Vol. 59.] GEOLOGY OF TATAGONIA. 177 



either side, occur along their lower reaches, often, as in the 

 case of the Chico de Santa Cruz, forming ' valley-pampas ' of con- 

 siderable extent. The great width and depth of the valleys is 

 strangely disproportionate to the streams which flow through them, 

 even taking into account their swift currents. In earlier times, 

 however, the accumulation of surface-drainage flowing through 

 these valleys was much greater, the subsequent decrease being due 

 to a diminution of rainfall, the lowering of the levels of the lakes 

 under the Cordilleras and the consequent diversion of their waters 

 (in some cases at any rate, to the Pacific), and the capture of streams 

 which originally flowed to the Atlantic, by others feeding the lakes. 



So great has been the decrease that now some river-beds are 

 quite dry but for pools, and others contain water only in the spring. 

 The great caiiadon of the old Rio Salado, north of the Chico de 

 Santa Cruz, is the best example of the former case ; the smaller 

 canadon of the same name, north of Camerones Bay, is a good 

 example of the latter. 



I was fortunate in crossing the head of the great Canadon Salado 

 on my way to Santa Cruz. The only water flowing now into it is a 

 small stream rising under a basalt-plateau, the Rio Olnie (or Olin), 

 which after a few miles loses itself in a series of pools. North of the 

 point at which the Olnie enters is a moraine-covered pampa, where 

 was once a sheet of ice moving a few degrees east of south. This 

 sheet apparently met, in the canadon, another coming from the 

 direction of Colle Belgrano, with the result that a barrier of morainic 

 material, in which obsidian figures largely, was formed across the 

 course of the old river. On riding along the top of this barrier it 

 was found that the river had never surmounted it, the only signs of 

 flowing water being small gullies formed by the melted snow in 

 spring running westward into a horseshoe-shaped lake about 1 mile 

 across, and quite 200 feet below the ridge. This little lake is an 

 ideal example of glacial ponding, and represents all that is left 

 of the Rio Salado ; that it is still diminishing was shown by the 

 series of perfectly defined beach-marks above it, more than ten of 

 which were counted. Mowing into it from the west were one or 

 two torrents, but I could not see any trace of an exit ; maybe that 

 surface-evaporation and filtration through the glacial debris account 

 for this. Interesting as this case of ponding is, it is nevertheless 

 evident that it cannot be responsible for the drying-up of the Rio 

 Salado, for had not other causes been at work, such as those 

 suggested above, the river would have filled up the cavity, sur- 

 mounted the barrier, and flowed on. That other agencies have been 

 at work may be also inferred from the fact that south of this lake 

 is another canadon, joining the Canadon Salado below the barrier, 

 and of considerable size, which is quite dry, except perhaps when 

 the snow is melting in spring. 



Senor Ameghino has expressed his opinion that the valleys of 

 the pampas are fault-valleys, and it would appear that Dr. Moreno 

 .is also inclined to this view. Valuable as are the works of these 

 authors, it is believed that, as yet, no convincing evidence on 



