538 Proceedings of Royal Society of Edinburgh. [sess. 
a more pronounced form, because it is inconceivable that any 
photospheric light can have been abnormally refracted on wave- 
lengths shorter than the normal. The whole of the bright band 
should have developed on the less refrangible side. This is clearly 
shown by the theoretical intensity-curve in Dr Ebert’s paper. 
I have mentioned before that the density of the expanding 
atmosphere, which may have been considerable at the moment of 
impact, must be assumed to decrease in course of time,, and I have 
pointed out the effect which this must have on the appearance of the 
absorption-bands. These bands, being very broad and hazy at first, 
12 
will gradually shrink into narrow lines. Suppose fig. 12 to represent 
a central section through the star and its atmosphere, A B indicat- 
ing the line of sight. It is clear that atmospheric particles, at 
one time distributed along the arc A A, will, by radial expansion, 
in course of time be distributed over the greater arc B' B'. Now 
all the particles within A A have contributed to the absorption- 
band at the first moment, but of these only those lying within the 
arc B B will absorb the photospheric light at the second moment. 
Hence the total number of gaseous molecules passed through by 
photospheric rays in the direction of the line of sight will be less 
at the second moment. Now, since the breadth and haziness of 
