THE SIGNIFICANCE OF SPECTROSCOPY 227 



without let or hindrance. As a consequence, we get a line 

 spectrum. The differences between flame, arc, and spark 

 spectra must be ascribed to changes in the radiating mechanism, 

 arising from the varied treatment it receives. We might say, 

 for example, that certain of our bells need the strong stimulus 

 of the spark to set them in vibration, while others respond readily 

 to the gentler influence of the flame. 



We have so far spoken of discontinuous spectra as " line 

 spectra," meaning thereby that they show a number of separate 

 lines instead of a continuous strip of radiation. The term 

 " line spectrum," however, has usually a more limited applica- 

 tion than this. There are some spectra containing what appear 

 at first to be a number of short, continuous portions, known 

 as " bands." The bands are sometimes sharply defined at 

 one or both edges, and sometimes gradually fade away. On 

 high magnification they are found to consist of a large number 

 of closely packed lines, arranged with evident regularity, and 

 approaching indefinitely close to one another as a sharp edge or 

 dense central portion is approached. Spectra containing 

 structures of this kind are known as " band spectra," and are 

 thereby distinguished from the " line spectra," which comprise 

 all other discontinuous types. It seems probable that bands 

 are produced by molecules, while lines are due to the workings 

 of separate atoms, so that there is a real difference between 

 them. To summarise, then, we may classify the spectra 

 we have considered as continuous and discontinuous ; the latter 

 are subdivided into band and line spectra, the line spectrum 

 of any one element being dependent on the conditions — flame, 

 arc, or spark — under which it is produced. 



All these different types are classed generically as " emission 

 spectra," because they are obtained from radiation emitted by 

 the body in which they originate. There is another kind of 

 spectrum, however, which is due, not to emission, but to 

 absorption. If we place a transparent or semi-transparent sub- 

 stance, such as didymium glass or iodine vapour, in the path of 

 a beam of continuous radiation, certain waves are absorbed by 

 it, and the resulting spectrum shows dark lines in the places 

 corresponding to those waves, on a continuous background due 

 to the unabsorbed radiation. Absorption spectra produced in 

 this way are characteristic of the absorbing substance. An 

 important type of absorption is that due to vapours which are 

 themselves incandescent. If such a vapour be traversed by a 

 continuous beam from a hotter substance, the absorption lines 

 in the spectrum will correspond exactly to the emission lines 

 which the absorbing substance would give if it were acting 

 alone. We shall see shortly that this phenomenon has an im- 

 portant bearing on the interpretation of solar and stellar spectra. 



