OF ARTS AND SCIENCES. 361 



tube in which the reduction was effected. It is stated by Stas,* that 

 " all kinds of glass, when long exposed to a red heat in contact with 

 the flame of alcohol or of gas, lose slowly, but continuously, in weight." 

 This loss would have apparently increased the amount of sulphur in 

 the sulphide, and so have led to a too low valuation of the atomic 

 weight. That this is not the only constant error is shown by the fact 

 that those determinations gave the highest equivalent in which, the 

 quantity of substance being the largest, the glass was longest exposed 

 to the action of the flame. 



Secondly, these experiments were made with a native sulphide of 

 antimony ; their success depended upon its purity, and upon its having 

 the precise atomic constitution attributed to it. On the first of these 

 points we have the opinion of Berzelius, that " native compounds are 

 never sufficiently pure to be used in such researches." f We know by 

 the experience of Dumas, of Erdmann and Marchand, and of Scheerer 

 in the determination of the atomic weights of calcium and magnesium, 

 how difficult it is to detect and make allowance for the presence of 

 minute quantities of foreign bodies ; and every one who has worked 

 with antimony will admit that the difficulty in this ca^ is still greater. 

 On the subject of the precise atomic constitution of these native com- 

 pounds I have long had doubts ; and the paper which I recently had the 

 honor of submitting to the Academy shows that they may not be with- 

 out foundation, even when the body can be obtained in well-defined 

 crystals. 



Whatever may be thought of these doubts, which it is indeed easy 

 to raise about the best-executed work, the fact of the dependence of 

 the atomic weight in Schneider's experiments upon the quantity of 

 matter operated upon, seems to me sufficient to destroy confidence in 

 his results. I have, accordingly, not calculated the probable error of 

 his determinations ; his extreme numbers are 120.08 and 120.53, the 

 mean of eight being 120.3. 



Professor Rose's work I was published rather as a confirmation of 



* " Sans exception aucune, tout verre chauffe longtemps au rouge dans la 

 flamme de ralcool ou du gaz eclairant, diminue de poids lentement, mais constam- 

 ment." — Recherches sur les Eapports Reciproques des Poids Atomiques, (Brux- 

 elles, I860,) p. 17. 



t "Les composes naturals ne sont jamais assez purs pour etre propres a des pa- 

 reilles recherches." — Traite de Chimie, (Paris, 1847,) Tom. IV. p. 529. 



t Pogg. Ann., Bd. XCVIII. S. 455. 



VOL. V. 46 



