But even these are not all the troublesome questions requiring an answer. It will be remembered that Franklin's corps, after forcing Crampton's Gap, had remained in Pleasant Valley between Rohrersville and Boonsboro until Tuesday night (16th September). McClellan then ordered Couch's division to be sent to occupy Maryland Heights and observe the enemy in Harper's Ferry, whilst Franklin with Smith's and Slocum's divisions should march to the battle-field at daybreak of Wednesday. Why could not Couch be called up and come on our left as well as A. P. Hill's division, which was the last of the Confederate troops to leave the ferry, there being nothing to observe after it was gone? Couch's division, coming with equal pace with Hill's on the other side of the river would have answered our needs as well as one from Porter's corps. Hill came, but Couch did not. Yet even then, a regiment of horse, watching that flank and scouring the country as we swung forward, would have developed Hill's presence and enabled the commanding general either to stop our movement or to take the available means to support it. The cavalry was put to no such use. It occupied the centre of the whole line, only its artillery being engaged during the day. It would have been invaluable to Hooker in the morning, as it would have been to us in the afternoon. McClellan had marched from Frederick City with the information that Lee's army was divided, Jackson being detached with a large force to take Harper's Ferry. He had put Lee's strength at 120,000 men. Assuming that there was still danger that Jackson might come upon our left with his large force, and that Lee had proven strong enough without Jackson to repulse three corps on our right and right centre, McClellan might have regarded his own army as divided also for the purpose of meeting both opponents, and his cavalry would have been upon the flank of the part with which he was attacking Lee; Porter would have been in position to help either part in an extremity or to cover a retreat; and Burnside would have been the only subordinate available to check Lee's apparent success. Will any other hypothesis intelligibly account for McClellan's dispositions and orders? The error in the above assumption would be that McClellan estimated Lee's troops at nearly double their actual numbers, and that what was taken for proof of Lee's superiority in force on the field was a series of partial reverses which resulted directly from the piecemeal and disjointed way in which McClellan's morning attacks had been made.