OF AKTS AND SCIENCES. 367 



error of the individual observations is 0.1, or twice that of my own 

 determinations. No account is given of the process by which the 



antimony ; d was formed by a second fusion of c with antimonic acid ; e by fusion 

 in a current of hydrogen. 



The probable value of the atomic weight deduced from the ten observations by the 

 method of least squares is 1529.2 for O = 100, or 122.336 for H = 1 ; with a prob- 

 able error, in the individual determinations, of 0.6 in the one case, and of 0.048 in 

 the other. This error corresponds to an error in the operation somewhat less than 

 0.0003 gramme. 



For the composition of the antimoniate of oxide of antimony the probable value 



is 



Sb 79.266 



* 4 20.734 



100. 



It remains now to be shown, that in the antimoniate of oxide of antimony the 

 antimony and oxigen exist in the exact proportion of 1 atom to 4, as has been 

 assumed in the foregoing calculations. 



This composition was inferred by Berzelius from his experiments upon the sul- 

 phide ; it seemed to me desirable that it should be proved by direct examination of 

 the oxides. For this purpose the oxide precipitated from chloride of antimony by 

 carbonate of potash or soda, which always retains traces of the alkali, was placed in 

 a small platinum boat in a combustion tube, and a stream of dry carbonic acid, gen- 

 erated in a self-regulating apparatus, led through the tube, till the air was completely 

 expelled. The part of the tube in which the oxide lay being heated by a lamp- 

 furnace, the oxide melted, and sublimed in beautiful and perfect crystals, shooting 

 across and filling up the tube. 1.52045 gr. of these crystals, which were proved to 

 be entirely free from any higher oxide of antimony, gave by oxidation with nitric acid 

 1.60385 gr. Sb04 : consequently 100 SbO* may be regarded as containing 94.8 

 SbOa. We have before found that 100 Sb04 contain 79.266 Sb ; hence the relation 

 of the oxygen of SbO^ to that of SbOg is that of 100 — 79.266 to 94.8 — 79.266, 

 or of 4.004 to 3.* 



It may be stated, then, as the result of this work : — 



1. That oxide of antimony cannot be determined quantitatively, as has been 

 asserted, by means of the double chloride of gold and potassium. 



2. That solution of chloride of antimony, after having been exposed to the action 

 of the air, will be found to contain a higher degree of oxidation of the metal. 



3. That the relation of the oxygen in the oxide, and in the antimoniate of the 

 oxide of antimony, as shown by direct analysis of these oxides, is that of 3 : 4 ; as 

 has been hitherto assumed. 



4. That the probable value of the atomic weight deduced from ten observations 

 is 122.34 ; the atomic weight of hydrogen being taken as unity. 



* If, conversely, the above experiment be used for the determination of the atomic weight, it 

 gives the number 121.84. But it is evidently unsuited to secure any degeee of accuracy ; for we 

 have only to admit an error in the operation of 0.00025, that is, less than the probable error of 

 the former observations, to derive from it the same number, 122.34. 



