524 



PROCEEDINGS OP THE AMERICAN ACADEMY. 



It seems probable that the exposure of the Wehnelt is at least as great 

 in proportion as the others, since the line at X 6708 comes out clearly 

 and lines of shorter wave-length than X 3233 are visible. 



The resemblance to the sj^ark spectrum is striking, but there are 

 considerable differences in intensities, X 4972 being much weaker and 

 A 0104 much stronger. Many of the strong arc lines are either absent 

 or greatly reduced in intensity. It is of interest to note that the lines 

 which retain a part of their intensity belong to the principal and first 

 subseries of Kayser and Runge, while those which lose all or a great 

 part of their intensity from arc to Wehnelt belong to the second sub- 

 series. The lines X 4972.0, X 4273.5, X 3986.0 are of this latter class. 



SODIUM. 



Spark. Sodium chloride on carbon electrodes. 

 Arc. Metallic sodium on carbon electrodes. 

 Wehnelt. Carbon point in solution of sodium hydroxide, 

 in lithium.) (Plate 1, Fig. 2.) 



(Impurities same as 



A rather close agreement with the spark spectrum is to be seen, with 

 a noticeable diflference in the relative intensities of the lines at X 5682 

 and X 5688. The lines which are present in the arc spectrum and do 

 not appear in the Wehnelt spectrum belong in this case also to the 

 second subseries, but it is quite probable that the exposure was insuffi- 

 cient to bring out these relatively weaker lines. X 5896 is at least as 

 strong as X 5890. Taking into account the rapid decrease in the sensi- 

 tiveness of the plates in this region, it must in reality be stronger. 



