Separation of the Least Volatile Gases of Atmospheric Air, &c. 395 



the discharge be continued without a jar, the resistance in the krypton 

 tubes increases rather rapidly, the tube becomes much less luminous 

 and finally refuses to pass the spark. With an oscillatory discharge 

 the passage of the spark and the brightness of the rays are much more 

 persistent. This seems to point to some action at the electrodes, which 

 is more marked in the case of krypton than in that of xenon. 



The wave-lengths of the xenon and krypton rays in the tables below 

 were determined, in the visible part of the spectrum, with a spectro- 

 scope having three white flint-glass prisms of 60° each, by reference 

 to the spark spectrum of iron, except in the cases of the extreme red 

 ray of krypton, which was referred to the flame spectrum of potassium, 

 and its fainter neighbour, which we saw but did not measure. The in 

 digo, violet, and ultra-violet rays were measured in photographs, taken 

 with quartz lenses and two calcite prisms of 60° each. The spectrum of 

 the iron spark was photographed at the same time as that of the tube, 

 the former being admitted through one-half of the slit, and the latter 

 through the other half. 



The xenon spectrum is characterised by a group of four conspicuous 

 orange rays of about equal intensities, a group of very bright green 

 rays of which two are especially conspicuous, and several very bright 

 blue rays. The only list of xenon rays we have seen is that published 

 by Erdmann, with which our list does not present any close agreement 

 except as to the strongest green lines. The number of xenon rays we 

 have observed is very considerable, and some of them lie very near to 

 rays of the second spectrum of hydrogen, but inasmuch as these rays 

 are more conspicuous with a jar in circuit than without, which is not 

 the character of the second spectrum of hydrogen, and, moreover, 

 many of the brightest of the hydrogen rays are absent from the 

 spectrum of the tubes, we conclude that these rays are not due to 

 hydrogen. Certain rays, which we have tabulated separately, have 

 been as yet observed in only one tube : they include a very strong 

 ultra-violet ray of unknown origin, and either due to some substance 

 other than xenon, or to some condition of the tube which has not 

 been repeated in the other tubes. 



Our krypton rays agree much more closely with Eunge's list, but 

 outnumber his very considerably, as might be expected when prisms 

 were used instead of a grating. Prisms, of course, cannot compete 

 with gratings in the accuracy of wave-length determinations. We 

 think that the krypton used by Eunge must have contained some 

 xenon, and that the rays for which he gives the wave-lengths 5419*38, 

 5292 "37, and 4844*58 were really due to xenon, as they are three of 

 the strongest rays emitted by our xenon tubes, and are weak in, and 

 in some cases absent from, the spectra of our krypton tubes. 



Our thanks are due to Mr. E. Lennox, to whose skill in manipula 

 tion we are much indebted. 



