To fix the fighting strength of Marmont’s army at Salamanca is comparatively easy. It consisted of the 49,636 officers and men accounted for by the return of July 15th printed on the next page, minus some 700 men lost at the combats of Castrillo and Castrejon [also called ‘combat of the Guarena’] on July 18, and such few hundreds more as may have fallen behind from fatigue during the long marches of July 20-1. Roughly speaking, it must have counted some 48,500 men, as opposed to Wellington’s 50,000. The French translators of Napier’s Peninsular War (Mathieu Dumas and Foltz) only give a table of June 15, which is of course a month out of date for Salamanca, and append a note that ‘deducting artillery, engineers, équipages militaires, officers, sergeants, and garrisons, as also losses between June 15 and July 15 they find the result of about 42,000 sabres and bayonets for the battle.’ Why any sane person should deduct officers, sergeants, and artillerymen from a fighting total I am unable to conceive, though contemporary British writers, including Wellington himself, often did so. But the results of adding to their ‘42,000 sabres and bayonets’ the list of 1,925 officers, 3,244 artillerymen and artillery train (both in the divisions and in the reserve), 332 engineers, and 742 équipages militaires, is to give the figure 48,343, which practically agrees with the total that I state above; if sergeants are added it would much exceed that total. We may take this, therefore, as fairly correct—bearing in mind that the 26 officers and 742 men of the équipages militaires cannot be counted as combatants.

These totals do not include the 23rd Léger (2 batts.) and the 2/1st Line, both from Thomières’s division, which were garrisoning Astorga, about 1,500 strong. Nor do they include the minor garrisons left at Toro, Zamora, Olmedo, Valdestillas, Tordesillas, Simancas, Cabezon, Medina del Campo, Puente de Duero, Tudela de Douro, Amagro, &c., which appear to have been altogether about 4,184 strong, nor the dépôts at Valladolid, 3,307 strong on June 15, but probably much less on July 15, when Marmont had remounted nearly 1,000 dismounted dragoons and picked up all detachments and convalescents that he could gather. Nor do they include the sick, who had been 8,633 on June 15th, and 8,332 on May 15th—probably the total in hospital was a trifle more on July 15, owing to the fatigues of the campaigns round San Cristobal in the latter days of June.

Parallel with the return of July 15th, I have printed that of August 1. The difference between the two—211 officers and 10,124 men—might be supposed to represent the losses in the campaign between those dates. It does not, however, because the total of August 1 represents not only the survivors from the battle of Salamanca, but all the men from garrisons evacuated after it, and from the Valladolid dépôt, who joined the colours after the disaster of July 22, in consequence of the district in which they were lying having been evacuated by the army. The garrisons of Toro and Zamora held out till they were relieved, that of Tordesillas surrendered to the Galicians: but the men from the other smaller garrisons and from the dépôts fell in to their respective corps before August 1. I imagine that we may take these additions to be some 5,000 men at least, but cannot give the exact figures, through being unable to say what the Valladolid dépôts (3,307 strong on June 15) amounted to on July 15.

After comparing the totals of the brigades and regiments shown under July 15 and August 1, we must proceed to show the reasons why, in individual cases, the regimental differences between the two sets of figures cannot be taken to represent the sum of the losses in the Salamanca campaign. The proof is clear.

THE ARMY OF PORTUGAL BEFORE AND AFTER THE BATTLE OF SALAMANCA

From two returns of effectives in the Archives of the Ministry of War, Paris,
dated July 15 and August 1, respectively.

N.B.—Guns, July 15, 78; August 1, 58; lost 7 12-pounders, 3 8-pounders, 9 4-pounders, 1 3-pounder. Horses, July 15, 4,278; August 1, 3,231. Draught horses, July 15, 2,037; August 1, 1,847. Équipages militaires, horses, July 15, 800; August 1, 331.

To these two tables we must append, as a side-light, the results of a compilation of the totals of officers killed and wounded at Salamanca, from Martinien’s admirable Liste des officiers tués et blessés pendant les guerres de l’Empire. This of course does not include unwounded prisoners.

General total 60 officers killed, 347 wounded at Salamanca and Garcia Hernandez; plus 7 officers killed and 27 wounded at the Guarena on July 18, and 2 wounded in minor engagements.

Loss in killed and wounded, not including unwounded prisoners, during the campaign, 67 killed, 376 wounded = 443 officers in all.

After arriving at this general loss in killed and wounded officers, so far as is possible from Martinien’s tables, which are not quite complete for all corps, it only remains to estimate the unwounded prisoners. I searched the immense volumes of rolls of French officers in captivity at the Record Office, and found 63 names of prisoners taken at Salamanca, the Guarena, and Garcia Hernandez. A few of these duplicate the names of wounded officers to be found in Martinien’s tables, the remainder must represent the unwounded prisoners. Wellington in his Salamanca dispatch wrote that he had 137 French officers prisoners—evidently the larger number of them must have been wounded, as only 63 were sent off to England that autumn. Probably many died in hospital. Prisoners are most numerous from the 101st, 22nd, and from Foy’s two regiments cut up at Garcia Hernandez, the 76th and 6th Léger.

In the Library of the Archives de la Guerre at the Paris Ministry of War I went through the regimental histories of all the French infantry regiments present at Salamanca. Like our own similar compilations, they differ much in value—some are very full and with statistics carefully worked out from regimental reports and pay-books; others are very thin and factless. Fourteen units give their losses, which I herewith annex:

Clausel’s Division: 25th Léger, 336; 27th Line, 159; 59th Line, 350.

Ferey’s Division: 70th Line, 111; 31st Léger, 340.

Sarrut’s Division: 2nd Léger, 202.

Maucune’s Division: 15th Line, 359.

Brennier’s Division: 17th Léger, 264; 65th Line, 359.

Thomières’s Division: 1st Line, 227; 62nd Line, 868; 101st Line, 1,000.

Bonnet’s Division: 120th, 458; 122nd, 527.

The total of this makes 5,560 for these fourteen corps; we leave fifteen others unaccounted for. As a rough calculation I suppose that we may hold that as these regiments lost, as we know from Martinien’s lists [which are not quite complete], at least 152 officers out of 5,560 of all ranks, then the other fifteen regiments with 181 officers killed or wounded must have lost something like 6,000. The vagaries of the proportion between officers and men hit are extraordinary in individual units, but these tend to rectify themselves on a large total consisting of many regiments. I therefore believe that 11,560 would be something very like the total loss killed and wounded in the French infantry. We have then to allow for some 40 unwounded officers taken prisoners, and corresponding to them perhaps 1,200 unwounded men. The total loss for the infantry would thus be 12,800. For cavalry and artillery, &c., 53 officers hit—as by Martinien’s tables—must imply something over a thousand men lost. We should thus arrive at a total of 14,000 for the casualties—the sum which I suggest in my text (p. 469).

To show the worthlessness of any attempt to deduce the French losses by a mere comparison of the official ‘morning states’ of July 15 and August 1, the following instances may suffice.

The 65th Line shows 59 officers and 1,527 men present on July 15, 52 officers and 1,302 men on August 1. The apparent loss is 7 officers and 225 men. But this unit’s regimental report shows 3 officers killed, 5 officers wounded, 204 men killed or prisoners, 106 men wounded, 39 missing; total, 8 officers and 349 men. Therefore, as is obvious, one officer and 124 men must have joined from somewhere (dépôt at Valladolid?) between the two dates, or the deficiency would be 125 greater between the ‘present under arms’ of the two dates than is shown.

A more striking case is the 62nd Line, of Thomières’s Division. It shows present on July 15, 47 officers and 1,076 men, on August 1st 45 officers and 1,048 men—the apparent loss is only 2 officers and 28 men. But Martinien’s lists show us that the regiment lost at least 15 officers, killed and wounded, and the regimental report gives 20 officers and 848 men killed, wounded, or missing! The real loss is 868 not 30! Therefore 18 officers and about 800 men, the equivalent of a strong battalion, must have joined between July 15 and August 1. This corresponds to the fact that the 62nd showed only 2 battalions[787] at Salamanca, while the ‘morning state’ of June 15th showed it as having at the front three battalions and 1,900 rank and file. Clearly the third battalion rejoined the colours after the battle—having presumably been quartered in the small garrisons of Castile evacuated after the disaster of July 22. Many men must also have rejoined the other two battalions.

But the most absurd case of all is that of the 47th Line, whose total figures actually go up from 1,625 to 1,712 of all ranks between July 15th and August 1st—in despite of the fact that it lost (as Martinien’s lists show), 18 officers and not less therefore than 360 rank and file (20 men per officer is a low allowance) at Salamanca. It must have picked up from Valladolid and the small garrisons 13 officers and 452 men at least[788].

Clausel, writing to King Joseph on July 25, said that of the whole Army of Portugal he could not yet show in the field on that day 20,000 men. This tallies well enough with the conclusion that we have already drawn, that the total loss from the army, which on July 15 had about 48,000 men, must have been some 14,000 killed, wounded, and prisoners, and over 10,000 men dispersed who were only just rallying.