Portal de Periódicos da CAPES

John F. Byrne's Chaocipher Revealed: An Historical and Technical Appraisal

2011; Taylor & Francis; Volume: 35; Issue: 4 Linguagem: Inglês

10.1080/01611194.2011.606751

1558-1586

Moshe Rubin,

Scientific Research and Discoveries

Abstract Chaocipher is a method of encryption invented by John F. Byrne in 1918, who tried unsuccessfully to interest the US Signal Corp and Navy in his system. In 1953, Byrne presented Chaocipher-encrypted messages as a challenge in his autobiography Silent Years. Although numerous students of cryptanalysis attempted to solve the challenge messages over the years, none succeeded. For 90 years, the Chaocipher algorithm was a closely guarded secret known only to a handful of persons. Following fruitful negotiations with the Byrne family during the period 2009–2010, the Chaocipher papers and materials have been donated to the National Cryptologic Museum in Ft. Meade, MD. This paper presents a comprehensive historical and technical evaluation of John F. Byrne and his Chaocipher system. Keywords: ACAAmerican Cryptogram Associationblock cipher encryption modesChaocipherdynamic substitutionGreg MellenHerbert O. YardleyJohn F. ByrneNational Cryptologic MuseumParker Hitt Silent Years William F. Friedman Acknowledgments This author would like to acknowledge and thank the following people who were instrumental and helpful in the writing of this article: Tony Bean (John Byrne's nephew) for generously sharing genealogical information about the Byrne family and for proof-reading from a unique vantage point; David Kahn for inspiring my life-long interest in cryptanalysis in general and Chaocipher in particular, and for facilitating the successful connection to the National Cryptographic Museum (NCM); David D'Auria for negotiating the donating of the Chaocipher material with Pat Byrne; Jeff Hill for the hundreds of absorbing Chaocipher-related hours and emails traded with this author, and for proofreading this paper; Jeff Calof for providing the highest level of proofreading of this and other Chaocipher papers written by this author, and for valuable leads and ideas; Mike Cowan for coming aboard and contributing to the ongoing attack on Chaocipher; and my wife Rochelle, for her excellent proofreading of this paper and constant support. This author has a special debt to repay to three women without whose help this paper, and the story of Chaocipher, could not have been written: Pat Byrne, for magnanimously heeding the call of history and donating her father-in-law's entire collection of Chaocipher-related artifacts to NCM. Chaocipher researchers and historians will owe her a debt for generations to come. Cheryl Needle who, as Pat Byrne's associate in parallel with her own illustrious career as a book antiquarian, was highly instrumental in making the Chaocipher donation to NCM a reality. Rene Stein, NCM Librarian, who responded to every request I had from NCM with exactly the material I needed, providing it promptly, efficiently, and always with a smile. Special thanks are due to the National Cryptologic Museum of the NSA, Ft. Meade, MD, USA, for permission to reproduce pictures and photographs from the Chaocipher archives currently residing in the library. Thanks are also due to the Curran Collection for permission to reproduce photographs of John F. Byrne. The Curran Collection—Photographs. Reproduced with kind permission of Helen Curran Solterer, from the originals held in UCD Library Special Collections. Digital image is a copyright of UCD Irish Virtual Research Library and Archive. Notes 1Marcellus Bailey (1840–16 January 1921) was an American patent attorney who, with Anthony Pollok, helped prepare Alexander Graham Bell's patents for the telephone and related inventions (from Wikipedia). 2Calculated using http://www.dollartimes.com/calculators/inflation.htm, assuming an annual inflation rate of 2.87%. 3Lieutenant Commanders Samuel M. Tucker and F. R. Furth coined the acronym ‘radar’ in November 1940, see http://www.meteor.iastate.edu/~jdduda/portfolio/571_write_up.pdf. 4Byrne refers to the two younger Commanders as Wagner and Tucker. I believe Byrne incorrectly refers to Lieutenant Commander Joseph N. Wenger. As a Communications Intelligence career officer and a cryptanalyst, Wenger headed OP-20-GY, the Navy cryptanalytic agency in Washington, during most of World War II. By 1937–1938 Wenger was a critical player in the field of machine processing and aided in the development and refinement of cipher devices, which were adopted by the U.S. Navy. He would, therefore, have been supremely qualified to sit in on this meeting with Byrne. See Weller, Robert, 1984. Rear Admiral Joseph N. Wenger USN (Ret) and the Naval Cryptologic Museum. Cryptologia, 8(3):208–234. This theory is upheld by David Kahn in The Codebreakers (page 768) who writes “negotiating apparently with Commander Joseph N. Wenger …”. 5On 4 May 1938, the day following Byrne's aborted demonstration, the US Navy submitted a patent for a rotor machine known as the ECM Mark III (U.S. Patent 4,143,978). It is safe to assume that Tucker and Wenger had already decided on adopting the ECM Mark III. From Byrne's memorandum to G. M. Campbell [Citation5] it would seem that Byrne at least began his demonstration to the Navy. See http://www.quadibloc.com/crypto/ro020703.htm for information about the ECM Mark III. 6Colonel Charles H. Hiser, 29 December 1915–28 August 1988. 7This occurred almost two years after the Army Corps, led by Friedman and Rowlett, deduced the nature of the Japanese PURPLE machine. One can understand why Friedman was not enamored by a cipher system rising or falling on a single submitted message. 8A worksheet dated 25 February 1947 and found in the National Cryptologic Museum library, showing alphabets derived from the keyword “COMPREHENSIBLEX” using the disk pattern “LRRLLRRLRRRLLRL,” adds the text “To encipher Friedman test stuff: …”. This seems to indicate that Byrne was preparing material for Friedman as late as 1947. It is significant that Byrne does not recount any of this in Silent Years. 9Byrne used text from General Douglas MacArthur's address to the US Congress, 19 April 1951. There are differences between the different transcript of this speech; Byrne obviously used one of the many versions. 10The word ‘holocryptic’ is defined as ‘wholly or completely concealing; incapable of being deciphered.’ Thus, a holocryptic cipher is a cipher so constructed as to afford no clue to its meaning to one ignorant of the key. 11Although one theoretically cannot prove that a given message is not genuine (“proving a negative”), we can assume Yardley examined evidence other than first-order frequencies, enabling him to declare strongly that no known cipher system could display the same randomness in all observed aspects. 12In a private email communication between David Kahn and this author on 21 November 2010, Kahn was unable to recall where he got the theory of an autokey as part of his research. 13The meeting took place at Deavours's home in New Jersey, with John, his wife Pat, Tony Bean (John's nephew), and his wife Brenda attending. John and Tony showed Kruh and Deavours the drawings and components of John's rudimentary model. As Tony relates, “I remember both Cy and Lou getting very excited about this disclosure and after dinner Cy retired to his study and wrote a small [8080] assembly language program that replicated the Chaocipher algorithm. […] I do not recall any agreement (other than non-disclosure) ever came although we knew articles would be written for Cryptologia” [Citation1]. 14Private communication from Jeffrey Hill to Wesley Horton, 23 July 1998. Hill tells how Kruh, at the 1997 American Cryptogram Association (ACA) conference, told him how he asked John Byrne “for a disclosure article to be published in the 20th anniversary issue of Cryptologia (January 1996). Byrne seemed to agree, but the deadline for the article came and went and no disclosure was forthcoming.” 15According to the Irish Virtual Research Library and Archive, the back of the photo says “La Bergerie 1917.” In this photograph, however, Byrne is most visibly older than the photograph to its left (i.e., circa 1924) so one of the datings may be inaccurate. For information about the photos, see [Citation28]. 16Patricia Byrne is the former Patricia Neway (born 30 September 1919, Brooklyn, New York). Patricia is an American operatic soprano and musical theatre actress who had an active international career during the mid-1940s through the 1970s. She is particularly remembered for creating roles in the world premieres of several contemporary American operas, most notably Magda Sorel in Gian Carlo Menotti's ‘The Consul.’ On Broadway, she won a Tony Award for her portrayal of the Mother Abbess in the original production of Rodgers and Hammerstein's The Sound of Music [Citation62]. 17See http://www.mountainvistasoft.com/chaocipher/chaocipher-016.htm for description of how this author contacted Pat Byrne. 18It is perfectly logical to alternate between locating the plaintext letter in the right or left alphabet based on some prearranged pattern. As will be shown in this paper, Byrne used this alternating alphabet method for deriving the starting alphabets. 19Tony Bean relates [Citation1] that he had many conversations about ‘the model’ with his uncle John (the son of John Francis Byrne). He writes “… to the best of my knowledge, my grandfather never constructed any device that was “too heavy and cumbersome.” The only models I ever heard of were the “cigar box” one and the two disk model depicted in [Figure 9].” 20In actuality, Greg Mellen's hypothetical system [33, pp. 149–152], included in his 1979 paper, may be the first recorded example of a dynamic substitution cipher (thanks to Jeff Hill for pointing this out). 21In an email thread this author had with Terry Ritter, Ritter could not think of a pre-90's example of Dynamic Substitution. In a later conversation, Ritter wrote “it seems to me that it [ed. Chaocipher] could well be related to DynSub. The idea of combining RNG [ed. random number generator] and combiner would be an interesting approach for a system not based on computing.” 22I am indebted to Jeff Hill for this interesting insight. 23This may explain why “MA” consists of all capitals while the others consist of a capital latter followed by a small letter, i.e., Byrne retained the upper/lower case when borrowing the letters. 24I am indebted to Jeff Calof for raising this possibility. Additional informationNotes on contributorsMoshe Rubin Moshe Rubin is a software engineer residing in Jerusalem, Israel. He received his B.Sc. in Computer Science from the Jerusalem College of Technology in 1979. His interest in cryptanalysis began at age fifteen after reading David Kahn's The Codebreakers from cover to cover multiple times, and it was Kahn's description of Chaocipher that led to a life-long fascination with this cipher.