200 MESSRS. A. SCHUSTER AND G. HEMSALECH ON 



eye observations, proves that the cloudy luminosity is due to metallic vapours. The 

 presence of metallic lines in fig. 2 does not prove the presence of metallic molecules in 

 the main discharge, as it is impossible to obtain an image of the latter free from that 

 of the luminous cloud lying in front and behind it. 



Fig. 14, Plate 9, November 12, 1897. Spark between iron poles, showing the 

 luminous filaments preceding the discharge. 



Figs. 15 and 16, Plate 9, November 13, 1897. Sparks taken under similar 

 circumstances as fig. 14. In fig. 16 a current of air is blown through the spark gap. 

 The effect of this current of air is remarkable, as the time of luminosity of the air column 

 was found in subsequent experiments to be less' than 10~ 6 second, and the actual 

 displacement of the air during that time must have been quite inappreciable. The most 

 probable explanation is that the spark passes through those portions of the air which 

 have been made conducting by preliminary invisible discharges. The air put into 

 the sensitive state by these first partial discharges has time to move over a sensible 

 distance before the main spark passes. , 



It is remarkable how the metallic vapours in this photograph seem to be drawn 

 into the glass tube. 



Fig. 17, Plate 9. Photograph of a succession of sparks taken with a small jar and 

 large induction coil. This photograph illustrates the fact that, in the ordinary 

 arrangement adopted to produce metallic spectra, the air discharge is predominant, 

 while the metallic cloud is chiefly confined to the neighbourhood of the electrodes. 



5. Measurement of Molecular Velocities. 



We may now pass to the description of the results obtained when the spectrum of 

 a single spark is taken on a moving film. A preliminary trial with various metallic 

 electrodes had shown us that the sharpest results were obtained with zinc, and we 

 chose, therefore, that metal for investigation under various conditions. Fig. 18 

 (Plate 10) gives the spectrum as we obtained it on the stationary film. The lines 

 of zinc which appear on it are* : 



4Q19-1 f a ' Zinc doublet not visible in the arc spectrum. 



48107 "I A 



4722-3 >y. Zinc triplet, strong both in arc and spark. 

 4680-4 J 8. 



4058'0 c. Not mentioned as seen in the spark by previous observers, but 

 given by KAYSER and RUNGE as a strong line in the arc. 



* All wave-lengths are given on ROWLAND'S scale in air, and are taken from the most reliable available 

 determinations. To identify the lines on the photographs we have attached Greek letters to the principal 

 metallic lines shown in the figures, and these letters are also attached to the wave-lengths as given in 

 the text. 



