Category Archives: Historical Ciphers

A few days ago I posted a list of open questions about the dead cipher pigeon, really as a way of externalizing the annoyance I felt from knowing so few basic facts. To my great delight, Mike Moor from Melbourne and (well-known military history buff) Christos T. stepped forward with a whole wheelbarrowful of answers. And here they are…

“Did any British pigeon handlers ever use “lib.” as an abbreviation?”

Mike Moor points out that the first message sent back on D-Day was by Reuters reporter Montague Taylor, attached to the eg of the war-seasoned (and subsequently Dickin-Medal-receiving!) carrier pigeon Gustav [NPS.42.31066]. At the bottom of the image (clearly on an RAF pigeon message pad), it says “Liberated 0830” (click to see the full message):-

“Why can’t I find a single other message written on the same printed pigeon service pad?”

For this, Mike Moor points to a message sent by Major General Roberts on a page talking about the Canadian armed forces’ involvement in World War Two. [Incidentally, the abortive Canadian raid on Dieppe was known as “Operation Rutter”, I wonder if Stu R knew that?] Even though the quality of the scan is frankly diabolical, it’s very much better than nothing at all, and tells us that this our pigeon message was (without any real doubt) an Army Pigeon Service message pad.

Mike also notes that this was an “Army Book 418B”, the updated version of the Army Book 418 used for pigeon messages in the First World War. It turns out that the National Army Museum near Sloane Square tube in London has an Army Book 418B in its collection described as “Army Book 418B, Pigeon Service Message Book, 1942”, accession number “1975-06-35”: it would be cool to ask the curators there to have a closer look.

“Was the pigeon message we have a hectograph or a carbon copy?”

Mike Moor notes “It is a carbon copy pad with 1 original retained in the book and 2 carbon copies made – which lines up with what you’d expect from the message i.e. 2 copies sent and the blue text of the cipher looks a lot like a carbon copy + black amendments by a second hand presumably prior to sending.” Excellent, thanks! 🙂

“When did Slidex change from having one letter per horizontal key slot to two letters per slot?”

The (plainly utterly indefatigable) Mike Moor points us to some December 1944 Slidex instructions available on Rob van Meel’s site (a copy will cost you two euros plus international postage from the Netherlands), by which time it had changed to two letters per key slot on the horizontal cursor. That narrows the range down dramatically to ‘sometime in 1944’… we’ll just have to keep digging to find out exactly when in 1944. At least this is a question that we can reasonably hope to get a solid answer on!

“When were Slidex Series B code cards introduced?”

In the Series A “RE No. 2” (Army Code No. 14070) Card 35 that I got from the excellent royalsignals.org.uk website, the three columns have had their shape changed to break up the columnar structure somewhat, which I believe may point to a rethink & upgrade of the Slidex code during WW2.

At the same time, another Series A card has two versions, one with an Army code and another with a different W.O. (War Office) code, which I suspect points to a post-WW2 handover from the Army to the War Office. But that’s as good an answer as this question has for now.

“Did Bletchley Park / GCHQ ever catalogue the tons of files brought back by the TICOM teams?”

Christos replies: “There are many TICOM file categories: I, IF, DF, M, D. Captured German documents had to be catalogued and then translated. This must have taken years. The question is whether there is a full list of those files. There is a DF list but I don’t know about any document covering the other files.”

Incidentally, p.38 of TICOM I-109 (a report by Lt Ludwig of Chi Stelle OB.d.L) says:

B. Slidex system.

Bigram substitution system.
In use in the army (front line units) and in air support networks (tentacle networks).
The system was known from the monitoring of exercises in Great Britain before the invastion, e.g. “Spartan”. The cryptanalytic detachments in army and GAF were able to get so much experience on these exercises that decoding worked well right at the start of the invasion.
Recovery was done in the army again at NAA St 5, in the GAF in 14/3 (W control 3).
Decoding was often done with so little delay that the messages could be dealt with like clear text in the evaluation.
The results were of more importance to the army than to the GAF, but theu provided the latter too with valuable indications, e.g. elucisation of the individual corps tentacle networks, reconnaissance operations (e.g. 400 and 414 Squadrons) etc.
The messages decoded daily were exchanged between Army and GAF in the form of written reports.

As far as just about any cipher I blog about goes, I have days when a solution seems so comically close I could almost accidentally breathe it in. But I also have days when one seems so tragically far away that it may as well be in a sealed box. On the moon. Guarded by killer robot ninjas. All voiced by James Earl Jones.

What’s at issue isn’t anything like pessimism on my part: rather, it’s the question of why anybody would think these might be solvable without doing a whole load of basic, grindy, grafty research first. Really, I think you almost always have to break the external history of these things before you stand much of a chance of extracting their internal contents.

So, if I list some of the open questions that are bugging me about the dead cipher pigeon story right here right now, perhaps you’ll see why that might be the case:-

* Who owned either of the two pigeons listed on the message?
* Were there any pigeon lofts in Bletchingley itself during WW2?
* Did any British pigeon handlers ever use “lib.” as an abbreviation?
* Why hasn’t a single record from a pigeon loft around Tunbridge Wells or Dorking turned up yet?
* Why can’t I find a single other message written on the same printed pigeon service pad?
* Was the pigeon message we have a hectograph or a carbon copy?
* From its skeleton, how old was the dead pigeon? [What a basic question to have to ask!]
* True or false: “many WW2 pigeon messages were sent encrypted”?
* When were Slidex Series B code cards introduced?
* When did Slidex change from having one letter per horizontal key slot to two letters per slot?
* How many syllabic / bigram ciphers were in use in WW2?
* Where is a copy of the Army syllabic cipher book BX 724 or BX 724/RE?
* Did Bletchley Park / GCHQ ever catalogue the tons of files brought back by the TICOM teams?

Personally, when I look at this fairly long list, I don’t feel hugely confident that we genuinely know even close to enough to enable us to solve this mystery, however engaging and intriguing it may be.

But then again, a single answer to any one of these questions from an unexpected corner might well be enough on its own to turn this miserable tide around. So perhaps we should just try to remain optimistic, for a tiny bit of clarity for any one of these might be enough to get us started. Fingers crossed that we shall see (and very soon!)… 🙂

Having got hold of a WW2 Slidex manual, I’m starting to see what a travesty of a coding system it was – and how a smart German decrypter could (with a bit of practice) decipher it almost in real time.

But Slidex was never intended as a highly secure coding system: it was only supposed to be a convenient way for people to discuss very short-term matters mostly in the clear over the radio or telephone, encoding any individual items or details that should not be overheard by the enemy. In fact because of its obvious lack of security, I don’t believe it was even classified as a “low grade” cipher.

Yet for a whole host of pragmatic reasons, it seems that Slidex was chosen to be used on D-Day for all non-machine cipher traffic. So if the pigeon cipher is in Slidex, can we crack it?

The first big problem we face is that “Slidex as described in the manual” and “Slidex as used in the field” seem to be quite different beasts. For example, though both employ a 12 x 17 (= 204-cell) table, manual-Slidex used a 12 letter horizontal key at the top, whereas field-Slidex (as evidenced by various pictures) seems to have used a pair of characters per key horizontal cursor cell, hence a 24 letter horizontal key. Similarly for the vertical key, manual-Slidex used only 17 characters whereas field-Slidex seems to have used all 26 alphabetic characters. So unless someone kindly comes forward and tells us how Slidex was actually used circa D-Day, we’re kind of stuck in a no-man’s land between manual-Slidex and field-Slidex, uselessly trying to guess what’s inside the Baggins’s nasty little pocket, yessss.

The second big problem is that unless something rather miraculous emerges from GCHQ’s archives, it now seems fairly unlikely we will ever get retrospective access to the daily keys used. Key pairs were tightly controlled and never used for more than 24 hours at a time (keys were normally changed over at midnight each day).

And yet… Slidex is designed to be quick and easy to use, with exactly the same code table and key pair used by both encoder and decoder. And in practice, all the symmetries and shortcuts that yield all that convenience also compromise the security to the point of uselessness.

For example, every 12-column-wide code table is arranged into three groups of four columns: and in each one I’ve seen, each group includes all 26 letters of the alphabet in alphabetic order, as well as a third of the numbers from 00 to 99, and a few SWITCH ON and SWITCH OFF cells. Moreover, in the code half of the cells, all the words are arranged in alphabetic order from top to bottom.

For example, the first four columns of the Royal Engineers Series A code table “No. 1” proceed like this:-

[?] 08  N  T
0/? G   16 24
OFF 09  O  25
00  H   P  U
A   I   17 26
01  1/? 18 ON
B   J   19 V
C   10  2  27
02  11  Q  W
D   K   20 X
03  12  R  28
E   13  S  29
04  L   ON 3
05  14  21 Y
06  M   22 Z
F   ON  T  30
07  15  23 31

(Original picture from Jerry Proc’s Slidex page, second image from bottom. Note there’s also a photo of an Op/Sigs code table there that closely follows the same kind of layout pattern).

Moreover, the German cryptologists interviewed by TICOM after the war noted that before September 1944, most people using Slidex tended to use the leftmost group of columns almost exclusively, which compromised yet further what was already a poor system. And the widespread habit of using Slidex for entire messages made the daily keys easier to get to rather than harder. What a mess!

And so if our mysterious dead pigeon message is in Slidex, all those flaws and poor enciphering practices might give us enough to decrypt it without a daily key, or even without a code table at all! After all, if the Germans could do it (albeit usually with more depth to work with), surely so can we?

Looking again at the bigram, if we precede each bigram with the number of times the first half of the pair occurs, I suspect we can predict fairly reliably which part of the message is in code and which part is in cipher:-

6HV 3PK 3DF 3NF 2JW 2YI 3DD 2CR
4QX
... 1SR 3DJ 6HF 3PG 1OV 3FN 4MI
4AP 1XP 4AB 1UZ 1WY 2YN 3PC
... ... ... ... ... ... ... 4MP
3NW 6HJ 4RZ 6HN 2LX 4KG 4ME 4MK
4KO 3NO 2IB 4AK 1EE 4QU 4AO 4TA
4RB 4QR 6HD 2JO 3FM 4TP 1ZE 6HL
4KX 3GH 4RG 3GH 4TJ 4RZ 2CQ 3FN
4KT 4QK 2LD 4TS 3GQ 2IR

Because so many single instance “1nn” pairs are clustered in the middle section (“1SR … 3PC”), I’m pretty sure that this is in code, and the last part (“4MP … 2IR”) is in cipher. The first part I’m unsure about.

If we now concentrate purely on the final section and look at the frequency counts and patterns there, plenty of other interesting things jump out:-

... ... ... ... ... ... ... 3MP
2NW 4HJ 4RZ 4HN 2LX 4KG 3ME 3MK
4KO 2NO 2IB 2AK 1EE 3QU 2AO 4TA
4RB 3QR 4HD 1JO 2FM 4TP 1ZE 4HL
4KX 3GH 4RG 3GH 4TJ 4RZ 2CQ 2FN
4KT 3QK 2LD 4TS 3GQ 2IR

From the way they cluster, I think that M and Q probably refer to the same column: and from the few single-instance “1nn” bigrams in there, I suspect that “EE” and “ZE” probably both encipher “E” (while “JO” could well encipher “T”).

What’s interesting is that it seems likely that the four columns encipher letters in alphabetic order: so (say) A-F, G-M, N-T, and T-Z in the case of the Royal Engineers code table #1 (there are a couple of extra E and T characters inserted around the table). It may be that this is enough for someone to try to solve this directly without any other information!

I’ve been wondering whether the pigeon cipher might be based around bigrams, i.e. where you split the cipher into pairs. If you disregard what seems to be the ‘AOAKN’ key indicator at the start and end, there are two basic ways to split the message into pairs. What is immediately interesting is that if you start immediately after AOAKN, you get three repeated pairs in the message (rather than just one), and that four of those six repeats occur every other pair along a short stretch towards the end of the message.

AOAKN
HV PK DF NF JW YI DD CR QX SR DJ HF PG OV FN MI AP XP AB UZ WY
YN PC MP NW HJ RZ HN LX KG ME MK KO NO IB AK EE QU AO TA RB QR
HD JO FM TP ZE HL KX GH RG GH TJ RZ CQ FN KT QK LD TS GQ IR U
AOAKN

AOAK
NH VP KD FN FJ WY ID DC RQ XS RD JH FP GO VF NM IA PX PA BU ZW
YY NP CM PN WH JR ZH NL XK GM EM KK ON OI BA KE EQ UA OT AR BQ
RH DJ OF MT PZ EH LK XG HR GG HT JR ZC QF NK TQ KL DT SG QI RU
AOAKN

The WW2 “Slidex” code works by having a table of codes particular to each arm of the armed forces (i.e. Royal Engineers, Operations/Signals, etc) where each bigram pair encodes a particular word often used by that arm. To encipher numbers or words, you bracket them inside a pair of SWITCH ON and SWITCH OFF bigram codes (there are several of each), and pick bigrams that correspond to each letter or digit or digit pair.

Hence I strongly suspect that what we are looking at in “GH RG GH TJ RZ CQ FN” is a sequence of letters enciphered in the Slidex SWITCH mode, i.e. where each letter is enciphered as a bigram.

Unsurprisingly, I would really like to read the 1944 Revised Instructions for using the Slidex R/T Code, and to have copies of the various Slidex code sheets and coordinate offsets used by the Royal Engineers and Infantry Brigade on D/Day, as I suspect these will allow us to read this message (does anyone have copies of these I can see?) Fingers crossed!