1883.] on the Ultra-Violet Spectra of the Elements. 247 



which exhibit the greatest variety in their chemical relations, and 

 produce combinations of the greatest number of types, and the most 

 comjilicated spectra, are also those which produce the most highly- 

 coloured compounds. In marked contrast to the thick-set ranks of 

 iron, manganese, and chromium lines, are the few scattered rays 

 exhibited by those metals which form their combinations each chiefly 

 on a single type, such as aluminium, and the alkali and alkaline earth 

 metals. These spectra are probably even simi^ler than at first sight 

 they seem to be. That of lithium is the simplest (Plate II. fig. 3) : 

 a series of single lines succeeding one another at decreasing intervals, 

 and with diminishing intensity, closely resembling in these respects 

 the spectrum of hydrogen. In the case of hydrogen, we know that 

 the oscillation frequencies of some of its rays are related in a simple 

 I harmonic ratio. We are not able to say that the relation is so simple 

 j in the case of lithium ; but still the whole series are probably over- 

 \ tones of a fundamental vibration, not so simply related as the harmo- 

 [ nics of a uniform stretched string, but, like the overtones of a string 

 ) which is not of uniform thickness, or is loaded at different points, 

 I similarly related in origin, though not exact harmonics. That the 

 j different rays are in many cases so related as overtones of a fundamental 

 I vibration appears more plainly, perhaps, when not single lines but 

 f groups of two, three, or four lines recur. Potassium shows a series of 

 j pairs to which the well-known violet pair, and perhaps that in the red 

 j also, belong. Calcium, magnesium, and zinc, each show a series of 

 triplets, which are alternately sharply defined and diffuse (see photo- 

 graphs 4 and 5 and Plate II. figs. 5 and 6). In other cases the same 

 characters may be traced, though less readily, because there is some- 

 times more than one such series of lines or groups. The alkali metals 

 have each one such series in the visible sj)ectrum, and another in the 

 ultra-violet. It may happen in other cases that two or more such 

 series overlap, and it may then be very difficult to distinguish and 

 -separate them. 



In some cases elements show at a lower temperature a far more 

 complicated spectrum than they do at higher temperatures further 

 removed from their points of liquefaction. This has been observed 

 by Roscoe and Schuster in the case of the alkali metals potassium and 

 sodium, which give at temperatures only a little above their boilino^- 

 points absorption spectra which consist of closely-set fine lines, pro- 

 ducing an appearance of shaded bands quite unlike their emission 

 spectra at higher temperatures. In some few cases we have observed 

 similar " fluted " or " Venetian blind " spectra, as they have been 

 called, in the ultra-violet, as, for example, one produced by tin ; 

 but in general the temperature of the arc, which we have chiefly 

 used in our observations on metals, is high enough to carry the 

 i| metals beyond the stage in which their vibrations are constrained by 

 the state approaching to liquefaction. 



But though metals do not often show spectra of this class at 

 the high temperature of the arc, it is otherwise \vith metalloids and 



