136 LYMAN. — TIIE SPECTRUM OF HYDROGEN. 
observed. At the very beginning of the investigation the author was confronted by a 
puzzling and persistent phenomenon — the absorption of the air appeared to be 
selective, not total. For a broad absorption band appeared between A 1790 and 
\15")0 and remained undisturbed even when the pressure had been reduced to 
.17 mm. At this point the air permitted the remainder of the spectrum to pass 
nearly down to the limit of transparency of the fluorite window. It was only after 
the receiver had been frequently washed with carefully dried air that the absorption 
band disappeared. The phenomenon is therefore due to some impurity — possibly 
something which comes from the brass of which the receiver is made, and which only 
persistent pumping will remove. 
It is not perfectly satisfactory to compare the values obtained by Schumann, which 
are given in terms of the absorption of a column of air at atmospheric pressure, with 
obtained by 
Tt may be of some interest to point out, however, th 
if the lengths of two equivalent air paths are to each other inversely as their 
corresponding pressures, the column of air in the receiver at .17 mm. pressure 200 
cm. long is about equivalent to a column at atmospheric pressure 0.4 mm. in 
length. Now, when the receiver was at this low pressure, light of wave-length 
trifle below X1400 was recorded on the photographic plate. It appears, therefore, 
that a column of air 0.4 mm. long will permit light of this short wave-length to 
pass in sufficient intensity to affect a photographic plate under the conditions of 
the experiment. 
The expression of the absorption of the air in anything like an absolute system is 
a very d.fflcnlt matter. The point of practical interest in this part of the research is 
the advantage of an atmosphere of hydrogen in the receiver. It is not easy to 
pparatus to a sufficient degree of transparency, but by successive wash- 
largely eliminated 
ydrogen the last traces of air can be removed and its absorpt 
•y 
PURITY OF THE SPECTRUM. 
tables w J ^? r0ge "' £r ° m Wl " Ch the wave -k«gths recorded in the following 
1 sLw Co" 16 ?'. Ve r been 0Uained Under ">™ *** of <*^ition, but they^ 
wee7 t ;:;; dera t b,e r formity of appearance - Th ° *«** ***»«» °-- 
between th se spectra obtamed from an internal capillary discharge 
o etri ;: whe r th r ube com ~ tea *^-* *£— * - 
t ;r e p : : it,r:^° tam , e ? ^ ^ ^ — *■ are mU cb ^ j: 
tube closed by a 
Contrary 
the first method yields. No .natter with what care the closed d 
