In order to answer that question let us consider what other points in the valley were at his disposal.

Reading we have discussed.  The chalk spurs in the gorge by Goring and Pangbourne are not isolated (as is that of Chateau Gaillard, for instance), and are dominated by the neighbouring heights.  The escarpment opposite Henley offered a good site for an eleventh-century castle—­but the steep cliff of Windsor had this advantage beyond all the others—­that it was at exactly the right distance from London.  Windsor is the warden of the capital.

If the reader will look at a modern geological map, he will see from Wallingford to Bray a great belt of chalk in which the trench of the Thames is carved.  Alluvials and gravels naturally flank the stream, but chalk is the ground rock of the whole.  To the west and to the east of this belt he will notice two curious isolated patches, detached from the main body of the chalk.  That to the west forms the twin height of the Sinodun Hills, rising abruptly out of the green sand; that to the east is the knoll of Windsor, rising abruptly out of the thick and damp clay.  It is a singular and unique patch, almost exactly round, and as a result of some process at which geology can hardly guess the circle is bisected by the river.  If ever the chalk of the north bank rose high it has, in some manner, been worn down.  That on the south bank remains in a steep cliff with which everyone who uses the river is familiar.  It was the summit of this chalk hill piercing through the clays that the Conqueror noted for his purpose, and he was, to repeat, determined (we must presume) by the distance from London.

The command of a great town, especially a metropolis, is but partially effected by a fortress situated within its limits.  In case of a popular revolt, and still more in case the resources of the town are held by an enemy, such a fortress will be penned in and find itself suffering a siege far more rigorous than any that could be laid in an open country-side.  On this account the urban fortresses of the Middle Ages are to be found (at least in large cities) lying upon an extreme edge of the walls and reposing, as far as possible, upon uninhabited land or upon water, or both.  The two classic examples of this rule are, of course, the Tower and the Louvre, each standing down stream, just outside the wall, and each reposing on the river.

But in an active time even this precaution fails, and that for two reasons.  First, the growth of the town makes any possible garrison of the fortress too small for the force with which it might have to cope; and, secondly, this same growth physically overlaps the exterior fortress; suburbs grow up beyond the wall, and the castle finds itself at last embedded in the town.  Thus within a hundred and fifty years of its completion the Louvre was but a residence, wholly surrounded, save upon the water front, by the packed houses within the new wall of Marcel.