556 REPORT— 1897. 



and will be of the same order of importance as tlie other. Now the effective elas- 

 ticity of the gaseous medium surrouHding the vibrating molecule, when thus 

 treated as continuous, varies inversely as its dielectric constant. We should thus 

 expect on the above hypothesis that increase of pressure would lower the free 

 periods roughly in the same ratio as it raises the square root of the dielectric 

 •constant. To reduce to figures : a shift of ^ of the distance between the D lines 

 •would correspond to 5a/X = ^.10~*, while the dielectric constant of air at 0° C. 

 and atmospheric pressure is I'OOOG. Thus this shift towards the less refrangible 

 ■end would indicate a change of density of the surrounding air of the order of 

 that due to a pressure oi^ of an atmosphere at 0° 0. 



This would make the eflect about 10- times too large for the observations : 

 thus the main seat of the fether strain maintaining the vibrations of the molecule 

 is the free aether immediately surrounding it, and the loss of stiffness due to the 

 other molecules which are some way otf diminishes the free periods only about 

 10 ~- times as much as if it were averaged right up to the vibrator. With 

 eimilarly constituted lines, it is the relative shift SA/\ that is proportional to the 

 change of density of the medium. 



10. Changes in the Wave-frequencies of the Lines of Emission SjKctra of 

 Elements. By W. J. Humphreys. 



For more than two years the best spectroscopic equipment of the Johns Hopkins 

 University has been devoted chiefly to the study of changes in the wave-frequencies 

 ofthe lines of emission spectra. It was found, soon after the investigation was 

 begun, that a change in atmospheric pressure about an electric arc, in which a 

 substance was being volatilised, caused a change in the wave-frequencies of the 

 spectral lines so produced. In studying this phenomenon a concave Rowland 

 grating of the largest size was used, and the electric arc was formed in a closed 

 cylinder provided with a quartz window, the pressure being obtained by pumping 

 air into this cylinder to any extent desired — usually till the gauge registered from 

 six to twelve atmospheres. 



Besides a number of eye observations several hundred photographs were taken, 

 rand a large number of lines carefully measured. In fact, the spectrum of almost 

 •every known metallic element has been examined at various pressures. 



No lines were found to shift more than a fraction (usually less than the tenth) 



of an Angstrum unit, but the shifts are of such regularity that as the wOrk pro- 

 gressed several interesting relations between the shifts of the lines, the conditions 

 under which the lines are produced, and the elements producing them became 

 evident. Some of these relations (given below) may be more or less accidental, 

 while doubtless others of as great importance have been overlooked. However 

 the labour of the investigation was spent iu determining the facts in regard to the 

 lines examined, and not in hunting after empirical relations. 



It is impossible, of course, in a mere abstract to enter into details of any 

 ■description, and I shall therefore confine myself to the following summary of 

 results. These are : — 



1. Increase in pressure around the arc causes all isolated lines to shift towards 

 the red end of the spectrum. 



This is entireh' independent of the manner of the lines spreading out, and is 

 the same for a line when reversed as when it is fine and sharp. Even those lines 

 which, like the sodium lines X3302 and X3303, spread to the violet give reversals 

 that shift to the red. 



2. The shift is directly proportional to the increase of pressure about the arc. 



3. It does not depend upon the partial pressure of the gas or vapour producing 

 tihe lines, but upon the total pressure. In other words, it is not affected by 

 quantity of material iu the arc. 



4. The shift of the lines seems to be nearly or quite independent of tem- 

 perature. 



