We must not confuse with these dead valleys the long depressions, with no outlet, which are scattered over the granite and sandstone tableland (_bajos_, _valles_, _cuencas_). Some have obstinately, but wrongly, sought in these the traces of rivers that have disappeared; and the _bajos_ of Gualicho and Valcheta have wrongly been regarded as the former bed of the Rio Negro and the Limay. Erosion by wind seems to have had something to do with these depressions. Their persistence, at all events, is one of the effects of the aridity which prevents normal erosion from moulding the surface of the tableland. The chief of them are centres for collecting running water. There is a group of valleys all round them, and alluvial beds accumulate in them. The climate determines the character of the soil in Patagonia. The rounded pebbles of granite and eruptive rock, so often described since the time of Darwin, sometimes free and sometimes embedded in red sand or limestone,[51] are spread over the tableland like aureoles round the masses of rock, and they are particularly abundant in the coast region. On the Rio Negro they seem to be confined to the vicinity of the valley; they disappear as one goes away from it. The progressive reduction in the volume of the Rio Negro gravels, as one goes downward, has been observed to begin in the Andean zone, and it is from the Andes that they come. South of Santa Cruz, in a moister climate, in which the circulation of the water is less localized, the bed is more continuous, and it covers the Tertiary sandstones and clays. It is of fluvio-glacial origin, and comes from the destruction of the old moraines, before the excavation of the actual valleys. But it is the wind that explains the concentration of the gravel at the surface. It separates the pebbles from the more mobile material about them. Wherever the outcrop-strata contain pebbles, the wind eventually converts the place into a field of shingle. It has done this with the terraces of the Limay. The Tertiary marine deposits of the coast region also are rich in pebbles torn from the rocky promontories of the shore; hence the extent of stony soils in the coast region. The wind similarly strips naked the angular stones, of local origin and incompletely worn, round the isolated rocks of the desert tableland or on the flanks of the secondary ravines. [51] The calcareous flag-stone of La Tosca, which is characteristic of the south-west province of the plain of the Pampa, stretches in the south as far as the Rio Negro in the coast-district. On the other hand, it is almost entirely absent a hundred miles to the west, between the Colorado and the Rio Negro, along the line of the railway from Fortin Uno to Choele Choel.