184 Prof. W. Ramsay and Dr. M. W. Travers. 



of argon into a light and a heavy portion passing by diffusion ; but 

 without definite results. While the density of the portion passing 

 first through the pipe-clay septum was 19*93, that of the last portions 

 was 20-01. We remarked on these figures in the following terms: 

 " These numbers show that no important separation has been effected. 

 The difference in density of the two portions may possibly be attri- 

 buted to experimental error. . . . As it stands, the difference is an 

 extremely minute one, and it may, we think, be taken that any separation 

 of argon, if effected at all, is very imperfect." 



It thus remained uncertain whether argon was a mixture or not ; 

 although the balance of evidence went against the supposition. 



When helium was discovered in 1895, a new light was thrown on 

 the question. Its density, determined by Ramsay, and independently 

 by Langlet, closely approximates to 2*0. Experiments, in conjunction 

 with Dr. Collie, and subsequently by ourselves, showed that no great 

 difference in density could be brought about by fractional diffusion ; 

 and, indeed, the latter revealed no impurity except traces of argon. 

 On grounds similar to those from which a corresponding conclusion for 

 argon was drawn, the atomic weight of helium must be somewhat 

 below 4'0. The difference between the atomic weights of helium and 

 argon is consequently about 36 ; and this is approximately the differ- 

 ence between the atomic weights of manganese and fluorine (36), 

 chromium and oxygen (36'3), vanadium and nitrogen (37'4), and 

 titanium and carbon (36*4). But between each of these pairs of ele- 

 ments, there exists another, exceeding in atomic weight the lower 

 member of each pair by approximately 16. It was therefore to be 

 expected that another element should exist, with atomic weight about 

 20, and on the assumption that, like helium and argon, it too would be 

 a monatomic gas, its density would be 10. It was with the object of 

 attempting to discover this unknown gas that the experiments on the 

 fractional diffusion of helium were made; and the gases evolved on 

 heating some fifty minerals were investigated in order to find out 

 whether a new spectrum could be observed, but with negative results. 

 Seven meteorites were also examined, as well as six mineral waters, but 

 no new lines could be found. Sixteen of the minerals, two of the 

 mineral waters, and one of the meteorites were proved to contain 

 helium ; but if the gases extracted from them contained any gas other 

 than helium or a trace of argon, it must have been in quantity too 

 minute to have revealed itself to the spectroscope. 



The gas which we were in search of was ultimately found in argon. 

 The argon, amounting to about 15 litres, was prepared by means of 

 apparatus, of which a description is given in the sequel. From the air 

 employed in a liquid state for the purpose of liquefying and fraction 

 ally distilling the argon, two other elementary gases have been 

 obtained, besides one yielding the " Swan " spectrum. A preliminary 



