EFFECT OF PRESSURE UPON ARC SPECTRA. NICKEL. 



235 



In Table VIII. the values of the mean displacements at 10 atmospheres have been 

 divided by the square and cube of the wave-lengths, and in Diagram 7, figs. 1 and 2, 

 these are shown in statistical form. Many more values are now included, and each 

 diagram is now very irregular. They are not as trustworthy as the previous 

 diagrams for the purpose of ascertaining the rate of displacement with wave-length, 

 but they serve as a guide to the classification of lines which cannot be measured at 

 pressures higher than 10 atmospheres. 



The separation into groups is clearer in the df\ 2 diagram, which consequently 

 favours this ratio. If, however, the lines be sorted out from .the dj\ 3 diagram in 



Diagram 7. 



Fiq.L 



Jli? 247 lot 

 > 74 245 10! 



91 40 It 



55 ta<4 66 2? J2 5841 42 3c il 26 67 M] 



21518 

 I48>7& 



71 '70 2!2 190 194 164 !'9 m JI3 507 J|4 ! 

 ||4C 164 29I68'" IB74I7J 141 



_H 



IF 



Fiq.2 



17 M 18 <a 81 30 3^ 87 Til 



2002202J5 II 



I6S9I 13774101)61 >K[_ 

 \11 14S 66 66 135 +0 81 4S AI J'3| 

 |l81 55 ;356J4* 75 19 2] 32 : 



U7| Jllil tr IS' 1 '901'4 IB7I772I4206 

 67 155 140 !04 176 168 170 WITH 16? 189 l8 

 30 31 87 le 171 I4e 151 159 171 



~w 



15 



25 



, displacements per atmosphere 

 Fig. 1. Frequency distribution of spectrum lines according to valu (wave-length)- 



2. 



(Readings are at 10 atmospheres only.) 



:lisplaccmcnts per atmosphere 

 (wave-length) 3 



accordance with their grouping at pressures above 10 atmospheres, it becomes much 

 clearer, as in fig. 2, and we see that the confusion between values 8 and 10 011 the 

 horizontal scale is due to the overlapping of two frequency curves with rather large 

 variations. The distributions for d/\\ d^ now appear about equally probable for the 

 readings made above 10 atmospheres. 



The rate of increase of the displacement with the wave-length has been a matter 



