I’ve noticed a number of things about the Paul Rubin ciphertext, which are definitely beyond what the FBI’s cryptanalysts managed to find. Please feel free to advance these yet further!

The Paul Rubin Ciphertext

Here’s my transcription, as posted on the Cipher Foundation website, but with the individual lines numbered:

[01] digIs sawthn'g mathUlley-Dulles crancklavn' meteore iElli
[02] zheaopfvamn greA'Lltenmn
[03] kKiqtu albawmnabs dzhjellEiE matel ungdreabozvmie oie
[04] sprekln meIktrene fodroscolmn oeir
[05] *driEk Conant astereantol Iyvondiolon
[06] desceth megleagna mAlzbourgnion grele

[07] newtdo sfoatzdexklagh 2pont ¼ly asgestaltverbensdi

[08] 7469921
[09] 100.011x100.10x.10011.1.xx0.101.x.001011.101x1011.1001..10x1

[10] 01.001011x10.1x.11101.x1.001x1.001001

[11] 0.101.x.101110.x101.1101101.0101x1.1011

[12] Want: datum Tywood Janossey Ketelle

[13] R-QR6
[14]                aliacaui PER

The Last Two Words

It seems almost certain that “PER” stands for “Paul Emanuel Rubin”, which gives us a reasonable amount of confidence that this is a real ciphertext Rubin himself had made.

Rubin liked reading science fiction, and repeatedly claimed to be a member of a “Brooklyn Astrophysics Society”: however, despite asking a lot of people, the FBI were unable to turn up any reference to any such society – Rubin seems to have dreamed this up completely. So there is (I think) good reason to suspect that we might find obscure references to science fiction and/or imagined science embedded in this ciphertext.

And in fact this is exactly what we find on line [14]. The word immediately before PER – “aliacaui” – comes from a 1950 novelette by Poul Anderson called “The Helping Hand“, that first appeared in Astounding Science Fiction:

Valka Vahino sat in his garden and let sunlight wash over his bare skin. It was not often, these days, that he got a chance to aliacaui… What was that old Terrestrial word? “Siesta”? But that was wrong. A resting Cundaloan didn’t sleep in the afternoon. He sat or lay outdoors, with the sun soaking into his bones or a warm rain like a benediction over him, and he let his thoughts run free. Solarians called that daydreaming, but it wasn’t, it was, well — they had no real word for it. Psychic recreation was a clumsy term, and the Solarians never understood.

In the story, “Aliacaui” is a central concept to the Cundaloans, one which the hard-working Solarians (machine-culture humans) find difficult to grasp:

“For instance, just this matter of the siesta. Right now, all through this time zone on the planet, hardly a wheel is turning, hardly a machine is tended, hardly a man is at his work. They’re all lying in the sun making poems or humming songs or just drowsing. There’s a whole civilization to be built, Vahino! There are plantations, mines, factories, cities abuilding — you just can’t do it on a four-hour working day.”

So this is the single word from the Science Fiction universe that Paul Rubin valued so much that he used it as a farewell in his note: “Aliacaui”, meaning lying around “making poems or humming songs or just drowsing”. Really, you couldn’t make it up.

The Codebook Indices

I’d also immediately point out that lines [01]-[07] almost certainly each use a different codebook entry, and that the seven codebook indices for them are on line [08] – 7469921. It seems likely (though not yet proven) that these are the indices in the order the lines appear.

Furthermore, lines [09]-[12] each almost certainly use a different codebook entry, and that the four codebook indices for these are on line [13] – R-QR6.

Hence, the three “binary”-like lines would seem to be codebook entries “R-QR”, while all the other lines’ codebook indices are digits: 7469921 and 6. I pointed this out to Craig Bauer before his book came out, but it was (alas) too late to update his Paul Rubin chapter.

Still, knowing this should help us avoid many pointless cryptanalytical tests: for example, there would seem to be no point in carrying out any test that combines a binary-like line with a text-like line, because they would seem to be using completely different types of codebook.

The Three Binary-Like Lines

If the 0s and 1s in these lines are binary, I noticed something a little odd about them: specifically, that they contain are no instances of ‘000’, and only two instances of ‘111’.

I therefore wondered whether the ‘.’ and ‘x’ character be standing in for ‘000’ or ‘0000’, and ‘111’ or ‘1111’? I converted various permutations for line [09] (the longest binary-like line) into the corresponding streams of pure binary digits, and then ran them through index of coincidence tests online.

[09] 100.011x100.10x.10011.1.xx0.101.x.001011.101x1011.1001..10x1

However, I don’t have a positive result yet (the IoC probably isn’t the most reliable test for this kind of thing, but I thought it was worth a try), but if you happen to be looking at this part of the ciphertext, I think this currently seems like the most likely route to an answer.

The Letter Ciphers

According to Cipher Mysteries commenter Thomas, “Conant and B.H. Ketelle were members of the Manhattan Project. Janossy was a Hungarian nuclear physicist at the same time. Tywood is a professor of Nuclear Physics in Isaac Asimov’s short story ‘The Red Queen’s Race’ from 1949.”

Indeed, Asimov describes Elmer (Pop) Tywood in his short story as “Ph.D., Sc.D., Fellow of This and Honorary That, one-time youthful participant of the original Manhattan Project, and now full Professor of Nuclear Physics.”

Unfortunately, “Elmer Tywood was dead. He lay next to the table; his face congested, nearly black. No radiation effect. No external force of any sort. The doctor said apoplexy. […] In Elmer Tywood’s office safe were found two puzzling items: i.e. twenty foolscap sheets of apparent mathematics, and a bound folio in a foreign language which turned out to be Greek, the subject matter, on translation, turning out to be chemistry.”

Mysterious deaths and Manhattan Project physicists were therefore at the forefront of Paul Rubin’s mind: my suspicion is therefore that Rubin’s book of words that would drive his code project may well turn out to be a list of names of members of his imaginary Brooklyn Astrophysics Society. We’ll probably never see it, of course: but it is what it is.

Back in 2015, I blogged about the ciphertext that was found taped to Paul Rubin’s stomach, and also wondered whether Rubin might have suffered from paranoid schizophrena, and whether the FBI would ever release his code-tables. On a larger scale, given that we only had a single scratchy newspaper photograph of his cipher to work with, it seemed that we were unlikely to make huge progress.

Well, a lot of that has now changed.

Craig Bauer’s “Unsolved!”

Craig Bauer’s (2017) “Unsolved!” covers a good number of cipher mysteries with a particular focus on Americana, and so his book covers the Paul Rubin case on pp.289-304. Very kindly, he passed me the following (much clearer) scan to work with, that his book had reproduced at a fairly small size:

Hence I’ve added a page on the Paul Rubin Cipher to the Cipher Foundation website: this also includes my transcription of the cipher, as well as (thanks to Albert Mock) a copy of Rubin’s death certificate and details of his grave.

Arguably even more importantly, however, Craig Bauer also received a 160-page document set from the FBI following a Freedom of Information Act request (though this arrived too late for him to use in his book), which is now also linked on the Cipher Foundation webpage – and this is where our real research begins.

Paul Rubin’s FBI File

A number of suggestions as to the possible contents of the cipher appear in the FBI file: that it might be written in cable language, that it might contain cribs for a chemistry examination, and so forth. It also lists (p.25) possibly the first nutty theory about the cipher, courtesy of Mrs L. Rohe Walker, 2 Beekman Place, New York 22, NY.

For me, though, one particularly interesting aspect of the file is that it details (on p.66) the specific sequence that the FBI’s cryptanalysts followed to try to understand Rubin’s ciphertext (though without success). It starts by listing the languages FBI linguists tried: “Yiddish, Hebrew, Russian, German, Hungarian, Finnish, Latinair [?], Lettuishuan [?], Turkish, Portuguese, Rumanian, Italian, Spanish, Dutch, Danish, Norwegian, Swedish, Malayan, Albanian.” They also “[t]ried to develop words phonetically and as abbreviations, with no success.”.

It then moves onto cryptanalysis, firstly listing the “Direct Cipher” methods they looked for (all with negative results):

a. Monoalphabetic Subst.
b. Transposition – uniliteral
c. Partial encipherment with & without nulls
d. Typewriter displacement
e. Combination subst + Transp + nulls + partial encipherment
f. Commercial word codes
g. False language
( 3 books from Library of Congress:
( * “On the Choice of a Common Language”
( * “Method [?] to Esperanto”
( * “A Planned Auxiliary Language”
h. Binary substitution as superencipherment

It then lists the “Open Codes” they looked for, firstly for letters mapped across the whole specimen:

(1) Constant key positions 2-25 (Entire specimen)
(2) 1st, 2nd, 3rd …. final letters of words, initial letters – forward and reversed alphabets
(3) Beginnings & endings of lines (+10 and -10 letters)
(4) Numerical key 7469921 as letter positions
(5) Capital letters, including letters to right & left

Next they looked for words embedded in the entire specimen:

(1) Constant key positions
Also 1st, 2nd letters of constant keyed words 2-10

They moved on to search for possible “Distorted Words”:

2PONT = Dupont
1/4ly = Quarterly
KETELLE = Catelle (psychologist) ?

They also looked for possible names, cross-referencing them in the FBI files (e.g. for B. H. KETELLE):

TYWOOD-JANOSSEY-KETELLE [lists FBI document references]
IVAN DIOLON (negative)

Next, they list various observations made by the cryptanalysts. This for me is the most interesting part by far (note that ‘Q5’ is the reference names for this ciphertext within the bundle of evidence made available to the FBI):

1. Peculiar letter combinations of contacts.
a. Phonetic & pronounceable but unintelligible. Hebrew/Yiddish influence observable.
b. Doubled letters – with “KK” at beginning of one word.
c. “MN” digram, occurring 3 times at end of words.

2. General letter frequency:
A = 21, B = 5, C = 4, D = 11, E = 35, F = 3, G = 10, H = 6, I = 15, J = 1, K = 7, L = 23, M = 10
N = 21, O = 17, P = 3, Q = 1, R = 13, S = 12, T = 15, U = 5, V = 5, W = 3, X = 1, Y = 3, Z = 5
2 = 1

3. “Digits” at bottom of Q5 not in organized form. “X” and “.” appear haphazardly. Makes no sense in attempts to convert binary to digits.

1. Letter frequencies and pronounceability indicate phonetic composition.
a. May be syllable, phonetic, or artificial word code. If so, material is insufficient for analysis.
b. letter material looks like irrational or unsystematic composition

2. Digits have irrational composition. Positions of “X” and “.” do not assist solution. Concentrated efforts on “Binary”, produced no significant results. If digit code on cipher underlying binary, material is insufficient.

Is It A Real Cipher?

On the one hand, the parents “did state, however, that there were a number of papers of a similar nature in their home in Brooklyn. They stated also […] that Benjamin Birnbaum […] would undoubtedly be able to furnish information concerning the ‘coded note.’ ” Later: “Mrs Rubin stated that her son and his friend, Benjamin Birnbaum, often exchanged “notes” similar to the one found taped to the subject’s abdomen when he was found.” (p.17). Yet “BIRNBAUM advised that he never sent to nor received a coded message from deceased” (p.29).

Rubin was also fascinated by binary. Birnbaum “stated that the deceased talked of using a word unit code with numbers for each word. The numbers would then be transmitted into the binary code. BIRNBAUM advised that the binary code is a code used on all calculating machines. He stated that the deceased was going to use another stage of transmitting this code unknown to BIRNBAUM.” (p.35)

I think it is abundantly clear from the file that Paul Rubin and Bernard Birnbaum did communicate by means of cipher, even if Birnbaum strenuously denied this under interview. Rubin and Birnbaum thought that everyone else was not only stupid, but deserved to be treated as stupid (and said so in the interview): so if Birnbaum treated the FBI as stupid, we should perhaps not be surprised.

In my opinion, the cipher would seem to be completely genuine and that Paul Rubin’s parents and Bernard Birnbaum did initially have access to Paul Rubin’s codebook. However, I strongly suspect that they chose to destroy it rather than give it to the government, lest the ciphertext reveal something unsavoury about the dead student’s end – better for him to die in mystery than in possible ignominy.

Here’s something a little unusual for you all – a feature-length 1971 episode of a Sci-Fi series called “Name of the Game” on YouTube, and set in “A.D. 2017” (also the name of the episode):

Interestingly, the director was a young Stephen Spielberg (it was one of his earliest pieces of work); while its author was American writer Philip Wylie, whose career moved from screenplays to slushy novels to non-fiction to dystopian fiction (and then dotted around between them for several decades).

But more about Wylie (and his connection to a cipher mystery) in a separate post. For now, on with the show! Enjoy! 🙂

Here’s a partial update on the Paul Emanual Rubin cipher mystery for you, followed by my updated thoughts on the ciphertext, and then finishing with my current best guess as to what happened. But… please don’t build your hopes up too high, there’s no sign of hugely happy ending to it all just yet. 🙁

The Update Part

After my recent post on the Paul Rubin cipher, I discovered that Craig Bauer (whose long-standing interest in the case was the starting point for that post) is currently planning to write and publish a Cryptologia article on it.

Needless to say, I’m really looking forward to his article: and I really hope that he gets to peek just a little behind the veil of the scanty external documentation to reveal the real secret history just behind it. There’s certainly very little of substance currently in the public domain: yet there’s a good chance that many of Rubin’s friends – if Bauer could manage to identify them – may well still be alive. I’m sure that they would be very happy to help Bauer get closer to the truth, even at this distance in time.

At the same time, I’m assuming that the FBI – which was so clearly involved right from the start of this case – will likely not be contributing to the picture Bauer will build up. As a matter of general policy, the Bureau seems uninterested in contributing or collaborating with external codebreakers (and don’t get me started on the Ricky McCormick case), so the FBI side of things seems to be an avenue that will remain blocked for many years yet.

The Cipher Part

Doubtless Craig Bauer will have his own conclusions about the ciphers used (and he may even have access to enough primary material to be able to break the ciphertext). But here are mine.


Even on the low resolution photo of the cipher note, we can see “PER” at the bottom right, which is surely Paul Emanuel Rubin’s typewritten signature. Hence there is no real doubt that this whole thing was made directly by him.

I also have no real doubt about the section with the numbers on (outlined in blue above): that this is a home-brewed Morse Code variant. Only an amateur / kid code-maker would use something like this: while visually impressive, it’s hugely long-winded and impractical.

Finally, I have very little doubt about the section that seems to have “Conant” and “Dulles” in the clear: there is only one cipher system that has sufficient latitude to achieve this, and that is where a plaintext message (or a simple substitution cipher message) gets interspersed with euphonic or orthographically pleasing nulls to generate a covertext. Trithemius proposed this 500 years ago, and I feel certain that this is what is being used here.

e.g. the underlying cipher message might say “C.N.N”, but after nulls get added, the cover message becomes “C.o.N.a.N.t”.

As a more general observation, though, I suspect that each line of the text uses a separate code-table, because there seems to be very little consistency from line to line, even if you take every other letter of each line. But that fits with the whole amateur code-maker thing: it’s almost impossible to break extremely short ciphertexts, unless you have a lot of a priori knowledge about what was going on in context (which seems not to be the case here).

Finally, the other short patches of text seem likely to me to be code-table references or offsets etc. When Rubin’s friend said at the time that he (I presume he) would be able to decrypt it in a few hours if he had access to Rubin’s code-tables, this is without much doubt what he meant. Hence my prediction is that each code-table is relatively simple in itself, but that each one is used only in short bursts, to make it nearly impervious to codebreakers.

The Secret History Part

As to what was going on Rubin’s life and head in the days and weeks up until his death, I don’t honestly know: as I mentioned before, I strongly suspect that he may have been suffering from paranoid schizophrenia (for surely only someone in a distressingly paranoid place would even consider taping a coded message to their abdomen), but many other causes and rationales are still in play.

But I do have a guess about what went wrong after his death.

And this is based on cryptanalysis: that if “Conant” and “Dulles” were, as I believe, both part of a covertext created by filling between genuine cipher letters with nice-looking Trithemian nulls to form additional words, then I don’t honestly believe that the plaintext will ultimately have anything remotely to do with “Conant” and “Dulles” at all. In short, if you were paranoid about Conant and Dulles, you really wouldn’t leave them in plain sight.

But this was the diametric opposite of the assumption that the FBI seemed to be working on in this case. This was 1953, after all, the height of the HUAC and Cold War era, and Reds were perceived to be under every goshdarned Bed. And Rubin was a student, and therefore – so the mythology goes – could well have been exposed to one of numerous radicalizing factors and ideologies. So in fact it seems to me that poor Mr Rubin may not even have been the most paranoid element in the whole setup: by which I mean that the FBI seems to have been institutionally paranoid at that time, paranoia on an almost industrial scale.

Personally, I believe that Rubin’s friends – with whom he had exchanged numerous coded notes, according to various newspaper accounts, and doubtless knew most of his Trithemian tricks well – could easily have broken the heterogeneous set of micro-ciphers that made up his final essage, had they been given access to his code-tables.

However, my suspicion (as things stand) is that Rubin’s friends never got to see the message. Instead, I suspect that the FBI collected up all Rubin’s code-tables and documentation, tried to break them (and failed), and then – because they perceived that National Security was somehow at stake, even if they couldn’t prove it – kept everything quiet. The paranoid logic being, of course, that if Rubin was a goddamn Commie, then all his code-breaking chums were probably all the same distressing shade of Red, and therefore not to be trusted anywhere near his note.

(To my eyes, there are institutional echoes of the way the Ricky McCormick case was later handled, with the (minor) difference being that the institutional paranoia at play there was a weed (ha!) that grew vigorously on top of America’s civil War On Drugs.)

If I’m right, all Rubin’s code-tables are probably still sitting in an FBI archive, having never been shown to the very people who could have helped crack it without much trouble: and if so, what a sad waste of time the whole affair was.

Let’s hope Craig Bauer gets to tell us the whole story.

I’ve been trying to post about the cipher that was attached to Paul Emanuel Rubin’s stomach for over a year now: so it’s about time I pushed what I’ve found out onto the web, to see if anyone else has more luck.

The story starts with a pair of 35-minute videos of a 2013 talk on unsolved ciphers given by Craig Bauer, managing editor of Cryptologia. Part 1 covers the usual suspects (Voynich, Beale, Dorabella, Somerton Man, etc), while Part 2 moves on to largely American cipher mysteries (Zodiac, Kryptos, etc). The slides are also available here.

One American cipher that has long fascinated Bauer (right at the start of Part 2) is the case of Paul Emanuel Rubin, who on the morning of 20th January 1953 was found dead (from cyanide poisoning) with a cipher taped to his stomach. Bauer showed a picture of the cipher taken from the Philadelphia Evening Bulletin:-


Newspaper Accounts

The best newspaper account I found was from the Schenectady Gazette, Friday 23rd January 1953:-

Dead Youth Sent Friends Coded Notes

The parents of a chemistry student who was found dead of cyanide poisoning with a coded letter taped to his stomach told police today that the youth had the curious habit of sending coded letters to friends.

One unidentified friend in Paul Rubin’s Brooklyn neighborhood told police he might be able to decipher the typewritten note that has baffled the FBI. Two words in it were recognizable – “Dulles” and “Conant”.

But these developments only heightened the puzzle for Coroner Joseph Ominsky. “I am keeping in my mind the opinion that this is no suicide,” he said. Yesterday Ominsky said there was a possibility it was murder.

The body of the 18-year-old New York University student was sent home to Brooklyn today for an Orthodox Jewish burial while local and federal authorities tried to learn how he came to die Tuesday in a ditch at International airport.

Ominsky said the FBI should explore contacts the victim had in New York and here to determine if he had been a messenger for some unusual activity.

“Until this question is answered,” the coroner said, “I cannot set a time for an inquest.”

The FBI here, informed of Ominsky’s statement, said it had “no comment”.

A number of curious objects were found in Rubin’s pockets and the appearance of the words “Dulles” and “Conant” on the coded message lent a cloak-and-dagger air of mystery to his death by cyanide poisoning. It was assumed the names referred to Secretary of State John Foster Dulles and James Bryant Conant, U.S. high commimissioner in West Germany.

The FBI here has sent the message to its cryptographers in Washington.

Neither the parents, who said Rubin left home last Monday morning ostensibly for classes, nor the friend, could see any motive for suicide.

Moreover, according to this report

Ominisky [sic] said that if the cyanide was self-administered, the body would not have been in its orderly position, with the youth’s thick lense [sic] glasses undisturbed. He pointed out that cyanide kills quickly, and no vial or other container was found at the scene.

However, the Shamokin News Dispatch reported the next day (23rd January 1953) that:-

Investigators disclosed today that a five-inch long test tube was found near the ditch where the body of a New York University chemistry student was discovered last Tuesday.

The vial was turned over to city toxicologists for tests to determine if it had contained the cyanide which killed 18-year-old Paul E. Rubin, of Brooklyn, N. Y., a sophomore at NYU.

Police said the test tube was found about five feet from the 12-foot deep embankment near the International Airport in which the body was spotted by a soldier en route to catch a plane.

A thorough search was made of the area, but nothing else was found, police reported.


Police reported that an unidentified friend of the victim said that he could probably decipher the message found on the body if he could find the proper code books, but it might take him a week.

“Curious Objects”

Bauer (who either has access to far more newspapers than me or has seen the Coroner’s Report) noted that the “curious objects” included:

* “an image of an airplane with a Nazi swastika on its tail assembly” in a wallet: on the photo’s rear was written “France Field, Panama”.
* Another picture in the same wallet was of “The Thinker”, Rodin’s well-known sculpture.
* “a plastic cylinder containing a signal fuse” (a magic prop)
* “the casing of a spent .38 caliber bullet (found in a pocket of his topcoat)”
* “a fountain pen gun”
* “forty-seven cents” (though he had started the day with a rather more substantial $15)
* the February 1953 issue of “Galaxy Science Fiction”

Attempted Followups

To me, this is an unsatisfactory cipher mystery because we can’t even see the cipher clearly. I asked the FBI if they would release a better quality scan of the cipher, but got no reply (which is a shame).

I also asked the Special Collections Research Center at Temple University (who have a large archive of the Philadelphia Evening Bulletin’s photos, many of which won awards) about the photo. Unfortunately, the scan of it they very kindly sent me from their microfilm archives was slightly worse than the one we already had, so no luck there either.

Also, I should point out that William Friedman retired in 1956, and without any doubt would have been shown the cipher at the time: so perhaps there’s a mention of it in Friedman’s papers at the George C. Marshall Foundation. I’ve had no luck there (though the archive does have lots of great photographs, if that floats your boat), so feel free to pick up my Friedmanian dropped baton and carry it further yourself.

Any other suggestions?

So… What Do I Think?

Firstly, I have to say that I think anyone who types out a mysterious note and tapes it to their stomach to be found later must (a) be suffering from some complex mental illness and (b) be about to consciously embark on a pre-planned perilous action. I’m sure up-to-the-minute politically correct healthcare professionals have a whole wealth of ways of pussyfooting around the term, but to me this has plenty of features that suggest paranoid schizophrenia.

Secondly, looking at the cipher itself, it’s clearly a mess: or rather, it doesn’t appear to be the product of an ordered, controlled mind or even a single cryptographic system. There are fragments of binary 0s and 1s (interpersed with dots and x’es), German, made-up words, an odd-looking signature, all kinds of stuff.

As a result, my overall opinion is that what we are looking at here is probably a product of delusion and steganography in equal measures: and so it is probably a kind of unhappy cryptogram rather than a tricky ciphertext per se. Hence my conclusion: that even though I would very much like to see the note in its entirety, I don’t currently believe that we will ever solve it.

Some things are just lost to the world.