2 Stromeyer, Grouping of Chemical Elements. 



scientific value, that it is purely empirical, and might per- 

 haps with advantage have been replaced by a fair curve 

 drawn by hand. 



The discovery had been made that in general, the 

 differences between the actual atomic weights and those 

 found by the calculation from the group positions — being 

 whole numbers — were practically of the same value for 

 each one of MendeleefP s groups. In other words, if the 

 formula could have been reversed and the group positions 

 calculated from the ascertained atomic weights, it would 

 have been found that the group positions were not only 

 not whole numbers, but also that the intervals between 

 them were very irregular. The process actually adopted 

 in the calculation of Table I. was to determine the average 

 differences between the real and estimated atomic weights 

 for each group, and thus fix its average position, and then 

 to recalculate the atomic weights with these new values. 

 The first group, headed by Lithium, may serve as an 

 example. The preliminary calculation had fixed the 

 position of this group at 1*04 (see table), the value of N 

 for the individual elements was therefore as follows : 

 Lithium rc>4, Potassium iy'04, Rubidium 33*04, Caesium 

 49/04. The calculation of the atomic weight of Potassium 

 is as follows : — 



^°-^ = i7'94, 375° x i7"°4 = 63°-54' ) 7-5° x 17-04= i27°48'„ 

 17-04 



i5°x 17-04 = 255°36'. and the sines of these angles are: 



4-0-8980, +0-7902, -0-9686. 



Then we have 



2-788 . . N ... .. ... ... = +47-526 



Less 6 - o ... ... ... — 6"ooo 



+ 41-526 



