86 Harriry—An Investigation of the Connexion between 
spectra, one composed of lines, the other of bands. The band spectra of the 
metals were observed to be subject to regular variations in structure, some of 
which were evidently connected with the periodic law in such a manner as to 
point to a similarity in molecular constitution; and, furthermore, certain spectra 
which had been attributed to oxides were in reality the spectra of the metals. 
Elements belonging to the same group emit band spectra with characters in 
common, e.g. copper, silver, and gold in the sub-group of the monad metals ; 
calcium, strontium, and barium in the sub-group of the dyad metals; also tin and 
lead in the tetrad group. ‘The bands of silver in the flame-spectrum appeared 
to be related to the principal lines of the element in the same way that a band 
spectrum of a gas at 760 mms, pressure becomes a line spectrum when the 
pressure is reduced to 7 mms. or less. 
In fact, it was found that in the spectra of lead, silver, and copper, the lines 
seen in the heads of bands, or overlaid by bands, were, in nearly every case, the 
only parts of the complex band spectrum remaining, after any one of these metals 
had been diluted with some other metal. 
The effect of reduced pressures upon the are spectrum of iron affords* an 
illustration ; it is a simple line spectrum at a pressure of 17 mms. of mercury, but 
as the pressure increases, a continuous spectrum is also seen in the region of rays 
less refrangible than \ 5300 and in a less intense degree, between d 4700 and 
4300. The same effect is seen in the oxyhydrogen flame spectrum of iron, and 
in the Bessemer flame-spectra, when in either case the vapour-pressure of the iron 
is reduced. 
In nearly every case the lines seen in flame spectra after the bands have 
disappeared correspond with reversed lines in the arc, and are the most persistent 
lines in the spark spectra of the same element. Although it was believed that 
they are not absolutely identical, they must have a common origin, for they 
occur as constituent lines of homologous groups, and they have so very nearly the 
same waye-lengths and persistency as the corresponding lines in the are and 
spark, that no substantial difference has been found between them. 
In endeavouring to account for two distinct spectra being obtained from the 
same metal, by heating in the same manner, and in some instances to 
approximately the same temperature, it was observed that the phenomena were 
such as would arise from the presence of two distinct substances in the flame, 
such as a metal and its oxide,f or by the reduction of the vapour-pressure of the 
substance, the bands being caused by the greater vapour-pressure in the flame 
* J. E. Petavel and R. 8. Hutton, Phil. Mag., 6, 569, 1903. 
| Allotropic forms of the same substance according to Pliicker and Hittorf, the allotropy depending, 
however, solely upon temperature, 
