Abstract
The JN-25 series of IJN operational codes was the most important source of Allied Sigint throughout the Pacific War. From early 1942, intelligence supplied by this source was significant in determining the operational decisions made by the Allied commanders. This intelligence was supported by some derived from other IJN systems, but only after mid-1943 was significant intelligence obtained from reading IJA intercepts. This chapter explains how the weaknesses in the structure and use of the early versions of JN-25 were successfully exploited by Allied codebreakers.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
NARA RG 38. CNSG Library. Boxes 16 and 18. File 3222/65. In general the USN WW2 cryptological archives are in Record Group 38 in NARA. The reports on JN-25 form a high proportion of the technical material there.
- 2.
The report by Sir Edward Travis is TNA HW 14/142. A report there dated 29 September 1944 states that this current problem with JN-25 ‘is now having the attention of Brigadier Tiltman and the Research Section. As you will remember, Brigadier Tiltman broke the cipher in 1939 and we are hoping this re-inforcement of our strength will do the trick’. In fact, JN-25N62 was never broken.
A quick examination of the list of WW2 signals intelligence material in the American National Archives confirms that JN-25 was the principal series of ciphers of interest.
- 3.
NAA Canberra item A10909 2 may be read via Recordsearch and the web. The spelling reveals its American origins. It would not have been compiled by Frumel alone. Some of its contents would have come from using captured code books, such as the JN-25C material captured on Guadalcanal in August 1942, to decrypt old JN-25 messages.
Unfortunately no work has been done to compare A10909 2 with Japanese archival records of the IJN in 1944.
- 4.
TNA HW8/102, page 1.
- 5.
The three were Richard Pritchard, Malcolm Burnett and Neil Barham.
Apparently Pritchard was an Army officer who had been with GCCS for about ten years. His speciality was German Army material. He was commandant of a short-lived GCCS unit based in France before the German invasion.
Malcolm Burnett was a naval officer who also had substantial experience with GCCS. Some information on his career, including a wedding photograph, is given by Ian Pfennigwerth in his book A Man of Intelligence. Burnett was later to be instrumental in getting FECB and then Op-20-G up to speed in working on JN-25.
Neil Barham was a naval officer working for GCCS and involved with IJN ciphers at various stages.
Curiously the commendation that GCCS Director Alistair Denniston wrote about Tiltman (29 April 1940, now in TNA HW 14/4) does not mention his work on JN-25.
- 6.
Nave, by then developing serious health problems, did some work on JN-25A at FECB. Autobiographical notes written by Nave, which are now preserved in the Australian War Memorial, indicate that he believed that JN-25A was insecure due to the patterns in the allocation of book groups in the code book. This may refer to the ‘multiples of three’ patterns, which he mentions elsewhere, or to lesser patterns or (quite likely) both.
Deidre Macpherson’s book Betty Archdale (2002) describes how H. E. Archdale, a former captain of the English women’s cricket team and later headmistress of Abbotsleigh school in Sydney, led a group of young women out to Singapore to do interception and clerical work necessary for breaking into JN-25A and JN-25B. The key letter of 5 March 1941 from the Asian Fleet to Op-20-G and quoted in Sect. 5.5 and also in this chapter mentions them: ‘British employ three officers and 20 clerks on this system alone.’
The FECB still worked on diplomatic systems in 1941. An interesting report by Professor Room survives in the British Archives.
- 7.
The function, if any, of this digit was not discovered by either the GCCS or Op-20-G at the time. However it appears not to have been a null, that is a randomly chosen digit inserted as padding and hopefully also to confuse the other side.
- 8.
Here one more complication is overlooked. Messages would usually be cut in two and transmitted with the second piece coming before the first piece. So the message in the text might well end up as: NEVER EXISTED IN JAPAN HE IS CERTAINLY RECKONED TO BE A TRUE PHILANTHROPIST START A MIKADO MORE HUMANE THAN HIROHITO.
- 9.
Peter Donovan’s paper The Indicators of Japanese Ciphers 2468, 7890 and JN-25A2 was published in the journal Cryptologia, July 2006. It does not mention how the reading of JN-25A1 and JN-25A2 indicators made possible the detection of the practice of tailing, which became most useful in breaking later indicator systems.
- 10.
In view of what would be at stake, it would have been appropriate to create a really difficult indicator system. For example, it would have been feasible to pad out the indicator 3567 by a null (always 2 in the examples) in varying positions, such as on the left on the first 6 days of the month making 23567, in the second place on the next 6 days of the month marking 32567, in the middle on the next 6 days of the month making 35267, etc. And the encryption could be achieved by adding to the padded indicator group the group for the day and the third encrypted group of the message itself. It is unlikely that the Tiltman team would have broken such a system using only a couple of thousand intercepts.
- 11.
Likewise a number leaves a remainder of one (two) on division by three if and only if the sum of its digits leaves a remainder of one (two) on division by three. These are useful in understanding the process of stripping at least the earlier versions of the code JN-11 and the process of stripping the indicators of JN-25A1 and JN-25A2.
- 12.
Op-20-G appears to have interpreted the page numbers as 001, 002, …, 299, 300 and to have brought in the rule ‘add 1 and divide by 3’ to get page numbers from the indicator. Of course neither GCCS nor Op-20-G could see a copy of the original additive table. There is no surviving evidence that the GCCS numbering did give Tiltman the hunch mentioned by Whelan 12 lines further on in the main text, but it seems likely that this was the case.
- 13.
The Whelan account of tabulators at Bletchley Park is in TNA HW 25/22.
One can only speculate about the ‘special procedures’ that were carried out for Tiltman.
- 14.
This was reprinted in Cryptologia in 2000 and is available in facsimile on the Turing web site. The report was mostly on the machinery being made at the NCR plant in Dayton, Ohio, to handle German submarine Enigma codes.
Mavis Batey on page 96 of Dilly draws attention to a meeting held at Bletchley Park on 1 November 1939 involving Knox, Twinn, Jeffreys, Turing and Welchman. Equipment needed from BTM for work on Enigma was discussed. Travis then fixed up the purchase. This may have led to the same people designing the first JN-25 decryption device.
Carrying out large numbers of arithmetical operations in one’s head is tiring and prone to error. In practice some machine would be essential for the processes described in this chapter and Chap. 10.
The British GCHQ, successor to the GCCS, cannot find documentation of the Letchworth device. It is difficult to understand why Turing, Welchman and others did not think up this use of colours while they were at it.
Cryptanalysis of JN-25 (July 1943) includes on page 231 a sketch of the Dayton version of the machine and states that various others had been tried earlier. This sketch closely resembles the device known as the ‘fruit machine’ and preserved in the Wenger Display at the Naval Station in Pensacola, Florida. Instructions for its use are given on pages 230–233 and 501–502 of the above-mentioned document.
- 15.
A 1945 reference to the effect that at least a sprinkling of the JN-25B7 and JN-25B8 additives had been evaluated incorrectly by Op-20-G is NARA RG457 Entry A1 9032, Box 578, file 1391.
This file refers to a proposed exercise of reworking the decryptions of JN-25B7 and JN-25B8 intercepts using the improved method discussed in Appendices 2 and 3 of Chap. 10. Presumably the motivation was to obtain more complete and accurate decryptions and decodings of more intercepts and so determine whether the Pearl Harbor raid could have been anticipated. Not much came of this. In any case Chap. 7 showed that more than enough information about the risk of such an attack in December 1941 was available at that time.
- 16.
The Ely report is to be found in NARA RG38, CNSG Box 7, item 2500/7 and also RG457 Box 600, item 1551. This refers to an earlier report by Lt. (jg) Tompkins dated 14 November 1942. The GCCS file on JN-25 (TNA HW 8/102) contains a terse summary dated 5 September 1942 of a method developed by Thomas Room and Richard Lyons in Melbourne.
Ely’s use of the word ‘additive’ may cause some confusion here. The matter has little intrinsic importance but has been explained in Note 19 of Chap. 8.
More sophisticated methods of handling Phase 1 would consist of taking various columns of depths 14 or more, selecting those with relatively few decrypting groups and making optimal use of the information they provided together.
- 17.
The Barham Japanese Ciphers—Notes (in TNA ADM 223/496) state that the Tiltman team, of which he was a member, was able to obtain intercepts in depths of up to 20. This would not have happened if the starting places used for JN-25A1 messages had been chosen by lot!
- 18.
When JN-25A5 was replaced by JN-25B5 the FECB cryptologists had already worked out some of additive table 5. They could thus effortlessly decrypt some of the JN-25B5 messages and thus get initial statistics on the most commonly used book groups in code book B. When additive table 5 was replaced by additive table 6 the British and American teams could use these statistics. Thus Phase 1 was by-passed for JN-25B.
- 19.
The GYP-1 Bible states on page 48 and elsewhere that collateral information in what is in the messages being used to recover additives may influence the decision about which common groups should be tried. Of necessity parts of this book oversimplify some of the issues. Thus in practice garbled GATs were quite common and so the decrypter may well have been happy to accept eight groups out of nine being divisible by three as good enough.
- 20.
The item is in NARA RG38, CNSG Library Box 115, 5750/119. Although it was written soon after the war ended, the author could not guarantee completeness of the account of the use of captured documents by Op-20-G. There is some considerable difference between documents captured in conflict and those obtained by sleight of hand or theft earlier but the material on the ‘S’ code was of considerable importance.
It is not clear whether Tiltman had memories of this same code as that called ‘S’ by Op-20-G.
Stephen Budiansky’s paper in the April 2000 issue of Cryptologia refers to archival material to the effect that Op-20-G was decoding some JN-25A messages in October 1940 and expected to be reading it more or less in full by April 1941. This hope died when JN-25A was replaced by JN-25B on 1 December 1940. Budiansky has written another extremely relevant paper on the early American work on JN-25. This is Too Late for Pearl Harbor, published in the Proceedings of the US Naval Institute, December 1999.
Jeff Bray interviewed the veteran Arthur Pelletier in 1994 and obtained information supporting this account of Agnes Driscoll’s role. This is published as a footnote on page xx of Bray’s version of SRH-009, Ultra in the Atlantic, Aegean Park Press.
Stephen Budiansky in his paper in Cryptologia (April 2000) draws attention to the History of GYP-1 in RG38 of NARA. The reference is CNSG 5750/202, pages 14–18. The American break into JN-25A was indeed as late as September 1940.
- 21.
In Sect. 14.2 some text is quoted from Part G of the CBTR about how the corresponding discovery was made for the Water Transport Code. The process appears to be quite similar to the Driscoll ‘break’ on JN-25A.
- 22.
NARA RG38, CNSG Box 115, 5750/119, page 12, quotes this from page 100 of the Op-20-GYP history.
The rules involved in some Japanese naval and army ciphers could be quite complicated. Thus the Jamieson report (B5554 in the NAA) mentioned in Note 46 below gives some minor IJN systems which had seven different variations, one for each day of the week.
- 23.
This is taken from the Barham report previously cited in Note 17. It was found courtesy of the book Action this Day by Michael Smith and Ralph Erskine (Bantam Press, London, 2001).
- 24.
The paper by Peter Donovan in Cryptologia, October 2004, gives some reasoning why, with JN-25 having book groups which were known to be multiples of 3, extra tricks were available for the cryptologists when either
-
the additive table was changed but not the code book, or:
-
the code book was changed but not the additive table.
Changes of both types happened with JN-25B.
In fact this chapter, Chaps. 10 and 15 reveal that the use of only multiples of three as JN-25 book groups was appreciably more pernicious than as suggested in the 2004 paper.
Once again attention is drawn to the Swedish experience with a Russian additive cipher in 1940. Bengt Beckman, on page 54 of Codebreakers, quotes a contemporary report to the effect that the Russian failure to change both code book and additive together saved the Swedish unit ‘several months’.
‘But for this mistake on the part of the Japanese the form of the book may have taken a couple of months to discover.’ (This is from the Barham quotation given earlier, see Note 23 above.) Would JN-25 have been broken in time for the Coral Sea and Midway battles if there had not been an overlap in additive tables for JN-25A and JN-25B?
-
- 25.
The Fabian interview is one of about 35 oral history records available from the National Cryptological Museum.
The NCM oral history is somewhat different from the version printed in the booklet The Quiet Heroes of the South-West Pacific Theater: An Oral History of the Men and Women of CBB and Frumel edited by Sharon Maneki. In particular, Fabian devoted some considerable time to explaining how having the book group digits summing to a multiple of three was exploited. This had been crudely blacked out in the original version available from the NCM. An uncensored copy has now been released by the NSA. The paper by Peter Donovan (Cryptologia 28(4), 2004, pages 325–340) appears to be the first reconstruction of this key aspect.
The NAA Melbourne file MP1185/8 1937/2/415 shows that Nave in March 1941 confirmed that the ‘5-figure code’ (JN-25B) and two other diplomatic codes were the only Japanese codes intercepted in reasonably large quantity at FECB.
- 26.
The timely transfer of key technical information is rather analogous to that of 27 July 1939, when a GCCS delegation, including Dilly Knox, was in Poland meeting the Polish cryptanalysts. Marian Rejewski was asked Quel est le QWERTZU?, that is in which order the 26 letters were channelled into the military Enigma machine. (The standard German typewriter has QWERTZU on the left of the top row of letters whereas English keyboards have QWERTY there.) After being told he responded Nous avons le QWERTZU—nous marchons ensemble!, that is ‘we have the QWERTZU—we will walk together’. In Singapore Op-20-G acquired information about how to exploit the use of multiples of three in the JN-25B code book and so could walk with FECB.
Chapter 6 of Mavis Batey’s Dilly: The Man Who Broke Enigmas gives much information on the July 1939 meeting.
- 27.
Evidence of the existence of the OTP link is to be found on page 20 of Robert Benson’s History of US Comint in WW2 published by the Center for Cryptologic History of the NSA.
- 28.
The Whitlock interview is in the same Cryptologic Museum series as that with Fabian.
- 29.
Barham is quoted on page 96 of Michael Smith’s The Emperor’s Codes. Newman’s account is available in the NAA Melbourne file B5555 2.
- 30.
The quoted text is taken from NARA RG38 CNSG Library Box 115, 5750/119. See also page 100 of the history of Op-20-GYP in the SRH series at NARA.
- 31.
The source is Laurance Safford’s A Brief History of Communications Intelligence in the United States, SRH-149 in NARA RG457. An edited version is published by Aegean Park Press under the title U.S. Naval Communications Intelligence Activities. Although there are some errors in matters of detail, Safford makes the key point: ‘If—and it is a big if—the Japanese Navy had changed the code book along with the cipher additives on 1 December 1941, there is no telling how badly the War in the Pacific would have gone for Australia and the U.S. or how well for the Japanese in the middle stages.’ (The correct date was 4 December 1941, not 1 December 1941.)
Fabian elsewhere states that there was an unenciphered message intercepted from a ship under attack near the Philippines some 12 h after the raid on Pearl Harbor. It would have been sensible to wait until several depths of 12 or more of new system messages were available and then test them to see if the known common JN-25B book groups were being used. If this was the case it would then be responsible to announce that CODE REMAINS UNCHANGED. However the second unencrypted message must have removed most doubt.
- 32.
Parker’s paper is in Cryptologia 24 (July 2000) 212–234. Parker also wrote two booklets Pearl Harbor Revisited: USN Communications Intelligence 1924–1941 and A Priceless Advantage: US Naval Communications Intelligence and the Battles of Coral Sea, Midway and the Aleutians. Both were published by the Center for Cryptologic History, National Security Agency. All three of Parker’s publications are highly recommended.
SRH-268 in NARA RG457 is given in www.history.navy.gov/library/online/srh268.htm and contains the full text of Redman’s letter.
See also Eliott Carlson, Joe Rochefort’s War, Naval Institute Press, 2011.
- 33.
One reference is Graydon Lewis: Setting the Record Straight on Midway published in Cryptologia, volume 22(2), April 1998, 99–101. This debunks theories that JN-25B was broken by means of recovering the code book from the wreck of the submarine I-124, sunk off Darwin early in 1942. The book Sensuikan I-124 by Tom Lewis on this submarine contains several high-quality photographs of it. Of course the first-hand accounts of Biard and others, coupled with overwhelming documentary evidence, make the truth of the matter quite clear. See also Biard’s paper The Breaking of Japanese Naval Codes: Pre-Pearl Harbor to Midway in Cryptologia 30(2) (April 2006) 151–158.
The myth of JN-25B being broken courtesy of the I-124 is attributed in Ken Kotani’s Japanese Intelligence in World War II, Osprey Publishing, Oxford, 2009 to a semi-published account of Commander Motonao Samejima. See page 87 of Kotani’s book, where it is stated that ‘in the lead-up to the Battle of Midway, it was clear that the IJN’s operational codes had been compromised, since the Allies had obtained the code books from the I-go 124 submarine sunk to the north of Australia in January 1942’.
In an interview given in 1969 and preserved in the US Naval Institute, Rochefort explains how his unit was dependent on IBM cards and associated machinery. His colleague Thomas Dyer had been responsible for putting together this early large-scale cryptanalytical data processing system.
Donald ‘Mac’ Showers, the ‘last man standing’ of Rochefort’s team, has recorded his experiences on a DVD prepared and sold by Shoestring Educational Productions of San Diego. He mentions various key figures in the Dungeon and confirms that JN-25B was not providing useful intelligence until late February 1942. This DVD is strongly recommended.
- 34.
The report by Fabian to Safford of June 1942 is found in NARA RG38, Crane NSG Library Box 19, and explains his difficulty in working out the intended functions of British naval signals intelligence units in Colombo and Kilindini. At that time Frumel was sending them progress work on JN-25 by means of a one-time pad supplied by Nave. Some of the unreadable OTP messages have survived in the Melbourne NAA.
- 35.
This paragraph is based on item N2 30/68/3 in the New Zealand Archives. This most valuable source was found by Alan Bath and is mentioned in his Tracking the Axis Enemy: The Triumph of Anglo-American Naval Intelligence, University of Kansas Press, 1998. However Bath does not place great importance upon its relationship with spreading the news about how JN-25B could be broken.
An unexpected security consequence of the Beasley visit is discussed in Sect. 19.3.
- 36.
The invaluable Aegean Park Press Cryptographic Series includes a version of SRH-180 edited by Sheila Carlisle. The original is available in NARA RG457. An internal Op-20-G memorandum of 23 January 1942 on this matter is mentioned on page 53 and the quoted text of the cable message is on page 54.
The second evacuation group from Cast was ordered on 12 March 1942 to report to the senior naval officer ‘for further transfer to Melbourne, New South Wales, Australia’. Likewise Fabian’s group initially had been to Exmouth Gulf hoping to catch a train to Melbourne. They all got there in the end.
Those interested in such things may find some amusement in Melbourne NAA file B5555 14, which is a ‘collection of comic relief’ assembled by Commander Newman. Some more were accumulated in the data base used for this book.
- 37.
The message of 19 February 1942 may be found in the Research Centre of the Australian War Memorial, AWM124 4/132. The response of 22 February 1942 is in the NAA Canberra file A816 43/302/18.
- 38.
The report is dated 8 October 1945 and is now SRH-197 in NARA RG457. It is reprinted by Aegean Press in booklet 65 of its Cryptographic Series; see page 63.
- 39.
The Newman material is in the Melbourne NAA as B5555 1. Earlier material is to be found in NARA.
It is of interest to compare a report sent by Captain Shaw of the FECB to GCCS in mid-December 1941. The source is TNA HW 8/102. The text is ‘New naval recyphering tables introduced 4th December, books and system unchanged. Progress expected to be similar to last period of 4 months, viz tables broken into after a month and all the 1,000 indicator subtractors and about 10,000 recyphering groups solved after 3 months.’ In fact there was massive disruption at FECB and Cast but reinforcements were provided at the Pearl Harbor Combat Intelligence Unit, later called Hypo. The process of handling JN-25 was well understood by December 1941.
The NAA Canberra file A6923 SI/2 includes on digital page 249 a report of 5 April 1942 for the Deputy Chief of the General Staff containing the cryptic sentence ‘It cannot be sufficiently stressed that Japanese Naval traffic is totally different in character from the Army’. As both the Japanese Army and the IJN used additive cipher systems, this would appear to be asserting that it had become possible to handle JN-25B (by exploiting its flaws) but high-level Army codes remained unreadable.
- 40.
SRH-020, Narrative Combat Intelligence Center JICPOA may be consulted in NARA RG457 but is also available in the book Listening to the Enemy: Key Documents on the Role of Communications Intelligence in the War with Japan by Ronald Spector. This source notes that call sign ciphers became more difficult from the introduction of JN-25C.
- 41.
Despite all the Sigint, the decision to risk so much at Midway must have been taken with some trepidation. The appropriate quotation from Shakespeare is to be found in Julius Caesar (iv 3 218):
-
There is a tide in the affairs of men
-
Which, taken at the flood, leads on to fortune,
-
Omitted, all the voyage of their life
-
Is bound in shallows and in miseries,
-
On such a full sea we are now afloat.
-
And we must take the current when it serves
-
Or lose the ventures before us.
-
- 42.
The reference is NARA RG38, CNSG Library, Box 115, item 5750/119. In fact the JN-25C code book was recovered from a fire. See also RG38, Box 42 RIP 83 for the JN-25C code book.
- 43.
TNA HW8/102, page 24.
The different systems were generally called ‘channels’. They had the same code book but different additive tables. The less commonly used channels tended to be secure enough.
- 44.
Michael Smith’s The Emperor’s Codes mentions on page 159 how Tiltman in March 1942 managed to break into the Japanese Military Attaché Code—an additive system—by putting a complete run of messages from an active tailer into depth by simply counting the groups used! The point is he would have guessed that the number of groups in the additive table was 5,000, 10,000 or 20,000 and so obtaining depth from a long run is easy.
- 45.
NAA Melbourne item B5554, known as the Jamieson report, gives a general account of minor IJN ciphers. The section on JN-11 includes a description of the Ransuuban, a substitute for an additive table. Another reference is page 152 of Norman Scott’s paper Japanese Naval Ciphers 1943–1945, published in Cryptologia 21(2) (1997).
- 46.
The following report by Hugh Foss was found by Ralph Erskine in record group HW 37 in the British Archives. It shows how sophisticated JN-25 decrypting became:
‘The text cards and additive masters are then put through an NC-4 machine which has been specially devised for Op-20-G and may be new to Freeborn.
It has two hoppers for master (read) and detail (punch) cards respectively. In this case the additives are masters and text cards details. It compares five columns of the master card with five columns of the detail card (cell number and last two digits of page in this case) and if they are different it takes the detail card and passes it through unpunched. It would do the same for the master, if it were lower, but in this case the detail cards are from a complete series, only the masters can be missing. Incidentally the machine can compare anything up to ten columns. When the master and detail cards are the same it takes both cards, punches the five columns of the additive group onto the detail card (cols 76–80), adds the additive group to the text group and thus gets the code group which it also punches on the detail card (cols 31–35). Then it works out whether the code group is divisible by 3 and if so, column 25 of the detail card is punched with the 11 punch. The detail hopper can be used without the master hopper so that two five-column fields of one card can be added and punched on it. Though this machine can only deal with 5 columns at a time it is used in preference to a multiplier, except when there are a large number of columns to be added. The reason is that this machine works at the rate of 100 cards per minute providing all detail and masters match, or only one feed is used, and the multiplier (type 601) can only handle 25 per minute.
The text cards are next sorted on column 25 into two decks accordingly to whether or not they have the code groups divisible by three. The deck with scanning groups is then sorted into numerical order (cols 35–31) and collated to verify the sorting.
All three decks:
-
1.
headings
-
2.
with scanning code groups
-
3.
with code groups that don’t scan or are absent for lack of additives
are then taken over to GS 1-A where a special unit is waiting to receive them.’
-
1.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2014 Springer International Publishing Switzerland
About this chapter
Cite this chapter
Donovan, P., Mack, J. (2014). JN-25 and Its Cryptanalysis. In: Code Breaking in the Pacific. Springer, Cham. https://doi.org/10.1007/978-3-319-08278-3_9
Download citation
DOI: https://doi.org/10.1007/978-3-319-08278-3_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-08277-6
Online ISBN: 978-3-319-08278-3
eBook Packages: Computer ScienceComputer Science (R0)