448 PROCEEDINGS OF THE AMERICAN ACADEMY, 



two of these are probably accurate to one part in 80, while the two last 

 are much less accurate, it seems fair to say that these differences are 

 probably much smaller than the value for the usual series (17.2). 



Many unsuccessful attempts have been made to find a simple formula 

 which would express Lenard's series. The formulae of Kaj^ser and 

 Runge, Kydberg, Fowler and Shaw,* and Ritz,f and modifications of 

 these, have been tried. Unless the formula contained four adjustable 

 constants, it could not be made to fit the observations with any degree 

 of precision. As almost any series, if not too accurately known, could 

 be represented by a four-constant formula, no results of this work are 

 given. 



In the spectrum of potassium, two new lines were found, one in 

 almost the position predicted for it.| It is one of the hitherto missing 

 first pair of tlie first subordinate series, which is, for some obscure 

 reason, very faint. (Ritz has given reasons for believing this to be 

 the first rather than the second subordinate series, as Kayser and 

 Runge classified it.) The other member of this pair was not found, 

 owing to the broadening of the line at 6939, near which it doubtless 

 lies. The other new line (7931.8) was so faint that its companion, if 

 it is a member of a pair, could not be seen. The first term of the 

 principal series was excellently photographed on several films and the 

 writer feels considerable confidence in the value of the wave-lengths 

 given for these lines. They are recorded to the second place of decimals, 

 as the difference between the values could be determined more accurately 

 than the values themselves. The best measurements were taken on 

 photograplis where the potassium was present as an impurity and the 

 lines were fine and sharp, though measurements on heavily reversed 

 images of these lines gave concordant results. 



The line at 46-42 seems to be outside the series formation, and, along 

 with the funtness of the first subordinate pair, offers a very odd peculi- 

 arity in this spectrum. 



Rubidium. 



Three new lines were found in the spectrum of rubidium. One is a 

 line at 3158 (not seen as a pair) belonging to the principal series; 

 another, a very diffuse and faint line at 5171, which, with 5234, forms a 

 pair in the second subordinate series. The thiid is a line at 7759.5, 

 wh'fh is a companion to the first subordinate series line 7757.9. The 



* Astrophys. J., 18, 21 (190.3). t Ann. der Pliys., 12, 2(34 (190.3). 



t See Kitz, Ann. der Pliys., 12, 264 (1903). 



