182 Proceedings of Royal Society of Edinburgh. [sess. 
perature in the surface layer above 20 metres depth. This factor 
can only be convection, or, let us say, diffusion of liquid. As 
already shown, this convection will set in as the sun sinks and the 
day cools towards night, and will continue till early morning. No 
doubt also surface waves and ripples due to wind will aid this con- 
vection ; nor can we leave out of account the vertical migration of 
fish and other denizens of the deep. Convective movements may 
also occur during the day in bodies of salt water, the surface layer 
of which, in virtue of evaporation and consequent increase of 
salinity, may become denser than the slightly cooler water immedi- 
ately below it. This last-named factor we should not expect to 
find in the case of fresh-water lakes. That the main causes are, 
however, the same in fresh-water lakes as in salt-water seas is 
proved by the general resemblance in the law of variation of 
temperature with depth in the two types of cases. From the data 
furnished by Professor Richter in the memoir already referred to, 
and from similar data supplied by W. F. Ganong, who studied the 
vertical distribution of temperature in certain American lakes, we 
notice, however, one striking difference between the fresh-water 
lakes and the Mediterranean Sea. In the Mediterranean Sea the 
most rapid vertical variation of temj)erature occurs at a depth of 
30 metres ; in the fresh-water lakes, on the other hand, the corre- 
sponding maximum gradient occurs at much less depths — namely, 
from 6 to 12 metres. The reason for this difference may probably 
be found in the following considerations. In the first place, the 
somewhat higher temperature of the Mediterranean Sea will no 
doubt mean a greater depth of the layer of quickest variation ; but 
this can hardly explain the magnitude of the difference. It must 
be remembered, however, that in the case of the fresh-water lakes 
the vertical distribution of temperature experiences a complete 
change during the winter months when the mass of water is at or 
below the temperature of maximum density. Hence the summer 
distribution of temperature, which resembles in type the distribu- 
tion throughout the whole year in the waters of the Mediterranean, 
has just time to establish itself before the autumn and winter 
conditions set in again, and finally overturn the whole type of 
distribution. On the other hand, in the Mediterranean the waters 
are never cooled sufficiently so as to come within sight of the 
