360 
ME. W. CEOOKES ON EEPULSION EESULTING EEOM EADIATION. 
The dots scattered about the upper portion of the figure, when measured from the 
upper horizontal line bounding the spectrum, give the millimetres traversed by the 
index ray of light when the portion of the spectrum corresponding to the dot fell on 
the black pith. Observations taken the same day are connected by lines ; the date 
of observation is written at the beginning and end of the line. The curves thus formed 
may be regarded, therefore, as a measure of the intensity of the radiation. 
Although there is a general resemblance between the curves taken on separate days, 
inasmuch as they all approach a maximum at a point somewhat below A, and then 
diminish on each side, the incline being steeper on the ultra-red than on the blue side, 
yet there is scarcely any resemblance between the heights of the curves on separate 
days. There are two causes for this. The first is that the weather was not always so 
favourable as I could wish ; but having waited so long, I took observations at first whenever 
I could get an opportunity. Thus the first four days, July 26 and 29, and August 4 and 5, 
were not quite clear, and the curves consequently have but slight gradients ; they form 
a group by themselves. Afterwards I became more critical, and would not attempt 
spectrum observations unless there was a good chance of a very bright sun. Consequently 
the curves on the next three days, August 13, 16, and 17, show an improvement in 
altitude. Having got these results, I determined not to try any more unless the weather 
were exceedingly favourable, and then to work principally at the extreme ends of the 
spectrum. The curves of August 31, September 1, 10, and 16 show, in consequence, 
much greater elevations. 
The second cause which may possibly account in some measure for the differences in 
the curves, is that, as the apparatus was kept exhausted week after week with the pump 
working almost daily, the pith may have become drier, the aqueous vapour may have 
become more perfectly removed, and the vacuum consequently more nearly absolute*. 
191. To obtain the nearest approach to the theoretical curve of intensity from my 
results, it will not be right to take the mean of all my observations. The error can only 
be on one side — that of deficiency. Therefore the lower observations must be rejected, 
and the highest results only taken. 
Bearing in mind that the spectrum formed with glass prisms is condensed at the red 
end and expanded at the blue end, it would appear from an inspection of the curves, 
that, with a normal spectrum, such as a diffraction-grating would give, the maximum 
of action should be at a little below A in the ultra-red, and that the curve should descend 
equally on each side. 
192. This appears to be the most appropriate place to discuss the question which, 
from its being almost invariably asked the first, seems to be of the greatest interest 
to most inquirers : — “ Is the effect due to heat or to light ” ? 
I cannot answer this question. The terms heat and light are not definite enough. 
The physicist has no test for light independent of heat. Light and colour are physio- 
logical accidents, due to the fact that a small portion near the middle of the spectrum 
* I could not exhaust this apparatus with charcoal (131) owing to the cement joints. 
