THE CODEBREAKERS

by DAVID KAHN

  The redcoats used invisible ink even earlier than the Americans. Only a few days after the Battle of Lexington, British headquarters in Boston received a secret-ink letter which revealed some of the military plans of the New England patriot forces. “… the first movement will be to make a feint attack upon the Town of Boston,” the invisible portion read in part, “& at the same time to attempt the castle with the main body of their Army.” The handwriting shows the document to be from Benjamin Thompson, a hated Tory, who, after a series of colorful adventures, became Count Rumford of the Holy Roman Empire and a widely known scientist. He used gallotannic acid for his ink, which the British developed by ferrous sulphate—a procedure described by Porta, from whose Natural Magick Thompson, who had been avid for science since his teens, had probably borrowed it.

  When it came to ciphers, the British provided themselves with a veritable menagerie of systems. Sir Henry Clinton, British commander in New York, had a small one-part nomenclator; he also had a monoalphabetic substitution in which a = 51, b = 52, c = 53, and so on. He had a truncated alphabet tableau of twelve lines. He even had a pigpen cipher. Still other specimens inhabited this cryptographic zoo, but the only one Clinton is known to have used in the early part of the war is a degenerate form of grille called the dumbbell cipher, from the hourglass shape of its one large hole.

  In the summer of 1777, Clinton had to inform General John Burgoyne, driving south down the Hudson in an attempt to cut the colonies in two, that he would have trouble pushing north to a meeting because his superior, Sir William Howe, had taken most of his troops to Philadelphia. On August 10, Clinton wrote as part of his secret message a heartfelt Sir W’s move just at this time the worst he could take. His cover-text for this portion, which necessarily had to include many of these words, stated just the opposite: SIR W’S MOVE JUST AT THIS TIME HAS BEEN CAPITAL; WASHINGTON’S HAVE BEEN THE WORST HE COULD TAKE IN EVERY RESPECT. But it was patently absurd for a commander to assert that the loss of his troops was “capital”; the example throws a sharp light on the weakness of the grille. Whether the message got through or not, and if it did whether it disheartened Burgoyne, is unknown. It is known that, deprived of the help of Clinton’s column from the south, he lost the Battle of Saratoga, which helped determine the ultimate outcome of the Revolution.

  While code and cipher systems proliferated throughout the Revolution, cryptanalysis hibernated. The basic reason seems to be that, with the exception of an infrequent episode like that of the Church cipher, no cryptograms were intercepted. It was not until the war neared its end that enough messages were captured to make recurrent cryptanalyses possible. Most of the messages were solved by James Lovell, a member of the Continental Congress who may be called the Father of American Cryptanalysis.

  Lovell, born in Boston on October 31, 1737, graduated from Harvard in 1756 and then taught for eighteen years in his father’s South Grammar School in Boston. His father was a fervid Loyalist, but James was named as the first orator to commemorate the Boston Massacre, and in 1775 was arrested by the British as an American spy. After his exchange, he was elected a delegate to the Continental Congress. He took his seat in 1777, and promptly distinguished himself for zeal and industry, particularly on the Committee on Foreign Affairs. It is said that he never once in the next five years visited his wife and children. He offered a design for the Great Seal of the United States, which, however, was rejected. He quit Congress in April of 1782 and was appointed receiver of continental taxes in Boston, and, in 1789, naval officer for the district of Boston and Charlestown, the post he held until his death in 1814.

  He was regarded as gifted in intrigue and as a lover of mystery. Where he learned cryptology is not known, but as early as 1777, he was endorsing Arthur Lee’s proposal that the Committee of Secret Correspondence use a dictionary as a codebook. Two years later, he urged Major General Horatio Gates, whom he preferred over Washington as commander in chief, to “Ask Dr. Joseph Gardner, one of my best earthly friends, to let your clerk copy an alphabet which he had from me.” The system was a Vigenère using numbers instead of cipher letters; Lovell keyed a letter in it to Gates with the name JAMES. The same system, with key CR, served him in enciphering letters to John and Abigail Adams in 1781. The following year, after a mail robbery had compromised the nomenclator of 846 elements used by Virginia’s delegates to Congress, one of them, the acid-tongued Edmund Randolph, proposed to another, James Madison, that they employ “the cypher which we were taught by Mr. Lovell. Let the keyword be the name of the negro boy who used to wait on our common friend Mr. Jas. Madison.” This name was CUPID, the system a numerical Vigenère. It is significant that Lovell was here popularizing a system that was relatively obscure and little used, but that was then the only type that lay beyond the known limits of cryptanalysis. Later, however, errors compelled its abandonment.

  Lovell’s successes in solution came at the most opportune time. In the fall of 1781, Lord Cornwallis, Britain’s second-in-command in America, had moved his troops north from the Carolinas to Virginia. He was convinced that that colony had to be taken before the southern colonies could be held, and he marched down the James River toward the coast in the hope of receiving reinforcements by sea from his chief, General Clinton, in New York. He planned to reduce Virginia, conquer the Carolinas, and quell the revolution for His Britannic Majesty, King George III.

  It was at this juncture that the American commander in the South, Nathanael Greene, sent to Congress, as he had done before, some intercepted redcoat cryptograms that no one in his headquarters could read, enclosing them in a general report. The British correspondence was among Cornwallis and several of his subordinates.

  Greene’s report was read in Congress September 17. Four days later, Lovell had solved the enclosures. A few were in a simple monalphabetic substitution, but most were in a bastard system that combined the poorest features of mon- and polyalphabeticity. A single numerical cipher alphabet enciphered four to ten lines monalphabetically, and then shifted to provide new ciphertext equivalents. For example, the positions were as follows for lines 1, 10, and 14 of the first page of the first letter:

  Any number above 30 was a null, and these were sprinkled freely throughout the message. Changes in alphabets were signaled by both a bracketlike mark and a series of four to seven nulls. No pattern appeared in the shifting; presumably it followed a list prearranged by the correspondents.

  Unfortunately, the tactical situation had changed too much for the information in the Carolina intercepts to be of much good. But the keys that Lovell had recovered might possibly prove valuable some time in the future, and so he took the precaution of writing Washington: “It is not improbable that the Enemy have a plan of cyphering their letters which is pretty general among their Chiefs. If so, your Excellency will perhaps reap Benefit from making your Secretary take a Copy of the Keys and observations which I send to General Greene, through your Care.”

  Lovell could not have been shrewder. The system that he had solved was, as he had guessed, also in service between Cornwallis and Clinton, who was the commander in chief of all British forces in America. Cornwallis had by now retired to Yorktown to await Clinton’s reinforcements. But Washington had encircled the town with 16,000 men, while the French admiral, the Count de Grasse, with 24 French ships of the line, barred relief by sea. On October 6, just after the French and American allies had driven a parallel close to the British lines, Washington wrote Lovell, “My Secretary has taken a Copy of the Cyphers, and by the help of one of the Alphabets has been able to decipher one paragraph of a letter lately intercepted going from L’d Cornwallis to Sir H’y Clinton.” The letter presumably gave Washington insight into conditions inside the British fortifications.

  Clinton, meanwhile, managed to maintain contact with Cornwallis by small boat. But the vessels that he sent out from New York on September 26 and October 3 were captured by the rebels. One of them had been driven ashore near Little Egg Harbor, where the Tory who was carrying
one set of dispatches hid them under a large stone before he was captured and brought to Philadelphia. “By means of a little address and a promise of a pardon,” as one American put it, he was persuaded to recover them. The search took at least two days, either because “the beach is so extensive and so many places like each other,” as the president of Congress, Thomas McKean, wrote Washington, or because the man was delaying. He still had not returned with them to Philadelphia by 3 p.m. October 13, nor, apparently, by the next morning. At that time, Lovell sent to Washington through a French officer what appears to have been a supplementary British system: “I found, as I had before supposed, that they sometimes use Entick’s Dictionary marking the Page Column and Word as 115.1.4. Tis the Edition of 1777 London by Charles Dilley.”

  James Lovell’s solution of a 1781 letter to Lord Cornwallis

  The Tory returned with the dispatches some time during October 14. Lovell attacked them at once and with immediate success, since he found to his joy that they were written in the same alphabets as the rest of the Clinton-Cornwallis correspondence. The more important message of the two that were apparently intercepted was the one sent in duplicate by Clinton on September 30 and received by Cornwallis on October 10. “My Lord,” it began, “Your Lordship may be assured that I am doing every thing in my power to relieve you by a direct move, and I have reason to hope, from the assurances given me this day by Admiral Graves, that we may pass the bar by the 12th of October, if the winds permit, and no unforeseen accident happens: this, however, is subject to disappointment, wherefore, if I hear from you, your wishes will of course direct me, and I shall persist in my idea of a direct move, even to the middle of November…”

  By the evening of October 14, Lovell was writing to Washington: “Since I wrote that Letter [of the morning], I have been happy in decyphering what the President of Congress sends by this Opportunity. The use of the same Cypher by all the British Commanders is now pretty fairly concluded. The Enemy play a grand Stake, May the Glory redound to the Allied Force under your Excellency’s Command!”

  This went out with a letter of the president of Congress, who told Washington: “My intelligence was true: the inclosed copies of two original letters from Sir Henry Clinton to Lord Cornwallis which I have in cyphers, and which have been faithfully decyphered by Mr. Lovell (whose key I had the honor to forward to you about a fortnight ago) more than prove the fact.”

  At the same time, McKean also sent the solutions to de Grasse, whose ships were to prevent Graves and Clinton from relieving Cornwallis. “The British General and Admiral seem to be desperate, and willing to risque all on the intended attempt,” he wrote de Grasse, adding prophetically, “If they fail it appears here that they are disposed to give up the contest for North America.” De Grasse continued to blockade Cornwallis and to watch for the British fleet. Five days after Lovell had completed his cryptanalytic exposure of British plans, Cornwallis surrendered. But victory was not quite complete. Washington recognized this when, on the following day, October 20, he received the copies of the solutions that McKean had sent him and lost “not an instant” in forwarding them to de Grasse. Doubly warned, the French admiral prepared for the British attack. On October 30, he scared off the English fleet and set the seal of final victory on the American War for Independence.

  With the coming of victory, the difficulties attendant upon the establishment of a new nation compelled the Founding Fathers not only to continue their secret communications, but to extend and improve them. In the fall of 1781, Robert A. Livingston, Secretary of Foreign Affairs, had forms printed that bore on one side the numbers 1 to 1700 and on the other an alphabetical list of letters, syllables, and words. They served as a convenient basis for correspondents to produce individual nomenclators by assigning the code numbers to the plaintext elements in whatever order they wished.

  They were widely used. Madison and Thomas Jefferson constructed a code on one of them in 1785, using it at least until 1793. It was in that year that Madison, vacationing in Fredericksburg, found himself staring at this en­lightening passage in a letter from Jefferson because he had left his key in Philadelphia: “We have decided unanimously to 130 … interest if they do not 510 … to the 636. Its consequences you will readily seize but 145 … though the 15…. ” Another code composed on the Livingston forms, endorsed “Mr. Monroe’s cypher,” was used by Monroe in 1805 when he was minister to England, by James A. Bayard in 1814 when he helped negotiate the treaty that ended the War of 1812, and as late as 1832 by President Andrew Jackson in letters to a diplomatic agent. It therefore seems to have been one of the first official codes of the United States under the Constitution.

  The nomenclator compiled in 1785 by Jefferson for use with Madison and Monroe

  Other emissaries used systems of secret communication while the America that they were representing was still little more than thirteen united colonies. Benjamin Franklin, in France in 1781, assigned consecutive numbers to each of the 682 letters and punctuation marks in a long passage in French to concoct a homophonic substitution cipher:

  One message began, I HAVE JUST RECEIVED A 14, 5, 3, 10, 28, 76, 203, 66, 11, 12, 273, 50, 14, … the numbers deciphering to neuucmiissjon. The double u was necessary because the French passage has no w. Plaintext e was represented by more than 100 different numbers. Another early representative, William Carmichael, minister in Madrid, seems to have made the first recorded suggestion for a standard American diplomatic cryptography. In a letter to Jefferson in Paris on June 27, 1785, he wrote: “It has long been my surprise that Congress hath not instructed those they employ abroad on this head [ciphers]: For this purpose a common cypher should be sent to each of their Ministers and Chargé Des Affaires.”

  Still other systems were used. Before they settled on the Livingston-form nomenclator, Jefferson and Madison agreed to use a French-English lexicon as a codebook. The Lee brothers, Arthur, Richard Henry, and William, corresponded from 1777 to 1779 in a dictionary code, probably the same Entick’s of 1777 that Clinton had used and Lovell discovered.

  The most far-reaching cryptogram in domestic American history used not one but three systems of cryptography. It served as evidence in the sensational trial for treason of the man who had lost the Presidency by a single vote in 1800 and who became Vice President instead—Aaron Burr.

  After killing Alexander Hamilton in a duel in 1804, Burr headed west, fired with his dream of carving out a colonial empire in the Southwest at the expense of Spain, with whom war then seemed imminent. Whether this empire was to be the United States’ or Burr’s was never clear. His military accomplice in this grandiose scheme was General James Wilkinson, who, unknown to Burr, was a paid Spanish agent. Though Burr had used a cardboard cipher disk with numbers for polyalphabetic substitution in 1800 and in a letter to his son-in-law in 1804, he and Wilkinson decided to combine into a single system of cryptography for their great work a symbol code in which, for example, a circle stood for “President,” a symbol cipher in which a dash represented a, and a dictionary code based on the 1800 Wilmington edition of the ubiquitous Entick’s. On October 8, 1806, as Wilkinson waited in camp at Natchitoches, Louisiana, a messenger arrived with a cipher letter in this system from Burr dated July 22, in which he outlined his final plans for the great adventure.

  Its exact wording will never be known. Wilkinson erased, altered, and redeciphered it time and again to suit his varying conveniences. In its final version, it began: “Your letter post marked 13th May is received. I have at length 263.13ed 176.3. and have 35.3 93.10ed….” It went on to tell how Burr was planning to move westward down the Ohio and the Mississippi with about 500 or 1,000 men to meet Wilkinson and “there to determine whether it will be expedient in the first instance to seize on or pass by Baton Rouge.” Wilkinson used it, not to meet Burr, but to double-cross him. He sent one of the decipherments to President Thomas Jefferson, who promptly ordered the breakup of Burr’s expedition.

  The former Vice President was arrested
and tried for treason, with Chief Justice John Marshall presiding. The letter formed one of the chief exhibits. Under cross-examination, Wilkinson brazenly admitted that he had changed the document to save himself from implication. At one point he averred that the decipherment was hasty, inaccurate, and done piecemeal; at another, that it was a careful, tedious, and lengthy bit of work. This sort of vacillation by the chief prosecution witness threw a reasonable doubt on Burr’s guilt, and the jury acquitted him. But the court of public opinion, roused by the evidence of the cryptogram, convicted him. For the remainder of his life Burr could never expunge the stain on his name that his enciphered message had helped place there.

  During these formative years, the black chambers of Europe did not disdain to read the communications of the fledgling nation just because it was weak and far away. As early as 1777 Britain’s black chamber was developing American letters in secret ink: the British chemists marked two of them, apparently sent between Paris and London, with “all written in white ink” and “R15th.” One has Benjamin Franklin’s name in the margin.

  The following year, a letter from an American businessman in London to Franklin’s secretary in Paris was solved. In 1780, Francis Willes, Bishop Willes’ son, solved a packet of letters from the Marquis de Lafayette, then in Philadelphia, to France’s Minister of Foreign Affairs, the Count de Vergennes. One dispatch, of May 20, in an extensive two-part nomenclator, proved to be a long and informative report summarizing the overall situation as Lafayette saw it—the Continental currency has greatly depreciated, New York can be taken if the French troops arrive in time, Washington is thinking of conquering Canada, and the ability, honesty, and constancy of “mes amis Améri cains.” The packet had been thrown overboard when the vessel carrying it was captured by the British, but some tars jumped in and retrieved it. The solutions were shown to King George III, who may have obtained thereby some valuable clues as to how to conduct his American war.