1004 
PROFESSOR J. W. MALLET ON A REVISION 
parts of oxygen (the atomic weight of sulphur being then taken = 20176—0 = 100), 
100 parts of alumina must include 46726 parts of oxygen. This last number should 
be 467207. In a later paper'" by the same author, it is calculated from the results of the 
above experiment, that 100 parts of “sulphuric acid” are saturated by 427227 parts 
of alumina, and that the earth contains 467047 per cent, of oxygen; the earth is 
assumed to be Ah0 3 , and consequently the atomic weight of A1 is found =171'667 
(0=100), or 2779 (H=l). These numbers, correctly calculated from the percentage 
of oxygen taken in this second paper, should read 171767 and 27'39 respectively. 
Or, if the percentage of oxygen of the former paper, corrected as above, be taken as 
467207, the atomic weight, for Ab0 3 , will be 171 '057 (0=100) or 27’37 (H=l).t 
Finally, if Berzelius’ direct results of experiment be taken, and recalculated with 
Stas’ atomic weights for O (15’96) and S (3T98), the atomic weight of aluminum, 
its oxide being assumed AhO s , will be 27'237 in reference to that of hydrogen as 
unity. 
Berzelius^ a ^ s0 attempted to obtain a pure hydrate of aluminum by precipitating 
the sulphate and nitrate with ammonia, but found that highly basic salts only were 
thrown down. Using the chloride instead, a hydrate was obtained which, after being 
dried in the sun, gave only water on heating, but there was a little loss from mechani- 
cally-carried-over alumina. This sun-dried hydrate left 64‘932 per cent, of alumina 
free from acid. Berzelius therefore calculates that 100 parts of anhydrous alumina 
had been combined with 54 parts of water—this amount of water containing 47'65 
parts of oxygen; while the alumina contains, as shown by the above-quoted analysis 
of the sulphate, 46726 parts of oxygen. He remarks: “I cannot affirm that either 
the determination of the amount of water or that of the oxygen in the alumina is 
sufficiently exact; both are, however, so far so as to sufficiently show us that alumina, 
like the preceding bases, combines with an amount of water whose oxygen is equal to 
that of the earth itself.” 
In the decomposition by heat of aluminum sulphate, as thus used to furnish data 
from which to calculate the atomic weight of the metal, the following possible sources 
of error may be noticed :—The hydrate precipitated by ammonia from a solution of 
(presumably) potash alum might carry down with it traces of fixed alkali, and this 
latter be retained in the sulphate afterwards prepared from the hydrate. The ten¬ 
dency of aluminum to form basic salts suggests the possibility of traces of sulphuric 
acid being lost in the preliminary drying over the simple alcohol lamp, even at a tem¬ 
perature at which possibly the last traces of water might not have been removed. I 
have found from my own experiments that a trace of basic sulphate may, on the other 
hand, be obstinately retained even after prolonged exposure to a very high tempera- 
* Poggendorff’s ‘ Annalen der Physik u. Ckimie,’ viii. (1826), 187. 
f A. C. Oudemans, Jr. (in his ‘ Historisck-kritisch Overzigt van de Bepaling der kEquivalent-Gewigten 
van twee en twintig MetalenLeiden, 1853), calculates, from Berzelius’ figures, Al=171 - 02. 
J Gilbert’s ‘Annalen,’ loc. cit. 
