The Dorabella Cipher Explained


In 1897, British composer Edward Elgar sent an apparently encoded message to a young lady named Dora Penny.

She did not successfully decipher it and only later published it in a memoir in 1937, four years after Elgar died. It is 
known generally as the Dorabella Cipher, after a nickname the composer gave to Miss Penny.  Over the years that 
have followed its publication, it has defied all attempts to convincingly crack its code.

I will establish in this article that the Dorabella Cipher is a nonsensical message by demonstrating that it contains 
two separate traits, each one of which is so statistically improbable that it would cast serious doubt on this being a 
legitimately coded message.

I worked for four years at the National Security Agency after 9/11 as an Arabic linguist. Even though I was not a 
cryptographer, I was formally trained there in the principles of code making and breaking.

As you can see, the Dorabella Cipher is constructed from symbols of one, two, or three semicircles, each of which has 
eight potential orientations.

Years after sending this message to Miss Penny, Elgar included a description of a similar code in one of his journals. 
You can see that he lists the symbols by orientation, and assigns these symbols to the letters of the alphabet in order, 
making I and J use the same symbol, as well as U and V, to fit the 26 letters of the English alphabet into the 24 
symbols of this code.

Now, when you apply the letters of the code from his journal to the Dorabella Cipher,  you get absolute gibberish:


So, clearly, either the Dorabella Cipher assigns different letters to the symbols or something else is going here.

I will now demonstrate that the something else going on is that Edward Elgar used these symbols to intentionally 
create a nonsensical message. This may have been intended as a playful joke. Perhaps after she had tried and failed 
to solve the puzzle, he planned to eventually tell her that it was undecipherable. In her memoir, however, she does 
not indicate that they ever discussed the matter again after it was sent.

Like many other well-intentioned scholars, I too attempted a decipherment of this puzzle. But in the course of my 
efforts, I noticed two very peculiar things.

Number One. In each of the three lines of the Dorabella Cipher, there is a long series where there is no repeat of a 
symbol with the same number of semicircles.

Again, recall that there are three types of symbols in this message, symbols with one, two, or three semicircles. 
Because the letters are supposedly assigned to the arbitrary code in alphabetical order, these long series would 
appear to be statistically improbable.

To explore this issue, I encoded an actual English text with two different versions of the Elgar cipher—the one he 
showed us in his journal and another formed by listing all eight orientations of one semicircle, followed by two 
semicircles, and so on.

Using these two codes, I encoded the first three lines of the first stanza of Lord Byron’s poem “She Walks in Beauty,” 
because it has roughly the same number of characters as the Dorabella Cipher.

I created two versions of this test because I very much wanted to discover what we should expect an actual encoded 
message to display in terms of stretches without repeated symbols of the same number of semicircles.

Here’s what we find. The longest series without a repeat in my two controls is six long. The other control had its 
longest series only five long.

So now let’s look back at Edward Elgar’s message to Dora Penny. At the very beginning of the cipher, there are a 
baffling twelve symbols in a row without a repeated number of semicircles. But if that weren’t enough, the second 
line of the cipher starts with nine symbols in a row without such a repeat! And if even that weren’t enough, line three 
has a series eight long with no repeated symbol of the same number of semicircles.

Now, in a randomly generated message, the chances of there being a series of twelve symbols in a row without 
repeating a symbol of the same number of semicircles would be one in 532,793.  Our two samples certainly 
demonstrate that a series of twelve in an actual encoded message is not expected. To then see an additional such 
series of nine in line two and then one eight long in line three becomes statistically improbable to say the least.

But then note something else quite peculiar about these three series. Two of them occur at the very beginning of 
their lines, and the third is found in the first half of its line.

Keep in mind, the potential that the Dorabella Cipher is a real message and yet is employing encryption beyond a 
simple substitution cipher does not explain away this observation. Quite the contrary, additional encryption would 
make the symbols on the message even more random, and thus render these series all the more statistically 

This is consistent with my theory that he was playing a joke on Dora, creating a message that could not be 
deciphered because there was no message here. And so, he starts each line working to create pure randomness. As 
he creates this joke, however, he succumbs to normal human laziness, and he then begins producing a second trait 
that only further proves there is no real message here at all.

And so, on to my second observation. There are simply far too many mirrored symbols in the Dorabella Cipher.

There are two ways that a symbol could potentially appear to our eye to be a mirror image of itself. Using symbols 
from my controls you can see that the symbols pointed directly north, east, south, or west have only one symbol that 
is their mirror image.

The other four orientations have potentially two symbols that seem to mirror one another.

There is the symbol that is a true reflection:

And there is the symbol that mirrors by inversion:

Let’s calculate the rate at which we should expect to see mirror images in the Dorabella Cipher.

There are 24 total symbols. Twelve of them have only one mirror image. Twelve have two. When we calculate the 
possible mirrors that could occur after each symbol in each separate line of the Dorabella Cipher, we predict either 
type of mirror to occur 5.5 times total in the message. (Adding up total possible mirrors to the end of each line and 
dividing by 24, the number of possible symbols.)

Instead, there are 13 times in the Dorabella Cipher when a symbol is followed by what appears to be a mirrored 

For a window into the significance of this development, let’s look back at my two controls on this very point.

The message using Elgar’s code predicts 5.3 mirrors. Six happen. That’s not a statistically significant difference.

The alternate code predicts five mirrors. That’s exactly what we see there.

And yet the Dorabella Cipher includes far more mirrors than we should have expected.

I assert this happened because, after he started each line, avoiding symbols of the same number of semicircles, he 
then understandably got a bit lazy and finished out lines by occasionally dropping in various types of mirrored 
symbols. And in the second line, he got so casual with this exercise that he produced something truly statistically 
impossible—four mirrored pairs in a row.

And if all that weren’t enough, look at this. With the single exception of line three, all of these mirrors occur only 
after the curiously long stretches of no repeats on semicircle number.  Indeed, can it really be a coincidence that 
every time he ends these statistically improbable series, he immediately includes mirrored pairs? And two times he 
has two mirrored pairs in a row?


In the end, there are indeed two statistically improbable or even impossible things happening in the Dorabella 
Cipher. One of the statistically improbable things occurs only in the first half of each line, the other tends to occur in 
the second half of each line. Either one of these traits, happening anywhere in the lines, would cast serious doubt on 
this being a real message. The fact that instead they appear grouped as they do on separate sides of each other 
becomes a final and convincing proof that Edward Elgar was constructing a nonsensical message as a playful joke for 
a friend.

The Dorabella Cipher was indeed a joke that Edward Elgar once played on a young lady. Because she published it 
years later, it then turned into a joke he accidentally played on all of us who love a good cryptological puzzle. But I 
believe I have demonstrated in this video that it was indeed just a joke, and therefore it deserves no further time or 
energy from any of us.

No comments:

Post a Comment