428 GEOLOGY. 



ciated hemisphere toward the warmer one, and that this shifting would 

 turn a large part of the warm equatorial waters away from the cooler 

 hemisphere, intensifying the direct astronomical effect, while the warm 

 water thus carried in excess into the warmer hemisphere would intensify 

 the evaporating effects, and induce a mild and moist climate. Croll 

 urged that this shifting would be peculiarly effective in the Atlantic, 

 because of the angular form of the eastern coast of South America, 

 and the critical position of Cape St. Roque relative to the equatorial 

 currents. He held that a few degrees of southward shift of the trade- 

 wind belts would throw a large part of the equatorial current soutli of 

 Cape St. Roque, and turn it into the South Atlantic, greatly reducing 

 both the existing contribution to the Gulf Stream and its auxiliary 

 climatic effects, while, on the other hand, a northward shift, when the 

 southern hemisphere was passing through its cold period, would throw 

 nearly all the equatorial current north of St. Roque, and thus intensify 

 the ameliorating conditions of the North Atlantic, and give a mild, 

 moist interglacial epoch to the northern hemisphere. On this account 

 especially he held that glaciation preponderated about the North 

 Atlantic, and was less pronounced in other high latitudes. 



A peculiarity of the hypothesis is that (1) the glacial epochs it 

 postulates alternate between the northern and the southern hemi- 

 spheres, and (2) that they are limited in duration to an appropriate 

 fraction of the precessional period (21,000 years). This appropriate 

 fraction is probably about that which effective winter bears to the 

 whole year, for in the course of the precessional period, which may 

 be conceived as an astronomical year, the attitude of the earth would 

 pass through a stage of neutral distribution of heat between the gla- 

 cial and the deglacial stages, very similar in nature to the con- 

 ditions that produce our spring and fall. In the middle latitudes, the 

 effective winter would perhaps occupy 5000 or 6000 years; in the high 

 latitudes, one half or more of the precessional year, while in the equa- 

 torial belt, there would probably be little or no glaciating effects. 

 These peculiarities of the hypothesis afford a means of testing it. If 

 it be true, the glacial episodes should bear evidences of equal length; 

 they should all be short, and they should be equally distant from each 

 other in the same period of eccentricity. If the -computations of the 

 periods of eccentricity published by Croll are founded on adequate 

 data (which has been questioned), there could only be a few alternations 



