194 Profs. Liveing and Dewar. [May 12, 



which were reversed was not in the middle. When the jar was used 

 the pressure could not be carried beyond 10 or 12-J atmospheres, as the 

 resistance became then so great that the spark would not pass across 

 the smaU distance of about 1 millim. between the electrodes. At 

 a pressure of 1\ atmospheres, with a jar, the ultra-violet magnesium 

 triplet near L was very well reversed, and the two pairs of lines on its 

 less refrangible side (shown in fig. 3) were expanded into two diffuse 

 bands. 



In nitrogen and in carbonic oxide the general effects of increased 

 pressure on the magnesium lines (not the magnesium-hydrogen bands) 

 seemed to be much the same as in hydrogen. Without a jar the blue 

 and yellow lines were enfeebled, and at the higher pressures dis- 

 appeared, while the b group was very brilliant but not much ex- 

 panded. With the jar all the lines were expanded, and all three lines- 

 of the b group strongly reversed. The bands of the oxide (wave- 

 length 4930 — 5000) were not seen at all in hydrogen or nitrogen j 

 they were seen at first in carbonic oxide, but not after the sparking 

 had been continued for some time. 



The disappearance of certain lines at increased pressure is in har- 

 mony with the observations of Cazin ("Phil. Mag.," 1877, vol. iv, 154), 

 who noticed that the banded spectrum of nitrogen, and also the lines, 

 grew fainter as the pressure was increased, and finally disappeared. 

 When a Leyden jar is employed there is a very great increase in the 

 amount of matter volatilised by the spark from the electrodes, as is 

 shown by the very rapid blackening of the sides of the tube with the 

 deposited metal, and this increase in the amount of metallic vapour 

 may reasonably be supposed to affect the character of the discharge, 

 and conduce to the widening of the lines and the reversal of some of 

 them. Without a jar the amount of matter carried off the electrode 

 also doubtless increases with the pressure and consequent resistance, 

 and may be the cause of the weakening, as Cazin suggests, of the lines 

 of the gas in which the discharge is passed. It is to be noted, moreover, 

 that the disappearance of the hydrogen lines depends, in some degree, 

 on the nearness of the electrodes. The lines C and F which were, as 

 above stated, sometimes invisible in the spark when the electrodes 

 were near, became visible, under circumstances otherwise similar, 

 when the magnesium points had become worn away by the discharge. 



M. Ch. Fievez (" Bull, de l'Academie Royale de Belgique," 1880, 

 p. 91) has investigated the variations in the appearance of the spark 

 spectrum of magnesium under certain different conditions. Using a 

 Rutherford grating of 17,296 lines to the inch, he has noticed certain 

 lines about the b group which increase in number with the order of 

 the spectrum observed. He has also noticed dark lines in the solar 

 spectrum corresponding to these lines of magnesium when the two 

 spectra were superposed (fig. 5). We have noticed similar lines in the 



