560 CAPTAIN W. DE W. ABNEY AND MAJOR-GENERAL E. R, FESTING 
white light would therefore be extinguished together. It may be more than a 
coincidence that this ray does not differ much in wave length from that ray which, as 
stated by one of us in a paper on the Transmission of Sunlight through the Earth’s 
Atmosphere (see ‘Phil. Trans.,’ present volume) was found to be affected to the 
same degree as the integrated light of the whole spectrum, no matter what was 
the thickness of the atmosphere through which it had passed. 
It will be seen, however, from the diagram, that the other curves become straight 
lines when certain degrees of intensity, different in each case, are reached ; and if 
these straight lines are produced to cut the axis the ordinates of the rays which lie 
towards the blue end of the spectrum above 46'3 have a negative value at the zero of 
white light, whilst those which lie toward the red side of 46’3 have a positive value ; 
showing that the blue part of the spectrum is extinguished last, and the red part first, 
as we have already seen to be the case. 
It is, moreover, evident, and this has been demonstrated by experiments described 
above, that for low intensities the luminosity curve of the spectrum will vary with 
difference of intensity, but that a degree of intensity is soon reached, when all the 
curves have become straight lines, and that the distances from the origin at which they 
cut the axis are so small compared with the distance where the curves of all the rays 
become straight, that the relative luminosities of the different rays in spectra of 
ordinary intensity are practically the same. In the experiments last described, the 
D light on the screen when not reduced by the sectors was equivalent to '027 of an 
amyl lamp at one foot. This would bring it far beyond the point where its curve, and 
indeed those of all other rays, would become straight. * 
The following table shows the agreement of the results of these last measurements 
with those of the observations, from which the luminosity curve for the central part of 
the eye was constructed. The quotient of the difference of two abscissse in the straight 
part of each curve divided by the difference of the corresponding ordinates evidently 
is the tangent of the inclination to the vertical, which, as stated above, is a measure 
of the luminosity of the corresponding ray. In Column V. of the table the first five 
of these quotients are multiplied by 28'2 in order to make them easily comparable 
with the ordinates of the normal curve which are given in Column VI. In the case of 
the last three entries in the table, the beam of white light was necessarily diminished 
in intensity before it passed through the sectors, the quotients have therefore to be 
multiplied by 5‘03. 
* It must Tbe remembered that we are only dealing with light reflected from a white screen, and it 
does not follow that the lines may continue straight indefinitely when the light is of the brilliancy 
seen when looking direct at a bright spectrum, such as that of the sun, with a fairly wide slit to the 
collimator. 
