430 
CAPTAIN ABNEY AND MAJOR-GENERAL FESTING 
draughts, the light emitted was remarkably uniform. The mere fact that we could 
place curves taken on different days one over the other, and that they should prac¬ 
tically coincide, shows that the errors due to variations of the candle must be very 
slight. With the Siemens’s unit lamp similar results were obtained. It may be 
concluded that the figures given are reliable within a very small percentage. (See 
Addendum, p. 453.) 
§ VIII. Determination of the Illuminating Value of Different Parts of the Spectrum 
by the Two Authors. 
When experiments of this nature are undertaken by two people working together, 
it is clearly necessary that there should be a careful comparison of the effects 
on their eyes of the same quantity and quality of light. It was a matter of great 
good fortune that in our case the readings of intensity for each part of the spectrum 
were identical. As we have not so far found anyone who had a keener perception of 
the intensity of the whole range of colours than ourselves, we have ventured to call 
the curve of luminosity, shown in fig. 3, Plate 24, the “normal” curve, as it gives the 
illuminating value, as determined by us, of the spectrum of the crater of the positive 
pole. That is to say, the illuminating power of any part of the spectrum is expressed 
by the ordinate of this curve at the part in question. 
We assumed that the light from the arc itself (a part of which must be in front 
of the crater of the positive pole) would have no appreciable value. That we were 
justified in our assumption the following experiments will show. With the light we 
used the strongest part of the spectrum of the arc itself is the green band which 
commences about “ h.” If the total intensity of the light of the arc and points 
together at this part of the spectrum be taken as 100, that due to a whole section of 
the arc alone is only '42. Now the image of not more than one-half of the total 
section of the arc is thrown on the slit with that of the crater, we may therefore safely 
say that the results are affected to a not greater extent than *2 per cent., a quantity 
inappreciable compared with errors of observation. 
In the violet band, which lies about “ G ” and on the side of “ H,” the proportion 
is somewhat larger,—if the intensity of crater and arc be taken at 100, that of arc 
alone is 2‘5, and not more than half that amount would have to be taken into 
account; but at this part of the spectrum this difference is practically immeasurable. 
It should be noted that the curve which we have given is the mean of a great 
many, of which none differ 2 per cent, from it and most much less. 
To make the diagram more complete we have attached to it the curve of com¬ 
parative energy as found by the thermopile. 
