1904 - 5 .] On Prof. Seeligers Theory of Temporary Stars. 539 
the spectral lines, as experiments show, increase with the density, 
and perhaps also with the temperature, of the emitting gas ; and 
since both density and temperature are more considerable at A 
than at B, we must conclude that the broad and bazy absorption- 
band is gradually reduced to a narrow line, and finally fades away 
altogether. 
This peculiar shrinkage of the absorption-bands lias indeed 
been noticed during the spectral evolution of Nova Persei. But, 
curiously, the band resolved into two lines instead of one. To 
explain this duplicity we have to make a further assumption, but 
fortunately one which seems not improbable. We have indeed 
only to suppose that in this special case the dark body was a double 
star. We are quite familiar with double-star systems in which 
one of the components is invisible (stars of the Algol type). There 
is, however, no reason that might debar us from assuming double 
stars in which the surfaces of both components have cooled below 
the range of visibility. Now, in such a case it is very unlikely 
that both stars should have the same mass. But if the masses are 
different, then the gravitational effects on the cloud particles should 
also be different, and hence the heat-development at the surfaces 
and the orbital velocities of the encircling rings. In other words, 
we should then obtain an intensity-curve of the bands which is 
found by combining two curves of the shape of fig. 1 1 drawn on 
different scales. The resultant curve is shown in fig. 13, which is 
indeed typical of the first stage of development in Nova Persei. 
The following stage is characterised by fig. 14, where the broad 
hazy absorption-band has already been resolved into two compara- 
tively distinct absorption-lines. At a still further stage, when the 
density of the expanding atmosphere has become extremely small, 
the absorption has practically disappeared, and there remains only 
the radiation of the two rings, giving rise to a bright band with 
four more or less pronounced maxima, its centre lying at normal 
wave-length (fig. 15). All these conclusions are well borne out 
by the observed facts. 
I may be allowed here to quote the following remark from a 
paper by Father Sidgreaves on the spectrum of Nova Persei in 
Monthly Notices , vol. xii. p. 141, descriptive of the gradual 
changes in the dark hydrogen-bands : — “ At the beginning these 
