John Hay Wrote the Bixby Letter

By E. L. Core

January 10, 2021

The Letter in Question

This letter of condolence to Mrs. Lydia Bixby, Nov. 21, 1864, has been called America’s Most Famous Letter:

Dear Madam, — I have been shown in the files of the War Department 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.
I feel how weak and fruitless must be any words of mine which should attempt to beguile you from the grief of a loss so overwhelming. But I cannot refrain from tendering to you the consolation that may be found in the thanks of the Republic they died to save.
I pray that our Heavenly Father may assuage the anguish of your bereavement, and leave you only the cherished memory of the loved and lost, and the solemn pride that must be yours, to have laid so costly a sacrifice upon the altar of Freedom.
Yours, very sincerely and respectfully, A. Lincoln.

Massachusetts governor John A. Andrew had requested a presidential acknowledgement of the circumstances of Mrs. Bixby’s family, as he understood them to be, through the agency of the Adjutant General of Massachusetts, Gen. William Schouler; the letter of condolence was hand-delivered to Mrs. Bixby by Gen. Schouler, after which he provided copies to newspapers.

But did Abraham Lincoln actually write this letter? Or was it written by his assistant private secretary, John Hay? The controversy over authorship has raged for nearly a century now, arguments on one side or the other pointing to personal testimonies, judgements about style, and a whole mess of guesswork.

Recently, though, a new computerized algorithm for determining the authorship of short texts has been aimed at the Bixby Letter, and it comes up with John Hay as the author. We will look at that after some other considerations.

We Know that Pres. Lincoln Did Not Write All of the Letters He Signed

At first glance, one might wonder why there is any doubt whatsoever about Lincoln’s authorship of the letter: after all, he signed it. Right?

True. But we know that Pres. Lincoln signed letters and other documents that he had not written himself. The most famous of these documents — and this may come as a schock to many people — is his oft-quoted 1863 Thanksgiving Day Proclamation. We know with rock-solid certainty that it was written by William Seward, the Secretary of State, as indicated in the footnote in Roy Basler’s 1953 nine-volume edition of The Collected Works of Abraham Lincoln (Volume VI, p. 497):

The original draft of this proclamation has not been located, but a letter from John G. Nicolay to John Hay from New York, April 1, 1864, relates that “the Mss. [manuscript] of the President’s Thanksgiving Proclamation, which was written by Seward and is in his handwriting” had been sent by the State Department to Leavitt Hunt....

Nevertheless, John G. Nicolay (Pres. Lincoln’s private secretary) and John Hay included that proclamation in their two-volume 1894 edition of Complete Works of Abraham Lincoln (Volume II, pp. 417f), and it was still included in the 12-volume New and Enlarged Edition (Volume IX, pp. 151ff) published in 1905.

And John Hay himself is known to have written letters signed by the president: these two letters, for instance, are in John Hay’s handwriting: June 9, 1863 and June 25, 1863; and Hay recorded in his diary that he had written this October 24, 1863, letter.

But we have known since the beginning of Lincolnian scholarship that the president did not himself write all of his letters; Lincoln’s law partner and biographer William Herndon quotes an 1866 letter from John Hay about the presidential correspondence:

He wrote very few letters, and did not read one in fifty that he received. At first we tried to bring them to his notice, but at last he gave the whole thing over to me, and signed, without reading them, the letters I wrote in his name. He wrote perhaps half-a-dozen a week himself — not more.

Perhaps Hay was engaging in hyperbole. More than 50 documents, though, are said to be in “John Hay’s handwriting” in the Collected Works: fewer than ten, though, are more than perfunctory one- or two-sentence notes.

(Of course, handwriting does not necessarily prove or disprove authorship. One long letter, for example, exists as a draft written and signed by the president, though the final version, excepting the signature, is in another hand — probably Hay’s, who wrote about it in his diary.)

On the Question of Writing Style

So, we know that Pres. Lincoln did not write all of the letters and other documents that he signed. Scholars, therefore, on both sides of the debate concerning authorship of the Bixby Letter have tried to analyze the style in which the letter was written, to compare and contrast with Lincoln’s and Hay’s writing styles.

In general, the debate goes this way: Lincoln never used this word, but Hay used it often —or vice versa. And this phrase is used here and there by Lincoln, never by Hay — or vice versa; and this other phrase is almost-but-not-quite used by one or the other writer in one or another circumstance — or never, ever; and, so forth, et ceteraad nauseam.

Unfortunately, that kind of analysis is inherently subjective and necessarily selective; it involves the conjoined traps of finding only what one wants to find while missing what one doesn’t want to notice.

Some times, the speculation devolves into the personal: that young whipper-snapper John Milton Hay simply couldn’t have had at age 25 the depth of feeling and understanding necessary to write the profound, compassionate Bixby Letter. But, we cannot really know that, can we?

Moreover, both Nicolay and Hay lived in the White House. Does not this kind of judgement ignore the influence that having lived and worked daily for years with Abraham Lincoln — whom Hay revered as “the greatest character since Christ” — might have had on the young man?

No, stylistic and personal analysis is just so much gussied-up guesswork — especially coming from people who really, truly, deeply want to believe the letter was written by one man or the other.

Already mentioned above, an objective kind of stylistic analysis would avoid the subjective pitfalls. Before we look at this new computerized algorithm, let’s look at a couple of other objections to Hay’s authorship.

Did Hay Deny Writing the Bixby Letter?

Some scholars have called attention to two passages in letters written before 1920 as settling the issue in favor of Lincoln’s authorship of the Bixby Letter. Let’s look at each in turn.

1) A 1904 Letter from John Hay

In the 1890s, engraved copies of the Bixby Letter began to be sold, with the claim that they were true copies of the genuine original in Abraham Lincoln’s handwriting with his signature. The authenticity of these claims was disputed, and that dispute gave rise to questions about whether the Bixby Letter originally published in newspapers on Nov. 25, 1864, might have itself been some sort of hoax.

Responding to an inquiry, John Hay wrote the following letter, Jan. 19, 1904:

The letter of Mr. Lincoln to Mrs. Bixby is genuine, is printed in our edition of his Works, and has been frequently re-published; but the engraved copy of Mr. Lincoln’s alleged manuscript, which is extensively sold, is, in my opinion, a very ingenious forgery.

That Hay called the letter “genuine” is seized upon as an admission that he did not write it, that Abraham Lincoln wrote it. But this is a tendentious reading.

For Hay also pointed to the letter’s inclusion in Nicolay’s and Hay’s 1894 collection of Lincoln’s Works as evidence of their acknowledgement that the Bixby Letter is “genuine”. Yet, we know for certain they included other documents written by Hay, including those already mentioned above; and, they included the 1863 Thanksgiving Proclamation — and that document was certainly written by Sec. William Seward.

Thus, by “genuine”, Hay must have meant — or, at the least, may have meant — no more than the Bixby Letter was not a hoax, but was indeed a document sent out over the president’s name with the president’s approval.

2) A 1917 Letter from Robert Todd Lincoln

[In this section, the Lincolns, father and son, are referred to by first name to readily distinguish them.]

Robert Todd Lincoln was the eldest son of Abraham and Mary Todd Lincoln, their only child to survive to adulthood. From 1903 to 1926, Robert corresponded with retired journalist Isaac Markens, who published several small studies of Abraham in that timeframe. Among their numerous topics of conversation was the Bixby Letter.

Markens’s letters to Robert have not, as far as I know, seen the light of day, but most of Robert’s dozens of letters to Markens have been published. Robert discussed various aspects of the Bixby Letter in his letters to Markens dated Jan. 5, Feb. 19 & Feb. 24, 1917; and, dated Jan. 4 & Jan. 25, 1918. These were the topics of discussion:

Authorship of the Bixby Letter was not discussed by Robert in the published letters. It is clear, though, from other sources that Robert did firmly believe his father had been the author.

Here is the passage cited by scholars disputing Hay’s authorship of the letter, the opening paragraph of Robert’s letter of Feb. 24, 1917 (emphasis added):

I think I have not acknowledged your letter of February 20th, in regard to the Bixby 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; when I took the matter up Nicolay had died and it was he who had compiled the collection of papers. It is entirely possible that neither of them knew of the letter at all; my father had no letter books and copies of his letters and documents were only made in special cases, many such copies being in the papers I now have, mostly drafts in his own hand — it is entirely possible that my father wrote this letter at his desk, folded it, addressed it and gave it to General Schouler without anybody else about him knowing of it.

“Hay himself told me so” is proferred as a specific denial by Hay of having any knowledge of the Bixby Letter at the time it was written — which is, naturally, taken as a denial by Hay that he had written it. As Jason Emerson says, “John Hay himself told Robert he’d had nothing to do with it”. This too is a tendentious reading.

For we do not know the exact nature of Markens’s “suggestion”, only Robert’s paraphrase of it; nor do we know what Hay had actually said, only Robert’s re-telling of it; nor do we know when or why Hay had been prompted to say whatever he said — we only know that Robert said that Hay said he had “no special knowledge” of the letter “at the time”.

Indeed the rest of the paragraph undercuts Emerson’s peremptory claim, does it not? “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.... It is entirely possible that neither of them knew of the letter at all....” If Hay had meant that he and Nicolay “had nothing to do with it” (Emerson), why would Robert Lincoln have to opine just a sentence later that “It was entirely possible neither of them knew of the letter at all”? Clearly, Robert’s conversation with Hay had not given Robert a clear notion of the letter’s genesis.

Besides, what is this “special knowledge”? (Robert’s phrase.) One may presume “special knowledge” is to be distinguished from mere knowledge. How? The topics of discussion between Markens and Robert were the existence and location of the original letter, and the genuineness of the engraved copies.

Hay, then, might very well have meant only that he did not know or remember whether the letter sent to Mrs. Bixby was in Abraham’s handwriting, or whether the president had only signed it, or whether anybody in the White House had made an official or unofficial copy at the time the letter was written. Or, Hay may have meant neither he nor Nicolay knew how Mrs. Bixby had been brought to the president’s attention, nor had been involved in discussions about the letter with Pres. Lincoln and Gen. Schouler, or knew how the letter got into the newspapers, or what had become of the original. Or, all of these, or any number of other circumstances that might qualify as “special” knowledge.

Moreover, what time is meant by “at the time”? (Robert’s phrase.) Considering the topics of discussion between Markens and Robert, for all we know, Robert had been talking with John Hay, at some time or other, about whether the original still existed and/or whether the engraved copies were genuine or forgeries — and by “at the time”, Robert may very well have meant at the time Robert and Hay were discussing those topics.

Computerized Stylometric Analysis

Since all of the judgements of the authorship of the Bixby Letter discussed so far are subjective, tendentious, or otherwise inconclusive, a more objective approach is needed.

A New Methodology Especially for Short Texts

Recent developments in stylometric analysis (stylometry) have been applied to attribute authorship of the Bixby Letter. The scholars involved were all academics at universities in the United Kingdom, publishing their study as Attributing the Bixby Letter using n-gram tracing (26 October 2018).

Their first step was to determine the corpus of works by Lincoln and the corpus of works by Hay against which the Bixby Letter could be analyzed. Lincoln’s texts were limited to those he had written before he hired John Hay as his assistant private secretary, to avoid unwitting inclusion of a work actually written by Hay (or somebody else, such as Sec. Seward or Lincoln’s other secretaries); no such cutoff point was required for Hay’s works (p. 497).

For Lincoln, 1,085 works were selected, comprising 400,747 words, with texts ranging in length from 5 to 17,003 words and with a median length of 125 words per text; for Hay, 577 works were selected, comprising 261,126 words, with texts ranging in length from 9 to 8,954 words and with a median of 159 words per text. Thus, the total number of texts was 1,662, comprising 661,873 words (p. 497).

Then, to test the accuracy of the new methodology in attributing texts to Lincoln or to Hay, each of the 1,662 texts was extracted one at a time for comparison against the other works of both authors. For all 1,662 texts, the algorithm was able to attribute the correct author every time. That is, every work by Lincoln was attributable to Lincoln, and every work by Hay was attributable to Hay (p. 504).

Finally, the algorithm attributes authorship of the Bixby Letter to John Hay (p. 506).

A Closer Look at the Algorithm

The researchers call their new methodology n-gram tracing. By n-grams, the authors mean a certain number (n) of words or characters (grams); their search for certain n-grams in sample comparison texts is tracing.

For instance, “shown” is a 1-word gram and a 5-character gram; “I have been shown” is a 4-word gram and a 17-character gram (counting punctuation and spaces).

Here is the authors’ summary of the steps taken in their analysis (p. 498):

  1. Extract all n-grams of a particular length and level from the questioned document.
  2. Take a random sample of texts of equal size from each possible author writing sample.
  3. Measure the percentage of n-gram types found in the questioned document that also occur at least once in each possible author writing sample.
  4. Attribute the questioned document to the possible author who uses the highest percentage of these n-grams.

For example, using the opening words of the Bixby Letter, a 2-word analysis would use the following grams: “I have”, “have been”, “been shown”, “shown in”, “in the”, “the files”, etc. And, a 3-character trace would use the following grams: “I h”, “ ha”, “hav”, “ave”, “ve ”, “e b”, “ be”, “bee”, “een”, “en ”, “n s”, “ sh”, “sho”, “how”, “own”, etc.

Here’s how the algorithm works in our case of the Bixby Letter and the works of Lincoln and Hay. Each of the grams would be compared to like n-grams in the works of the possible authors. For example, for 2-word grams, the algorithm searches for occurrences of “I have”, “have been”, “been shown”, “shown in”, “in the”, “the files”, and so forth, in random samples of text in Lincoln’s and Hay’s works. The same is done, mutatis mutandis, for character-level grams. The length of the sample texts is random and can be very large, limited only by the length of the shorter corpus of works — in our case, 261,126 words, the number of words in Hay’s corpus (compared to Lincoln’s 400,747 words).

The researchers discovered that analyses using 4-12 characters and 1-3 words provided the most accurate attribution of the total 1,662 texts from both authors — that is, those traces attributed each and every text to its known author. Shorter and longer grams are inapt: shorter, because too many matches are found in the comparison texts; longer, because not enough matches are found (pp. 505f).

Further analyses using 3-16 characters and 1-3 words — 17 distinct traces — attributed the Bixby Letter to John Hay every time (pp. 506ff).

The Objective Nature of the Methodology

How is n-gram tracing more objective than other kinds of stylistic analysis? The authors of the study explain the subjective nature of typical stylometric analysis (pp. 497f):

In forensic linguistics, short texts are often attributed by manually selecting linguistic features from the questioned document that appear to be relatively distinctive or rare and by then searching for these forms in the writing samples of each possible author. Although this method is logical and is regularly applied in casework, there are at least three potential issues with its application.
First, it is unclear how to select an exhaustive or at least an unbiased feature set, as the debate around the style of the Bixby Letter illustrates: different analysts can identify different sets of seemingly distinctive features and consequently come to different attributions of the same questioned document.
Second, it is unclear how to control for variation in the amount of material in the possible author writing samples, which often varies tremendously, as is the case here: if more text is available for one of the possible authors, then the forms extracted from the questioned document have an increased chance of being found in that author’s sample regardless of authorship.
Third, it is unclear how to judge whether differences in the use of forms in the possible author writing samples are sufficient in the aggregate to attribute the questioned document: because this approach relies on the judgment of the analyst and therefore cannot be consistently or mechanically applied, it is difficult to systematically evaluate the reliability of such methods.

N-gram tracing is more objective than other methodologies on two counts.

First, the amount of sample comparison text can be tremendous: in the case of Lincoln and Hay, as already mentioned, up to 260,000 words can be, and have been, used for comparison. This avoids subjective judgements about what texts to use for comparison.

Second, texts can be attributed accurately without regard to the meaning — or lack of meaning — of the textual fragments being analyzed. The researchers discovered that 7-8 character grams (e.g., “ed glor”, “artment”, “reavemen”, and “le you f”) and 2-word grams (e.g., “and the”, “who have”, “of a”, and “be any”) provided the most consistently accurate traces for Lincoln’s and Hay’s works. Such short fragments of text often, if not usually, have no meaning in and of themselves.

Why Does This Debate About Authorship Exist?

We have now seen that subjective arguments about authorhip of the Bixby Letter are inconclusive — but a more objective methodology determines the author to have been John Hay. We will now take a look at one more aspect of this issue, which explains why there is any debate at all about who wrote the letter.

In the first half of the 20th Century, reports surfaced that several individuals claimed John Hay told them he had written the Bixby Letter; among this handful of persons are the following:

These reports have been dismissed for decades by proponents of Lincolnian authorship as being too remote and sketchy to be reliable — if not outright fantasies or falsehoods.

But such dismissals, no matter how well-argued, left a question unanswered: why in the world would anybody claim Hay had said he wrote the Bixby Letter if Hay had not actually said he wrote the Bixby Letter? Especially since these sources, as far as we can tell, are completely independent of one another.

The results of n-gram tracing now answer that lingering question: John Hay told people that he wrote the Bixby Letter because he had, indeed, written the Bixby Letter.


© 2021 E. L. Core.