Edward Elgar’s Rose-Shaped Key

Abstract: A symmetric assembly of the symbols of Edward Elgar’s Dorabella Cipher yields a rose-shaped key, which can be populated and rotated at each cipher character for a systematic decipherment. The proposed key layout contains a dedication to Elgar’s wife, and the deciphered message refers to the geographic region where the cipher originated, along with a “fern walk” list of phonetically-spelled plants. An extensive survey of Elgar’s personal and professional letters provides examples of the same phonetic substitutions used in the deciphered message. The Rose Key is also used to propose a solution to Elgar’s Enigma, with an audio sample and supporting evidence presented.

1. Introduction

1.1 Origin

Composer Edward Elgar wrote an encrypted message to his friend Dora Penny on July 14, 1897 (Powell 1949, 129). It is now known as the Dorabella Cipher (Figure 1), referred to here as the Dorabella.

Figure 1. The Dorabella, from Powel (1949).

Elgar and his wife, Caroline Alice Lady Elgar, who went by Alice, visited Dora and her family in Wolverhampton earlier that July (Powell 1949, 7). Alice wrote a letter of thanks to Dora’s stepmother, and Elgar placed the Dorabella inside (Powell 1949, 129). It comprises curved symbols, each with one of eight orientations and one, two, or three arcs — for 24 possible symbols total. Given that Dora was English and not trained or experienced in codebreaking, it is generally assumed that the Dorabella’s message is in the English language and encrypted as a simple monoalphabetic substitution cipher (MASC), with each symbol representing one letter of the alphabet. The relatively short 87-character length and lack of spacing between words add decryption difficulty, but it is unusual that a simple cipher of this length would go unbroken.

1.2 Previous Work

Most proposed solutions to the Dorabella have traits that place them on the wrong side of some difficult-to-define threshold of believability. These traits can include assumptions of encryption mistakes, omitted or extraneous characters, word reversals, abbreviations, phonetic spelling, shorthand, and deviation from a systematic decipherment process. Solution proposals that show various combinations of some of these traits have been offered by Sams (1970), Gaffney (2008), Roberts (2012), ShadowWolf (2019), and Packwood (2020). The work described in this paper does not entirely avoid these traits, either, but bolsters their presence using historical context and supporting evidence.

Hauer et al. (2021a) apply MASC solving techniques including a reliable algorithmic solver without producing a readable solution, and conclude that the Dorabella may not be a simple MASC. When they assume it is a MASC, however, they find evidence the message is written in English. Detective work by Schmeh (2018) finds indications that the cipher is a MASC and in English, but without luck in finding a solution when applying largely-forgotten but effective MASC-solving techniques. Pelling (2013) has pointed out potential steganographic features but has not proposed a full solution. Mulliss (2022) used Pelling’s work as a starting point for a decryption based on the Dorabella’s symbol spacing, but the result appears to suffer from a similar collection of shortcomings as the solutions already mentioned. Hauer et al. (2021b) have explored potential encoded musical structures, though they have not claimed a definitive solution. Bauer (2017, 133) suggests that the lack of a convincing solution may point to an encipherment that is more complicated than a MASC.

At this point, it is fair to say that if the Dorabella were a simple MASC and concealed a clear English message, it would have been deciphered by now. However, its MASC-like characteristics and inexperienced recipient support a process that is not terribly complicated. Elgar wordplay is also not wholly unexpected; according to Bauer (2017, 154) automated solvers may be failing due to intentionally misspelled and invented words. The solution proposed in this paper bets on a modest increase in encipherment complexity and a somewhat obscure message, while providing supporting evidence for the elegance of the former and the context of the latter.

2. Methods

2.1 The Rose Key

A symmetric assembly of the symbols of the Dorabella, with proposed key letters, is given in Figure 2. As is common in historical cryptography, the letters I and J share a space, as do U and V.

Figure 2. The Rose Key of the Dorabella Cipher

To the author’s knowledge, the rotationally-symmetric, octant Rose Key has never been explicitly proposed. The potential for rotation has been noted before, however. Hartmeier (2006) insightfully discussed the symmetry of the symbols and their potential for rotation, pointing out that “… every symbol can be rotated and mirrored, always producing another valid symbol”. Pelling (2009) also pointed out the possibility for rotation based on symmetry “where you rotate each of the positions around after each letter”. Bauer (2017, 147) notes that a symbol on Elgar’s Cryptogram Card — one of the few other places his cryptic symbols appear — “starts to lean, then falls over and flips over, as if doing a forward roll.”

2.2 Decipherment

The decipherment begins by taking off the letter from the key that corresponds to the first symbol of the cipher, an I. The key is then rotated an octant — an eighth of a turn clockwise — just enough for every symbol to replace the one adjacent — and the second symbol is read, an M. The key is again rotated an eighth of a turn and the third symbol is read, an A. The decipherment proceeds in this fashion. When a dot is encountered in the Dorabella, additional eighth rotations are performed according to the digits of the date. This means that the first dot requires one extra turn, as indicated by the dot near the 1 in July 14, and the second dot requires four extra turns, as indicated by the dot near the 4 (Figure 3).

Figure 3. The dots of the Dorabella cipher indicate additional key rotations.

The dot on line two is relatively unknown and is necessary for the proposed decipherment. It is in contact with the 12th symbol on the second row, and appears only in the 3rd edition of Dora’s book, published by Methuen & Company (Powell 1949, 129). It does not appear in the 1st, 2nd, 4th, or 5th editions, all by other publishers. While it is common for book publishers to edit images, and easy to imagine the propagation of edited images into later editions, the publication of a more lightly-edited image in a middle edition is strange. Attempts by the author to inquire with Methuen & Company about this have shed little light.

The solution proposed in this paper was found using the hillclimbing function of CrypTool 2 (2022) and pen-and-paper follow through, but not beginning with a direct transcription of the Dorabella. Instead, an empty Rose Key was populated arbitrarily and a transcription was then produced by applying the key rotations. The following is the transformed ciphertext , with rotations included:

BMYSN RAFKU ZHDNY MROBF CBKAY TAWF
DYXHT WAYML UOYYT AAUFO BLFAK AKGQW F
OZLUP LGUSA IRBOY QPASC QZLCA QS

3. Results

Completing the decipherment yields the plaintext:

IMATH PENDO CKWHA MPSIN VIDEA FERN
WALKF REAMB OSAAF EEONS IBNED EDGYR N
SCBOQ BGOTE ZPISA YQETU YCBUE YT

The cleartext of which is:

I’m Ath. Pendock, Whamp sin vi de. A fern
walk: freambosa, afeeons ib., neded gyrn,
scb, oq, bgotez, pisay, qet, uyc bueyt

Expanding abbreviations, translating the Latin phrase and phonetic spellings, and notating common names:

I’m Athenaeum Pendock, Wolverhampton without force. A fern
walk: framboesa (raspberry), afions ib. (poppies, in the same place), netted gern (cantaloupe),
scab (fungus), oak, bogotas (amaryllis), picea (spruce), qat (khat), euc beaut (eucalyptus).

This can be interpreted as an intentionally-obscured list of plants from Pendock Old Church — a tiny but historic place with connections to geology and botany, and with an organ which Elgar is rumored to have played. Section 4 describes Elgar’s connections to Pendock and Wolverhampton, along with a survey of his letters demonstrating his use of abbreviations for cities, Latin phrases, and the same phonetic substitutions. Other promising traits of the proposed decipherment are also discussed.

4. Supporting Evidence and Discussion

4.1 Message content

Pendock is a village about 18.5 kilometers from Forli Malvern, Elgar’s home at the time the Dorabella was written. According to Moore (1984, 224), upon returning home from visiting Dora in Wolverhampton, Elgar stayed in the region into early August completing the orchestration of his work Te Deum and Benedictus. This indicates that he was within range of Pendock — about an hour by bicycle — on July 14, 1897 to write the Dorabella.

Further, Elgar and Dora were both connected to Pendock. Elgar’s wife Alice had studied geology alongside Dora’s stepmother, Mary Frances Baker, at Pendock Old Church under Rector William Samuel Symonds (Powell 1949, 1). There is also a small Georgian organ in the church, the type for which Elgar wrote two pieces of music, and it is thought that he may have played the instrument (Philipson-Stowe 2005, 7).

The botany of the Malvern Hills in mid-July would make it pleasant for a fern walk — a popular Victorian undertaking. Pendock Old Church has a strong historical connection to botany: Rector Symonds’ daughter, Lady Hyacinth Jardine, married famed botanist and director of Royal Botanic Gardens, Kew, Joseph Dalton Hooker, and a plaque in his memory hangs in the Church to this day (Philipson-Stowe 2005, 6). Between the botanic and geologic connections, a tongue-in-cheek reference to Pendock Old Church as an athenaeum — essentially a library for learning and research — is not a stretch, especially in teasing Dora by comparison to her more cosmopolitan home city of Wolverhampton.

Coming from Elgar’s pen, Whamp is a reasonable abbreviation for Wolverhampton. A survey conducted by the author of 426 of Elgar’s letters produced 39 instances of his abbreviations of cities when obvious to the letter’s addressee. Examples include WHampton (Wolverhampton), Hford (Hereford), Lpool (Liverpool), Glouc (Gloucester), Mvn (Malvern), Birm, Bham, B (Birmingham), Bkhptn (Brockhampton), and Mchester (Manchester), among others. A full list is provided in Table 1.

Table 1. City abbreviations by Elgar from a survey of 426 of his letters, in alphabetical order with spaces, apostrophes, and periods removed.

Table 2 lists 21 occurrences of Latin words and phrases in his English letters, including inter alia, imprimis, solus, sui generis, and passim, among others, making the use of sin vi de — without force — entirely plausible.

Table 2. Latin words and phrases used by Elgar, from a survey of 426 of his letters, in alphabetical order.

The survey of 426 of Elgar’s letters yielded many instances of phonetic spelling, a number of which use substitutions that match those in the proposed solution. These include a superfluous e (freambosa, negligeible, newes), using o to represent double-vowel /ō/ (freambosa, oq, wo, woful), ee for /ē/ (afeeons, deleerious, exspeerience), q to represent /k/ (oq, qopy), s in place of c to represent /s/ (pisay, serremony), z for the /z/ sound of s (bgotez, wizdum), the use of y in place of vowels (gyrn, thys, spellynge), and removal of vowels (scb, bgotez, sd, shd). Of course, Elgar’s motivations were more playful than rigorous, and a number of examples deviate significantly from reasonable phonetic representations (e.g. score: skore, skoughre, skourrghe and excuse: xqqq, xqqqq). The list in Table 3 gives an idea of his approach, range, and willingness to experiment.

Table 3. Phonetic and other unusual spellings from a survey of 426 of Elgar’s letters. Words that use the same phonetic substitution as found in the proposed solution are indicated with an asterisk.

According to Royal Botanic Gardens, Kew: Plants of the World Online (2022), raspberry (framboesa), poppy (afion), amaryllis (bogotas), oak, spruce (picea), and eucalyptus can all at least currently be found in England, though not all are native. Cantaloupe (netted gerns) are not listed, but can be grown as described by the Royal Horticultural Society (2022). Scab is a common fungus that plagues melons, and qat is more difficult to assess because of its amphetamine-like influence, according to the European Monitoring Center for Drugs and Drug Addiction (2022).

4.2 Other message traits

The proposed message exhibits some encouraging traits aside from the content. No mistakes seem to be made by Elgar. The phonetic spelling begins with the plant list and carries to the end, with all plants having phonetic alterations, and with no preceding words subjected to it. No words span between lines, which fits nicely with the different line lengths of the Dorabella: 29, 31, and 27 characters respectively.

Letter frequency counts of the solution are shown in Figure 4 along with expected values for English (Fauché Gaines 1956, 219). The agreement is reasonable, especially considering the obscurity of the message and that more than half is spelled phonetically — an aspect which explains deviations such as larger-than-expected counts for Q and Y.

Figure 4. Frequency analysis of the proposed Dorabella solution.

4.3 A closer look at the Rose Key

Elgar’s Rose Key has eight octants, each of which are divided into three, giving 24 slots for letters. Figure 5 shows the Rose Key next to another rose with the same structure, knitted by the author’s grandmother, Sylvia Belanger, around 1960.

Figure 5. Elgar’s Rose Key next to another rose of 8 x 3 structure.

The number eight is an important structural component in music; octaves are the basic frequency intervals that delineate different musical pitches — an appropriate choice of cipher key structure for a composer whose life centered on music.

Any Elgarian will tell you that the only force that rivaled music in Elgar’s life was his love and devotion to his wife, Caroline Alice, Lady Elgar. Note the sequence beginning at the center of the lowest octant and moving outward, then clockwise and inward:

FOR CALE…

The remaining letter sequence could be based on a pangram. Guesses could be ventured, but many are possible and none would be definitive, save the discovery of a phrase of documented importance to Elgar that matches perfectly.

4.4 The structure of the Dorabella

The codebreaker’s usual practice of direct transcription of arbitrary symbols as a first step (Fauché Gaines 1956, 68) foils decryption attempts of the Dorabella because the encipherment rotations visually decorrelate its symbols from their corresponding letters. Rotated symbols become indistinguishable from other symbols.

For example, one symbol occurs five times in the first line and, per the solution proposed in this paper, represents the letters K, A, P, S, and I. If those five symbols are considered identical then the cipher is polyalphabetic. However, with knowledge of the rotating key, the five instances are known to be distinct symbols, each from a different octant, that rotate in and out of position and only appear identical. The Dorabella has only a single alphabet, described by the Rose Key, making it inherently monoalphabetic, but it is cloaked as polyalphabetic by the key rotations. The work of Schmeh (2018) and later Hauer et al. (2021a) in finding indications of MASC structure offers evidence that the rotational cloaking did not entirely obscure the statistical monoalphabetic properties of the cipher.

From the polyalphabetic view, all symbols that look the same are considered identical, and a tableau can be constructed (Figure 6). The tableau begins with a linear representation of the key. Each row that follows simply applies a clockwise eighth rotation to all symbols in the previous row, until all eight key positions are represented. To use in a decipherment, start at the first row of symbols, find the first symbol of the Dorabella, and take off the associated letter above. Then move to the next row for the second symbol. When a dot is encountered, simply skip the designated number of rows. Encipherment would reverse the process.

It is worth noting that the symbol sequence in the U column of the tableau matches the “forward roll” sequence noted by Bauer on Elgar’s Cryptogram Card.

Figure 6. A tableau for the Dorabella.

5. The Enigma

A question still haunts the Dorabella: Why would Elgar give Dora a cipher that was difficult to decrypt and never help her solve it? One possibility is that the assembly of the Rose Key could have given away more than Elgar was willing. Elgar left another mystery — The Enigma — that refers to the unknown melody that underpins his Enigma Variations.

In the author’s opinion, the Rose Key points strongly to Elgar’s 1892 setting of the poem Like to the Damask Rose as the solution to the Enigma, the hidden melody. For an excellent overview on the Enigma with clear solution criteria, see Rushton (1999).

Like to the Damask Rose and Enigma Variations have an uncanny alignment of their respective 14 subjects; the structure of the poem mirrors the linking notes of Variations I-II, V-VI, and VIII-IX, and XII-XIII (Rushton 1999, 62–63); and Elgar gives seven hints that align with the poem’s subjects in his notes for pianola rolls published by the Aeolian Company (Elgar 1946). Most importantly, the two fit elegantly in counterpoint, and the result sounds very Elgarian:

The fit of “Like to the Damask Rose” with “Enigma Variations” in counterpoint, arranged by Steve Marlow.

If Like to the Damask Rose is the solution to the Enigma, then Dora was more revered than she knew, immortalized as Variation X not for her speech patterns (Powell 1949, 112), but as a bird that’s here to-day, akin to Elgar himself, the singing swan that sounded into death. Elgars wife Alice, of course, was the Damask Rose.

If Dora had found the encrypted message of the Dorabella, she might have noted the parallel between her 1897 youth and the vibrant botany of the Malvern Hills in mid-July. Even then, the scab was beginning to taking a morbid hold on the netted melon, just like the gourd that Jonas had in Like to the Damask Rose. Dora was the youngest of the Variations, the Intermezzo who outlived them all, and she may have been given the greatest chance of solving the Enigma. But youth and the luxury of time were not to be a free pass. When she pressed Elgar about the Enigma in November of 1899 (Powell 1949, 23) he replied “Oh, I shan’t tell you that, you must find it out for yourself.” and when she pushed more, she only received “Well, I’m surprised. I thought that you, of all people, would guess it.” It could be that Elgar offered Dora a private path to a solution to the Enigma, if she had only taken the first step and assembled the Rose Key.

The author acknowledges Nicholas DeMaison for his work on the fit of Like to the Damask Rose to the Enigma Variations theme in counterpoint, and for recognizing the number 14 in each piece.

References
Bauer, C. P. 2017. Unsolved: The history and mystery of the world’s greatest ciphers from ancient Egypt to online secret societies. New Jersey: Princeton University Press.

CrypTool 2. 2022. CrypTool Portal — cryptography for everybody. Accessed March 5, 2022. http://www.cryptool.org/.

Elgar, E. 1946. My friends pictured within: The subjects of the Enigma Variations as portrayed in contemporary photographs and Elgar’s manuscript. London: Novello.

European Monitoring Center for Drugs and Drug Addiction. Khat drug profile. Accessed December 29, 2022. https://www.emcdda.europa.eu/publications/drug-profiles/khat_en.

Fouché Gaines, H. 1956. Cryptanalysis: A study of ciphers and their solutions. 1st Dover ed. New York: Dover.

Gaffney, T. 2008. Dorabella Cipher (14th July 1897) Key. Accessed December 10, 2022. https://i262.photobucket.com/albums/ii112/tony59b/doabell.gif.

Hartmeier, D. 2006. Clues or red herrings to the Dorabella Cipher. April 27, 2006, https://www.benzedrine.ch/dorabella.html)

Hauer, B., C. Choi, A. Hindle, S. Smallwood, G. Kondrak. 2021b. Dorabella Cipher as musical inspiration. Proceedings of the Workshop on Speech and Music Processing, 33–38.

Hauer, B., C. Choi, A. S. Sundar, A. Hindle, S. Smallwood, G. Kondrak. 2021a. Experimental Analysis of the Dorabella Cipher with Statistical Language Models. Proceedings of the 4th International Conference on Historical Cryptography.

Moore, J. N. 1984. Edward Elgar: A creative life. New York: Oxford University Press.

Mulliss, J. 2022. “The Dorabella Cipher — the complete solution,” https://jamesmulliss.wordpress.com/the-dorabella-cipher-the-complete-solution/)

Packwood, W. 2020. Elgar as Cryptographer. Musical Opinion Quarterly, July — September 2020, 34–37.

Pelling, N. 2009. “Is Dorabella a rotating pigpen?” https://ciphermysteries.com/2009/03/19/is-dorabella-a-rotating-pigpen)

Pelling, N. 2013. “Dorabella Cipher first two words decrypted perhaps” https://ciphermysteries.com/2013/02/24/dorabella-cipher-first-two-words-decrypted)

Philipson-Stow, R. 2005. Pendock old church. London: The Churches Conservation Trust.

Powell, Mrs. R. 1949. Edward Elgar: Memories of a variation. 3rd ed. London: Methuen.

Reed, W. H. 1949. Elgar. New York: Pellegrini and Cudahy

Roberts, T. 2012. Solving the Dorabella Cipher. Accessed December 10, 2022. https://cryptologicfoundationorg.presencehost.net/file_download/33703a5a-9bbc-4183-85ff-16310eada267.

Royal Botanic Gardens, Kew. 2022. Plants of the world online. Accessed December 25, 2022. https://powo.science.kew.org/.

Royal Horticultural Society. 2022. How to grow melons. Accessed December 29, 2022. https://www.rhs.org.uk/fruit/melons/grow-your-own.

Rushton, J. 1999. Elgar: ‘Enigma’ Variations. Cambridge, UK: Cambridge University Press.

Sams, E. 1970. Elgar’s Cipher Letter to Dorabella. The Musical Times, 111(1524), 151–154. https://doi.org/10.2307/956733.

Schmeh, K. 2018. Examining the Dorabella Cipher with three lesser-known cryptanalysis methods. Proceedings of the 1st Conference on Historical Cryptography, 145–52.

ShadowWolf. 2019. The Dorabella Cipher solution. Accessed December 11, 2022. https://drive.google.com/file/d/1f4kssBlzrPkp5V1L677npTrbNgVmAsA8/view.

Wikipedia. 2022. Like to the damask rose. Accessed January 3, 2023. https://en.wikipedia.org/wiki/Like_to_the_Damask_Rose

Young, P., ed. 1956. Letters of Edward Elgar and other writings. London: Geoffrey Bles.

Young, P., ed. 1965. Letters to Nimrod: Edward Elgar to August Jaeger 1897–1908. London: Dennis Dobson.