4 A STUDY OF THE ABSORPTION SPECTRA. 



nomena for many of the bands of the emission spectra of fluorides and 

 chlorides of calcium, strontium, barium, and silicon. These give in some 

 cases a normal and in other cases an abnormal longitudinal Zeeman (light 

 being parallel to the magnetic field) doublet, the normal doublet usually 

 being considered as originating from a negative charge and an abnormal 

 doublet as due to a positive charge. The only difference in these two 

 effects is that the light is circularly polarized in opposite directions for 

 corresponding components of the doublet. Dufour considers that so far 

 all spectra (emission or absorption) that show the abnormal Zeeman effect 

 have their centers in the molecules. If one considers the explanation to 

 be due to positive and negative electrons, then the value of e/m for these 

 will be about the same except in the case of some of the bands of xenotine. 

 At the University of Manchester it has been shown that the Humphrey- 

 Mohler pressure-shift is to be observed for bands that give the Zeeman 

 effect. With the exception of the few bands described by Dufour, the 

 wave-length of bands is unalterable by physical and chemical changes. 



The band-spectra are very complex indeed. In Watts's " Index to 

 Spectra" the wave-lengths of over 5,000 bands are given for sulphur between 

 X 6400 and X 3600, over 2,700 lines for iodine between X 6300 and ^ 5100, 

 over 2,800 for bromine between X 6200 and X 5100, and over 2,600 for alumi- 

 nium oxide between X 5200 and X 4400. Complex as these spectra are, the 

 so-called line-spectra of the elements are even more complex. The same 

 author gives the wave-lengths of over 2,300 lines for chromium, 3,000 for 

 iridium, 2,300 for iron (spark), 3,000 for tungsten (spark), and 5,200 for 

 uranium (spark). For most of these elements the greatest number of lines 

 lie in the regions of shorter wave-lengths, and in most cases the maximum 

 number of lines lie between X 4000 and X 3000. For example, Watts gives 

 1,100 iron lines between X 2000 and 3000, over 1,400 lines between X 3000 

 and 4000, 1,100 lines between X 4000 and 5000, over 600 lines between X 5000 

 and 6000, and only a little over 300 between X 6000 and 6750. A similar 

 distribution holds for vanadium, osmium, etc. The work of Schumann 

 and Lyman shows that many more lines exist in the ultra-violet down to 

 X 1000, but it seems quite probable that most of the spark and arc lines lie 

 either in the visible or in the adjacent ultra-violet regions of the spectrum. 



When the source of the line-spectrum is subjected to physical changes 

 the width and relative intensities of the bands change enormously. Ray- 

 leigh ' and Michelson 2 have shown that the Doppler effect accounts for 

 the width of the lines when the pressure is small. Michelson gives a formula 

 for the breadth (6) of the spectrum lines, 



b = ^/OmXia + bdX) 



where 6 is the absolute temperature, m the molecular weight, a and d 

 constants. 



At present only two physical causes are known to change the frequency 

 of vibration of the emitters or absorbers of the line-spectrum. One of these 

 is the Humphrey-Mohler effect that an increase of pressure about the source 



1 Phil. Mag., 27, 298 (1889). * Ibid., 34, 280 (1892). 



