French codes having an extensive vocabulary are called grands chiffres (great ciphers), as opposed to petits chiffres (small ciphers) having only a couple of hundred entries. Great ciphers date back to the age of Louis XIV (see another article). The present author partially reconstructed some great ciphers based on published materials. The present article describes various codes and ciphers during the Napoleonic age including these.
Table of Contents:
A Great Cipher in the Archives
Great Ciphers for Diplomatic Correspondence
"Great Paris Cipher" (1812-1813)
Napoleon-Maret Code during the Russian Campaign (1812)
Napoleon-Berthier Code (December 1812)
Great Ciphers of Marshal Davout (1813)
Vilcoq (1969) reproduces part of an original great cipher under the First Empire preserved in the archives. It is an encoding table (arranged alphabetically) of a two-part code of 1200 entries.
It has a feature typical in French ciphers: a code number may represent variant forms. For example, an entry "ecrit,s,e,s,re ... 37" means that the figure "37" may be used to represent various forms of the verb ecrire (write) including "ecrit", "ecrits", "ecrite", "ecrites", "ecris", "ecrire". An entry "differ,e,nt,s,es,ce,s ... 14" encompasses "differ", "differe", "different", "differents", "differentes", "difference", "differences." Such alternatives are not always grammatically or etymologically related. An entry "ave,c,z ... 558, 469, 295" encompasses a syllable ave as well as words avec (with) and avez ("have", second-person plural). Similarly for "dan,s,t ... 371, 583" and "don,t,c ... 54, 909". An entry "fa,brique,cant 513" means the figure may represent not only words "fabrique", "fabricant" but also a syllable "fa" (see an example for a small cipher in another article).
While le and les as well as de and des have separate entries, "il,s" covers both il and ils. There is no entry for "mes". Unexpectedly, "et" has no entry, while "est.es" is given three figures. Possibly, an entry "ete ou t" covers "et".
As for the irregular verb avoir, the entry "avoi,r,s,t" encompasses the infinitive as well as avois [avais], avoit [avait] (not avoient, at least in the table), which may also be used as endings. There are separate entries for "ave,c,z" encompassing avez and "ayant", not to speak of syllable-like "a","ai,t", "as", "on", "eu,x", but not future aurai, auras, etc. and other complex forms. As for faire, there is a separate entry "fait, s, e, s" encompassing fait, faits, faite, faites as well as one for the infinitive. Such particular knowledge would be helpful in reconstructing other French ciphers.
Groups "il y a" and "de la" have their own entries.
As was still typical then, "i" and "j" as well as "u" and "v" were written differently but were not treated as different letters of the alphabet. So, entries "au,x", "avan, t, ce", "aucun, s, e, s", "ave,c,z", "augment,e,ation,s", "avoi,r,s,t", "aussi" come in this order. The entry "J" is followed by "ja", j'aye", "ici", "idee,s", "je", .... Of these, "je" would have also been used for "ie" (see below).
Although it is not known specifically when and where this code was used, its vocabulary seems to include many words related to Mediterranean trade: "Adriatique", "Cap Finistere", "Constant,inople", "Levant", "car,gaison,s", "Chambre du comce", "barque,s", "Batiment,s", "Chebeck,s", "galere,s". Such particularity is not seen among military terms such as "Amiral,aux,te", "artillerie", "bataille,s,on,s", "bombarde,r,ment", "canon,s." Entries "Barbar,ie,esque,s", "Corsaire,s", "kam des Tartares" may be inherited from much older codes but "Etas unies de l'Am [illegible but apparently not erique]" indicates this code does not date earlier than the 1780s. The entry "Aix La Chapelle, Tte de" may give a further clue to the field of usage of this code.
Tant provides pdf of three great ciphers with 1200 entries from unknown periods (code inconnu 1, code inconnu 2, code inconnu 3) (The first and the second have an extensive gap, which may be an omission of some pages in the transcription. For example, the first lacks an entry for the letters P and S.). Of these, internal evidence indicates at least the second and the third belong to the First Empire. (Bessières and Bernadote became marshals of France (Wikipedia) in 1804. The title "Le Prince de PONTE CORVO" was created in 1806 (Wikipedia).)
As with the above, these three codes also include entries reminiscent of older ages. Some entries are similar to the code of the same size (1200) which formed the basis of the Great Paris Cipher below.
The first has many figures (10, 19, 48, 96, 112, 123, 716, 778, 796, 806, 831, 872, 882, 903, 910, 967, 985, 1022, 1031, 1033, 1053, 1128, 1163, 1166) for "nulle". It also has 755 for "annullant le precedent" and 1189 for "annullant chiffre compris entre".
The second has 29, 69, 83, 87 as "nulle" and 1002 for "annulle le precedent."
The third has 235 for "annullent le precedent." (Although there is no significant gap, there seem to be no nulls.)
Diplomatic correspondence is the foremost in requiring protection of secure codes/ciphers. It is known that a great cipher of 3500 entries was used with Vienna in 1807 and one of 3000 was used with Rome in 1808 (Vilcoq). Laforest, French ambassador in Madrid, also appears to have had a great cipher in 1809 (see below).
While great ciphers appear to have been in common use among diplomats, the army was slow to adopt great ciphers.
A simple substitution cipher of Eugène de Beauharnais and General Grammont (1809) is given by Tant (pdf). It assigns each letter one or two figures, letters, or other symbols.
A code of about 600 entries (code numbers up to 576 for alphabetically arranged syllables and words (multiple readings) and numbers 592-611 for single letters) is given by Tant (pdf). Another pdf provides a small fragment of this cipher but has additional entries: 616 (albanie), 617 (albanais), 618 (adriatique), 619 (brindisi), 620 (barlette), 621 (bari), 622 (butrinto).
Entries 1-17 are for names (or titles) such as:
5: S?M. le Roi de Naples [Napoleon's brother Josseph was made King of Naples in 1806; Marshal Murat was made King of Naples in 1808]
6: Le Vice Roi [Eugène de Beauharnais was made Voiceroy of Italy in 1805]
7: Ministre Directeur de l'adm[inistration] de la Guerre (Wikipedia)
8: Ministre des finances de Naples [post occubied by Pierre-Louis Roederer (Wikipedia) from 1808-1810]
9: Ministre du tresor public d'Italie (Wikipedia)
10: General Grenier (Wikipedia)
11: General Ottavi
12: General Decour
16: Ali Pacha (Wikipedia) [who allied with Napoleon but changed sides to ally with Britain in 1807]
17: General Marmont
A great cipher of 1400 entries was used by King Joseph in 1812-1813 during the Peninsular Campaign. The code was broken by a British officer George Scovell and was called "Great Paris Cipher" by Wellington (Urban p.173). The feat is described in Urban (2001) and Oman (1914).
King Joseph, Napoleon's elder brother, and his generals did not employ ciphers, still less extensive great ciphers, for several years after Joseph was set up on the Spanish throne in August 1808 (Urban p.68-69).
It was in April 1811 that the first enciphered French dispatch fell in the hands of the British under Wellington. The cipher was of the simplest kind: a monoalphabetic substitution that represented letters of the alphabet by symbols ("A" by ".", "B" by "1", ..., "x" by "+", "y" by "-", z by "="). It was decoded within the day. (Urban p.97)
Marshal Marmont, who assumed command of the (French) Army of Portugal in the Peninsula in May 1811, used a code of 150 entries with the six divisional commanders. It assigned several code numbers to frequent letters (e.g., nine for "e"). Letters "u" and "v" were not distinguished, as was still common at the time. In one portion of a dispatch captured in August or September, 131 out of the 711 code numbers stood for "e" but use of nine different figures must have made it less salient than in a simple substitution cipher. Scovell, who was appointed in August to superintend all the communications of the army (Urban p.102), could decipher it owing to the careless use of the code in mixing plaintext words here and there in the encoded text (Urban p.117-119). Although the French took care to update the code, at least three different versions were broken (Urban p.323).
Colonel Jardet's report to Marmont in early 1812 was also broken, which included a passage as follows:
A reconstructed key of this cipher is preserved with a deciphered letter of Marmont dated 6 March 1812 in the Spanish archives (PARES (ES.28079.AHN//DIVERSOS-COLECCIONES,133,N.29)), as I found years after first posting this article. Its deciphering table shows that the assignment was derived from a key phrase "a tous les coeurs biennes que la patrie est chere".
Even after a great cipher was introduced, a simpler code was still used lower in the hierarchy. When Colonel Lucotte wrote to Madrid on his way back from Paris on 16 March 1813, he used a cipher consisting of arbitrary symbols. Though it was not the simplest substitution cipher but "its encoding table involved lines of letters that would suggest the next few substitutions", Scovell solved it in 6 hours (Urban p.282).
When Cassan, commander of the garrison of Pampelona under siege (Wikipedia), wrote to Marshal Soult in 1813, the number of figures was not much over 60. There were six figures for "e", four for "a" and "i", three for "t", "s", "n", etc. The following is an example of its usage.
This particular letter was decoded in a few hours owing to the lucky guess on the initial words: "10(M)45(o)23(n)21(s)16(i)2(e)41(u)25(r) 5(l)24(e) 10(M)4(a)25(r)24(e)3(c)9(h)8(a)5(l)", which was confirmed by two-letter words such as "23(n)24(e)" and "10(m)2(e)" (Oman).
Napoleon did provide Joseph with a cipher when the latter had been made King of Spain in August 1808.
Maret is the person who was to provide King Joseph with the Great Paris Cipher in 1811, when he was a foreign minister (April 1811 to November 1813). Laforest was French ambassador in Madrid and must have had an extensive diplomatic cipher. Even if the code was indeed provided at this time, it appears to have just gone into oblivion.
King Joseph did not like marshals reporting directly to Napoleon and, in May 1811, personally urged his brother that he also receive those reports. It was arranged that he would be given information, if nothing else. This led to introduction of a great cipher for enhanced security (Urban p.111-113). In July, the code was supposed to be distributed to people concerned.
However, when Berthier, Napoleon's Chief of Staff, used it in a dispatch of 12 July 1811 to King Joseph, the key was not at his court. Joseph asked the ambassador Laforest with no avail.
Joseph even sent copies to Marmont and Soult on 10 August to seek help (Du Casse, Joseph, vol.7-8, p.58). In the meantime, correspondence continued to be carried on without being encoded. It was only in late 1811 that King Joseph received the Great Cipher from the cabinet of Foreign Minister Maret (Urban p.112-113, 136).
The Great Paris Cipher was a two-part code, which assigns code numbers to words etc. out of alphabetical order and thus requires separate tables for encoding and decoding. Its size of 1200 entries was soon enlarged to 1400. (One of the first captured dispatches in this code dated 16 April 1812 already used high numbers such as "1238 [que]", "1207 [Berthier]", "1264 [the Emperor]", "1328 [Joseph]", "1327 [Army of the South]", "1333 [Army of Portugal]" (Urban p.174, 188).) However, such ad hoc addition resulted in concentration of terms peculiar to the Peninsular Campaign in the addenda: "1201 Malaga", "1202 Valladolid", "1279 Talavera", "1280 Soult", "1265/1345 cavalerie", "1282 division(s)." On the other hand, there is a term "837 Crimea", which would be of little use in Iberia. This was because the code was modelled after a diplomatic code made in the 1750s. (Urban p.176, 137)
Images of the original of the Great Paris Cipher (designated B No30) preserved in the archives of the French army (SHD) is presented on Jean-François BOUCHAUDY's Crypto page.
An example of a passage in this code reads as follows.
Users of the Great Paris Cipher include King Joseph, Marmont (commander of the Army of Portugal), Soult (commander of the Army of the South), Dorsenne (commander of the Army of the North), Suchet (commander of the Army of Catalonia), Jourdan (Joseph's adviser), Napoleon, Berthier (Chief of Staff), Clarke (Minister of War in Paris) (Urban p.267, 345, etc.). Clarke became Joseph's new contact and was given the Great Paris Cipher when Napoleon took Maret and Berthier to the Russian Campaign.
Clarke's dispatch to General Caffarelli in Spain dated 19 October 1812 also used the Great Paris Cipher (I learned of it at Cipherbrain). Karsten Hansky acquired it at an auction and described it in "A dispatch from the French Minister of War Clarke to General Caffarelli, the commander of the Northern Army in Spain, from the year 1812" (17.06.2021) (pdf), according to which the cipher is preserved as code table #24 in the archives of the SHD (Service Historique de la Défense) in Vincenne (SHD box 1M-2352).
The marginal note (beginning with "Les parties de cette lettre qui sont chiffrées peuvent etre traduites comme il suite avec la Table C D no.29.") also includes coded message, which appears to be a transcoding of the ciphertext in the dispatch in another code (#29).
From the range of numbers, the code #29 appears to have about the same size as #24, but the number of codegroups is different (that is, there is no one-to-one correspondence between the code groups.)
Despite the much larger vocabulary compared to the low-level codes, the careless use in mixing plaintext words in the encoded message still allowed the British officer George Scovell to gradually identify code groups of the Great Paris Cipher (Urban, Chapters 11-12). In the summer of 1812, he could read intercepted letters fairly well (Urban p.206-210 ff.). The information allowed Wellington to ascertain the French dispositions and to choose the right moment to take a strategic movement. When Wellington beat the French at the Battle of Vitoria on 21 June 1813, King Joseph's copy of the Great Paris Cipher was captured (Urban p.297).
Ironically, the capture of the original code meant a loss of the advantage for the British, who had been reading the most secret messages of the French highest command for many months without their adversary noticing it. Now, King Joseph reported the loss of the Great Paris Cipher.
Early in July, an intercepted letter from Clarke to Suchet, who had the Great Paris Cipher, used a different code.
The few materials captured did not allow Scovell to break the new great cipher. For once, Wellington was misinformed and was surprised by attacks from unexpected directions. The British were forced to retire to the north of Pampelona in late July (Strawson, The Duke and the Emperor (Japanese edition), p.240-241). But it was only a temporary setback. The last French strongholds in the Peninsula, San Sebastian and Pampelona, were taken by the end of the campaign.
Correspondance vol.12 (2012) provides invaluable materials in its appendix "La correspondance en chiffre, Les lettres à Maret en 1812". It provides Napoleon's letters to Maret from 18 November 1812 to 1 December 1812 in both code and plaintext. The volume also includes images of original manuscripts (apparently, encoding sheets). With these materials, the present author partially reconstructed the code with entries up to 3525.
Although the encoded text has a code number as large as "3525 on/ons/ont", the code numbers greater than 1200 are extremely biased in its distribution (numbers in 1240-1350, 1370-1470, 1520-1630, etc. do not appear in these letters). So, most of the high numbers may have been blank.
The vocabulary includes letters, syllables, and words/names, of which frequently used ones are given multiple figures. For example, "et" may be represented by "181", "613", "1084", "1509", or "2375". The word "artillerie" could be encoded as:
"687(ar)1782(ti)372(l)700(l)2671(er)449(ie)" or
"687(ar)1782(ti)710(l)3011(le)391(ri)238(e)".
This code lacks basic military terms such as "artillerie", "chevaux", "brigade", and "division" (it seems improbable that the encoder chose to take pains in spelling these words if they had their own code numbers defined in the code), though there are 681(infanterie) and 630(cavalerie).
Compared with the Great Paris Cipher, the code used by Napoleon's letters had, despite having much higher numbers, a striking weakness in that the arrangement of entries is not completely random. The partial (blockwise) regularity can be seen in code numbers such as 151(peu), 153(pi), 154(pla); 441(un), 442(deux), 443(trois), 444(quatre), 445(cinq); and 1162(dou), 1164(dra), 1165(dre(s)). In view of the absence of such regularity above 1200, the portion above 1200 or 1300 may have been an addenda (as was the case for the Great Paris Cipher).
(Actually, such a regularity was helpful in reconstructing the code. When "988.447" is identified as "froid", there still remain several possibilities in division such as 988(fr)447(oid), 988(fro)447(id), 988(froid)447(-), etc. The regularity in the arrangement allows a safe guess of 988(fro)447(id).)
As with other French codes, the same code number may represent variant forms. Thus, 220 may represent "donner", "donnez", "donne"; 1145 may represent "tout", "toute", "tous"; and 1232 may represent "envoyer", "envois", "envoyez." More confusing may be variants in syllables: "653" and "3525" appear to represent either "on", "ons", or "ont" and "52" appears to represent either "dans" or "dant." (The latter is consistent with an entry of the great cipher for Mediterranean commerce noted above, which, however, has separate entries for "on", "ons", and "ont.")
As was common at the time, "i" and "j" as well as "u" and "v" are not distinguished. Thus, "je" (the first person pronoun) and syllable "ie" as in "artillerie" are both encoded as "449" and "587 ur/vr" may be used in words such as "rigoureux", "autour", and "vivres." "W", which is not used in French but is used in Foreign names such as Abramowicz or Wrede, was also represented by the same code with "v".
Partial encoding is sometimes used to give a delusively bright outlook for the French: "L'armée est nombreuse, mais débandée d'une manière affreuse" (in which the underlined words are not encoded) (p.1404-1405). Such a practice has been pointed out by Urban (2001) (p.210 etc.) and others.
This section notes some minor details about Napoleon's (or his aide's) encoding printed in Correspondance (2012).
The encoder is supposed to encode the text other than the underlined portions in the manuscript. But "le prince Schwarzenberg" is encoded as "738" (p.1398). (While this per se may be nothing of note, it seems strange that "738" is written under "le"; "1194" for "?" is written under "prince"; and "Schwarzenberg" is underlined as if not encoded. There are many other instances in which code numbers are not aligned with plaintext.)
What may look like "1040" in the manuscript should be "1041" (eux) in "rigoureux" (p.1398).
The word "cette", though underlined, is encoded as "540" (fourth line, p.1399).
The number "585" should be "595" (qui) in "inquieter" (p.1399).
The number "591" should be "595" (qui) in "ce qui lui ..." (p.1399).
The number "12", though the underline stops short of this word, is omitted in encoding (p.1399).
What may look like "540" in the manuscript should be "541" (en) in "enlevent" (p.1399).
What may look like "3720" and "[4491]" in the manuscript should be "372" (l) and "449" (ie) in "artillerie."
The word "et", though underlined, is encoded as "1084" (p.1399).
The word "generale", though not underlined, is omitted in encoding (p.1400).
The word "sur" in "sur Minsk", though not underlined, is omitted in encoding (p.1401).
What may look like "813" in the manuscript should be "823" (nt) in "pont" (p.1401).
The uncertain number "[1030]", which should correspond to "reçu", may be "1080", which occur three times in other instances. (p.1402)
The phrase "beaucoup de glaces", though underlined, is encoded (p.1402).
The number "1012" should be "2012" (la) in "la stabilite" (p.1402).
The number "120" should be "1200" (v) for "f" in "Brisof" (p.1402).
The phrase "à Vilna", though not underlined, is omitted in encoding (p.1403).
The phrase "construire en", though not underlined, is omitted in encoding (p.1404). (For this particular case, blank below these words may indicate that these words are not encoded, as with the words "assurer les ponts et en" in the preceding words)
The number "105" should be "1005" (ces) (fourth line, p.1405).
The number "690 armée", which is not in the plaintext, is inserted before "masse" (1405).
The phrase "en les faisant partir a jour nomme", though not underlined, is omitted in encoding (p.1406).
The word "isolés", though underlined, is encoded as "803(is) 1152(ol) 859(es)" (p.1406).
What may look like "782" in the manuscript should be "1782" (ti) in "fatigues" (p.1406).
The word "peuvent", though (faintly) underlined, is encoded as "151(peu) 3130(ve) 3370(nt)" (p.1406).
The number "219" should be "2019" (est) (first line, p.1407).
The phrase "Voilà 45 jours" seems to be encoded as "3521(v) 625(oi) 2012(la) 899(quarant) 445(cinq) 835(jours)", as partially suggested by the editor (p.1407). (The intention of the line through "3521.625.2012.899" or under "Voilà 45 jours" is not clear.)
What may look like "1191" in the manuscript should be "1190" (y) (p.1408).
The number "5560" is a transcription error for "556" (mi) in "Ochmiana" (p.1409).
Apparently, Napoleon did not have to personally rely on the code early in the campaign. Late in August 1812, he was yet to establish a cipher with Foreign Minister Maret.
Here, Méneval was Napoleon's most trusted secretary. As of 1806, he was the only secretary who could enter the Emperor's office and had the keys to the secret drawers and the emperor's portfolios, though he had many vice-secretary under him in later years. (napoleon.org; Wikipédia; cf. Meneval, History of Napoleon I, vol.3 p.38-39) Many letters signed by Napoleon were actually in his hand (Correspondance générale vol.5 (2008), "Claude-François Méneval, Secrétaire de Napoléon, Une Biographie" p.956).
Foreign Minister Maret resided in Vilna during the campaign. Vilna (now Vilnius, capital of Lithuania) was the French rear base (where the headquarters of Czar Alexander had been located a few days before the French entered the city). Napoleon lingered in Vilna from 28 June to 16 July and he set up a provisional government of Lithuania under control of the French Foreign Minister. The Grande Armée kept contact with the rest of Europe through Vilna. (The Period of French Rule in Vilnius; Meneval, History of Napoleon I, vol.3, p.35-36)
Shortly after Napoleon entered Moscow on 14 September, he received a dispatch in code from Maret but he did not have the code.
This must have added to the Emperor's irritation after the fire of Moscow that raged to that very day (18 September). If it was Méneval rather than Maret that prepared the code (which may explain its weakness as compared with the Great Paris Cipher, which was furnished by Maret), there was nothing to blame Maret for and the Emperor could not simply locate the code in the confusion of the fire. Indeed, when the fire had been threatening to surround the Kremlin and cut him off from the corps outside the city, he had had to temporarily evacuate the palace and spend two days at Petrowskoi (Meneval, History of Napoleon I, vol.3, p.61-62). As far as can be known from Correspondance (2012), Napoleon's dispatch includes only one short note for the Empress (about the grandeur of Moscow) on the 16th and none on the 17th.)
By mid-October, the Emperor or his staff received (or just found) the code. On 15 October, regarding an envoy who had come all the way from King Joseph in Spain to Vilna, the Emperor told Maret to report the content of his mission in cipher (Correspondance (2012) no.31906; Tajan catalogue p.23, No.64).
Joseph had fallen out with Marshal Soult and demanded his recall (Urban Chapter 16, esp. p.255-256; Joseph's letter is in Du Casse, Joseph, vol.9-10, p.86). Napoleon dictated his reply to Joseph in his letter to General Clarke, Minister of War in Paris, dated 19 October 1812 (Correspondance (2012), p.1206). The present author confirmed the code used in this letter is the same as the above code used between Napoleon and Maret. One passage reads as follows:
King Joseph's plea came at a most inopportune moment for the Emperor, who was just about to abandon Moscow. On the same day (19 October), Napoleon wrote to Maret that he would decide the next day to blow up the Kremlin and make a decision on the route of retreat (Correspondance vol.24 (1868) p.276). He had already arranged a code for Marshal Mortier, duc de Trévise (Wikipédia), who was to remain in Moscow to camouflage the Emperor's intention but was secretly instructed to blow up the Kremlin.
Napoleon's letter to Maret (then in Vilna, not in Paris), dated 20 October 1812, which notified the minister of his decisions, attracted much attention in the bicentenary year of the fatal campaign by fetching a huge sum in an auction. The letter itself had been known from catalogues of earlier auctions and substantially the whole text had been printed in Correspondance (2012), p.1210 and other sources (e.g., Tajan catalogue p.24, No.65). The present author confirmed the code used in this letter is the same as the above.
The French army did blow up the Kremlin but the rain reduced much of its effects (Wikipedia).
During the Russian Campaign, Savary, Minister of Police since the disgrace of Fouché, had a cipher with Napoleon as well as with Maret (Correspondance (2012) no.32021, 32062), which may have been the same as the above.
The retreat of the Grande Armée was calamitous because of the bitter cold and persistent attacks by the Russians. In November, the baggage of Marshal Davout, tasked to support the rear-guard of Ney after leaving Smolensk, was seized at Krasnoie (Nicolson p.238). (Urban found Berthier's instructions for use of a great cipher in the Russian State Military Historical Archive in Moscow but apparently not the code itself (Urban p.324).) At Orcha, Napoleon ordered burning of secret papers in the hands of Meneval as well as almost all his carriages and baggage (Meneval, History of Napoleon I, vol.3, p.83).
After the crossing of the Berezina, the Emperor left the remnant of the Grande Armée on 5 December to get back to Paris on the 18th. (One of the first despatches of Napoleon on his return, dated 19 December 1812, was to give direction to assure the vice-reine [spouse of Eugène de Beauharnais] by (optical) telegraph of his arrival and her husband's safety.)
A letter in code from Berthier (then in Koenigsberg, where the remnants of the Grande Armée were headquartered at the time) to Napoleon dated 22 December 1812 appears to use a great cipher of about 1200 entries different from the above.
While the message itself is wholly in code, it bears a note: "Duplicata, Chiffre du Prince de Neufchâtel, La Primata a été déchiffrée." From this, it appears to indicate that Napoleon had Berthier's code apart from the one he used with Maret, not that this was a replacement of the latter. This may be the predecessor of the "code of the Chief of Staff" noted below.
This may correspond to "votre note chifrée", which is acknowledged, together with Berthier's letter of the 21st, in Napoleon's letter to Berthier dated 30 December 1812.
In the spring of 1813, when the Emperor was going on a German campaign, Empress Marie-Louise was appointed nominal regent and it was decreed that she might use the code of the Foreign Minister, which may have been the same as the above or its update.
Napoleon used it at least in September in conveying a speech the Empress should give at the Senate.
(By the way, the word chiffre (cipher) can also mean a symbol made of interwoven initials. As an example, here is a cipher of Marie-Antoinette. So when Napoleon wrote to his wife "Il faut avoir des bagues avec ton chiffre." (Correspondance (2012) no.30901, 12 June 1812), it had nothing to do with cryptography. So did the "snuff-box ornamented with her cipher in diamonds" that Mèneval was given by Marie-Louise when he left her in May 1815 in Vienna (Meneval, History of Napoleon I, vol.3, p.441).)
From late February to early March 1813, Napoleon gave directions for replacement of codes. Although Bazeries deplores that it is too late to arrange for ciphers at the beginning of a campaign (Bazeries (1896), p.19), it may have been necessitated by change of the military situation.
Once across the Niemen into the Prussian territory, the French army on a way back from Moscow was relieved of the Russian pursuit for a time (Nicolson p.290). But the Prussian detachment, isolated at the rear of the Grande Armée, made a settlement with the Russians and General Yorck declared neutrality on 30 December 1812 (Wikipedia). In defiance of the King of Prussia, East Prussia was stirred against Napoleon (Paret p.343-345). At the end of January, the Prussian court removed from Berlin under French rule to Breslau in Silesia (Paret p.345, Wikipedia). Prussia secretly entered into formal alliance with Russia on 28 February 1813 (Wikipedia).
In the meantime, the French army had been pressed by the Russians. In mid-January, after crossing the Vistula, Commander Murat, King of Naples, returned to Italy, leaving Prince Eugène (Eugène de Beauharnais, vice-roy of Italy) in command in Posen (now Poznań), which town he had to give up in mid-February. In late February, Cossack patrol appeared even near Berlin. (Correspondance, vol.24, p.587, 605) On 26 February, Napoleon's distrust of the Prussians, which had been aroused by their inaction earlier in the month, led him to order to disarm them if Spandau, an important place to defend Berlin, were to be invested by the Russians (Correspondance, vol.24 p.634, 633). In the end, Berlin had to be abandoned early in March. Defection of Prussia was finalized by their declaration of war against France on 17 March 1813, right after Czar Alexander appeared at the court of the King of Prussia in Breslau.
It was in such a background that Napoleon gave directions to replace code. Although there is not conclusive evidence, he may have considered "all the ciphers of the army" were compromised because of the change of allegiance.
While waiting for replacement code, Napoleon told General Lauriston, then at Magdeburg, which was still safe for the French, to inform vice-roy Eugène in code if he had one with the vice-roy (2 March 1813) and asked if he had a code that the Emperor might use in case the communication route with Magdeburg is threatened by the enemy (6 March 1813). The latter shows that Napoleon considered encoding necessary only when the line of communication was threatened.
Upon this occasion, Napoleon appears to have made two kinds of (great?) ciphers: a code of the Chief of Staff with commanders and another between the Emperor and the commanders.
Despite the disaster in 1812, Napoleon was still more than a match for the allied forces of Russia and Prussia. He won the Battle of Lützen and the Battle of Bautzen in May, though the victories were not as complete as could have been. A truce was agreed, not out of a sincere desire of peace but in order to gain time on either side. By the expiry of the armistice on 10 August, the allies succeeded in persuading the Austrian Emperor, who declared war on France on 12 August (Nicolson p.309, Takagi p.218). During the armistice, Napoleon had to change the arrangement of his troops because he knew Austria had to be considered an enemy (Wikipedia).
In late August, Napoleon gave alert that the code of the Chief of Staff [Berthier] with the commanders of the army might be in the hands of the enemy. Considering that Napoleon does not appear to have had concern on 12-13 August about the "Code of the Chief of Staff" shared by Oudinot, Davout, General Lemarois [i.e., governor of Magdeburg], General Lapoype, General Durosnel, and the governor of Torgau (Correspondance, vol.26, no.20340, 20352, 20365, 20374), the cause of the replacement would not have been the defection of Austria, which could have been foreseen before that.
Right after beginning to move on 20 August, Napoleon had to stop to immediately despatch several officers because the bearer who carried a cipher to one of his marshals had been taken prisoner (Odeleben, vol.1, p.263-264).
The code given to Davout and Lemarois at this time is probably the code prepared in the spring for use between Napoleon and commanders (as opposed to the code of the Chief of Staff with commanders made at the same time). It probably corresponds to the great cipher used by Davout and Napoleon in November-December 1813 as well as by Oudinot (see below). Prince Eugène had a particular code to be used with the Emperor (which is mentioned as early as 15 April 1813 (Du Casse, Eugene, vol.9)) and thus was simply told not to use the compromised code.
It appears "the small cipher of the army", separate from the "code of the Chief of Staff with the commanders of the army" above, was also replaced at this time.
The latter quotation indicates Macdonald had a small cipher separate from "the small cipher of the army."
The compromised (small?) cipher of the army had also been used between King Jerome of Westphalia and the Emperor. But Jerome appears to have had a diplomatic great cipher and Napoleon asked Foreign Minister Maret to write to him in code.
After making these arrangements for security of correspondence, Napoleon had a limited victory at the Battle of Dresden on 26-27 August but after the defeat of Oudinot, who failed in taking Berlin, Napoleon had to withdraw across the Elbe between late September and early October (Wikipedia).
The following letters, with one immediately before the decisive Battle of Leipzig, indicate small ciphers (which would have been similar to the one with the commander at Danzig (see another article)) were commonly used in the field.
The loss in the Battle of Leipzig on 16-19 October forced the French army to retreat across the Rhine in November. On 11 November, the French army in Dresden surrendered (Takagi p.219-220). Danzig under General Rapp surrendered on 17 November (see another article; General Rapp's letter in cipher dated in 4 August 1813, found in Dresden, uses the same cipher (posted on Klaus Schmeh's facebook and solved instantly by Geoge Lasry without relying on the known cipher).). Marshal Davout in Hamburg was now isolated.
A small cipher (code numbers up to 284) of marshals of Npoleon at the time of the Battle of Nations (October 1813) is given by Tant (pdf). Fragments of this cipher is also given in other pdf, pdf. This may correspond to one of the ciphers mentioned in the above.
Entries up to 177 are mainly for letters, syllables, and numerals. Entries 210-284 list place names.
Bazeries (1896) prints some cipher letters to and from Marshal Davout (Duke of Auerstaedt, Prince of Eckmühl). Since the codes which he must have reconstructed are not in print, the present author re-reconstructed these and confirmed different great ciphers were used in May and November-December 1813.
Instructions from Napoleon's Chief of Staff Berthier (Prince of Wagram, Prince of Neuchâtel) to Davout dated 7 May 1813 (Bazeries (1896), p.23 ff.) use a great cipher, with entries up to at least "1197 ver".
It is similar to the Great Paris Cipher and does not show regularity as in the Napoleon-Maret code above. The word "et" is given numbers 197, 413, 534, and 821.
Davout had to report on 11 May that he could not decipher the instructions with any of the keys he had (ibid. p.31). The cipher arrived a few hours after he dispatched the report. Considering that Napoleon had asked Berthier about a cipher with Davout on 16 March (Correspondance (2012), no.30231), the delay seems to indicate a serious problem in handling of codes in the French army.
In the report, Davout explains the ciphers he had at the time were: (i) a cipher he made before the last campaign, which he used for correspondence with l'Oden [the Oder?] and Danzig when he had been in Hamburg and (ii) a cipher he made in Thorn for correspondence with governors, of which he gave a copy to the vice-roy [of Italy, Eugène de Beauharnais].
The present author confirmed the cipher used in Berthier's instructions to Davout is the same as the "grand chiffre 34" used in a letter of Marshal Oudinot (Duke of Reggio) (Bazeries (1896), p.19 ff.).
Three letters from Davout to Napoleon dated 14 November, 19 November, and 1 December 1813 (Bazeries (1896), p.37 ff.) use a different great cipher, with entries up to at least "1200 me". Use of this great cipher (in the archives) started on 23 August (Bazeries (1896) p.14) and this must be the one given to Davout in August (see above). These letters were written soon after Davout evacuated, on 13 November 1813, Ratzebourg to fall back to Hamburg, which he did not surrender until after the fall of Napoleon in April 1814.
This code assigns figures 10, 18, 834, and 1128 to "et".
The present author confirmed it is the same as the great cipher used in an unsigned letter printed in Bazeries (1901), p.165.
Napoleon was in Düben on 10-14 October 1813 so the timing coincides with Davout's use of this code.
Davout did not use any code in his letter of 4 December because he believed the defection of the King of Bavaria (on 8 October 1813 (Wikipedia)) must have compromised the code.
The King of Bavaria had said to Prince Eugène, his son-in-law, that he returned the code without making a copy in a letter explaining his change of allegiance.
His chivalry might have been true but Davout might have been right in acting on the assumption that the code had been compromised.
In November 1813, Foreign Minister Maret was replaced by Caulaincourt but remained with the Emperor as a secretary (Wikipedia). Caulaincourt had been a contact with foreign courts and was considered to be more suitable to peace talks (Takagi p.229-236), which, however, did not begin until February 1814.
In January 1814, Joseph Bonaparte, ousted from the Spanish throne in 1813, was named Lieutenant General of the Emperor and was entrusted with defense of Paris. Napoleon repeatedly wrote about security of correspondence. (Again, these show that Napoleon was forced to use cipher because of the military situation threatening his lines of communication.)
At one time, after a military setback in February, Napoleon allowed Maret and Berthier to write instructions authorizing Caulaincourt to make substantial concessions. However, his subsequent local victories over the allies made him change his mind (Takagi p.252-257; Meneval, History of Napoleon I, vol.3 p.169).
Despite Napoleon's precaution, the allies obtained intelligence from captured letters on some occasions. Napoleon's letter to the Empress of 23 March, notifying his intentions to march to Saint-Dizier behind the Austrians to cut the enemy's line of communication, was captured by the Cossacks. On the other hand, captured letters to the Emperor from the Empress and Savary revealed that Paris had no means of defense; anti-Napoleon sentiment was growing; and people were disturbed by the news that Bordeaux welcomed a Bourbon prince (the Duke of Angoulême), who entered the city with Wellington from the Spanish front. Based on the intelligence, the allied forces determined to march on Paris, with a detachment of cavalry dealing with the Emperor. Napoleon found it only on the 27th from captured letters and confession of a captured Russian trooper. (It is unknown to the present author whether these letters captured on either side were not encoded.) But it was an encoded letter from Minister of Posts Lavalette, who was his confidant and handled censorship to watch royalist activities, that made Napoleon decide to return to Paris. (Takagi p.277-278; Meneval, History of Napoleon I, vol.3, p.189, 190)
The Emperor's decision was too late. In the evening of the 28th, the council under Joseph decided to remove the court to Blois. The Empress, Joseph, and ministers (except for Talleyrand) left Paris on 29-30 March. On the 30th, Marshal Marmont, commanding the last army between the capital and the allied forces, agreed to ceasefire. On the 31st, the allies entered Paris. (Takagi p.279-292)
At Fontainebleau, Napoleon tried to stage a counterattack but finally agreed to abdicate on 6 April (Takagi p.308-323).
Exchanges in the meantime show that ciphers were forgotten in these last minutes.
(It had been arranged to burn the most important papers but the order was only partially fulfilled (Meneval, History of Napoleon I, vol.3, p.198-199). At least, Meneval kept his cipher and received a ciphered letter dicated by the Emperor on 8 April (Meneval, History of Napoleon I, vol.3, p.231-232).)
Specific nature of the code used by Napoleon during the Russian Campaign in 1812 as well as ones used in 1813 is revealed.
Napoleon was aware of the importance of information security. However, he was inclined to think code/cipher was required only when line of communication was threatened by the enemy. Combined with his often optimistic outlook on the situation, it meant arrangement for codes and ciphers was made in the last minute. Further research is desired as to specific timing and cause of replacement of codes.
Some contemporary publications say Napoleon used a Porta-like polyalphabetic cipher in a proclamation to the army in preparing for his comeback from Elba. See another article.
A code "F 18" used by Napoleon's troops during the Hundred Days was found by Jean-François Bouchaudy in the archives of the French army (SHD - 1M 2352 - MEMOIRE DE CORRESPONDANCE REPUBLIQUE ET EMPIRE DOCUMENT GENERAUX CHIFFRES DE CORRESPONDANCE) and is described on his website. This particular copy belonged to General Thouvenot, commander in Bayonne (Wikipedia).
It is a two-part code consisting of two sheets for encoding (marked "C") and decoding (marked "D") and contains 1200 elements for letters, syllables, endings ("ition", ...), words and place names, and punctuation marks. High-frequency elements have several numbers assigned to them ("du": 241, 407, 960, 1150; "cet, te": 247, 968, 1175).
Cancellation symbols are provided (one to delete the preceding group; one to cancel the current line; one to cancel the rest of the line and the rest of the page; and a pair to cancel the groups between them). Many groups are defined to indicate the end of the message.
The code comes with instructions for using the codebook.
Correspondance de Napoléon Ier; publiée par ordre de l'empereur Napoléon III, vol.23 (1868) [12 November 1811- ], vol.24 (1868) [1 July 1812- ], vol.25 (1868) [1 March 1813- ], vol.26 (1868) [1 August 1813- ], vol.27 (1869) [1 January 1814- ] (Internet Archive)
Correspondance generale de Napoleon Bonaparte, Tome 12: 1812 - La campagne de Russie (2012)
Du Casse (ed.), Mémoires et correspondance politique et militaire du prince Eugène, Vol.9 (1860) (Internet Archive)
Du Casse (ed.), Mémoires et correspondance politique et militaire du roi Joseph, Vols.7-8 (1854), Vols.9-10 (1854) (Internet Archive)
The confidential correspondence of Napoleon Bonaparte with his brother Joseph (vol.1, vol.2)
Memoires du Marechal Marmont, duc de Raguse (Complete), vols. 7-9 (Gutenberg) (HathiTrust)
Méneval, Napoléon et Marie-Louise: souvenirs historiques de M. le baron Méneval, vol.1 (1843) (Google)
Méneval, Mémoires pour servir à l'histoire de Napoléon Ier depuis 1802 jusqu'à 1815 vol.3 (1894) (Google)
Méneval, Memoirs to serve for the history of Napoleon I; from 1802 to 1805 (1895, London) (vol.1, vol.2, vol.3) (Internet Archive)
Baron von Odeleben (Lieutenant-Colonel of Royal Saxon Cavalry, Adjutant on the General Staff), A circumstantial narrative of the campaign in Saxony, in the year 1813 (1820) vol.1 (Google), vol.2 (Internet Archive) (HathiTrust) ... circumstantial description of Napoleon's dictating on vol.1, p.147-149, 198, etc.
The Online Books Page ... links to further sources
Mark Urban, The Man Who Broke Napoleon's Codes (2001)
Charles Oman, A History of the Peninsular War, Volume V (1914) (Internet Archive), Appendix XV, "The Scovell Ciphers", pp.611-618
The Great Paris Cipher; General George Scovell (1774-1861) (UK National Archives)
Daniel Tant, "inventaire des messages codés, cryptés ou chiffrés dans les services d'archives français" (online, accessed on 8 January 2018) provides several Napoleonic codes/ciphers in pdf, references to which I included in the above. Note that "Code napoléonien III" should belong to the age of Napoleon III (see another article).
J. Vilcoq, "Le Chiffre sous le Premier Empire", Revue Historique de L'Armée No.4 (1969)
Etienne Bazeries, Les <<Chiffres>> de Napoleon Ier (1896)
Etienne Bazeries, Les Chiffres Secrets Dévoilés (1901)
Nigel Nicolson (1985), Napoleon 1812 (Japanese translation by Shirasu Hideko)
Peter Paret (1985), Clausewitz and the State (Japanese translation by Shirasu Hideko)
Takagi Yoshio (1997), Napoleon and Talleyrand (in Japanese), vol.2
Tajan (catalogue before 2012)
65 NAPOLEON Ier (the 2012 Osenat auction catalogue)
Napoleon's secret coded Kremlin letter on sale (Yahoo! News)
'I am going to blow up the Kremlin': Napoleon's secret coded letter on sale (NBC News)
'I am going to blow up the Kremlin': Encoded letter from Napoleon sold at auction (The Independent)
Secret code letter by Napoleon boasting forces would blow up Moscow's Kremlin sold at auction for £152,000 (Mail Online)
Museum buys Napoleon coded letter on blowing up the Kremlin (The History Blog)
Napoléon: 'Je fais sauter le Kremlin' (dailymotion video)