1006 
PROFESSOR J. W. MALLET OR A REVISION 
From the amount of silver chloride found and silver obtained from it in this one 
experiment, and from the atomic weights of silver and chlorine adopted by Berzelius 
and Thomson respectively, Mather calculated values for the atomic weight of aluminum 
ranging from 1‘82274 to 1 ‘85430 (0=1, and the formula of the chloride being taken 
as AlClg), or 29‘16384 to 29‘66880 for 0=16; but from the amount of alumina 
obtained and the amount of aluminum therein (the latter deduced from the chloride 
taken for analysis minus the chlorine found), he calculated the atomic weight for 
aluminum as 1‘3188017 (0=1) for alumina taken as A1,0 3 , or 21T008272 for 0=16. 
He does not seem to have been struck by the evidence of some error in his own work 
which these discrepant numbers afford, but suggested that the figures given by Ber¬ 
zelius for the aluminum and oxygen in alumina might have been accidentally inverted, 
which would explain the disagreement between himself and the great Swedish chemist. 
In reality it is pretty plain that Mather’s alumina was not pure, either from fixed 
matter of some kind left behind from the acids and wash water used, or from absorp¬ 
tion of moisture before weighing. If his most direct result only be taken as the basis 
of calculation, namely, the weight of aluminum chloride used and silver chloride 
obtained from it, using Stas’ numbers for chlorine (35‘37) and silver (107‘66), the 
atomic weight of aluminum found will be 2 8‘77 8 for the formula A1C1 3 . 
5. Experiments of Mallet, 1857.—In 1857 the writer of this paper attempted to 
use metallic aluminum, which had not long before begun to be manufactured and 
sold, for the determination of the atomic weight. At the meeting of the British 
Association held in that year at Dublin, 4 ' he gave a brief account of his experiments, 
which had been made with the metal of commerce, containing, as he found, only from 
93 to 96 per cent, of pure aluminum. The exact nature and amount of the foreign 
substances present, chiefly iron and silicon, having been determined, the crude metal 
was dissolved in hydrochloric acid, the solution precipitated by ammonia, and from the 
amount of alumina left from the precipitate on ignition, after allowing for the impuri¬ 
ties, the atomic weight was deduced. The results obtained from a few experiments 
were not satisfactory enough to warrant any proposal to modify the then received 
number. The probability that this number needed correction was, however, pointed 
out, with reasons for such an opinion; the desirability of obtaining for the purpose of 
new experiments really pure metallic aluminum was noticed; and it was suggested 
that difficulties connected with the accurate determination of alumina by the method 
which had just been tried might make it eligible to determine instead the hydrogen 
given off during the solution of the metal in acid. 
6. Experiments of Dumas, 1858.— Dumas! redetermined the atomic weight in 
question by dissolving in water known weights of aluminum chloride, and ascertaining 
the quantity of silver, used as nitrate, which was required in each case for precipita¬ 
tion of the chlorine. The aluminum chloride had been carefully prepared on the large 
* ‘ Report of British Association Meeting at Dublin, 1857 : Transactions of Sections,’ p. 53. 
t ‘ Annales de Chimie et de Physique’ [3], lv. (1859), p. 151. 
