There were, of course, plenty of towns and castles susceptible of defence.  At one time or another every important settlement upon the Thames was capable of resistance:  Oxford was walled, Wallingford was a fortress, Abingdon or Reading could be defended.  But these were all, so to speak, artificial.  The settlement came first, and after the settlement the necessity of guarding it from attack, and it was so guarded, not by natural means, but by human construction.  The castle at Oxford, for instance, stood upon a mound of earth raised by human work.  The only considerable place in which the river itself suggested defence from the earliest times appears to have been at Dorchester.

The curious importance of Dorchester in the very origins of English history and the still more curious way in which it sinks out of sight for generations, to revive again in the tenth century, is one of the puzzles of the history of the Thames.

It is useless to pursue an archaeological discussion as to the origin of the place, and still more useless to try and determine why, though certainly the most easily defended, it should originally have been the only heavily fortified spot in the whole of the valley.  We know that it was Roman:  we know that it was a place of pre-historic fortification before the Romans came:  we know that a Roman road ran northward towards Bicester from it, and we also know, or at least we can make a very probable guess, that though it was continuously important, and that the interest of early history is continually returning to it, it can never have been large.

Perhaps the best conjecture upon the origin of Dorchester is that the stronghold grew up as an out-lier to the great fort over the river at the top of Sinodun Hill.  The exact and regular peninsula between the bend in the Thames and the mouth of the Thames is obviously suited for fortification:  the tributary flows just to the east of this peninsula, exactly parallel with the main river beyond, and covers the peninsula not only with a stream on its east flank, but with a marsh at the mouth.  One can imagine that the conspicuous heights of the Sinodun Hills were held, from the very beginning of human habitation in this district, as a permanent fortress, into which the neighbouring tribes could retire during war, and one can imagine that when the river was low in summer, and perhaps fordable, the spit of land before it, which formed an exception to the marshes round about, needed to be protected as a sort of bastion beyond the stream.  This theory will at least account for the two great ridges of earthwork going from one water to the other and completely cutting off the peninsula, since it is agreed these works are earlier than the Roman invasion.  Whatever its origin, the part which Dorchester plays in the early history of England is most remarkable.

The conversion of England was effected by a process of which we know far more than of any other series of national events before the Danish invasions.  That process is more exactly recorded, less legendary, and more consecutively told because it was (to all contemporary watchers) the capital event of the time, and to all posterity the one thing that explained men to themselves.