THE SCOVELL CIPHERS
By the very great kindness of Mr. G. Scovell of Brighton, I have had placed at my disposition the papers of his great-uncle, General Sir George Scovell, G.C.B., who served during the Peninsular War in the Intelligence branch of the Quartermaster-General’s department. In the beginning of 1812 the number of intercepted French dispatches in cipher which came into Wellington’s hands, through the happy activity of Julian Sanchez and other guerrillero chiefs, began to be so considerable that the Commander-in-Chief thought it worth while to detail a member of his staff to deal with them. Captain Scovell was selected because of his ingenuity in this line, and became responsible for attempting to interpret all the captured documents. They were made over to him, and, having done what he could with them, he placed the fair-copy of the ‘decoded’ result in Wellington’s hands, but seems to have been allowed to keep the originals—which were, of course, unintelligible because of their form, and therefore useless to his chief. The file of documents which thus remained with him is most interesting: they range in size from formal dispatches of considerable bulk—eight or ten folio pages long—down to scraps of the smallest size written on thin paper, and folded up so as to go into some secret place of concealment on the bearer’s person. Some of them look as if they had been sewed up in a button, or rolled under the leather of a whip handle, or pushed along the seam of a garment. I take it that these must all have been entrusted to emissaries sent in disguise, Afrancesados or peasants hired by a great bribe. Presumably each of these scraps cost the life of the bearer when it was discovered—for the guerrillero chiefs did not deal mildly with Spaniards caught carrying French secret orders. The large folio dispatches, on the other hand, must no doubt have been carried by French aides-de-camp or couriers, whose escorts were dispersed or captured by the partidas at some corner of the mountain roads between Madrid and the head-quarters of the Armies of Portugal and Andalusia.
The cipher letters are of two sorts—in the first (and more numerous) class only the names of persons and places, and the most important sentences are in cipher—invariably a numerical cipher of arbitrary figures. In the other class the whole dispatch is written in figures, not merely its more weighty clauses. The reason for adopting the former method was that it saved much time; the transliterating of unimportant parts of the dispatch (such as compliments, and personal remarks of no strategical import) would have taken many extra hours, when it was necessary to get a letter sent off in a hurry. But, as we shall see later on, there was grave danger in using this system, because the context might sometimes allow the decipherer to make a good guess at the disguised words, after reading that part of the letter which was not so guarded.
Occasionally a French dispatch is ciphered after the same infantile system that readers of romances will remember in Poe’s Gold Bug or Conan Doyle’s Sherlock Holmes, where letters or numbers are merely substituted for each other—where, for example, 2 always means letter e, or 25 letter r. This sort of cipher is dangerously easy to an expert reader, especially if the words are separated from each other, so that the number of letters in each can be counted. Take, for example, a letter sent to Soult in 1813 by Cassan, the blockaded governor of Pampeluna[797]. Only one precaution had been taken in this cipher-epistle, viz. that elaborate care has been taken to defeat the attempt of the reader to arrive at results by counting what figures appear most frequently, and so deducing by their repetition that these must be e (the most frequently used letter in French, as in English), s, i, a, t, and other common letters. This is done by having six alternative numbers for e, four each for a and i, three for t, s, and n. Taking the simple phrase
47.50.40.41.14.26
58.24
3.51.10.36.44.23.17.24.10.50.53.27
47.46
11.18.39.17.46.21,
which deciphers into ‘depuis le commencement du blocus,’ we see that e appears five times, but is represented by both 50, 24, and 44; u three times, but varied as 14 and 46; m thrice, varied as 10 and 36. This made the reader’s work harder, but not nearly so difficult as that required for certain other ciphers: for the whole set of signs, being not much over 60 in number, there was a limited amount of possibilities for each figure-interpretation. And the words being separated by spaces, there was a certainty that some of the two-letter units must represent et, de, ce, eu, du, and similar common French two-letter words. As a matter of fact this particular dispatch was deciphered in a few hours owing to the lucky guess that its initial words
10.45.23.21.16.2.41.25
5.24
10.4.25.24.3.9.8.5
might be ‘Monsieur le Maréchal,’ the preliminary address to the intended recipient. This hypothesis was verified at once by finding that this rendering made good sense for the two-letter words 23.24 = ne, and 10.2 = me, lower down in the letter. After this all was plain sailing.
But the usual French cipher, the ‘Great Paris Cipher’ as Scovell called it, was a very much more complicated and difficult affair, as the list of figures, instead of being only a few score, ran to many hundreds. And of these only some few represented individual letters, more were parts of a syllabary: ma, me, mi, mo, mu, for example, had each a figure representing them, and so had ab- ac- ad- af- ag- &c. Moreover, there was a multitude of arbitrary numbers, representing under a single figure words that must often be used in a dispatch, such as hommes, armée, général, marche, ennemi, corps d’armée, canons. In addition there was a code of proper names, e.g. 1216 meant the River Douro, 93 Portugal, 1279 Talavera, 585 King Joseph, 1391 General Dorsenne, 1327 the Army of the South, 1280 Soult, 1300 Wellington, 400 Ciudad Rodrigo, &c. If the King wished the Duke of Dalmatia to send 9,000 men of the Army of the South to Talavera, he had only to write
‘585 désire que 1280 dirige 1156 (neuf) 692 (mille) 1102 (hommes) de 1327 sur 1279.’