250 PROCEEDINGS OF THE AMERICAN ACADEMY. 



In addition to the eleven pairs of lines in the fluorescence spectrum 

 excited by cadmium 5086 there are two strong lines at wave-lengths 

 5305 and 5341. These seem to belong to the same series, and the 

 former has a faint companion, the two forming a doublet. The line 

 5341 is also accompanied by a companion, which, however, is so faint 

 as to be barely distinguishable. 



The spectrum excited by the more refrangible of the two blue cad- 

 mium lines 4678 is reproduced only on the chart. It consists of a 

 regular series of five lines in the blue region, and a large number of 

 irregularly spaced lines of widely different intensities in the yellow- 

 green region. None of these lines appear to be represented in the 

 magnetic spectrum. 



This spectrum illustrates well the characteristic peculiarity of the 

 sodium fluorescence spectrum, that stimulation at the more refrangible 

 end excites powerful fluorescence at the opposite end. The lines which 

 form the regular series we may call directly excited, the others in the 

 yellow region indirectly excited. The latter in all cases seem to be 

 irregularly spaced. The great problem to solve is to determine the 

 nature of the mechanism and find out how the low-fi:equency vibrators 

 are set agoing by the stimulation of the high-frequency ones, while they 

 remain quiescent when the stimulation is at the middle of the spectrum. 

 Speculations on these points must be deferred for the present. 



Zinc-arc Excitation. 



The complete fluorescent spectrum excited by all three of the zinc 

 lines (i. e. the total radiation of the lamp) is shown near the top of the 

 folding chart, just below the cadmium fluorescence. It is scarcely 

 possible to find a coincidence of two lines. The two spectra placed 

 side by side make a striking picture of the variation produced by 

 different excitations of the same fluorescing medium. A photograph 

 of the spectrum is reproduced on Plate 2, c. 



Exciting with the zinc line 4811 alone gives us the spectrum shown 

 on Plate 2, b. The other two zinc lines are of course present on account 

 of diff'used light. Referring now to the chart, we find that the violet 

 end of the spectrum agrees pretty well with the fifth magnetic series, 

 though other lines are present. The two strong lines at 5188 and 5225 

 also appear to belong to the same series. The lines on the whole are 

 much less regularly distributed than in the case of the cadmium 480 

 excitation. The three wide pairs between 523 and 535 are peculiar to 

 this excitation. 



The fluorescence excited by the other two zinc lines, 4680 and 4722, 

 is also recorded on the chart. These lines are so close together that it 



