The first letter from Jack the Ripper, referred to as the Dear Boss letter. Photo: John Bennett

A forensic linguistics expert who has worked on some of the most famous cold case mysteries in history has published new evidence about the Jack the Ripper letters. Specifically, Andrea Nini has connected two of the most famous letters together via a common author, as well as linked them to a third letter that was largely believed to be a hoax.

Jack the Ripper is the name given to the unidentified serial killer who is believed to have murdered at least five women in the Whitechapel district of London in 1888. In the approximately three months spanning the first five murders, letters discussing the homicides were sent to the Central News Agency, purportedly from Jack the Ripper, the self-named serial killer.

However, many believe the letters did not come from the murderer, rather from journalists trying to drum up newspaper interest. Once the first four letters were published in the paper, hundreds of copycat letters were received. Therefore, Nini focused his work on two of the earliest letters—the “Dear Boss” letter, in which the name Jack the Ripper first appeared, and the “Saucy Jacky” postcard.

This past summer, Nini, a professor at the University of Manchester (UK), used a method called n-gram tracing to solve a longstanding mystery about a letter Abraham Lincoln may have written. N-gram tracing looks at the presence or absence of n-gram sequences—a sequence of one or more elements, such as words, characters, grammatical structures, etc.—to establish authorship, rather than just words’ numerical frequency.

The method Nini used on the Jack the Ripper letters is similar, but not exactly the same. He utilized what is called the Jaccard coefficient, a statistic used for comparing the similarity and diversity of sample sets, often in biology.

“It has already been used in forensic psychology to link crimes based on similarities of features,” Nini explained to Forensic Magazine. “Recently, it’s been used in forensic linguistics as well, and it seems to be very useful in the cases of short texts [like these letters]. I used the Jaccard coefficient to see the similarities in the text, and then I used the cluster analysis to organize, visualize and explore those similarities.”

Of the four letters received before mass publication, Nini found two for which there is evidence of common authorship.

One specific combination of words allowed Nini to connect authorship—the phrase “letter back till I.” The combination of words is so rare Nini could not find it in any other authored text, even when he googled it.

In the Sept. 25, 1888, Dear Boss letter, the phrase appears as the second-to-last sentence: “Keep this letter back till I do a bit more work, then give it out straight.” In the Oct. 1 letter, the author repeats the phrase: “thanks for keeping last letter back till I got to work again.” Both letters also share the use of the verb “work” to indicate the act of killing.

Altogether, Nini said he found eight similarities between the two letters.

In another text deemed the “Moab and Midian” letter—written post-publication—Jack the Ripper writes to say he did not kill the female whose body was found most recently, but that he is planning three murders tomorrow. The last line of the text reads, “Keep this back till three are wiped out and you can show the cold meat.”

Nini said this is the “only instance across all the [texts] to exactly match the syntactic structure” of the Dear Boss and Saucy Jacky letters—leading the professor to tie this letter’s authorship to the other two, as well.

The forensic linguist is still working on improving the n-gram tracing method he pioneered when working on the Bixby letter this past summer. But this Jack the Ripper research has only informed this work. More than anything, Nini is excited about the dataset created from his work on the letters.

“It’s a very good resource for forensic linguistics,” he explained. “It’s like an experiment to see how good people are at imitating others. We have a dataset of all of these hoax letters pretending to be Jack the Ripper. They are all copying and pretending to be someone else, which is the author of the first letter, of the Dear Boss letter. They are all trying, but you can see when you look at it computationally, they are still very different from the original letter. Even when there are similarities, they are very non-distinctive. From that, we can find evidence that it is very difficult to fake being someone else linguistically.”

Although Nini does not intent, at the moment, to go any further in his Jack the Ripper research, he’s adamant that the potential is there, given the dataset. For example, now that two texts are tied to each other, a researcher could use linguistic techniques to create a profile of the author, including social background information, education level, etc.

Although it’s been more than a century since the murders and letters, the great thing about linguistics is it can’t be degraded—not in the way DNA or fingerprints can.

“Any ‘new’ evidence found in the case is corrupted because it’s been too long,” Nini said. “But not linguistics. I’m not looking at the paper or the ink or anything physical. I am just looking at the words.”