44 Proceedings of tlie liot/al Jrisli AcaJcmy. 



differences in tlieir intensities, and the nearness of the conii>onents, the 

 accuiacy of the measurements varied, but the error would seldom lie over 

 .005 mm. From the measurements of the intervals in millimetres the 

 vibration differences per unit length of path (cm.) between the original 

 position of the spectrum line and the positions of the components were 

 calculated, i.e., 



1 , AX 



vA ) °^ - X^- 



^^\) 



was calcidated. For X = 2700 this probable maximum error would give an 

 error in the vibration difference, if the measurement were in the second order, 

 of 007, and for X = -4600 in the same order the error would be 0*02. 



II. EUXGE A^D ElTZ EfLES AXD EXPLANATIOX OF THE TABLES. 



Eunge and Eitz have each formulated rules connecting the intervals 

 bstween the comijonents of a spectmm line. In 1907 Eunge* stated that in 

 all the complicated separations which had been observed up to that lime the 

 intervals of the components of the spectrum line from the middle were 

 multiples of an aliquot part of the " normal interval." 



_ AX _ ^ ^ 

 X" m iirC 



f 



where — = the ratio of the charge to the mass of the election, H = the field 



m 



strength, and c = the velocity of light. The aliquot parts -which had been 

 obsei'S'ed were 



a a a a a a a a 



2' 3' i' 5' 6' r 11' 12' 



From this normal interval Paschen and Eunge, using the mercury lines, have 

 calculated 



— = 1*69 X 10' [aia gm~^ \- 



For some lines in tungsten and molybdenum it was necessary to consider the 

 separations as multiples of such a small aliquot part of the normal interval 



a 

 as ^. Now it has been pointed out that, if it l>e necessary to express the 



separations as multiples of aliquot part« with such large denominators, 

 Eunge's rule has not much value in these cases, for then, considering the 



* C. Ruiige, I'hys. Zeit., viii, 232, 1907. 



