132 LYMAN. — THE SPECTKUM OF HYDROGEN. 
reached about .45 mm. and an equal amount of hydrogen was for the third time 
admitted. It was usual to repeat this process of washing at least four times before 
a photograph was tried. 
In making the exposure the end on tube was excited by the transformer described 
in the previous paper. For the best results that pressure was chosen which gave a 
brilliant discharge in the tube without being so low as to permit the glow to spread 
from the tube into the spectroscope. The best value for the pressure under the condi- 
tions was in the neighborhood of 1.5 mm. The receiver was generally pumped to 
this pressure before the tube was excited. 
I n a plate such as that previously published, where several spectra appear upon one 
negative, it was usual to allow a fresh supply of hydrogen to enter the receiver be- 
tween each exposure. Thus after the plate-holder has been lowered by the magnetic 
device the receiver is repumped. It is just to suppose that the gas in the apparatus 
is much purer during the last exposure than during the first. The effect of this in- 
creased purity upon the nature of the spectra themselves has been noted in the former 
article, unfortunately the reproduction did not show the effect at all well, though it 
was extremely clear upon the original negative. 
The process of washing, pumping, and rewashing is of necessity a tedious one and 
generally occupied the better part of a day. Schumann has observed that the appear- 
ance of the hydrogen spectrum in its visible part was no criterion of its purity as 
observed in the region of short wave-length. It may be of interest to add, neverthe- 
less, that the discharge tube when properly washed with hydrogen showed the many 
line spectrum of that gas in a state of considerable purity. The appearance of air 
lines was always a sure warning, if a discharge tube without a window was used, that 
the spectrum on the photographic plate would be extremely feeble. 
In work of this kind it is found almost impossible to make the receiver absolutely 
air-tight. In fact some of the most successful of the early plates were obtained in the 
presence of a slight leak. Under the circumstances the magnitude of this leak 
1— s of importance. For example, the plate of the previous article was obtained 
surpnsmgly large leak of 0.2 mm. in an hour, showing that perfect tig 
g 
with 
of the apparatus was not necessary when prope 
ofZmuof ' h rr T ""^ W " t0 be exhailsfced Wow .1 mm. the leak must 
and wl las I I T^ ^ "" b "*"« the face plate secured this result 
oeen said about the purity of the gas in the receiver and the 
