by DAVID KAHN
The T-55 is an on-line device that enciphers teletypewriter pulses instead of letters, either by an appropriate modification of the Hagelin cage principle or by a straightforward one-time tape Vernam principle. To make sure that the tape is truly one-time, this machine slices it in half after use! It is much larger and heavier than the other machines.
In addition, Hagelin tempts his customers with a full line of accessories, which in their way are not unlike those that hi-fi addicts or yachtsmen find so hard to resist. A base with a keyboard and an electric motor fits under the C-52 to convert it to rapid, typewriter-like operation—for $1,000. (This gadget replaces the wholly separate electrical printing machines that the firm used to manufacture.) Attachment PE-61 will perforate a teletypewriter tape with the C-52’s ciphertext. To facilitate setting the keywheel pins, the pin-setter SRP-58 is available. Arabs, Burmese, Thais, and other users of non-Latin alphabets may purchase machines with their own scripts; these usually serve only for the plaintext, the ciphertext using the Latin letters which are more acceptable in international communications. Units to produce one-time tapes may also be purchased. (These usually generate the random keys required from one of the most random processes known: the decay of a radioactive element. A Geiger counter causes the unit to punch a hole in the tape whenever the disintegration exceeds a certain level in a given period of time; it leaves a blank when the rate falls below this level. Thermal noise, which is equally unpredictable, is also used.)
Nearly all Crypto Aktiengesellschaft’s production goes to its approximately 60 governmental customers, whose military services buy substantially more than their diplomatic. Complete installations typically cost from between $30,000 and $50,000. When the purchasers squawk about the price, as they invariably do, Hagelin’s representatives ask them whether they ever send messages whose value is much greater than that. This quiets them. A minute portion of the firm’s output goes to commercial users, nearly all of whom have international interests, usually highly competitive, such as oil and mining, or highly confidential, such as finance.
The firm explains that the “tremendous number of variable elements”—more than 24 quintillion quintillion quintillion quintillion—enables each customer to select an individual set of keys. It advises which key procedures are good and which bad, but carefully refrains from recommending specific keys because it does not want one customer to think that it is giving them instructions that it could also give to others. “It is not good business practice for us to be knowledgeable of the details of the customer’s machine and usage, which should be truly national secrets,” says the company in its instruction brochure, “any more than it would be for the safe manufacturer to know the combination of his customer’s safe.”
Hagelin, whose house in Zug stands a few dozen yards behind his factory, has retired only partially. He does not spend all day at his unique operation, but he still does most of the firm’s development work. He says, “I don’t understand electronics but I know what it can do.” He handles his old customers, though he leaves the new ones and many administrative details to his general manager, Sture Nyberg. A white-haired man of greater than average height and of average build, with firm, pleasant features, Hagelin has a quiet humor and a gentleness about him. Is it true that he speaks five languages fluently? “Only one at a time,” he smiles. In his pockets he carries peanuts to feed the birds that fly to his window; one day one perched on his head and another on his arm as he was walking up the steps from his office to his home. At night they leave their “calling cards” on his bedroom light.
His interests are wide. He discusses food like a gourmet, takes good amateur photographs, enjoys sailing, and talks knowledgeably about the flowers in the beds behind his house cultivated by his wife, the former Annie Barth, a distant relation of theologian Karl Barth. He lives very well. Twice a year he returns to Sweden, either to his estate outside of Stockholm or to a log cottage in the north. A soufflé served at luncheon by his cook will be as high and fluffy and tasty as any in a fine restaurant in New York or Paris. The smell of the tan leather upholstery in the white Mercedes-Benz that he drives is almost overpowering. His guest book contains signatures from all over the world—the United States, France, Egypt, Iran, Germany—and he himself is an extremely gracious and considerate host. Hagelin has enjoyed greater material rewards from cryptology than any other person in the world, and it might well be said that it couldn’t have happened to a nicer guy.
* The code used four-digit groups, but the French divided the codetext into clusters of five figures, and divided each cluster into two pairs, which were enciphered by the tables, and a single figure, which was left unenciphered. The encipherment thus straddled from one codegroup to another. The French furthermore cut the clusters in three ways: with the single figure at the beginning, with it between the two pairs, and with it following them. Kunze began working on the superencipherment in 1921 and had reconstructed it by 1923. He returned to the system in 1927-28 and solved the code, which the French were still using. By then they were combining the single digit of one cluster with that of another and enciphering them as a pair. For instance, the imaginary codetext 8975 4263 … would be divided 8 97 54 2 63 …, and the 8 and 2 would be enciphered together, as well as the 97, the 54, etc. Kunze solved that variation as well.
* In its original form, the ciphertext included the stunt characters. This made it difficult to record the ciphertext on paper. The sudden appearance of a figure shift would abruptly convert a literal cryptogram into one of numbers and punctuation marks. A carriage return without a paper feed would result in an overline. To prevent this, Vernam added some circuits that would cause the stunts to print as two-letter groups.
* In the original C-36, only 25. The machine described here is the M-209. Similarly, where the M-209 had six keywheels and moveable lugs, the C-36 had only five keywheels and fixed lugs. The increase in the number of keywheels and the number of bars, and the moveability of the lugs, is due to Yves Gyldén. The operation of the two models is the same, however.
14
DUEL IN THE ETHER: THE AXIS
SHORTLY AFTER NOON on the tense 31st of August, 1939, the last day of peace that the world was to know for six years, Swedish businessman Birger Dahlerus met with Hermann Göring at the Nazi leader’s large and richly furnished town house at 2 Leipzigerstrasse in Berlin. Dahlerus had been trying desperately to avert the onrushing cataclysm of war by flying between England and Germany as Göring’s unofficial mediator. Britain had pledged to aid Poland if Hitler attacked her, and, in an effort to stave off actual warfare had proposed to both Germany and Poland that they negotiate their differences directly. At a few minutes past one, as Dahlerus and Göring were discussing the situation, an adjutant brought in a red envelope of the kind used for especially urgent affairs of state. Göring ripped it open. When he read its contents, he leaped from his chair and, striding angrily up and down, raged at Dahlerus that he had in his hands proof that the Poles were sabotaging every move toward negotiation.
After a few minutes he calmed down enough to tell the Swede what had been in the envelope. It was a telegram from the Polish government in Warsaw to its ambassador in Berlin. It was in code, of course, but the cryptanalysts of the German Foreign Office, who had long ago cracked the Polish diplomatic code, had reduced it to plaintext at once, translated it into German, and sent a copy to Göring via messenger. The entire process had taken less than an hour.
At the end of the telegram came a “special and secret message” to the ambassador: “Do not enter under any circumstances into any factual discussions….” To Göring this proved so conclusively that the Poles had no intention of negotiating in good faith that he copied the translation in his own hand for Dahlerus to show the British ambassador. The German Air Minister told Dahlerus that he was taking a great risk in doing this—he undoubtedly meant jeopardizing Germany’s possession of the Polish code—but felt that Britain should know how faithless the Poles were.
&n
bsp; In fact this was not a reason for going to war, but just another excuse to do so. The Germans were using Dahlerus as a cat’s-paw, for at the very moment that Dahlerus entered Göring’s home, Adolf Hitler was signing his “Directive No. 1 for the Conduct of the War.” At daybreak the next morning German troops invaded Poland. And although the Foreign Office solution of the Polish message had no role in that attack except to confirm the Nazis in their perfidy, it did demonstrate the keenness and efficiency of one of Germany’s major intelligence weapons as she embarked upon what she fondly thought would be her blitzkrieg of conquest.
The cryptanalytic service of the German Foreign Office was created early in 1919, apparently at the suggestion of Kurt Selchow, a 32-year-old former captain in the Army intercept service. Selchow became its administrative chief and staffed it with cryptologic acquaintances from the war. His organization was at first known as Referat I Z, the Z section of Division I, Personnel and Budget, of the Foreign Office. It included both the cryptanalytic service (the Chiffrierwesen) and the cryptographic (the Chiffrierbüro), the latter twice as large as the former. Around 1936 a reorganization of the Foreign Office renamed I Z as Pers Z (pronounced “pers-zed”), the Z section of the Personnel and Administrative Division. The Z meant nothing—the division did not have 26 sections—and it may have been chosen because it seemed appropriate to cryptology. Much later, Foreign Minister Joachim von Ribbentrop took the Chiffrierbüro under his own office, presumably to restrict access to his own coded telegrams.
By 1939, Pers Z had divided the Chiffrierwesen into two groups—one that dealt with ciphers, either as primary systems or as superencipherments, and that was heavily mathematical in personnel and approach; and one that dealt with codes and emphasized the linguistic.* Three senior cryptanalysts headed them—Rudolf Schauffler and Adolf Paschke as joint chiefs of the linguistic section, Dr. Werner Kunze as chief of the mathematicians. All were veterans of the military cryptanalytic bureaus that Germany had belatedly started in World War I; all joined the Foreign Office in 1919 when they were close to 30. Schauffler and Kunze participated in the development of the one-time pad.
Kunze had his doctorate in mathematics from the University of Heidelberg, where he also studied physics and philosophy. A cavalryman for most of World War I, he began cryptanalysis in January of 1918, solving some English ciphers but working for several months without success on a British code. During his first years in the Foreign Office, he studied cryptanalysis, emphasizing theory and applying his mathematical knowledge. Kunze may well have been the first mathematician employed in a modern cryptanalytic office. About 1921 he opened his first major assault—on the superencipherment of a French diplomatic code. He solved it in 1923, thus learning early the need for persistence and patience in cryptanalysis. His theoretical studies helped him in the joint development of the one-time pad for German diplomatic systems.
In the spring of 1936, he undertook his finest work: ascertaining the system of and ultimately solving a Japanese machine cipher, apparently known to the Americans as the ORANGE system that preceded the RED machine that preceded PURPLE, the same system solved by the U.S. Navy’s Lieutenant Jack S. Holtwick, Jr. Kunze thought that the solution would take only six weeks, but it was not until the day before he went on his vacation in July that he made his entry. The machine enciphered vowels into vowels and consonants into consonants by separate sets of rotors, and by September Kunze had recovered all the alphabets used in the machine. He also solved the later RED, in which vowels and consonants were enciphered indiscriminately through the two rotor arrays. But neither he nor anyone else in Pers Z ever solved the final system of this development, the PURPLE machine.
Paschke and Schauffler served as joint heads of the linguistic group because it was so large. Paschke was the nominal head, handled more of the administration, and was in charge of European languages; Schauffler, an expert in Asiatic languages who was in charge of them, also had a good grounding in mathematics. He concentrated more on the substantive work. As one colleague described it, “Paschke said he was in charge and Schauffler was modest and didn’t object.” Paschke, a slight, erect man, with a little mustache and a small smile, was sensitive and touchy, but courteous and a good family man. Born in St. Petersburg, he got into cryptology in 1915 because of his expertise in Russian. Though he was a lawyer, he liked cryptology so much that he stuck with it, working on Russian, British, and Italian codes. He was a natural linguist. One of his specialities was to establish the meanings of the first 500 or so codegroups of a code and then to turn it over to a less able cryptanalyst for the much easier task of completing the solution.
Schauffler, a nervous and high-strung man who had studied at Tübingen and Munich, and had taught school before the war, started his cryptanalytical career at Army headquarters in 1916. A thorough student who felt that a comprehensive theory would pay off in practice, he probed much more deeply into the core of cryptology than the others, who mostly limited themselves to practical results. Thus he tried to systematize the science, sought to impose a uniform terminology, kept up with the work going on in all areas, and either wrote or encouraged the writing of reports on important topics. Pers Z’s theoretical investigations into the mathematical structure of the Enigma and into the regularities in a stack of codewords that are necessary to correct garbles but that correspondingly help cryptanalysts were probably inspired by Schauffler, as well as its texts on cryptology, its “Introduction to Probability Theory” with applications to cryptanalysis, and its preparation of graphs and nomograms. Schauffler bridged the linguists and the mathematicians—he knew the main Asiatic languages well enough to provide the linguistic data to help Kunze reconstruct the alphabets of the Japanese cipher machines. After the war he got his doctorate in mathematics.
These three were chiefly assisted by three other old-timers, Erich Langlotz, the third inventor of the one-time pad; Ernst Hoffmann, who held the title of Counsel for the High Cipher Service; and Hermann Scherschmidt, a specialist in Polish and other Slavonic codes. All usually held the same rank of Regierungsrat that Kunze, Schauffler, and Paschke did. In 1933, when Hitler came to power, Pers Z employed about 30 civil servants. As Germany rearmed, Pers Z expanded, though slowly at first. Recruiting was subtle: prospective recruits did not know that they were being considered for the highly secret work of cryptanalysis. One woman, Asta Friedrichs, who had taught school in Bulgaria and knew that language, which Pers Z needed, was simply asked if she would like to learn Serbo-Croatian and do some work involving it; she accepted, and not until after a probationary period was she told about the codebreaking. She began solving Serbo-Croatian codes, then some Bulgarian, then helped with others.
With the outbreak of war, Pers Z’s growth became explosive. Among the brightest of its new members was Dr. Hans Rohrbach, a 37-year-old mathematician who later became editor of the oldest mathematical journal in the world, the Journal of Pure and Applied Mathematics. Another mathematician was Dr. Gottfried Köthe, later rector of Heidelberg University. The agency needed people and it made exceptions. Ottfried Deubner, whose father, Ludwig, had solved Russian military cryptograms for Germany in World War I, was partly Jewish, but he was allowed to join and work on Italian cryptograms because of his father’s earlier contributions; the Nazis made him an honorary Aryan.
For several years, Pers Z had been situated on the top floor of the library building just behind the Foreign Office main building in Berlin’s Wilhelmstrasse. But by early 1940, it had burst out of these quarters. The mathematicians moved out first, into several flats in an apartment house at W-8 Jaegerstrasse that had been entirely taken over by the Foreign Office. Their departure relieved the crowding in the original office only temporarily, and soon the linguistic codesolvers found new offices, first in an anthropological museum, where they were surrounded by artifacts from Siam, and then in Dahlem, a suburb of Berlin. Here some worked in a garden apartment on a street called ImDol, some in a nearby girls’ boarding school, where they were joined
in 1943 by the mathematicians. The combined group, the Chiffrierwesen arm of Pers Z, called itself the Sonderdienst Dahlem (“Dahlem Special Service”). While there, during the middle period of the war, it consisted of about 200 staff members—20 to 25 mathematical cryptanalysts, probably the same number of linguistic cryptanalysts, the rest clerks and support staffers. Later it grew to 300.
Heavy bombings—the workers had to spend nearly every night in air-raid shelters—forced still another move in the summer of 1944. The linguistic branch moved 150 miles southeast to Hirschberg in Silesia, where they installed themselves in another school; the mathematicians moved to the nearby town of Hermsdorf. The odyssey of Pers Z did not end even there, however. In February, 1945, the advance of the Russians compelled each group to move about 150 miles west. The mathematicians evacuated to Zschepplin Castle, near Eilenburg, about 80 miles south of Berlin. The linguists, joined by a few mathematicians to strip current superencipherments, moved into a wing of Burgscheidungen Castle near Naumburg, northwest of Wiemar. Here, as wartime guests of the Count von der Schulenburg and his five daughters, the 90 cryptanalysts, some with their wives, lived and worked amid art treasures and ancient furniture, handicapped by the almost total lack of liaison with the mathematicians, about 50 miles away.
The ever-present problems of security added to the difficulties of Pers Z. Ink was not permitted because it required blotting paper. Each night all papers had to be locked away. Waste paper had to be burned, and the ashes broken up to make sure that no cinder would float away. Later Pers Z got a machine to shred the paper before it was incinerated. None of the codesolving groups was allowed to know what the others were doing—but these artificial barriers dissolved in the camaraderie of the Dahlem bomb-shelter.
Security also meant political security, and even before the war the Nazis planted a spy in Pers Z to watch for any signs of anti-Hitler activity. In 1942 Selchow became a Nazi. He took the honorary rank of Sturmführer, which gave him access to three or four cars. The next year he became an Obersturmführer because this gave him “a certain authority with the drivers.” However, he insisted, he never wore the uniform. Among the cryptanalysts, Paschke, Schauffler, and Kunze, at least, also joined the Nazi party.
