114 Mr. H. M. Vernon on the 



quantity of samaria as an impurity, most of this being found 

 in the fraction S y : also it is quite possible that a very small 

 quantity of some other earth may be present, and that it may 

 be to the presence of this that the variations in the spectra of 

 the new earths are due. Crookes has shown that the merest 

 trace of a certain earth, unable to be detected by chemical 

 means, can cause great variations in the spectrum of another 

 earth. Perhaps another experiment of Crookes (Chem. News, 

 1887, p. 87) will serve to still further confirm the point at 

 issue, namely, the unreliability of the radiant-matter test. 



Marignac presented Crookes with what was supposed to be 

 a pure specimen of a new earth, to which he had provisionally 

 given the symbol Ya. This earth was found to possess a 

 radiant-matter spectrum almost identical with that given by 

 a mixture of 61 parts of yttria and 39 parts of samaria, which 

 had been ignited with sulphuric acid. The only difference 

 was that in the yttria-samaria spectrum there is present a 

 citron band wdiich is absent from the Ya-spectrum. " Hence 

 the earth Ya is shown to consist of samaria with the greenish 

 blue of yttria, and some of the other yttria bands added to it. 

 It proves further that the citron band, which w r as hitherto 

 regarded as one of the essential bands of the yttria spectrum, 

 can be entirely removed, whilst another characteristic yttrium 

 group, the double green band, can remain with heightened 

 brilliancy. If, now, it were possible to remove the citron- 

 band-forming body from this mixture, the earth Ya would be 

 left behind ; in fact, the earth Ya would have been recom- 

 posed from its elements. - " This, Crookes says, he has no 

 doubt he will ultimately accomplish. 



Now the earth Ya must either be identical with the yttria- 

 samaria earth, or it must be distinct from it. If it is identical, 

 then the radiant-matter test is not reliable, as the same bodies 

 have been found to give spectra which do not coincide in all 

 their lines. If it is distinct from it, then the radiant-matter 

 test is equally unreliable, as different bodies have been found 

 to give spectra the larger portions of wdrich are absolutely 

 identical. 



It has thus been shown how very little reliance can be 

 placed in this new physical method of differentiating bodies. 

 Time alone will show the degree of reliance that should be 

 placed in it. Even Crookes himself seems to feel this in 

 some parts of his papers, though in others he states that the 

 further he goes the more reliable he finds it to become. 

 In Phil. Trans. 1885, p. 722, he says :— " One important 

 lesson taught by the many anomalies unearthed in these 



