By Gordon Rugg and Gavin Taylor
The story so far: We’re working on the D’Agapeyeff Cipher, a short ciphertext that’s never been cracked. It’s a good testbed for codebreaking methods. In this series of articles, we’re collecting together resources about the Cipher for other researchers. In the previous episode, we looked at the methods that D’Agapeyeff described in the first edition of his book Codes and Ciphers, where the D’Agapyeff Cipher appeared. In this episode, we look at the solutions to the worked examples in the same book, to see what insights they might give.
When you’re trying to crack a code, you look for any clues that might possibly give you some insight into the content of the ciphertext. If you know what the text is about, then you can work backwards from that, and improve your chances of finding a solution. That’s why, for a while in World War II, one of the safest postings for a soldier on either side was in an Afrika Korps forward observation post in the Western Desert (hence the title for this episode).
A couple of years ago Gordon visited Bletchley Park with two leading codebreakers and a musician, as one does. Bletchley Park is where the British codebreakers were based in World War II, the codebreakers who cracked the German Enigma code. We ended up having dinner with the tour guide, a fascinating man who told a lot of stories that aren’t part of the standard tour.
One of those stories was about a German observation post during the North African campaign – Rommel and the Afrika Korps, Montgomery and the Desert Rats, the battle of El Alamein, and all the other legendary names from that struggle. This observation post was a story on a different scale, just a few men in an outpost in the middle of nowhere, surrounded by empty desert where every day was as quiet as the day before and the day after.
It would make a great movie, if it had the right scriptwriter, because there was a reason that those soldiers’ lives were so uneventful while they were in that post. The reason was that Allied troops were under strict orders to leave them completely alone. Why? Because the Bletchley Park codebreakers loved them, for the predictability of their reports. Every day, those soldiers dutifully sent in their reports to base, using the Enigma code system. And every day, they dutifully reported the same thing, that nothing was happening. Which meant that as long as those soldiers were left safely undisturbed, the Bletchley Park codebreakers knew what those daily reports would say, so whenever the Afrika Korps changed the settings on its Enigma machines, the Bletchley Park people could simply work backwards from the observation post’s reports, and crack the new settings.
Did those soldiers ever find out that their dutiful reports changed the course of history? How would they have felt about it if they had known the truth? We don’t know. For the right scriptwriter, it could make a deeply insightful movie about truth and war and a whole batch of other things. For us, it’s the motivation for collecting together the solutions to the worked examples in D’Agapeyeff’s book.
A body of text like this is known as a corpus, plural corpora. Corpora are typically useful for two reasons.
One is the words within the corpus. If you know what the individual words are in a corpus, then you can use those to work backwards towards possible solutions for a ciphertext. In the case of the German forward observation post, the Bletchley Park codebreakers knew that the daily reports would be along the lines of “All quiet, nothing to report”. There are only so many ways that you can phrase that message, and most of them will include the word “quiet” and/or the word “nothing”. In the more famous case of the German Kriegsmarine, the U-boats would routinely send in weather reports, where the Bletchley Park codebreakers would know what the weather was in the areas where the U-boats were operating, and could therefore work backwards from that.
Another use for a corpus is working out the relative frequencies of different letters within the corpus. This type of frequency analysis has been a standard method among codebreakers since it was first described by Al-Kindi in the 9th century. Al-Kindi isn’t exactly a household name today, but he’s been famous among codebreakers for centuries. For example, the sixteenth century polymath cryptographer Girolamo Cardano, who invented the Cardan grille cipher method, was familiar with the work of Al-Kindi, and thought he was one of the greatest intellects of the Middle Ages.
For ordinary English texts, the most commonly occurring letters are so well known that they have their own phrase, etaoin shrdlu. It’s occasionally used as an in-joke in publishing and related fields.
For some specialist areas, however, the letter frequencies are different, which has implications for codebreaking. Telegraphic English, for instance, typically leaves out a and the where possible, which affects the frequency distributions.
So, that’s the background to today’s article. The next section contains our current draft of the corpus of solutions that D’Agapeyeff used in the first edition of Codes and Ciphers. It’s a prefinal draft of a work in progress, so there may be some minor errors in it, but it gives a good general overview of the type of language he uses. His language is mainly military, and often telegraphic, with one or two excursions into famous historical examples. Some examples are incomplete or partially gibberish, for instance because they show a partial decipherment in progress. Here’s an example of that.
One interesting point about this example is that it uses plaintext in French. That’s perfectly reasonable in a book about codes – the reader needs to get used to cracking codes written in a foreign language. However, it raises the question of whether the plaintext for the D’Agapeyeff Cipher might be in a language other than English. We’ll return to that question in episode 6.
Another of his examples was originally written in French, and is famous in codebreaking circles; he shows it in English translation.
If you’re wondering about the mention of the man in the iron mask – this solution is from a real historical ciphertext, the Great Cipher, referring to events that really happened, and it has been suggested that de Bulonde was The Man in The Iron Mask. Similarly, D’Agapeyeff uses the phrase “the enemy” several times, which may be a deliberate allusion to the corresponding French phrase that the French codebreaker Bazeries used as the way in to crack the Great Cipher.
The Great Cipher quote is the longest text in the corpus, and D’Agapeyeff’s phrasing is different from the Wikipedia phrasing. Among other things, the D’Agapeyeff version has “Coni” where Wikipedia has “Cuneo”. These issues will re-surface in a later article in this series, with regard to possible insights into what text the D’Agapeyeff Cipher might contain, and whether it might be possible to work back from those possibilities.
Our strategy is to get the overall picture of our D’Agapeyeff work published soon in these blog articles, and then to put a collated final version onto the Hyde and Rugg website when we upgrade the website. For the corpus, we’re producing a properly annotated document that explains the context of each solution – for instance, the NKOY example below relates to African talking drums, and in the solution that refers to “aeroplane(s) defence” there’s a reference on the next page to the different phrasing “aerial defence”. Since D’Agapeyeff shows partial solutions and makes comments in the text of some solutions, producing a corpus of his solutions isn’t a simple case of typing in the words; the surrounding context needs to be included. However, the corpus below should be enough to give readers a reasonable idea of the types of text that D’Agapeyeff uses.
In next week’s thrilling episode, Gavin finds a significant typo. Who needs Indiana Jones movies when you can have real life?
The corpus (the page number is shown for each solution; some solutions occur across several pages during worked examples)
P28 Do not use bearer
P35 Food supplies running out
P43 Letter sent to the emperor giving full details
P47 It is not necessary to explain to you in detail how ill-pleased is His Majesty having learned that against your orders M. de Bulonde has raised the siege of Coni and that he retreated even in a disorderly fashion. H.M. knows very well the consequences of such disaster and the prejudice with which the news will be received that this fortress was not taken, and that we shall have to repeat this operation during the winter.
H.M. desires that you arrest M. de Bulonde and accompany him to the Citadel of Pignerole where H.M. wishes him to be incarcerated during the night, being allowed to walk on the ramparts during the day but wearing an iron mask.
As the Governor of Pignerole is under your orders will you send him the necessary authority for execution.
P50 Re-union to-morrow at three p.m. Bring arms as we shall attempt to bomb the railway station. Chief
P68 King Henry the 8th was a Knave to his Queen
P86 GALLANTLY and FURIOUSLY he fought AGAINST the foe at WATERLOO.
IVY creeping along the ground suggests HUMILITY
The JURISDICTION of the NOBLE LEGISLATOR was offensive to the BARBARIAN.
A PHOTOGRAPHER saw SEVERAL KANGAROOS on the MOUNTAIN
I will arrange for (CC) XYKCD MQRAD GVK (CC) and after that (CC) XZALG MSAQW QAL (CC) Instructions requested.
PP111-112 I will arrange for (CCFI) 23301 89972 41 (CC) and after that (CCFI) 29330 61100 465 (CC) Instructions requested
P117 ENEMY ATTACKS AT DAWN
P118 ENEMY ATTACKS
P121 Hold your brigade in readiness to move
P124 Advance brigade to-morrow
P126 Meeting brigade village cross roads
P134 ‘Le prisonier est mort – il n’a rien dit’ (‘The prisoner is dead; he has said nothing.’)
P138 -141 General ordered the second brigade to attack at 3.30 A.M. the third brigade one hour later the fourth keep in reserve
P145 The new plan of attack includes operations (EA) (BE) those (EA)o(AD) (EA) I s (DE)u(AB)ions over (BA)actor(BE) are(BE)a south west o? the river.
P146 No one here has deciphered the three latest dispatches, please discontinue these ciphers as the ones used hitherto were better.
P152 The third battery to take up north position
PP154-156 The reconnaissance of the route to the sea has revealed that the aeroplane(s) defence was over the town
For more about letter frequencies in codebreaking, the Wikipedia page is a good introduction:
There’s more detail in Wikipedia here:
Wikipedia also has an article about Al-Kindi, who first described the frequency approach to codebreaking:
For de Bulonde and the Iron Mask: