by DAVID KAHN
T.O.D. also drew upon the pool of scientific knowledge of the Office of Scientific Research and Development. Each month, Shaw and his technical aides met with a bevy of physics and chemistry Ph.D.’s in the board room of Harvard’s Mallinckrodt Chemical Laboratory. Arthur B. Lamb, professor of chemistry at Harvard and editor of the Journal of the American Chemical Society, presided over a group that included Robley D. Evans, a physicist at the Massachusetts Institute of Technology, Harris M. Chad well, a physical chemist at Tufts University, Warren C. Lothrop, an organic chemist at Connecticut’s Trinity College, S. Edward Eaton, a chemist at Arthur D. Little, Inc., and George Richter, an expert in inks and papers with the Eastman Kodak Company. In addition, Sanborn C. Brown, a young faculty member of M.I.T. who worked as a “free physicist” for the O.S.R.D.’s National Defense Research Committee, unraveling puzzlers that had baffled other agencies, such as ways of de-arming booby traps and the causes of mysterious accidents to dive-bomber pilots, attacked some of the toughest technical problems faced by Censorship.
The primary detection of clandestine communications took place in the censorship field stations. Largest of all was New York’s, filling a huge building on lower Eighth Avenue. About 4,500 postal examiners scanned the snowdrifts of mail that piled onto their desks each day. They excised all matter that might have injured the Allied war effort, and they looked closely for traces of hidden messages. Censorship had catalogued the occupations and hobbies of its examiners. A balance sheet would be given to an accountant to see whether it made financial sense; an amateur horticulturist could tell whether a discussion of tulip beds rang true. Once an examiner in New York was perturbed by a letter from Germany to a prisoner of war in the United States, saying that Gertrude was developing into a swimming champion and listing the times of her victories. He consulted an amateur swimmer in the office, who reported that the speeds were impossible. Further investigation revealed that the times actually indicated the speed of a new fighter plane, given by a war worker who could not resist bragging. The factory was later bombed. Censors in a political section looked for clues to hoards of vital material that might be bought by the Allies to preclude the Axis from getting it. An economic section extracted remarks about shortages and living conditions to help build up pictures of national economies. Letters in uncommon languages went to a language identification section, which obtained translators for such esoteric tongues as Ladino, a mixture of Hebrew and 15th-century Spanish spoken only by the 30,000 Sephardic Jews in colonies in Spain, the Balkans, and Latin America.
Floor examiners passed all messages with peculiar wording, odd-looking marks, or other suspected indications to the security division, which had two sections to examine steganograms concealed in the two basic ways—linguistically and technologically. These were the code and cipher section for the linguistic steganograms and the laboratory section for the technological. Both were linked to T.O.D. by a security assistant who implemented T.O.D.’s instructions and passed the more recalcitrant problems back to Washington. The 70 examiners in the New York code and cipher section occupied about half the 14th floor, with some of the more expert people constituted as a specialist group. About 30 technicians tested for secret inks in the laboratory next door.
Linguistically concealed messages fall into two general categories, the semagram and the open code. There are three kinds of open code: the jargon code, the null cipher, and geometrical systems like the Cardano grille. In the jargon code, an apparently innocuous word stands for the real term in a text contrived to seem as bland and as innocent as possible. Jargon codes can range from the most informal sort of code to a full code list. They begin with mere allusions to mutually known events and persons—“I visited the man you had dinner with last week.” They continue through double meanings that would be easily understood by the recipient, as one criminal’s referring to another’s arrest by saying “Joe went to the hospital.” They culminate in a prearranged table of artificial meanings. Jargon has been popular since the dawn of cryptography. The Chinese employed it; the oldest papal code is the 14th-century use of EGYPTIANS for Ghibellines and SONS OF ISRAEL for Guelphs; in the 17th century a French code consisted entirely of such jargon expressions as GARDEN for Rome, ROSE for the pope, PLUM TREE for the Cardinal de Retz, WINDOW for Monsieur the king’s brother, and STAIRCASE for the Marquis de Coeuvres. It is clear that skillful application of jargon’s literary veneer requires no little finesse!
Censorship defends itself against this ruse by a feel for stilted or heavy-handed language and by a healthy skepticism concerning subject matter. The standard story about jargon comes from World War I. A British censor grew suspicious of the enormous orders for cigars wired each day—mostly from port towns—by two “Dutch businessmen.” One day Portsmouth called for 10,000 Coronas; the next day Plymouth and Devonport craved large quantities of stogies; then Newcastle succumbed overnight to the tobacco habit. It seemed as though all the males in the coastal population of England had suddenly and simultaneously developed an irresistible addiction to the weed, so inexhaustible was the demand for cigars. At the suggestion of the censor, a check was made; the two businessmen proved to be German spies, and their orders an open code, in which, say 5,000 Coronas for Newcastle meant five cruisers lying in that port. On July 30, 1915, the two—Haicke P. M. Janssen and Wilhelm R. Roos—were executed at the Tower of London by a firing squad whose triggers were really pulled by an alert censor.
So long as jargon codes can pass unnoticed, they are, naturally, safe. But they nearly always give up their secrets soon after detection. Paradoxically, the less suspect they are, the easier they are to solve when discovered: because the more they throw the burden of communication on the context, the more information the context contains that may be used as a lever to pry out the secret meaning. Thus in World War II, a series of letters evidencing a legitimate if somewhat neurotic concern for dolls apparently eluded censorship. Suspicion was drawn to them when one was returned from Buenos Aires, marked “Unknown at this address,” to the person marked as the sender, a woman in Portland, Oregon. Not having sent it, she brought it to the F.B.I. “I just secured a lovely Siamese Temple Dancer,” the letter said in part, “it had been damaged, that is tore in the middle. But it is now repaired and I like it very much. I could not get a mate for this Siam dancer, so I am redressing just a small ordinary doll into a second Siam doll….” Then other doll letters, all in the same disconnected feminine style filled with typographical errors, were intercepted by censors. “A broken doll in a hulu grass skirt will have all damages repaired by the first week of February,” and “The broken English dolls will be in a doll hospital for a few months before repairs can be completed. The doll hospital is working day and night.”
T.O.D. and F.B.I. cryptanalysts soon determined that the dolls represented warships in a jargon code, each kind a specific type of ship. The innocent chatter assumed sinister overtones: “Light cruiser Honolulu will have all damages repaired by the first week of February.” “The damaged English warships will be in a shipyard for several months before repairs can be completed. The shipyard is working day and night.” “I just secured information of a lovely aircraft carrier, it had been damaged, that is torpedoed in the middle. But it is now repaired [the “and I like it very much” just adds to the air of innocence]. A second carrier could not be obtained, so another ship is being converted to an aircraft carrier.” But no clue as to the writer emerged until she used as a return address a woman with whom she had had a spat. When an F.B.I. agent checked with this woman to ask if she knew who might have been using her address, she named Mrs. Velvalee Dickinson, who ran an exclusive doll shop on New York’s Madison Avenue. A Japanophile and friend of prominent Japanese, she was later found to have received thousands of dollars from Japanese officials. She was charged with espionage, which could have brought the death penalty, but was allowed to plead guilty to the lesser charge of violating wartime censorship regulations by illegally using codes in international com
munications and was sentenced to ten years in jail and a $10,000 fine.
A second type of open code is the null cipher. Only certain letters or words of a null cipher’s text are significant; for example, every fifth word or the first letter of every word, with all the other letters and words serving as nulls to produce the disguise. These usually sound even more strained than the jargon code. Even two of the better examples, sent by the Germans during World War I, have that “funny” sound that invariably accompanies them. The first, disguised as a press cable, read:
PRESIDENT’S EMBARGO RULING SHOULD HAVE IMMEDIATE NOTICE, GRAVE SITUATION AFFECTING INTERNATIONAL LAW. STATEMENT FORESHADOWS RUIN OF MANY NEUTRALS. YELLOW JOURNALS UNIFYING NATIONAL EXCITEMENT IMMENSELY.
The initial letters spell out Pershing sails from N. Y. June 1. The second message, apparently sent as a check on the first, beaded the same content on the second letters of each word:
APPARENTLY NEUTRAL’S PROTEST IS THOROUGHLY DISCOUNTED AND IGNORED. ISMAN HARD HIT. BLOCKADE ISSUE AFFECTS PRETEXT FOR EMBARGO ON BYPRODUCTS, EJECTING SUETS AND VEGETABLE OILS.
Whoever the sender was, his ingenuity was expended in vain, since Pershing actually sailed May 28.
Most null ciphers in World War II were used not by spies, but by otherwise loyal Americans who could not resist trying to “beat the censor.” Servicemen in particular attempted to sneak information about their whereabouts to families who otherwise would quite literally not know where in the world their soldier boy was—even though such attempts endangered the serviceman’s own life.
One such system, though elementary, brought deserved bewilderment instead of clarification to its intended recipients. A young GI, following a prearranged system with his parents, tried to tell them he was in Tunis by using first T, then U, then N, I, and S as his father’s middle initial in successive letters home. Unfortunately, he forgot to date them and they arrived out of order. The frantic parents wrote that they had looked and looked through their atlas but couldn’t find Nutsi anywhere! Attempts of this sort grew so frequent by 1943 that the Navy had to warn its sailors that these “family codes,” which were usually solved quite easily, could lead to severe penalties for the users.
The third kind of open code is the geometrical. A Cardano grille places the message-bearing words in fixed positions on a page. The significant words can be placed at intersections of the lines of a geometrical diagram of specified dimensions. In the 1600s, Sir John Trevanion, a Cavalier awaiting trial and almost certain execution by Cromwell’s forces, noted the third letter after each punctuation mark in a letter that his jailers had scrutinized before giving him and learned that Panel at east end of chapel slides. He disappeared during vespers. And in World War II, captured U-boat officers spelled out secret messages in their letters home by breaking the flow of script after each significant letter. An alert censor noticed that the minute gaps did not occur in natural places, as after syllables. The hidden messages described Allied anti-submarine tactics and technical U-boat faults. Some outlined escape plans—which were, of course, foiled.
The second category of linguistically concealed messages is the semagram (from the Greek “sema,” for “sign”). A semagram is a steganogram in which the ciphertext substitutes consist of anything but letters or numbers. The astragal of Aeneas the Tactician, in which yarn passing through holes representing letters carried the secret message, is the oldest known semagram. A box of Mah-Jongg tiles might carry a secret message. So might a drawing in which two kinds of objects represented the dots and dashes of Morse Code to spell out a message. The New York censorship station once shifted the hands and altered the positions of the individual timepieces in a shipment of watches lest a message be concealed in it.
The examination of the linguistically concealed messages—or, more correctly, those suspected to be such—was largely a frustrating experience. Often the examiner could not tell whether or not a message was hidden beneath the awkward or illiterate or misspelled writing. And even if he felt certain, solution often eluded him. He usually had only one message to work on, and no probable words. Early in the war, censorship practice even forbade working on a suspected cryptogram more than half an hour, on the theory that if the cryptanalyst hadn’t gotten it by then, he’d never get it. These unsolved messages posed a difficult problem to the censors. Presumably they were carrying contraband information and so should be banned. But, in the absence of solution, no proof of this existed, and so the letter could not be mutilated. Sometimes this was done anyway, to destroy the suspected code.
Technological steganography early in the war consisted almost exclusively of invisible inks. This is truly an ancient device. Pliny the Elder, in his Natural History, written in the first century A.D., told how the “milk” of the tithy-mallus plant could be used as a secret ink. Ovid referred to secret ink in his Art of Love. A Greek military scientist, Philo of Byzantium, described the use of a kind of ink made from gall nuts (gallotannic acid), which could be made visible by a solution of what is now called copper sulfate. Qalqashandi described several kinds of invisible ink in his Şub al-a’ shā. Alberti mentions them. The Renaissance employed them in diplomatic correspondence. About 1530 a book was printed with panels in invisible ink; if these pages were dipped in water, the message would appear; this could be repeated three or four times. Porta devoted Book XVI of his Magia Naturalis to invisible writing.
The common inks are of two kinds: organic fluids and sympathetic chemicals. The former, such as urine, milk, vinegar, and fruit juices, can be charred into visibility by gentle heating. Despite their antiquity and their minimal protection, they are so convenient that they were used even during World War II. Count Wilhelm Albrecht von Rautter, a naturalized American who was spying on his adoptive country for his native Germany, ran out of his good secret ink and had to use urine.
Sympathetic inks are solutions of chemicals that are colorless when dry but that react to form a visible compound when treated with another chemical, called the reagent. For example, when a spy writes in iron sulfate, nothing will be visible until it is painted over with a solution of potassium cyanate, when the two chemicals will combine to form ferric ferrocyanide, or Prussian blue, a particularly lovely hue. The colorless writing of lead sub-acetate will turn into a visible brown compound when moistened with sodium sulfhydrate. Copper sulfate can be developed with ammonia fumes, and it may have been this chemical that was used for the secret writing on the handkerchief of George Dasch, leader of the eight Nazi spies who landed by submarine on Long Island in 1942 to blow up American defense plants, railroad bridges, and canal locks. The red letters that appeared as if by magic when the pungent ammonia reached it spelled out the names and addresses of a mail drop in Lisbon and of two reliable sources for help in the United States. Each of the eight saboteurs had also been given a watertight tube containing four or five matchsticks tipped with a grayish substance that served as a ready-made pen-and-secret-ink. The trick in concocting a good secret ink is to find a substance that will react with the fewest possible chemicals—only one, if possible, thus resulting in what is called a highly “specific” ink.
A drawing of the San Antonio River that conceals a secret message (solution in Notes)
To test for secret inks, censorship stations “striped” letters. The laboratory assistant drew several brushes, all wired together in a holder and each dipped in a different developer, diagonally across the suspected documents. The developers were wide-spectrum, picking up even such substances as body oils, so that fingerprints and sweat drops often showed up. On the other hand, they missed some specific inks. A bleaching bath removed the stripes. Letters were also checked by infrared and ultraviolet light. Writing in starch, invisible in daylight or under electric light, will fluoresce under ultraviolet. Infrared can differentiate colors indistinguishable in ordinary light and so can pick up, for example, green writing on a green postage stamp. The censorship field stations tested all suspicious letters and a percentage of ordinary mail picked
at random, and sometimes all letters to and from a certain city for a week to see if anything suspicious turned up. During the war, about 4,600 suspicious letters were passed along to the F.B.I. and other investigative agencies; of these 400 proved to be of some importance.
Problems that would not yield to the crude approach of the field stations went back to the T.O.D. laboratory. Here, amid Bunsen burners and retorts, Pierce and Breon, aided by an expert photographer and laboratory technicians, cooked up reagents that would reincarnate the phantom writing. Better equipped and more deeply versed in the nuances of sympathetic inks than the mass-production workers of the field stations, they had received a great stimulus from contact with one of the great secret-ink experts of the world, England’s Dr. Stanley W. Collins, who had conducted this battle of the test tubes in two World Wars; he spoke at the Miami Counter-Espionage Conference in August, 1943. T.O.D. soon learned that Nazi spies were taking countermeasures to frustrate the iodine-vapor test and the general reagent.
One was to split a piece of paper, write a secret-ink message on the inner surface, then rejoin the halves. With the ink on the inside, no reagent applied to the outside could develop it! The technique came to light when one German spy used too much ink and the excess soaked through. Sanborn Brown, the M.I.T. physicist, got two inmates of a local jail to explain how two sheets of parchment could be used to do the splitting. They had been caught misapplying the talent to one- and ten-dollar bills, pasting one half of the tens to one half of the ones and passing them with the ten-dollar side up. The method is more an art than a science, for if the sudden tear is not done just right, the paper will shred. To read the message, the paper must be resplit, but it comes apart much more easily the second time.
