264 PROFESSOR L. BECKER ON 
degrees of intensity 18, 11, 9, 7, etc., has at corresponding points the degrees, say, 5, 
4°1, 3°1, 2°2, etc. (Also see § 9.) 
The bands contained on the first seven photographs here under discussion give no 
evidence as to the second question. On the later plates there are two bands which can 
with certainty be identified as consisting each of two bands, while the structure of the 
standard band is independently determined from detached bands. From Table XIL., 
where the later observations are compiled along with the calculated bands, it will be 
seen that the sum of the calculated degrees of intensity due to radiations of the same — 
wave-leneth nearly agrees with the observed intensity. I adopt this additive rule here 
as a working hypothesis, the accuracy of which will be investigated in § 9. 
I deduce the common structure of the bands by successive approximation. Choos- 
ing first the H,-band of Table IV., I calculate the band for A, = 4922, and employing 
Table V1., reduce the degrees of intensity so that the degree of intensity of the maximum — 
of this band agrees with the observed intensity of the maximum. I then subtract 
the calculated degrees from their observed values, and find the degrees of intensity at 
the different points of the H,-band freed at its less refrangible end from the superposed 
band 4922. From the corrected H,-band I calculate the band for \>= 4265, which a 
preliminary discussion had shown to overlap the more refrangible end of the H,-band, 
and proceeding as before, I obtain the intensities at different points of the more re- 
frangible end of the H,-band. In second approximation I combine these results, and 
repeating the calculation, find the mean structure as contaimed in Table V., the values 
a, being the means, with regard to weights, of the measurements given in Table 
lu ; 
8. Resolving of the Spectrum mto Bands.—I set myself the problem to find the 
wave-lengths , of the zero of each band, and the degree of intensity of its maximum, 
which I shall call the intensity of the band, so that the superposed bands represent the 
observed intensity curve. I found this research on the following basis:—1. The 
continuous spectrum is faint and may be neglected. 2. The bands are similar to the 
band given in Table V., and determined by formula (1), A, being unknown. 3. The 
intensity curve of each band is defined by the unknown maximum intensity and the 
data contained in Tables V. and VI. 4. At places where bands are superposed, the re- 
sultant degree of intensity is the sum of the degrees of intensity which the radiations 
would singly produce on the photographic plate. : 
The last assumption is merely a convenient rule, which, though not strictly correct, 
is sufficient for our purpose, as will be proved in § 9. I may mention here the con- 
siderations which induced me to undertake a research which at first sight appears to be 
hopeless. I suppose that two bands have been identified in the spectrum, and draw 
their intensity curves as calculated from Tables V. and VI., together with the observed 
intensity curve of the spectrum. The length, in the direction of the axis of wave-lengths, 
of the area bordered by the three curves is independent of the manner in which the 
ordinates of the two bands are deducted from those of the observed intensity curve. 
