230 Scientific Proceedings, Royal Dublin Society. 
under examination, and at first it was supposed that the solutions 
had sprayed on to the fresh electrodes; but on keeping the fresh 
electrodes in another room, no difference was observed, and in the 
case of an element such as iron or calcium, the dilution spectra 
could not be followed beyond the ‘1 per cent. solution, as the elec- 
trodes then gave as strong spectra asthe solutions. It was then seen 
that the atmosphere was charged with the element, and remained 
charged for a considerable time. In the following investigations 
the difficulty was got over by beginning with the most dilute 
solution, and working backwards towards the strong solutions, 
finally sparking the metal when it could be procured. 
The photographs of spectra extend from 2 5900 to A 2500; 
but the plates were not very sensitive below X 4792°8, nor was the 
instrument in perfect focus beyond A 2590°2. 
It is a remarkable fact that the residuary lines of an element 
differ greatly with the method of excitation, and there is no 
guarantee that the residuary lines here tabulated would be the 
most persistent lines if the substances were vapourised by something 
other than the condensed spark; certainly, in the case of the 
oxyhydrogen flame, there is a notable difference; thus, with 
manganese, we have shown in a previous paper’ that, when the 
condensed spark is used, the residuary lines are AX 2605°8, 2594-0, 
and 2576°2; but if the oxyhydrogen flame be employed to 
vapourise the element or its compounds, the residuary lines, as 
shown by Professor Hartley,’ are AX 4034°6, 4033°2, 4030-9; 
and, in general, we note that, with the oxyhydrogen flame, the 
residuary lines tend to the less refrangible end of the spectrum ; but 
with the condensed spark they tend to the more refrangible end. 
Apparently the nature of the dilutant has no effect on the 
residuary lines; thus the same residuary lines would be obtained 
whether the metal was in the form of a dilute solution or alloyed 
with another metal; but we have not yet investigated whether 
the degree of persistency is affected; probably it would be 
influenced by the relative volatility of the diluting metal in the 
alloy, and the sensitiveness greatly reduced owing to the vapour 
of the dilutant being itself a conductor, so that in an alloy one 
would not readily detect the presence of less than -1 per cent. of a 
substance. In tabulating the results, when the intensities were 
other than those usually accepted, they are enclosed in brackets. 
1 Op. cit., p. 217. * Hartley, Phil. Trans., vol. clxxxv., Part i., 1894, pp. 161-212. 
