The Somerton Man code has never been read, and the reason is not that the right cipher has yet to be tried. It is that the object carrying it is gone, the letters on it cannot be transcribed with certainty, and the only official record of the case never mentions it at all.
That last point is the one that gets missed. Evidence closed on 21 June 1949, when the coroner adjourned. The book with the pencilled letters was not handed to police until 22 July 1949 (State Records of South Australia). That is the most famous piece of evidence in the case, and it arrived a month after the evidence closed.
Key Takeaways
- State Records of South Australia holds the inquest file as 71/1949, in GRG1/27, and states that the code and the phone number "are not mentioned in the Coroner's files relating to the case."
- The inquest opened 17 June 1949 and adjourned 21 June 1949. Police did not receive the book until 22 July 1949.
- The working transcription runs to 44 letters using 16 different letters, with 8 of the 44 being the letter A, one Q and no U. The same research site elsewhere describes 50 letters in five lines.
- The University of Adelaide's cipher cross-off list carries 37 entries and strikes through 32 of them. One of the five survivors is a statistical technique; of the four remaining hypotheses, two are not ciphers at all.
- The book itself was lost, and the pencil was so faint that police needed an ultra-violet lamp to read it.
What is actually written on the Somerton Man code page?
Nobody can say for certain, and that is the first obstacle rather than a quibble.
Every published attempt works from the same four-line transcription: MRGOABABD, MTBIMPANETP, MLIABOAIAQC, ITTMTSAMSTGAB. That is 44 letters drawn from 16 of the 26 English letters. Eight of the 44 are the letter A. There is one Q and no U anywhere in it.
Those are not decorative statistics. University of Adelaide project teams used each of them to kill a hypothesis. Sixteen distinct letters rules out any scheme whose output alphabet is smaller, which is how the ADFGVX field cipher went. The lone Q rules out anything keyed to a rotary telephone dial, because the dial face of 1948 had no Q on it.
Eight A's does the same work against the flat-frequency hypothesis. Set up a hypothesis that every letter is equally likely, work the binomial probability of eight A's turning up in 44 letters, and the number is small enough to abandon the idea. Project teams ran exactly that calculation.
Then the transcription itself wobbles. The university's own page on details of the case that are commonly misreported says of that four-line string: "Some of the letters are in fact ambiguous and cannot be transcribed with certainty. Also there are spaces that are missing." It adds that the original carries an extra crossed-out line, an "x" marked above the final O, and a hooked S.
Its own case timeline describes what police actually received as "a sequence of 50 letters presumed to be a secret code. The sequence is in 5 lines and contains 3 spaces. One line is possibly crossed out."
| Description |
Letters |
Lines |
Spaces |
| Transcription used in the cipher analyses |
44 |
4 |
0 |
| Timeline account of what police received |
50 |
5 |
3 |
One research group, two counts, and no way to settle it, because the physical page is not available to check.
Was the Somerton Man code ever evidence at the inquest?
No, and the archive says so directly.
State Records of South Australia describes the copy handed in as one that "contained a handwritten note believed to be a secret code and a telephone number, later traced to a woman nicknamed 'Jestyn'. The police interviewed both the man and Jestyn, but they are not mentioned in the Coroner's files relating to the case."
The exhibit list bears that out. Exhibit C.9 is the torn slip itself, described by the coroner as "torn from a copy of the second edition of FitzGerald's translation of the Rubaiyat of Omar Khayyam". The book that slip came out of appears nowhere in the exhibits, and the word "code" appears nowhere in the transcript.
Chronology explains how that happened. Cleland's inquisition names its own sitting days: the 17th and 21st June 1949, and then, nearly nine years later, the 14th day of March 1958. Evidence closed on 21 June 1949, and the book did not reach police until 22 July 1949, a month later. A copy of the letters went to Army Headquarters in Melbourne on 25 July 1949 and to a Navy codebreaker on 27 July 1949, both after the 1949 hearing had closed.
Two further things follow from the archive's own description of what it holds. The 1949 file is open to the public only because coronial files are restricted for sixty years and this one has aged past that limit. A later inquest file is in the archive's custody but still restricted, with access requests directed to the Coroner's Court. And police files are not there at all: "The Somerton Man case is still an open case, despite its age, and therefore we do not have police files relating to the case in our custody."
What did the coroner actually find?
Nothing, formally, and he said so in the plainest possible words.
The cover page of the inquisition, signed by Coroner Thomas Erskine Cleland on 14 March 1958, reads: "I am unable to say who the deceased was. He died on the shore at Somerton on the 1st December, 1948. I am unable to say how he died or what was the cause of death."
His summary of the evidence is more revealing than the finding, because it shows exactly where he stopped. "There was no indication of violence, and I am compelled to the finding that death resulted from poison. But what poison?" He was willing to go as far as this: "I would be prepared to find that he died from poison, that the poison was probably a glucoside and that it was not accidentally administered; but I cannot say whether it was administered by the deceased himself or by some other person." Then: "The evidence is too inconclusive to warrant a finding."
The analyst's testimony explains the gap. R. J. Cowan, the Deputy Government Analyst, tested stomach and contents, liver, urine and blood, and told the court: "I feel quite satisfied that if death were caused by any common poison, my examination would have revealed its nature. If he did die from poison, I think it would be a very rare poison." He also said he knew of no poison that could kill and then decompose beyond detection.
Worth separating two men with the same surname, because the case notes routinely merge them. Professor John Burton Cleland, a pathologist, found the slip of paper. Thomas Erskine Cleland was the coroner. They were distant cousins.
Nor did the court agree on what the slip meant. The coroner wrote that the words "support this conclusion, and indeed put its accuracy beyond reasonable doubt". Detective Brown, who traced the phrase to a Persian-English dictionary with a librarian's help, told the same court the opposite: "It does not seem to have any bearing on the case, or any meaning as to the cause of his death. As far as this death is concerned, there is no context into which the words can fit."
When the court reconvened on 14 March 1958, it heard no new witnesses. Its one product was the signed inquisition above, which recorded that the coroner was unable to say who the man was or how he died. That document, and not the book, is where the official account of the case stops.
Which ciphers have been ruled out for the Somerton Man code?
Almost all of them, one at a time, over several years of undergraduate project work at the University of Adelaide.
Those project teams keep a public cross-off list. It carries 37 entries and 32 of them are struck through as eliminated. Among the struck: Vigenere, Playfair and double Playfair, Beaufort, Porta, Gronsfeld, Trithemius, Bifid, Trifid, four-square, two-square, Hill, Baconian, Nihilist, ADFGVX, VIC, pigpen, book cipher, rail fence, plain transposition, homophonic substitution, and plain shift and Atbash.
Method matters more here than the tally. Where a cipher could be brute-forced, it was. An affine cipher has only 312 possible keys over a 26-letter alphabet, so all 312 were run and none produced readable text.
Where a statistical fingerprint was available, that was used instead. Index of coincidence figures were calculated for every block length up to and including 22, which is the largest meaningful value for a 44-letter text. A separate run used the Rubaiyat itself as a key stream and produced "some ten thousand decrypts, of which none in the shortlist yielded results".
Two honest caveats. The list strikes through the null cipher, but its own reasoning is softer than that: "While the Null Cipher cannot be completely ruled out, it is unlikely." And the keyword cipher and the one-time pad sit on the list unstruck, meaning untested to elimination rather than tested and survived.
What if the letters are not a cipher at all?
That possibility is not a fringe reading. It is where the university's own analysis points.
Strip out the index-of-coincidence entry, which is a technique rather than a hypothesis, and the cross-off list has four surviving candidates. Two are ciphers. The other two are "Initial letters of a sentence" and "Initial letters of an unordered list".
Language statistics support them. A 2013 project report ran letter-frequency comparisons across 266 languages, using the opening words of the Rubaiyat as a control text of the same length, and concluded that "the most likely language is English if the code were the initial letters of words in a sentence or a list".
Derek Abbott, who led the identification work, has offered his own version of the same idea. He told the ABC that the letters probably represent the first names of horses that Carl Webb, the man identified in 2022, bet on.
That is a hypothesis, not a solution, and it is important not to dress it as one. Here is the null result stated plainly. No source consulted for this article sets out a tested method that turns those 44 letters back into words, and none claims one exists. An initialism is not a cipher with a key you can search for. Forty-four initials could stand for an enormous number of English sentences, all of them equally consistent with the evidence, and there is no second message, no crib and no plaintext fragment to choose between them. If the letters really are initials, the honest position is that they are unrecoverable rather than unsolved.
One thing the letters are probably not is idle scribble. Project teams in 2009 and 2010 surveyed people, sober and intoxicated, asked them to produce strings of letters, and compared the resulting frequencies with the page. Neither group matched.
What would it take to read the Somerton Man code now?
Something that no longer exists.
The book was lost by police, so no fresh photography, no raking light and no modern imaging can be applied to the original page. Those marks were made in pencil so faint that officers needed an ultra-violet lamp to see them, and every transcription in circulation descends from a police photograph of that faint text rather than from the paper.
Genetic work of recent years answered a different question entirely. Writing in August 2022, Smithsonian Magazine noted that "the state coroner has yet to confirm the pair's findings", and quoted Abbott saying "It's up to the cops to make the legal determination of who this guy was." South Australia's archive, updating its own case page the following month, still described the matter as an open case.
A name is not a key. Knowing who wrote something tells you nothing about a 44-letter string unless you also have his handwriting habits, his reading, or a second document in the same scheme. None of those survive.
Frequently asked questions
Has the Somerton Man code ever been decoded?
No published decryption has ever been demonstrated and checked. University of Adelaide project teams keep a cross-off list recording 32 of its 37 candidate schemes as eliminated, including brute-force runs over every key of the affine cipher and roughly ten thousand decrypts using the Rubaiyat as a key stream, none of which produced readable text. Of the five entries left standing, one is a statistical technique rather than a hypothesis; the other four are two ciphers that have not been tested to elimination and two accounts in which the letters are not a cipher at all.
How many letters are in the Somerton Man code?
The transcription used in the published analyses has 44 letters across four lines, using 16 of the 26 English letters, with eight instances of the letter A. The same research site's case timeline instead describes what police received as a sequence of 50 letters in five lines containing three spaces, with one line possibly crossed out. That difference is not resolvable, because the university's page on commonly misreported details states that some letters "cannot be transcribed with certainty" and that spaces are missing from the usual rendering.
Why do some analysts think the letters are not a cipher?
Because the statistics fit an initialism better than they fit an encryption. A 2013 project report compared the page against 266 languages and found that English is the most likely language if the letters are the initial letters of words in a sentence or a list. On the project's cross-off list, "Initial letters of a sentence" and "Initial letters of an unordered list" are among the few entries never struck through. Derek Abbott has suggested the initials are the first names of horses that Carl Webb, the man identified in 2022, bet on. None of these accounts recovers the underlying words, and none claims to.
Where is the book the code was written in?
It is gone. University of Adelaide case notes state that police appear to have lost the copy containing the code. That matters more than the loss of an ordinary exhibit, because the pencil marks were faint enough to require an ultra-violet lamp when police first examined them, so the surviving photographs are now the only witness to what the page said, ambiguous letters and all.
Does the 2022 identification help solve the code?
Not directly. Identifying the man is a question about human remains and family trees; reading the letters is a question about a lost sheet of pencil marks with no key, no crib and no second sample. Smithsonian Magazine reported in August 2022 that the state coroner had yet to confirm the identification, and State Records of South Australia described the case as still open when it updated its page in September that year. The coronial record, which is the only official document in the case, still says the coroner was unable to say who the deceased was or how he died.
This story ends with a name and no reading. If you want to feel why 44 letters is so little to work with, put a short string through the cipher identifier and see how many schemes it cannot tell apart, or take a longer run at the rest of the codes.
Sources
- State Records of South Australia, "But what poison?" Mystery of the Somerton Man - the archive's own account of inquest file 71/1949 in GRG1/27, the sixty-year restriction on coronial files, the open status of the case, Cowan's poison testimony, and the statement that the code and phone number do not appear in the Coroner's files.
- University of Adelaide, School of Electrical and Electronic Engineering, The Taman Shud Case Coronial Inquest - the transcript, the cover-page finding signed 14 March 1958, the "But what poison?" summary, and the exhibit list including Exhibit C.9.
- University of Adelaide, Cipher Cross-off List - the 37 entries, the 32 struck through, the 16-distinct-letter and eight-A arguments, the affine brute force, the index-of-coincidence block-length testing, and the null-cipher caveat.
- University of Adelaide, Semester B Final Report 2013, cipher cracking - the 266-language comparison, the initial-letters conclusion, and the ten thousand Rubaiyat decrypts.
- University of Adelaide, List of facts on the Taman Shud Case that are often misreported and Timeline of the Taman Shud Case - the ambiguous letters and missing spaces, the crossed-out line and hooked S, the 50-letter description, and the July 1949 dates for the Army and Navy referrals.
- University of Adelaide, List of facts we do know about the Somerton Man - the lost book and the ultra-violet lamp.
- Smithsonian Magazine, Meilan Solly, Have Scholars Finally Identified the Mysterious Somerton Man?, 8 August 2022 - the coroner's unconfirmed status and Abbott's remarks on the legal determination and on horse names.
