Mysteries Department
Sign inRequest access
MYS0CRZAKFP1TJ5·Mystery · document

The Dorabella cipher

In July 1897 Edward and Alice Elgar stayed for a few days with the family of the Reverend Alfred Penny at Wolverhampton. On 14 July, back in Great Malvern, Alice wrote to thank Mrs Penny; inside the letter was a slip in Edward's hand addressed on the back to "Miss Penny", covered with 87 small marks made of one, two or three semicircles⁠[1]⁠[2]. Dora Penny, then 23, kept it for forty years. Elgar later gave her the tenth of his Enigma Variations (1899), "Dorabella".

Dora printed the note as an appendix to her 1937 memoir: "I have never had the slightest idea what message it conveys; he never explained it and all attempts to solve it have failed. Should any reader of this book succeed in arriving at a solution it would interest me very much to hear of it"⁠[1]. The original has since been lost. Eighty-nine years after that appeal there is still no accepted reading⁠[3]⁠[4]. Of the 3 questions on this site, 0 are answered. The most likely answer to “What kind of message is the Dorabella note?”, not yet confirmed: english in Elgar's alphabet with irregularities (phonetic spelling, abbreviations, private words, mirrored….

Key questionsset by the editors

What kind of message is the Dorabella note?Most likely: english in Elgar's alphabet with irregularities (phonetic spelling, abbreviations, private words, mirrored…
All 3 questions

The object

The note has three lines of 29, 31 and 27 symbols. Each symbol is one, two or three nested semicircles whose open side faces one of eight directions at 45-degree steps, giving 24 possible symbols, of which 20 appear; a small dot follows the fifth symbol of the last line⁠[4]⁠[2]. The orientation of some symbols is ambiguous in the facsimile, and published transcriptions differ at a few places; Hauer and colleagues built a majority transcription from five, which this project uses⁠[4]⁠[5].

The Dorabella note (Elgar to Miss Penny)
The Dorabella note (Elgar to Miss Penny)unknown (from the 1937 facsimile, via ciphermysteries.com), 1897. Public domainArtifact

Elgar used the same symbols elsewhere. In April 1886 he pencilled 18 of them into a Crystal Palace concert programme beside a musical example from Liszt's Les Préludes; they appear again in an exercise book of about 1920, with diagrams like clock faces; and on a card headed "Cryptogram" that belongs with his working of a cipher challenge printed in the Pall Mall Magazine in 1896⁠[6]⁠[7]⁠[8]. The programme and the exercise book were reported in 2007 to be at the Elgar Birthplace Museum⁠[7].

Map of the regionOpen the full map
LondonBirminghamBristolLeicesterCoventryReadingLuton123450 kmNBristolLeicesterCoventryReading123450 kmN
1
Wolverhampton (the Penny family's rectory)Town
Within 1 km
2
Elgar Birthplace (The Firs), Lower Broadheath, WorcestershireMuseum
Within 100 m
3
Great Malvern, WorcestershireTown
Within 10 km
4
Crystal Palace, LondonConcert venue (demolished 1936)
Within 1 km
MappedWithin 1 kmA regionNatural Earth

Timeline

Timeline
Squeezed where little happened. Each mark is a published claim, by how it stands:Accepted: Passed its tests and two independent reviewsAcceptedLikely: Best supported so far, not yet confirmedMost likelyStill possible: Not tested enough to call either wayStill possibleDisputed: Reviewers disagree about itDisputedRuled out: The evidence contradicts itRuled out
What happenedPublished claims188618961899193719702007201820202023Elgar pencils 18 semicircle symbols into a programme at a Liszt concert, 1886Liszt fragmentElgar works on a cipher challenge printed in the Pall Mall Magazine, 1896Pall Mall challengeAlice Elgar's thank-you letter encloses Edward's cipher note to Miss Penny, 1897Note sent to DoraEnigma Variations performed; Variation X is "Dorabella", 1899Enigma VariationsDora Powell's memoir reproduces the note and invites solutions, 1937Note publishedElgar Society Dorabella competition for the 150th anniversary; no entry accepted, 2007Elgar Society competitionSchmeh presents a manual-methods analysis at HistoCrypt 2018, 2018Manual cryptanalysisHauer and colleagues publish solver, language-identification and music tests, 2021Solver and language testsWase shows solvers fail on Dorabella but succeed on English and Latin controls, 2023Substitution ruled outDora Powell (née Penny), 1937: The note is a message in Elgar's cipher that she never read and he never explained.Penny) 1937Eric Sams, 1970: The note is a substitution of English in which phonetic shorthand expands 87 symbols into 109 letters, beginning "Starts: Larks!".Eric Sams 1970Klaus Schmeh, 2018: Manual methods for simple substitution pick out likely vowels in Dorabella, as they do in an English control text of the same length.Klaus Schmeh 2018Wayne Packwood, 2020: Rearranged by its dots and with each symbol's value shifted, the note reads "A woman is like chess" with a hidden "rats".Wayne Packwood 2020Bradley Hauer and colleagues, 2021: The note is closer to language than to music, most likely English, and its 13 mirrored symbol pairs are structure, not chance, though it is no simple substitution.Hauer et al. 2021Viktor Wase, 2023: Dorabella is not a monoalphabetic substitution of English or Latin, because solvers that break such ciphers of the same length fail on it.Viktor Wase 2023

What is known

Elgar was a lifelong maker of puzzles, puns and ciphers; Dora's memoir opens its appendix by saying so, and his 1896 work on the Pall Mall challenge shows he could break a cipher as well as make one⁠[1]⁠[6]. The exercise-book pages are the strongest clue to the system: a commenter on Nick Pelling's site reports that they pair the semicircle alphabet with test phrases such as "A very old cipher", "Do you go to London" and "Marco Elgar", which suggests a key that assigns letters to the symbols in a fixed rotational order⁠[9]. If so, the system is a simple alphabet of Elgar's devising, and the difficulty lies in the message, not the cipher.

The frequencies of the 20 symbols resemble a letter distribution in a short English text⁠[10]. The Liszt fragment has been read as "GETS YOU TO JOY, AND HYSTERIOUS" (Anthony Thorley, 1977), a reading Jerrold Northrop Moore printed in his biography of Elgar but which others find unconvincing⁠[7]⁠[8].

What has been tested

Klaus Schmeh applied Helen Fouché Gaines's manual methods for monoalphabetic substitution (vowel detection, digram analysis, consonant lining) to the cipher and to a control text of the same length. Vowel detection worked well on the control and gave candidate vowels for Dorabella, but no reading⁠[10].

Hauer, Choi, Sundar, Hindle, Smallwood and Kondrak ran state-of-the-art automatic solvers that reliably break 87-letter English substitutions; none produced readable text from Dorabella. A ciphertext language identifier ranked English as the most likely underlying language; character-model classifiers placed the cipher closer to language than to music; and Dorabella has 13 pairs of mirrored symbols where random English enciphered with these symbols has 3.6 on average, which happens by chance about one time in a thousand. They read the mirrored pairs as structure, perhaps a way of writing double letters, and as evidence against a meaningless hoax⁠[4]. Viktor Wase showed that decipherment algorithms solve English and Latin substitutions as short as Dorabella and fail on Dorabella itself, and concluded it is not a monoalphabetic substitution in either language⁠[11].

In 2007, for the 150th anniversary of Elgar's birth, the Elgar Society ran a Dorabella competition. Several entries came in and none was accepted: entries that matched symbols to letters "invariably ended up with a fairly arbitrary sequence of letters" read as "a disconnected chain of bizarre utterances"⁠[2].

The readings

Eric Sams proposed in 1970 a substitution reading in which phonetic shorthand turns 87 symbols into 109 letters: "Starts: Larks! It's chaotic, but a cloak obscures my new letters..."⁠[12]. Later claims include simple substitutions with nulls (Richard Henderson, 2011; Tim Roberts, about 2012) and a reading by Wayne Packwood (2020) that rearranges the text by its dots and shifts each symbol's value without a stated rule⁠[2]⁠[13]. Each needs free choices that allow other readings equally well, and none has been accepted⁠[3]⁠[9].

Three broader readings remain. The note is English in Elgar's alphabet but not a plain substitution: it may use phonetic spelling, abbreviations, private jokes, double letters marked by mirroring, or a reading order other than line by line⁠[4]⁠[12]. It encodes music: the eight orientations as notes of a scale and the arc count as a modifier, a reading proposed by Javier Atance and tested, without support, by Hauer and colleagues⁠[2]⁠[4]⁠[5]. Or it means nothing that could be decoded, a joke made to look like a cipher; the non-random mirroring weighs against this⁠[4].

Published claims

Published claims and their status
YearWhoClaimedStatusReason
1937Dora Powell (née Penny)⁠[1]The note is a message in Elgar's cipher that she never read and he never explained.Accepted: Passed its tests and two independent reviewsAcceptedIts statistics look like language rather than random marks, and no reading has been accepted since⁠[4]⁠[3].
1970Eric Sams⁠[12]The note is a substitution of English in which phonetic shorthand expands 87 symbols into 109 letters, beginning "Starts: Larks!".Disputed: Reviewers disagree about itDisputedThe reading needs free choices that allow other texts equally well, and cryptologists judge that it makes little sense⁠[3]⁠[9].
2018Klaus Schmeh⁠[10]Manual methods for simple substitution pick out likely vowels in Dorabella, as they do in an English control text of the same length.Disputed: Reviewers disagree about itDisputedAutomatic solvers that break English and Latin substitutions of the same length produce no readable text from it⁠[4]⁠[11].
2020Wayne Packwood⁠[13]Rearranged by its dots and with each symbol's value shifted, the note reads "A woman is like chess" with a hidden "rats".Disputed: Reviewers disagree about itDisputedThe shifts follow no stated rule, so the method can produce other texts as easily, and no specialist has accepted it⁠[2].
2021Bradley Hauer and colleagues⁠[4]The note is closer to language than to music, most likely English, and its 13 mirrored symbol pairs are structure, not chance, though it is no simple substitution.Likely: Best supported so far, not yet confirmedMost likelyWase's tests agree that it is not a plain substitution, but no irregular reading has yet produced a text⁠[11].
2023Viktor Wase⁠[11]Dorabella is not a monoalphabetic substitution of English or Latin, because solvers that break such ciphers of the same length fail on it.Accepted: Passed its tests and two independent reviewsAcceptedIt agrees with the independent solver tests of Hauer and colleagues⁠[4].

What would settle it

A key in Elgar's hand that maps his semicircle alphabet to letters, tested on the Liszt fragment and the exercise-book phrases, and then producing a sensible English text from Dorabella without further free choices, would settle it. High-resolution images of the exercise book, the Cryptogram card and the 1886 programme, published with transcriptions, are the most useful single step; the 1937 facsimile is the only image of Dorabella itself, so any lost original resurfacing would also resolve the ambiguous orientations.

Questions

0 of 3 questions are answered.

Key questionsAll 3
0
Open: 3 answers, none ahead
0
0 of 1 sub-questions answered·0
Most likely: english in Elgar's alphabet with irregularities (phonetic spelling, abbreviations, private words, mirrored…

Evidence 16 findings

Artifacts

The films, photos, papers and objects this mystery rests on, each with its own key and its distance from the original.

See also

References

  1. 1.Powell, M. 1937. Edward Elgar: Memories of a Variation. Oxford University Press, London.doicited 8 times hereSRC0CRZAG2A7KEB
  2. 2.Wikipedia contributors n.d.. Dorabella Cipher. Wikipedia.cited 10 times hereSRC0CRZAK7W5CAT
  3. 3.Schmeh, K. 2017. The Top 50 unsolved encrypted messages: 26. The Dorabella Cryptogram. Klausis Krypto Kolumne (scienceblogs.de), 3 August 2017.cited 3 times hereSRC0CRZAJ0T38SC
  4. 4.Hauer, B., Choi, C., Sundar, A. et al. 2021. Experimental Analysis of the Dorabella Cipher with Statistical Language Models. Proceedings of the 4th International Conference on Historical Cryptology (HistoCrypt 2021), Linköping Electronic Conference Proceedings 183: 70-79.doicited 10 times hereSRC0CRZAGSR16WS
  5. 5.Hauer, B., Choi, C., Hindle, A. et al. 2021. Dorabella Cipher as Musical Inspiration. Proceedings of the Workshop on Speech and Music Processing (SMP-2021), pages 33-38.cited 2 times hereSRC0CRZAH1J0G38
  6. 6.Bauer, C. 2017. Unsolved! The History and Mystery of the World's Greatest Ciphers from Ancient Egypt to Online Secret Societies. Princeton University Press (ISBN 978-0691167671).doicited 6 times hereSRC0CRZAHS00E5Y
  7. 7.Hartmeier, D. 2017. Clues (or red herrings) to the Dorabella Cipher. benzedrine.ch.cited 6 times hereSRC0CRZAJR87G5P
  8. 8.Schmeh, K. 2018. Two more unsolved cryptograms created by Edward Elgar. Klausis Krypto Kolumne (scienceblogs.de), 13 July 2018.cited 5 times hereSRC0CRZAJ8M69J2
  9. 9.Pelling, N. n.d.. Dorabella Cipher. Cipher Mysteries.cited 5 times hereSRC0CRZAJGE1ZDJ
  10. 10.Schmeh, K. 2018. Examining The Dorabella Cipher with Three Lesser-Known Cryptanalysis Methods. Proceedings of the 1st International Conference on Historical Cryptology (HistoCrypt 2018), Uppsala, Linköping Electronic Conference Proceedings 149: 145-152.cited 5 times hereSRC0CRZAGHY464C
  11. 11.Wase, V. 2023. Dorabella unMASCed - the Dorabella Cipher is not an English or Latin Mono-Alphabetical Substitution Cipher. Cryptologia (online 20 November 2023; 49: 15-24).doicited 4 times hereSRC0CRZAHH64CP5
  12. 12.Sams, E. 1970. Elgar's Cipher Letter to Dorabella. The Musical Times 111(1524): 151-154.doicited 6 times hereSRC0CRZAGA45BG6
  13. 13.Packwood, W. 2020. Elgar as cryptographer - Tuning and Turing. Musical Opinion 143: 34-37.doicited 1 time hereSRC0CRZAK022C12
Last edited 27 Oct 2026
Private beta Downloads are for beta members

We're in a private beta while we get things right. Everything here is free to read; downloading datasets, sources and code is open to members for now.

It takes a couple of minutes. We'll reply by email.