MORSE. — SPECTRA OF WEHNELT. 533 



and his measurements have been used for comparison, the agreement 

 being as close as could be expected from the small dispersion and poor 

 definition of the plates measured. 



These bands have been for many years attributed to the oxide, but 

 there exists a considerable amount of evidence contradictory to this 

 view. Arons * found this same banded spectrum in the arc between 

 aluminium points in an atmosphere of nitrogen or hydrogen, and con- 

 cluded that the banded spectrum corresponds to the metal and not to the 

 oxide. Hemsalech f agrees with Arons, and he has shown that by in- 

 troducing inductance the line spectrum of the spark between aluminium 

 terminals in nitrogen changes into this same banded spectrum. Berndt $ 

 concludes from his experiments that the presence of oxygen is necessary 

 for the production of the bands. Lockyer also attributes the bands to 

 the oxide, Wiillner to the metal, Kayser to the oxide, etc.§ Simple ex- 

 periments of the author in hydrogen in closed tubes have shown that 

 the band spectrum, which is present in considerable strength when the 

 discharge is first sent through the tube, decreases rapidly and disappears 

 in a short time, to appear again after the tube has been allowed to 

 recover for a time. The simplest apparent explanation is that the oxide 

 coating of the aluminium terminals reacts with the hydrogen under the 

 influence of the spark with the formation of water vapor, probably until 

 equilibrium between metallic aluminium, aluminium oxide, hydrogen 

 and water vapor is reached. Ou standing the oxide is re-formed and 

 the equilibrium at the lower temperature re-established. 



The tendency already spoken of under barium, toward a composite of 

 arc and spark spectra, is very evident in the case of aluminium. Not 

 only are the bands and lines of the arc spectrum present, but also several 

 lines which do not belong to it. || Most of the lines of the spark spectrum 

 are present in the Wehnelt, some of them with changed intensities. 

 The general appearance is as though a rather weak spark spectrum had 

 been superimposed over a stronger arc spectrum. 



It is the intention of the author to return to this and other band spectra 

 of the Wehnelt in a later paper. The intensity of the light from an 

 aluminium point in hydrochloric acid is sufficient to permit of a photo- 

 graph with a grating of higher dispersion without an excessively long 

 exposure. The spectrum being free from the overlapping carbon bands 



* Drude'B Ann., 1, 700 (1000). t Ibid., 2, 331 (1900). 



} Ibid 4, 788 (1901). § Vid. references, Hemsalech, 1. c. 



II Vid. Kayser and Runge,W. A., 48, 126. 



