by DAVID KAHN
One way was called the “Swedish Roundabout” by the British. Sweden, which was neutral in favor of Germany, had since early in the war helped the German Foreign Office get messages past the British cable blockade by sending them as her own. British censorship detected this practice. When Sweden complained in the summer of 1915 that Britain was delaying her messages, Britain informed her that it had positive knowledge of the unneutral practice. The Swedish government admitted this and promised that it would no longer send any German messages to Washington. It did not. Instead, it sent them to Buenos Aires. Here they were transferred from Swedish to German hands and then forwarded to Washington. This was a circuitous route of about 7,000 miles, half of them in flat violation of the prerogatives of a nonbelligerent.
But the cable from Stockholm to South America touched at England. Germany feared that British censorship might recognize the German codegroups in the Swedish messages and would stop the dispatches. So the German Foreign Office disguised the codegroups by enciphering them. This was done with Code 13040 in messages to Latin America and to Washington. Unfortunately for the Germans, the superencipherment did not obliterate all traces of the underlying code, which employed a distinctive mixture of 3-, 4-, and 5-digit codegroups. These traces aroused the suspicions of the ever-alert Room 40; it resolved the superencipherment, and Code 13040 reappeared. Room 40 then looked closely at other official Swedish messages. Many of them proved to be German as well; concealed under one superencipherment, for example, they found Code 0075. But this time England entered no protest. Hall perceived that it was more advantageous to listen to what the Germans were saying than to stop them from talking.
The second route that Zimmermann used was of such simplicity, perfidy, and barefaced gall that it probably remains unequaled in the annals of diplomacy. It had its inception in the pompous mind of Colonel Edward M. House, President Wilson’s alter ego and a major exponent of personal diplomacy. On one of his missions to Europe in 1915, House arranged to have coded reports from the embassies cabled directly to him, bypassing the State Department. When, on December 27, 1916, Ambassador Bernstorff discussed a new peace attempt by Wilson with House, he pointed out that the chances would be improved if his government could communicate directly with Wilson through House. House checked with the President. The next day Wilson permitted the German government to send messages in its own code between Washington and Berlin under American diplomatic auspices—an arrangement that was, at best, simpleminded, and that, furthermore, contravened the accepted international practice of requiring the messages to be submitted in clear for transmission in American code.
Germany availed herself of this arrangement to make America seal her own doom by letters she herself bore. Under the aegis of American sovereignty, Zimmermann sent his message striking at that sovereignty. It was delivered to the American embassy in Berlin at 3 p.m. January 16. It could not go direct to Washington, but had to be sent first to Copenhagen—and then to London. Only from there could it go to Washington. Consequently Britain seized this copy as well. Room 40 was “highly entertained” at the sight of the German code in an American cable, but again did not protest.
With two copies of the same text helping to eliminate garbles, Montgomery and de Grey rammed into the cryptogram. De Grey, though at 30 the younger of the two, had been in Room 40 the longer. Slightly built, rather handsome, with dark hair and brown eyes and chiseled, movie-star features, an Eton graduate, he was descended from the peerage as the grandson of the fifth Baron Walsingham (no relation to Sir Francis Walsingham). He had worked for the prestigious publishing house of William Heinemann for seven years before the war, when he joined the Royal Naval Air Service. He came to Room 40 in 1915.
Soon after his work on the cryptogram that became known as the Zimmermann telegram, he left 40 O.B. to serve as head of the naval intelligence mission that Hall had sent to Rome. After the war, he became director of the Medici Society, a publishing house specializing in art prints. In 1939, his government remembered his World War I services, and he joined the cryptanalytic division of the Foreign Office, soon becoming deputy director. A man who listed as his recreations the odd threesome of shooting, gardening, and acting, he also enjoyed carpentry and was useful around the house. He died May 25, 1951, leaving two sons and a daughter.
Montgomery was 45 at the time of his work on the Zimmermann telegram. A Liverpool shipowner’s son who studied in private schools or under tutors in England, France, and Germany, he took a bachelor of divinity degree at Presbyterian College, London. But his health prevented an active pastorate and he became a member of St. John’s College at Cambridge University. He specialized in early church history, editing the Confessions of St. Augustine for the Cambridge Patristic Series and writing a study on the life and thought of the African father. His most memorable work, however, was as a translator. It was said of his translation of Albert Schweitzer’s The Quest of the Historical Jesus in 1910 that “no German work has ever been rendered into English so idiomatically and yet so faithfully.” A modest, reticent man, Montgomery entered the censor’s office in 1916, and later that year transferred to Room 40. Cryptanalysis so suited his aptitudes that after the war he continued the work in the Foreign Office, remaining there until his sudden death in October, 1930.
While in Room 40 his familiarity with Scripture unriddled a problem that had baffled most of the other staffers. A Sir Henry Jones had received a blank postcard from Turkey addressed to him at 184 King’s Road, Tighnabruaich, Scotland. Sir Henry knew that the card was from his son, who had been captured by the Turks, but Tighnabruaich is a small village, with no King’s Road and so few houses that no number would have been needed in any case. The card found its way to Room 40, where nobody seemed able to ascertain what Sir Henry’s son was trying to tell him. Finally Montgomery suggested a reference to chapter 18, verse 4, of one of the books of Kings. Second Kings shed no light, but First Kings revealed that “Obadiah took a hundred prophets, and hid them fifty in a cave, and fed them with bread and water.” Montgomery interpreted this to mean that Sir Henry’s son was safe with other prisoners but in need of food—and this proved to be the case.
But the solution of the Zimmermann telegram required more than a flash of inspiration. It demanded the reconstruction of Code 0075, a two-part code of 10,000 words and phrases numbered from 0000 to 9999 in mixed order. Since a code is, in a sense, a gigantic monalphabetic substitution, the establishment of plaintext equivalents is the “only” task involved. But where the cryptanalyst of cipher deals with only 26 such elements, the cryptanalyst of code must keep his eye on hundreds or thousands, whose characteristics, moreover, because of their reduced frequency, are much scantier and more diffuse than the sharply defined traits of letters.
Solution usually begins with the identification of the groups meaning stop. Groups that recur near the end of telegrams are likely candidates. The identification of stop or period is often aided because often only a few of the many code equivalents are employed. Code clerks, referring frequently to stop, come to memorize one or two of its codegroups; they then simply use these groups in encoding instead of hunting up a different one in the codebook. Indeed, cryptanalysts familiar with a given embassy’s messages can often tell when a new code clerk has been hired by the sudden efflorescence of new equivalents for stop!
The identification of the stops outlines the structure of the message. In English messages, nouns, as the subjects of sentences, will often appear directly after stops. In German, where the predicate often comes at the end of the sentence, the codegroup immediately preceding a stop may be a verb. Other clues come from the stereotyped expressions that diplomats so love in their dispatches: “I have the honor to report to Your Excellency….” Collateral information is of very great value.
The first tentative identifications are usually written in pencil for easy erasing, and such are called “pencil groups.” Eventually, further traffic confirms them and they become “ink groups.” Solution proceeds mu
ch more rapidly if a code is one-part. If codegroup 1234 represents a word beginning with d, then 5678 must represent one farther back in the alphabet; this both rules out some guesses and suggests others. Sometimes the meaning of a codegroup can be indicated rather precisely by its location between two ink groups. This is not possible with a two-part code, where the code and plain equivalents are matched in an absolutely arbitrary fashion. Code 0075 was of this type. It required more traffic for its solution than a one-part code, and the identifications came more slowly and with greater difficulty. It had been in service on the Continent for only half a year—not a very long time for a diplomatic code—and portions of many messages remained unreadable.
As more traffic came in (including now the messages to and from Bernstorff), Montgomery and de Grey, working night and day, filled in more and more groups, ever more rapidly. On January 28, de Grey brought Hall part of Bernstorff’s protest against Zimmermann’s plan of unrestricted submarine warfare, which, to the ambassador’s dismay, had been announced to him in message No. 157, the first part of the double-decker. Bernstorff argued vigorously against this plan, for he felt that it negated all his efforts to bring about a détente between the two countries and that it would drive the United States into the war on the side of the Allies.
And in fact, on February 3 Wilson announced to Congress that he was breaking diplomatic relations with Germany, as he had said he would the previous April if Germany continued its course of submarine warfare. Though he added that “only actual overt acts” on Germany’s part would make him believe that she really would sink neutral vessels on the high seas, it must have seemed to the war-weary Allies that now, at last, within a few days or a fortnight at most, the United States would enter the war. Day by day, they awaited the final inevitable step.
While waiting, Room 40 continued its work on Code 0075. De Grey had taken to Hall Bernstorff’s message giving details of his interview with Wilson severing relations. Recovered codegroups were substituted into the Zimmermann telegram, and on February 5 Hall was able to show a more fully solved version of it to Lord Hardinge at the Foreign Office.
Hall had realized from the first day that Montgomery had brought him the first sketchy solution of the Zimmermann telegram that he had in it a propaganda weapon of titanic proportions. Exposure of this German plot directed against the United States would, in the present circumstances, almost certainly compel that nation to declare war on Germany. This was an immensely strong argument for showing it to the Americans. But for the moment, at least, even stronger considerations militated against it. First, Room 40 and its cryptanalytic capabilities was one of Britain’s darkest secrets. How could she disclose the message without Germany’s guessing that her codes were being read? Britain might minimize the risk by hinting that the plaintext had been stolen, but the danger would still remain that Germany would suspect the truth, change her codes, and deprive Britain of her most valuable intelligence. In the second place, to reveal the message, Britain would have to admit that it had been supervising the code telegrams of a neutral: Sweden. It would not require much wit for the Americans to surmise that England might also be supervising the code telegrams of another neutral: the United States, which, like Sweden, was working as a messenger boy for the Germans and had, in fact, transmitted this very message. This realization would both embarrass and anger the United States and would not conduce to pro-Allied feelings. In the third place, the solution was still not complete. The missing portions would inevitably raise doubts about the validity of the solution and so weaken its impact. Perhaps the British had failed to solve a word like “not” that would completely alter the sense, the arguments would run. Perhaps the British had not even correctly solved the portion that they were offering as evidence of German duplicity. Moreover, the gaps would shout “codebreaking,” preventing any subterfuges about captured codes or a stolen message and exposing the very secret Britain sought to conceal.
But the most powerful argument against disclosure of the German plot, with all the attendant difficulties, was that events might make it unnecessary. Relations had been severed between Germany and the United States. American public opinion seemed to be turning increasingly against Germany. Shipping dared not sail; ports were congested; men were laid off; business languished. Bitterness was growing. It seemed only a matter of a short while until the declaration of war. And so the British continued to wait, and to hope.
Hall, however, while waiting for events to dictate, did not remain idle. His job was only half done if he merely solved the Zimmermann telegram without making it ready for use by his government. Consequently, he conceived a plan that at one-stroke might resolve the three difficulties connected with the telegram’s exposure, in what still appeared the unlikely event that that might be necessary. He reasoned that the telegram as received in Mexico would differ in small but significant details from the telegram as sent from Berlin. The date would almost certainly be different, and probably the serial number as well. The preamble addressed to Bernstorff ordering him to forward the message would of course be omitted. If Hall could produce the copy from Mexico, perhaps the Germans would spot these slight variations and infer that the plaintext had been betrayed on the American continent and would not change their codes. Other collateral details might confirm a tale of a Mexican theft to the Americans. Moreover, Room 40 perhaps knew, from its numerous solutions of German messages via the Swedish roundabout, that the German mission in Mexico had not used Code 0075 and probably did not hold it. Bernstorff might then have had to re-encode the Zimmermann telegram in another code, which Room 40 might have solved more completely than 0075 and which might therefore enable it to fill in the missing portions in its solution.
On February 5, therefore, Hall began trying to get a copy of the Zimmermann telegram as received in Mexico. An English agent known only as T obtained from the Mexico City telegraph office a copy of the message that Bernstorff had sent to Eckardt by Western Union. Soon Hall had it.
It proved him right in every one of his assumptions. Eckardt did not have Code 0075, and so Bernstorff had had to recode the dispatch in one that Eckardt did have. This was Code 13040, which was an older and simpler code than 0075 and whose superencipherment had led to the discovery of the Swedish roundabout. It had been distributed to German missions in Central and South America between 1907 and 1909 and to Washington, New York, Havana, Port-au-Prince, and La Paz in 1912. Its basic repertory contained about 25,000 plaintext elements with a fair number of homophones—Bernstorff’s telegram alone employed six different groups for zu—and proper names took up a huge section of 75,000 codenumbers. But Code 13040 was a cross between one-part and two-part codes. In the encoding section, blocks of several hundred codenumbers in numerical order stood opposite the alphabetized plaintext elements, but the blocks themselves were in mixed order. A skeleton code, made up from a few groups from Bernstorff’s encoding, will illustrate this:
Encoding decoding
13605 Februar 5144 wenigen
13732 fest 5161 werden
13850 finanzielle 5275 Anregung
13918 folgender 5376 Anwendung
17142 Frieden 5454 ar
17149 Friedenschluss 5569 auf
17166 führung 5905 Krieg
17214 Ganz geheim
17388 Gebeit
4377 geheim
4458 Gemeinsame
The solution of such a hybrid code stands midway in difficulty between the two pure types: harder than a one-part code but easier than a two-part. The large orderly segments considerably help the cryptanalyst, though his guesses are not as delimited as in a one-part code. For example, the cryptanalyst could not assume, as he could in a one-part solution, that a codegroup for Krieg will be higher in number than the codegroup for Februar. But if he knows that Februar is 13605 and finanzielle is 13850, he will know that the codegroup for fest must almost certainly fall somewhere between the two. His identifications thus come with greater speed and certainty.
Owing to t
his weakness, and to the fact that they had had all of the war to work on a great volume of messages, the codebreakers of Room 40 had recovered most of Code 13040’s commonly used groups. They could consequently read all or nearly all of Bernstorff’s message to Eckardt, and in those few places where a rare proper name or syllable might have been used for the first time, the partial alphabetical arrangement afforded a strong check on their guesses. This eliminated the problem of having only a partial solution. In addition, it confirmed their almost-complete solution of the original Berlin-to-Washington message and added a few new values to their reconstruction of Code 0075.
The Zimmermann telegram as re-encoded in Washington into Code 13040 and forwarded to Mexico
The cryptanalysts also found the slight changes in heading that Hall had foreseen. Bernstorff had deleted the Foreign Office preamble and substituted one of his own: “Foreign Office telegraphs January 16: No. 1. Most Secret. Decode yourself.” He replaced the Berlin-Washington serial number with a Washington-Mexico City serial number, which was 3. And finally, his message was dated January 19, which, due to the numerous steps in the complicated transmission routes, differed from the January 16 date that the original German text bore.
Fairly early in February, it seems, Hall was ready. With a stroke bordering on genius, he had done his job. His must stand as one of the most subtly dissembling moves in the whole history of espionage. It was now possible to give the message to the Americans, should that prove necessary, with as little risk as possible to Britain’s intelligence sources. But though Hall had covered his tracks fairly well, it remained possible that the Germans might guess the truth. Events might yet make it unnecessary to chance this. So Britain held the message and waited.
And waited. The days passed. On the Western Front the lifeblood of the Empire and of the French republic trickled into the earth. The armies shuddered in mortal combat. Still there came no sign that America was going to enter the war. Though it seemed that Germany’s announcement of unrestricted torpedoings of American ships had made, as Bernstorff himself had warned in cables read by Room 40, “war unavoidable,” the American President seemed unable to do what the British thought that honor, self-respect, and the whole course of recent actions made obligatory. Even Ambassador Page, a long-time friend of the President and a wholehearted sympathizer with the Allied cause, was irked enough to note in his diary, “The danger is that with all the authority he wants (short of a formal declaration of war) the President will again wait, wait, wait—till an American liner be torpedoed! Or till an attack is made on our coast by a German submarine!” Evidently Wilson was waiting for the “overt acts” that he had mentioned in his address to Congress. But perhaps Germany would not actually be so rash as to torpedo American ships and thereby—Britain thought—cut her own throat. More days passed. The Germans did nothing. Tension mounted. The situation was, a British diplomat in America reported, “much that of a soda-water bottle with the wires cut but the cork unexploded.”
