254 Scientific Proceedings^ Royal Dublin Society. 



they are faint lines not likely to cause confusion in an analysis, and 

 need not be considered ; so that if three or four prominent lines are 

 obtained in a vacuum tube spectrum, that cannot be identified as the 

 ultimate lines of a known element, one may safely conclude that they 

 are the ultimate lines of an unknown element. With bands it is quite 

 otherwise. Different compounds appear to give rise to different bands, 

 not in any way related to the spark spectra of tlie elements of whicli they 

 are composed, but due to the molecular aggregation of the elements or com- 

 pounds present, so that, from an analytical point of view, no importance can 

 be attached to the appearance of a new band in a spectrum. On investigation 

 it will certainly prove to be the most prominent band in the banded spectrum 

 of some compound of the elements present. The only danger of confusion 

 with bands is when only a small quantity of the compound that gives rise to 

 them is present, when the band fades away and leaves only one or two pro- 

 minent lines of the head of the band, which then have mucli the appearance of 

 the lines of a line- spectrum, and may possibly cause the analyst to erroneously 

 suspect the presence of some new body. 



As a general rule, the strongest lines in the spark spectrum are the 

 strongest lines in the vacuum tube spectrum of an element ; and, as a general 

 rule, the ultimate lines in the vacuum tube spectrum appear to be identical 

 with the ultimate lines of the spark spectrum, but this is not always the ease, 

 and it is very desirable that the ultimate lines of vacuum tube spectra sliould 

 be carefully investigated and tabulated at an early date, for without a 

 knowledge of the ultimate lines of tlie elements it is often difficult, if not 

 impossible, to arrive at any certainty in the identification of the lines of a 

 spectrum. Manganese gives a very characteristic group of lines that are 

 surprisingly persistent in the vacuum tube A 2801'3, A 2798'5, and X 2795'3, 

 and these are quite different from the ultimate lines of the spark spectrum 

 of manganese A 2605*8, A 2594'0, A 2576-2. I have seen this group in the 

 vacuum tube spectra of metallic lithium, and of potassium chloride, and 

 barium ciiloride, no other lines of manganese showing. 



The tenacity with which metals adhere to a tube when once they are 

 introduced is very surprising, and would indicate that very small quantities 

 are all that are necessary to give a spectrum. 



Photographs of the various spectra are reproduced in Plate XXII., and 

 an approximate scale of wave-lengths is given to facilitate the identification 

 of the lines, but it is not exactly in position over the lead spectrum. For 

 the purpose of accurate measurement the cadmium spectrum has been 

 photographed through the middle of some of the spectra. 



I have to thank the Q-ovcrnment Grant Committee of the Royal Society 



