178 THE ATOMIC WEIGHTS. 



The next determinations were made by Dumas,* who re- 

 turned to the original method of Berzelius. Pure selenium 

 was converted by dry chlorine into SeCl^, and from the gain 

 in weight the ratio between Se and CI was easily deducible. 

 I include Berzelius' single experiment, which I have already 

 cited, and give in a third column the quantity of chlorine 

 absorbed by 100 parts of selenium : 



1.709 grm. Se absorb 3.049 grm. CI. 178.409 



1. 810 " 3.219 " 177-845 



1.679 " 3-003 " 178.856 



1.498 " 2.688 " 179.439 



1.944 •' 3.468 « 178.39s 



1.887 " 3-3^2 " 179.226 



1-935 " 3-452 " 178-398 



1 79.000 — Berzelius. 



Mean, 178.696, ± .125 



The question may here be properly asked, whether it 

 would be possible thus to form SeCl^ and be certain of its 

 absolute purity ? A trace of oxychloride, if simultaneously 

 formed, would increase the apparent atomic weight of 

 selenium. In point of fact, this method gives a higher 

 value for Se than any of the other processes which have 

 been adopted, and that value has the largest probable error 

 of any one in the entire series. A glance at the table which 

 summarizes the discussion at the end of this chapter will 

 render this point sufficiently clear. 



Latest of all, we come to the determinations made by 

 Ekman and Pettersson.* They tried various methods of 

 investigation, and finally decided upon the two following : 



First. Pure silver selenite, AggSeOg was ignited, leaving 

 behind metallic silver in the subjoined percentages : 



* Ann. Chem. Pharin., 113,32. i860. 



f Ber. d. Deutsch. Chem. Gesell., 9, 1210. 1S76. Published in detail by the 

 society at Upsala. 



