In shape, the alluvial fragments vary very much, according to their mineral nature and the amount of rolling they have received in transport. Foliated rocks, like schists and slates, produce typically flaky fragments which seldom attain any high degree of rounding; the same is true of serpentines which have been shattered by crushing. Granitic rocks produce generally more symmetrical fragments, though granitic sand is typically angular unless it has been transported for considerable distances. The larger blocks are generally rounded in the case of granitic rocks, even when they have not travelled for any great distance; for if a cubical block of granite be exposed to weathering, disintegration takes place most rapidly at the corners, and the block eventually becomes more or less spheroidal without having moved at all. This can be well seen in such masses as Gebel Selaia, where the granite boss is thoroughly well rounded _in situ_. The same phenomenon can be noticed in certain dolerites, where portions of the rock detached by joint planes have weathered _in situ_ into forms like cannon balls. The more homogeneous rocks like felsite and serpentine, as well as the foliated rocks like gneisses, schists, and slates, do not show much of this rounding _in situ_, as they yield to the stresses of expansion and contraction by breaking along planes of crushing or foliation, producing separate masses which preserve most of their angular shape. In the lower reaches of the wadis and on the plains, of course, all the materials are more or less rounded, owing to the attrition to which they have been subjected during transport. Of the thickness of the alluvial sands and gravels very little is known. But that it exceeds eight metres in many of the wadis is proved by wells sunk to that depth. On the coast-plain it may be vastly thicker, for Grabham[122] mentions that borings made in the maritime plain near Port Sudan have been carried down to 1,000 feet (305 metres) in the deposits. Only a very small proportion of the alluvial detritus brought down the wadis enters either the sea or the Nile; for the storm-formed streams, though violent in the mountainous tracts, commonly cease by being absorbed in the thirsty wadi floors and plains before reaching either the sea or the river. Bearing this in mind, and the ages through which the actions of denudation and transport have gone on, we can well imagine how great must be the total quantity of detritus accumulated. In some wadis there are high banks of old detritus through which the streams have cut their way. In the Wadi Meneiga, for instance, there are terraces of alluvium, some eight or ten metres high, on either side of the present water channel; these terraces, which are about two kilometres lower down the wadi than the wells, are covered with rude rubble ruins, the remains of dwellings erected out of reach of the occasional streams which rush down the wadi after rain.