of Simple Bodies, according to Thomson and Berzelius. 163 



(a.) Aluminum. — The number of Berzelius for this metal 

 is influenced by his Canon, of the proportions observed by 

 negative substances in combinations ; which makes 2 atoms of 

 aluminum combine with 3 of oxygen (Al 2 *) to form alumina. 

 In following the simpler views of British chemists, as better 

 suited to the sliding scale, no inference is intended hostile to 

 that Canon; which has perhaps not been justly appreciated 

 in this country generally f. Berzelius deduces his number 

 from an experiment (Essai sur le Theorie des Proportions 

 cJiimiques, p. 148), where in 100 sulphate of alumina gave, by a 

 red heat, 29*934 of alumina. Neglecting the Canon, this will 

 make alumina |^:f x 5 = 2*14; the mean, with 2*25, Thom- 

 son's number, = 2*2, and hence aluminum 1*2. 



(b.) Antimony. — The experiments of Thomson giving very 

 direct results, confirmed by Phillips's analysis of tartar emetic 

 (Ann. Phil. 2nd Ser. ix. 373), his number is retained ; although 

 the statements of Berzelius (Essai, p. 123), and more at large 

 Ann. Phil. iii. 248, throw doubts upon its accuracy. 



(c.) Bismuth. — The experimental reasons for the alteration 

 of the number for bismuth in Berzelius's last table (Ann. Ck. 

 xxxviii. 427) not being given, his construction of the oxide 

 only, Bs 3 , can be referred to for the explanation. Hence two- 

 thirds of his number == 8*869 must be taken to compare with 

 Thomson's ; when they approached so nearly, that under the 

 circumstances no change seemed desirable. 



(d.) Boron. — According to Berzelius's former views (Essai, 

 p. 126), the atom of boron was 69'655: the experiments on 

 which the change is founded are not given. It is probable 

 they are of importance, as they have led him to such a curious 

 construction of boracic acid as Bo 9 6 ; but until the 4th 

 volume of his French edition of Chemistry comes out, in which 

 they will probably appear, we can hardly reckon upon them. 

 To compare his new number with that of Thomson, we must 

 take two-thirds of it = 93*988. 



Thomson makes fluoboric acid = 4*25; and fluoric = 1*25 

 (First Prin. i. 161) ; but if fluoric acid be taken at a mean of 

 their numbers 1*3, boracic acid will turn out 4'25 — 1*3 = 2*95 



* The exponent - is substituted for the bisecting line, to facilitate the 

 printing ; as the latter would require type on purpose. 



t It seems to have been regarded as a mere rule drawn from the com- 

 binations of protoxides with substances containing plural atoms of oxygen. 

 {Vide First Principles, and Turner's Chemistry.) But this leaves out of view 

 the numerous cases wherein a sesqui- or bin-oxide requires a propor- 

 tionate number of atoms of acid • as for instance, sulphate of iron, in which, 

 by the absorption of oxygen, two-thirds saturate the acid, and one-third 

 precipitates : as well as other cases still more complex. 



Y 2 instead 



