ON THE SPECTRUM OF HYDROGEN. 
series of plates taken throughout the spectrum, the series lines of both elements and 
also most of the secondary Hydrogen spectrum, being recorded. The wave-lengths of 
the first six members of the Balmer series, and of the Parhelium fines included in the 
range required, were measured on these plates, together with those of about 100 secondary 
Hydrogen lines. 
The deviations of the series line determinations from the values given by Curtis 
(‘ Hoy. Soc. Proc.,’ A, 46, p. 147, 1920) for Hydrogen and by Merrill (‘ Astrophys. 
Journ.,’ 46, p. 357, 1917) for Helium enabled the shifts in the comparison spectrum 
to be eliminated, and the corrected values obtained for the selected secondary Hydrogen 
lines were then used as standards in measuring the remaining lines of the spectrum 
from plates taken with pure Hydrogen tubes. The plates were measured on a Hilger 
travelling micrometer, with a screw-pitch of 1 nun. and reading by a vernier on the 
drum to 0-001 mm. Three plades were measured for each region, each plate, following 
the usual procedure, being measured in both directions to eliminate personal errors in 
setting. Each series of readings so obtained was repeated before altering the position 
of the plate on the stage of the micrometer. The two values rarely differed by more 
than 0-003 mm., and their mean was adopted in subsequent reduction. In all, 
twelve settings were made on each line, except in the case of some of the faintest lines 
which were not visible on all the plates ; these exceptions have been noted in the 
tables given. 
The reduction of micrometer readings to wave-lengths was simplified by the fact that 
the dispersion was almost exactly 10 A per millimetre. In the region A6560-/14860 
the readings were reduced to approximate wave-lengths by the addition of a constant, 
and the final adjustment was made from an error curve drawn either b}^ means of the 
iron standards or the Hydrogen standards prepared from them. In the blue and ultra¬ 
violet regions a preliminary linear correction was applied so as to reduce the slope of 
the error curve. The arithmetic mean of the six values calculated in this way was 
adopted as the final value. The wave-lengths and wave-numbers in vacuo have also 
been tabulated, the corrections for this purpose being taken from the tables of Mergers 
and Peters (‘ Bureau of Standards Publications,’ p. 698, 1918). 
For all but a few of the weakest lines and a few diffuse lines the probable error of 
the mean wave-length adopted was less than 0-02 A, and for most lines it was consider¬ 
ably smaller than this. It is believed that the values given can be relied on to two 
hundredths of an Angstrom unit. The weakest lines were much more sharply defined 
in the visible regions than in the ultra-violet, where they were superposed on the con¬ 
tinuous background, and it is possible that this may slightly affect the accuracy of some 
of these wave-lengths. Our measurements agree in general very closely with those 
of Watson ( loc . tit.), when the latter are transposed into International Units, but 
Porlezza’s (loc. cit.) measurements differ from ours, in extreme cases by almost an 
Angstrom unit. 
Several hundred new lines have been recorded, these lines occurring for the most 
