390 Transactions of the Boyal Society of South Africa. 
of dew-point temperature. Here it is to be observed that the dew-point 
numbers would be antecedently expected to exhibit a sort of illusory 
influence on the temperature differences of our two thermometers. For 
take any grade of relative humidity, say the one comprising 50 per cent, 
to 54 per cent. Granting that the temperature differences fall as the 
relative humidity rises, there would be a small fall of temperature dif- 
ference inside the limits of the grade — the temperature difference being 
less for 54 per cent, than it is for 50 per cent, of relative humidity. But 
there would also be in general, at the same time, a rise of dew-point 
corresponding to the rise of relative humidity. The mean results at the 
foot of the Table show this illusory influence quite clearly. Yet even here, 
as we see from Table 9, which gives the corresponding velocities of the 
wind, a great part of the effect shown, such as it is, is due to the wind 
movement. After allowing for the gradual illusory fall of temperature 
difference running parallel with the rise of dew-point in each grade of 
relative humidity, it is evident that the irregularities in the mean values 
of Table 8 agree fairly well with the irregularities in the mean wind 
velocities ; the maximum points in one Table opposing the minimum 
points of the other. 
To summarise the results it appears that after allowance has been 
made for the state of the sky and the movement of the air, the only factor 
of real importance determining the magnitude of the radiation-temperature 
gradient is the relative humidity. The absolute humidity, as such, is 
unimportant. The non-recognition of the distinction between the two 
elements of humidity (granting, of course, the justice of my argument) 
has vitiated many of the so-called classical experiments in the domain of 
radiant energy : the experiments out of which the theory of the wet bulb 
thermometer has been manufactured, for instance. Or take the seemingly 
simple case — more simple at any rate than the meteorological problem — 
of the cooling of a hot ball in a closed room filled with atmospheric air. 
In dealing with this case the "laws" commonly cited are Newton's and 
(indirectly) Stefan's. But it must be evident that the relative humidity 
in the vicinity of the ball must be considerably greater at the beginning 
than it is at the end of the experiment when the temperature of the ball 
has fallen almost to that of the room. Hence on that account alone the 
radiation of heat will be relatively greater at the higher temperature. 
Probably not one single text-book has ever hinted that no such experi- 
ment can be valid unless it be performed in absolutely dry air ; or, shall 
we say, unless due allowance be made for moisture conditions. 
Moreover, these considerations seem to help towards explaining the 
discrepancies between the results obtained by those who have experi- 
mented upon the absorption of heat by aqueous vapour. Had Tyndall 
