THE CODEBREAKERS

by DAVID KAHN

  It may be that Dee had somehow obtained the mysterious manuscript ×possibly from the Duke of Northumberland, who pillaged many religious houses when Henry VIII broke up the monasteries, and with whose family Dee was associated), was told or assumed that it was Bacon’s, tried to solve it, and, failing, made a gift of it to Rudolf, perhaps on behalf of Elizabeth, for whom he was serving at Rudolf’s court as a secret political agent. The English physician and writer Sir Thomas Browne ×who, incidentally, first used the word “cryptography” in English) related that Dee’s son, “Dr. Arthur Dee ×speaking about his father’s life in Prague) told about … book containing nothing but hieroglyphicks, which book his father bestowed much time upon, but I could not hear that he could make it out.” The comment may refer to this very manuscript.

  This is conjectural, however. What is certain is that Kircher deposited the manuscript in the Jesuit Collegium Romanum, and that in 1912 an American rare book dealer named Wilfred Voynich purchased it for an undisclosed sum from the Jesuit school of Mondragone in Frascati, Italy.

  Eager to read the manuscript, Voynich generously supplied photostats to anyone who seemed likely to solve it. Many tried. Botanists thought they could read it by identifying the plants and assuming their names as probable words; one difficulty here was that most of the flora were imaginary. Astronomers recognized stars such as Aldebaran and the Hyades but could not force a solution. Philologists tried the methods used for reading lost languages and failed. Cryptanalysts observed characteristics in common with ordinary ciphers and found that it resisted their well-tried techniques. Voynich heard from many specialists who were interested in the problem: palaeographer H. Omont of Paris’ Bibliothèque Nationale, who had written a learned article about a 15th-century cryptographic manuscript on alchemy; Professor A. G. Little, a foremost authority on Bacon; a Harvard professor of anatomy; George Fabyan of the Riverbank Laboratories; the vice president of the Royal Astronomical Society in London; even Dom Aidan, Cardinal Gasquet, prefect of the Vatican Archives, who offered to help get any documents from those archives that might throw light on the problem. Almost certainly many of these and others tried to solve the cryptogram. Among the others in 1917 was Captain John M. Manly, then second in command of Yardley’s MI-8. He had cracked the Lother Witke cipher that had baffled all his colleagues but, like the others, with the Voynich manuscript he failed. And so did Yardley.

  In 1919, some of Voynich’s reproductions found their way to William Romaine Newbold, a professor of philosophy and former dean of the Graduate School at the University of Pennsylvania. Newbold, 54, a brilliant man who had stood first in his class of 1887 at Pennsylvania, had wide-ranging interests, many of which had in common an element of the occult—spiritism, the Gnostics, the Great Chalice of Antioch, supposed by some to be the actual chalice of the Last Supper, which is known in legend as the Holy Grail. He knew many languages and later became proficient in crypt-analysis: in 1922, Theodore Roosevelt, Jr., then Assistant Secretary of the Navy, thanked him for his “time and trouble in deciphering espionage correspondence that had baffled the Departments here in Washington.”

  Newbold saw microscopic shorthand symbols in the macroscopic characters of the manuscript text and began his decipherment by transliterating them into Roman letters. A secondary text of 17 different letters resulted. He doubled all but the first and last letters of each section: the secondary text oritur would become the tertiary text or-ri-it-tu-ur. Any of these groups that contained any of the letters of the word conmuta, plus q, underwent a special substitution. The resultant quaternary text was then “translated”: Newbold replaced the pairs of letters with a single letter, presumably according to a key, which, however, he never made clear. Newbold regarded some letters of this reduced quinary text as equivalent to one another because of phonetic similarity. When required, therefore, he interchanged d and t, for example, b, f, and p, o and u, and so on. Finally, Newbold anagrammed the letters of this senary text to produce the alleged plaintext in Latin.

  In April, 1921, Newbold announced the preliminary results of his solution according to this method before brilliant and learned audiences. These results stamped Roger Bacon as the greatest scientific discoverer of all time. According to Newbold, Bacon had recognized the Great Nebula in Andromeda as a spiral galaxy, identified biological cells and their nuclei, and come close to seeing the union of the sperm with the ovum. He had therefore to have not merely speculated about but to have actually constructed a microscope and a telescope and used them to make discoveries that anticipated the 20th century. Newbold’s cryptanalysis of a caption on a sketch that somewhat resembles a pinwheel and that he took to represent the Andromeda nebula reads in part: “In a concave mirror I saw a star in the form of a snail … between the navel of Pegasus, the girdle of Andromeda, and the head of Cassiopeia.” Newbold asserted that his solution could not be subjective because “I did not know at the time [of solution] that any nebula would be found within the region thus defined.”

  Newbold’s solution created a sensation in the world of scholarship. Many scientists, though declining to pass upon the validity of the cryptanalysis, which they did not think themselves competent to do, accepted the results with alacrity. One eminent physiologist went so far as to specify that some of the drawings probably represented the columnar epithelial cells with their cilia, drawn to a magnification of 75. The public at large was fascinated. Sunday supplements had a field day. One poor woman came hundreds of miles to beseech Newbold to use Bacon’s formulas to cast out the demons that possessed her. The cipher itself drew mixed reviews. Manly, back at the University of Chicago, half accepted, half rejected it. “Professor Newbold’s theory and system now seem much more reasonable than they did a year ago when he first explained them to me,” he wrote in Harper’s Magazine. But a writer in Scientific American Monthly, J. Malcolm Bird, observed acutely, in relation to the tertiary text of interlocking pairs, as or-ri-it-tu-ur, that “Professor Newbold has not in any of his public utterances explained satisfactorily how, in the original encipherment, it is possible to … get letter-pairs that interlock as in the above example.” In other words, although the system might work in deciphering it did not seem to work in enciphering. Many oneway ciphers have been devised: it is possible to put messages into cipher, but not to get them back out. Newbold’s seemed to be the only example extant of the reverse situation. For this and other reasons, Bird rejected the solution.

  The excitement simmered down. Newbold went back to continue his solutions; other scholars weighed his conclusions. In 1926, Newbold died. But his working notes, his solutions, and the chapters for the book that he had projected were faithfully edited by his friend and colleague Roland Grubb Kent. In 1928, they were published as The Cipher of Roger Bacon. An important French philosopher, Étienne Gilson, later one of the 40 “immortals” of the Académie Française, though bewildered by the method, accepted the results; a French specialist in Bacon, Raoul Carton, enthusiastically endorsed both method and results. American and British historians of medieval science were cooler.

  In 1931, Manly, who had studied the Newbold method in detail, concluded that it “is open to objections of so grave a character as to make it impossible to accept the results.” Warning that these results “threaten to falsify, to no unimportant degree, the history of human thought,” he demolished them in a 47-page article. He pointed out that the cipher postulated by New-bold permitted many different “solutions.” The encipherer could never be certain that his message would get through correctly; the decipherer would never know whether he was reading the intended message. The chief cause of this flexibility lay in the anagramming process—the one that finally produced the Latin plaintext. Anagramming rearranges letters of one text into another; it is a kind of unkeyed transposition. Often many anagrams are possible: live, veil, evil, vile, and Levi are all anagrams of the “ciphertext” EILV, each as valid as the next. As the number of letters involved rises, the possible anagrams increase in geometric propor
tion. The 31 letters of the angelic salutation, “Ave Maria, gratia plena, Dominus tecum,” have afforded thousands of different anagrams, all perfect in spelling, diction, and syntax. One zealot turned out 1,500 pentameters and 1,500 hexameters; another 3,100 anagrams in prose and an acrostic poem; another composed a “Life of the Virgin” in 27 anagrams—all these of the salutation. Newbold tended to anagram Bacon’s message in blocks of 55 or 110 letters. How certain could he then be that his anagram was the right one? The answer is that he could not be certain at all.

  Manly also showed that the alleged shorthand signs were nothing more than the breaking up of the thick ink on the rough surface of the vellum into shreds and filaments that Newbold had imagined were individual signs. New-bold himself conceded that “I frequently, for example, find it impossible to read the same text twice in exactly the same way.” Manly pointed to different solutions from the same text. Finally, he criticized the texts of the solutions themselves on the ground that they “contain assumptions and statements which could not have emanated from Bacon or any other thirteenth century scholar.”

  How, then, to explain Newbold’s cryptanalyzing information that he said he never knew, such as the position of the spiral nebula? The answer is that he must have known it, though subconsciously. Newbold, a scholar of immense erudition who casually learned the Catalan language and read a thousand pages in it in pursuit of a minor point of the solution, must have swept up that detail in his extensive studies and slipped it into the depths of his brain, where it lay hidden from his active mind until the solution drew it forth. No one ever questioned Newbold’s integrity; he was a victim, Manly said, “of his own intense enthusiasm and his learned and ingenious subconscious.”

  The spectacular collapse of the Newbold theory has not deterred other scholars from attacking the manuscript, though it has made them a bit more cautious in publishing their “solutions.” In 1943, however, a Rochester, New York, lawyer, James Martin Feely, recklessly exposed to the world—and to its ridicule—a solution that makes little sense in Latin and not much more in English: “The feminated, having been feminated, press on the fore-bound; those pressing on are moistened; they are vein-laden; they will be broken up; they are lessened.”

  Two years later, Dr. Leonell C. Strong, a highly respected cancer research specialist, concluded that the Voynich manuscript was the work of one Anthony Ascham, an English scholar of the 1500s and author of an herbal. Strong cryptanalyzed out of the manuscript several texts in alleged medieval English, including a contraceptive formula, by means of a “double reverse system of arithmetic progressions of a multiple alphabet,” by which he apparently meant some form of polyalphabeticity. The contraceptive works, and anyone who wishes to prove it may do so, since Strong published it; but he has not seen fit to do the same with his method of cryptanalysis, and it therefore remains unproved and unaccepted. His published texts have been severely attacked on linguistic grounds, and the formula has been explained on the same basis of subconscious knowledge as Newbold’s spiral-nebula solution.

  There have been many more attempts that did not result in publication because the would-be solvers honestly admitted defeat. Scores of persons have worked at home on the illustrations in the Newbold volume without success. In 1944, from among specialists in languages, documents, mathematics, botany, and astronomy then doing war work in Washington, William F. Friedman organized a group to work on the problem. Unfortunately, by the time they had, working after hours, completed the task of transcribing the text into symbols that tabulating machines could process, the war was over and the group disbanded.

  Their preliminary results had the effect of deepening the mystery. For they found that words and groups of words repeat more often in the manuscript than in ordinary language. This fact alone differentiates the manuscript from all other cryptograms, for all known cipher systems seek to suppress repetitions, not to intensify them.

  What causes this difference? Friedman thinks that the manuscript represents a text in an artificial language that has divided all existence into categories, assigned each a basic symbol, and indicated subclasses by additional symbols tacked onto the first. The first artificial language, that of the Scot George Dalgarno, was of this kind. He distributed knowledge into 17 main classes and labeled each with a consonant: for example, K stood for political matters, N for natural objects. He subdivided these into subclasses and assigned a vowel to each. Thus Ke was “judicial affairs,” Ku “war.” Finer divisions were represented by alternating consonants and vowels. Many other artificial languages of this type have been invented, one by Bishop John Wilkins, who wrote the first book on cryptology in English. Obviously a text in such a language would repeat its “roots” over and over while its suffixes would vary—and this phenomenon is very common in the Voynich manuscript. Friedman planned to test this hypothesis ×in which the English cryptologist Brigadier John H. Tiltman concurs) on an R.C.A. 301 computer, but the work did not progress very far.

  Another explanation for the great redundancy is that it reflects the many repetitions of pharmaceutical formulas that are likely to occur in an herbal or any medical tract. This is the view of the late Father Theodore C. Petersen, Ph.D., of St. Paul’s College in Washington, D.C., an expert in ancient documents who made a 40-year study of the Voynich manuscript. He thought that minute variations in the shape of the characters and in their hooks and other appendages might represent the syllables of a medieval shorthand. He never did collect the statistical evidence he needed to confirm or refute this hypothesis.

  Yet men have solved mysteries far more abstruse. Why hasn’t anyone unriddled this? The reason, Manly said, is that “the attack has proceeded on false assumptions. We do not, in fact, know when the manuscript was written, or where, or what language lies at the basis of the encipherment. When the correct hypotheses are applied, the cipher will perhaps reveal itself as simple and easy….”

  Is it, then, just a gigantic hoax, like the Cardiff giant or the Piltdown man or the fossils of Professor Beringer? Nobody involved with it seems to think so—and this includes those who have been rebuffed by it. The work is too well organized, too extensive, too homogeneous. Nothing repeats larger than a group of five words, whereas in actual hoaxes, such as the fake hieroglyphic papyri sometimes sold to tourists in Egypt, much longer phrases are repeated. Moreover, the words in the text recur, but in different combinations, just as in ordinary writing. Even if it were a hoax, there seems to be no point to having made it so long. Most critically, the medieval quasi-science that was seeking the philosopher’s stone and the elixir of life while the manuscript was being written was too credulous to entertain the concept of a hoax.

  Voynich died in 1930. His wife, Ethel, kept the manuscript in a safe-deposit box at the Guaranty Trust Company in New York for 30 years, until her death in 1960, aged 96.* Her estate sold it to Kraus. He priced it at $160,000 because he believes that the manuscript contains information that could provide new insights into the record of man. “The moment someone can read it,” he says, “this book is worth a million dollars.” Others do not think so. They contest the attribution to Bacon, observing that the manuscript looks much more like a 16th- than a 13th-century work. They feel, as did an American foundation that turned down Friedman’s application for funds to attack it, that it contains nothing new, that it may be, after all, only some kind of fanciful herbal.

  But no one yet knows, and the book lies quietly inside its slipcase in the blackness of Kraus’ vault, possibly a time bomb in the history of science, awaiting the man who can interpret what is still the most mysterious manuscript in the world.

  * Mrs. Voynich deserves a footnote. Her novel, The Gadfly, has sold more than 2,500,000 copies in translation in the Soviet Union, where critics revere her as one of the all-time greats in English fiction. The patriotic romance, a best-seller when it was published in England in 1897, is read by most Russian schoolchildren, forms the subject of Soviet doctoral theses, and has been made into a movie and an
opera. The Russians think so highly of it that they paid Mrs. Voynich one of the very few royalty fees they ever gave to an American.

  24

  THE PATHOLOGY OF CRYPTOLOGY

  SICKNESS APPEARS IN CRYPTOLOGY as cryptanalytic hyperactivity. Its victims overcryptanalyze documents, and they bring forth invalid solutions. A case in point is William R. Newbold’s “solution” of the Voynich manuscript. It is not necessary, however, that the cryptic text be patent, that it be, in other words, a cryptogram. The text subjected to the excessive cryptanalysis may be a steganogram, which conceals the presence of the real, the secret message beneath an innocent cover-text. Steganography opens a much wider field of search for those seeking documents on which to vent their excessive cryptologic energies, for they may postulate that any text contains a hidden communication beneath a dissembling surface. The literary excellence of the outer text will then only attest to the superiority of the steganography; the less suspicious the outer text by virtue of its literary qualities, the better the steganogram. And, indeed, those who suffer from the most virulent form of this mania attack what on this basis would be the greatest steganogram of all time—the plays of William Shakespeare, seeking to draw forth solutions demonstrating that the real author was Francis Bacon.

  They are not entirely without cryptologic warrant. Just as systems of cryptography have transmitted valid messages despite abuses like Newbold’s, so systems of steganography have preserved legitimate messages beneath an innocent camouflage. Among these are steganograms of authorship. In 1897, the eminent philologist Walter W. Skeat was editing The Testament of Love, which had been attributed to Chaucer in its only known copy, a printing of 1532, when he noticed that the initial letters of the various chapters were intended to form an acrostic. With some emendation, they spelled out Margarete of Vitrw, have merci on thin[e]—Usk—indicating as some other scholars had suggested, that the real author was not Chaucer but Thomas Usk. Other cases are known, probably the earliest of which is in cuneiform—the only case in which a cuneiform author gives his name. The most famous authorship steganogram involves the famous Hypnerotomachia Poliphili, published by Aldus Manutius at Venice in 1499 with no author listed. It is regarded as one of the most beautiful books ever printed, and its typeface has inspired many of the ones used today. The title is made up of five Greek roots and has been translated as “The Strife of Love in a Dream.” As early as 1512, however, readers discovered that the first letters of the 38 chapters spelled out Poliam frater Franciscus Columna peramavit (“Brother Francesco Colonna passionately loves Polia”). Colonna was a Dominican monk, still alive when the book was published, and the reason for the secrecy was clear. Polia is still unknown.