their Determination in general. 375 



last few years, only gives too much support to this view. The 

 equivalent of antimony was until quite lately, as I showed in 

 1856, taken 8 entire hydrogen equivalents too high. The deter- 

 mination of the equivalent of tantalum, whatever care was de- 

 voted to it, could not be exact, as long as we knew nothing of 

 the existence of niobium. The number found by Berzelius, 

 92'2 (or 1152'87 for = 100), was necessarily inexact, for this 

 reason. This number, however, as H. Rose * has lately shown, 

 contained a still greater error, introduced by the supposition of 

 Ecrzelius that sulphide of tantalum and tantalic acid had the 

 same constitution. From the composition of the chloride of 

 tantalum, II. Rose has now determined the equivalent of tan- 

 talum as 68'8 (that is, approximately, 69). 



Dumas appears to have taken no notice of this ; he puts f the 

 equivalent of tantalum = 92 (or 93) in the same series as that 

 of tin (56) and titanium (25), and finds in the common differ- 

 ence of 34 which these numbers show, the law which connects 

 them. That which here appears as a law is clearly only an ac- 

 cident. The number which apparently accords with the law, lies 

 24 entire units away from the truth. 



In such instances, which might be easily multiplied, where 

 there are individual departures from the general law, there is 

 sufficient reason not to find at once new laws, but to attribute at 

 first something to our limited knowledge. As long as we are 

 liable to commit lai'ge errors, it seems to me to be an unsafe 

 undertaking to establish laws which assume the most accurate 

 knowledge of equivalent numbers up to 9. fourth of a unit. 



From this point of view, even the assumed regularity first 

 poiuted out by Maumene, and afterwards fux'ther developed by 

 Dumas, according to vt^hich a great number of equivalents are 

 multiples of half an equivalent of hydrogen, rests upon insuffi- 

 cient foundation. 



I must agree fully with Strecker J, that most equivalent deter- 

 minations are not yet exact to 0*25 (forH = l), which would have 

 to be the case if this regularity had to be established as a general 

 truth. 



Of the seven equivalents v/hich Dumas formerly § arranged 

 in this group (CI = 35-5; Mg=12-5; Mn = 27-5 ; Ba = 68-5 ; 

 Ni = 29-rj; Co = 29-5; rb= 103-5), that of magnesium (12-5) 

 hjis already had to be retracted. It docs not speak much for the 

 law in question, that it is decidedly inapplicable for the lowest 



* Ber. tier Berl. Acad. 1B56, p. 385. Ann. der Chetn. und Pharm. 

 vol. c. p. 24(>. 



t Comptes Rendus, vol. xlv. p. 730. 

 X Strcckei', Iluudworlerljuch, vol. ii. p. 51C, 

 § Comptes Rendus, vol. xlvi. p. 951. 

 T2 



