THE SPECTRUM OF NOVA PERSEI AND THE STRUCTURE OF ITS BANDS. 267 
duced the spectrum not only possible, but probable, for these reasons :—(1) Towards. 
the more refrangible side of H.,, lines coinciding within admissible limits with prominent 
_H-lines, were found without any bias, and they belong to that region of the spectrum 
which from the outset offered the least difficulty to the splitting up of the intensity 
curve into bands. (2) The intensity curve of the band belonging to 4637 or 4634, 
also detected without bias, follows the intensity curve of the March 25 plate in that 
part of the spectrum without requiring the introduction of another bright band, and 
this line is, together with H,, H,, H;, the most prominent line of the hydrogen spectrum. 
(8) If, instead of the line 4570 found under A, which does not occur in the hydrogen 
spectrum, the prominent H-line 4581 is introduced, the lines then required to represent 
the intensity curve agree in position more closely than before with those of Nova 
Aurigee, 7.e. a star whose spectrum is, as I shall show, identical with that of Nova Persei 
after August. (4) The lines of the hydrogen spectrum, which I have been forced to take 
from Witstne’s table as being within 3 t.m. of the zeros of bands actually obtained, 
include all the hydrogen lines whose intensities exceed 2 between 4341 and 4861, 
though no heed was taken of their intensities. 
9. Proof that the Error of the Additive Rule does not affect the Result of § 8.—The 
results just derived rest on the assumption that the radiations of intensity 2, . . . 7, which 
individually give on the photo-plate a blackness of degree m,... m, for the same exposure t 
produce if acting together during the same time ¢ a blackness equal to 2m. It is, how- 
ever, well known that this cannot be correct for all degrees of blackness. With the view 
of determining the error introduced by the use of the additive rule, | exposed several 
plates on a continuous spectrum, each plate containing five spectra, the exposures of 
which were proportional to 1, 2, 4, 8,16. I estimated the degrees of blackness of the 
spectra in the same way as done on the star photographs. Any two degrees of blackness 
could then be superposed, and compared with those estimated for another exposure. [ 
find that for the degrees of blackness occurring on the star photographs, the blackness of 
| the film is about proportional to the time of exposure for a constant intensity of radia- 
tion, and that the degrees of my scale are about proportional to the blackness. Since 
| this relation cannot hold good for the highest degrees of blackness, I take it to be only 
approximately true for the lowest degrees, and put 
(2) ey 1) for a radiation of constant intensity 7, 
ee er) 
and choose 
(3) F(m) — 10%" — i 
where m is the degree of blackness, ¢ the time, and @ a constant. 
SCHEINER'S Die Photographie der Grestirne contains on p. 246 a table, the results 
of experiments by Micuakg, which gives the times of exposure for intensities of radia- 
tion varying from 1 to 36, to produce the same degree of blackness on the photographic. 
film. I find this table is sufficiently well represented by 
(4) ti? =constant b=1:08, for a constant blackness m. 
TRANS. ROY. SOC. EDIN., VOL. XLI. PART II. (NO. 10). 4] 
