of Edinburgh, Session 1873 - 74 . 367 
behave like chalk, in this respect, to some of the invisible rays. In 
fact, Melloni’s experiments show that this is the case. 
2. The second difficulty is, that in employing the eye, a source 
of error is introduced by the fact that a certain intensity of radia- 
tion is necessary before a light becomes visible to the eye. This 
intensity ( i ') is evidently dependent upon the w r ave-length (A) of 
the light considered. Hence 
i'=p(A). 
The difference between i and i\ or 
/(*, *)“*>(*) 
gives a third curve, 
I = if/ (<9, A) , 
which shows the apparent intensity of any colour in terms of the 
temperature. 
In this paper I wish to pay attention to the luminous portion of 
the spectrum, with the object of determining the nature of the 
function <p (A) as much as that of / ( 6 , A). 
I shall enumerate the different experimental facts which throw 
some light on the forms of these two curves. 
1. Mossotti has shown* from the experiments of Frauenhofer 
that the curve of apparent intensities ( i.e ., the curve if/ (0, A) ) is, in 
the case of sun-light, a sinuous line, symmetrical about the mean 
wave-length. 
2. Draper f has shown that the radiation from the parts on either 
side of the mean wave-length are, with visible radiations equal. 
Thus we are led to conclude that i and i - i' are both symmetrical 
about the wave-length of mean visibility; and hence i—f or <p (A) 
is also symmetrical about that line. 
3. Dewar J has shown that several methods combine to prove that 
the temperature of the sun is about 16,000° C. 
Hence the isothermal on our diagram, corresponding to 16,000° 
C., has a maximum value of i for the mean visible wave-length. 
* Atti Scienz. Ital., 1843. 
t Pliil. Mag., 1872. 
f Proceedings of the Royal Society of Edinburgh, 1871-72. 
