OF THE ATOMIC WEIGHT OF ALUMINUM. 
1011 
either admitted by the authors themselves or shown by an examination of their 
methods and the inconsistency of their conclusions. Our knowledge of the atomic 
weight under consideration rests therefore upon the investigations of Berzelius, 
Tissier, and Terreil, each of whom made one experiment, and those of Dumas, who 
made seven. 
General results of former determinations. —If all the results be taken as I have 
re-calculated them, using Stas’ atomic weights for the other elements concerned, and 
equal value be given to all, we shall have the following arithmetic mean— 
Berzelius 
Dumas . 
55 
55 
Tissier 
Terreil 
27-237 
27-447 
27-696 
27-435 
27-318 
27-522 
27-327 
27-489 
27-068 
27-030 
Mean . .. 27-357 
If, however, the results of Dumas, all depending on repetition of the same process, 
be viewed as possibly affected by some constant error, and be thrown together, taking 
into the calculation the mean only of his seven experiments, the general mean will be— 
Berzelius. 27 "237 
Dumas (mean). 27-462 
Tissier. 27 - 068 
Terreil. 27*030 
Mean.27-199 
If we separate Dumas’ results, and take the mean of the other three, we get in 
contrast— 
Berzelius. 27-237 
Tissier. 27-068 
Terreil. 27-030 
Mean.27T12 
Dumas (mean). . . . 27"462 
6 o 2 
