1887.] on Light as an Analytic Agent. 89 



Strips of copper, electroplated with nickel, brought out almost 

 all the strong nickel lines in the ultra-violet between K and Q ; 25 

 were photographed. When nickel oxalate was put in so as to give a 

 powder of metallic nickel after the first explosion, the same lines 

 were developed, and three additional lines in the ultra-violet. Only 

 one line was seen in the visible part of the sj)ectrum, and that was 

 the yellowish-green line (w.l. 5476). 



Coj)per wires electroplated with cobalt gave in the flash 22 lines 

 in the violet and ultra-violet, between G and P ; no lines beyond 

 those limits. Cobalt oxalate gave no more. 



No other metal gave anything like so many lines as iron, nickel, 

 and cobalt ; and it is remarkable that almost all the lines of these 

 metals develoj)ed in the flash lie in the same region between G and P. 



We exj)ected that manganese would have given several lines in 

 the flash ; but it was not so. Neither metallic manganese, nor any 

 of several compounds which we tried, gave us any lines of that metal 

 except the violet triplet, and this was generally given by the iron 

 tube alone, and was merely stronger for the manganese put in. The 

 green channellings characteristic of manganese, and ascribed to the 

 oxide, were, however, well seen when metallic manganese was used. 



Chromium, introduced as bichromate of ammonia, which of course 

 became chromium oxide at the first flash, gave three triplets in the 

 green, the indigo, and the ultra-violet near N respectively, but no 

 other lines. , 



Bismuth, antimony, and arsenic gave no lines, nor did mercury 

 spread over a sheet of copper lining the tube. Several metals were 

 tried as amalgams spread over such a piece of copper, but with no 

 fresh results, except in the case of thallium, which gave the green 

 line strongly, the strong line between L and M, and two lines between 

 N and O. 



On the whole it does not appear that the form in w^hich the metal 

 is introduced into the tube makes much difference. The merest 

 traces of those which gave lines were sufficient. Generally when a 

 metal had been put into the tube, its lines continued to show after 

 the strip or lining had been removed. Thus, after the nickel strips 

 had been taken out, and the tube cleaned out as completely as it 

 could be mechanically, the nickel lines still came out in the flash, 

 and the same was the case with other metals. 



The strongest part of the water-sjDcctrum, from s to near R, 

 generally impressed itself more or less on the photograj)hic plate ; 

 but, with the exception of T, which was only developed once or 

 twice, no lines made their appearance in the region more refrangible 

 than s. 



Thus far the experiments had been made with the gases at the 

 atmosj)heric pressure, or nearly so, before ignition. The proportions 

 of hydrogen and oxygen were nearly two to one ; but an excess of 

 either gas to the extent of one-fifth did not sensibly affect the results. 



Other explosive mixtures were tried. Carbonic oxide with 



