by Mark Urban
With each league of his journey back to Spain, Joseph should have seen that the 17 May memorandum was worthless. It began promisingly enough from the point of view of His Catholic Majesty: ‘The King of Spain, as the emperor’s lieutenant, commands the French armies of Spain and Portugal in accordance with the direction he receives from the emperor, through the Chief of Staff the Prince of Neufchatel.’ Joseph would be assisted in these duties by a military figure, ‘responsible for the details of the overall direction of troops garrisoned in the provinces’.
The body of this Rambouillet accord then proceeded to undermine, line by line, all that had been confidently stated at the beginning. Marmont’s Army of Portugal and Soult’s Army of the South would ‘correspond’ with Madrid, a clause which those two commanders took to mean keeping the king informed rather than accepting any orders from him. The armies of the North and Catalonia would be run under different arrangements: military governors appointed by Napoleon ran administrative matters and he also supplied the orders for operations. A small Army of the Centre, operating around Madrid itself, would be the only corps directly controlled by the king.
One week after their meetings at Rambouillet, Napoleon began to rethink even his limited concessions. Above all, the emperor was loath to appoint a single marshal as primus inter pares directing operations from Madrid. However logical that might be militarily, it carried all sorts of risks to the emperor’s personal authority. Writing to Berthier from Normandy on 27 May, Napoleon revealed the extraordinary depth of his own egotism:
I cannot give away that supreme command, because I do not see any man capable of managing the troops … it is in the nature of things that if one Marshal were placed in Madrid, and directed all operations, he would want to have all of the glory along with all of the responsibility.
The emperor added lamely, ‘I want to do all that I can to give the king a new prestige on his return to Spain.’
Berthier had to make sense of this sibling rivalry and emotional insecurity masquerading as military directive. As Napoleon’s right-hand man for more than a decade, Berthier lived on his nerves. When contemplating some conundrum of army administration, he paced up and down in his office and gnawed at his fingers until they bled. This was a real dilemma. The emperor wanted to dictate the overall direction of the war, that was clear. These gentlemen – Marmont, Soult and the like – wanted to decide the details and Berthier knew what difficult customers they could be. Now, on top of all this, he needed to make Joseph feel important, when in fact he had been given almost nothing. He must be kept informed, that was vital. The note agreed at Rambouillet would lead Joseph to expect a constant flow of letters from the provinces. That carried risks, of course, Berthier knew that. If Marmont and Soult were writing to Madrid as well as to him in Paris, there would be sensitive bits of parchment flying all over the Iberian Peninsula. A grand chiffre, that was the answer. That was prestige all right, a great cipher for a great monarch.
*
Restored to his palace in Madrid, Joseph began his courtly life once more. He had vented his frustrations at Rambouillet and, for the moment, felt better for it. A military chief of staff was to join him soon. Reports began to arrive from the far-flung armies. There appeared to be some hope of concerting military operations with his plans for the edification of the Spanish nation. Once he could dedicate more lycées, eradicate the superstitious influence of the Church in matters of law and trounce a few of those guerrillas, then things would begin to turn.
On 12 July, Berthier wrote to Joseph, briefing him on the latest military developments. The letter was in a grand chiffre, a complex code. Not the sort of thing the Army had played about with in Italy or the Balkans; this was a proper diplomatic cipher that Berthier had obtained from Hugues Maret, the Secretary of State, and his civilian opposite number in matters both of routine and sensitive imperial administration.
The arrival of Berthier’s missive caused consternation at Joseph’s palace. Nobody could decipher it. The king’s secretary, seeing the cipher for the diplomatic one that it was, sent for the comte de Laforest, the French ambassador in Madrid. A few days later the reply came back that the Embassy had no knowledge of this particular grand chiffre. On 27 July, Joseph had to write back to Berthier, ‘we are not in possession of this cipher’. Could the prince possibly send his message again, only this time, ‘in the French ambassador’s actual cipher, and send me one that can be used by the two of us in particular’. A couple of days after he wrote to the chief of staff in Paris, the thought seems to have occurred to Joseph that perhaps Berthier had slipped up in his dispatch of 12 July and revealed a cipher that Paris had in common with the army commanders but not with Madrid. Was this code, far from being a tool for his authority, something which was being used to circumvent it?
On 5 August, Joseph sent a messenger to the commander of the Army of the North with a letter about various administrative matters but which included the telling ‘Mr Frochot will inform you about a mishap that prevents me understanding a dispatch that must be most important, judging by the means in which it is ciphered. If you are able to let me know any more about it, you would give me pleasure.’
Five days later, abandoning his earlier coyness about the sender of the 12 July letter, the king wrote to Marshal Marmont, asking for enlightenment, ‘on the contents of a ciphered dispatch that I cannot read’. On the same day, Soult was sent a letter also containing a copy of Berthier’s original message and ‘hoping you might be able to shed light on the contents’.
By now, the arrival of the grand chiffre in Spain, despite Berthier’s hopes, had become a farcical episode. By failing to ascertain exactly which deciphering tables were in the possession of the king or French ambassador, Berthier had simply compounded Joseph’s insecurity. Further months had been lost in which Wellington had continued to receive intercepted high-level dispatches en clair – uncoded, revealing much about the state of the French armies and their operations. The British Staff had also learned something of the king’s unhappiness over his powerlessness and his visit to Paris. Wellington’s military secretary wrote: ‘[we] greatly pity the poor King Joseph for I am convinced nothing but force could have made him return to this country.’
*
As the matter of granting Joseph a semblance of power by trying to give him a great cipher played itself out, Marshal Marmont found himself operating once more in the plains of Leon and Castile, close to northern Portugal. Having returned from helping Marshal Soult in Estremadura, he put into effect his earlier plans to improve the security of his communications. This had become all the more important as he faced new threats from the British and their allies.
French troops still held Ciudad Rodrigo, the fortified town on the River Agueda that was now the subject of Wellington’s attentions. This strong point was in the border highlands, sparsely inhabited and studded with forests of black and pygmy oak. Around Rodrigo itself, to a radius of a couple of miles, was a verdant pasture which allowed sufficient cultivation and husbandry to sustain the people of the town but not invading armies. A siege, or indeed a defence, of this place posed considerable difficulties of supply. Wellington understood that if an army of 20,000 or 30,000 troops was to canton the area around it for weeks while the engineers conducted their regular approaches, large amounts of food and ammunition would have to be brought up to them.
Marmont, on the other hand, was obliged by considerations of supply to keep his army around Salamanca, about forty miles to the north-east in an arc down towards the River Tagus which stretched most of the way to Badajoz. Although the Salamanca plain had once been an area of agricultural abundance, the war had ravaged its farms, and the Army of Portugal had been left chronically short of the supply wagons needed to provide for this dispersed host. All the same, forces in Salamanca were still close enough to Ciudad Rodrigo for a movement of three or four marches to allow them to drive off any British force blockading it. Marmont had resupplied the fortress in July, but by August it
was clear to him that Wellington was preparing to attack it. As he meditated his response, Marmont’s generals began cloaking their orders in the Army of Portugal cipher.
The British general’s plan involved a blockade by Spanish irregulars, British cavalry and light troops while heavy guns were moved ponderously up to the frontier. It was in the struggle to seal off Ciudad Rodrigo from resupply or information that the guerrilla chief Don Julian Sanchez came into his own. If this blockade could be maintained, the garrison might be starved out, or at least reduced in numbers and spirit.
Don Julian was a native of the lands between Rodrigo and Salamanca. He and many of his horsemen had grown up riding the borderlands, hunting in its forests. Their favourite chase involved the pursuit of jabalies, wild pigs. Roasted, these were a local delicacy. The animals shot through the trees at a fearful pace, defying their pursuers to skewer them with their lances or picos. In August 1811, the don’s quarry was human but his hunt was pursued with the same tactics and often with as little mercy.
Sanchez had served as a non-commissioned officer in the Spanish army until the death of his father in 1803 forced him into civilian life. He had re-enlisted in a volunteer cavalry regiment in 1808, following the outbreak of what the Spanish called their War of Independence against the French. He had then risen swiftly through the ranks. In the summer of 1810 his men surprised a company of about a hundred French dragoons. Eighty had been killed, the few survivors testifying to an attack of merciless ferocity. After this, one British officer noted in his journal, ‘the French promise to hang him and he in return gives them no quarter’. The don and his men sported curled moustaches, carried fearsome lances, tucked pistols into their gaudy red sashes and in general resembled the worst nightmare of every French convoy commander.
As a former soldier, the don was keener than many other guerrillas to give his men some semblance of uniform and order, but even so, one British officer drew this memorable pen portrait:
a more verminous set of fellows you never beheld. The infantry in English clothing and the cavalry, both horse and man, completely armed and equipped in the spoils of the enemy, so that it is next to impossible to distinguish friend from foe. The Don himself wears a pelisse like the 16th Dragoons with an immense hussar cap and the Eagle of Napoleon reversed. In this dress, [he is] accompanied by two aides-de-camp equally genteel in appearance, twelve lancers, a trumpeter on a grey horse.
Sanchez had been brought under British pay in October 1810, receiving silver coin and weaponry from swords to light cannon. The following year his troops (perhaps 250 cavalry and twice as many on foot) had been attached to Scovell’s corps of Guides. They were vital as scouts and in collecting the secret intelligence scrawled on scraps of paper by agents in Ciudad Rodrigo or Salamanca.
In late August and early September, several dispatches were brought in by Don Julian and some other guerrillas in the curious new cipher of the Army of Portugal. Wellington and Somerset looked them over. They were not the simple codes that had been used earlier in the summer, that was clear. Scovell was brought in to help.
Marmont’s code consisted of numbers from 1 to 150. To understand how it defeated the simple analysis prescribed by Conradus in Scovell’s notebook, one needs to examine the patterns in just one example.
A portion of a message sent by one of Marmont’s Staff is written out in 711 code numbers. The great majority of these figures represent a single letter, although with the 150 possibilities in this cipher, some whole words like ‘Marmont’ or ‘the enemy’ could be written as a single coded number. The rules of French composition described by Conradus hold true to the original text, but the cipher succesfully hides them. Fully 131 of those 711 code numbers stand for e, the most frequently used letter in French. The cipher, however, allocates nine different code numbers to the writing of this most common letter. Other vowels each have several alternatives, obscure consonants only one or two. In this way the cipher balances out the patterns detected by earlier decipherers and conceals them.
Looking at this particular letter from the chief of staff, Scovell would have seen that the most common code was 14, which had been written thirty-five times in the 711 characters. Needless to say, it does not stand for a common vowel, but for r. Little wonder that Scovell wrote on one of the facing pages of his decipherer’s notebook ‘the art of writing in cypher is so much improved since Conradus wrote as to render it next to impossible (when knowing the language made use of) to unravel what it conceals without being in possession of the key’.
He was not a defeatist, though, and what may be ‘next to impossible’ may equally be possible. Scovell believed the cipher could be broken by two principal methods: the writers usually mixed code with clear text to save time and evidently the context could tell you much. Secondly, once more messages had been brought in, the small fissures opened up by examining context could be widened into great cracks by making comparisons.
A message from General Montbrun, commanding Marmont’s cavalry division, to the Governor of Ciudad Rodrigo provides some sense of how Scovell attacked the cipher as he sat at his table in Fuente Guinaldo, surrounded by scraps of paper brought in by the guerrillas. In its first paragraph, Montbrun acknowledged the previous communication from the besieged town and went on:
I am making haste to pass on the contents to 25. 13. 8. 9. 38. 19. 18. 37. 14. 10. 33. 28. 17. 34. 17. 26. 5. 19. 21. 23. 31. 32. who has ordered me to open communications with you.
This was a very good passage to begin the attack, since it had been stupidly enciphered by Montbrun or, more likely, one of his ADCs: it was a good bet that he was referring to orders from his chief, Marmont. It was then a matter of trying every possible way that the marshal might be referred to in French and seeing whether it had the right number of letters: ‘M le Maréchal Marmont’, not long enough; ‘M Le Maréchal, Duc de Raguse’, one letter too short. Perhaps ‘S. E. Le Maréchal, Duc de Raguse’ (the ‘S. E.’ standing for His Excellency): that was the right number of letters. It was correct too, since the two code numbers that occur twice (19 and 17 for a and d) fit exactly. With this partial knowledge of the cipher, revealing the meaning of twenty-one of the coding numbers, the decipherer could continue with his attack. The last paragraph contained a long body of code, running across several lines. But how did its beginning, where it switched from clair to code in full flow, look with the knowledge of those characters used to make up Marmont’s name?
It would seem that there are not many people you can count on there. 30. r. 15. 30. 15. s. 55. 33. L. h. 15. m. m. e. 27. u. 49. 47. e. u. 15. u. s. e. 16. u. 29. e.
It was evident to any decipherer that he still had to find two vowels, since he had only discovered a, e and u. In any case there were various ways of writing vowels, so perhaps it was not the best way to proceed. What about consonants, since some of those might only have a single code number and several (like b, d, f, j, k, n, p, q, t, v, x, y and z) were unknown to him? With the knowledge that the second letter of the first word is r, it is then a matter of methodical puzzle-solving. The repetition of the codes 30 and 15 eventually suggests ‘proposez’ to the patient decipherer. The deduction proceeded in stages: for example, does 15 signify o? Comparison with other 15s later in the text quickly allows some sense of whether the supposition is right once the letters derived from higher up in the message had been added in. For the above passage, the decipherer will eventually discover a peculiarity of Marmont’s cipher, that u and v are interchangeable in the coding and that where he has assumed a u occurs in this later portion, the codes 23 and 34 actually stand for v in this different context.
After long hours of study, then, the last paragraph reveals itself as a most interesting item in the battle for information: ‘It would seem that there are not many people you can count on there. Proposez à l’homme que je vous envoie …’ Translating the whole of this last paragraph from French into English,
It would seem that there are not many people you can count on there. Suggest to
the man I am sending you that he should search out the English in Gallegos and Fuente Guinaldo, and come back through El Bodon and you will send him back to me forthwith. Tell him I will pay him well if he wants to make this trip, but if he refuses, I ask you not to force him.
The fate of Montbrun’s unnamed spy is not known. Being locals, Don Julian’s men were able to sniff out anyone who did not belong in the area fairly quickly. A thorough search would then be conducted of his clothing and possessions. The discovery of some ciphered scrap of paper usually meant death for the collaborator. It is evident from Don Julian’s operations in his native region, though, that some of these terrified messengers declared themselves to the guerrillas as soon as they met them, in order save their lives. This set Scovell thinking about the uses such a man might be put to.
Apart from showing the quality of Marmont’s cipher, Montbrun’s message also revealed that a large convoy of supplies was being collected in Salamanca in readiness to throw into the fortress. This confirmed the reports of spies. The one problem with the deciphering of the French general’s text was with numbers. Only five occurred and it was not possible to deduce what the code numbers stood for. In this way, a critical sentence about the date when the supply convoy might set out for Ciudad Rodrigo was not deciphered. More messages in the same code would be needed before these numbers could be made out. On 18 September, Wellington wrote to the secretary of war in London: