I recently read an article on Nautilus about the Dorabella Cipher scribbled by Edward Elgar in 1897, before he became a famous composer. The cypher is made up of geometrically regular glyphs, and was given by Elgar to his young friend Dora on what seems to have been a whim. Some people claim to have decrypted it, but these solutions cannot be verified and there is no reason to expect that definitive decryption is possible.
My own thoughts and illustrations follow. It’s not that I have much to say, but a blog post is a good medium for integrating text and pictures. Also, all images embedded in this blog post are public domain, so if you want to play with the cipher yourself you’re welcome to make use of them.
First, the original cypher:
As you can see, each glyph consists of one, two or three cusps, facing in any of eight directions. Four of those directions (left, right, up and down) I will call upright and the others (the diagonals) I will call slanted.
The geometric regularity is part of the intrigue, but when you start examining the cypher, you soon notice that Elgar’s handwriting is not exactly regular. It isn’t always clear which direction a particular glyph is supposed to be facing. For example, take the fifth glyph on line two, or the third glyph from the end of line three — in isolation these look mostly upright, but if you interpret them that way then other glyphs, ever so slightly more tilted, become ambiguous. No matter where you draw the line there are always edge cases, and I think the only solution is to document how you have interpreted the glyphs so that if other people want to change anything they can look it up and see precisely what to change.
Below is the cypher again, with the glyphs I’ve interpreted as upright in black, and the glyphs I’ve interpreted as slanted in red.
When I decided to play with the cipher, I began by replacing each glyph with a perfectly regular design, on the grounds that this might make any patterns become more obvious to the eye. I then compiled these designs into the following animation.
The outer circle represents glyphs with three cusps, the middle circle those with two cusps, and the inner circle those with one. The eight sectors represent the different directions the glyphs can face: the mapping is arbitrary, but I took the direction faced by the open side of the cusps and mapped it to the first sector clockwise from the corresponding line.
Here are the 87 frames in sequence (left to right, top to bottom, as you would read).
The table below shows the number of times each element occurs. You should be able to translate these back to the original glyphs.
Of course it’s possible that Elgar’s scribble isn’t a substitution cipher at all. It could represent a piece of music, or dance steps, or the location of buried treasure. Music seems unlikely, though, as nothing in the tally suggests a pattern indicative of a bias toward dominant notes in the scale, or other regularities you’d expect.
The directions most often faced by the open sides of the glyphs — using the standard map convention — are: northwest (corresponding to the column where the numbers are 8, 11, 4), southeast, south, and then a tie for fourth place between north and east. Curiously, if we give precedence to north, then the most and least common orientations alternate in pairs (probably a coincidence but worth noting nonetheless). Below is the sequence of frames again, with the four most common orientations shaded maroon.
The first pattern I noticed after watching the animation you saw earlier is that a particular glyph often has either the same orientation as its immediate predecessor or the exact opposite orientation. This occurs 42 times (take note, Douglas Adams fans) out of a possible 86, whereas by chance it would occur a quarter of the time. But taking into account the distribution of the four most common orientations — which I had not tallied at the time — increases the chance prediction from 21.5 to 34 occurrences (I think), so the discrepency is greatly reduced. The 42 occurrences are highlighted below.
Examining the above further shows that 23 of the 42 cases (more than half) involve the northwest and southeast directions. Of these, the most common are a northwest followed by a southeast (9 occurences), or a southeast followed by a northwest (8 occurrences). A northwest follows a northwest 5 times, and there is only one pair of consecutive southeasts.
That’s basically all I have. I just wanted to mess around with the cipher for a while, to see what I could make of it. I didn’t expect to make a historical breakthrough; I expected to have fun. I hope this has been fun for you too, and perhaps you are inspired to explore further.