THE IIOYAL ARCHAEOLOGICAL INSTITOTE. 397

•which divides the Avon from the Bourne. Tliese gi'avels consist of Btones washed out of the geological deposits which occur up-strcani, the upper greensand and tlie chalk. Blocks of ' sai-sen ' and beach pebljles are also present, indicating the former existence of Tertiary beds which rested upon the ciialk. " The flints, which form such a considerable part of these gravels, are sub-angidar, that is, tliey have not been subjected to sufficient rolling- action to reduce them to the condition of bcach-jjebbles, their general appearance being that of stones to be found in a river l)ed. Indeed, it apjjears that these gravels at Fisherton and Milford uctially formed the beds of rivers which long since flowed in the same direction as our Avon and Bourne, and that the present valleys have been excavated to the depth of from 70 to 80 ft. l)y the eroding action of these streams. " The excavating power of these rivers was doubtless formerly much greater than it is at present, and, in attemjiting to measure the requisite period which has elapsed for the excavation of these valleys to their present dejjth, this circumstance must be taken into account. From the character of the fauna, the comparative al)undance of the reindeer, the presence of the musk-sheep, etc., we know that the climate of the Quaternary Period was of great severity, and, consequently, that the accumulations of ice and snow upon our Wiltshire downs must have been considerable. " Proliably, towards the spring of each year, torrents, like the Shrcwton flood of 1841, resulting from a sudden and ra})id tliaw, swept down our valleys witli almost resistless force. But anotlier cause of floods has to be considered, and this is the formation of ground-ice, which played an important part in the transport of some of the larger blocks of ' sarsen,' and indeed of the gravels themselves. In rapid streams, when the water becomes sufficiently cold to reduce the tcmi)erature of the bed of the river to the freezing point, ground-ice is fretjuently formed. The gravel of the river-bed becomes coated with ice, and this ice, being lighter than water, after acquiring certain dimensions, rises to the snrface, carrying with it large stones and the gi*avel to which it adheres. Ground-ice is a great cause of floods in the upper part of the Rhine and the Danube. These livers have a rapid current, and do not freeze over their entire breadth, but large blocks of ice float upon the surface. Tiiese blocks are huiried along by the stream, impinge up<ni each other, become heaped together, and ultimately barricade the river. This accumulation of ice- drifts, however, is not itself the immediate cause of floods ; these take ]»lacc when a thaw 'commences in the ujiper part of the river, above the point where the latter is com])letely frozen, the masses of ice, drift- ing with the current and unable to pass, are hurled upon those already soldered together ; thus an enormous barrier is formed, which the water, arrested in its course, caimot pass over, and hence overflows to the right and left, breaking the dykes, inundating the plains, and spreading devas- tation far and near.'* " It is at such times that the ground-ice gi-catly increases the mischief, for, becoming detached from the bottom and rising towards the snifacc, it unites itself to the under side of the masses already in place, and rcndei-s the b.irrier more difficult to remove. • KriRplhardt, " Ann.iles de Chiinie ct de riiysique," 18(36. Translated in " Sniith- aonian Report " for 1866, p 425.