108 Prof. F. W. Clarke on the Results obtained 



General mean, Yt= 97-616. Excluding Popp's work as 

 worthless, the general mean becomes Yt= 89*816 ; +'067. 

 Ytterbium.— mison. Yb = 172-761; ±'038. 



Erbium. — For this metal the most probable results are those 

 of Cleve, which give, recalculated, Er= 165*891. Previous 

 workers undoubtedly studied material rich in ytterbium and 

 other metals of the group. 



For terbium, samarium, phillipium, decipium, thulium, &c. 

 there are no satisfactory data. Cleve gives 170*7 for thulium, 

 while Delafontaine puts Ph= 123 to 1 25, and Dp = 171. These 

 figures assume that all the earths of the group are sesquioxides. 



Columbium*. — Marignac's results give approximately 

 Cb=94 ; Blomstrand regards 95 as the most probable value. 

 New estimations are needed. 



Tantalum. — All of Marignac's figures give a general mean 

 of Ta= 182*144 ; ±*166. Probably 182 may be safely used 

 as the true value. 



Platinum. — By a recalculation of the results lately pub- 

 lished by Seubert, I get Pt= 194*415 ; ±*049. 



Osmium. — Berzelius's figures give me Os = 198*494. 



Iridium. — The general mean calculated from Seubert's 

 weighings is Ir= 192*651; + *033. 



Palladium. — Berzelius's last results are the only ones worth 

 considering. They give Pd = 105*737. 



Rhodium. — Data by Berzelius. Bh = 104*055. 



Ruthenium. — A single analysis of potassium ruthenio-chlo- 

 ride by Claus gives Bu = 104*21 7. Plainly the values for Ru, 

 Rh, Pd, and Os need scrupulous redetermination. 



Conclusions* 



A careful scrutiny of all the data upon which the foregoing 

 atomic-weight calculations depend will reveal various sources 

 of error. Of course, each series of results must be considered 

 by itself and weighed on its own merits ; but a few general 

 errors are important enough to warrant mention here. 



First, every value after oxygen, with one or two partial 

 exceptions, involves whatever error may attach to the atomic 

 weight of oxygen. If the latter is 16 instead of 15*9633, this 

 error in some instances becomes multiplied to a serious ex- 

 tent, as a glance at the tabulated results will show. Other 

 similar errors are repeated continually. The value assigned 

 to any element is necessarily affected by whatever errors may 

 attach to the atomic weights of those other elements through 

 whose medium it is referred to the standard, hydrogen. 



* This name has priority over the generally accepted " niobium/' and 

 therefore is entitled to preference. 



