64 Transactions of the South African FhilosopMcal Society. 



below the brass rims ; for in spite of close attention the water level 

 sometimes did sink below the rims. 



In both (A) and (B) the depth of water evaporated occupies the 

 same order relatively to the colours, albeit the ratios are not quite 

 identical. The temperatures of (A) do not greatly differ, inter sc, 

 and seem to indicate that at XVII. they are falling to a nocturnal 

 equality with that of the tank. In (B) we see that the greatest 

 difference at XIV. is between the green and purple, and is exactly 2°. 

 It is not likely that the mean difference would be much greater than 

 this at any hour. We get, then, for a maximum increase of 2° of 

 temperature an increase of evaporation in ten weeks of about half an 

 inch, or, say, 2^ to 3 inches per annum. This is in fair accordance 

 with the previous result, considering the circumstances ; and it is 

 also not antagonistic to Latham's, in which, however, it happened 

 that the gauge which was coolest by day was also warmest by night. 

 On the other hand the Camden Square results work out one to a 

 rate of increase of nearly 13 inches per annum for an increase of 2"^ 

 of temperature, the other to a rate of increase of nearly 24 inches for 

 an increase of 5°. 



A further test was attempted in w^hich the gauges should stand in 

 air instead of water, and the temperatures should be taken at XI. 

 instead of later in the day. The gauges were placed on a board 

 nearly 5 feet above the ground, and their sides were sheltered from 

 the sun's rays. The glass plates were fixed a little higher- — about 

 two inches — above the gauge-rims for the sake of variety. These were 

 kept in the same place, the gauges being permuted this time. The 

 experiment lasted from July 14 to September 21, 1901 — seventy 

 days. There were earlier observations beginning early in May, but as 

 a slight leak was developed in one of the gauges, it was thought best 

 to begin again after making the necessary repairs. The results are: — 



Colour. 



Mean Temp, 

 at XI. 



Total Evap. 



Purple 







62-8 

 61-9 

 60-1 

 61-8 

 630 

 67-2 



Inches. 

 10-08 



Blue 



Green 



9-78 

 9-32 



Yellow 



9-73 



Red 



9-83 



Air 









The net result here, again, is an increase of evaporation of about 

 2?,- inches per annum for an excess of temperature of about 2"^. 



