454 
CAPTAIN ABNEY AND MAJOR-GENERAL FESTING 
nately so as to make the illuminated shadows on one side too light and too dark. 
With rapid alternations the point of what we may call equilibrium is found, and the 
current read off by suitable means. The same method of procedure may be carried 
out to compare any light with an incandescence light, and the results show great 
concordance one with the other. 
Addendum. 
(Added June 9, 1886.) 
§ XXV. Elimination of the Variations in the Comparison-Candle. 
Since our last communication was made on this subject, we have introduced some 
improvements into our mode of using the instrument described therein, which we 
think will meet any objections which may be made to our use of a candle or other 
similar source of light in the comparison-light. Mr. Vernon Harcourt, F.R.S., 
whose colour perceptions we tried, found that the brand of candle we employed burnt 
more uniformly than the standard candle, and he gave us the results of some careful 
measurements which he made. Except for the first five minutes, when the candle he 
tested evidently was not in its normal condition, and readings with which we should 
have rejected, the light varied but slightly ; and we have no doubt that the results 
we have already given are very nearly correct, particularly as we find our curves agree 
almost absolutely with those taken by our improved methods which we describe 
below. It may be remarked that fluctuations in the candle light are readily dis¬ 
tinguished in plotting the curves, since observations are taken on each side of the 
maximum; and it is readily seen that if the pair of observations are too high or too 
low the candle has varied. Their variations disappear when the readings for a curve 
are taken in three sets. To meet, however, the possible objection that may be taken 
to the candle method, we have employed an incandescence lamp giving about 10 
candles, and instead of shifting this comparison-light have kept it stationary, and 
altered the intensity of its light falling on the screen, by means of rotating discs in 
which are movable sectors to cut off more or less light. These discs are set to give 
various apertures, and the reading taken as before. We drive the discs by means of 
a small electromotor, which causes them to rotate about 40 times a second. We are 
glad to state that the results for the electric light do not vary from those originally 
obtained. 
§ XXVI. Measurement of Sunlight. 
Owing to the variability of sunlight during a set of experiments to determine the 
luminosity of its spectrum, we considered that it would be advisable to make the 
comparison-light vary in the same degree as the light which formed the spectrum. 
