But I could find no evidence of any such tampering on my scrap of paper, so I set aside the leaf and took up the first of my books on decipherment.

Well, perhaps our age with its scientific spirit was not quite rid of the old trickeries after all. I sold an alarming number of books on decipherment, most of which titles were also on the shelves at Pontifex Hall. Indeed, had not a whole shelf been devoted to the art of steganography? Now, as I sat with a pile of the books spread before me like braces of grouse ready for plucking, I saw that many of them had been reprinted in London during the past twenty years. Yes, ours was evidently an age that prized the preservation-and the revelation-of secrets. And who could blame us, I suppose, after so many years of war and intrigue?

I had discovered on my shelves the Steganographia of Johann von Heidenberg, alias John Trithemius, a Benedictine monk who had supposedly raised the spirit of the dead wife of the Emperor Maximilian I. There was also the Magia naturalis of the occultist Gian Battista della Porta, who had founded an 'Academy of Secrets' in Naples, along with De cifris, written by Leon Alberti, whose greatest invention was a 'cipher disk', two copper wheels, one inside the other, that rotated forwards and backwards. I owned also the work of an English author, John Wilkins, whose wife was Oliver Cromwell's sister. And I had a copy of the most famous cryptographer's manual of all, Blaise de Vigenère's 600-page Traicté des chiffres, ou secretes manières d'escrire, first published in Paris in the year 1586. A copy of this particular work, I recalled, was likewise on the shelves at Pontifex Hall.

For two whole hours I sat hunched over the paper, shaking my head in dismay as I tried to make sense, first of the volumes, then of the cipher, to which I applied their obscure precepts. The concept of a cipher is simple enough. It consists of a series of masqueraders: a number of characters beneath which others, the true characters, hide their faces. The faces of these hidden characters have been changed according to some arbitrary and prearranged convention called a code, the 'language' in which the cipher is written. Like any language, a code consists of a network of connections governed by its own particular rules and conventions. Deciphering therefore involves knowing or else discovering these rules and conventions in order to reveal the true identities of the impostors occupying their places. The problem, naturally, is through what method these masqueraders should be unmasked. Ordinarily the recipient solves the mystery by means of a key, a sort of grammar explaining the language in which it is written. The key might stipulate, for instance, that the true characters are replaced by those two places down the alphabet, thus:

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

C D E F G H I J K L M N O P Q R S T U V W X Y Z A B

In this case-a two-letter shift to the right-the cryptographer simply replaces the letters in the top line with those below, while the decipherer moves two letters backwards instead of two letters forward. This fairly crude system is known as the 'Caesar Alphabet', since it was first used by Julius Caesar in his communications with his troops in Spain and Syria. Such a system may be cracked, the books on my desk explained, by a little guesswork. For example, according to type-founders' bills the most common letter in the English alphabet is E, while the second most common is A, then O, then N, and so forth. The most common word is, of course, the definite article, 'the'. Now, given this small bit of information, the decipherer should first determine whether one particular letter occurs more than the others. Presumably the letter will not be E, because, like the other letters, the E will have been occulted beneath its impostor. Should he find one-the letter X, shall we say-it will become his candidate for the letter E. And should this letter frequently occur in conjunction with two others, he will have reason to suspect the trio together represents the definite article-and he will moreover have solved the identities of two more letters.

Or so he hopes. But he must proceed carefully. Snares may have been laid for him as he blunders along. The word may be spelled backwards or otherwise transposed. Or perhaps nulls-letters of no value-will have been inserted to throw him off the scent. The key might stipulate, for example, that the letter Y is a null and therefore is paired with nothing whatsoever in the plain-text. Or else the key might stipulate that every fifth letter in the cipher should be ignored, or that only the second letter in each line should be accounted. Or perhaps the definite article, or even the letter E itself, will have been omitted from the cipher altogether.

My mind was beginning to spin at the thought of these duplicities, so I turned from the books on covered writing to the cipher itself. By this time the sun had turned bullfinch-orange in the casement and the watchman was passing up and down the carriageway, ringing his bell. The most common letter in the text, I discovered, was K, of which I counted eleven. I made substitutions based on the assumption that K represented E, which meant, therefore, that the cipher alphabet would consist of a six-letter shift to the right of the plain-text alphabet. But after I made these simple changes the cipher was no clearer than before. It appeared that my cryptographer was a subtler creature than Julius Caesar.

I therefore decided that he must have used what was known to cryptographers as le système Vigenère, a more complex method in which a keyword is used to occult and then expose the letters in the plain-text. According to Vigenère, the keyword was the clue to the labyrinth of letters: the golden skein the decipherer unspools as he winds his way backwards and forwards. Its purpose is to explain which cipher alphabets-often as many as six or seven-have been substituted for the plain-text one. Usually it will be a single word, but occasionally two or three, or possibly even an entire phrase. Vigenère himself recommends a phrase, because the longer the keyword, the harder the cipher will be to solve.

Once more I felt daunted by the task confronting me. I had creaked open Vigenère's Traicté and was stumbling through passages of archaic French, trying to make sense of the long columns and tables of letters that filled page after page. Without the keyword it seemed that the cipher would be all but impossible to crack, since as many as a dozen codes might have to be solved in a single cipher.

At length, though, I discovered that le système Vigenère was really not as mysterious as all that, at least not in conception, and as a method of enciphering texts it was ingenious, not to mention dismayingly effective. As I studied the Traicté I came to see the great Vigenère as a wizard or conjuror whose medium was words and letters rather than chemicals or flames-words and letters whose shapes he transformed with the incantation of a spell or the gesture of a wand.

His method consists, like Caesar's, of polyalphabetic substitutions, but substitutions of a more complex variety, ones whereby the plain-text letters can be replaced by those in any one of twenty-five cipher alphabets. The plain-text letter A might be replaced in the cipher alphabet by C, as in the Caesar Alphabet. But it does not therefore follow that plain-text B will then be replaced in the cipher by the letter D: it could be replaced with equal probability by any one of the twenty-four other letters. Neither does it mean that when C reappears in the cipher it will once again represent the plain-text letter A, because A, too, might have changed its value. For in Vigenère's substitution table, any plain-text letter along the horizontal axis may be replaced by any one below it in the vertical or left-hand one, which becomes its cipher: