in this kind of writing. The second implementation reproduces the attribution and finds the register effect in the same direction and smaller: 19 of 63 pieces of Lincoln's own elevated wartime prose (cuts of about 140 words, matched to the letter's kind of writing) go to Hay there, 30 here, and the difference lies in the choice of Lincoln texts, not in the code (section 4). What remains is documentary, and it is close to a balance. No new document was found; everything new here is analysis of evidence already in print. The table gives the confidence for each claim separately.
| Claim | Confidence and what it rests on |
|---|---|
| Lincoln signed and sent the letter, and the Transcript text of 25 November 1864 is the letter as sent. | Beyond reasonable doubt. Governor Andrew's request, Schouler's delivery, the printing on the day of delivery, and Hay's 1904 statement. The 1891 facsimile and its copies are forgeries (Hay 1904; Bullard 1946). |
| No claim by Hay to have written it survives in his own hand or from his lifetime. | Established. Five reported claims, none first-hand, the earliest written down in 1922. Bullard (1946) and Emerson (2008) showed this; confirmed here from the sources. |
| The 2019 attribution to Hay reproduces under its own design. | Confirmed by replication: all seventeen string lengths point to Hay, the shortest by a margin of four parts in a hundred thousand. The study's own validation reproduces too: under its design every one of the 1,011 known texts of 50 to 400 words in the corpora built for this report (the paper's own set was 1,662) is attributed correctly by the majority vote (each string length casts one vote for the man whose known writing shares more of the text's strings, and the majority decides). |
| Under added tests, the 2019 design's attribution is sensitive to period, register and edition. | Reproduced in direction by the second implementation. Under that design 19 to 30 of 63 register-matched pieces of Lincoln's wartime prose go to Hay, the Gettysburg Address among them in both runs, the rate depending on the size of the Lincoln pool, the set of known Lincoln texts the letter is compared against (section 4); Lincoln's war texts in his own hand of 100 to 200 words go to Hay 14 percent of the time; print versions of his texts sit nearer Hay than manuscript versions at every string length. This report puts the chance at about 90 percent that the same direction would hold on the paper's own corpus (its collection of known texts), which neither run could obtain. The Hay side of the register set is three documents, which limits what the test can show (section 4). |
| Controlled stylometry cannot decide the question. | About 85 percent, conditional on the corpora and designs used here. Under the own-hand design (Lincoln's writing in his own hand from his presidency against Hay's letters and diary of the same years) the n-gram lengths split eight to nine; a classifier (a statistical model trained to sort texts by author, its probabilities adjusted so that a stated 60 percent is right about 60 percent of the time) and Burrows's Delta (a measure of stylistic distance), both described in section 4, lean to Lincoln; and the methods call 18 of the 19 documents in Basler's edition that survive in Hay's handwriting, three of Hay's four known letters for Lincoln's signature, and Hay's own condolence "Lincoln". Weight in the final reckoning: none (a likelihood ratio of 1, with 1:3 to 3:1 as the range; the term is explained in section 5). The second implementation finds the same under period-matched pools of its own. |
| Lincoln composed the letter. | About 53 percent, Hay about 42, another hand about 5, conditional on the weights given in section 5, which are judgments. Changing any one weight moves Lincoln between 26 and 75 percent; taking the 2019 stylometry at face value puts him at 11. This is a lean. It does not establish that Lincoln wrote the letter. |
This report does not show that Hay did not write the letter, or that Lincoln did; nothing found excludes either man. It shows that the stylometric case for Hay depends on the comparison design, and that the rest of the evidence is a near balance that leans, on the office's practice and Hay's own words, to Lincoln.
| Line of inquiry | Finding | Weight | Status |
|---|---|---|---|
| Hay's reported claims | Five chains of report, none first-hand, none written down before 1922, two prompted by framed copies of the letter. The one datable occasion, a visit by the British statesman John Morley in November 1904, is contradicted by Hay's own letter of January 1904 and by his diary. For Hay. | ●●○○ moderate | Bullard (1946) and Emerson (2008) established the hearsay character; the chains are dated and cross-checked here against the memoirs of the men involved and Hay's diary |
| Hay's own words | "The letter of Mr. Lincoln to Mrs. Bixby is genuine" (to Chandler, 1904); he told Robert Lincoln he "knew nothing whatever of the source" and had "never seen any draft or copy" (reported by Robert's secretary in 1908 and by Robert in 1917). Nothing in his diaries, letters, essays or family. For Lincoln. | ●●○○ moderate | Bullard printed the Chandler letter; Emerson (2008) found Robert Lincoln's letters; weighed here against Burlingame's narrow reading |
| The office's practice | Of the documents in the three volumes of Basler's edition that could be read, dated from May 1860, 1,887 survive in Lincoln's hand and 31 in Hay's; both surviving presidential condolences are in Lincoln's hand; Hay's known compositions for Lincoln are routine. Hay wrote bodies for Lincoln's signature four times that week, and a letter of the same day to John Phillips, a 104-year-old Massachusetts voter, survives in an unidentified hand. For Lincoln. | ●●○○ moderate | Argued by Bullard, Emerson and Robert Lincoln; counted here from the handwriting notes in Basler's edition |
| The week's workload | Basler's edition has 15 documents for 21 to 27 November, against a thirteen-week mean of 21, 23 in election week and 35 in the week of the annual message to Congress. Hay's letter of 25 November to Charles S. Spencer, saying that the "crush" of business had stopped the President writing a note he had meant to write himself, shows Lincoln reserving a letter to himself. Removes a premise of the Hay case. | ●○○○ weak | Bullard argued it; counted here |
| The scrapbooks | A clipping of the letter in two Hay scrapbooks, beside other Lincoln letters and Lincoln material Hay did not write; the copy of the Phillips letter kept beside it is not in Hay's hand. For Hay. | ●○○○ weak | Burlingame (1995) found them; Emerson (2008) examined them; the Phillips-hand point added here |
| Vocabulary | "Beguile" is Hay's (none in 338,000 words of Lincoln). The office idiom ("tendering to you", "the files of the War Department", "should attempt to") is Lincoln's alone; the elegiac idiom ("the altar of", "solemn pride", "gloriously") is Hay's alone, and mostly from after 1865. Split. | ●○○○ weak | Burlingame, Nickell and Jean Pival, the linguist who examined the letter's grammar for Nickell's 1989 article, supplied the examples; the full table across four corpora is new |
| The 2019 n-gram tracing | Reproduces (all seventeen lengths to Hay, in both implementations) and passes the validation its authors ran. Under the added tests the same design misattributes 19 to 30 of 63 register-matched Lincoln pieces, the Gettysburg Address among them, and is sensitive to period and edition. Neutral. | ●○○○ weak | Grieve and colleagues (2019). The authors named register sensitivity as untested in their 2016 presentation; the replication and the three specific tests are new; a second implementation of the register test found the same direction and a lower rate |
| Controlled stylometry | Own-hand n-gram tracing splits eight lengths to nine; the calibrated classifier gives a probability of Hay of 0.29 to 0.41 against 0.66 for Hay's own pieces; Delta says Lincoln; and 18 of 19 Hay-hand items, three of four known Hay-for-Lincoln letters and Hay's own condolence are called Lincoln. Neutral. | ●○○○ weak | New; the second implementation's period-matched design agrees |
Weight is our judgement of how much each line moves the question on its own; the direction is given at the end of each finding. The 2019 study was the only line that had seemed decisive, and under the added tests it is the one that proves design-dependent.
Section 1 gives the text of the letter as printed and traces the request from Boston to the White House. Section 2 sets out the documentary record: what Hay is reported to have said, what he wrote himself, and what he told Robert Lincoln. Section 3 reconstructs the President's desk in the week of 21 November 1864 from Basler's edition, item by item. Section 4 is the stylometry: the 2019 study replicated, then tested in ways its authors did not run, then rerun by a second, independent implementation. Section 5 weighs the documentary items as likelihood ratios, a way of giving each item a number that is explained there, and shows how far the result moves when any one of them changes. Section 6 states the case for Hay at full strength, in ten parts, with what was found against each. Section 7 lists what would overturn the verdict in either direction. Section 8 separates what was known from what this report adds and says where it falls short. Section 9 lists the sources and the saved code and data.
The text below is the one printed in the first edition of the Boston Evening Transcript on Friday 25 November 1864, the day the letter reached Mrs Bixby. It is the only authority for the wording. No manuscript, draft, copy or letter-book entry has ever been found, and the "facsimiles" sold from 1891 onward were traced in pencil and inked over on the wrong paper, as the autograph dealer Charles Hamilton showed in Great Forgers and Famous Fakes (1980, pp. 29 to 37). Basler reprinted the Transcript text in his edition, The Collected Works of Abraham Lincoln (1953), volume VIII, pages 116 to 117 (cited below as CW VIII, 116 to 117), with the note that "there is no reason to question the authenticity of the text of the letter which appeared in the Transcript and other contemporary sources." Every stylistic analysis, including ours, therefore rests on the 139 words as the Transcript's compositor set them in type.
The facts in the letter were wrong. Two of Lydia Bixby's sons were killed (Charles, 3 May 1863; Oliver, 30 July 1864); one deserted, one was honourably discharged, and one was captured and either deserted to the enemy or died in prison (Basler, CW VIII, 117 n.). The error was Schouler's, and it has no bearing on who composed the sentences.
The paper trail from Boston to the White House is complete and was printed by Schouler himself in 1868 (History of Massachusetts in the Civil War, I, 576 to 578) and re-examined in the National Archives by Bullard in 1946 (pp. 14 to 26). On 24 September 1864 Schouler reported to Governor Andrew on the case of Otis Newhall of Lynn, a father of five soldiers who had asked for one son's discharge, and added "a word in regard to a still more remarkable case": a widow who had shown him "five letters from five different company commanders, and each letter informed the poor woman of the death of one of her sons." Andrew forwarded the papers to the War Department with the sentence that set the letter in motion: "This is a case so remarkable, that I really wish a letter might be written her by the President of the United States, taking notice of a noble mother of five dead heroes so well deserved." The War Department received the papers on 26 and 27 September; Major Thomas M. Vincent, of the Adjutant General's office in Washington, asked Schouler for names and regiments on 1 October; Schouler replied on 12 October; Colonel Samuel Breck of the same office summarised the two cases for Stanton, the Secretary of War, on 28 October, ending "Governor Andrew desires that a letter may be written Mrs. Bixby by the President." The file was signed out to Stanton that day. On 5 November Schouler called on Lincoln in Washington and "passed an agreeable half hour with him" (Schouler's report for 1864, p. 606); nothing records that the Bixby case came up. The letter is dated Monday 21 November. Schouler placed it in Mrs Bixby's hands on the morning of Friday 25 November, and the Transcript had it in type the same day (Bullard, pp. 26, 50 to 51). How the letter travelled from Washington to Schouler's office is not documented.
Two things in this trail matter for authorship. The request reached Lincoln through the Secretary of War with a governor's explicit wish that the President write, which is the kind of request that came to the President personally rather than to the mail clerks. And the letter left no trace in the White House: no draft in Lincoln's hand, no draft in Hay's hand, no copy. Both facts are consistent with either author, as section 3 explains.
The case for Hay rests on things he is reported to have said in private. The case for Lincoln rests on things Hay wrote down himself and on what he told Robert Lincoln. We read every source that prints either kind of statement (the 1940 memoir of Nicholas Murray Butler, the president of Columbia University; Bullard 1946; Sherman Day Wakefield's two pamphlets of 1947 and 1948; Burlingame 1995 and 1999; Emerson 2006 and 2008; Nickell 1989) and set the two kinds side by side.
| Hay allegedly told | When | Chain to print | First written down | What survives a check |
|---|---|---|---|---|
| John Morley, the British statesman and biographer of Gladstone, at the State Department | November 1904 | Morley to Nicholas Murray Butler (London, 9 July 1912); Butler's memoir Across the Busy Years II (1940), 390 to 393 | 1926 (Butler to the Lincoln biographer William E. Barton, veiled); 1940 in print | The one claim tied to a datable occasion, recorded 36 years after it. Butler's circumstantial dating of Morley's movements is wrong (Bullard, pp. 113 to 114). Morley's own Recollections (1917) describe the White House week and never mention Hay. Hay's diary for 12 and 13 November 1904 records Morley's visits and his brother Leonard's death, and nothing about the letter (Bullard, pp. 116 to 117). |
| W. C. Brownell, literary adviser to the publisher Charles Scribner's Sons | Undated, before July 1905 | Brownell to the newspaper editor Rollo Ogden "at the time"; Ogden's New York Times editorial of 14 May 1933 and letter to Bullard of 15 May 1933 | 1933 | Ogden: "I know of no scrap of documentary evidence." Brownell died in 1928; nothing in his hand records the confidence (Bullard, pp. 107 to 108). |
| Walter Hines Page, the publisher, later ambassador to Britain | Undated, before July 1905 | Page to Cora, Lady Strafford, an American-born countess (1913 to 1918); she to the Rev. G. A. Jackson, an English clergyman; Jackson to the essayist E. V. Lucas, 16 January 1922; Lucas, Post-Bag Diversions (1934) | 1922 | Five links. This version has Lincoln order a draft and send it unchanged because Hay "had so perfectly captured his style"; the Morley version has Hay explain why no manuscript survived. The two stories do not describe the same event (Bullard, pp. 110 to 112). |
| An informant of Louis A. Coolidge, a Boston journalist | Unknown | Coolidge "emphatically told one person, who in turn told us" (Bullard, pp. 109 to 110) | Never | Anonymous at two links. Not certainly a claim for Hay rather than a denial for Lincoln. |
| Spencer Eddy, Hay's State Department secretary | c. 1898 to 1901 | Eddy to his sister Catherine Beveridge, wife of Senator Albert J. Beveridge; her memorandum of 22 July 1949; Burlingame 1995, p. 64 | 1949 | The memorandum itself says the story was "presumably heard (or inferred)" from Hay or Henry Adams, the historian and Hay's closest friend (Emerson 2008, p. 91). |
Nothing Hay wrote in his lifetime claims the letter. His diaries and letters as printed in 1908, William Roscoe Thayer's biography of 1915 and Tyler Dennett's edition of his Civil War diary and letters (1939) do not mention it. His children told Barton in 1926 and Bullard in 1946 that he never spoke of it to them. J. A. Spender, the English editor who "knew [Hay] well", told Bullard in 1928 that the rumour "could not be raised on anything that Hay himself said to me" and in 1941 that he had it "possibly from Lord Morley" (Bullard, pp. 106 to 107, 114 to 115). The earliest written record of any claim is Jackson's letter of 1922; the earliest publication is Ogden's editorial of 1933. Burlingame, the most careful advocate of Hay's authorship, concedes that none of these reports "clinch the case" and relies on their number (1995, p. 71). Their number is not independent evidence: Morley, Page and Brownell moved in the same circles, the story was "afloat in Washington" by 1904 (Bullard, p. 127), and none of the three first recipients left a record.
Burlingame reads "genuine" narrowly: the printed text is authentic, the facsimile is not, and Hay says nothing about who composed it (1995, p. 68). Bullard read it broadly, and pointed to Hay's letter of 4 June 1903 to a Mrs Langford, who had asked about two Lincoln letters, in which he volunteered that "the signatures, of course, are genuine, but the body of the letters was written by one of the clerks in the White House. It was not Mr. Nicolay or myself" (pp. 129 to 133). Both readings are possible. The 1903 letter shows that when Hay knew a body was not Lincoln's he said so unasked; the 1904 letter shows that ten months before he allegedly confessed to Morley he described the Bixby letter to Chandler as "the letter of Mr. Lincoln".
Robert Lincoln's testimony, published by Emerson in 2008 from Robert's letters to Isaac Markens, a New York writer on Lincoln, now at the Chicago History Museum, is the nearest thing to a first-hand record. Robert showed Hay a lithograph of the letter at the State Department around 1902. "Colonel Hay knew nothing whatever of the source of the printed copy of the letter in the Nicolay and Hay book. He suggested a curious thing; namely, that my father's handwriting was very easy to imitate ... and that perhaps this lithograph had been made from a forged document" (to Markens, 5 January 1917; the "book" is the edition of Lincoln's works that Nicolay and Hay published in 1894, which printed the letter). "Your suggestion that neither Nicolay nor Hay probably had any special knowledge of the letter at the time is correct. Hay himself told me so" (24 February 1917). Robert's secretary had written in 1908 that "both Mr. Nicolay and Mr. Hay stated that they had never seen any draft or copy of this letter" (Towers to a correspondent named Tweedale, 22 April 1908; Emerson 2008, pp. 103 to 105, n. 68). Robert also wrote that his father "never dictated correspondence; he sometimes wrote a document and had his draft copied by either Nicolay or Hay", and "I do not now recall having seen any letter written by either Nicolay or by Hay which was personally signed by him" (13 February 1918; Emerson 2008, pp. 106 to 107). The last statement is wrong: Basler's notes record dozens of letters and telegrams in Hay's hand signed by Lincoln, several of them in the weeks around the Bixby letter (section 3). Robert had spent the war years away from Washington, so his knowledge of office routine is second-hand, while his report of what Hay told him is first-hand.
Burlingame quotes the first four sentences; Emerson points out that the fifth is usually left out (2008, p. 88). Both halves are true. Hay composed and signed for the President as a matter of routine, and Lincoln wrote letters himself every week, the ones he cared about. The question is which kind the Bixby letter was, and the 1866 letter cannot answer it.
Burlingame's strongest documentary find (1995, pp. 70 to 71) is that a clipping of the Bixby letter is pasted into two of Hay's scrapbooks, one at Brown University filled mostly with Hay's own poems and their reviews, one at the Library of Congress (vol. 54) filled with his anonymous journalism of 1860 to 1865. "It is difficult to understand why Hay would have pasted the Bixby letter in those scrapbooks, full of his own literary creations, unless he had composed it himself." Emerson examined the same volumes (2008, pp. 94 to 96): the Brown scrapbook also holds other letters signed by Lincoln and a French newspaper notice about Mary Lincoln; the Library of Congress volume holds clippings of Lincoln's letters to John Phillips, L. B. Wyman, F. B. Loomis and a New York committee; a third oversized volume is half filled with Lincoln's writings including photostats of the Gettysburg Address and the Second Inaugural. Hay had been collecting material for a history of the administration since 1861. The scrapbooks show that Hay kept a clipping of the letter, and nothing in them records why he kept it. We weigh them as weak evidence for Hay, and note that the Phillips letter kept beside it survives only in a copy "in an unidentified handwriting" that is not Hay's (Basler, CW VIII, 118 n.), which cuts against reading the scrapbook as a record of Hay's own compositions.
Two arguments about that week recur in the literature. Burlingame holds that Lincoln was "unusually preoccupied" after the election and had no time for a condolence letter, citing Hay's note of 25 November to Charles S. Spencer of New York, who had asked for a letter from the President to read at a banquet: "I regret that the President was literally crowded out of the opportunity of writing you a note for yr. banquet. He fully intended to do so himself, and for that reason, I did not prepare a letter for him. But the crush here just now is beyond endurance" (Burlingame's edition of Hay's Civil War letters, At Lincoln's Side, 2000, p. 101, as quoted in Burlingame 1999). Bullard and Emerson hold that a request of this kind, sent by a governor through the Secretary of War, was exactly what Lincoln reserved for himself.
We counted what survives. Basler's edition lists every known Lincoln document with its manuscript source. For the thirteen full weeks from 3 October to 1 January the mean is 21 items a week. The week of 21 to 27 November has 15, the same as the week before and fewer than election week (23) or the week of the annual message (35). In that week Lincoln wrote in his own hand a 154-word letter to General Nathaniel P. Banks with a draft, five telegrams and several endorsements. The Spencer note shows that Hay drafted routine acknowledgments and that Lincoln kept some letters for himself even when he then failed to write them; it does not say which kind the Bixby letter was. The "crush" was callers, General Grant's visit on 23 November, the resignation of the Attorney General, Edward Bates, on 24 November, and the annual message to Congress read to the cabinet on 25 November; it was not correspondence.
| Dated 21 November 1864 | Source Basler found | In Lincoln's hand? |
|---|---|---|
| To Mrs Lydia Bixby, 139 words | Boston Transcript, 25 November 1864; no manuscript | Unknown |
| To John Phillips, the 104-year-old voter of Sturbridge, about 95 words | Copy in the Nicolay Papers "on Executive Mansion stationery in an unidentified handwriting"; New York Tribune, 9 December 1864 | No surviving text in his hand |
| To Augustus R. Wright, telegram, about 40 words | Autograph letter signed (Basler's code ALS), National Archives, Presidential Telegrams | Yes |
| To William P. Fessenden, the Secretary of the Treasury, endorsement for Joseph G. Bowman, 70 words | Illinois State Historical Library | Yes |
| Three endorsements to Stanton, 15 to 75 words | A 1912 dealer's catalogue; Nicolay and Hay's edition of Lincoln's works (1894); Gilbert Tracy's Uncollected Letters of Abraham Lincoln (1917) | Manuscripts not located |
The two composed letters of the day are the two for which nothing in Lincoln's hand survives, and both went to Massachusetts civilians, the class of letter (all those "of a non-political kind", he said) that Nicholas Murray Butler claimed Hay wrote for Lincoln (Bullard, pp. 81 to 82). The coincidence is consistent with Hay's authorship but on its own does not establish it. The Phillips copy is not in Hay's hand either, and Hay's own drafts for Lincoln in these months usually survive in the Nicolay or Lincoln papers (to Charles Butler of New York, 9 November; to the Shakers, 8 August; to Thomas J. Durant of New Orleans, 24 December; to William Lloyd Garrison, the abolitionist editor, 7 February), which the Bixby letter does not. Hay was in Washington all that week, back since the morning of 15 November from City Point, Grant's headquarters in Virginia, and he wrote for Lincoln on 23, 25, 27 and 28 November (CW VIII, 122, 568 to 569). Nicolay was ill in St Louis from 17 to 30 November, so the two-candidate framing is sound for the secretaries; the clerks Hay mentioned to Mrs Langford are a third possibility nobody has examined, and a 139-word text cannot exclude them.
Basler's notes let us sort the war-era documents by hand. Our copy of volumes IV, VII and VIII holds 3,125 items. Of those dated from 18 May 1860 onward, 1,887 survive in Lincoln's own hand (Basler's codes ALS, ADfS and AES, for an autograph letter, draft or endorsement, signed), 1,094 survive in other hands or only in print, and 31 survive only in Hay's hand, with Lincoln's signature or without it. The Hay-hand items are telegrams, orders, a draft to the Shakers, acknowledgments of gifts (a congressman's salmon trout: to John F. Driggs, 3 February 1865), a declined invitation (Charles Butler, 9 November 1864) and a letter of thanks to Garrison (7 February 1865). Lincoln's two surviving letters of condolence, to the parents of Elmer Ellsworth, the young colonel and friend of his killed in May 1861 (25 May 1861), and to Fanny McCullough, the daughter of an Illinois friend killed that month (23 December 1862), are in his own hand, and so is his letter of religious reflection to Eliza Gurney, a Quaker minister who had visited him in 1862 (4 September 1864). Hay's known compositions in Lincoln's name are courteous and flat: "I could not ask, and no one could merit, a better reward" (to the Philadelphia poet George H. Boker, 26 October 1863); "I beg that you will convey to Mr. Williams my cordial thanks for his kind thoughtfulness" (to Driggs). Nothing Hay is known to have written for Lincoln is a letter of condolence. This record is the starting point of the weighing in section 5, and section 4 asks whether the stylometry changes it.
The only published stylometric study of the letter is by Jack Grieve and six colleagues (Isobelle Clarke, Emily Chiang, Hannah Gideon, Annina Heini, Andrea Nini and Emily Waibel), "Attributing the Bixby Letter using n-gram tracing," in the journal Digital Scholarship in the Humanities 34 (2019), pp. 493–512. It concluded that Hay "was almost certainly its primary author." No replication has been published since, as far as we could find (section 8). We rebuilt the method from the paper's description, ran it on corpora we built ourselves, and then ran the tests the paper did not run: whether the method still works when the known texts resemble the letter in date and register, and whether its answer depends on which texts of each man are put in the comparison pool.
Every Lincoln text comes from Basler's Collected Works, volumes I, IV, VII and VIII (the volumes for 1849 to 1859 and November 1861 to August 1863 are not available in full text; the 1849 to 1860 gap was filled from Arthur Brooks Lapsley's edition of 1905, an older printed collection). Basler's source codes let us keep only what survives in Lincoln's own hand (ALS, ADS, ADfS and the like) and set aside the items Basler says are in Hay's handwriting. Hay's writing comes from the three volumes of his letters and diary that his widow had printed privately in 1908 (Letters of John Hay and Extracts from Diary), from Thayer's biography, and from his books; only his 1860 to 1865 letters and diary are contemporary with the letter. Twenty other writers of the period supply a control set, used to check that a method separates Lincoln from Hay rather than from writers in general: William H. Seward, Salmon P. Chase and Gideon Welles of Lincoln's cabinet; Senator Charles Sumner; the secretaries John G. Nicolay and William O. Stoddard; the journalists Noah Brooks, Horace Greeley and George William Curtis; the orator Edward Everett; the writers Ralph Waldo Emerson, James Russell Lowell, John Lothrop Motley, Bayard Taylor, Mark Twain, Walt Whitman and Henry Adams; Generals Grant and Sherman; and Frederick Douglass.
| Pool | Texts | Words | What it is |
|---|---|---|---|
| Lincoln, own hand, 18 May 1860 to April 1865 | 1,868 | 120,000 | Every Basler item with an autograph source code and at least five words: letters, telegrams, endorsements, memoranda, a few addresses. The full set is 1,887 items and 125,600 words; the table gives the pool as used, after the copies of the test texts and the full documents from which the register pieces were cut were removed |
| Lincoln, all autograph items, 1832 to 1865 | 2,197 | 187,000 | The above plus volume I, after the same removals |
| Lincoln before 18 May 1860 (Grieve's cut) | 580 | 243,000 | Basler volume I and Lapsley for 1849 to 1860, after the same removals; Grieve's own pool was 1,085 texts and 400,747 words |
| Hay, 1860 to 1865 | 150 | 53,000 | Letters and diary from the 1908 edition |
| Hay, everything | 891 | 495,000 | Letters, diary, essays, his travel book Castilian Days, his novel The Bread-Winners, addresses, verse, 1858 to 1905 |
| Register-matched pieces, Lincoln | 63 | 9,600 | Thirty personal, elevated or ceremonial texts of 1861 to 1865 cut into pieces of about 140 words: the Gettysburg Address, the Second Inaugural, the Ellsworth, McCullough and Gurney letters, and letters to the Kentucky editor Albert G. Hodges, General Joseph Hooker, Horace Greeley, General Carl Schurz, the actor James H. Hackett, the New York politician Thurlow Weed, General Sherman and others |
| Register-matched pieces, Hay | 57 | 11,000 | His two war-time eulogies, "Ellsworth" (Atlantic, July 1861) and "Colonel Baker" (Harper's, December 1861), attributed to him by both magazines' own indexes, plus his 55-word condolence of 9 June 1864 to his friend Manning Leonard |
| Controls | 5,055 | 1,180,000 | Twenty writers, up to 60,000 words each, in pieces of 100 to 300 words |
The questioned text is the Transcript printing, 139 words after the salutation. Every copy of a test text was removed from the pools before it was tested, and the full document from which each register piece was cut was removed too.
N-gram tracing takes every distinct string of one, two or three consecutive words and every distinct string of 3 to 16 consecutive characters (spaces included) in the questioned text, its word and character n-grams, and asks what share of them occurs anywhere in an equal-sized random sample of each candidate's writing. The candidate with the larger share wins, one vote per n-gram length, and the authors take the majority of the character 4- to 10-gram votes as the result. We reimplemented it exactly as described (capitals lowered, punctuation kept, no string running across a sentence boundary, samples of 0.95 times the smaller pool, and the random sampling repeated 10 times for validation and 50 times for the letter, with the shares averaged).
On known texts it behaves as the paper says. Under Grieve's design, Lincoln before May 1860 against all of Hay, leave-one-out attribution (each known text removed from its pool and attributed as if unknown) of 423 Lincoln and 588 Hay texts of 50 to 400 words is correct by the majority vote for every text. At single lengths the only misses are one Lincoln text at 3 characters and one or two Hay texts at 3 characters and at each length from 10 to 16. Under our own-hand design, Lincoln's autograph writing of 1860 to 1865 against Hay's letters and diary of the same years, the vote is right for 643 of the 645 Lincoln texts and for all 93 Hay texts. No single character length from 5 upward misses more than four Lincoln texts, and no length misses a Hay text. The 182 Lincoln and 31 Hay texts of 100 to 200 words are all attributed correctly by the vote. A 139-word text is not too short for the method when the text is an ordinary letter, telegram or diary entry. Grieve's confidence rests on this validation, and the validation holds.
The letter also comes out as the paper reports. Under Grieve's design all seventeen analyses give it to Hay: the 3-character analysis by four parts in a hundred thousand, the rest by margins of 0.01 to 0.14, with 0.04 to 0.10 for character 5- to 12-grams and 0.14 for word 2-grams, where the margin is Lincoln's share minus Hay's. (The paper reports the 5-word n-gram analysis for Lincoln on one phrase; we did not run 4- or 5-word n-grams.) Figure 1 plots the margin at each of the seventeen string lengths under this design and under the own-hand design described below.
The authors had anticipated the second of these tests, on register. Their first presentation of the work (Aston University, 13 April 2016) noted that they had "not looked at detail at the effect of ... register variation" and that they "also need to test sensitivity to register variation", and its last slide listed testing "across different registers" as work still to do. The published paper says that "a systematic analysis of the writings of many authors and many registers is needed" and was being pursued. The tests below are new, and a second implementation of them, written independently, is reported at the end of this subsection.
Period. Grieve's Lincoln pool ends in May 1860 because the presidential correspondence is contaminated by secretaries. But Lincoln's style changed. When his own-hand writing of 1860 to 1865 is tested against Grieve's pool, 14 percent of the 182 texts of 100 to 200 words are attributed to Hay by the majority vote, and 18 percent of the 377 texts of 50 to 100 words; the figure for his pre-1860 texts is zero. Tested against each other, Lincoln before 1860 and Lincoln in the White House are separated by the method with 100 percent accuracy at character lengths 5 to 16. The method sees a difference between the two Lincolns as clearly as it sees a difference between Lincoln and Hay. The Bixby letter, tested against the two Lincoln periods alone, sits between them.
Register. Both pools are dominated by business prose, and the letter is not business prose. On the register-matched pieces the picture changes. Under Grieve's design the majority vote gives 30 of the 63 Lincoln pieces to Hay, including the Gettysburg Address, the Ellsworth letter and the McCullough letter, while 56 of the 57 Hay pieces go to Hay. The second implementation, on the same pieces with a smaller Lincoln pool, gives 19 of the 63 to Hay; the cause is traced below. At character lengths 4 to 6 the method is right about Lincoln's elevated prose only 21 to 38 percent of the time. Under the own-hand design the vote misses 17 of the 63 Lincoln pieces and none of Hay's. Short character strings, which carry function words (the, of, and), word endings and spelling, are the ones that fail; they mark the register, and the register of an elevated Lincoln text looks like Hay to a pool of Lincoln's telegrams. Long strings, which carry phrases, recover: at character lengths 11 to 16 the own-hand design is right about 89 to 94 percent of Lincoln's pieces. Figure 2 plots, for each string length and each design, how often the method attributes the register-matched pieces correctly, beside its rate on ordinary known texts.
The register set has limits of its own. The 63 Lincoln pieces come from 30 documents, and the 57 Hay pieces from three: two eulogies of 1861 and one short letter. The validation used below, which leaves out one source document at a time, therefore tests the classifier on only three Hay sources, and a Hay eulogy is not a Hay condolence letter. The test shows that the 2019 design misreads between a third and a half of these pieces of Lincoln's elevated prose, the share depending on the pool. It cannot show how that design would treat a wider range of Hay's writing in this register, because no such range survives from the 1860s.
Edition. Lincoln's own-hand texts are transcribed from manuscripts and keep his habits ("can not," "it's," dashes); the letter survives only in print, and the Lapsley texts used for 1849 to 1860 and for the 1862 to 1863 register pieces are print. We found 95 Lincoln texts that appear in both editions. The print version of the same words scores 0.010 to 0.016 nearer to Hay than the manuscript version at every n-gram length (at character 4-grams, in 85 of 95 pairs), and its majority-vote accuracy falls from 0.93 to 0.74 at 4-grams. In every design whose Lincoln pool is built from manuscripts, a printed Lincoln text therefore scores slightly nearer Hay than the same text in manuscript would.
A second implementation of n-gram tracing was written from the paper alone, by a different hand, without sight of our code or our results. It rebuilt the two pools from our corpus files under the paper's design and reran the tests. Its report is cited in section 9 and its files are saved beside ours. On the letter it agrees with the paper and with us: Hay at all seventeen lengths, by margins of 0.002 at 3 characters to 0.145 at word 2-grams. Its leave-one-out validation of all 1,397 texts in its pools is right by the vote for 1,380, with the paper's accuracy curve (its F1 score, a standard accuracy measure that counts both wrong attributions and missed ones) at every length (the paper: all 1,662 of its texts; ours: all 1,011 of 50 to 400 words); its misses are Lincoln's verse, his 1842 eulogy, his temperance address and reported speech. On our 63 register pieces it sends 19 to Hay by the vote, 30 percent, against our 30, 48 percent; on 97 pieces of its own from 38 Lincoln documents of 1861 to 1865, 27, 28 percent. It agrees with us on which lengths fail (4 to 6 characters, at 25 to 44 percent accuracy) and on the Gettysburg and Ellsworth pieces, and it disagrees on the McCullough letter, which it gives to Lincoln four votes to three. Under period-matched pools of its own it finds what we found under the own-hand design: the letter splits, five lengths to two for Lincoln on the character vote and nine to eight for Hay over all seventeen, and all four of Hay's letters for Lincoln's signature, and Hay's own condolence, are called Lincoln.
The gap between 30 and 19 is in the Lincoln pool, not in the code. The two implementations agree piece by piece on 48 of the 63; on twelve of the fifteen they disagree about, one side or the other is a vote of four to three. Our code on the second implementation's exact pool gives 19 of 63 under the paper's rule for drawing the random samples of each man's writing, repeated 50 times, 19 under our rule repeated 20 times, and 20 repeated 10 times. Its code on our pool gives 26. Repeating our own run with three different random draws gives 30, 30 and 28, and under the paper's sampling rule (the whole smaller pool, repeated 50 times) 29. The one difference that matters is a set of 19 items from Lapsley's third and fourth volumes: 83,000 words of Lincoln's 1858 campaign speeches and his side of the seven debates with Stephen A. Douglas, his opponent for the Senate, which the second implementation dropped because Lapsley prints Douglas's letters and some interrogatories inside a few of them (a few hundred words at most that are not Lincoln's). With those 19 items removed, our code gives 18 to 20 of 63; with its selection of texts applied to our own copies, 18; and when the second implementation restored the substantive Lapsley items to its own pool, with Douglas's words cut out, its code gave 27 of 63. On its 97 pieces, our code gives 28 with its pool (its own figure is 27) and 40 with ours. Nothing else moved the figure.
The mechanism is the method's sampling rule. Each candidate is represented by a random sample the size of the smaller pool, so adding the debates raises Lincoln's pool from about 159,000 to about 243,000 words and Hay's sample by the same amount. Both halves of that matter. At an equal sample of 159,000 words, a Lincoln sample that includes the debates in proportion misses 24 of the 63 pieces where one without them misses 20: stump oratory supplies few of the strings of a condolence letter. Raising both samples to 243,000 takes the figure to 29: the extra words of Hay's mixed prose cover more of those strings than the same number of extra words of Lincoln's oratory do. (Raising Hay's sample alone, with Lincoln's held at 159,000, sends 54 of the 63 to Hay, which is why the method insists on equal samples.) Accuracy at 4 to 9 characters falls by 0.06 to 0.14 at every length when the debates are in. Which pool is the right one is a question about the paper's design, not about either implementation. The paper's Lincoln corpus, 1,085 texts and 400,747 words from Basler's first four volumes with texts up to 17,003 words long, included his speeches, and its samples were 261,000 words each, the size of its Hay corpus; our pool with the debates, 243,000 words, is the nearer reconstruction on both counts, and the second implementation's, 192,000, is the more cautious about Douglas. Neither is the paper's corpus, which was not obtainable. We therefore report a range: under the paper's design between 19 and 30 of the 63 register-matched Lincoln pieces go to Hay, 30 to 48 percent, and between 28 and 40 of the second implementation's 97, and the rate rises with the amount of Lincoln's oratory in the pool. The attribution therefore depends on which Lincoln texts are in the pool, and the range above measures the size of that dependence.
Under the own-hand design the seventeen analyses split, eight to Hay and nine to Lincoln. Character 3- to 8-grams, 10-grams and word 1-grams go to Hay, by margins of 0.002 to 0.046; character 9-grams, 11- to 16-grams and word 2- and 3-grams go to Lincoln, by margins of 0.000 to 0.018. The 4- to 10-gram vote is Hay, six lengths to one. A rerun with punctuation removed gives the same split. Read against the register-matched pieces, the short-string margins are typical of Hay's eulogies and outside the range of Lincoln's elevated pieces (at 4-grams the letter is below every one of the 63 Lincoln pieces), and the long-string margins are typical of Lincoln's pieces and outside Hay's (at 15-grams it is above every one of the Hay pieces). A logistic regression model (a standard statistical classifier) trained on the full set of margins of the 120 register pieces, validated by leaving out one source document at a time, classifies 92 to 95 percent of the Lincoln pieces and 89 to 90 percent of the Hay pieces correctly, and gives the letter a probability of Hay of 0.62 with all Hay pieces and 0.50 when the sixteen Hay pieces that are more than a quarter quotation are dropped. The same model trained under Grieve's design gives 0.98, and under a third design (all of Lincoln's autograph writing against all of Hay's) 0.61 to 0.73, a design that also gives the Gettysburg Address to Hay at 0.79 to 0.87.
Three methods that do not depend on shared strings were validated the same way and then applied.
| Method (own-hand design) | Accuracy on ordinary texts of 100 to 200 words | Accuracy on the register-matched pieces | The letter | Known Hay writing in Lincoln's voice |
|---|---|---|---|---|
| Character 3- to 5-gram logistic classifier (a classifier trained on short character strings, tested only on documents it was not trained on, with its probabilities calibrated so that a stated 60 percent is right about 60 percent of the time) | Lincoln 0.95, Hay 0.94 | Lincoln 0.92, Hay 0.95 | Probability of Hay, written P(Hay), 0.36 to 0.41; the Lincoln pieces average 0.37, the Hay pieces 0.66; only 2 percent of Hay pieces score as low | Hay's own 1864 condolence 0.25 to 0.30; the four letters Hay composed for Lincoln's signature 0.14 to 0.34, all called Lincoln |
| Word 1- and 2-gram classifier | about 0.95 | Lincoln 0.95, Hay 0.93 | P(Hay) 0.29 | 0.16 to 0.27, all called Lincoln |
| Burrows's Delta (a standard measure of stylistic distance based on the frequencies of the 150 most common words) | Lincoln 0.87, Hay 0.88 | Lincoln 0.94, Hay 0.89 | Closer to Lincoln (0.54 against 0.59) | All closer to Lincoln, the 1864 condolence included |
| Syntactic markers (the grammatical features Jean Pival examined in 1988) | Lincoln's share of shall and should among the four modal verbs shall, should, will and would is 0.34 in his own-hand war writing and 0.36 before 1860; Hay's is 0.21 in 1860 to 1865 and 0.27 overall. Lincoln uses the passive voice ("I have been shown") twice as often as Hay (138 against 74 per 10,000 words) and the perfect tense ("have died") less (63 against 86). | One "should," two passives, three perfects: consistent with either man | Not tested | |
A fourth check trains the same character classifier to choose among Lincoln, Hay and the twenty control writers. It ranks Lincoln first for the letter (0.12, with Hay at 0.05), first for all four letters Hay composed for Lincoln's signature, and first for Hay's own condolence; it also ranks Seward first for the Gettysburg Address and Lowell first for the McCullough letter, so it has nothing reliable to say about elevated prose and is not counted.
The last column is the important one. Hay wrote a condolence letter of his own in June 1864, and he composed at least four letters that went out over Lincoln's signature in 1863 to 1865 (to Boker, Garrison, Driggs and Charles Butler). The classifiers and Delta call all five Lincoln's. Own-hand n-gram tracing calls three of the four letters for Lincoln's signature Lincoln's by the majority vote; Garrison goes to Hay by margins of 0.004 to 0.021 at 4 to 9 characters and to Lincoln above that. It calls Hay's own condolence Lincoln's, four votes to three. The 19 documents in Basler's edition of 50 to 200 words that survive in Hay's hand, most of them telegrams and orders that Hay may have composed or only copied, go to Lincoln 18 times out of 19 under the own-hand design and 13 times under Grieve's. The methods are good at telling Hay's diary from Lincoln's telegrams. They have not been shown to detect Hay writing as the President, which is the hypothesis in question.
The margins can be read directly. The strings of the letter that occur in Lincoln's own-hand war writing and in none of Hay's are the office's: "in the files of the War Department," "a statement of the Adjutant General," "that may be found in," "should attempt to," "of mine which," "and leave you," "tendering to you," "and respectfully." The strings that occur in Hay's writing and in none of Lincoln's are the elegy's: "the altar of," "solemn pride," "gloriously on the field of battle," "the anguish of," "the consolation that," "from the grief of," "any words of mine," "refrain from," "memory of the loved." Nearly all of Hay's are in his post-war books, not his war-time letters. The letter's business idiom is Lincoln's office idiom, which any drafter in that office would have used; its elegiac idiom is one that Lincoln's surviving writing hardly ever had occasion for.
| Word in the letter | Lincoln own hand 1860–65 (126k words) | Lincoln all autograph (193k) | Hay 1860–65 (54k) | Hay all (500k) |
|---|---|---|---|---|
| beguile (all forms) | 0 | 0 | 1 | 8 |
| gloriously | 0 | 0 | 2 | 4 |
| cherished | 0 | 0 | 1 | 21 |
| solemn | 2 | 2 | 5 | 41 |
| altar | 0 | 0 | 0 | 20 |
| anguish | 1 | 1 | 0 | 20 |
| consolation | 2 | 3 | 0 | 11 |
| bereavement | 1 | 1 | 0 | 2 |
| assuage | 0 | 0 | 0 | 0 |
| tender as a verb (tender you, tendering to) | 22 | 24 | 0 | 1 |
| attempt | 25 | 37 | 0 | 25 |
| Heavenly Father | 0 | 0 | 0 | 0 |
| Father in Heaven | 1 | 1 | 0 | 0 |
Raw counts. Lincoln's "our Heavenly Father" occurs in the proclamation of 10 April 1862 and the reply to Mrs Gurney of October 1862, neither of which survives in his hand and neither of which is in the volumes available to us; Bullard prints both. "Beguile" does not occur in Lincoln's Collected Works at all (Burlingame 1995); Hay used it at least thirty times in his life (Burlingame 1999), once in his war-time letters and twice in the "Colonel Baker" essay.
The word counts are single digits on both sides. "Beguile" is the strongest single item for Hay and has been since Burlingame noticed it, and it remains one word in a register in which Lincoln's surviving prose amounts to three letters. "Tender" as a verb, which Burlingame also claimed for Hay, is Lincoln's: 22 instances in his own-hand war writing against none in Hay's contemporary letters and one in all of Hay's half-million words, where the word appears 74 times as an adjective.
On ordinary texts the answer is "a great deal": at 100 to 200 words the methods are 95 to 100 percent accurate. On texts of this letter's register the same methods are 73 to 95 percent accurate, the two designs give opposite answers for the letter, and the genuine Hay condolence and the four genuine Hay-for-Lincoln letters pass as Lincoln under every method but one, which catches one of the four. Taking the methods together, the likelihood ratio the stylometry supports for Hay against Lincoln lies somewhere between one to three and three to one. We treat it as one. On our tests the 2019 result depends on the corpus design as much as on the letter. A second implementation, written independently, agrees on the direction; the size of the register effect depends on the Lincoln pool, as shown above. Neither implementation used the paper's own corpus.
The question has three answers: Lincoln composed the letter; Hay composed it and Lincoln signed it, perhaps after touching it; or someone else in the office did. The third is a residual. Nicolay was ill in St Louis from 17 to 30 November, and the clerks Hay mentioned to Mrs Langford are recorded only as writing the bodies of ordinary letters. We give it about 5 percent and do not analyse it further.
Between the first two the evidence is documentary, and each item is weak. We set out the items with the weight we give each, as a likelihood ratio: a number saying how many times more likely the item would be if one man wrote the letter than if the other did (a ratio of 2.4 for Hay means the item is 2.4 times more likely under Hay's authorship). A reader who weighs an item differently can see what changes. The ratios are judgments. The starting point, the prior before any item is counted, is the practice of the office as the record shows it: Lincoln wrote the letters that mattered to him personally, and the two condolence letters that survive from his presidency, to the Ellsworths in 1861 and to Fanny McCullough in 1862, are in his hand, while Hay wrote most of the rest and was writing routine letters for him that same week. We start at 65 percent for Lincoln.
| Item | What it is | Ratio | Running probability, Lincoln against Hay |
|---|---|---|---|
| Starting point | The office's practice: a governor's request through the Secretary of War, for a letter of a kind Lincoln wrote himself, in a week when Hay was drafting the routine ones | — | 65 |
| Hay's reported claims | Three chains (Morley to Butler; Brownell to Ogden; Page to Lady Strafford to Jackson to Lucas), none first-hand, none written down before 1922, two triggered by framed copies of the letter, one carrying a checkable error; plus two anonymous ones | 2.4 for Hay | 44 |
| Hay to Chandler, 1904 | "The letter of Mr. Lincoln to Mrs. Bixby is genuine." Consistent with either reading; Hay was answering a question about the forged facsimile | 1.2 for Lincoln | 48 |
| Hay to Robert Lincoln | Hay "knew nothing whatever of the source of the printed copy" and had "never seen any draft or copy," reported by Robert in 1908 and 1917. A man who drafted the letter would have known its source. Hay could have chosen not to tell Lincoln's son | 2.2 for Lincoln | 67 |
| Silences | Nothing in Hay's diaries, letters, memoir essay or family; his co-edited edition of Lincoln's Complete Works (1894) prints the letter as Lincoln's with a note on "its evident sincerity"; Spender, who discussed it with Hay, could not trace the story to him | 1.4 for Lincoln | 74 |
| The scrapbooks | A clipping of the letter among Hay's own poems and reviews at Brown, and another at the Library of Congress beside four other Lincoln letters; the same books hold material Hay did not write | 1.5 for Hay | 66 |
| "Beguile" | A word Hay used at least thirty times and Lincoln never, in a register Lincoln's surviving writing hardly touches | 1.5 for Hay | 56 |
| Stylometry | Section 4: the confound-controlled methods disagree and cannot detect known Hay writing in Lincoln's voice | 1 | 56 |
| Result | Lincoln against Hay, 56 to 44. With the 5 percent held back for a third hand: Lincoln composed it, about 53 percent; Hay, about 42; someone else, about 5 | 56 |
With the stylometry set at a ratio of 1, the case for Hay rests on the reported claims and the scrapbook clippings, the case for Lincoln on the office's practice and on what Hay told Robert Lincoln, and the two are close to a balance.
| If you instead… | Lincoln % | Hay % |
|---|---|---|
| take the 2019 stylometry at face value (a ratio of 10 for Hay) | 11 | 84 |
| give the own-hand stylometry its most Lincoln-ward reading (3 for Lincoln) | 75 | 20 |
| believe Butler's account of Morley exactly (a ratio of 8 for Hay in place of 2.4) | 26 | 69 |
| discount every reported claim to nothing | 72 | 23 |
| read Hay's statements to Robert Lincoln as discretion, not information (ratio 1) | 35 | 60 |
| start from an even prior instead of 65 percent | 39 | 56 |
| start from 75 percent, on the strength of the two autograph condolences | 64 | 31 |
Each row changes one item and keeps the others; the residual 5 percent for a third hand is held fixed. The first row is where the debate stood after 2019; the second is what the stylometry would say if only the classifier and Delta results were kept.
Michael Burlingame built the case for Hay in 1995 and 1999, and the 2019 study added a statistical argument to it. Burlingame's articles and the 2019 study were read in full; his 2000 chapter in At Lincoln's Side could not be obtained and is cited through Emerson and through Burlingame's own later summaries of it. The case has ten parts. Each is stated here as its advocates state it, followed by what we found.
The case for Hay is cumulative and circumstantial. Its testimony is hearsay throughout. Its two strongest items, the scrapbooks and the 2019 study, each fail a specific test. It remains a respectable hypothesis: nothing we found excludes it, and the verdict above gives it about two chances in five.
The verdict is a lean of 56 to 44 between two men, on documentary evidence that is weak on both sides. Small additions to the evidence would move it; the following would move it substantially.
Two things would not change the verdict. More reported confessions of the same kind, at two or more removes and written down decades later, would add little, because the story was in circulation by 1904 and the chains share a source. A further n-gram study under the 2019 design would add little, because both implementations of our tests find that design misreading between roughly a quarter and a half of Lincoln's elevated prose (28 to 48 percent, depending on the pieces and the pool). A rerun on the paper's own corpus, by a third party, is the check that remains.
No new evidence is presented here: no document was found that was not already in print. The documentary case on both sides was assembled between 1946 and 2008. What is new is analysis of old evidence, mainly the added tests of the 2019 design. A second implementation of those tests, written independently, reproduces the attribution, finds the register effect in the same direction, and finds it smaller (section 4).
We found no published replication, re-analysis or rebuttal of the 2019 attribution between its appearance and September 2026, in the journals, in the 41 works citing it that the Semantic Scholar database lists, or in the Lincoln periodicals we could reach. The only independent test of the method is Proisl and colleagues (2018), on literary texts, published before the journal version of the study. A new book, Jan Cigliano Hartman's Lincoln's Speechwriter (2026), argues for a large Hay hand in Lincoln's prose generally; the Lincoln scholar Harold Holzer's review of it (on the history website Emerging Civil War, 23 July 2026) calls the Bixby question open and names the scrapbook as the principal argument for Hay. We did not read the book. The journal and book databases Project MUSE, JSTOR, HathiTrust and Google Books could not be searched, so the search is not complete.
collectedworksof015581mbp, collectedworksof015582mbp, collectedworksof015583mbp, collectedworksof015584mbp. Vols II, III, V and VI are lending-only on archive.org and the University of Michigan online edition refused the connection; their absence is a stated limitation of the Lincoln corpus. The Bixby letter and note: VIII, 116 to 117.lettersofjohnhay01hayj, 02, 03; William Roscoe Thayer, The Life and Letters of John Hay (1915), 2 vols, lifelettersofjoh01thay, 02; John Hay, Castilian Days (Project Gutenberg 7470), The Bread-Winners (16321), Pike County Ballads (6062), Poems (10518), Addresses of John Hay (1906, archive.org addressesofjohnh00hayj).massinthecivilwar01schorich; Annual Report of the Adjutant General of Massachusetts for 1864, archive.org annualreportofad1864mass.acrossbusyyearsr02butl; John Morley, Recollections, II (1917), archive.org recollections0002morl.caseoflydiabixby00bann.abrahamlincolnwi00bull.abrahamlincolnwi00wake, abrahamlincolnbi00wake.myfirst, items 1946 and 3761), which serves page images without a text layer; we ran optical character recognition on the issues and read the articles from the output.independent-rerun/.Everything below is saved with this report. The corpus builder (corpus/scripts/parse_basler.py) reads Basler's source notes to tag every item with its manuscript code (ALS, ADfS, LS, Copy and so on) and to flag items whose surviving text is in Hay's hand. The analysis scripts (analysis/ngt.py, run_ngt.py, designs.py, run_other.py, features.py) reproduce every number in this report from the JSONL corpora; analysis/results/ holds the raw outputs; analysis/diag_gap.py and analysis/results/diag/ hold the runs that trace the gap between the two implementations' register results. The dossier folder holds the reading notes on the documentary record with page citations for every quotation used above.
The documentary record was read in full from the editions named above; the stylometry was re-implemented from the 2019 paper and validated on known texts before the letter was tested. Every number in sections 3 to 5 can be regenerated from the saved corpora and scripts.