CHAPTER 10

Australian Signals Intelligence, 1914–29

By 1914, Britain was well versed in the utility of SIGINT, and, as we have seen, the RAN, as part of the Admiralty’s worldwide intelligence system, played a significant role in the SIGINT successes in 1914. Following on from this were the successes of the Admiralty’s Room 40, Old Building, and the other groups exploiting transmissions by radio or cable. These groups, and particularly Room 40, quickly grew from adjuncts to naval or Military Intelligence to centres of strategic and political intelligence serving the highest policymakers in Britain and the Empire.

Suddenly, civilian politicians and officials could see inside the thinking of their opposite numbers in other governments. Government realised how useful SIGINT could be in dealing with international affairs. On 1 November 1919, the British government established the Government Code and Cypher School (GC&CS), the first national SIGINT organisation in world history. The decision to establish GC&CS is significant because it was not made to meet the dire and immediate necessity of war, but to meet the day-to-day needs of peacetime government. The age of the spy had arrived.

Despite its national role, GC&CS remained an Admiralty responsibility in which Room 40 personnel, such as Commander Alexander (Alastair) Denniston, RN, Alfred (Dilly) Knox, William (Nobby) Clarke, Frank Adcock and Walter Horace Bruford, held important posts. In April 1922, though, the Foreign Office finally took administrative control of GC&CS. The change did not result from any reappraisal of GC&CS’s role, but from the embarrassment of the Foreign Secretary, Lord Curzon, whose indiscreet comments to the French Ambassador criticising his Cabinet colleagues were intercepted, decrypted and circulated by GC&CS to those colleagues.1 Curzon was not going to let that happen again, and brought GC&CS under the Foreign Office and thus under the sway, if not control, of the Secret Intelligence Service, also known as MI6.

In the case of Australia, Jellicoe saw the ACNB and its Naval Intelligence Department as the future centre for all intelligence collection in the Far East. This fitted in with his brief, but what did not was his recommendation that Australia be encouraged to create its own intelligence organisations, particularly a SIGINT organisation. Jellicoe had justified this recommendation by pointing out that Australia, being physically in Asia, would be a very keen collector of intelligence on Asia and would provide Britain with material that she could not otherwise obtain. This raised some complaint within the Admiralty that Jellicoe was going beyond his brief,² but the real threat to Jellicoe’s recommendations would come from the Australian government’s antipathy to paying for a national intelligence system on this scale.

The idea of Australia as a major centre for the collection of intelligence for the Admiralty was not new, but SIGINT had now changed the face of intelligence, and intelligence organisations had to deal with a flow of reporting on a scale far beyond anything a HUMINT system could generate. This was a problem because, to be effective, SIGINT had to be kept extremely secret and yet produced huge amounts of material that had to be transmitted by telegraphic or other means that also had to be kept secret. All of this meant a secret communications infrastructure on a scale never before contemplated by government. This was the beginning of what we today call the internet.

There is no point collecting intelligence if it cannot be provided to the decision-maker who needs to see it. In the early 1920s, as Jellicoe was writing, it could take an important telegram up to seven days to get from Australia to Britain, and this was without any additional reporting of SIGINT. The problem was a lack of bandwidth, the curse of all SIGINT, and, in the early 1920s, having to compete with everyday telegraphic traffic on the commercially operated network of transoceanic telegraphic cables. The only logical answer to this was to build a long-range wireless system or lay more and larger cables. Both of these projects were subject to serious commercial and international blocks, due to the need to negotiate with the United States, Canada and other interested parties.3

Not only would Australia and Britain have to build a suitable communications system covering the earth, but they would also need to recruit and train specialist operators to conduct intercept operations. This was not going to be easy, because these operators needed to work at higher speeds than ordinary telegraphists and had to be conversant with Asian signalling and languages. All of this was expensive.

The modest program put in place to teach Japanese could not meet these needs. The Japanese did not even use standard international Morse, because this could not carry the pictographs and syllables that make up the Japanese language. Japanese communications used Japanese Morse, also known as KANA or, more correctly, Wabun code. Telegraphists working as intercept operators had to be trained to listen to and take this down, but to do so they needed to speak some Japanese. Professor Murdoch’s program could only train a few officers, not the large numbers of intercept operators, clerks and traffic analysts that would be required.

None of this really mattered, because with the death of James Murdoch, the last vestiges of the Japanese-language program were brushed out of existence. The only Japanese-language training available was that provided commercially by Japanese teachers living in Sydney.

In the meantime, the ACNB looked to its existing capabilities as a means of collecting SIGINT. Foremost among these were the yeomen telegraphists aboard ships and working the coastal signal stations, which controlled signal traffic between ships at sea, ports and other authorities. The signal stations had become part of the wartime censorship system, charged with ensuring no sensitive information was passed by ship’s radio. During the war, this role had expanded to include the active interception and reporting of enemy communications during the periods of radio silence between schedules. This was relatively simple work as the enemy used standard Morse. It may not have been seen as such at the time, but this was SIGINT.

All of this activity was centrally controlled, with each station’s attached censor reporting directly to the Naval Intelligence Section at Navy Office while also reporting information of immediate and local tactical importance to the relevant DNO. This centralisation of control at a national level gave the signal station system one of the characteristics of a SIGINT system, even though censorship did not include cryptography, traffic analysis (TA) or cryptanalysis. The ‘system in force in peace’ was, to paraphrase the Admiralty, simply modified and extended for war. It demonstrated a very real interest in obtaining intelligence from enemy and other radio transmissions, while preventing the leakage of intelligence from friendly radio networks.4 It was SIGINT and it met the Admiralty’s demand for ‘prompt and accurate operational intelligence’.⁵

The fact that the Australian system did not include extensive cryptanalysis or codebreaking has fooled many historians into thinking that there was no SIGINT effort in Australia when in fact there was an extensive system operating throughout the 1914–19 period. The issue appears to be the modern habit of equating SIGINT solely with cryptanalysis, when, in fact, cryptanalysis is only one part of the SIGINT system.

The most important intelligence in SIGINT, especially at the tactical level where time is of the essence, is geolocation through DF and tactical information derived from TA. DF is vital for telling commanders the rough location of enemy forces, while TA provides information on who is communicating at what times and frequencies, what level of organisation they are and how active they are. In fact, TA can provide an enormous amount of intelligence without a single word of a message being understood.

TA not only provides the date and time of the message and the frequency it was sent on, but also the call signs that are communicating, allowing identification of the units involved. It also provides the number of messages in the series, allowing accurate collation of the traffic, and can provide insights into standard preambles and sign-offs that serve to identify the sender. Added to this, TA tells you just what grade of message you are dealing with, while the level of urgency and activity can be deduced from the speed of reply and total numbers of messages being sent compared to other equivalent periods. TA also tells you if the activity is routine or special. In a lot of instances, the only thing TA does not tell you is what the target is saying inside the message.

The value of TA is greatest at the tactical and operational levels, not at the high policy level, where the actual communication contains the gold. TA provides the foundation from which all cryptanalytical attacks are launched on a cipher or code.

Another misconception that has been incorporated into the history of British intelligence in Asia during this period is that it was of limited scope and value. The truth is that Britain was making full use of its technical capacity to spy on all the countries in Asia that used British telegraphic cables. This activity was compromised in 1925 when Vice Admiral William R. (Blinker) Hall, the Director of Naval Intelligence (DNI) between 1914 and 1918, handed 10,000 Room 40 decrypts to the American lawyer Amos Peaslee, so that Peaslee could prosecute a case against the German government for acts of sabotage in the United States.6 Why Hall did what he did and how he had secreted away so many of Room 40’s messages is unknown, but a National Security Agency (NSA) report into this event credits Hall with providing Germany with the evidence needed to develop ENIGMA.⁷ It also makes Hall the forerunner to Edward Snowden.

British cable companies in all British-controlled areas, including China, copied messages sent by any government or organisation of interest, including those of the United States, China, USSR and Japan, and passed them to Room 40 for further distribution to the Foreign Office, Ministry of Munitions the Board of Trade, and, it should be remembered, to the Indian government in New Delhi, which shared Australian concerns. This type of activity is the bread and butter of SIGINT. It is dry, boring and mundane, except to the economists, trade commissioners and those charged with running a wartime economy. It is not known how many messages from Asia were intercepted and passed along, but there is little doubt many were.

Following Jellicoe’s report, in January 1920 the Admiralty announced that a conference would be held in Singapore for the flag officers commanding the East Indies Squadron at Trincomalee in Ceylon (now Sri Lanka), the China Station at Hong Kong and the Australian Fleet at Sydney. The aim of the conference was to develop the strategic advice the Admiralty would provide to the 1921 Imperial Conference in London. Part of this advice would concern the arrangements for the future coordination of intelligence collection and wireless communications between the stations in the Far East and the Admiralty in London.⁸

The conference was finally held at Penang, Malaya, in March 1921. Australia accepted the majority of the recommendations arising from the conference, but refused to endorse the automatic subordination of the Australian Squadron to Imperial Command at the outbreak of war. Instead, Australia would retain control and only consider this sort of transfer if a crisis arose.

On the subject of intelligence, Australia agreed to contribute to the development of the wireless telegraphy network in the Pacific and to the tweaking of the Admiralty’s intelligence system. The importance of this to Australia lay in the need to keep Britain committed to Asia and to ensure she maintained a fleet and all of the support systems that fleet needed, including intelligence systems.9 The price was Australia’s commitment to building part of the necessary infrastructure, including the high-capacity, high-power beam radio stations.

The threat posed by Japan was now being seen as more serious. Not only was the IJN becoming increasingly capable, but the RN was becoming increasingly less so. The assessment of the conference was that good intelligence was essential for the success of any future operations against Japan. Early intelligence, it was hoped, could provide sufficient warning of Japanese aggression against Britain or her imperial holdings in Asia, to provide a ‘defensive period’ of around 38 days, the time it would take the Home Fleet to sail to the East.¹⁰ In order to achieve this, the conference agreed a Pacific Naval Intelligence Centre (PNIC) should be co-located with the fleet HQ at Singapore. This PNIC would provide intelligence to the Admiralty in London and the governments of Australia and New Zealand, as well as the administrations in India, Burma, Malaya and the NEI.¹¹

The intended PNIC was seen as a combined entity, with representation from all of the services as well as, one supposes, from the SIS and the Security Service (MI5). This latter inclusion is implied by the agreement that one requirement of the PNIC was that it would conduct both HUMINT and SIGINT collection operations. The PNIC would be responsible for the establishment and running of agents in Asia, particularly in security intelligence matters, and for the study of wireless telegraphy, cable and telephone networks.¹² In addition, the conference proposed two high-frequency direction-finding (HFDF) networks be established, one covering the Indonesian Archipelago from stations at Seletar on Singapore, Kuching in Sarawak and a station in North Borneo, and the second covering the south-west Pacific and the archipelago from stations located at Nauru, Rabaul on New Britain and one in New Guinea.¹³ These shore-based networks would be complemented by shipboard HFDF capabilities on all HM ships in the squadrons.

As events would prove, the Penang conference was highly optimistic. Britain and her Empire just did not have the resources in place. The HFDF system had to be built, as did the intercept sites, and this would be a costly exercise for governments in Britain, Australia and everywhere else who were now battling to bring military spending down from the massive highs of the war. Both the existing and future budgets were under strain, as Britain owed enormous sums of money to the United States and other lenders, money she had borrowed to fund the war.

The other problem was manpower. Like Australia, Britain had no Asian-language capability of any size. The existing program could send a few commissioned officers to Tokyo from Britain, India and Australia every year, but no soldiers, sailors or airmen.14 The issue was not as acute in relation to Chinese, because many British soldiers and service personnel learned Chinese while stationed in China as garrison troops. In an emergency, the population of Hong Kong could also be drawn upon to provide the linguists needed. Japanese was a problem of a completely different order.

Still, a sailor has to make do with what is available, and this is what the ACNB did. In June 1921, it ordered all ships and stations to ensure telegraphists practised taking Japanese Morse in a secret program called the ‘B Telegraphic Code’.¹⁵ Japanese Morse Code manuals were distributed to all RAN telegraphic stations and major units, and commanding officers warned that neither the activity nor the fact that the RAN had access to this code were to be discussed. This training of RAN operators in KANA was to be supervised by warrant officer telegraphists and, in an effort to ensure the Japanese did not intercept the RAN using KANA, practice was restricted to internal buzzer sets.¹⁶

The idea behind this program was to make use of suitably trained telegraphists to intercept Japanese Morse during quiet periods on their shifts while they maintained guard, a listening watch, on their assigned frequencies. Now, rather than listening to the universe’s background radiation or their favourite music, RAN telegraphists would scan the spectrum and report what they heard.¹⁷

This activity was reported monthly to Navy Office, which also detailed the progress being made by the trainees and the level of proficiency being attained.¹⁸ The ACNB also ordered that all recruit telegraphists be trained in both international and Japanese Morse.¹⁹ This was a step too far. The average telegraphist rating under training was not up to the task of learning International Morse, the technical aspects of telegraphy and Japanese Morse in the time allocated to training them, and the idea had to be dropped.20

By 1 July 1921, 47 blueprints containing copies of Japanese Morse systems and other SIGINT information had been distributed to ships and shore stations for the training program.²¹ All of this activity was in accordance with the decisions of the Penang Conference, but by October 1921 the Admiralty had concerns about the training impost on its own wireless ratings.²² Similar issues were arising in Australia, and by March 1922 the scheme had run into real trouble. At the end of that month, following a letter from the Rear Admiral Commanding HMA Fleet, the program was abandoned.²³

The abandonment of the listening watch and local training program did not end the ACNB’s attempts to develop the RAN’s SIGINT capability. As 1922 progressed, SIGINT in the RAN increased slowly through the artifice of using telegraphists aboard HMA ships to conduct intercept activity during the long quiet periods as they transited the region. The rationale for this lay in the practical experience of the telegraphists in intercepting radio transmissions from around the world at particular times of the day. The official view, though, was that ships needed to be in close proximity to a target to collect SIGINT.

In June 1922, HMAS Sydney proved that intercept operators afloat could conduct SIGINT operations against Japanese high-frequency (HF) transmissions in the Mandated Territories during her visit to Noumea.²⁴ It was a small advance, but it was an advance all the same.

In 1922, another step forward in the creation of an effective SIGINT attack on the Japanese was taken when Paymaster Lieutenant Eric Nave, RAN, went to Tokyo for formal Japanese-language training. At the time, it was unlikely that anyone understood the contribution Eric Nave would make to the Admiralty’s SIGINT operations in Asia, but he was destined to become one of the RN’s leading Japanese cryptologists.²⁵

Eric Nave was born in Adelaide in 1899 and gained employment as a clerk with the South Australian Railways in 1915, most likely due to the good offices of his father, a senior official there.²⁶ Nave remained with the railway until February 1917, when he left to join the RAN as a Paymaster Clerk on 1 March 1917. Exactly three years later, Nave was promoted to Paymaster Sub-Lieutenant, backdated to 1 September 1919, and it was during this time that he found himself a Japanese teacher in Sydney and developed the skills that would see him ‘loaned’ to the RN and posted to Tokyo on 16 February 1921.27

The teacher of Japanese that Eric Nave found had been ‘brought out from Japan especially to teach the language in schools here’.²⁸ The teacher was Miyata Mineichi, one of James Murdoch’s recruits. This action directly connects Nave with James Murdoch, Edmund Piesse and, at a bit of a remove, Atlee Hunt. Later, Professor Murdoch tutored Nave and told him about his plans to take captains J.R. Broadbent and G.H. Capes to Tokyo in September 1920 for further training.²⁹ Nave was keen to be sent to Japan as part of this group, but the wheels of Navy Office turned slowly, particularly as Professor Murdoch was difficult to communicate with, and Nave had to wait.

As Nave waited to find out if he would be sent to Japan, he arranged to take lessons from Professor Miyata. The cost of educating Nave was carefully managed, and Miyata’s fee of £4 5s ($AU1986.00 labour value in 2017)³⁰ was paid for two students, Nave and Paymaster Lieutenant Eric Kingsford-Smith (the brother of the famous aviator Charles Kingsford-Smith).³¹ The aptitude of both men was such that, on 20 March 1919, Miyata reported to Navy Office that they had made ‘remarkable progress and that they can speak, write and read short fundamental sentences with considerable fluency’.³²

Nave’s further education in Japan was approved by the ACNB on 31 December 1920, and the naval attaché at the British Embassy was informed via the RAN representative in London.³³ Eric Nave departed Australia aboard the SS Eastern on 28 February 1921. Now Australia had three students in Tokyo: Nave, Broadbent and Capes.³⁴ In Tokyo, Nave met another language student, Lieutenant R. Chichester, RN, who may have passed on his interest in cryptography to Nave.³⁵ Eric Nave soon came to the attention of the Admiralty and his posting to HMS Hawkins, on the China Station, was arranged while he was still on the course in Japan.³⁶

While Nave was in Japan, the RAN continued to develop its capabilities using the opportunities afforded by routine cruises in the islands to Australia’s north. In 1923, the Winter Cruise saw the first combined intelligence operation involving concurrent SIGINT and HUMINT operations when two RAAF officers, E.A. Mustard, DFC, and A.E. Hempel, joined HMAS Adelaide and Brisbane respectively to survey the islands for suitable airfield locations.³⁷ They paid particular attention to the Deboyne and Admiralty islands, and to the area around Aitape on the north coast of New Guinea. In 1932, another operation using a RAAF A9-2 aircraft for aerial reconnaissance and photography was conducted from HMAS Australia, while her specialist SIGINT team conducted intercept operations below deck.38 These operations display a surprising level of sophistication and commitment by the RAN and RAAF, particularly given the financial stringency that existed at the time.

By 1924, the pace picked up as the Admiralty increased pressure on the China and Australian stations to send more SIGINT intercept to the newly formed Japanese Naval Section at GC&CS. This new section had been created because of tension between the Foreign Office and the services over restrictions imposed on the type of intelligence GC&CS was to produce. The Admiralty finally buckled in 1924, and financed and staffed the Naval Section under the leadership of Paymaster Commander E.P. Jones. Experienced Japanese linguists and intelligence officers from the China Station were then posted to this section.³⁹ This meant Admiralty had a dedicated team looking at IJN traffic. All it needed was the raw material of SIGINT to work on, and this required the China and Australian stations to start aggressively collecting SIGINT.

The Admiralty instructed the C-in-C, China Station, to establish the first of the ‘Y’ Stations in the Far East at Hong Kong—‘Y’ Procedure was the early covername for SIGINT operations. The Hong Kong station was to collect and conduct TA and low-level decrypts of IJN traffic for local analysis while forwarding all encrypted IJN traffic to GC&CS.⁴⁰ The result was a slow and painful increase in capability through 1924 until 1925, when Hong Kong began to produce better results. Part of this increase in capability may have been due to the arrival of Eric Nave there on 12 June 1925 to take up ‘specialist duties (W/T Procedure ‘Y’)’ and as interpreter in Japanese under the direct supervision of the C-in-C China Station.⁴¹ Within a short period, the flow of intercept increased to a point where the Naval Section at GC&CS was unable to cope.⁴²

In Australia, in July 1924, the army intruded into SIGINT by suggesting that a voluntary civilian radio intelligence bureau be established to conduct radio interception activity. The main proponent of this idea was a Major H.A. Corbet in Perth. His idea was odd, and not because he thought a voluntary organisation could conduct SIGINT. The odd bit was that he thought that wireless specialists working for Amalgamated Wireless Australasia Ltd (AWA) would be allowed to spend their ‘down-time’ intercepting and reporting foreign communications.43 The more experienced ACNB replied to this proposal by pointing out that AWA staff had plenty to do even when they were not taking messages. Undoubtedly, the ACNB knew from direct experience that AWA did nothing for free, and moved to kill off Major Corbet’s idea as quickly as they could by proposing no further action be taken.

It was not the end of the matter. It seems others, maybe even politicians, took up the idea. On 2 December 1924, Commander Frank G. Cresswell had to reinforce the initial rejection of Corbet’s idea by pointedly informing the head of Naval Branch that the ACNB’s preferred option was the provision of modern recording receivers at certain shore stations.⁴⁴ Cresswell also made it clear that this issue was strictly a naval matter, and that the army and others should stay out of it. Cresswell and the ACNB were not interested in using amateur radio enthusiasts; they were looking to build an organisation of professionals equipped with modern high-speed recorders like the Dictaphone.⁴⁵

And this brings us to technology, an important part of the Australian SIGINT story and one of the least considered factors in much of the history of SIGINT during this period. One of the most difficult areas for the radio operators and specialists of the time was propagation, particularly in the tropics, the area of most interest to Australia.⁴⁶ It has to be remembered that Professor Edward Appleton’s proof of the Kennelly–Heaviside layer’s ability to reflect radio waves was only confirmed in 1924.⁴⁷ Until this happened, the Admiralty and everyone else thought that SIGINT interception of radio messages could only be done on a ship passing close to an enemy coast.

Even simple tasks like getting accurate recordings of the actual message were an issue. The usual method at the time was for the intercept operator to write down the message as he heard or misheard it. This left the analysts at a loss, because they could not identify what was operator error or actual message content. The answer lay in the invention of devices that recorded the actual sound of the message as it was intercepted: the Dictaphone Company’s electronic recording equipment.

In 1924, the ACNB acquired two Creed Relay high-speed wireless telegraphic recorders from the Admiralty to record Japanese Morse for the small group of expert telegraphists and linguists capable of working on them.⁴⁸ These recorders were found to be highly suitable for radio signals emanating from close to Australia. For the much more distant and very faint IJN traffic, however, the Creed Relay was too crude to pick up the signals, and the audio recorders were not sensitive enough.49 The answer was the new US Dictaphone, which had been developed to record music, specifically symphony orchestras for later broadcast by commercial radio stations.

The Dictaphone was a high-speed electric recorder and was one of the most important advances ever made in capturing sound electrically. Western Electric designed and built the first such recorder, which comprised a Bell microphone connected to a valve that supplied the amplified signal to a Bell electromagnetic cutting head. This new technology extended the frequency spectrum that could be recorded from 200 Hertz at the bottom to 6000 Hertz at the top, which was 3600 Hertz higher than any mechanical recording system.⁵⁰ This was the technology now sought by the ACNB. The value of this technology lay in its capacity to record the actual sounds so they could be replayed repeatedly to cryptologists and other operators. This was a major advance in SIGINT capability.

By 1925, Dictaphone had developed a similar device designed specifically for capturing telephone messages. The ACNB’s interest in this technology most likely explains why the Dictaphone Company opened its first office in Australia in 1926, the same year the Dictaphone was enhanced by the addition of piezoelectric featherweight styli, and the company sought an Australian patent for these ‘improvements in phonographic machines’.⁵¹ The success of the Dictaphone in the ACNB’s SIGINT operations may also be the reason the company took out further patents relating to ‘improvements in shaving and resurfacing of phonograph records’ in 1928, something the RAN needed to do during ship deployments.⁵²

These developments improved SIGINT collection in three ways. First, it enabled raw signals to be sent back for further technical analysis; secondly, it meant that the number of intercept operators required for tasks could be reduced; and thirdly, it meant that intercept operators did not have to be as highly trained as first thought.

The addition of this technology paid dividends as RAN intercept operators began to realise that they could hear IJN traffic emanating from the Home Islands of Japan. The experience of these operators was confirmed with Professor Appleton’s proof at the end of 1924 that radio beams bounced off the ionosphere to land at great distances from their origin.⁵³ This propensity of radio waves made Australia a very important part of the SIGINT system, when it was realised the nation was perfectly situated for intercepting HF transmissions originating in Europe and the Americas (see Figure 10.1).⁵⁴

The significance of Appleton’s discovery and of the experience being gained in Australia and Asia meant that long-range interception could be systematically exploited from the safety of Australia and other British territories. ‘Y’ Procedure had become even more attractive as an intelligence technique, and the Admiralty’s insistence on only using ships for ‘Y’ Procedure was no longer as significant as once thought.⁵⁵

The importance of this work should not be underestimated. At the time these developments were being achieved, neither the Admiralty nor the ACNB knew how the IJN operated. They did not even know if it was using International Morse or a special syllabic Morse.56 The new technology and the active program of listening and recording began adding slowly to the intelligence picture of Japan’s wireless communications. The contribution made by the American entertainment and electronic industries and academic research into the behaviour of radio waves had produced greater advances in SIGINT than codebreaking. These developments, along with the recruitment and training of specialist SIGINT intercept operators, crystallise a truth that is often overlooked in the story of SIGINT: it is an industrial activity in which intelligence is only produced by large, well-run organisations made up of numerous specialities and work groups of which the cryptanalysts are only one part. A factory is required to produce good SIGINT.

Figure 10.1: How radio propagation made Australia important in intercepting long-range Japanese diplomatic high-frequency transmissions

In October 1924, cryptanalysis of IJN messages was lagging behind the advances in technology, and Lieutenant Nave was now starting to insist on getting better access to the raw materials. One example is his insistence on being given the updated copy of the Japanese Naval Telegraph Code that Captain Broadbent had brought back from Tokyo. This appears to have been the first time an Australian cryptologist demanded to see the raw material himself in order to form his own judgements as to its value. The fact that Nave was given the copy he demanded indicates that the RAN and Navy Office were more sophisticated than is sometimes assumed.

Nave’s plan of attack on the IJN’s codes appears to have been to obtain an intercept of their plain-language messages to learn their operating procedures. This was an important first step, because the idiographic nature of Japanese made abbreviation difficult. Once he had mastered this, Nave then proposed attacking the Japanese Economic Code in preparation for the attack on more heavily enciphered traffic. This is where the other specialists came in, because Nave needed experienced wireless ratings skilled in reading Japanese messages. It seems that the ACNB and the RAN were looking at concentrating this capability on the flagship, HMAS Australia.⁵⁷

These plans changed because the Admiralty needed to find qualified personnel to work on the China Station so that the newly created Naval Section at GC&CS could be kept supplied with intercept. Eric Nave was an obvious candidate, and in late 1924 the Admiralty requested that he be made available for service on the China Station. In early 1925 the ACNB agreed.⁵⁸

As Eric Nave was being sent to the China Station, the ACNB was obtaining four portable Marconi medium-frequency direction-finding (MFDF) kits from the Admiralty. These kits were positioned at RANVR bases—in Port Melbourne, Edgecliff in Sydney, Fremantle in Western Australia, and Brisbane—under the command of the DNO in each port. The primary objective was to train the RANVR in DF operations, thus providing the RAN with a portable DF capability that could be quickly assembled and deployed.59

The year 1926 would prove a good one for SIGINT on the China Station and in Australia. At the end of 1925, Eric Nave had doubts there would be enough intercept taken to keep him fully employed.⁶⁰ He feared that the difficulties of dealing with Japanese traffic might result in the whole program being wound down. This did not happen and instead more resources, although of a limited nature, were committed to the work, and Nave was posted to HMS Titania, the Submarine Depot Ship, where SIGINT operations were being extended and new equipment, including Dictaphones, was being installed.⁶¹

The problems in training enough specialist operators that had eroded confidence in the ability of the China Station to cover the IJN effectively had been resolved by the Dictaphone, and the concerns about the unbreakability of the IJN codes were also put aside.⁶² Instead, the Admiralty informed the C-in-C, China Station, and the ACNB that SIGINT was ‘a subject to which the Admiralty attached the greatest importance’.⁶³ The C-in-C, China Station, was instructed to increase the urgency and effort being put into the SIGINT attack on the IJN.⁶⁴ These orders, along with the arrival of Dictaphones and other equipment seemed to have worked, and by July 1926, the Naval Intelligence Division (NID) was reporting to the Admiralty that considerable progress had been made.

The Admiralty added to the attack on the IJN’s codes by posting Lieutenant Commander Stephen Lushington to HMS Ambrose. Lushington was a SIGINT specialist with close ties to GC&CS, and appears to have been its man in the East.⁶⁵ To provide Lushington and GC&CS with material, the C-in-C, China Station, was instructed to ensure that any HM ships venturing into waters near Japanese locations were manned and equipped to undertake SIGINT. This approach had just been proven in the Mediterranean, where a single destroyer with a dedicated SIGINT team aboard had collected more material in a few weeks than the entire Mediterranean Fleet had in six months.⁶⁶

This telegram appears to have been what prompted Commander Cresswell to continue the schedule of small SIGINT operations by the RAN.67 The sending of a single destroyer with a small group of specialists on board was well within the RAN’s capability, as Cresswell knew. He also knew his operators were intercepting IJN traffic from Funabashi and Tokyo at the Navy Office’s own communications centre in Melbourne. This was long-range interception indeed, and Cresswell reported that the Admiralty-supplied receivers had been highly effective in picking up these transmissions.⁶⁸

China Station had also passed a report from Nave on Japanese communications that identified the call sign ‘AB’ as Imperial HQ. AB carried all IJN Admiralty traffic and broadcast routinely on even hours, while the Major Stations, the next level down, broadcast on the odd hours to all ships and minor stations within their geographic area of responsibility.

The work on AB also provided a major crib into the IJN’s codes, the realisation that the hierarchical IJN forced subordinate call signs to pass messages intended for one another via their higher control station. This meant that two different operators were sending the same message twice in quick succession, and this dramatically increased operator error. This duplication enabled the rapid identification of call signs and their relationships to one another. More importantly, the errors stood out when the recording of each version of the message was compared. This sort of crib seriously undermined the security of the IJN’s codes.

Another important characteristic of IJN wireless procedure was also identified: the IJN’s telegraphists were not very good. Their lack of uniformity, frequent duplication and mixing of code with clear transmissions enabled the British to quickly build up their net diagrams, the paper charts of call signs, their frequencies, opertating times and other technical data. Net diagrams are the first output of traffic analysis and provide an indexed file of all activity noted on a network.

Even AB, Tokyo, was guilty of coupling international call signs with secret ones. The result was that the British ‘Y’ Organisation, although small, quickly identified the secret call signs of eleven of the seventeen major shore stations, as well as those for the battleships HIJMS Mutsu, Nagato, Yamashiro and Kirishima; the cruisers Nagara, Tone, Yubari, Iwate, Kasuga, Komahashi, Kamakaze and Tanikase; and ten destroyers and six fleet tankers.⁶⁹

By June 1926, the SIGINT attack on the IJN was going so well that the Admiralty expressed its pleasure at the great progress made by the China Station. It had good reason to be pleased, as in December 1926 alone, China Station intercepted and recorded more than 1000 Japanese messages.70 The difficulties had been overcome ‘by great zeal and ability on the part of those concerned’.⁷¹ One officer singled out for individual praise by their lordships was Paymaster Lieutenant T.E. Nave, RAN.⁷² One of Atlee Hunt and Edmund Piesse’s tyros was beginning to make an impact.

In Australia, the loss of Eric Nave to the RN did not mean the end of the ACNB’s efforts in cryptanalysis, as the work started by Nave was taken over by a civilian employee, R.A. Ball,⁷³ who had just completed the Japanese-language course in Tokyo.⁷⁴ On his return to the ACNB, Ball’s contribution to SIGINT was just as important as the one Nave was making in Hong Kong at the time.

Australia’s SIGINT effort was now benefiting from the newly improved Dictaphone equipment and the arrival of better recorders that now meant small vessels could undertake quite extensive SIGINT operations to supplement the listening watches kept at the RAN W/T Service shore stations. Electrical Commander Cresswell, the staff officer responsible for communications, now argued for a SIGINT mission to Rabaul, where transmissions from the IJN stations at Funabashi, Yap, Truk (Chuuk), Jaluit and the Bonin (Ogasawara) Islands were being picked up.⁷⁵ The mission would last three months and would be conducted under the cover of the regular goodwill mission by an Australian destroyer to the Mandated Territories.⁷⁶

Ball supported Cresswell’s suggestions, arguing that as he was now available to fill the role of Japanese linguist at Navy Office, the other Japanese linguists could be released to take part in SIGINT operations. Ball believed that the RAN could provide different intelligence to that obtained on the China Station because the RAN would be operating against the Japanese in the Mandated Territories. He also believed that giving the telegraphists practical experience would result in more cribs being detected in Japanese operating procedures.⁷⁷

All of these ideas were contained in a report Ball wrote expanding on some of Nave’s suggestions. Among Ball’s recommendations was that all intercept be recorded using Dictaphone to ensure that no loss of syllabic groups occurred during ‘Y’ operations. His procedure placed a heavy emphasis on Dictaphone recordings, with manual cover only being used when the Dictaphone records were being changed over. Ball’s procedure also provided a bi-graph system for noting Japanese syllabic groups corresponding to English letters and a table into which this information could be entered. Another of Ball’s improvements was ensuring messages were numbered consecutively according to the order in which they were intercepted, rather than relying on the less dependable original message externals (the message numbers and other operating information attached by the original Japanese telegraphist).

Other issues, such as atmospheric breaks in transmissions and losses due to technical reasons, were to be marked by ‘?’ to enable linguists and cryptanalysts to understand the reasons for gaps. All of this would make it easier for the linguist to read the intercept. It removed obvious errors and suggested possible groups for identified gaps. This initial analysis would then be compared with the Dictaphone record if necessary. During this activity, the linguist would identify and exclude plain-language private messages and messages of no intelligence value. The final intercept, or ‘take’ as it is called, of encrypted messages and plain-language official messages would then be forwarded to the DNI.

Ball was also a keen advocate of linguists working closely with intercept operators to establish station identities and construct net diagrams, forming the basis for further exploitation through TA. Ball ‘earnestly’ submitted that this routine be put in place and allowed to run for a period to establish whether it worked.78 For the time, this was quite a sophisticated state of affairs, and the resulting improvement in the RAN’s SIGINT capability was noted in the Admiralty.

Commander Cresswell’s proposal for a SIGINT mission was now agreed, and the Assistant Chief of the Naval Staff (CNS), Commander H.T. Baillie-Grohman, took responsibility for it. Baillie-Grohman informed the CNS, Rear Admiral William Napier, RN, that not only did he accept the importance placed on ‘Y’ Procedure by the Admiralty, but he had already approached General E.A. Wisdom, the Administrator of New Guinea, to request the use of his steam yacht SY Franklin in a ‘Y’ Procedure mission.⁷⁹

The Franklin was selected because, as the Administrator’s official yacht, she was a known entity in the waters near the Japanese Mandate. She was also big enough, the ACNB hoped, to house the SIGINT team and its equipment. Franklin would prove problematic, however, because no one had realised that the wind would create vibrations within her rigging that would cause background noise and electrical interference from static. Her electrical system also turned out to be a major source of unwanted interference. All of this was unknown when the mission was put in place.

Using the cover story that she was to undertake transmission and reception trials in the seas around and to the north of New Britain and New Ireland, the Franklin was fitted out in Sydney in early 1927. The mission lasted from April until July 1927. Warrant Telegraphist B. Harding led the team of four telegraphists, which included Leading Telegraphist H.J. Barnes, of whom we will read more later.80

This mission was not a local or spur-of-the-moment operation. The support that Navy Office and the intercept team received not only included Eric Naves’ intelligence, but also British HUMINT, most likely from SIS, as serendipity cannot explain how the detailed report on Japanese wireless developments from W. Turner, the General Manager of Reuters in the Far East, arrived in Australia just before the Franklin mission.⁸¹

This report, passed from the highest levels of the British government to Australia for use on the mission,⁸² described Japan’s telegraphy system in detail. It also outlined how the Japanese Communications Department operated a dozen home stations and an overseas network including Fukkikaku (Fuguijao) on Formosa (Taiwan), Dairen in Kwantung, Seoul in Korea, Otomari (Korsakov) in Sakhalin and Paramushiro (Paramushir) in the Kuril Islands during the fishing season.⁸³ Turner’s report provided information that the IJA and IJN operated their own networks independent of the Communications Department and one another. Further, it described the dockyard network of three private stations that coordinated berthing and repairs of vessels.⁸⁴ The only information Turner had on the Japanese Mandated Territories was that there were seven stations, but he had no details on them.⁸⁵

The intercept team on Franklin were setting out well prepared and provided with excellent intelligence. It was only when they were at sea that they discovered the Franklin’s unsuitability.

The team installed improvised short-wave transmitting gear and Navy Office receiving gear, selected by Harding. Two-way simplex working was established and maintained with Navy Office on two frequencies, 33.5 and 35 metres, from the date of departure until Franklin reached a point 700 nautical miles north of Sydney. From there, she continued north to New Guinea, while her intercept team installed and tested their intercept equipment and initiated the interminable and only partially successful routine of fault-finding, noise reduction and vibration abatement.86

In the areas around New Britain and New Ireland, two-way communication was maintained between 1730 and 0830 hours (K and L time zones). The intercept operators encountered a number of serious technical problems caused by the swaying of the yacht, and noise and vibration caused by the wind passing through the rigging. The vibration was serious enough to make reception very difficult, and the team had to use a smaller antenna condenser to reduce oscillation and lessen these problems. They even had to begin screening all the electrical wiring on board in an effort to reduce induction issues.⁸⁷

On top of these issues, the team also had to contend with the failure of the motor in the Dictaphone, which, luckily, they were able to replace in Port Moresby. Then, on arrival in Rabaul, they found out that the Administrator had been forced to surrender his copy of the naval cipher to the Defence Department when the Prime Minister’s Department had taken responsibility for Papua New Guinea. This meant they could not communicate with Navy Office on ‘Y’ Procedure matters or with the intercept operators who had left the now immobilised Franklin to conduct operations ashore.⁸⁸

Navy Office was not amused and signalled ‘The seriousness of this failure is being represented in the right quarters’, particularly as the Secretary of the Department of Defence had not replied to Navy Office’s original request for action.⁸⁹ Baillie-Grohman also ordered the wireless telegraphic group not to conduct ‘Y’ Procedure operations from Government House in Rabaul for security reasons.⁹⁰

The SIGINT operation conducted by the Franklin team was well thought out and well organised. Security was maintained at the highest level, with good cover being provided by the vessel itself, and when matters slipped, by senior officers in Melbourne. Despite all the setbacks, the mission was not only a procedural and technical success, but also an intelligence success. Harding’s team had identified the call signs of HIJMS Tone (call sign JLF), Katata (call sign JWH), Minekaze, Nagato, Kongo, the submarine tender Chogei, submarines i52, i53, and two other submarine call signs unattributed to vessels. Effectively, the Franklin mission identified the IJN’s 1927 submarine net, as described in Figure 10.2.

Source: NAA BC 505956

Figure 10.2: The IJN submarine net identified by SY Franklin, 1927

Another significant success of the mission was that the Dictaphone worked very well and 100 recordings were made, 97 of which were of Japanese transmissions.⁹¹ This data allowed the identification of 99 call signs as well as their frequencies and schedules. This was all the information needed to begin the job of constructing net diagrams and developing an order of battle.⁹²

The mission also identified a number of new procedure signs and their meanings, and the secret call signs used by a number of IJN stations and vessels, such as the submarine flotilla. The success of the mission was due to the cribs provided by the numerous errors committed by Japanese operators, and by the intercept operators aboard Franklin identifying operator and equipment idiosyncrasies in messages from the international call sign and the secret call sign of the targeted transmission.⁹³

The Franklin’s Dictaphone records were received at Navy Office on 3 August 1927 and copies of these, plus 424 message transcripts taken by hand, along with Warrant Officer Harding’s report, were immediately forwarded to the Admiralty. The originals were forwarded to Ball, and Petty Officer Telegraphist L.G. Porter, who was expert in reading Japanese Morse.⁹⁴

The only fly in the ointment was provided by the Acting Secretary of the Department of Defence, who complained that it was ‘quite impossible, without lamentably disorganising the current work of the office, to make Mr Ball available for full time on this work, at any rate while the temporary arrangements in the Secretariat continue’.95

His superior released Ball for two hours every afternoon to look at the material, and it was hoped that the return of Paymaster Lieutenant Nave in September 1927 would enable the latter to take the material to London with him.⁹⁶ It appears that the Acting Secretary of the Department of Defence did not really understand what was being done.

Further planning on ‘Y’ Procedure activity was undertaken in September 1927, to see if the inaugural voyage of the newly acquired cruisers HMAS Australia and HMAS Canberra back to Australia could be used as a collection opportunity. This voyage provided excellent cover while the two ships crossed the Pacific. It was also recommended that Nave be asked to assist in these arrangements.⁹⁷

On 21 February 1928, the Admiralty wrote to the ACNB expressing appreciation of the ‘effort at interception on board Steam Yacht Franklin and ashore at Rabaul’, but advising that the lack of live cryptographic traffic made the activity ‘unremunerative’.⁹⁸ The bulk of the traffic was plain-language commercial and civilian messages, with almost no operational coded traffic. This was a gloomy assessment given the operation proved the concept of mobile SIGINT collection by a small team. However, the China Sea and Pacific Ocean operations were not producing enough encrypted traffic for the Naval Section at GC&CS to make an effective attack on the IJN’s ciphers.

Australia faced its own resourcing problems, not least of which was a lack of qualified and experienced personnel for SIGINT operations, made worse by the loss of experienced staff as their period of engagement came to an end. In July 1928, the RAN lost Petty Officer Porter and two of the leading telegraphists who had served with him on Franklin. This was a quarter of the RAN’s specialist SIGINT operators and three-quarters of their most experienced team. The last man in this team, Leading Telegraphist Barnes, was posted to HMAS Canberra. This loss was so serious that Commander Cresswell informed the CNS, and the ACNB even considered passing this information to the Admiralty.⁹⁹ It was a serious blow to the Navy’s nascent SIGINT organisation.

Yet 1928 saw its breakthroughs. Technology was now available that enabled a lot more to be done by fewer people, and the senior officers of Australia’s navy were strong advocates of the importance of good foreign intelligence collection, especially SIGINT. Despite the limitations imposed by falling budgets and the lack of linguists and other specialists, enough intelligence was being collected to deduce the IJN’s order of battle from TA and the creation of the first net diagrams. In addition, Eric Nave broke the IJN’s nine-letter general-purpose ‘43’ code, and although the exploitation of this breakthrough was limited by the lack of linguists, it represented a real step forward in the attack on the IJN.100 The RN certainly thought so. In his annual officer’s appraisal, Nave was described as a ‘diligent and efficient’ officer of ‘exceptional’ zeal and energy, who was ‘an exceptionally able Japanese interpreter according to Japanese Naval Officers, with whom he is very popular’.¹⁰¹ He was also an excellent spy.