POSSIBLE REVERSAL OF DEEP-SEA CIRCULATION 369 



the actual humidity runs roughly parallel to it on the grand average. 

 Poleward movement of the atmosphere leads therefore to a lower 

 content of moisture; equatorward movement, to a higher. As the 

 acquisition of moisture lags behind the capacity to hold it, it is a rather 

 firm inference that precipitation exceeds evaporation in the high lati- 

 tudes, and that evaporation exceeds precipitation in the low latitudes, 

 on the grand average. The bearing of observational data is of the 

 same import. The result of these ratios of precipitation and evapora- 

 tion is a raising of the ocean surface by fresh waters in the polar 

 regions and a lowering of it in the low latitudes accompanied there 

 by concentration of saline constituents. Considered alone and 

 ideally, this should give a slight equatorward gradient and a flow of 

 fresh surface waters in that direction. These waters, however, min- 

 gle with the superficial sea waters, and involve a movement of these 

 also toward the equator. So far as these affect abysmal movement, 

 they antagonize the present circulation. 



3. In so far as evaporation exceeds precipitation in the low lati- 

 tudes, it results in an increased salinity of the superficial waters, 

 and a tendency of these to sink and flow poleward to replace the salt 

 waters carried equatorward by the fresh waters as just observed. If 

 these were the only factors, it seems clear that the deep circulation 

 would be poleward. 



4. On the other hand, the lower temperatures of the high lat- 

 itudes increase the density of the water, and tend to cause it to sink 

 and flow equatorward. But the low temperatures affect primarily 

 the superficial stratum which is freshened by the superior precipita- 

 tion of the high latitudes, and both computation and observation 

 show that cold fresher waters may float upon warmer saline waters. 

 A large part of this cold superficial water flows away in surface cur- 

 rents to lower latitudes. The actual mode by which polar agencies 

 control the deep circulation is therefore by no means so simple as it 

 might seem. 



There is ground to suspect that the formation and melting of ice 

 is an important factor. In freezing, the salt and gases of the surface 

 layer are largely forced out into the underlying layer. If the sur- 

 face layer has an average degree of salinity, the underlayer is super- 

 charged and, being^also cold, must tend to sink. On. the borders of 



