248 PROCEEDINGS OF THE AMERICAN ACADEMY. 



obtained with silver stimulation is shown on Plate 2, g. The series in 

 this case has no gaps in it, the line at 5170, which is absent with 

 cadmium excitation, coming out strong (see chart as well). The 

 monochromatic illuminator, with its slits reduced to the width 

 of a hair, was arranged to furnish light of wave-lengths cor- 

 responding to other lines in the series, and photographs obtained 

 which are recorded on the chart. It will be seen that faint or missing 

 lines occur in each case, but that their position varies with the point 

 of excitation. If we consider each line caused by a single electron or 

 vibrator, the phenomena suggest that the vibrators are united in some 

 way, perhaps in a closed ring, and that when the system is set in 

 vibration there are nodal points, the position of which depends upon 

 the point in the chain where the periodic force is applied. Moreover, 

 as has already been pointed out, if the force is applied at the " high 

 frequency" portion of the chain, the regions excited are those of 

 highest and lowest frequency, the intermediate portion appearing to 

 be at rest. This is especially noticeable in the case of the bismuth 

 excitation. (Plate 2, e, and chart.) 



In addition to the Hues enumerated above, there are a number of 

 others at the upper end. These do not appear to be distributed with 

 the same regularity, though some of them may form an extension of 

 the series, or more probably may be the beginnings of other series. In 

 general it has been found that in the simple spectra the lines are reg- 

 ularly spaced between the extreme violet end and a point at about 

 X = 5350. Above this point the spacing is generally very irregular, 

 and it is difficult to unravel the spectrum. Of this more will be 

 said later. 



"We will next take the fluorescent spectrum excited by the green 

 cadmium line 5086. 



This spectrum is remarkable in that it is made up of eleven pairs of 

 lines regularly spaced (Plate 2, j and chart). The other two cadmium 

 lines appear on the plate, as the spectrograph was not shielded from the 

 diffused light from the lamp. 



A series in the magnetic spectrum coincides with the series formed 

 by the shorter I member of each pair. The wave-lengths and differ- 

 ences are given in the following table : 



A. A differences. K. A differences. 



5165.85 ... -7J*6A 38.82 



5126.54 •:;^-^.^ 4970.85 



5087.31 • • • • ':^ti 4932.64 " " " " f^l^ 



5048.49- • • •^^•^'^ 4894.58- ' ' ' ^^"'^^ 



