THE SPECTRUM OF NOVA PERSEI AND THE STRUCTURE OF ITS BANDS. 265 
If this length agrees with the extent of one band, which is almost constant for the 
brighter bands, the position of a band can be adj usted in this area, 7.e. the wave-length, 
),, of the zero of the band can be found. Fainter bands are shorter in length, and two 
or more fainter bands may be combined so that their total extent is that of a bright 
band. On the understanding that a band selected in the manner described may be 
replaced by two or more fainter bands whose values of A, differ little, the wave-length 
of the zero of the band is independent of the rule according to which the degrees of 
intensity are combined, and depends only on the wave-lengths A, of the zeros of the two 
adjoining bands. The error of the rule will appear only in the residuals. To a lesser 
degree, this holds also good for a space belonging to two or more bands, and the greater 
the space to be covered, the more important it will be that the intensities be correctly 
compounded. For this reason I have divided the spectrum between H, and H, into 
four parts, introducing three bands, whose maxima I make to coincide with pronounced 
maxima of the observed intensity curve. The intermediate spaces were explained by 
as few bands as possible, with the intention that each could afterwards be replaced by 
two or more fainter bands, should this improve the agreement of the spectrum with one 
which otherwise resembled it. 
A. For the discussion of the spectrum between H, and H, I employ the intensity 
curves of photo-plates Nos. 3 and 4. As before, I add 5 to the degrees of intensity of 
No. 4 to refer them to the same scale as was chosen for No. 3. The spectrum on No. 4 
differs somewhat from those on the other plates, of which No. 3 is the best representative. 
On plate No. 4 there are four prominent maxima between H, and H,, viz., 4459 to 
4480, intensity 6; 4560 to 4589, intensity 5°5 ; 4627 to 4648, intensity 8; 4682 to 
4691, intensity 8; all of them fading off several degrees towards both sides. I assume 
that they are due to bands whose A, is respectively 4470, 4574, 4637, 4687, and I 
calculate the tensity curves of these bands from the data given in Tables V. and VI. 
The calculated degrees of intensity were written out at intervals of two tenth-metres, 
and subtracted arithmetically from the observed intensities. To explain the residuals, I 
chose, in accordance with the above, as few bands as possible, and introduced further the 
condition that the same bands be selected for both photo-plates. Between 4341 and 
4470 at least three bands were required for plate 3 and two for plate 4; between 4470 
and 4574 three for plate 3 and two for plate 4; between 4574 and 4637 one each 
for plates 3 and 4; between 4637 and 4687 one for plate 3 and none for plate 4 ; 
between 4687 and 4861 two bands for both plates. By a lengthy process of trials 
in which the wave-lengths and the intensities of the bands were altered, including 
those of the above bands, I found the wave-lengths \, of the zeros of the bands and the 
degrees of the intensity of their maxima, as given in Table VII. under the heading A. 
The three intensity curves calculated from these data at intervals of two tenth-metres 
are represented on Plates I. and II. under A, together with the observed curves, which 
are dotted. The straight lines drawn at the top of the plates show the extent of each 
band and the number of superpositions at each point. 
