1008 
PROFESSOR J. W. MALLET ON A REVISION 
I.- 
-Atomic weight of A1 . 
. =27-447 
II. 
3 3 3 3 
53 * * * 
. =27-696 
III. 
5 5 3 5 
33 ... 
. =27-435 
IV. 
5 5 55 
33 ... 
. =27-318 
V. 
55 55 
33 ... 
. =27-522 
VI. 
55 35 
33 ... 
. =27-327 
VII. 
5 3 3 3 
33 ... 
. =27-489 
Mean 
. . 27-462 
It is to be remarked that the tendency of the chief causes of error connected with 
these experiments is in the same general direction. The presence of any iron in the 
chloride used ; the action upon it, though but to a minute extent, of any trace 
of moisture in the hydrogen in which it was finally sublimed ; any loss occurring with 
the fumes formed on introduction of the chloride into water; and the retention of 
traces of silver chloride in solution in the liquid from which the main mass of this 
compound had been thrown down; any or all of these would tend to diminish the 
quantity of silver chloride obtained, and therefore to make the atomic weight of 
aluminum appear greater than it really is. In discussing results which we owe to 
the labours of such experimenters as Berzelius and Dumas, it is of course to sources 
of error likely to inhere in the method itself that attention should especially be given. 
Dumas also tried dissolving aluminum (containing iron and silicon in considerable 
quantity) in hydrochloric acid, adding nitric acid in excess, evaporating to dryness, 
igniting and weighing the alumina (and other oxides) left behind. From an analysis 
of the crude metal employed, so as to allow for the impurities present, and on the basis 
of three experiments made as above, he calculated the atomic weight as 
1374 (=27-48 for Ah0 3 —0=16) 
13-87 (=27-74 „ „ — „ ) 
13-89 (=27-78 „ „ — „ ) 
but he was dissatisfied with these results, having found that the impurities of the 
metallic aluminum were unequally distributed throughout its mass, and having been 
unable to obtain the metal in a pure state. He considered the results furnished by 
the experiments with the chloride as accurate, and concluded that the atomic weight 
of A1 is represented by the number 13"75 (for A1^C1 3 )—this becomes 27"5 for A1C1 3 or 
AhCl 6 . 
7. Experiments of Tissier, 1858. — Ch. Tissier* prepared aluminum by reducing 
very pure fluoride of aluminum and sodium—probably cryolite, although this is not 
stated—by means of purified sodium in a carbon crucible, re-fusing the metal several 
times in order to free it from any of the flux which might have been retained. No 
* ‘ Comptes Rendus des Seances de l’Acad. des Sciences,’ xlvi. (1858), p. 1105. 
