78 Froceediijg.s of the Royal Society of Victoria. 



its couipletion, and Caesium a pair of normal series. This may 

 account for some discrepancies in my table of wave lengths and 

 likewise reinstate several wave lengths rejected in my first paper. 

 The atomic weights, taking Sodium = 23, may be calculated 

 from proposition No. 3 as follows: — Helium = 3-64; Na = 23; 

 K = 39-05; Rb = 85-15; Cs= 133-47. Kaiser and Runge calculated 

 the atomic weights from the differences of the vibration numbers 

 and obtained the values, viz., Na = 23; K = 40-6; Rb = 82'6; 

 Cs=126. 



I may state here that the same proportionality prevails in the 

 second group of elements, and even in the other groups, where 

 series have been discovered. In the case of triplets the first and 

 the third wave length yield the proportional value. My investi- 

 gations on those groups have, however, not been concluded. I, 

 therefore, refrain from giving the numbers. 



Rydbei-g's proposed law connecting the roots of the subordinate 

 series with the headlines and root of the principal series, although 

 not strictly true for Helium and Potassium, I have made use of 

 for the approximate determination of the modulus of the head 

 series of Caesium. 



T. Balmer has quite recently published a formula similar to 

 mine with three constants. Its application to the Helium lines 

 shows good accordance, excepting the c^ and r series, where it 

 fails to the extent of more tlian 10 Angstroem units. His jjaper 

 is published in Wiedemann's Annalen der Physik and Chemie. 

 The same periodical is the source from whence the experimental 

 data of my paper have been derived, excejjt the Helium series, 

 which have been published by Runge and Paschen in the 

 Sitzungsberichte der Academie in Berlin last year. 



In my table of wave lengths the experimental values and the 

 calculated ones are placed one above the other, those on a line 

 with E, E being the nvimbers derived from observations and 

 corresponding to those inunediately above. 



If I may allow myself any conjectures regarding the nature of 

 the curves and the origin of the spectrum lines, I consider the 

 curves to belong to the conical order — hyperbolas in preference 

 — while the lines are caused by the interference of two light 

 waves, the distance of one line from the next following being 

 consistently 2/z + 1. 



