One such route starts from Southampton and passing Winchester, crosses the Kennet at Newbury, where it was met perhaps by a route from Chichester. Thence it passed by the head waters of the Thames to a point on the Cotswolds not far from Cirencester, where it may have been joined by other routes from the south-west. It descended the scarp slope of the Cotswold at or near Broadway, crossed the Avon near Evesham, and the Severn at Bevere Island above Worcester. Thence it passed up the west side of the valley, crossing the river again below Shrewsbury. Its course across north Shropshire seems to have lain on the watershed between the Tern and the Perry, if we may judge from evidence of a later date,[117] thence passing from Ebnal towards Llanarmon-dyffryn-Ceiriog, it crossed over to the Dee Valley, where we can pick up fresh evidence near Corwen. From the head of Bala lake it seems to have turned slightly north of west, instead of passing down the Mawddach Valley, and it reached the coast somewhere to the north of Harlech, perhaps by the so-called Roman steps at Cwm Bychan. This is the best attested route so far traced out, but further work is required to establish its course with precision all the way. Another route from the Yorkshire coast through York to the Aire gap has been described by Colonel E. Kitson Clark,[118] while some years ago I traced several from the borders of the Fens into Leicestershire, where they met at Bardon Hill; thence the route passed through Ashby-de-la-Zouch as far as Burton-on-Trent, where it seemed to be pointing to the Peak district.[119] There appears to be a route running thence by Macclesfield and Knutsford towards Warrington, while there are signs that the route through the Aire gap also turned south towards the same spot. Near Warrington a number of flat axes have been found,[120] some on the north and some to the south. The northern settlement was in the parish of Winwick, and among the things found there and dating from this time is a battle-axe of the so-called boat-axe or _batyx_ type.[121] This type and the flint of which it is made both indicate Denmark as the place of origin. The fact that both these trade routes run to Warrington, which seems then to have been an island in the middle of the Mersey, shows, I think, that here we have a port, from which in the early bronze age Baltic traders set sail for Dublin Bay.[122] Warrington, therefore, rather than Chester, was the first predecessor of Liverpool, and the Mersey holds its own as the earliest estuary used for the western trade. Crawford’s map also shows that a similar trade route must have crossed Scotland from the Firth of Forth to the Firth of Clyde and the Mull of Galloway, but no details of such route have been worked out.