I spent last Saturday at the National Archives in Kew, accompanied by fellow programmer Stu Rutter who is just as fascinated by the whole pigeon cipher mystery as I am. Between us we did a kind of “Extreme Programming” two-man research team thing, where every time one found something unexpected or cool or had an insight, he would call the other over to see, or we’d both go downstairs to the café and discuss where we’d got to over a sandwich or whatever (reminder to self: don’t choose the double chocolate muffin again, it wasn’t very good).
The big result is that we ended up (I’m pretty sure) working out the secret history of this dead pigeon… and we didn’t even need to break its cipher (though Stu’s still on the case, more on that below). I’ll divide the overall argument down into a series of individual steps so that any passing Army historian who wants to take me to task over any detail can do so nice and easily. 🙂
1. Percy was an Army pigeon
I was already pretty sure of this: when I looked up all the “A Smith”s in the armed forces, every single “Serjeant” [with a ‘j’] was in the British Army. But what emerged at the National Archives were two widely-distributed pigeon-related documents (one from 1941, the other from 29th Jan 1944) that made it absolutely clear what different colour pigeon-carried canisters meant:
WO 205/225 – see pages 1/2/3 (at the back of the folder)
* Red = US Forces + British Army
* Blue = US Forces + British RAF
* Blue with coloured disk = British RAF
* Blue with white patch = RAF
* Red with coloured disk = British Special Service
* Grey = British Special Service
* Green = British Special Service
* Black = British Civil Police
* Yellow = British Commercial
Hence our dead pigeon was an Army pigeon; or (to be more precise) a NURP pigeon commandeered by the British Army.
2. The signature is that of Lance Serjeant William Stout
Knowing for sure that we’re looking at an Army message helps us narrow down the list of suspects to (I’m quite certain) one and only one individual – Lance Serjeant William Stout of 253rd Field Company of the Royal Engineers (as predicted here before), whose war grave says he died on 6th June 1944, the day better known as D-Day. You’ll read far more about Stout further down…
3. The pigeon was sent from France by a French speaker
I’m 99% certain that the “lib. 1625” writing on the cipher message was short for “libéré” (released) in French, and hence it seems overwhelmingly likely to me that the pigeon was released in France. But because there were no Royal Engineers at all in France (or indeed in Holland or Belgium) between the Dunkirk Evacuation (27th May 1940 – 4th June 1940) and D-Day (6th June 1944), the pigeon can therefore only have been sent either on or before 4th June 1940, or on or after 6th June 1944.
This gives us two “bubbles” of historical possibility to consider, the first ending with Dunkirk, the second starting on D-Day. How can we possibly tell which one of these was the right one? And can we be even more specific?
4. The pigeons were not released on or before Dunkirk
We’re helped here by the two pigeons’ ring tag identifications. Pigeon “NURP.37.DK.76” was the 76th pigeon to be ringed in 1937 by the DK (probably Dorking, we think) group of NURP [National Union of Racing Pigeons] pigeon fanciers, while “NURP 40 TW 194” was the 194th pigeon to be ringed in 1940 by the TW (almost certainly Tunbridge Wells) group. Yet having now read up on war-time pigeon-breeding administrivia at the National Archives, I know that (a) pigeons don’t normally breed in Winter; (b) pairs of pigeons will typically produce up to three pairs of eggs in a year, starting in Spring; (c) a young bird aged 6-8 weeks is called a “squeaker” and needs a fair bit of training before it is ready to race; and (d) pigeon races held in the first part of the summer almost never involve pigeons born earlier that year.
Hence if pigeon “NURP 40 TW 194” flew on 5th June 1940, it would have had to have been born in the middle of Winter right at the start of 1940: and the chances of there having been 193 other birds born and ringed earlier in 1940 among a single group of lofts around (say) Tunbridge Wells are extremely close to zero. Once you look at it like that, I think that there is no real chance that this pigeon was sent back from the Dunkirk evacuation, because it would simply have been too young.
5. The pigeons were released on D-Day itself – 6th June 1944
Here, the letter and number groups in the ciphertext itself give us the clues we need. Having also read up on the multitude of Army ciphers used in WW2 at the NA, I’m 99% certain how the structure of the wrapper around the cipher was contructed. Firstly, whatever system was employed for the cipher itself, the AOAKN letter group (which appears at the start and at the end of the message) is very likely an obfuscated or enciphered key reference for the message as a whole. And if this is right, then 1525/6 must surely hold the time and day of the month the cipher was sent at. “1525” = 3.25pm, “6” = “6th June 1944″… i.e. D-Day itself. But as we will see with Lance Serjeant Stout, this is the only day he could have sent it… though not quite as simply as you might expect from his gravestone.
6. Lance Serjeant Stout was mortally wounded on D-Day, and died on 28th June 1944
Right before the National Archives closed on Saturday, Stu & I managed to sneak a few minutes with the 1944 War Diaries for 253rd Field Company in the locked room at the back (someone else had put these diaries aside for photocopying this precise page, which made Stu and me both wonder if he or she might be a Cipher Mysteries lurker? Well, a big hello to you if that’s you!). Here’s what the entry for D-Day says:-
Well… not really very informative, you might think at first glance. Yet here’s where Stu Rutter really shone: having taken a photo of every page for June in these War Diaries, he then checked them all that evening and was pleasantly surprised to find an informative entry discussing Stout on the 28th June 1944:-
From this, we know for sure (a) that Stout was indeed in Normandy on D-Day; (b) that he was mortally wounded near Hermanville-sur-Mer (where he was later buried in the War Cemetery); (c) that he died of his wounds on the 28th June 1944; and that (d) he was in No. 2 Platoon. Stout was an NCO (“non-commissioned officer”, i.e. someone who had advanced through the ranks, rather than parachuted in from a public school fast-track), and at 37 was doubtless older than most of the men in in his Field Company. As my friend Ian suggested, perhaps this helped make Stout something of a father figure to many, for I think there’s definitely a warm combination of respect and fondness at play in this latter entry, quite a contrast to the consciously dry detachment evident in most of the others.
Combine what we know from this with the entry for 6th June 1944, and we can see precisely what Stout was doing on D-Day: assisting the tanks of 185th Infantry Brigade as they tried (unsuccessfully, as it turned out) to push inland to take the town of Caen by nightfall. But… what happened to 185th Inf Bde on D-Day?
7. Meet the unnamed NCO who spiked them all.
Having gone looking for a history of 185th Infantry Brigade, I found this web-page from a group of historical re-enactors who have a particular interest in that very brigade on that very day. Essentially, according to their accumulated history of events on D-Day, what happened is that around about 2pm, 185th Infantry Brigade’s advance was being held up by a group of Germans shooting at it from the cover of some woods. It needed help to make progress against these very well dug-in defences. The web-page continues:-
Eventually a Pole was captured who knew the way through the wire at the back of the battery. The gunners fled into the woods, harried for some hundreds of yards by the Company. The guns were then blown up by an unnamed N.C.O. of the Divisional R.E.s, who, though badly wounded, succeeded in “spiking” them all.
Who was that “unnamed NCO”? Unless anyone can demonstrate otherwise, I think it’s not being overly romantic to believe that this man was Lance Serjeant William Stout – he was an experienced NCO in the Royal Engineers, he was in No. 2 Platoon assisting 185th Infantry Brigade outside their temporary base in Hermanville, he was badly wounded, yet he did the right thing in obviously difficult and trying circumstances. If anyone can be said to be “Stout-hearted”, it was surely him.
8. What kind of cipher did he use?
According to diagrams in the documents we found in WO 193/211, Royal Engineers were only supposed to use low grade ciphers in the field. But which?
One likely candidate is the Double Transposition Cipher (introduced 5th Nov 1943 in document 32/Tels/943). Yet one problem with Double Transposition as a candidate here is that encipherers were required to finish up the message with a pure number group containing the number of letters in the cryptogram followed by a forward slash and the day of the month (i.e. “179/12”), which plainly wasn’t what was used here. For a number of reasons like this, Double Transposition was thought to be “difficult and slow to operate”… possible but not ideal for use in the field.
The other major possibility here is the low grade “Syllabic Cipher”, a system I unfortunately failed to find described in any of the National Archives documents I looked through (which, let’s face it, do tend to be more administrative than operational). However, I do know that this cipher used a book marked “BX 724”, while the Royal Engineers had their own specific version marked “BX 724/RE”: these presumably comprised tables of syllables, which were then offset / obfuscated using a key from a Daily Key Allocation List. Stu has already gone off looking for anything like this and/or any other information on the Army’s Syllabic Cipher, but please email me or leave a comment here if you know where in the archives to find more information on these!
9. After all that… what did Stout’s pigeon message say?
It’s 3pm in the afternoon of D-Day. Lance Serjeant William Stout has just destroyed the arms of a German battery to help clear the way for 185th Inf Brde to move on towards Caen. He’s wounded. He wants to get a message back to his field company, but radio traffic has to be kept to a minimum. My best guess? 185th Infantry Brigade have brought with them some pigeons and a bilingual (but English-born) translator who is also a pigeon fancier. Despite his pain, Stout writes down his message and he (or someone else) rapidly enciphers it using a Syllabic Cipher (or possibly a Double Transposition Cipher, though I somewhat doubt it). This contains 27 groups of five letters, i.e. up to 135 plaintext letters – roughly 25 to 30 words. So, perhaps we can guess it says something along the lines of… SPIKED ARMS OF GERMAN BATTERY OUTSIDE HERMANVILLE 185 INF BDE NOW MOVING FORWARD AM BADLY WOUNDED TELL WIFE AND CHILDREN ETC. He starts copying the enciphered message onto the form at 15:22 (British Time), finishes copying it at 15:25, passes it off to the bilingual pigeon fancier, who signs them off at 16:25 French Time, places them in red Army canisters, attaches them to a pair of commandeered pigeons and then releases them.
However, it’s worth remembering that of the 16,000+ pigeons released on the Continent by SOE, I believe that only around 1,250 returned safely. So perhaps unsurprisingly, it could well be that one of these two pigeons failed to make it back at all; while the other did make the 136 or so miles back to Bletchingley, which I suspect will turn out to be remarkably close to its home loft: flying at 45mph or so, the bird likely took about three hours. Yet as it briefly rested there on top of a roof at about 6.30pm at the end of what had been a thankfully clear day (or else D-Day would have been an unmitigated disaster!), could it be that someone in the house below lit an evening fire in a hearth, unknowingly sucking the poor pigeon to its death down in the chimney?
* * * * * * * * *
All of which makes a great dinner-table story, with the added bonus that a fair proportion of it is certainly true… but will this turn out to be the whole story? Perhaps we’ll find out before too long… fingers crossed that we do! 🙂
Incidentally, has anyone tried to trace Lance Serjeant William Stout’s son (also William Stout) or daughter (Urula Stout)? Perhaps they already know even more about this than we do… I for one hope they do!
Pingback: Pigeon Code - Fantastic Scholarship from Cipher Mysteries - Enigmatic Ape
Incredible stuff!
I’ve been writing a bit about the Pigeon Code at the linked address, and following your research with interest. Double Transposition is on my list of things to rule out before declaring (temporary!) defeat.
This trove of information should provide fertile ground for all sorts of cribs – as well as being thoroughly fascinating in its own right, so a huge thank you from me to yourself and Stu for your excellent scholarship.
I’m not at all familiar with syllabic ciphers but the thought occurs that we ought to see more repeated sequences ? I’d love to be wrong about that, so if you could give me any pointers to examples of such ciphers I would be eternally grateful.
Steve: well, the good news is that I managed to find an Army document describing exactly how Double Transposition works (with some worked examples) – I’ll post that here shortly, as it’s an interesting bit of cipher history in its own right. Oh, and thanks for the kind words on your blog, much appreciated! 🙂
Nick: I shall look forward to that!
Note that the encrypted message, and the Serjeant’s signature and “Time of origin” are in blue, blue that suggests to me that a hectograph or gelatine duplicator was used to make copies. They were small and cheap and required no power. Then, the copies were passed to a French soldier who added in pencil the “NURP” numbers and time of liberation, stating that two copies had been made and sent.
A p.s. to my previous message — this page shows that hectographs were indeed used for intelligence operations at the time because of their cheapness and the fact that they could easily be destroyed after use. The analine dye, and the telltale signs that some parts of some letters are oddly darker than others, suggest this strongly.
Russel, I think we covered that one: http://ciphermysteries.com/2012/11/24/dead-ww2-cipher-pigeon-timeline
also: http://www.kevinlaurence.net/essays/cc.php
I think a hectograph far more likely than carbon paper, assuming the color of the image is accurate. Carbon paper nearly always leaves traces and smears outside the letters; there are none here. A hectograph master required only a special pencil; the gelatine for the apparatus could be prepared and disposed of in the field, and it could make many more copies — dozens if needed — whereas carbons could only make a very few. Also you can see that the letters of UATA in line five have been gone over — this could be done with the same sort of pencil used for the master; no way this could have been done with carbon.
I am working it with analysis of Playfair cipher. Best Solving! Tom
Tom: good luck!
Hi Steve / Nick, I have added a bit of information regarding the “Linex” Cipher – this was to be a replacement to the existing low grade double-transposition cipher after 1943. I don’t believe that it was used for this message, however, it is an interesting and (I think previously unheard of, at least to me anyway!) ciphering method. It is very similar to a book cipher although the main difference is the how they generated the “Indicator Key” with a Playfair square. Anyway, here is a write up: Linex Low Grade Cipher
Some good stuff here. The inclusion of a French-speaking, but British-born secretary in the transmission of the message is the weak part, not least because it is not really suggested by the facts, but required by the conclusion.
I am reading your previous page on the note, and see better your reasoning on the Frenchman being involved.
Pingback: Pigeons - Page 8 - World War 2 Talk
Pingback: Pigeon Code Almost Certainly Not Broken - Enigmatic Ape
Great article. Excellent research & reasoning. Thanks for sharing it.
Kerry: thanks for dropping by! If you were me, what (presumably cipher-history-related) thing would you be looking for next? Thanks! 🙂
It is not in Playfair and it is not a double transposition.
Playfair is a digraphic cipher system in which two plain text letters are taken in combination to produce two cipher text letters. Because of the particular construct of a Playfair the plaintext letters in each pair can not be the same letter. When the plaintext has a letter in an odd position in the message followed by the same letter in the next position the letters are separated by the insertion of a “waste” letter (typically X). The resulting cipher text is also devoid of repeated letters in odd/even positions. The message contains odd/even repeats in the 10th group (YY) the 14th group (KK) and the 22nd group (GG). In addition, the message contains 135 letters and every Playfair must of its nature contain an even number of letters.
In double transposition the usual method is to write out the message so that each line contains the same number of letters except possibly the last line. The message is then transposed by re-writing the message by columns based on a transposition key that gives the order in which the columns are to be taken. The result is written out in the same format as the original and then transposed yet again. Note that no enciphering occurs. The message is concealed by re-arrangement of the letters. The letter frequencies of the language the message is written in are preserved and a simple frequency count of the message shows that it cannot be a transposition cipher. J appears 5 times, K 10 times, Q 6 times (but U only 4 times), while X and Z both appear 4 times.
If the June 6th, 1944 timeline is correct then this was very likely a tactical communication and more likely to employ a cipher as opposed to a code (which would require a code book which if captured would compromise messages by making them instantly readable by the enemy). In addition, codes that employ letter code groups usually have a pattern to the letters that is meant to ensure readability given potential errors in transmission or reception by radio. This message’s code groups do not exhibit those characteristics although this is not conclusive proof that the message is not in code.
My own suggestion is that the cipher method may have involved the US M-94 cipher device. Sending tactical messages while under fire is extremely difficult and I doubt that a code would have been employed as it would require a code book and quite a bit of effort to use effectively. The M-94 however, is an ideal tactical device. It consists of 25 alphabets, each inscribed on the rim of a half dollar sized disk. The disks are mounted on a spindle based on a pre-arranged key. The plain text message is encoded, 25 letters at a time, by lining up the appropriate letters running in a line across the disks and then taking off any other line of letters. The next 25 letters of the message are then enciphered in a similar manner (and not necessarily taking off the same line as before). Deciphering is effected by setting an M-94 with disks arranged to the same key to the first 25 letters of the message and then looking for plain text on one of the other lines of the cylinder, etc. In the event of imminent capture, all that needs to be done is to remove the disks from the spindle and scatter them, thereby destroying the key currently in use. The enemy is thus confronted with a cryptanalysis problem whose time of solution is longer than the usefulness of the information in the message.
On a different (but related) note I am in possession of the…
“B A B” TRENCH CODE No. 4. Serial number 924 issued to one Maj. Hathaway 2nd Bn.
This is a British code book from WW I and based on two of the pages never having been cut apart I believe that while it was issued to Major Hathaway it and quite possibly he never saw action. If anyone is interested in some historical research regarding Major Hathaway I would be delighted to provide further details about the code book.
Fred Brandes: 1. Thanks for ruling out Playfair and Double Substituion. 2. I am also inclined to think this is a mechanical cipher. But a M-94? Would a M-94 cipher have an indicator group? The message in question clearly has, but I have not found anything on the internet that the M-94 procedure would.
But would the British Army be using the US Army M-94 ?
Geoffs: That does not sound likely to me either. Since there are two indicator groups (at the beginning and end) I lean towards a M-209 or similar machine. The M-209 is a military version of a Swedish Hagelin machine. It exists in different versions and were used by not only american and swedish army.
Surely someone must know what kind of cipher machines were available at the time in the British armed forces? And more importantly: What were the encipherment procedures? Not all ciphers used indicator groups that look like that or…?
I am not convinced this is a low grade tabular type cipher, more of a medium grade one.
Micke: as far as I can tell, any one of about ten or more Allied cipher systems could reasonably have produced what we see. Which is not to be negative about the hunt at all, but rather to point out that none of us seems to yet be building on a particularly nuanced view of what was in use, particularly as evidenced by actual examples of WW2 ciphertexts. For example, the M-209 is possible but seems an unlikely travelling partner for a Serjeant: similarly for the M-94. And many alleged disproofs I’ve seen based on flatness of frequency counts fail to take into account things like the various substitutions that were used for numbers and punctuation signs. And we still haven’t seen a single example of a “low grade” syllabic cipher!
Have to agree, probably not M 209 either. Machines just not there, also regularly broken at BP!
Frequency counts on a short message can be misleading – I’ve seen enough raw Enigma messages to know that.
I think your are on the right path with the name Stout. Stot does not seem to occur as a name but Stott does and I can’t see the last T being dropped in a signature. However we only have access to WW2 casualties, there being no Medal Index like there was for WW1. There were around 40 W. Stout’s of military age according to the birth index for England and Wales. One of these could have served as a ‘Sjt’, but I do like the proposed solution!
Double transposition can probably be ruled out as the frequency count is a bit flat. in DC the letters and frequencies are preserved, as Fred has stated previously.
However SOE used a Delastelle cipher from around 1943, before settling on OTPs. This is a double transposition using a numbered grid key which does not preserve the original letters and can destroy the letter frequency distribution, most probably flattening it.
Also noted in Stu’s blog that Syllabic ciphers should not use an indicator, but we have probably AOAKN at the start and end.
REF:
Secret Warfare, Pierre Lorain, Orbis Pub. 1972.
Cryptanalysis, Helen Fouché Gaines. Dover Publications.
Elementary Cryptanalysis, Abraham Sinkov. Mathematical Association of America. 1998
I’m a little late to the party here but thought I’d contribute this little insight I stumbled upon.
Discussion about M-209 not being practical seems a little pre-mature, so I went looking to see whether it was actively used in June ’44.
Have a look at http://omahabeach.vierville.free.fr/258-PC1erBat116.html (in French) which details the find of an interesting mix of equipement from the Omaha invasion:
– M-209-A, serial No. 9982
– bound notebook containing the “Signal Operation Instruction” 1st U.S. Infantry Div. (with the handwritten words: “Lt. Hackett , Commo, 1st BN, 116Inf. “)
– Combined Assault Code 1, in 2 volumes: CCBP 0130-A1 (Reg. No. 6946) and CCDI 0130-A2 (Reg. No. 7005)
– 1 block “RT Slidex Code” (system of British origin?), With instructions, grid handwritten notes and U.S. ARMY HQ 116 CT of 22 May 1944 specifying the key for days D to D +6
So at the very least, on D-Day the American’s were using 3 coding schemes.
And while we could assume that a RE Serjeant would carry such equipment, is it unreasonable to assume he may have had access to it within the local group?
Is it a stretch to guess that the British forces would be using a common set? Certainly the Slidex and Common Assault Code were in common use by the British
Mike: from what the official documents seem to say, only a few ciphers were allowed for inter-force communication traffic – ciphers were an area where it was (rightly) believed that each armed force should have its own ciphers, so that a crypto disaster in one battle arena should not have catastrophic failure in another.
So I’d be somewhat surprised to find that a British Army Serjeant was using an American cipher machine: because there wouldn’t be a corresponding deciphering department in Wing House or elsewhere with a set of M-209-A machines and matching daily keys etc.
Slidex, however, is a “medium grade” (but actually very weak) Army cipher systems I’ve been looking for a good reference for. The Germans broke it easily and regularly, so we should stand a pretty good chance too.
http://chris-intel-corner.blogspot.co.uk/2012/07/the-slidex-code.html
However, Germans cryptanalysts had the advantage over us in that they had a large volume of traffic to work with, rather than just a single example. Moreover, “Each card was valid for a specific period of time, from several days to a month. The coordinates strips changed daily.”
Interestingly, this page from Jerry Proc claims that “SLIDEX was apparently adopted for use by all the units involved in Operation OVERLORD. All cards issued prior to 1970 are no longer classified.” If that’s right (and Jerry is normally extremely solid), then all we’d be looking for is the cards and coordinate strips used on D-Day! Simples! 🙂 http://www.jproc.ca/crypto/slidex.html
Surely we are looking for an American seconded. The Diary refers to Sgt as the British commander would, but the soldier referred to himself as sjt (an Americanism). The lack of Stott in the Kew database? does not therefore permit the researcher’s corruption to Stout.
The name seems quite clearly Stet or Stit if the pen is followed in the mind as it writes. To look for the nearest name that can be found in lists is flawed. That person may not be in a list.
British Sgt American Sjt.
Donald: please feel free to look for American Serjeants called Stet or Stit, let me know how you get on! 🙂
I think the header and footer must identify the cipher sheet in some way, and the deciphered message must then reference a code book or Slidex sheet. 25 groups = 125 characters including dummies to conclude the 5 letter groups (all or some of GQIRU will be a reference to a dummy character or space).
Andy: other British Army ciphers I’ve seen described employed various ways to obfuscate a key reference to the cipher settings used in the message, so it seems likely to me that Slidex would incorporate something like this too. However, the data encoded in a single five-letter indicator group wouldn’t be enough to define all the settings, so this is only going to be a small step towards breaking the cipher. Still, a small step would be better than no step at all! 🙂
The original text is written on a wireless operator’s message pad and then he’s annotated it with the pigeon numbers (presumably). This suggests to me that it was intended to send by wireless.
In the Slidex archives on the web I’ve seen a PRO document on Dday that confirmed Slidex was used and it would be reasonable to look at the coordinates for the day.
According to my copy of the manual of the M209 “The indicators will always be [placed in the order shown above ] and will be inserted before the first group of cipher text. They also appear (in the same order) as the last two groups of the message, following the last group of cipher text “(page 20/21).
I’m not arguing it was a M209 in this procedure (because it had two groups heading and footing) but the repitition of the header and footer suggests to me a similar system.
Andy: given that the top right of the sheet says “PIGEON SERVICE”, are you suggesting that this was copied over from a wireless operator’s message pad? There is also an open question about the sheet, given that it seems to be different to all the other pigeon message pads in archives and museums I’ve seen.
I’ve only seen it on video and it looks like a standard signals pad to me.
Andy: here’s a link to a reasonably-sized (865 x 1536) scan of the message, very slightly enhanced from the image posted by GCHQ (as I recall):-
http://ciphermysteries.com/wp-content/uploads/sites/6/2012/11/Pigeon-message-large-enhanced.jpg
Thanks. I was looking just at the lower boxes – but the document is obviously a pigeon service.
http://www.royalsignals.org.uk/misc.html has links to the army signals documents – you have to register.
Nice site! i’d like to comment on the following:
‘Since there are two indicator groups (at the beginning and end) I lean towards a M-209 or similar machine.’
One problem. The M-209 used 10 letter indicator. In the message only the first 5 letters are repeated at the end.
Chris: that’s correct. However, it’s also true that we currently have only sketchy information about how most Allied crypto systems (whether machine or hand) were actually used in the field, in particular how indicator groups were encoded. Unless you know a definitive book or article on the subject…?
Perhaps more importantly, may I also ask if you happen to know if there are any records of Slidex-coded wireless intercepts from D-Day in the German archives? I think we’re looking at a Slidex-enciphered message, but we may well have insufficient depth in a single message to break it. I’m always impressed by the references you dig up on your website! 🙂
‘Chris: that’s correct. However, it’s also true that we currently have only sketchy information about how most Allied crypto systems (whether machine or hand) were actually used in the field, in particular how indicator groups were encoded. Unless you know a definitive book or article on the subject…?’
Well for the M-209 I used the manual, a US army instructional video showing how to set it up and use it plus German reports that explain the indicator system to their cryptanalysts. I think the indicator system was indeed 10 letters. Check: http://chris-intel-corner.blogspot.gr/2012/06/american-m-209-cipher-machine.html
‘Perhaps more importantly, may I also ask if you happen to know if there are any records of Slidex-coded wireless intercepts from D-Day in the German archives? I think we’re looking at a Slidex-enciphered message, but we may well have insufficient depth in a single message to break it’
The reports I’ve used are mainly TICOM interrogations. Most of the German crypto archives were destroyed at the end of the war (especially those of the army and airforce sigint agencies) so I doubt whether you can find entire decoded messages.
‘Perhaps more importantly, may I also ask if you happen to know if there are any records of Slidex-coded wireless intercepts from D-Day in the German archives?’
The reports I’ve used are mainly TICOM interrogations. Most of the German crypto archives were destroyed at the end of the war (especially those of the army and airforce sigint agencies) so I doubt whether you can find entire decoded messages.
Yes the M209 did use two 5 letter indicator groups according to the standard procedure in the manual. I don’t think it was necessarily a M-209 because of that. But for all I know there was another procedure. Enigma changed its procedure for example. A second indicator group could have been enciphered.
If the signal came form a generic signals unit then you might well expect a double structure.
Otherwise was it really one man and two pigeons?
My suggestion is that it is a standard or machine cipher of an underlying system, not of underlying plain text. For this underlying system I think slidex is the best candidate.
My apologies for coming so late to the party, but I wasted a lot of time at other (website) parties before discovering this one.
Firstly thanks Nick for the website and all the useful information you have gathered, as well as your subscribers. The first website I found where uneducated guesses are extremely rare.
Some observations by a novice and aged cypher geek:
1. My first thought on seeing an image of Percy’s note was that ‘lib. 1625’ was a catalog reference to ‘Book #1625’ or ‘Page 1625’ after the Latin, which would have indicated the location of the key code, but apparently ‘Liberated xxxx’ is/was common pigeon terminology for the time when the bird was let loose.
2. My expert cold-war Morse interpreter Brother-in-law informs that 5-character blocks is a standard radio procedure, the number of columns is merely a convenience to fit the size of paper the R.O is using. He also seems to think the ’27’ is a standard parity check for the number of blocks, although it more commonly should have indicated the number of characters. Unless, of course, Percy’s message is only 27 chracters. Uh Oh!
3. Before educating myself (mostly from this website) I had the idea that the message was encoded using a simple Caesar-shift, but after trying over three hundred words all I get is gibberish, with the exception that ‘AOAKN’ translates to ‘JACKC’ using the key ‘ROYAL’ regardless of the number of shifts. So I think ‘JACKC’ was the (radio) sender’s (usual) call sign which, in my naive opinion, would have informed the receiver who the message was from AND how to decode it. The only other curiosity is that ‘FNFJW’ translates to ‘NURPD’ (NURP Dispatcher?) using the key ‘STOUT’ but I think that’s just an odd coincidence.
4. If I’m right about ‘JACKC’ then the rest of the message was encoded with a different key or system, since after nearly a quarter of a million tries I haven’t found a single English or French word of five or more characters.
5. So, I’d like to continue to attempt to crack this code, but I need help. I’m a PC programmer with spare hardware, so if anybody wants to suggest a method or point me to an example they think might be relevant I’ll happily try to write the relevant software and let the PC run until we recognize something. Unless Jack used a standard book, which means we’re all wasting our time until we find a copy.
Has this cipher been definitively cracked? Would it be useful to state one of my decode efforts resulted in the last three words being “Bearn En Fin”?
(Oops, I meant to spell it Bern. Anyway, here are three of my attempts at translating:
Attempt 1: As of A DE (German) Air RD (raid) by ALPS OP AC End is AID BRIT CUT. MUST TRAVEL IN A FORD (aircraft). (We are) ID (ing) THE FALLEN IN BUNACH (Germany) (and) LV (leaving) NOW. MAKE A CINDER ORDER FRAGO TO CEASE FIRE NW AND FIRM (from location). (signed in) BEARN, EN FIN.
ATTEMPT 2: AS OF A DE AIR RD BY BLOM OP ACE AGENT, AID BRIT CUT. PS TWO RAAF IN EAST AID THE FALLIN IN BAUNACH. LV NOW. MAKE A CINDER ORDER FRAGO CEASE NW AND FIRM. BEARN EN FIN.
ATTEMPT 3: AS OF ACE AIR RD A BY OPS MN (main) VIA AGENT AID BRIT CUT, MUST TRAVEL IN A FORD (or the other option) ID THE FALLEN IN BAUNACH LV NOW. NAMED BAND (friendly unit) IS NW AT FIRE CO CEASE NW AND FIRM. BEARN EN FIN
Mindy: it hasn’t yet been cracked, but given that it was written on a British Army pigeon pad, placed inside a British Army red pigeon capsule, and has a Royal Signals Corps X02 abbreviation at the top… I’d say that unless your cipher has something specific to do with British Army practices during WW2, you’re probably out of luck with this one. Just so you know.
Mindy: we also know (from the “X02”) that, despite its brevity, the pigeon message contains at least two codresses (coded addresses). Hence any claimed decrypt that doesn’t include at least two codresses is going to be just plain wrong.
Nick: I was actually assuming it was British…or possibly (less likely) Australian RAAF. I feel however, that I have accidentally offended you, and for that I’m sorry. I guess the whole “ford” thing probably makes it seem I thought it was American, but actually, there were some aircraft made by Ford in WWII in use during the war by the British, for example the XC-B24. However, they were not considered very good aircraft. I thought it was a jibe at american aircraft of the time, and therefore historically kind of funny. However, I guess another thing that might not have happened, is a British person probably would not havIe said “brit (british) aid cut”…but I guess an Australian or other person could have. I did not assume a british person would not have said that, when I did that.
Anyway, I acknowledge these are just guesses. I used a particular method for them. Basically, I assigned numbers from 1-9 to them, based on the assumption the letters could have been associated with numbers on a number cypher page. then i associated letters with these numbers. Hard to explain, because it doesn’t work quite like that, but it works quite well in excel. Anyway, those were just the guesses I got using that method. I was working on a separate method the other day, which did not involve numbers, but I didn’t get too far, because my primary focus is of course the voynich manuscript.
Mindy,
May I ask how long you’ve been working on ms Beinecke 408? Also – have you written anything somewhere else about it? I mean do you have a blog, or web-page?
You know, about 20 years ago, before it was acceptable in academe, I did some research into imagery on playing cards. The best – almost the only scholarly treatment of the subject – was in a book by the late Michael Dummett, who entitled his book ‘The Game of Tarot’. I couldn’t get a copy through the University bookshop or library, both of whom began to wonder if I was quite ‘sound’ for daring to mention such a word in their hearing!
For the same reason, I then wrote under a friend’s kindly-lent name, ‘dana mindon’, of which I’m always reminded when I see yours. But times have changed.
Mindy: my point is that we have specific, solid evidence that this is a precise, operational British Army cipher used in the field in WW2.
So if your starting point for this cipher is the idea that some kind of guesswork or interpolation is involved, I really don’t think you’re going to get anywhere with it.
Diane,
Of course you may ask, and hopefully you will not automatically assume my answer diminishes my position.
Although I am well aware most people working on the VMS have been working on it for several years, I only first came across it in May of 2013. I decided it was “awesome”, and I wanted to crack it. So, I did what I do, and I went at it from the assumption it was a real language whose alphabet had been lost. I also assumed it was an older manuscript, and must be a copy. My third assumption was regarding what I believed the content might suggest about who the culture was. Then, my only problem became figuring out the letters. So, I went through several ancient alphabets, compared and contrasted them, and began trying a variety of letters out.
I had two breakthroughs almost simultaneously about a week and a half in. One was that I had discovered actual names of people in the culture I suspected. The other was that I was able to decode folio 1R.
The thing is, although I know this could be some sort of a book deal, I don’t really care about that too much right now. I think the history itself is far more interesting. However, the history happens to be very politically sensitive in a few cases. For example, legends appear to be history. On the first folio, the data could to this day change the way we think about a particular culture, which is why I want to get in touch with the entity beforehand, and ask them what they would like to do about the information…give them some time to prepare people for the data.
I know I’m a newcomer, but I also know the code can be used on every page I have tried thus far (except the very last page, which I’ve already provided a while back, and which is not in voynichese), and the only problem becomes figuring out whether or not I’ve translated it correctly. (Although the code is pretty straight forward, this is actually somewhat difficult because of particular nuances of the language, such as words that have multiple meanings. An English example would be coming across a word such as RUN, and trying to figure out which definition was in use).
As far as a website, no I don’t have one. I posted some stuff on my facebook initially, but eventually took it off, in case someone might attempt to use my ideas as their own before I could release them. I have shown the data to several people over here, including people who are familiar with the manuscript and have attempted to decode it at some point, all of whom have encouraged me to submit the answer. I have also given the code to a good friend of mine, who is now working on the naked lady folios using my code. I have also sent a fully translated copy of the OTOLAL page (folio 70R) to Mr. Nick Pelling, with method, but I told him when I emailed him that I was invoking copyright, and he was not allowed to do anything with the information without my express permission. (Not sure what he thinks about it, but I only sent it to him this past weekend).
Also, because I know the culture of origin, I’ve been able to do things like find answers to questions such as why scorpio is a “croc like monster” (I think it was Thomas Spande who posed that question). While I don’t know WHY a lizard was used, I do now know the culture has other examples of using this type of animal for scorpio. I would not be able to do this, I don’t think, if I didn’t know where to look in the first place, because even in this one culture there are hundreds of manuscripts to look through. In fact, I was not sure which manuscript to look in, so I approached the search systematically, and began flipping through online manuscripts in order, until I found (quite quickly, considering) a manuscript which had the data I was looking for.
The one thing that I find probably most difficult is that I am able to get answers, but I’m also attempting to protect the data at least until I can get some sort of message to the culture most likely to be adversely affected by the contents of the book. Therefore, I hint at the answers, instead of giving them outright, unless I think giving the answer will not give away the source.
Oh, and as for the plants…the pages which have large plants on them, thus far, have not named the plants on the pages. However, the pages with roots have named the roots. There are a couple pages with the large plants I thought would name the plants (writing near the plants), but at least the one I translated does NOT give a plant name. Rather it says “from the border”. I agree the plants in the front have depictions of their roots in the back. This is fortunate, because it can help in decoding which plants are which. However, I also note several of the plants have hints right on them (eg the dragon’s blood tree has a dragon on it).
Well, I’ve spent a lot of time I usually reserve for decoding, on writing this answer. (I’m a full time working wife and mom, so I usually use the time after the kids are in bed/before they are awake, or my lunch hour to do my decoding). Hopefully this answer has provided you with some insight into me.
🙂
Dear all, Just a minor point on copyright. It hasn’t been necessary (in the USA anyway) to register a copyright anywhere. All that is necessary is proof that “you wrote it” and when. Seems to me that a copyright is automatically insured by just entering an idea on a web site (automatic date and time stamp) and maybe making a back up copy on a thumb drive and by that you have secured a copyright. Providing someone else visits the site, you will have at least one witness (important also). Nick’s sites are backed up by email addresses as well. So no one can contribute without a valid email address. Where things really count, as with a patent, a signed page of a lab note book where some discovery is described and the signature of a witness is good enough. As Diane once commented, many ideas are proposed on Nick’s pages or other web sites and after time passes, tend to become sort of common knowledge or get forgotten altogether. Compiling a book, as Nick has done (and for whom another is evidently under consideration), is a way of establishing who proposed a particular idea but a searchable data base would be even more useful. I would not bother anyone, most particularly Nick, with any copyright concerns. Spontaneously popping off in public about a VM decrypt is another matter.
Dear all, On the “Scorpio” rondel. The resolution of my computer monitor and printer is not great but the “croc-like” monster of Scorpio evidently has an infant in its jaws! If it is a “lizard” as Mindy opines then it almost has to be a Kimodo dragon found only in a few islands of the Indonesian archipelago. I am thinking it is closer to Asia minor, and may refer to the “nhangs” or crocs of Gobekli in southern Turkey, formerly in the ancient kingdom of Armenia. There is some imagery close to that of the zodiac figure and many think Gobekli was an ancient observatory that is several thousand years older than Stonehenge. Many of the petroglyph animals are used in the Mesopotamian zodiac such as the bull and crab but some like the boar and croc are not common. Many think these have nothing to do with the zodiac and just represent animals of the hunt but certain stone alignments support Gobekli being an ancient, maybe the world’s most ancient, astronomical observatory.
Thomas Spande,
Thanks for the info on the copyright. I don’t know a lot about copyright yet, aside of what I’ve researched over the past few days on the internet. I read it hasn’t been necessary since 1989 (for intellectual property rights), but is freqently advised anyway.
Oh, and I don’t necessarily think it’s a lizard, it just looks somewhat lizard like in that picture. I like your idea of it being a dragon. The ones in the manuscript I looked up looked more recent (even though still middle ages, maybe a couple or possibly a few hundred years newer?), and more like a dragon of some sort…close enough for me though, and definitely not scorpions.
Oh, and I’m not protecting the data from Nick Pelling (If I was I wouldn’t have sent him the decoded portion with decode key)…I’m trying to figure out how to keep portions of it out of the public until I’ve had time to contact the culture who might really want to know.
Thomas that ‘infant’ could be meant for a man. Medieval imagery is sometimes flexible about scale.
If a man, see e.g. Crocodile
http://bestiary.ca/beasts/beast146.htm
Actually it might still be meant for a crocodile if the infant were an error for reports of how the crocodile may carry its young, though this habit was not recorded by Pliny or the medieval bestiaries.
However:
http://pixdaus.com/a-nile-crocodile-gently-carries-its-newly-hatched-young-in-i/items/view/306705/
So do American alligators, apparently:
http://www.agefotostock.com/en/Stock-Images/Rights-Managed/DAE-15012361
As a general comparison, certain fish do too – e.g. the lungfish.
http://archive.org/stream/naturalhistory70newy#page/n231/mode/2up/search/gill
Cheers
Mindy,
I think it is sensible to keep the matter fairly close. It’s less an issue of formal copyright infingements ~ more an insidious habit of deliberately ignoring the source from which a given result, or plant identification was gained, the result or id being taken up and then the original researcher ignored either by (a) announcing the matter as if it were a commonly known ‘fact’ and by implication expecting to receive the credit for discovery or (b) taking the result or id and re-presenting it ‘with a twist’ with the specific aim of ignoring the original research. Neither practice is savory but the second is particularly low, in my opinion.
It is not uncommon, in the second case to present the stolen results ‘with a twist’. For example if someone identified a plant as a Japanese pumpkin, in a short time the same folio might be announced as a pumpkin, but not the Japanese variety.
Or a ‘Japanese pumpkin’ might be announced, but randomly ssigned to a different folio.
Dishonesty on that scale more than infringement of legal copyright is what you’ll need to be wary of. It has become endemic in the internet age, and we can hardly be surprised if some amateurs carry such ghastly habits over to this subject. I’ve seen whole blogs whose content is nothing but a re-worked version of some other person’s… rather boring to read imo.
Thanks Diane. 🙂
There were over a thousand pigeon breeders that gave up their birds for the war effort. How many young motorcycle riders were used to retrieve the messages and carry the messages to a top secret location? I think this bird should be acknowledged for keeping the official secrets act for so long. What was it’s name ? It has a band on the other leg.
Was looking at your site and have to differ with you. This code seems to be about two defectors, both in September of 1945. One was killed in Turkey and the other was taken into Camp X in Canada. He survived and it became known as the Corby case.
Louise: how can you tell any of that at all?
Hi Nick, Hi Stu,
I know I am (also) a bit late to the game, although I have been following this from the beginning, but only now do I have the some spare time to focus on this and have some fun – I started my company to solve difficult problems, so this definitely counts.
I read this post where you mention worked examples of Double Transposition (in the Army):-
“………….managed to find an Army document describing exactly how Double Transposition works (with some worked examples) – I’ll post that here shortly….”
and I was wondering if you managed to post them, as I couldn’t find them.
I mention this because I am not yet prepared to rule out Double Transposition, as I believe the reasons given as to why it probably isn’t Double Transposition, may be problematic. My reasoning this thus:-
1. Letter Frequency Count should be close to that of English.
There are a number of very good reasons why this need not be so. These are:-
(1a) The plaintext may contain a moderate-to-large amount of common or army abbreviations and acronyms, significantly skewing the frequency count analysis.
(1b) If we are correct that the plaintext contained one or more forwarding addresses, then we could expect further army a abbreviations.
(1c) I read in a book (purchased in 2013) that it was common practice in the army to pad Double Transposition plaintext at the beginning of the message with Q, K or Z in the belief that it make it harder to decrypt (although if that were known to the Germans, it may actually have made it easier, as you then have a known plaintext at a known position), however, the point being that this too would have an impact on the Frequency Count.
(1d) Having lived in Northern France and West Flanders, there are many village and town names that could very easily skew the letter frequency count (such as Beveren-IJzer, Kapelhoek, Zouafques and Rexpoede ), especially as you might include two or three of these place names to provide a precise triangulation of your location. Being quite a short messages would mean that there would be greater impact on the letter frequency count.
2. Army Double Transposition indicator groups were four letters, not five.
(2a) I note from an account by Leo Marks that SOE operatives used Double Transposition with five-letter indicator groups, thus a model does exist that would work for our AOAKN indicator.
3. Required to finish the message with a pure number group, but missing.
(3a) This was missing from our pigeon message, but given that we know there were variances in the operating procedures for enciphering a message, between the difference army sections and regiments, it is not inconceivable that the Royal Engineers were using Double Transposition in a different manner to the standard procedure (including a five letter indicator group).
My general feeling is that we should not yet rule out Double Transposition, for which I hope to have a crack at it using likely plaintexts, dragging them across the message space, although if the first part of the message really is a forwarding address, then we do have quite a bit to go on.
Anything you can tell me about how the army used Double Transposition, especially some examples, would be a wonderful starting point. I am currently looking at how Hemming and Leveinsheim distances can be used in the analysis of partially recovered words.
Many thanks,
Adrian Neal
Oxford Scientifica Ltd.
Harwell.