LYMAN. — THE SPECTRUM OF HYDROGEN. 137 
tube is pumped and repeatedly washed with hydrogen, certain characteristic bands 
are bound to make their appearance to a greater or less degree. The nature of these 
bands is unfortunately only too clearly seen in spectrum II of Plate VI 11. If. liowev r, 
the tube communicates directly with the receiver and is filled with hydrogen along 
with it, these bands maybe totally absent. Schumann has observed their presence 
and ascribes them to carbon monoxide. On this point the author cannot yet be sure ; 
certain it is, however, that they occur strongly in the spectrum of the air. (Compare 
spectra I and II, plate VIII.) 
The means used to produce the hydrogen for the discharge tube have been varied. 
Zinc and hydrochloric acid, and electrolytic action in both dilute sulphuric acid and 
on barium hydrate solution have served as sources for the gas. Various shapes of 
discharge tube have been tried, both closed and communicating directly with the re- 
ceiver. Aluminum, copper, and iron have been used as the material of the electrodes 
which in turn have been of various dimensions. The comparison of the plat - taken 
under the above conditions serves as an excellent test of the true source of the radia- 
tions supposed to be due to hydrogen. 
In addition, the spectra obtained by exciting the discharge tube when filled with 
air at pressures between 2 and .5 mm. have been compared with the spectra obtained 
when the same tube was filled with hydrogen. The lines found to be common to the 
two spectra have been eliminated as due to the air itself or to some impurity. Such 
a process may result in the loss of a few true hydrogen lines but what remain can be 
safely attributed to that gas. Finally, this matter has been checked by a rtttdy of the 
behavior of suspected lines as the purity of the hydrogen in the discharge tube is in- 
creased. It must be remembered that the elimination of lines due to impurities by 
comparison of the air and hydrogen spectra can only be applied to those rad 
for which fluorite 
table of wave-lengths given at the end of this paper 
The results are to be found 
The general appearance of the spectrum may be described as follows : Between 
X2000 and X 1675 the author can find no trace of radiation due to hydrogen, but he 
is not prepared to assert that a faint continuous spectrum may not exist. From 
X1675, however, the spectrum consists of a multitude of very fine lines with a maxi- 
mum of intensity near X 1600. Near X 1300 something very like an absorption band 
occurs, due, perhaps, to some slight trace of impurity in the gas, but always present no 
matter under what conditions the gas may be produced or examined. Lines are visibl 
in this band but they are very feeble. The lines beyond the region limited by the 
absorption of fluorite are some of them as strong as any in the spectrum. The low. t 
