The Secret War

by Max Hastings

  In the spring of 1943 an agreement was signed which became known as BRUSA, and was justly described as the closest intelligence-sharing pact in history between any two nations. Though not always immaculately observed on either side, its success was astonishing. In mid-August a scrambler was installed on the Washington–London phone line, enabling intelligence officers of the two nations to converse when an issue was important enough. A series of major conferences was held, at which British and American officers exchanged information and techniques.

  Cryptanalysis was a world of its own, using a language incomprehensible to those outside it, but American and British practitioners came to understand intimately each other’s doings and difficulties. In October 1943, for instance, the JIC in London noted that US forces signalled some weather reports in plain language, while the British encrypted the same information. This posed a danger that enemy codebreakers could exploit the match, and the Americans duly started to encrypt all meteorological information. Likewise a memorandum from the British SLU in Washington to Bletchley Park on 16 February 1945 offered practical advice of a kind which flew to and fro daily across the Atlantic in the latter part of the war: ‘noted ref JN-11 Ransuuban. For recovery of strip digits a straight additive attack has proved to be superior; with particular emphasis on the heavy use of Hatsu and related groups in the first position. Speed of recovery is dependent upon the condition of the code once the strips are identified and completed, recovery of the daily key on a complete day’s traffic proceeds rapidly by stripping.’

  It was agreed at an early stage that the Americans should major on Japanese traffic, while Bletchley Park maintained the lead role on German material, and trawled neutral states’ messages. An OSS officer who urged burgling safes in Vichy to secure its codes was politely informed that this was unnecessary – they had long since been given to the Allies by sympathetic French intelligence officers. London also gave a dusty answer in 1943 when Arlington Hall requested help with its discreet monitoring of some Soviet traffic: this would have breached Churchill’s prohibition on espionage against an ally. The British were always cautious, perhaps extravagantly so, about adopting any course of military or naval action that might betray the Ultra secret, while US forces were more willing to take the security risk involved in exploiting prospective targeting information.

  For the most part, however, the US displayed considerable sophistication in protecting Ultra. When a Japanese Purple signal was broken on 28 December 1942, requesting Spanish diplomats in Washington to retrieve on their behalf half a million dollars in cash left in their abandoned embassy safe, the Americans made no attempt to seize the money, in case the signal was a coat-trailing ruse to discover whether Tokyo’s traffic was being monitored. On 18 April 1943, after US fighters shot down the aircraft carrying Admiral Isoroku Yamamoto, the official American bulletin merely routinely reported the destruction of four Japanese planes in the North Solomons. No hint was given that the attackers knew the supreme importance of one passenger, an act of forbearance that proved decisive in lulling Japanese suspicions of a communications breach. It was left to Tokyo to announce Yamamoto’s death.

  In the spring of 1944 the Americans went even further to reassure the Japanese about their own security. On 1 April, two Imperial Japanese Navy flying-boats were damaged in a tropical storm, en route from Palau to Davao. One of them carried Admiral Mineichi Koga, commander-in-chief of the Combined Fleet. In the second was Vice-Admiral Shigeru Fukudome. When this plane ditched off Cebu island, Fukudome floundered ashore without his attaché case, containing Japanese codes and important strategy documents in plain language. Guerrillas on Cebu alerted the Americans, who got to the plane. A US submarine rushed the attaché case to the Australian Army’s intelligence department, where Fukodome’s codes and documents were photographed. Then the case was hastened back to the crash area for local people to hand over to the Japanese, claiming that they had chanced upon it. Fukudome himself eventually got home, to be forgiven and promoted. The Japanese navy never suspected that its haul of secrets had passed through American hands.

  There was a 1945 debate between Bletchley and Arlington about whether to commit Allied bombers in Europe to attack key German landline telephone exchanges, whose locations and importance Ultra had identified. An eventual decision was made to go ahead, but subject to consultation between the targeting officers and Bletchley’s Hut 6 – with a cut-out through SHAEF intelligence – before each such attack was launched. The security system within the armed forces protecting Ultra worked remarkably well: the most significant threats to it proved to be not Axis intelligence officers, but instead recklessly indiscreet American journalists.

  It would be wrong entirely to idealise the conduct of the vast Allied interception and codebreaking staffs, most of them very young men and women performing monotonous and repetitive tasks. In 1944–45 the British teleprinter intercept centre at Knockholt in Kent suffered severe problems from staff discontent: its six hundred staffers resented their low pay and poor working conditions; absenteeism rose sharply at weekends, and no manager bothered to explain to the girls why their work mattered. Meanwhile Lt. Ed Parks of the dissemination unit at Arlington Hall scrawled an earnest little note on 16 October 1944: ‘In the last few days our work has tailed off considerably … We turned out a total of 15,739 messages in the 15 days … But now that the work is not too pressing, we should make a special effort to keep all the messages going through promptly. I noticed last week a tendency to leave early, and to play around when there were messages waiting to be typed. I think the impression given is bad … It seems to me essential to remember that the work we are doing is of vital importance to the conduct of the war … [and] deserves our best efforts.’

  From late 1942 onwards, the British and Americans were processing enemy decrypts in industrial quantities, though inter-service rivalries and jealousies continued to hamper US activity. The ambitious and expensive bombes designed and built for the US Army were a failure, but the navy models attained higher technical standards than the British ones, and were produced in numbers unimaginable in Churchill’s straitened island: for a period National Cash Register of Dayton, Ohio, was completing three or four each day. Op-20-G bombes, twice the size of their British counterparts, began to achieve operational effectiveness only in August 1943, but by April the following year eighty-seven were in service, which proved subject to fewer technical failures than their Bletchley counterparts. Thus, they took over from Hut 8 an increasing share of the burden of reading Kriegsmarine traffic. Bletchley Park’s staff held Op-20-G and its cryptanalysts in the highest respect.

  Until 17 September 1941, when GC&CS made its first significant breaks into Wehrmacht traffic, most Ultra information came from the army-Luftwaffe keys, and from reading traffic dispatched through the Italians’ Hagelin-C38m machines. Between December 1941 and May 1942, some thirty to forty Ultra messages a day were transmitted to overseas commands by Hut 3. Thereafter, however, there was a rapid increase – to eighty a day between June and October, then to a hundred by April 1943, a level which it sustained thereafter, to achieve a cumulative wartime total of 100,000 decrypts circulated to operational headquarters. This, in turn, was only the small proportion deemed useful or relevant to commanders, among 90,000 decrypts a month processed at Bletchley in 1944–45.

  The physical and mental strain on the codebreakers never eased. While machines aided their labours, the foremost weapons in the battles waged by Bletchley, Op-20-G and Arlington Hall were always human brains. Reading enemy signals seldom became an easy or routine task – during the last two years of the war, Stuart Milner-Barry said it was impossible to aspire to read more than half of all incoming German intercepts, even under the most favourable circumstances. ‘There was a perpetual excitement about each day’s breaks, at whatever time of day or night they might come. To the chess-player’ – he himself had represented England before the war – ‘it was rather like a long-running tournament with several rounds bei
ng played each day, and never any certainty that the luck would continue to hold.’

  Efforts to read the Italian navy’s higher traffic, encrypted through book codes, were abandoned as a failure, a decision influenced in part by the fact that Italian warships had ceased to be a threat to Allied operations. Many of those engaged in codebreaking were obliged to take sick leave at intervals to obtain a respite. In July 1943 John Tiltman of Bletchley and William Friedman of Arlington Hall exchanged letters, in which beyond discussing the difficulties of breaking Japanese army traffic, both acknowledged the stresses of their role. There were never enough qualified staff available on either side of the Atlantic. Friedman mentioned that Commander Joseph Wenger, now chief of Op-20-G, had missed a scheduled meeting with him because ‘he was out at the Naval hospital. I suspect that he will have to take a rest for several weeks. As a matter of fact, I had been wondering how long he would be able to stand up under the strain … for a number of months now I have watched him go down noticeably.’ Never should it be thought that freedom from physical danger provided the codebreakers with a passport to a ‘cushy’ war.

  As for the impact of this huge effort on the battlefield, until the summer of 1942 the British Eighth Army’s commanders-in-chief and their staffs in North Africa were sceptical, even contemptuous, about intelligence. They could assert in self-justification how patchy was the service they received – for instance, Bletchley warned that Rommel would launch an attack in May, but did not give any clue where. Both the Park and senior officers in the Middle East were slow to grasp the scale of the Afrika Korps’ logistical problems. The appointment of Gen. Sir Bernard Montgomery as C-in-C in June coincided with an important increase in the Ultra flow. A Luftwaffe signal, for instance, stated that Axis day-bombing of Malta was to be abandoned. The theatre air intelligence chief, Group-Captain Harry Humphreys, immediately recognised this meant the Germans would transfer their Messerschmitt Bf109 fighters from escort duties to North Africa, and thus felt able to get Spitfires shifted from defending Malta to securing air supremacy over Egypt and Libya. Breaks in Italian traffic empowered the Royal Navy and the RAF to devastate the Germans’ Mediterranean supply line, sinking forty-seven ships totalling 169,000 tons between July and October, while German messages famously warned the British of the impending Axis thrust at Alam Halfa in August. This was among the most important intelligence breakthroughs of the war, and enabled Montgomery to achieve his first victory.

  Before the British launched their own Alamein offensive on 23 October 1942, on the 7th Gen. Georg Stumme, Rommel’s deputy, told his officers that the main axis of the forthcoming British attack would be between Ruweisat and Himeimat, confirmed on 20 October as ‘the northern part of our southern sector’, together with an advance along the coast road. This view was strongly influenced by elaborate British deception operations in the south, and Montgomery’s officers thereafter congratulated themselves on the cleverness of their scheme involving dummy tanks and pipelines; but it deserves notice that the Germans were able to switch 21st Panzer Division to the north three days later, before the British achieved their breakthrough. Deceptions were only serviceable until an attacker showed his real hand.

  After Montgomery’s victory his critics – prominently including Ralph Bennett of Bletchley’s Hut 3 – argued that the sluggish British pursuit was inexcusable when decrypts flagged in real time almost every movement the retreating Germans made. Bennett wrote of ‘the fierce indignation and dismay felt throughout the Hut at Montgomery’s painfully slow advance from Alamein to Tripoli, incomprehensible in the light of the mass of Ultra intelligence showing that throughout his retreat Rommel was too weak to withstand serious pressure … [Montgomery’s] delay seemed to cast doubt on the whole point of our work.’ From the beginning of 1943 onwards, matters got better: Eighth Army’s commander and his staff developed complete confidence in Ultra, and became more proficient about exploiting it on the battlefield.

  For months after the Americans arrived in North Africa in November 1942, they made the same mistakes that the British had done a year or two earlier. One of Rommel’s intelligence officers paid tribute to the value of German voice-monitoring of US Army channels: ‘they were still happy-go-lucky and careless of their signals procedure’. As for Ultra, a young British intelligence officer who visited Allied headquarters in Algiers lamented that Eisenhower’s staff ‘did not know what they ought to be doing and had learned a whole lot of wrong things they ought not to have been doing’. Bletchley’s input was scrappy in the early part of the Tunisian campaign, but more skilled and experienced intelligence officers in Algiers could have anticipated Rommel’s punishing February 1943 assault on the Kasserine Pass. The sacking of Brigadier Eric Mockler-Ferryman, Eisenhower’s British intelligence chief, was a just penalty for his failure. Thereafter in North Africa, Sicily and Italy, the Allies usually had an extensive knowledge of the Germans’ strengths and deployments.

  Bletchley Park became marginally less uncomfortable in 1942, when four steel and concrete blocks, designated ‘A’ to ‘D’, replaced some of the wooden hutments. A pneumatic tube system used in London department stores was introduced for shifting messages between sections, replacing the earlier tray-and-pulley operation – this was a brainchild of Hugh Alexander, who in his earlier life had been head of research for the John Lewis store chain. Alexander supplanted Turing as head of Hut 8, not because the latter was the victim of any persecution or palace coup, but because he was too disorganised a human being to administer anything. It was recognised that his astounding intellect was best left to roam free.

  Ralph Bennett has painted a vivid portrait of the daily routine of the team in Hut 3, which kept its name even when moved into a brick building. It was its inmates’ function to translate and render coherent broken or part-broken signals, with increasing input from officers of the US Special Branch who were now attached to several Bletchley sections. The codebreakers knew almost nothing about Allied operations – the context of campaigns – and thus saw the war through a peculiarly narrow, enemy prism. ‘We knew much more about most German divisions and some German generals,’ wrote Bennett, ‘than we did about any on our own side … [Rommel’s] 90th Light Division became so daily an acquaintance during the African campaign that there was even a sort of temptation to rejoice when it scored a success.’ He also emphasised an important constraint on Ultra’s practical utility: ‘No message had more authority than that of the officer who sent it, nor more reliability as a guide to his superiors’ intentions than the extent of the knowledge they allowed him to have.’ Again and again – for instance when decrypts caused the Allies to expect a German evacuation of southern Italy in September 1943, when instead Kesselring stood and fought – regular access to enemy traffic ended in a misreading of Hitler’s intentions, often because he changed his mind.

  In the second half of the war, however, the Allies could plan most of their operations with remarkable confidence that the enemy had no unpleasant surprises in store. Over three hundred Allied ships passed through the Straits of Gibraltar in the thirty-three hours before the ‘Torch’ landings, knowing that they were most unlikely to face enemy air interference. The January 1944 Anzio landings were launched with the assurance that Kesselring had no hint of them, and Ultra also flagged his big February counterattack against the beachhead, which belatedly convinced the US Army’s Gen. Mark Clark that Ultra was a reliable source of intelligence. The Allies knew before the August 1944 ‘Anvil’ invasion of the south of France that Hitler anyway intended to evacuate the region, which caused Gen. Sir Harold Alexander, commanding in Italy, to urge that it would make more sense to leave the invasion force fighting with his own army. Once the Americans were ashore in southern France, they could pursue the retreating Germans with unusual exuberance, because they knew they need fear no counterattack. Bletchley was breaking a substantial portion of traffic transmitted in some fifty different Wehrmacht and Luftwaffe Enigma keys, and was forwarding the fruits to forty subs
cribers in Allied headquarters everywhere that the Germans were being fought.

  Ultra never told all, however. A popular modern delusion holds sway that GC&CS, through the agency of Turing’s bombes and those of the US Navy, made the Allies privy to the enemy’s communications throughout the war, and that all German messages of significance were transmitted by Enigma. None of this is so. The Wehrmacht’s Enigma traffic posed ongoing difficulties, and until a late stage of the war Bletchley remained vulnerable to delays and blackouts in its decryption. In September 1944, Hut 6 solved only 15 per cent of army messages; in October 18 per cent; in November 24 per cent. By contrast, in September 64 per cent of Luftwaffe intercepts were read, and 77 per cent in October and November. Many decodes of all kinds were achieved too slowly to influence events on the battlefield.

  Moreover, from 1941 onwards the German high command transmitted an ever-growing proportion of its most sensitive traffic by teleprinter, of which there were several models. The most widely used were the Lorenz Schlüsselzusatz SZ40/42 ‘Tunny’, in Bletchley parlance, and the Siemens & Halske Geheimschreiber T-52 – ‘Sturgeon’. These systems worked on-line, in contrast to Enigma’s off-line operation, and employed the so-called Vernam cipher, a non-Morse language. When British interceptors recorded its incomprehensible stutter across the ether, in August 1941 a Bletchley team led by Col. John Tiltman began to probe its significance. They managed to break a single message from Athens to Vienna for which an obliging German signaller repeated a corrupt text, but this took them little nearer to reading the traffic. Tiltman, a decorated First World War soldier, was a cryptographic veteran who showed that not all the Park’s wizards were civilian eggheads. He was an unusual colonel: when a newly arrived private soldier recruited to Tiltman’s section stamped to attention before him and saluted, the officer set the tone for their subsequent association by his pained response: ‘I say, old boy, must you wear those damned boots?’ Thereafter the young man adopted plimsolls.