ON THE SPECTRUM OF HYDROGEN. 
373 
which have made it possible to establish the conclusion that the secondary spectrum is 
to be referred to the Hydrogen molecule. 
(2) Experimental. 
As the source of the spectrum vacuum tubes of the H type have been used, and have 
been designed on lines indicated by the investigations of Wood ( loc. cit.) so as to give the 
secondary spectrum as strongly as possible. The usual capillary tube of a few inches 
in length was replaced by tubes of from 20 to 50 cm. in length and of about 5 to 8 mm. 
internal diameter. The electrodes consisted of spirals of aluminium ribbon, and the 
tubes were provided with palladium tubes, which were sealed through the intermediary 
of short platinum tubes into side tubes in the usual manner. Pure Hydrogen could be 
admitted by heating these palladium tubes in a flame or in a current of Hydrogen, or 
alternatively the Hydrogen in the tubes could be removed by heating the palladium in 
an atmosphere from which Hydrogen was absent. In some cases the tubes were cleaned 
before exhaustion by washing them out with a very dilute solution of hydrofluoric acid, 
followed by distilled water ; this procedure was found to be very effective. The tubes 
w T ere exhausted by means of an oil pump, and the evacuation was completed by means 
of a bulb containing charcoal, which was cooled with liquid air. For some of the tubes 
a Gaede mercury pump was used and in all cases the tubes were washed out during the 
process of exhaustion by the frequent admission of Hydrogen through the palladium 
tubes. In the case of tubes containing Helium, this gas was prepared by heating 
powdered Thorianite in a fused silica tube, and was purified before entering the vacuum 
tubes by passage through a U-tube containing charcoal cooled with liquid air. 
The tubes were excited by the current from a large induction coil provided with a 
mercury jet interrupter, and in some experiments a 15,000 volt £ kilowatt step-up 
transformer was used. It should be mentioned that although the utmost care was taken 
to remove the carbon compounds and other impurities with which vacuum tubes are 
liable to be contaminated, before the tubes were sealed off, the highest degree of purity, 
as shown by the intensity of the secondary spectrum relative to that of the Balmer 
series, was never attained until a discharge had been passed for several hours and the 
aluminium mirrors, which were deposited on the tubes around the electrodes, had 
removed the last traces of impurities which had been present in such small quantities 
when the tubes were sealed off that they could not be detected by any characteristic 
bands or lines in the spectrum. 
(3) Wave-length Measurements. 
The earliest tables of wave-length of the secondary spectrum are due to Hasselberg 
(loc. cit.), and although his measurements were made visually, and are not accurate 
enough for modern requirements, they are more complete than later photographic 
