274 PROFESSOR L. BECKER ON 
Tur SPECTRUM FROM 1901 Aveust 1 To 1902 NovEMBER. 
10. The Mean Spectrum.—tThe results derived from the photo-plates Nos. 8 to 21,. 
1901 August 1 to 1902 January, agree closely with each other, and it is unnecessary 
that the results be given separately. Those derived from the later plates, Nos. 22 to: 
27, are considerably less accurate, owing to the faintness of the spectrum, but they suffice 
to show that the bands did not change in position. Their results are also not printed 
separately. The changes which the spectrum underwent belong to the intensity, and 
they appear for the whole period in Table XVII. 
I combined the wave-lengths and the estimates of intensity to mean values, which 
are given in the first columns of Table XII. For most bands seven to eight plates: 
contributed to the mean, and for the band near H, thirteen plates, all belonging to the 
period August to November. The two bands at wave-lengths 386 and 397 were outside 
the range of the plate until the beginning of October, and they rest on the results of 
the plates Nos. 18, 19, 21. In Table XII. the intensities of the bands, therefore, do not, 
belong to the same epoch. ‘The average error of a tabulated wave-length is 0°3 t.m. 
With reference to the faint bands, whose intensities do not exceed 1°5, and which 
were difficult to measure, most of the detail had to be discarded, because it was seen only 
on one plate. The neglected measurements are about five per cent. of the total number. 
The wave-lengths of these faint bands may be several tenth-metres wrong. 
11. The Common Structure of the Bands.—Of the detached bands, the first two: 
have the most pronounced intensity curve (see Plate III.). I shall show (see § 12) that 
their structure is similar, and further, that the similarity extends to all the other bands. 
The wave-lengths of the lines to which the bands belong being unknown, I introduce: 
Xm, the mean of the wave-lengths of the three minima, and determine a,, from 
(12) ey 2 oa 
Am 18 given in Table XIII. for the first two bands, and two other prominent bands, whose — 
zeros X,, equal 4364 and 4726. ‘The latter merge into fainter bands, which overlap their 
more refrangible ends; the intensities at these places are bracketed in the table, and I 
do not take them into account here. The adopted values of Table XIV. are the means of 
the figures contained in Table XIII., with the exception of a few which were corrected so 
as to represent other bands better. I have also calculated a,, for all the other bands. 
except one, and drawn on Plate III. their intensity curves with a,, as abscissa. 
12. The Calculated Spectrwm.—I decide whether all the brighter bands are of the 
same type by calculating the different points of the bands from «,, of Table XIV. by means 
of formula (12). About six well-defined points contributed to the final X,,, and its 
calculated error is on an average 0°2 t.m. ‘The degrees of intensity have been obtained 
from Tables VIII. and XIV. I have given the calculated bands in Table XII. Including 
all points, I find that the calculated wave-lengths differ from the observed wave-lengths 
on an average 0°5 t.m., as compared with a calculated average error of 0°3 t.m. for A, of 
