OF THE ATOMIC WEIGHT OF ALUMINUM. 
1005 
ture, when it might be assumed that pure alumina alone was left. As Berzelius 
himself says, ignited alumina rapidly absorbs moisture from the air, involving risk of 
error in determining its weight. Traces of the light pulverulent alumina are liable to 
be mechanically carried away during the decomposition of the sulphate. It is observ¬ 
able that all these sources of error, except the last, tend in the same direction, to make 
the atomic weight of aluminum come out too high. 
2. Experiments of Sir Humphry Davy, 1812.—In Sir Humphry Davy’s ‘ Elements 
of Chemical Philosophy’—published in 1812, the same year in which Berzelius’ first 
paper on this subject appeared—it is stated 4 ' that no direct researches had then been 
made on the quantity of oxygen in alumina, but that, from some experiments by the 
author on the quantity of ammonia required to decompose saturated solutions of 
alumina in acids, “it would appear that the number representing alumina is about 48, 
and, supposing it to consist of one proportion of aluminum and one of oxygen, 33 will 
be the number representing aluminum.” The details of the experiments in question 
are not given, and the combining proportions of all substances having been very imper¬ 
fectly known at the time—the number 15 is taken above for oxygen—it is needless 
to say that this passage throws no light upon the exact atomic weight of aluminum. 
3. Experiments of Thomson, 1825.— Thomson! attempted to deduce the number 
representing this atomic weight from, (a) analyses by himself and others of sundry 
natural aluminous silicates, ( b ) analyses of potassium alum, and (c) analyses of hydrates 
of aluminum. He concluded from all his experiments that the true number for 
alumina is 2’25 (0 = 1), and, taking alumina to be AlO, he made Al=l - 25. This cor¬ 
responds to Al=30, if 0 be assumed =16, and alumina Ah0 3 —a result which can 
only be viewed as a rough approximation to the truth, since Thomson’s methods were 
far from accurate, and his experimental results agree but poorly with each other. 
4. Experiments of Mather, 1835.—W. W. Mather,^ Assistant Professor of Che¬ 
mistry at the United States’ Military Academy, West Point, prepared anhydrous 
aluminum chloride by Wohler’s process, dissolved a weighed portion of it in water, 
added silver nitrate in excess, filtered off, dried and weighed the silver chloride formed, 
threw down excess of silver from the filtrate by hydrochloric acid, filtered again, 
evaporated this second filtrate and washings to dryness, ignited the residue, and 
weighed it as alumina. "646 grm. of aluminum chloride gave 2'054973 1§ grins, of 
silver chloride (yielding on reduction 1’548161 grm. of silver) and * * * § 2975 grm. of 
alumina. 
* Vol. iv., p. 263, of the ‘ Collected Works of Sir H. Davy,’ edited by his brother Dr. John Dayt ; 
London,1840. 
t Thomas Thomson, M.D., ‘ An Attempt to Establish the First Principles of Chemistry by Experiment,’ 
vol. i., p. 285; London, 1825. 
J Sillihan’s ‘American Journal of Science and Arts,’ xxvii. (1835), 241. 
§ The seven decimal places are given, notwithstanding the statement by the author himself that his 
balance could weigh easily m grain, and was sensible to yoo grain ! 
