Country roads, however, existed between the Sierra de Culebra and the bend of the Douro between Zamora and Miranda, and the whole detail of Graham’s march depended on their practicability: Wellington had caused the whole region as far as the Esla to be explored by his intelligence officers, and the report had been that the movement of all arms was possible in the summer. The permanent bridges of the Esla, from Benavente down to its confluence with the Douro, had been broken long ago; but there was a certain number of fords, and convenient places for the laying of temporary bridges. By the end of May it was calculated that the spring floods due to the melting of snows in the Galician and Asturian mountains would be over, and that the river would be down to its normal low summer level. On these facts depended the success of the operation, which must be a rapid one, in order that the French might have no time for concentration. But Wellington’s provident mind had taken into consideration the possibility of unexpected high-water, and as a matter of precaution he had ordered up his main pontoon-train from the Tagus. To get the cumbrous pontoons, 33 of them, from Abrantes to Miranda de Douro by land was no small matter. They travelled slowly on specially constructed wheeled trucks[476] by Castello Branco, Sabugal, and Pinhel, crossed the Douro at Villa Nova de Foscoa, and were then brought up on the Miranda high road as far as Villa Velha, where they were halted for a short time, as Wellington was not quite certain whether he would not lay them at Espadacinta on the middle Douro, rather than farther up, beyond Miranda. For across the pontoons he intended to lay the main line of communication between Graham’s and Hill’s wings of the army, and if the latter failed to get forward on the Tormes and to open up touch with Graham via Zamora, it might be necessary to throw the bridge at some lower point, such as Espadacinta[477]. On the 20th he made up his mind that 14 pontoons should be sent to Espadacinta, while the remaining 19 should proceed to Miranda, and from thence follow the course of the Douro to the point where they would be laid--probably the ferry of Villa al Campo, a mile below the confluence of the Esla and the Douro. The nineteen all-important trucks with their burdens reached Miranda safely, and moved close behind the right-hand column of Graham’s army during its advance. Down to May 26 all the British troops had been kept behind the Portuguese frontier, nothing having been sent beyond it save D’Urban’s Portuguese dragoons, who formed a screen some little way on the Spanish side. Their appearance would, in the event of a French cavalry raid from Zamora, create no suspicion of there being in the rear anything more than the usual Tras-os-Montes militia. Nor could spies draw any deduction from their presence. But on the 26th the whole of Graham’s army[478] started out in three columns arranged as follows: