I9I3-] HINRICHS— TRUE ATOMIC WEIGHT OF BROMINE. 



557 



of intersection, as just done. Besides : each single point in the hne 

 of perturbation above the point of intersection is balanced by the 

 equal value of contrary sign belozu that line symmetrical with the 

 above point. 



Trusting that we have made this subject as clear as so difficult 

 a matter — at first reading — may be made, we shall add only that all 

 thought of the so-called " probable error " of the mean must be laid 

 aside in atomic weight determinations ; we have repeatedly shown its 

 utter absurdity in this field — the more careful laboratory work hav- 

 ing the desperate character of placing itself squarely far beyond the 

 field restricted for it by the so-called " probable error of the mean " 

 so as to leave it not a shred of probability. We expect as soon as 

 possible, by a most striking example (covering all the work done for 

 an important element) to show the utter fallacy of this "probable 

 error of the mean " introduced by Gauss a century ago. We shall, 

 at the same time, show that it is but a false and misleading substitute 

 for the simple mean value of the actual distance of each determina- 

 tion from the mean of all. 



We here insert, from a most extended table carefully classified 

 by order of magnitude and for each individual element separately, 

 giving the most general results in a broader grouping, to show the 

 fact referred to above : that with progress in practical laboratory 

 work, the final departure has been greatly reduced. 



The four catalogues of work represent: (A) recent and classical 

 work (Dumas, Stas to present) ; (B) older determinations (except 

 the classical) ; (C) dominant elements only: i. e., O, CI, Ag; C, Na; 

 S, Br, H ; N, Ka. 



Per Cent, of Each Grade Specified. 



