During past ages, when the sea water was less salty, the circulation was presumably more rapid than now. This was because, in tropical regions, the rise of cold water is hindered by the sinking of warm surface water which is relatively dense because evaporation has removed part of the water and caused an accumulation of salt. According to Kruemmel and Mill,[103] the surface salinity of the subtropical belt of the North Atlantic commonly exceeds 3.7 per cent and sometimes reaches 3.77 per cent, whereas the underlying waters have a salinity of less than 3.5 per cent and locally as little as 3.44 per cent. The other oceans are slightly less saline than the North Atlantic at all depths, but the vertical salinity gradients along the tropics are similar. According to the Smithsonian Physical Tables, the difference in salinity between the surface water and that lying below is equivalent to a difference of .003 in density, where the density of fresh water is taken as 1.000. Since the decrease in density produced by warming water from the temperature of its greatest density (4 deg.C.) to the highest temperatures which ever prevail in the ocean (30 deg.C. or 86 deg.F.) is only .004, the more saline surface waters of the dry tropics are at most times almost as dense as the less saline but colder waters beneath the surface, which have come from higher latitudes. During days of especially great evaporation, however, the most saline portions of the surface waters in the dry tropics are denser than the underlying waters and therefore sink, and produce a temporary local stagnation in the general circulation. Such a sinking of the warm surface waters is reported by Kruemmel, who detected it by means of the rise in temperature which it produces at considerable depths. If such a hindrance to the circulation did not exist, the velocity of the deep-sea movements would be greater. If in earlier times a more rapid circulation occurred, low latitudes must have been cooled more than now by the rise of cold waters. At the same time higher latitudes were presumably warmed by a greater flow of warm water from tropical regions because less of the surface heat sank in low latitudes. Such conditions would tend to lessen the climatic contrast between the different latitudes. Hence, in so far as the rate of deep-sea circulation depends upon salinity, the slowly increasing amount of salt in the oceans must have tended to increase the contrasts between low and high latitudes. Thus for several reasons, the increase of salinity during geologic history seems to deserve a place among the minor agencies which help to explain the apparent tendency toward a secular progression of climate in the direction of greater contrasts between tropical and subpolar latitudes.