536 Proceedings of Royal Society of Edinburgh. [sess. 
of star and nebula. Certainly the two objects must have approached 
each other, otherwise a collision would, of course, have been 
impossible. But the rate of approach is here a question of no 
relevancy. In this point the theory may be clearly distinguished 
from the earlier attempts, in which two or more bodies were 
assumed to move through space in different directions, with speeds 
far transcending the average proper motions of celestial bodies. 
In illustrating these views on the physical processes connected 
with the phenomenon of temporary stars, I have discussed some 
of the more important facts brought to light by the observations 
of Nova Aurigae. I beg now to enter upon a brief discussion of 
the observational records of Nova Persei. Broadly speaking, the 
spectral phenomena noticed in this specially remarkable new star 
were in fair accordance with those of its predecessor of 1892. 
There are, however, some peculiar differences in the structure of 
the bands which seem to require an explanation. Most noticeable 
among these is the fact that during the time when the continuous 
spectrum was strong, the bright band, which in Nova Aurigae was 
strongly displaced towards the red (fig. 6), appeared in its normal 
position in Nova Persei. Fortunately, our theory is sufficiently 
flexible to explain this peculiar difference. We have seen before 
that the absorption-band on the violet side is caused by the rapid 
development of an expanding atmosphere at the moment of the 
collision. Now, obviously, the rate of expansion will depend on 
the temperature developed during the impact. If, therefore, on 
account of greater density of the impinging cloud, we suppose the 
catastrophe of Nova Persei to have been considerably more violent 
than that of Nova Aurigae — an assumption which is perhaps 
supported by the relative brightness of the two stars — then the 
displacement of the absorption-band would also be more consider- 
able in Nova Persei. On the other hand, if the masses of the two 
stars have been nearly the same, the two maxima of the bright 
bands which are due to gravitational effects would appear in the 
same positions. Thus, while the curve in fig. 4 would have to be 
extended in the horizontal direction (fig. 9), fig. 5 would remain 
unaltered (fig. 10). By combining the two curves in the same 
way as before we obtain the intensity-curve in fig 11. Hence, as 
the total effect of the combined radiations and absorptions, we find 
