406 Prof. W. Ramsay and Mr. M. W. Travers. 



with water is pure lias already been proved by the fact that Lord 

 Rayleigh found that the nitrogen obtained from it had the normal 

 density. The magnesia, resulting from the nitride, yields only a trace 

 of soluble matter to water, and that consists wholly of hydroxide and 

 carbonate. So far, then, the results have been negative. 



Recently, however, owing to the kindness of 'Dr. W. Hampson, we 

 have been furnished with about 750 cubic centimetres of liquid air, 

 and, on allowing all but 10 cubic centimetres to evaporate away 

 slowly, and collecting the gas from that small residue in a gas- 

 holder, we obtained, after removal of oxygen with metallic copper, 

 and nitrogen with a mixture of pure lime and magnesium dust, fol- 

 lowed by exposure to electric sparks in presence of oxygen and 

 caustic soda, 26*2 cubic centimetres of a gas, showing the argon 

 spectrum feebly, and, in addition, a spectrum which has, we believe, 

 not been seen before. 



We have not yet succeeded in disentangling the new spectrum 

 completely from the argon spectrum, but it is characterised by two 

 veiy brilliant lines, one almost identical in position with D 3 , and 

 almost rivalling it in brilliancy. Measurements made by Mr. E. C. 

 C, Baly, with a grating of 14,438 lines to the inch, gave the following 

 numbers, all four lines being in the field at once : 



Di 5895-0 



D 2 5889-0 



D 3 5875-9 



D 4 5867-7 



There is also a green line, comparable with the green helium line 

 in intensity, of wave-length 5568"8, and a somewhat weaker green, 

 the wave-length of which is 5560'6. 



In order to determine as far as possible which lines belong to 

 the argon spectrum, and which to the new gas, both spectra were 

 examined at the same time with the grating, the first order being 

 employed. The lines which were absent, or very feeble, in argon, 

 have been ascribed to the new gas. Owing to their feeble intensity, 

 the measurements of the wave-lengths which follow must not be 

 credited with the same degree of accuracy as the three already given, 

 but the first three digits may be taken as substantially correct : 



Violet 4317 Blue 4834 



, 4387 4909 



4461 Green 5560*6 



4671 5568-8 



Blue 4736 Yellow 5829 



4807 5867-7 



4830 Orange 6011 



