According to a nice little 2004 New Scientist article by Kevin Jones (Professor of Music at Kingston University, my most recent alma mater), even though Elgar composed his cipher note to Dora Penny in 1897, he appears to have reused the same 24-token cipher alphabet in an exercise book 30 years later. (Kevin Jones doesn’t mention in which collection the exercise book is to be found: there’s a nice listing of Elgar’s notes and immense collection of letters here.)
As with the majority of self-conceived ciphers, it was born of a simple idea:-
[Elgar] listed the symbols used in the Dorabella cipher matched against the letters of the alphabet. The cipher follows a simple pattern, with single, double and triple E-like characters, each in eight possible orientations – upright, rotated 45 degrees clockwise, 90 degrees clockwise and so on. This gives a total of 24 potential characters, and as with many ciphers, I and J share a single character, as do U and V.
Elgar then tries it out on some samples, which when deciphered read:-
M-A-R-C-O E-L-G-A-R (Marco was his pet spaniel) and A V-E-R-Y O-L-D C-Y-P-H-E-R. But when applied to the Dorabella cipher this key does not generate anything that makes obvious sense.
It certainly was “a very old cypher” (probably 30+ years old at that stage). But there’s something a bit back-to-front about this whole thing. If he was reusing an old cipher, why would he be going through the palaver of trying it out again? He would surely have gone through his experimenting phase decades before? But according to Kevin Jones’ subsequent notes to the 2007 BBC Proms:-
Elgar scribbled an 18 character code using the same cipher symbols in the column of printed programme notes for a concert he attended at Crystal Palace in April 1886 – opposite a musical example from Liszt’s “Les Preludes”. (Copy at the Elgar Birthplace Museum.) Annotations on other pages are not ciphered – so it’s possible that this may have been added at a later date.
And so even though this was used as a cipher circa 1886 (probably), and post 1927 (probably), was it also one circa 1897? All these scraps muddy the water once again – which is perhaps what Elgar was hoping to achieve. I just wish we knew what Dora Penny’s favourite song was…
Interestingly, one of the comments to this page was by Peter Brooks, who said he was “increasingly confident that the message consists of two parts separated by an evident period on the last line”, with a first apart in Latin and the second in some kind of vertically arranged English. Personally, I’m not sure how that would be any less obscure than the solution proposed by Eric Sams discussed here recently: but I’m sure Peter Brooks has plenty of sensible reasons to back his notion up.
Following on from the Proms post, “The Elgar Apostle” (“the Elgar on-line newspaper”) held a Dorabella cipher competition, which “seven individuals were brave enough to submit entries”.
The final Dorabella bombshell of the day comes from Peter Brooks, who noted (in his comment) that “there is a moderated Yahoo group Elgar-Cipher“. If you want to find out more about the Dorabella Cipher, this is surely the first place you’d want to head towards.
Incidentally, the “enigma” of the 1899 “Enigma Variations” was Elgar’s claim that they all played in counterpoint to a well-known melody (which he never disclosed, and which has never been worked out) – might the Dorabella Cipher be enciphering this tune, too? (The timing would be basically right.)
PPS: the German WWII Enigma machine was (apparently) specifically named after the Enigma Variations: yet another non-obvious connection between music and cryptography…