PROF. A. FOWLER ON SERIES LINES IN SPARK SPECTRA. 
231 
series thus resembling the Diffuse and Sharp series which occur in the spectra of 
other elements. Two such series can only be included in a single formula in the 
special case when the fractional terms associated with the parameter m differ by 
exactly 0'5, the Rydberg constant N being then replaced by 4N, if m takes successive 
integral values. In the general case, an objection to this combination would be the 
smaller intensities of the lines of the Sharp as compared with those of the Diffuse 
series. The intensities of the “ 4481 ” lines, however, are graded as in an ordinary 
series, and the chief objection to uniting them in a single formula in the first 
discussion was the undesirability of introducing a new type of series formula for 
what seemed to be a special case, so long as the older formulae could be employed. 
This objection is no longer valid, because, as will appear later, several other series of 
the same type, occurring in associated groups, have since been recognised. 
From the formulae previously given for the two divisions of the “ 4481 ” lines, it was 
evident that all the lines could be combined, at least very nearly, in one equation if 
desired. There were, however, small deviations, apparently systematic, which 
* 
suggested that the union was not quite exact, but these have been removed by the 
new measurements, which became necessary when it appeared that the lines were 
doublets. It can no longer be doubted that the lines form a single series, differing 
from the more familiar series in that the lines occur twice as frequently; i.e,, if the 
ordinary formulae involving the Rydberg constant N be employed, we find lines 
corresponding not only to the integral values m, but also to m + \. It will, 
however, be more convenient to employ formulae in which m takes successive 
integral values only, N being then replaced by 4N or by a number of that order of 
magnitude. 
As the positions of the lines have been determined with considerable precision, and 
the lines are fairly numerous, it seemed desirable to compare the accuracy of a variety 
of formulae, some withN assumed the same as for hydrogen, and others with this term 
calculated from the lines themselves. The fact that not all the lines have been 
resolved, and the slight uncertainty as to the absolute positions of the last two or 
three lines of the series, renders the comparison to some extent wanting in finality, but 
the results may nevertheless be of interest. 
For comparison of observed and calculated wave-numbers, the unresolved lines have 
been regarded as consisting of two components separated by 0‘99, one having a wave- 
number greater by 0'33 than that of the unresolved line, and the other 0‘66 less, since 
the more refrangible component is about double the intensity of the less refrangible. 
The wave-numbers adopted in the calculations for the less refrangible components, 
adjusted in this manner, are given in the first column of Table II. The wave- 
numbers are on the International scale, and have been corrected to vacuum. On this 
scale, the value of the Rydberg constant, according to an investigation by 
W. E. Curtis* is 109,679'3. 
* Not yet published. 
