530 Proceedings of Royal Society of Edinburgh. [sess. 
of the actual intensities of all the components implicated in the 
formation of the band renders it impossible to select from the 
infinite number of possible cases the one which corresponds to the 
actual conditions. But the fact I want to point out here will be 
shown under all circumstances. It becomes at once apparent if 
we compare the bright band of fig. 7 with that of fig. 6. 
While the continuous spectrum was present, the band appeared 
shifted towards the red (fig, 6) ; after the continuous spectrum 
had vanished, the band appears in approximately normal 
position, but the maximum of light lies on the violet side (fig. 7). 
y asr vet S 
It is readily noticed that the excess of brightness on the more 
refrangible side is due to the expanding atmosphere between star 
and observer. Since the density of this atmosphere must be 
supposed to diminish in the course of time, the same quantity of 
gas occupying more and more extended spaces on its outward 
journey, and since at the same time its temperature will be 
reduced by expansion, its contribution to the light of the bands 
will gradually lessen, and we may finally imagine a state in which 
the light of the star is mainly due to the gaseous radiations of the 
ring, whose temperature may be maintained for a more consider- 
able time by the doubtless frequent collisions between its in- 
dividual meteoric members. At this stage the intensity-curve of 
