The crossing at Staines was the main bridge over the middle river during the Roman occupation; no other spot on the banks (except London Bridge) is certainly the site of a Roman bridge.

But apart from these there are two unsolved problems in connection with the roads across the Thames Valley in Roman times.  The first concerns the passage of the upper Thames south of Eynsham; the second concerns the road which runs south from Bicester and Alchester.

As to the first of these, we know that the plain lying to the north of the Thames between the Cotswolds and the Chilterns was thoroughly occupied.  We have also in the Saxon Chronicle a legendary account of the occupation of four Roman towns in this plain by the Saxon invaders.  By what avenue did this wealthy and civilised district communicate with the wealthy and civilised south?

It is a question which will probably never be answered.  There is no trace remaining of Roman bridges; perhaps nothing was built save of wood.

The obvious short-cut from the Roman town of Eynsham across the Witham peninsula to Abingdon bears no signs of a ford approached by Roman work or of a bridge, nor any record of such things.

As to the second question, the road from Bicester southward runs straight to Dorchester.  At Dorchester, as we have seen, there was no ford, though just below it a Roman ferry has been guessed at.

There may have been a country road running down along the left or north bank of the river to the pre-historic crossing place at Goring and Streatley; but if there was, no trace of it remains, save perhaps in the two place names North Stoke and South Stoke.

A barrier has yet another quality in history, and that quality is perhaps the most important of all.  In so far as it is an obstacle it is also a means of defence.

All the great rivers of Europe prove this.  They are studded with lines of strongholds standing either right upon their banks or close by; and various as is the character of the different great rivers in their physical conformation, few or none have been unable to furnish sites for fortification.  For instance, the slow rivers of Northern France, running for the most part through a flat country, were able to afford fortresses for the Gaulish clans in their numerous islands; the origin of Melun and Paris, for instance, was of this kind.  The sharp rocks along the Rhone became platforms for castle after castle:  Beaucaire, Tarascon, Aries, Avignon, and twenty others all of this sort.

The Thames, curiously enough, forms an exception; it is an exception even in the list of English rivers, most of which can show a certain number of fortifications along their banks.

In the whole course of the great river above London there are but three examples of fortification, or at any rate of fortification directly dependent upon the river.  Of these the first, at Lechlade, is conjectural; the second, at Windsor, came quite late in history, and the only one which seems to have been a primeval fortified site was Dorchester.