by Kahn, David
It was probably World War I that made Scherbius succumb to the bacillus of cryptography. Yet that science was underdeveloped in German-speaking lands. The most recent comprehensive text in German dated from 1881, more than forty years before, and the author had had to publish it himself. The German and Austrian literature after that time consisted of a handful of scholarly historical articles and books, a few survey articles in scattered magazines, pamphlets telling how to shield love letters and telephone conversations from pryers, a booklet overview of elementary ciphers intended for businessmen, and studies of cable secrecy and codes. A few dozen cipher devices had been patented in Germany, Austria, and Switzerland, but they had merely mechanized systems that were hundreds of years old.
Perhaps the greatest activity, and that not very intense, was manifest in the publication in German of commercial codes. These thick books, sometimes produced privately for a firm, sometimes published for general sale, replaced business and personal phrases with codewords. “Do not exceed limit,” for example, might become JIWUL. Their chief purpose was to economize on cable tolls. But they did provide some secrecy, they were constructed in the same way as many secret governmental codes, and they had the word “code” in their titles, all of which brought them into the purview of cryptology. The codewords, sometimes taken from real languages, sometimes made up, were always “pronounceable,” because international telegraph regulations set lower rates for pronounceable codewords than for unpronounceable ones or for codenumbers.
Scherbius’s first cryptographic device sought to maintain this economy while making these mostly nonsecret messages secret. It enciphered codenumbers into pronounceable codewords by replacing the successive digits alternately with vowels and consonants. One of the first cipher mechanisms to employ electricity, it passed the input impulses through “multiple switch boards which connect each arriving lead with one of the outgoing leads and which are adapted to interchange this connection with great facility of variation.”
These switchboards formed the germ of the rotor. That concept may have come to Scherbius while he was at a concert, as his best ideas often did. He was said to be very musical, but his mind apparently wandered frequently from the melody, for he often jotted ideas and made calculations on his cuffs while the orchestra played. His first rotor enciphered numbers, presumably codenumbers, gaining security but losing pronounceability.
A rotor for letters followed, and it was this device that Scherbius submitted to the navy and the Foreign Office in the spring of 1918. That both rejected his machine did not diminish his confidence in it. He turned to the commercial market.
Scherbius & Ritter transferred the cipher patent rights to the Gewerkschaft Securitas. Though a Gewerkschaft was, in the German law of the time, a corporation for mining (and this one was indeed headed by a mining director), this one’s name, Securitas, and the fact that it had also been granted the rights to the Dutch rotor patent suggests that it may have been established to funnel risk capital into cipher machines. On July 9, 1923, Securitas founded the Chiffriermaschinen Aktien-Gesellschaft (Cipher Machines Stock Corporation), which began operating in August 1923 at Steglitzerstrasse 2, in central Berlin. Scherbius and Ritter sat on its board of directors.
The firm publicized its cipher machine—by now named the Enigma—as much as it could. It printed flyers and exhibited the Enigma at the 1923 congress of the International Postal Union. A number of articles about the machine appeared in German and foreign electrical and business publications. Many were illustrated with diagrams of rotors and photographs of the firm’s ponderous printing version of the Enigma—a 15-inch-high monster with knobs and handles on its right side that weighed more than 100 pounds. This was being tested by the Deutsche Reichspost. Another version worked directly from and to punched teletypewriter tape.
Gradually the simpler version that indicated its output by illuminating letters, the “Glow Lamp” Enigma, became the most widely known and, eventually, the only one produced by the firm. It was much more compact than the printing version, standing only 4½ inches high, 10 inches wide, and 10¾ inches deep, and it weighed only 15 pounds. At the front stood three rows of typewriter keys. Behind them lay the three rows of circular windows for the output letters. In back of these and to the right was a switch allowing the operator to choose battery or house current. On the left, the tops of four rotors and four toothed thumbwheels for setting them poked up through the closed lid of the machine. The lid also had little windows through which showed the letters on the rims of the rotors.
The mechanism incorporated three significant improvements by other people over the straightforward system described by Scherbius in his letter of 1918. Two came from Willi Korn, an engineer in Scherbius’s employ, and one from Paul Bernstein, a Berliner.
Korn designed rotors that were removable. Previously their order left to right was fixed, but now the operator could put them into the machine in any order. This made possible Bernstein’s improvement: a movable ring with indicator letters on it on each rotor. The ring rode the circumference of the rotor like a tire on a wheel; the ring could be turned to any position and locked in place with a pin. Previously, a particular indicator letter meant that the rotor was in a particular position; now the indicator letters bore no relation to the position of the rotor. The position of the alphabet ring on the rotor had to be known to the decipherer, so it became part of the key. In addition, Bernstein shifted from the rotor to the ring the notch or notches that caused the rotor to the left to move one space at a certain point or points in the rotor’s revolution. This disjoined the rotor moves from the rotor encipherment, throwing up a further obstacle to solution.
Finally, Korn converted the leftmost of the four rotors into a reflector. Although it was called a rotor, it did not turn. It had contacts only on one face, and it sent the current that had come from the three normal rotors back through them along a different path before it illuminated an output letter. The reflector was sometimes called a half rotor because its wiring went from one contact on the side facing the three main rotors to another contact on the same side; it consequently had only thirteen connections instead of the twenty-six of the main rotors. The current’s double traversing of the rotors meant that encipherment was like decipherment: if plaintext a became ciphertext X, plaintext x became ciphertext A. This reciprocity had the advantage of eliminating the need for any switch to shift from enciphering mode to deciphering and vice versa, thus precluding the error of enciphering a message in the deciphering mode. But it had the crypt-analytic disadvantage of yielding the knowledge of a second plaintext letter whenever a first was found. The double passage brought another advantage and disadvantage: it complicated the cryptosystem, but it meant that no letter could ever represent itself, a fact that might speed solutions by showing which possibilities could be rejected.
In 1924, the firm got the German post office to exchange Enigmaenciphered greetings with that year’s congress of the International Postal Union. Later a book on cipher machines by an Austrian criminologist, Dr. Siegfried Türkel, gave the Enigma extensive coverage, including a detailed description of the various models, many photographs, and praise from the Austrian cryptanalyst and author Colonel Andreas Figl. But, no more than any other cipher-machine inventor of the time who had dreamed of getting rich by selling protection for businessmen’s messages, no more than Alexander von Kryha or Edward H. Hebern or Arvid Damm, did Arthur Scherbius make money. By the end of 1924, his firm still had not paid dividends.
The situation, however, was changing. Behind the sandstone walls of the four-story headquarters of the Naval Command at Tirpitzufer 72–76, facing Berlin’s tree-lined Landwehr Canal, the cryptologic branch that had turned Scherbius down in 1918 was reconsidering the security of German naval communications. The reason was the shocking discovery that the British had been reading coded German naval messages for much of World War I.
The first clue came from the fiery builder of Britain’s Dreadnought navy, the ret
ired first sea lord, Admiral of the Fleet Sir John Fisher. In his Memories, published in 1919, he wrote:
The development of the wireless has been such that you can get the direction of one who speaks and go for him; so the German daren’t open his mouth. But if he does, of course, the message is in cypher; and it’s the elucidation of that cypher which is one of the crowning glories of the Admiralty work in the late war. In my time they never failed once in that elucidation.
Subsequent indications were even more specific. In 1923, the official history of the Royal Navy in the war revealed various instances when intercepts had given the British an advantage. At the same time a dramatic and authentic story drew the attention of all to Britain’s cryptanalysis.
In his best-selling The World Crisis, the first two volumes of which were also published in 1923, Winston Churchill, who had been the civilian head of the Royal Navy at the start of the war, revealed, in his flamboyant style and with some poetic license as to facts, the basis of Britain’s codebreaking successes:
At the beginning of September, 1914, the German light cruiser Magdeburg was wrecked in the Baltic. The body of a drowned German under-officer was picked up by the Russians a few hours later, and clasped in his bosom by arms rigid in death, were the cipher and signal books of the German Navy and the minutely squared maps of the North Sea and Heligoland Bight. On September 6 the Russian Naval Attaché came to see me. He had received a message from Petrograd telling him what had happened, and that the Russian Admiralty with the aid of the cipher and signal books had been able to decode portions at least of the German naval messages. The Russians felt that as the leading naval Power, the British Admiralty ought to have these books and charts. If we would send a vessel to Alexandrov, the Russian officers in charge of the books would bring them to England. We lost no time in sending a ship, and late on an October afternoon Prince Louis [of Battenberg, first sea lord, whose name was later changed to Mountbatten] and I received from the hands of our loyal allies these sea-stained precious documents.
Churchill followed this with some colorfully told stories of how solved German intercepts had enabled the British to fight better at sea. Soon a volume of the official German naval history acknowledged that “the German fleet command, whose radio messages were intercepted and deciphered by the English, played so to speak with open cards against the British command.”
Suddenly, the German navy saw that a mere change of codes was no longer enough. It needed to fundamentally transform its system of secret communications. It had to have a cryptosystem that would not give away any secrets even if captured. Perhaps a machine was the answer. The navy had been offered one half a dozen years before that promised security to messages whether or not the machine was in the hands of the enemy. The staff had rejected it as unsuitable, but now the navy saw things differently. It may have examined other cipher machines on the market, such as the wholly inadequate Kryha, but it turned back to Scherbius and began letting contracts.
By 1925, Chiffriermaschinen Aktien-Gesellschaft had started production of the first Enigma machines for the navy. They differed from the commercial model in several ways. The order of letters on the typewriter keyboard and on the illuminated panel was not the QWERTY of the commercial version but alphabetical. The rotor wiring was different. Though only three rotors were used in the machine at one time, five were supplied, providing a greater choice of keys and therefore greater security. Since the reflecting rotor could not be turned, only three toothed thumbwheels instead of four projected above the cover. Instead of twenty-six contacts, the naval Enigma had twenty-nine, adding to the normal alphabet the three umlauted letters ä, ö, and ü, included because the codebook in which plaintext was to be encoded before encipherment by the Enigma had umlauted codewords.
This pre-encoding and the extra codewheels were only two of the ways in which the navy sought to increase the security of messages enciphered in its new machine. Another measure sought to preclude the navy’s chief security concern: espionage. The navy required that only officers, whose honor presumably immunized them against the blandishments of money and women, could set rotor positions.
Another major security measure was aimed at blocking the only method that any German cryptanalyst could then conceive of for solving Enigma messages. Called superimposition, it would require having thirty or so messages, of which portions had been enciphered with the same succession of rotor positions; with very heavy traffic, this might happen. To avoid an accumulation of overlapping texts, the navy prescribed rotor starting positions that were far apart. These were listed in a booklet. The enciphering clerk would choose one and communicate it to the deciphering clerk by an indicator—a group of letters. The indicator was itself enciphered, and the randomness of the prescribed rotor starting positions eliminated the possibility of a cipher clerk’s making up a starting position that was not random, such as XXX or LIL.
A final security measure assigned messages different grades of security—general, officer, staff—with successively more complex cryptosystems and keys held by fewer people.
By the start of 1926, all of these systems had been prepared and Scherbius’s firm had delivered enough Enigmas for the navy to put the machine into service as its Funkschlüssel C (Radio Cipher C). The twenty-three-page manual for it, dated February 9, 1926, covered, in addition to a description of the machine and the method of enciphering and deciphering, such matters as how to test the bulbs and how to deal with ciphering errors.
The navy’s satisfactory experience with the Enigma during its first year became known to the army’s Chiffrierstelle, or Cipher Center. The officer in charge in 1926 and 1927 was Major Rudolf Schmidt, a World War I signals officer who had written the chapter on communications for a major study of the war. He and his cryptologists saw the merits of the Enigma. They made some changes to suit it better for army practice: twenty-six-contact rotors, only three rotors (perhaps to have less to carry in mobile warfare), a standard QWERTY keyboard, and a system of message keys that required no booklet, only a set of keys that enabled the cipher operator to make up a different key for each message. On July 15, 1928, the Enigma went into the army’s service.
That year a single Enigma cost 600 reichsmarks, or $144 ($900 in 1991 dollars); volume purchases may have reduced this price. But the firm’s sales remained low. A few machines were sold to businesses, but the commercial market never materialized (nor did it for other cipher-machine makers). By the end of the decade the navy had bought no more than a couple of hundred machines, and the army about as many. Still, it was a start.
Then, one spring day in 1929, the team of a horse-drawn wagon that Scherbius was driving at his factory shied and smashed the wagon against a wall. Scherbius suffered severe internal injuries. On May 13, he died, only fifty years old. But his business survived.
By the mid-1930s the firm was manufacturing a variety of cryptographic machines. The army experimented with an eight-rotor printing version for a while. The most important change had come in 1930 with the army’s addition of a plugboard on the front of its machine. This consisted of a plate with twenty-six sockets, each representing a letter, that could be connected with one another by short cables with jacks on the end. The sockets were connected by wires to the keyboard and to the lamps, so that the enciphering and deciphering current passed through the plugboard. It added an extra substitution that overlay the rotor substitution. If on the plugboard the C socket and the R socket were joined by a cable and if without the plugboard the cipher letter for a plaintext e was C, the plugboard would convert the C to R. If the plugboardless cipher letter was R, the plugboard would replace this with C. The army connected only six pairs of letters, meaning that twelve letters were enciphered through the plugboard, the others being enciphered only with the rotors. But even twelve encipherments increased the number of keys—and so, theoretically, the number of trials a cryptanalyst would have to make—by billions. The plugboard was an excellent improvement.
In 1935, Hitler den
ounced the Versailles treaty and began his enormous expansion of Germany’s armed forces. They needed cipher machines, and they continued to buy Enigmas. Other agencies also purchased them: the railroad administration, the Abwehr (the military espionage service), and the Sicherheitsdienst, or SD (the Nazi party intelligence service).
During those rearmament years, both the army and the navy continually improved the Enigma and developed their systems of secret communication.
The navy alertly scanned the cryptologic horizons for new ideas. In the summer of 1930, for example, its cryptographers reported on a cipher machine devised by one Dr. Ruckhaber. “In its mechanical construction the method resembles in many points the not very successful Kryha system,” they wrote. Its mechanism slipped or jumped and caused many enciphering errors. Its output letters were harder to read than those of an illuminating system. Changing its setting took longer than changing the Enigma’s. It appeared easy to solve. The navy turned it down.
The Reichsmarine (its name was changed in 1935 to Kriegsmarine) developed its own cryptosystems, mostly for specialized uses. Some naval attachés held Schlüssel A (Code A), a code with a numerical superencipherment. The Werftschlüssel (Dockyard Cipher), a pencil-and-paper system, served shipyards and small ships. Early in 1939 the navy reworked and reissued the Funkschlüssel H (Radio Cipher H), which enciphered in pairs the letters of the nonsecret International Radio Telegraph Code. One edition, of 1,400 books in gray binding, served the merchant marine (Handelsmarine); the other, 800 in red, was for warships and naval posts. Shortly after war broke out, the navy prepared a Wetterkurzschlüssel (Short Weather Cipher) to abbreviate weather information so it could be transmitted “in the shortest possible time.”