From some garrulous German U-boat POWs and the recovery of Prien’s unexploded torpedoes in Scapa Flow, Churchill and the Admiralty learned almost all there was to know about the organization and size of the U-boat arm, the characteristics, armament, and limitations of the three main classes of boats (Types II, VII, IX), the usual patrol routes and zones, the torpedo and boat defects, and the U-boat production rates. This wealth of information, together with the sharp decline in merchant-shipping losses from October to December 1939, and an absurdly optimistic estimate of U-boat kills, convinced Churchill and Pound—indeed, the entire Admiralty—that the U-boat menace had been checked, at least for the present.
But Churchill continued to fret about the future, the several hundred U-boats certain to appear in due course to savage British shipping. He therefore continued to push for ever larger numbers of convoy escorts, including “jeep” carriers, ASW surface vessels such as the Hunt-class small destroyers, and aircraft, as well as new merchant ships to be built at home and abroad, and to encourage R&D for improved sonar and radar, ASW bombs, and depth charges.
Of the British antisubmarine R&D programs, the three most technically difficult but promising—and urgent—were minesweeping, radar, and codebreaking. By happy coincidence for the British, all three programs began to pay the first dividends in the early weeks of 1940.
• MINESWEEPING. Immediately after the recovery of the misdropped German magnetic mine at Shoeburyness, a team of British scientists, engineers, and mine technicians commenced a crash program to develop ways to “sweep” (i.e., explode) these mines.
The team pursued several approaches, but that which proved most practical and efficient was called the “Double Longitude,” or “Double L,” method. Two wooden minesweepers, sailing on a parallel course about 300 yards apart, each dragged two buoyant electrical cables, one long, one short. One cable constituted a negative electric pole, the other a positive pole. When the pairs of cables of both ships were energized in precise synchronization by five-second pulses of DC current from shipboard batteries, the salt water completed the electrical circuit, creating an intense magnetic field almost ten acres in size.
The first full sea-trial of the Double L system was conducted on the day after Christmas, merely thirty-six days after the recovery of the German mine. The trial was not completely satisfactory, but by the end of January the defects in the system had been corrected. During February 1940, a total of seventy-four sweeps were performed, resulting in the explosion of dozens of magnetic mines and the clearing of numerous fouled areas. Developed in about ninety days, the British Double L method thoroughly defeated the first generation of German magnetic mines.
Parallel with the “sweeping,” the British initiated procedures for “degaussing” (neutralizing the magnetic fields) of ships to reduce their vulnerability to both magnetic mines and magnetic pistols in torpedoes. At first this was done by clamping a big, permanent, heavy electrical cable around the ship’s hull and continuously energizing the cable with shipboard DC current. Later it was discovered that the same effect could be achieved by laying the energized cables on the ship’s deck, or—better yet—placing them in steel tubes inside the ship’s hull. Still later, it was discovered that a ship could be satisfactorily degaussed for about three months by passing a very powerful charged electric cable along the length of the hull while the ship was still in port. Although this degaussing technique (called “wiping”) was temporary, it was preferred to all others because the permanent cables on shipboard were a nuisance.
• RADAR. The main British radar R&D was aimed at improving the reliability and accuracy of the extensive Chain Home network to give early warning of Luftwaffe bombers. However, work continued on miniaturized radar sets, which could be fitted in aircraft and small surface ships, such as convoy escorts. By the close of 1939, an Air Ministry research team, led by Edward G. (“Taffy”) Bowen, had made some progress toward miniaturized radar for aircraft (the ASV for U-boat detection and A-I for night bomber interception), but field tests of these laboratory-built sets merely served to emphasize that a scientific breakthrough of some kind was required.
The Air Ministry’s Robert Watson-Watt, who was supervising all British radar development, concluded that because of limitations imposed by the basic laws of physics, the avenues of miniaturization being pursued by Bowen’s team and others would never bear practical fruit. He therefore recommended that a separate group of scientists tackle the problem independently of Bowen, with the aim of finding a solution along theretofore untried avenues. For this purpose, Watson-Watt created a new committee, chaired by the scientist Fredrick’ Brundrett. Seeking a fresh research team, Brundrett gave the task to the physics department at the University of Birmingham, headed by the Australian—and Cambridge graduate—Mark Oliphant.
Oliphant, with a senior assistant, John Randall, and a graduate student, Henry Boot, approached the assignment with a sense of urgency. Simply stated, the aim was to invent from scratch an entirely new “electronic valve” that was capable of generating high-frequency radio waves with sufficient power to find objects as small as bombers and submarines. That is, a powerful electronic “booster” of a new kind which could sharply focus radio waves.
The initial research of Randall and Boot led them to some obscure scientific papers published by an American physicist, Albert W. Hull. While working at the General Electric Research Laboratory in Schenectady, New York, to develop an alternative to the radio vacuum tube in the 1920s and 1930s, Hull had invented what he called the magnetron, in which the flow of electrons was controlled magnetically, rather than electrostatically. For various reasons, General Electric did not pursue the magnetron commercially, but Hull continued to work on the physics of it for the next ten years, publishing several papers that were of no interest to anyone outside of the tiny handful of specialists in that arcane science.
The magnetron per se was not the answer Randall and Boot sought, but a cursory study of Hull’s papers led to an idea that was to make the practical miniaturization of radar possible and open the way to its installation on aircraft and ships.
The idea, called a “cavity magnetron,” is quite simple to a scientist but difficult for the layman to understand. Radar historian David Fisher explained it this way:
The basic concept of the magnetron is to use a magnetic field to herd the electrons along, to guide them where you want them to go. Randall and Boot combined this with a totally different concept—the concept of a penny whistle.
A whistle consists of a loose, hard object in a cavity into which you blow. The force of your blowing makes the object inside rattle around in the cavity, generating sound waves which reverberate inside and then escape. The frequency with which they bounce around inside governs the frequency of the sound emitted, and that frequency is in turn decided by the dimensions of the cavity, so that a large whistle emits a long-wave, low-pitched hoot while a tiny whistle gives out a high-pitched scream. Ergo, the cavity magnetron.
Randall and Boot constructed a small solid block of copper, which would conduct their variable magnetic field. In this they carefully scooped out a precisely measured cavity. When an electric current was made to flow through the copper and a magnetic field was applied, electrons were caught in the cavity—which served as an anode—and were made to bounce around in there by the magnetic field. As they bounced back and forth, resonating, they emitted electromagnetic waves just as the bouncing ball in a whistle emits sound waves. The size of the cavity was constructed so that the electrons moved only a few centimeters between bounces and therefore produced electromagnetic waves just a few centimeters long. It was brilliant and it worked.
Randall and Boot conducted the first test of the cavity magnetron on February 21, 1940. Having no idea what the power output would be, they hooked it up to a set of automobile headlights, hoping they might at least get a dim illumination. Such was the power output, it blew out the headlights and then the bigger headlights of a truck. Finally, they wired it up to a set of neon floodlights. These held, enabling Randall and Boot to measure the wavelength and power output. As expected, the wavelength was 9.8 centimeters (usually described as “ten centimeters”); the power output of this experimental little gadget was an awesome 400 watts, or nearly half a kilowatt, four times the power output of the existing airborne radar sets. Moreover, it was an easy step to increase the power a hundredfold. A prototype, generating 50 kilowatts, picked up the periscope of a submerged submarine at a range of more than seven miles!
The cavity magnetron, which made possible practical radar miniaturization, was one of the greatest scientific breakthroughs of World War II—the greatest in general utility. The British author Brian Johnson aptly wrote in 1978: “It is impossible to exaggerate the importance of Randall and Boot’s work. It lifted radar from an electronic stone age to the present day.”
• CODEBREAKING. The German Enigma encoding machine, employed by all military and paramilitary organizations in the Third Reich, was conceived by a Dutchman, Hugo Alexander Koch, in the years immediately following World War I. The patents for it were acquired by a German engineer, Arthur Scherbius, who attempted, unsuccessfully, to market it in the 1920s. Subsequently, two German firms (Heimsoth & Rinke and Konski & Kroger) obtained the rights and marketed the machine for commercial use (cost: $144 plus postage). By 1930 military codebreakers of all major nations had acquired the “commercial” version of Enigma for study. All professional codebreakers understood Enigma’s basic principles.
The Enigma was a small, portable, battery-operated machine about the size of a typewriter. It was compactly housed in a varnished oak box with a hinged lid. It looked something like a typewriter. It had a three-tiered typewriter keyboard of twenty-six letters—but no numerals or punctuation marks. In place of the platen and typewriter keys, there was a flat panel with the twenty-six letters of the alphabet repeated in the same order. Each of the letters on the panel had a tiny light bulb behind it. To encode a message, the sender pecked out the message, one letter at a time, on the keyboard. The machine automatically scrambled—or enciphered—that letter into another that appeared, lighted up, on the panel. In a reverse process, the receiver, operating an identical Enigma, pecked out the encoded message on his keyboard, one letter at a time, and the deciphered letters lit up on the panel.
The scrambling—or encoding—mechanisms inside the machine were fiendishly clever. The basic idea was to route the electrical impulses from the keyboard to the light panel by as devious and complicated a route as possible. The heart of the mixing system was a row of three turning drums or rotors, about three inches in diameter. Each rotor was fitted on both sides with twenty-six electrical contact points, which interconnected with the other rotors by means of spring-loaded and flush contact points. When the operator struck a letter on the keyboard, the first, or right-hand, rotor revolved one notch (or l/26th of a revolution), like an automobile odometer. With this movement the setup of the contact points with the other rotors completely changed, routing the electrical impulse on entirely new entry and exit paths. The second letter struck moved the first rotor another notch (or l/26th of a revolution), creating yet another entry and exit path for the impulses. After twenty-six letters had been struck, the second, or middle, rotor clicked in, extending the process with its twenty-six contact points. Then, finally, the third, or left-hand, rotor with its twenty-six contact points clicked in. Altogether the three rotors had 17,576 positions (26 × 26 × 26). Another complicated feature, known as a “reflector,” bounced all the electrical impulses back through the maze of rotor contact points, further scrambling them (in effect, creating the equivalent of six rotors) before they lit up the letter bulbs.
Nor was that all. The rotors were removable. They could be rearranged on the shaft or axle in six different combinations (1-2-3, 1-3-2, 2-3-1, etc.). These variables increased the possible rotor positions to 105,456 (17,576 × 6). Moreover, the rotors were fitted with ratchets on the outer rims, so that they could be set individually to start revolving from any one of the twenty-six different positions. The twenty-six optional rim-settings for each rotor, multiplied by the total 105,456 possible rotor positions, increased the possible electrical contact points available to astronomical levels.
Encoding and decoding messages on identical Enigmas was a fairly simple process for the operators, but it had to be done exactly according to a prearranged and distributed menu, which established what came to be called the “keys.” First, both sender and receiver had to insert the three rotors on the axles in the identical left-to-right order (1-2-3 or 1-3-2, etc.). Second, both had to set the rotors to identical rim numbers. Third, both had to turn the three rotors until identical letters embossed on the outer faces of the rotor appeared in tiny peepholes above the rotors. Assuming all three steps had been done correctly on identically wired Enigmas according to the prearranged “keys,” the encoder and decoder were then ready to communicate.
Sitting before his typewriter keyboard, the sender commenced the automatic encoding process. As he pecked out the letters of the message, one at a time, the electrical impulses traveled through the maze of circuits, switches, and contact points, automatically scrambling or encoding the letters and lighting the bulbs behind the encoded letters on the panel. (“A” might light up as “R,” for example.) An assistant copied down the encoded letters as they lit up on the panel. When the message was completely encoded, the sender gave it to a radio operator, who telegraphed it to the receiver in five-letter groups. In a reverse process, the receiver typed out the encoded message, one letter at a time on his keyboard. The machine automatically unscrambled the encoded letters, one at a time, lighting up letters on his panel in plain text. (The encoded “R” in the example would light up on the receiving machine as an “A.”)
Cryptologists quite rightly believed at first that messages properly encrypted by Enigma were unbreakable. The total of the permutations offered as a result of the many variables was mind-boggling: six thousand trillion by one calculation. Such huge possibilities ruled out known codebreaking techniques, such as statistical analysis (e.g., letter frequency), and appeared to defy solutions by higher mathematics. Possession of an Enigma wired exactly like that of the enemy was only half the battle. One also had to know the three easily changeable “keys”: left-to-right rotor order, rotor-rim settings, and rotor-peephole settings.
The German military was enchanted with Enigma. It was compact, easy to operate, rugged, cheap, and seemingly foolproof. Even if the enemy captured an Enigma, it would not do great harm; the keys could be changed at will. Accordingly, it was adopted by the Reichsmarine in 1926 and the Reichswehr in 1929. To increase its security beyond the “commercial” Enigma available to all, the Germans altered the wiring of the “military” version and added yet another layer of encryption: a “plugboard” on the front below the keyboard. This consisted of twenty-six holes, lettered A to Z (or, alternately, 1 to 26). When one (or more) combination of these letters was paired by cables, plugged in like an old-fashioned telephone switchboard, it rerouted the electrical impulses through yet another maze, raising encryption possibilities to figures almost beyond mathematical reckoning. To break this military version thus required knowledge not only of the three rotor keys but also of the plugboard cable arrangement.
Enigma was wondrously flexible. It enabled various Third Reich organizations to establish individual and completely different encoding setups. This was achieved in a simple way by distributing different rotor and plugboard keys (to be in force for a specific and limited time) or, in a more complicated way, by altering the internal wiring schemes, or by adding one or more rotors. Thus it was to transpire that the Enigma of the Reichsmarine and Reichswehr and the Luftwaffe evolved along different paths and none could read the other’s transmissions without obtaining identically wired machines and the keys.
A team of Polish codebreakers, led by Marian Rejewski and including Jerzy Rózycki and Henryk Zygalski, commenced an attack on German military Enigma in December 1932. Their tools were not inconsiderable: a “commercial” Enigma, acquired earlier; espionage materials (old keys, plugboard settings, etc.), obtained from a money-hungry German traitor, Hans-Thilo Schmidt (codenamed “Asche”), cultivated and exploited by the chief of the French codebreakers, Gustave Bertrand; and a wealth of Enigma messages intercepted by Polish stations.
The Poles were soon thoroughly familiar with German procedure for transmitting Enigma. They noted with special interest one procedural feature. After having set the four keys of their Enigmas according to the predistributed menu, the sender and receiver then engaged in yet another procedural step designed to enhance Enigma security. The sender embedded at the head of the message a three-letter encrypted key, selected at random, for that message only and repeated it twice to be certain the receiver got it, even in poor radio-transmission conditions.
The Germans believed that by instructing the sender to choose the three-letter message key at random, another level of security had been added to Enigma. But they were wrong. German Enigma operators were human and therefore predictable. At first all too many operators lazily—and predictably—chose AAA or ABC or a three-letter diagonal on the Enigma keyboard, such as QSC or ESY. Moreover, the standing instructions to repeat the message key (to insure reception in poor conditions) revealed to codebreakers vital information: the first and fourth, second and fifth, and third and sixth letters at the head of every message (the message key repeated) would always be identical. This known constant, together with the fact that no encrypted Enigma letter ever replicated itself (“R” never lit up as “R”), made Enigma vulnerable to penetration by certain very complicated higher mathematical processes.
Enigma was also vulnerable in another way. German military traffic was rigid and stylized, composed of military addresses, titles, salutations, and words often repeated. A close student of these messages could sometimes correctly guess a word or a phrase (“division,” “regiment,” “operation,” or “nothing to report”). Correct guesses from these official messages, or from idle “test” chitchat between operators, or from lapses were known in the codebreaking trade as “cribs.” In the early days, before the Germans became skilled at communications security, Enigma yielded numerous cribs.
Astonishingly, the Polish team broke German Enigma in a few weeks and were able to duplicate the machines without ever having seen the German version. Although the Germans tightened security procedures and made refinements to Enigma, for the next six years—to 1938—the Poles broke German Enigma at will and read it consistently and currently. To simplify and speed up the breaks, they developed a “cyclometer” (two sets of Enigma rotors, linked in a certain way) and an immense card file, listing possible keys and other data. In a two-week “test” in January 1938, the ten-man Polish team decoded 75 percent of German Enigma messages received and calculated that with more personnel it might have decoded 90 percent.
During the Munich Crisis of September 1938, and again in December 1938, the Germans delivered the Polish team two stunning setbacks. On September 15, the Germans abolished the procedure of distributing preset peephole settings for starting the message transmissions. Instead, Enigma operators were instructed to choose any three letters at random for the initial peephole settings and to transmit these unenciphered, or in the clear, to the receiver before transmitting the enciphered and repeated three-letter settings for the message itself. Beyond that, on December 15, the Germans distributed two additional rotors to all Enigma operators, making a total of five rotors from which to select three for insertion in the machine. The change in peephole-setting procedure—the random selection of the start position—rendered all the decoding work of the Poles to that time useless. The addition of two more rotors raised the encription possibilities, mathematically, to yet more mind-boggling levels.
The Poles were dismayed, but not discouraged. To attack the increased complexities introduced by the procedural change, they conceived two techniques. One, offered by Rejewski, was an ingenious automated machine, a sort of super-cyclometer, which the Poles called a “bomba.”* Its heart was comprised of three linked rotor sets from six Enigmas (eighteen rotors). When prompted in a certain way, the machine “tried out” various combinations of encrypted letters until the embedded message code was found. A total of six bombas (108 rotors) was required to conduct a full range of searching. The other technique, offered by Rejewski’s associate Zygalski, made use of large, perforated, heavy paper sheets with horizontal and vertical columns of letters, which when laid one atop the other on a light table in a certain way revealed the message keys by admitting light through the holes. Each sheet had about 1,000 holes. A total of 156 separate sheets (six series of twenty-six sheets), comprising about 156,000 holes, each hand-cut with a razor blade, was required for a full range of search on a three-rotor Enigma.
The addition of the two extra rotors in December 1938 posed even greater problems for the Poles. First they had to figure out the wiring arrangement of each of the new rotors. They were able to do so because of another careless communication lapse on the part of the Germans. The Nazi S.S. and S.D. nets adopted the five-rotor option but, inexplicably and foolishly, retained the old system of predistributed initial peephole settings, which the Poles had mastered earlier. By exploiting that lapse and by intuition, the Poles, in yet another astounding crypt-analysis achievement, were able to replicate the wiring of the new rotors. By January 1, 1939, the Poles could read five-rotor S.S. and S.D. traffic accurately and consistently.
However, the solving of the five-rotor military Enigma employing random initial peephole settings, as well as random-encrypted peephole settings for individual messages, defeated the Poles. Rejewski calculated that to search by automated means, each of the six bombas in operation would have to be fitted with thirty-six of the two new rotors (for a grand total of 1,080 rotors) and be operated twenty-four hours a day. Alternately, the laborious perforated-sheet method would require 1,560 different sheets (sixty series of twenty-six sheets) with a grand total of 1,560,000 hand-cut holes. It could be done, of course, but the Poles lacked the resources for so massive an undertaking.
That was where matters stood in August 1939, a few days before the outbreak of war, when the Poles turned over to the British and French copies of all their research materials and the Polish-built German military Enigmas.
The British codebreaking unit, GCHQ, located at Bletchley Park, was directed and staffed by many Veterans of the Admiralty’s famous Room 40 of World War I. These included the director, Alastair Denniston, his deputy Edward Travis, and the codebreakers A. Dillwyn Knox, William F. Clarke, and others. All were quite familiar with Enigma principles—the British had bought a “commercial” Enigma in the 1920s—but they had made only cursory attempts to break German Enigma and had had no success. The Polish “gift” was therefore immensely valuable and saved the British many months of tedious work.
Before the war, GCHQ had recruited and vetted a “reserve” force, drawing heavily on academics at Cambridge and Oxford universities. When war came, the reserve was called up to Bletchley Park and divided between the codebreaking and the specialized army, air force, and naval sections, which were located in temporary one-story outbuildings misnamed “huts.” Among the first call-ups were three brilliant Cambridge mathematicians: Gordon Welchman, Alan Turing, and John R. F. Jeffreys.
These three mathematicians were assigned different tasks. Alan Turing, who was fascinated by machines, set about designing a bomba which was to be more powerful and “general” in nature than the Polish one and theoretically capable of coping with whatever new complexities the Germans might add to Enigma. John Jeffreys was put in charge of manufacturing the huge numbers of perforated sheets required to solve five-rotor Enigma. Gordon Welchman was directed to study German call signs and to identify various Enigma nets.
The intellectual challenge presented by Enigma enthralled Gordon Welchman. Although GCHQ was strictly compartmentalized for security reasons, Welchman refused to be limited to a study of call signs. In a flash of brilliance, entirely on his own, he reinvented the perforated-sheet method and presented it to Dillwyn Knox, only to learn that it was already being vigorously pursued. Again straying from his limited area, Welchman offered a suggestion for Turing’s bomba. This idea, as Turing’s biographer put it, was so “spectacular” that Turing was “incredulous.” Incorporated into Turing’s design, the suggestion gave Turing’s proposed machine “an almost uncanny elegance and power.”
By December 1939, Jeffreys and his aides had completed two sets of perforated sheets, each set containing about one and a half million punched holes. One set was sent to Paris for the French and Polish codebreakers. Employing the British-made sheets, the Poles found the daily keys of a five-rotor Enigma net for October 28 and broke the net (called “Green” by the Allies) messages for that day only. This was the first Allied break into the five-rotor Enigma. The excitement that ensued was tempered by the fear that on January 1, 1940, the Germans would make changes in the keys or procedures, which would negate this achievement.
But the Germans made no changes on January 1. As a result, the codebreakers at Bletchley Park, employing the sheet-stacking method, broke into the Luftwaffe Enigma net (called “Red” by the Allies) of January 6. This was the first “all-British” codebreaking triumph over the Germans. Owing to the generally lax Enigma procedure and communications security in the Luftwaffe, which yielded cribs, and to cribs intuitively arrived at, thereafter British codebreakers read the Luftwaffe Red with fair reliability. The British also sporadically broke into the Wehrmacht Green and a Luftwaffe training code, Blue, and some other German nets that still employed three-rotor Enigma.
From January 1940 onward, the British at Bletchley Park played the dominant role in breaking Enigma. The Poles and French continued to make contributions, but the importance of their work diminished. Pending the building of the Turing bomba, incorporating Welchman’s “spectacular” improvement, the British relied on the sheet-stacking method. It was a tedious and work-intensive process, requiring ever greater numbers of clerks. Nor was it a sure thing. Often as not, the British failed to recover the daily keys.
No progress whatsoever was made in breaking Kriegsmarine traffic. Naval Enigma employed eight rotors rather than the five rotors of Luftwaffe and Wehrmacht Enigma, ruling out the sheet-stacking method. There was scant traffic to intercept—too little grist for the codebreakers’ mill. Kriegsmarine radio operators continued to exercise strict transmission discipline, offering no cribs. Attempts to break naval Enigma before Turing’s bomba was completed were bound to fail. And so they did.
To the technical achievements at Bletchley Park must be added another, no less important. That was the superb management of the information flow. The leaders centralized the gathering, storage, and distribution of all codebreaking intelligence, however trivial. They barred the “interservice rivalry,” the jealous withholding of bits and pieces to favor one military service or the other. All hands shared equally in all phases of the operation, from translators to analysts to librarians in the hugely growing data bank to the distributors of the information. And, moreover, they did so without a single leak, so far as was known in 1996.