1903 - 4 .] Prof. C. G. Knott on Ocean Temperatures, etc. 179 
culated in the previous paper. On page 299 in that paper a table 
is given from which we may estimate the amount of solar energy 
available in one day in the middle of September for localities in 
the latitude of 33° N. Taking the average declination of the sun 
during September at about 3°, we find for the solar energy supplied 
in one day the value 6 x 117 = 700. According to the present 
calculation we conclude that about two-thirds of the solar energy 
incident on the surface of the Mediterranean Sea heats the surface 
waters through a depth of nearly 20 metres. This, perhaps, is 
not an unreasonable result, and is an important correction upon 
the earlier result, as showing that the Austrian observations are 
from this point of view in sufficient accordance with Langley’s 
valuable investigations into the value of the solar constant. 
Dr Buchan has drawn attention to the importance of the obser- 
vations in relation to the manner in which the ocean waters (first) 
gain their heat in the day, and (second) lose it again at night. But 
here again their value would have been greatly increased if the 
observers had had this particular problem present to their mind when 
the observations were being made. Had the Pola, on one particu- 
larly quiet sunny day, in the centre of the Levant, far from land, 
made throughout a complete day of twenty-four hours a succession 
of complete sets of temperature readings at the various depths, 
at intervals, say, of two or three hours, a great deal of valuable 
information bearing on this question would have been obtained. 
The conditions of the survey undertaken quite precluded this. 
Fortunately, however, observations of the temperature of the 
surface waters at midnight were frequently, though not regularly, 
taken. By comparing these with the preceding afternoon tem- 
peratures and the succeeding morning temperatures, and taking 
into consideration the air temperatures at the same times, we gain 
distinct evidence of convection in the surface layers. The data 
are given in Table D, sixteen different cases in all. In only two 
cases was the early morning temperature lower than the immedi- 
ately preceding midnight temperature ; in two cases it was the 
same ; in all other cases it was higher, sometimes markedly so. 
In thirteen out of the sixteen cases the air temperature was lower 
than that of the water at early morning ; and in eleven of these it 
was lower even than the contiguous midnight temperature. We 
