348 PROFESSOR TYNDALL ON THE ACTION OF FREE MOLECULES ON 
He there describes the fiEthrioscope, an instrument used to measure these impressions. 
“ The sensibility/’ he says, “ of the instrument is very striking, for the liquor inces¬ 
santly falls and rises in the stem with every passing cloud. Under a fine blue sky, it 
will sometimes indicate a cold of 50 millesimal degrees ; yet on the other days, when 
the air seems equally bright, the effect is only 30°. The causes of these variations are 
not quite ascertained.” He might have said, not at all ascertained. The causes, 
I submit, are the variations of the quantity of transparent aqueous vapour in the 
atmosphere, which, without affecting the visual brightness of the air, is competent to 
arrest radiation from the earth. Precisely of the same character is the difficulty 
noticed by Lieutenant Hennessey in his paper on “ Actinometrical Observations in 
India.” Like Leslie he speaks of variations, the causes of which are not ascertained. 
“ Again,” he says, “ there is a change of intensity from day to day apparently not due 
to alterations in the sun’s declination, so that the average daily curve (about noon) 
is higher or lower without any visible reason.” 4 ' 8. The reason here is that applicable in 
Leslie’s case; namely, the variations of the invisible atmospheric vapour. 
In 1866 my friend Professor Soret of Geneva favoured me with a letter from which 
the following is an extract :—“ In two comparative experiments, made within a few 
days at Geneva and Bologna, the most powerful radiation was obtained at Geneva, 
although at Bologna the heavens were visibly 'purer. The result appears to me to 
support your views regarding the aqueous vapour of the air; for the tension of aqueous 
vapour at Bologna was 10’7 while at Geneva it was only 6’33.” 
Cautiously abstaining from drawing a general conclusion from a single fact, M. Soret, 
in 1868, made some further experiments on solar radiation. The intensity was measured 
by first allowing the rays to fall directly on the thermometer of the actinometer, and 
then by allowing them, prior to meeting the thermometer, to pass through 5 centi¬ 
meters of water. Calling the first temperature T and the second t the ratio p will be 
obviously greatest when the absorption by the water is least. And as we know that 
water exerts its chief absorbent power on the ultra-red rays of the spectrum, the 
variations in this ratio observed at different atmospheric thicknesses will enable us to 
infer the nature of the heat “ arrested ” by the atmosphere. 
M. Soret found the ratio to be greater in the middle of the day than when the sun 
is near the horizon. At 12.30, for example, on the 9th of March the ratio was 0'594, 
while at 5.10, on the same day, it was only 0*409. A smaller fraction of the total heat 
was absorbed by the water near mid-day than near sunset. At mid-day therefore the 
solar heat was more thoroughly sifted of its calorific rays, and more transmissible by 
water, than it was when the atmospheric thickness was much greater. It would seem 
difficult to reconcile this result with the notion that aqueous vapour is the absorbing 
constituent of the atmosphere. 
A year subsequently MM. Desains and Branley found, both at Paris and at 
* Proc. Roy. Soc., vol. xix., p. 228. 
