446 
CAPTAIN ABNEY AND MAJOR-GENERAL FESTING 
another direction. Taking the mean of the observations made in the four different 
directions of the lines, we arrive at the result that the candle value to A : candle 
value to G : : 75 : 52, that is that G has less perception of light than A. 
We have shown that to reduce G’s curve so as to lie on ours his ordinates had to 
be multiplied by '6, and that this corresponded to his candle value as derived from 
area of G’s candle curve T „ . ,, 
- - - - -. 11 we multiply 72 by '6 we get 48 as the relative candle value 
area ol normal curve 
of G, which differs from that obtained by the above method by 4, i.e., about 8 per cent. 
It seems probable, therefore, that the want of appreciation of red in G’s case is not 
made up for by an increased sensitiveness to the other colours. We do not wish to 
lay too much stress on the above quantitative determinations, but we think that the 
observations made undeniably show that the actual intensity of the total light of a 
candle as seen by G is inferior to that as seen by us. 
§ XIX. Curve of Luminosity of the Solar Spectrum. 
As a rule estimations of luminosity have usually been made by comparison with the 
light of the sun or sky. We, therefore, think it may be of interest to give a general 
idea of the illuminating value of the different parts of the solar spectrum as deter¬ 
mined by us, and we annex two diagrams made by independent methods of observa¬ 
tion. The first was our early plan of noting the amount of light required to enable 
us to distinguish closely-ruled parallel lines. The required amount of light was given 
by opening or closing the slit of the collimator, the width of opening being measured 
by a micrometer screw. The lines and observing telescope were fixed, the only variables 
being the width of opening of the slit and the parts of the spectrum dealt with. This 
spectrum was taken about 10 a.m. on the 6th July, 1885. The other was obtained on 
the 17th November of the same year by the method we have described in this paper. 
We do not lay any particular stress on these curves, as the first was by a method 
which, as we have said, is particularly difficult to manage, and the latter was taken at a 
time of year when sunlight is particularly apt to vary. As curves taken on subse¬ 
quent days agree witli this latter, we believe it to be fairly representative of the sunlight 
as it came to us on that day. The sky was blue, but there was a slight haze which 
would necessarily alter slightly the proportions of the different parts of the spectrum, 
particularly the violet and blue, but would not alter the characteristics of the curve 
in the red, orange, green, and green-blue. This we have been able to determine on 
several occasions, though from the date given to the present time on no day has the 
sun been visible long enough to enable us to fully complete a curve. (See Addendum.) 
We have put forward these curves to compare with that obtained by Yierordt 
(Annal. Phys. Chem., vol. 137 (1869), p. 200), which has, we find, been generally 
accepted. His method was to take a certain portion of the spectrum, and having 
extinguished the colour by white light, to calculate the amount of it employed at the 
