342 BELL SYSTEM TECHNICAL JOURNAL 



experimental point of which at least the abscissa Is certain. But at the 

 frequencies between the lines, there is no way of getting information; 

 and many a published curve is traced by guesswork right across the 

 regions of major importance where data are essential, simply because 

 Nature left those regions vacant of lines in the spectrum of the mercury 

 arc! 



If on the other hand the source has a continuous spectrum or one 

 crowded with bright lines, the device for resolving or filtering the light 

 will transmit to the ionizable gas photons not of a single wavelength, 

 but of a range of many Angstroms — dozens or scores of Angstroms, 

 perhaps even a hundred. A single measurement may be, and usually 

 is, plotted as if it belonged to the central wavelength of the transmitted 

 band; but actually the ionization results from waves of all the wide 

 range, and not even from a uniform distribution of energy across the 

 range, but from a distribution affected by the qualities both of the 

 source and of the resolving apparatus. With chemical filters, i.e. 

 with coloured absorbing liquids, bands of transmitted light may be 

 formed in various parts of the spectrum. They are likely to be broad 

 and hazily bounded, and limited in number; but the liquid filters are 

 inexpensive and easy to handle, and in tracing backward the sequence 

 of observations on any one substance, one often finds that the very 

 earliest were made with filters. Monochromators— which is to say, 

 spectroscopes — form bands of which the central wavelength and the 

 width may be varied at will. This sounds ideal; but in practice, of 

 course, the narrower the transmitted range of wavelengths, the 

 scantier the transmitted energy; and one must compromise as best 

 one may between a band too narrow to produce a measurable degree 

 of ionization, and one so broad that it is hard to apportion the credit 

 for the effect which it produces among the frequencies which make it 

 up. It will be evident that the ideal curve, drawn through experi- 

 mental points scattered thickly all through the spectrum and each 

 corresponding to a single wavelength, is difficult of attainment and 

 even of approach ! 



As the atmosphere of the earth prevents us from observing the 

 spectra of the stars at shorter wavelengths than some 280w/i, so the 

 opacity of all terrestrial solids prevents us from projecting quanta of 

 smaller wavelength than \2Smix, into an enclosure. Indeed, it is only 

 from fluorite and only from occasional samples of fiuorite that one can 

 make windows which are transparent so far out; the next best and much 

 the commoner substance, quartz, ceases to transmit at about 145m/z. 

 We are thus almost entirely debarred from observations on the noble 

 gases and on the common diatomic gases, which is deplorable. 



