Tuesday, 2 August 2016

Cryptography in Weird War 1

After hostilities commenced in August 1914 the Admiralty's secret intelligence unit, Room 40, stepped-up its monitoring and codebreaking operations against Germany, providing the British armed forces with tide-turning information about the enemy's plans. This second of a two-part series highlights Room 40's operations from the outbreak of hostilities to the war's end.

Room 40 was the Admiralty's secret intelligence-gathering and -processing department that provided vital intelligence to the British military commands and their allies during the First World War.

Based in the Admiralty Building in London, it was in several ways a predecessor to the more concerted and sophisticated code-breaking operations at Bletchley Park during the Second World War.

The British had already moved to protect its globally-encompassing telegraph cable network – the 'All-Red Line' – and had curtailed German international telegraph communications by cutting five subsea German cables off the Atlantic coast.

At the start of the war, codes and ciphers deployed by the military, naval, and diplomatic services on all sides were relatively primitive, derived from the age of the cavalry on land and of sail at sea. There was an awareness of the need for secrecy, so signals were first encoded using common codebooks; then, for additional security, signals were enciphered. The concepts underlying these techniques, however, were distinctly old-fashioned, and also increasingly vulnerable to innovations in code-breaking techniques.

The main German naval codebook, called the Imperial Navy Signal Book ('Signalbuch der Kaiserlichenmarine'), treated coding methods for wireless signals as an afterthought, added onto a publication that concentrated on traditional means of signalling at sea, i.e., using flags and other visual techniques, such as flashes of light.

The design of this and the other codebooks used by the Imperial German Navy and German Army was simple in the extreme. The codebooks listed plain words alphabetically, but also allocated codewords in alphabetic order, thus making the coded signals very vulnerable to the codebreakers, even if they did not have direct access to the codebooks being used.

Codes that are made to be broken

In any event, British Admiralty codebreakers working in Room 40 did not have to concern themselves initially with cracking the codebook structures: this was because all three of the main German naval codebooks were captured within a few months of the start of hostilities.

Similarly, the cipher techniques, used to 'super-encipher' the coded signals, they found were surprisingly primitive. In some cases they even used the simple 'Caesar Cipher', where there is a fixed transposition of one letter in the alphabet for another (so coded letter A is always transposed to enciphered letter D, for instance). These ciphers, even if slightly more sophisticated, were an easy technique to break.

So it was no surprise that French and British army codebreakers worked-out how to crack German army cipher techniques in the short period of mobile warfare up to the First Battle of the Marne (September 1914). This was to prove fortunate as, once the trenches had been dug and cables laid, the German Army used very little wireless signalling on the Western Front until later in the war. Information about this valuable technique was passed back to Room 40, which was then able to decipher all German naval wireless traffic.

For some time into the conflict, even when the German codebooks were replaced with new versions, they were structured in the same way using the same alphabetical plain word/codeword allocations, and thus took little time to break. Similarly, although the German forces changed their individual cipher keys fairly regularly, they stuck for two-to-three years with the same simple cipher systems, and it took some time for new keys to be broken by the British.

From August 1914 the British Royal Navy thus had an unmatched insight into the activities of its German counterpart, listening-in to instructions for enemy vessels to collect at a certain time, and for harbour lights to be turned on at the same time (thus revealing that the Imperial German Navy battle fleet was preparing to set to sea on a sortie): one of these led to the Battle of Jutland (30 May-1 June 1916).

Room 40's intelligence gathering was aided by the prolific use of wireless communications by the German Navy, with submarines and small surface vessels such as minesweepers, announcing departure, course and activities at frequent intervals.

To be sure, this harvest of prime intelligence was not always used to best advantage by the Royal Navy's admirals and commanders. The Battle of Jutland was effective by default, in that it scared the Imperial German Navy surface fleet into remaining in port for the remainder of the First World War – but the Royal Navy missed the opportunity to hammer its enemy's fleet. Indeed, it took a good two years for the Admiralty to understand how best to organise Room 40's operations effectively.

Contrary to expectations Room 40 founder, engineer and physicist Sir James Alfred Ewing, turned out to be not a particularly good leader of an operational unit, and in 1917 was replaced by the more dynamic (and devious) Captain (later Admiral Sir) William Reginald Hall (1870-1943), the director of Naval Intelligence from 1914 to 1919 (who was known as 'Blinker' Hall due to a facial twitch, which reportedly caused one of his eyes to blink like a flashing Navy signal lamp).

Security vulnerabilities

Progress in terms of intelligence gathering and processing was patchy on land, too, in the early years of the conflict. The opposing armies, bogged down in the trenches, laid-out dense networks of communications cables, rather than use wireless communications at the front. Both sides also learnt how to 'tap' into their enemies' telephone and telegraph communications; but ironically, they did not always pay sufficient attention to securing their own communications.

Arguably, the most important development in British codebreaking at this stage of the war was the way codebreakers British Military Intelligence (MI1b for the army, Room 40 for the Royal Navy) turned their attention to diplomatic eavesdropping. They began to take an interest not just in German diplomatic communications, but even those of friendly neutral nations, such as the USA.

The US diplomatic codebook was broken by a sound, if unoriginal, ruse. The British handed the US ambassador in London a diplomatic note that they knew would have to be transmitted by telegraph to Washington in full. Before crossing the ocean on a submarine cable, the now encoded message was sent on a telegraph land cable from London to Cornwall. The British were able to covertly intercept the signal en route to the West Country, and used it to start working-out the structure of the US code scheme.

This allowed the British military and political leaders to follow US diplomatic moves, such as the promotion of peace talks, and to keep tabs on Germany's efforts to coerce neutrals into supporting its point-of-view before it introduced unrestricted submarine warfare at the beginning of 1917. The codebooks used by that time were much more complex than the early alphabetic allocation codebooks. Codewords were now allocated randomly, making it much more difficult to break new codebooks if a physical copy had not already been captured.

It was necessary to work-out about half of the codeword meanings of a codebook before it would be practical to break enough of the individual messages to make any sense of them. It would take a lot of effort, therefore, to break a codebook with 10,000 codewords – which was not an unusual number at the time. As the First World War progressed the Germans changed their codebooks more frequently, so it was necessary to break a book quite quickly – or all the effort would be wasted when a new one was introduced. The codebreakers would use complex logical assumptions to identify, first, codewords representing numbers, punctuation and common terms such as names, units, call signs, and suchlike. From then they could start to focus-in on more and more of the content of coded messages.

It was a slow and arduous process, and needed people who worked with words – such as professional lexicographers, experts in ancient history, and other specialist linguists – whereas, by contrast, in the Second World War ciphers needed mathematicians.

Early code-breaking machines?

Codebreaking was a very labour-intensive process, and the volumes of data being collected was soon placing strain on the Room 40 team and its resources (an Allied intelligence report of 1918 notes that codebreakers "must possess the faculty of keeping anything from a dozen to 20 theories in [their] mind in order to build-up a chain of coincidence and reasoning until each link fits into its place and forms a coherent whole").

Sometime in 1916 one of their number – whose identity is unfortunately not recorded – came-up with the idea of using machinery to work out the sequence of logical steps of the code-breaking process, usually known as 'flow-charting'. The science of code-breaking and interpreting intelligence was about to enter a new and highly significant phase that would, it can be argued, result in innovations that would play into the development of the electronic computer, leading to Bletchley Park's Colossus.

Not much hard evidence about the nature of the Room 40 apparatus, or how it functioned, survives. From the somewhat scrappy extant information available in the UK National Archives at Kew we can guess that whatever physical form it took, it was almost certainly a type of punched card tabulator machine; but not much more else can be said for sure about the 'hardware configuration', as it were.

The tabulating machine was an electromechanical machine invented to help summarise fielded information and, later, accounting applications; computing giant IBM had its origins in tabulating machine technology. The only clue about the Room 40 hardware is mention of a 'pianola'. It was quite common for the pianola-type device to be mentioned in a description/introduction to the idea of punched card machinery in computer histories (pianolas are sometimes described as among the first example of mechanical automation). The key point is the 'holes/not holes' concept as a means of conveying machine commands and/or information, rather than the precise nature of the card/roll.

The codebreakers adopted the term 'hatted' for these randomly allocated codebooks, as if the codeword for any plain word had been drawn out of a hat. This led to the description of the team of machine operating women as 'grinding' codeword meanings 'out of the hat machine'.

There is, however, some information about how the new Room 40 machinery sped-up the process of working out codewords. In a document about diplomatic codebreaking there is a short account of the project: "It was not realised that this form of [randomly allocated] code required special treatment until May 1916 when leave was granted to set-up a special staff of educated women to work machinery by which the guessing process could be accelerated... By this method the [number of] guessed codewords rose at once to 20 daily, and by the law of increasing returns grew mechanically to a maximum of 100 per day by which time the code was approximately readable."

This contrasted with a handful of codewords that could be worked-out in a day by a practiced person. It is the sort of productivity increase that we typically associate with the introduction of modern information technology (IT) to a data-processing-intensive requirement. In another indication of the way that IT changes the nature of work, the report adds that "the reading of messages in such codes proved to be merely a matter of tedious drudgery for one or two experts and the staff of ladies trained by Miss Robertson".

It is, of course, highly regrettable and frustrating to contemporary technology historians that so few details are available about the 'hat machine'. Knowledge of its existence helps us understand why Second World War codebreaking machines such as Colossus (and many other less-well-known machines) were invented at Bletchley Park. The concept of using machines for codebreaking was not new in 1940, but was yet another way in which the First World War forerunners helped define the progress of the later conflict.

Most sensationally, Room 40 intercepted a January 1917 cable message from the German foreign minister, Arthur Zimmermann, which attempted to bring Mexico into the war on Germany's side – by militarily attacking the US. The cable was decoded, and its contents leaked to the US President: it played a part in bringing the US into the war, tipping the balance of power against Germany and its allies. The incident has gone down in cryptographic history as the interception of the 'Zimmermann Telegram'.

It was a triumph for the eavesdroppers and the codebreakers, but also for William Hall, who worked-out how to pass the intelligence onto the Americans while retaining two important secrets. The first secret was to make sure that the Germans did not realise that Britain was intercepting and decoding their secret communications; this was militarily vital. Fortunately, the Germans blamed their own people, not the enemy, for the leak, so this secret was safe. It was also politically vital that a second secret also be held very securely indeed. This time it was the Americans who had to be kept in the dark.

The Mexican connection

US President Woodrow Wilson had allowed the Germans to bundle their cables in with those ostensibly coming from the US embassy in Copenhagen, Denmark. The reason for this was that, as previously mentioned, the British had cut Germany's cables and had captured its international wireless stations, gaining a hold on its communications with the wider world.

President Wilson allowed Germany to use US facilities to communicate with the German embassy in Washington, supposedly about his peace proposals. Instead, Zimmermann used the channel to try and provoke a wider war that included an attack on the US mainland. Much as they would want to apprise the Americans of this diplomatic skulduggery, the British could not at the same time easily admit that they were intercepting and looking at US messages passing through British telegraph networks (which is how messages from Copenhagen were routed). Fortunately, the message had to be sent on to the German legation in Mexico City, and the British managed to spirit away a copy of the message on its arrival there – which was then shown to the American authorities.

After the Armistice in 1918 Room 40's necessity waned. The following year the unit – by now more properly known as NID25 – combined with its army equivalent MI1b to form the Government Code and Cypher School, later located at Bletchley Park; this in turn was renamed after the Second World War Government Communications Headquarters (GCHQ) after the Second World War, and relocated to Cheltenham.

No comments:

Post a Comment