Marmont had the easiest passage over the Tormes, namely, that by the ford of Huerta; the allies had the greatest number of passages and the shortest line of operations. Hence if Graham had been ordered vigorously to attack the French troops on the left bank, they must have been driven upon the single ford of Huerta, if not reinforced from the heights of Aldea Rubia. But the allies couldSee [Plan, No. 2.] also have been reinforced by the fords of Santa Marta and those of Cabrerizos, and even by that of Aldea Lengua, although it was not good at this early season. A partial victory would then have been achieved, or a general battle would have been brought on, when the French troops would have been disadvantageously cooped up in the loop of the Tormes and without means of escaping if defeated. Again, it is not easy to see how the French general could have avoided a serious defeat if Wellington had moved with all the troops on the right bank, against the divisions left on the hill of Aldea Rubia; for the French army would then have been separated, one part on the hither, one on the further bank of the Tormes. It was said at the time that Marmont hoped to draw the whole of the allies across the river, when he would have seized the position of Christoval, raised the siege and maintained the line of the Tormes. It may however be doubted that he expected Wellington to commit so gross an error. It is more likely that holding his own army to be the quickest of movement, his object was to separate the allies’ force in the hopes of gaining some partial advantage to enable him to communicate with his forts, which were now in great danger.

When the French retired to the heights at Aldea Rubia on the night of the 23d, the heavy guns had been already brought to the right of the Tormes, and a third battery, to breach San Cajetano, was armed with four pieces, but the line of fire being oblique, the practice, at four hundred and fifty yards, only beat down the parapet and knocked away the palisades. Time was however of vital importance, the escalade of that fort and La Merced was ordered, and the attack commenced at ten o’clock, but in half an hour failed with a loss of one hundred and twenty men and officers. The wounded were brought off the next day under truce and the enemy had all the credit of the fight, yet the death of general Bowes must ever be admired. That gallant man, whose rank might have excused his leading so small a force, being wounded early, was having his hurt dressed when he heard that the troops were yielding, and returning to the combat fell.

The siege was now perforce suspended for want of ammunition, and the guns were sent across the river, but were immediately brought back in consequence of Marmont having crossed to the left bank. Certain works were meanwhile pushed forward to cut off the communication between the forts and otherwise to straiten them, and the miner was attached to the cliff on which La Merced stood. The final success was not however influenced by these operations, and they need no further notice.

The 26th ammunition arrived from Almeida, the second and third batteries were re-armed, the field-pieces were again placed in the convent of San Bernardo, and the iron howitzers, throwing hot shot, set the convent of San Vincente on fire in several places. The garrison again extinguished the flames, and this balanced combat continued during the night, but on the morning of the 27th the fire of both batteries being redoubled, the convent of San Vincente was in a blaze, the breach of San Cajetano was improved, a fresh storming party assembled, and the white flag waved from Cajetano. A negociation ensued, but lord Wellington, judging it an artifice to gain time, gave orders for the assault; then the forts fell, for San Cajetano scarcely fired a shot, and the flames raged so violently at San Vincente that no opposition could be made.

Seven hundred prisoners, thirty pieces of artillery, provisions, arms, and clothing, and a secure passage over the Tormes, were the immediate fruits of this capture, which was not the less prized, that the breaches were found to be more formidable than those at Ciudad Rodrigo. The success of a storm would have been very doubtful if the garrison could have gained time to extinguish the flames in the convent of San Vincente, and as it was the allies had ninety killed; their whole loss since the passage of the Tormes was nearly five hundred men and officers, of which one hundred and sixty men with fifty horses, fell outside Salamanca, the rest in the siege.

Marmont had allotted fifteen days as the term of resistance for these forts, but from the facility with which San Vincente caught fire, five would have been too many if ammunition had not failed. His calculation was therefore false. He would however have fought on the 23d, when his force was united, had he not on the 22d received intelligence from Caffarelli, that a powerful body of infantry, with twenty-two guns and all the cavalry of the north, were actually in march to join him. It was this which induced him to occupy the heights of Villa Rubia, on that day, to avoid aConfidential official reports, obtained from the French War-office, MSS. premature action, but on the evening of the 26th the signals, from the forts, having indicated that they could still hold out three days, Marmont, from fresh intelligence, no longer expected Caffarelli’s troops, and resolved to give battle on the 28th. The fall of the forts, which was made known to him on the evening of the 27th, changed this determination, the reasons for fighting on such disadvantageous ground no longer existed, and hence, withdrawing his garrison from the castle of Alba de Tormes, he retreated during the night towards the Duero, by the roads of Tordesillas and Toro.

Wellington ordered the works both at Alba and the forts of Salamanca to be destroyed, and following the enemy by easy marches, encamped on the Guarena the 30th. The next day he reached the Trabancos, his advanced guard being at Nava del Rey. On the 2d he passed the Zapardiel in two columns, the right marching by Medina del Campo, the left following the advanced guard towards Rueda. From this place the French rear-guard was cannonaded and driven upon the main body, which was filing over the bridge of Tordesillas. Some were killed and some made prisoners, not many, but there was great confusion, and a heavy disaster would have befallen the French if the English general had not been deceived by false information, that they had broken the bridge the night before. For as he knew by intercepted letters that Marmont intended to take a position near Tordesillas, this report made him suppose the enemy was already over the Duero, and hence he had spread his troops, and was not in sufficient force to attack during the passage of the river.

July. Marmont, who had fortified posts at Zamora andSee [Plan, No. 3.] Toro, and had broken the bridges at those places and at Puente Duero and Tudela, preserving only that of Tordesillas, now took a position on the right of the Duero. His left was at Simancas on the Pisuerga, which was unfordable, and the bridges at that place and Valladolid, were commanded by fortified posts. His centre was at Tordesillas, and very numerous, and his right was on some heights opposite to Pollos. Wellington indeed caused the third division to seize the ford at the last place which gave him a command of the river, because there was a plain between it and the enemy’s heights, but the ford itself was difficult and insufficient for passing the whole army. Head-quarters were therefore fixed at Rueda, and the forces were disposed in a compact form, the head placed in opposition to the ford of Pollos and the bridge of Tordesillas, the rear occupying Medina del Campo and other points on the Zapardiel and Trabancos rivers, ready to oppose the enemy if he should break out from the Valladolid side. Marmont’s line of defence, measured from Valladolid to Zamora, was sixty miles; from Simancas to Toro above thirty, but the actual line of occupation was not above twelve; the bend of the river gave him the chord, the allies the arc, and the fords were few and difficult. The advantage was therefore on the side of the enemy, but to understand the true position of the contending generals it is necessary to know the secondary coincident operations.

While the armies were in presence at Salamanca, Silveira had filed up the Duero, to the Esla river, menacing the French communications with Benavente. D’Urban’s horsemen had passed the Duero below Zamora on the 25th and cut off all intercourse between the French army and that place; but when Marmont fell back from Aldea Rubia, D’Urban recrossed the Duero at Fresno de la Ribera to avoid being crushed, yet immediately afterwards advanced beyond Toro to Castromonte, behind the right wing of the enemy’s new position. It was part of Wellington’s plan, that Castaños, after establishing the siege of Astorga, should come down by Benavente with the remainder of his army, and place himself in communication with Silveira. This operation, without disarranging the siege of Astorga, would have placed twelve or fifteen thousand men, infantry, cavalry, and artillery, behind the Esla, and with secure lines of retreat; consequently able to check all the enemy’s foraging parties, and reduce him to live upon his fixed magazines, which were scanty. The usual Spanish procrastination defeated this plan.

Castaños, by the help of the succours received from England, had assembled fifteen thousand men at Ponteferada, under the command of Santocildes, but he pretended that he had no battering guns until sir Howard Douglas actually pointed them out in the arsenal of Ferrol, and shewed him how to convey them to the frontier. Then Santocildes moved, though slowly, and when Bonet’s retreat from the Asturias was known, eleven thousand men invested Astorga, and four thousand others marched to Benavente, but not until Marmont had called his detachment in from that place. The Spanish battering train only reached Villa Franca del Bierzo on the 1st of July. However the Guerilla chief, Marquinez, appeared about Palencia, and the other Partidas of Castile acting on a line from Leon to Segovia, intercepted Marmont’s correspondence with the king. Thus the immense tract called the Campo de Tierras was secured for the subsistence of the Gallician army; and to the surprise of the allies, who had so often heard of the enemy’s terrible devastations that they expected to find Castile a desert, those vast plains, and undulating hills, were covered with ripe corn or fruitful vines, and the villages bore few marks of the ravages of war.