PHOTOGRAPHS OF ARC SPECTRA. 



107 



increases, until, at saturation, the risino; points of all higher harmonics 

 become coincirlent with that of tlip first harmonic of the induction. 



For purposes of more critical investigation, a series oF h, \p, curves 

 such as those in Fig. V., together with corresponding ones such as 

 are given in Fitis. II F., I\^., for /x„ and 6 for the same speed, would 

 enable one to construct a series of induction waves for that speed 

 from which the experimental errors and slips in reduction (which we 

 can hardly hope are entirely absent from this work) would be partly 

 eliminated. 



10. With regard to tlie formula — 



I = (a + hn) av*- 



proposed by ow^ of us, wo tind that it is only an approximate i-epresen- 

 tation of "the facts x the index of SB is not independent of the 

 frequency, and, probably, not of the wave form of H. That it is a 

 fairly close approximation, however, when one type of exciting current 

 is used and the range of frequency not very large, is shown by Fig A"I., 

 in which locr I is jdoUed against log 3B from Tables I., II., and III. 

 The points for each frequency ar<> seen to fit closely to a straight line, 

 and the three lines for the three frequencies are very nearly parallel, 

 the slope showing a slight tendency to increase as the frequency 

 increases. 



The formula — 



I = (00257 + -0000(5 n) ^IV *8 7 

 gives the values of I in these three series with considerable accuracy. 



For frequencies higher than 50 per sec. the above formula does 

 not apply. For any one frequency, however, I is given by an expression 

 of the form cr"!^'' . Thus for series IV. in which — 

 a = 50 iq.p.) I = -00371 3B'% 

 for series V. m which — 



11 = 100, I = -00124 '■:B'^^'% 

 and for series YI. in which — 



11 = 179, I =00214 a?!"*. 

 Obviously further investigation on this subject is needed. 



11. The former conclusions as regards the actuality of such a 

 quantity as Kinetic hysteresis, and the general way in which it varies 

 with induction density and freijuency, are fully verified by the present 

 investigation, as a glance down the columns under K in tlie tables 

 will show. 



8._PH0T0GRAPHS OF ARC SPECTRA OF METALS UNDER HIGH 



PRESSURE. 



Ui/ W. GEOFFREY DUFFIELD, D.Sc. 



In 1896 it was discovered by Humphreys and Mohler* that when 

 a source .of light was subjected to pressure, the spectral lines were in 

 general broadened and displaced towards the red end of the spectrum; 

 that is to say, the rapidity of tlie vibration was diminished by 

 pressure. 



The present research was undertaken in the physical laboratories 

 of the Manchester University in 1904, at the suggestion of Dr. 



* Astrophy.sical Journal, vi., 101, 189(5. 



