670 TRANS. ST. I.OUIS ACAD. SCIENXE. 



meats Sc, Ti, and F^, fall on this curve, which may be the expres- 

 sion of some constant error or undetermined relation. 



The elements which conform as nearly as could be reasonably 

 expected are H, L/, Be, B, C, N, O, F, Na, M^, A/, S/, P, S, 

 C/, K, C«, Rd, Sr, Y, Zr, N^, a, Mo, Pd, Te, Ba, La, Ce, Er, 

 H_^, T/, P(5, B/, I, As, S3, Br, and Se, thirty-nine in all. 



The curves pass through maximum and minimum "character- 

 istics" of C, N, S/, P, S, S^, C/, and Te; through maximum cha- 

 racteristics of Lz, Be, B, Na, M^, A/, K, Ca, R6, Y, Zr, N(5, C6, 

 Mo, P<f, La, Ce, Er, and T/, in which negative valence has not 

 been observed ; through minimum characteristics of O and F, in 

 which positive valence has not been observed ; through charac- 

 teristics not maxima or minima of Sr, Ba, H^^, I, A5-, S3, Br; 

 through maxima or minima not characteristics of H, P3, and Bt. 



A few other elements have known, though not maximum, mi- 

 nimum or characteristic valences lying on the curves. Among 



+ 3 +3 +2 +2 +2 +5 +2 +1 



them are Co, N/, R«, R/i, Fe, U, W, and Au, No importance, 



however, can be attached to this fact. 



The first prominent exceptions we find to the law are the ele- 

 meats immediately following Ca. These are Sc, Ti, V, Cr, and 

 Fe. The deviation of Sc, Ti, and Fe, may be accoimted for in 

 the manner suggested on p. 669. 



The atomic weight of Cr should be 5 1 instead of 52, in order to 

 bring its maximum-minimum point on the curve. This confirms 

 the results of Siewert's* determinations. The deviation of Cr is 

 not at all surprising, when we remember that improved methods 

 of determining the atomic weight of Cr have steadily reduced it 

 from 56 to 52.009. 



Vanadium also may be considered as a candidate for the posi- 

 tion of Y^, PI. II. Its atomic weight and many of its properties 

 favor this view ; and it lacks only the power of forming a per- 

 vanadate, KVO4 ; or a hydrovanadic acid, HV, or some corre- 

 sponding salt. 



The fact that certain perissads, viz., R3, C3,N(5, and Y, whose 

 atomic weights are still uncertain, fall on the curve of artiads, 

 may be regarded as evidence that their position is entirely acci- 



* Zeitschrift Gesamml. Wissenschaften, 17, 530, 1861. 



