"In a paper published in 1926, an American engineer, G.S. Vernam, pointed out that it would not be difficult to convert a standard teleprinter into a cyphering machine. His theory was that the five-unit electrical impulses produced by a teleprinter, when directly connected with a wire, could be mixed at the moment of transmission with a second series of impulses to give a third and meaningless series. This last would appear upon the receiving teleprinter as a mere jumble of characters unless the second series of impulses were removed before the message was printed. It was, in effect, the principle that A plus B equals C and that C minus B produces A once more - the familiar principle of the one-time recyphering table. Vernam had shown how this could be applied by transmitting the impulses produced from a piece of five-unit teleprinter tape, perforated at random, simultaneously with an en clair message. At the opposite end of the wire the jumbled impulses were filtered through an identical piece of perforated tape and reappeared as en clair characters upon the receiving teleprinter.
 |
| Telekrypton photo from the book BEST-KEPT SECRET by John Bryden. (Virginia Military Institute, George Marshall Library, Lexington, Virginia). |
The idea was sound, and the Western Union Telegraph Company had gone so far as to manufacture a limited amount of the equipment necessary for Vernam’s cyphering process. However, it had no commercial success, and in December 1941, two Telekrypton cyphering machines (such was the trade name) were still lying in the Western Union warehouse.Telekrypton had some security weaknesses which were corrected in a complete redesign of the machine which was then called Rockex I. Canadian communications expert Benjamin deForest Bayly used the existing, patented Telekrypton enciphering device and modified the combining logic so that only the 26 letters of the alphabet appeared in cipher with all "functionals" suppressed and the output formatted in 5-letter groups separated by spaces. Specifically, the new design did not encipher carriage return, a line feed, a space, a letter shift, and a figure shift. The Teletype Corp of Chicago, Illinois supplied all the parts for Rockex. The output of the new machine could now be sent over conventional commercial communications circuits instead of private leased lines.In many respects the Western Union Telekrypton was well suited to BSC’s (British Security Coordination) need for secure local communication between Washington and New York. It had, however, two weaknesses. The code tape, which consisted of a string of random teleprinter characters, was joined at the ends to form a loop. This loop ran through the transmitter continuously as a message was being sent, and, although the mélange which passed down the wire was impressive, it would not have given much difficulty to a cryptographer. Secondly, the machines were far too complex for their purpose and would obviously be difficult to keep in running order.
Despite this, Telekrypton looked promising, and to the evident surprise of Western Union Telegraphs, BSC bought the two machines. First they had to be remodelled and unnecessary working parts eliminated. Then the loop of keystream tape required improvement. It was clear that what was needed was not a loop, but a limitless quantity of tape, perforated at random and in duplicate, and cut into convenient lengths. One set of these could then be used for enciphering, and its twin at the far end of the wire for deciphering. Individual lengths could be destroyed after a single use and the level of security would be the same as that of one-time tables.
Although the electronic tape-punch was not completed until the beginning of 1943, the Telekrypton cyphering equipment, was supplied with hand-punched coding tapes and came into operation on a line between BSC/New York and BSC/Washington in January 1942. After a few days, during which it was studied with suspicion by both its operators and the office at large, Telekrypton was accepted as part of the general furniture of life. It proved to be so successful that in February, at a meeting with representatives of the Canadian Military and Naval Intelligence and of the Department of External Affairs held in New York to discuss the improvement of communications between Ottawa, BSC and the United Kingdom, it was decided to install a line fitted with Telekrypton between BSC and Ottawa. The line came into operation in May 1942. By the following July, BSC was also connected by Telekrypton with its new transatlantic wireless transmitter at Camp X in Oshawa, Ontario Canada.
In the course of the next two years other lines were installed from BSC in order to connect it with the Signals Security Agency (SSA) of the US War Department, Arlington, Va. (January 1943), with the BSC Security Office in downtown New York (February 1943), with the offices of the Combined Chiefs of Staff, Washington (November 1943), and with the Communications Annex of the US Navy Department, Washington (February 1944). In each case Telekrypton was used, save with the Signals Security Agency. For administrative reasons it was found simpler that SSA should provide BSC with its own standard cyphering machine, a device similar to the BSC version of Telekrypton, except that it used multiple loops of coding tape instead of unique lengths. These loops, which had to be changed constantly during transmission to maintain security, caused considerable inconvenience and by mutual agreement the SSA equipment was eventually discarded in favour of the British Typex which was sure, though slow".
Though this was a great achievement, the real problem in any one-time system is the production (and distribution) of vast quantities of genuinely random keystream tape. It was shortly after WWII that engineer Don Horwood, working for GCHQ and ex-GPO COLOSSUS, produced a genuinely electronic random keystream tape generator. In the new design, twin keystream tapes were produced from a noise generator which fed 5 flip-flops randomly. (Don retired from GCHQ in the 1980's and passed away in 2004).
The Secret History of British Intelligence in the Americas goes on to say:
"By January 1943, BSC had supplied London with a machine of its own, and thenceforward Rockex I was used to the limit of its capacity. Rockex I remained the standard means of enciphering and deciphering almost all SIS telegrams passing between London and New York throughout 1943 and 1944, and it also proved invaluable for the transmission, via BSC, of an uninterrupted flow of Top Secret intelligence between GCCS and the Combined Chiefs of Staff, Washington".
Patent 4,332,977 dated June 1, 1982 and originally filed June 12, 1962, describes a Teletypewriter Privacy System which makes numerous references to Telekrypton.
"For commercial privacy use, a machine known as "TELEKRYPTON" has been widely used. This system has performed admirably for a number of years but has certain disadvantages relative to size, cost, complexity and portability. Telekrypton uses a cryptographic program generated by a paper tape which completely obviates the possibility of system portability. Also, the tape causes serious problems, with regard to the amount, temperature, and humidity of storage areas.
Telekrypton is complex since it uses multiplex equipment to time share one signal relay coil. The relay coil is connected in a three terminal delta network so that current flows in either direction through it dependent on the program signal and the input baud.
This [new] device may be activated during message reception when
the operator notices that the readout is gibberish. It represents a great
improvement over the Telekrypton in which the entire system must be shut
down to advance or retard the paper tape.
Ray White who served as a Communications Supplementary rate at HMCS Coverdale, Moncton N.B. recalls his experience with the Telekrypton device. "In our usage of this beast, we fed the output into a Teletype Model 19 - keyboard, page copy, chad tape punch. This was in the 1958-1959 period and I was told that our configuration was similar to that in NDHQ (National Defence Headquarters) and in the Halifax Communication offices of the Flag Officer Atlantic.
Before commencing the encryption, the operator would type the "header" and then start the process. The Model 19 would produce a tape which would then be sent to the line. We didn't actually transmit on-line from the crypto machine mainly because the Canadian National/Canadian Pacific (CN/CP) wire chief in Moncton would interrupt the transmission and call us to say something was wrong at our end, that he had checked the line for distortion etc and can't figure out what's up. We would explain that this was the intended transmission and please don't worry about it. One of their problems was their in-house requirement to record the number of groups. Our technical staff suggested they might wish to approximate the number of groups by taking into consideration the length of time it took for a given message to pass through their relay point. This would work for one Wire Chief but they must have had a rapid turnover of staff because the problem persisted. Anyway, we had produced a tape of the encrypted message and could retransmit it with only a slight loss of timeliness.
One problem that really irritated our technical branch was that the encrypted text was also printed on the Model 19 and, because such things as line feed, carriage return, etc, were eliminated, it would print at the right-end of the platen and the pounding of the keys at one spot tended to make a mess of the rubber-coated platen. I suppose we could have produced 'tape only' but for some reason we produced a page copy of every transmission. Incidentally, I always wondered how easy it would have been for someone to intercept the encryption process because there must have been arcing between the segments of the commutators.
The table in the photo found in "The Best Kept Secret" was not exactly as I remember it. We used a standard unwired wooden Canadian government table so we had to plug into the wall. The keytape was supplied by CBNRC (Communications Branch of the National Research Council) . Every 36 inches or so there would be an ink-stamped 'starting point' and when a message ended, a subsequent encryption would start from the next stamped indicator. Once it had been used the keytape was destroyed. We kept track of tape usage in much the same manner as Rockex tapes. Some of our messages were really enormous, consisting of locally-generated intelligence reports".
There is some reference to Telekrypton in the UK National Archives and some indications that it was being redesigned.
ARCHIVAL INFORMATION
Mr. Jones and I have recently had a discussion with Commander Travis and [redacted] about the Telekrypton instrument. This is a machine devised by the Western Union Telegraph Company. Its original purpose is to transmit direct to a telegraph cable, in cypher, a message which is typed on to it en clair. The occasions on which a machine could be employed in this manner are few; since to none of our posts is the volume of Foreign Office traffic sufficient to justify the leasing of a private wire. Luckily, however, this machine can also be employed as an ordinary cyphering machine similar to the Typex, from which, however, it differs fundamentally.
The Telekrypton instrument consists of a typewriter keyboard which produces a punched tape, the letters being represented by the ordinary telegraphic punchings. This tape is fed into a small cyphering machine together with another tape perforated with letters punched at random. They thus correspond to the random figures of the One-Time Pad.
When the two tapes are put through the machine, the original message tape and the "One-Time" tape combined produce a cyphered version which in the original instrument was transmitted direct to the telegraph line, but which can also be fed into a typewriter which types it out as a series of groups of letters. This typed cypher can then be handed to a wireless operator or a Post Office employee in the ordinary way. The speed of operation of the machine is limited by that obtainable from a good touch typist, say 50 words a minute.
To be sure of the capabilities of this machine it is urgently necessary that is should be submitted to test under working conditions obtaining in one of our larger missions, where the only maintenance available would be that supplied by, e.g., one of "C's" wireless operators. But our impression is that with tapes made of the right kind of paper and reasonably careful operation, its speed of operation will exceed that of the Typex machine.
Unless this machine can be adopted in some of our more important posts in the fairly near future, I am at a loss to see how cyphering staff can be supplied for the ever-growing traffic of the Foreign Office and for missions which will undoubtedly be sent to the re-occupied countries.
[Redacted] tells us that his representative in the United States had introduced some improvements into this machine which, among other things, would enable it to produce a cyphered text consisting of letters only instead of, as at present, a mixture of letters and figures. This representative is due in this country shortly.
The machine can be manufactures in [Redacted] workshops of standard components.
I consider it of great importance:-
(1) that his expert should complete the re-designing of the instrument within the shortest possible time;
(2) that [Redacted] should then make three prototypes in his experimental workshop;
(3) that these prototypes should be subjected to test, say between
London and Cairo, or London
and Algiers;
(4) that if found satisfactory [Redacted] main workshops should be instructed to produce a run of, say, three dozen machines as soon as possible; and
(5) that steps should be taken to ensure absolute priority for the Foreign Office in the testing and adoption of these machines. Otherwise, there is a risk that [Redacted] and Commander Travis, both of whom have intimate "sales connections" in the Services, may let other orders get in front of ours.
31st August, 1943.
-------
6th September, 1943
Thanks to Travis and [Redacted], we have had an opportunity of inspecting the Telekrypton Cyphering Machine; and I have discussed with all concerned the possibility of adopting it for use of the Foreign Office.
The provision of book-cyphering staff for the Foreign Office and our missions abroad presents an increasingly difficult problem. This will be accentuated when, as we hope may happen in the fairly near future, we have to supply at short notice cyphering staffs for missions dispatched to re-occupied countries. So serious is the position that when this occurs we may be faced with a danger of complete breakdown of cypher communications.
I have therefore come to the conclusion that all possible steps should be taken to push ahead, as a matter of urgency, with the adaptation, and test under working conditions, of the Telekrypton Machine. I understand that your representative, who has been working on this machine in the States, is shortly due here; and that he will be able to suggest certain modifications to it which will improve it in various respects and, in particular, enable it to produce a cyphered text consisting only of letters, instead of letters and figures mixed. I should be grateful if you could do anything to expedite this stage of the work in order that the production of, say, three prototypes in the experimental workshop may be proceeded with as soon as possible. Of these three prototypes, I suggest that one should be retained in the workshops for experimental purposes; that one should be installed at the Foreign Office, and that the other should be installed at one of our missions, say Cairo, Algiers, Lisbon, or Washington, where only elementary maintenance would be available.
As soon as working tests indicate that the machine is suitable for our purposes, we should like the manufacture of a batch of, say, three dozen machines to be put in hand with all dispatch. I take it that the Government Code and Cypher School will be able to undertake the cutting of the scrambling tapes.
As the Services and other Departments of State are well equipped with Typex machines, I think it important that the new machine should be regarded primarily as a Foreign Office instrument and that, in any case, the Foreign Office should have absolute priority in obtaining them until its requirements are met.
It was suggested that the adapted machine should be known as the "Rockex"; but to emphasize its primary association with the Foreign Office there might, i think, be advantage in endowing it with a name expressive of this fact, such as "F-O-X" or "F-OTP-EX".
I should be grateful if you would let me know whether you agree with the above and how soon it will be possible to demonstrate the improvements to the original Telekrypton instrument which have been devised by your representative in the States.