In the course of a paper by Professor Humpidge on the above subject, recently read before the Society, the author seeks to decide between the atomic weight 9·2 for beryllium, resulting from my comparison of the atomic heat of the element with that of silver and aluminium, and the value 13·8, arrived at by MM. Nilson and Pettersson by determination of specific heat.J The difference between the two possible atomic weights is so small, and the difficulties met with in attempting to prepare even a few decigrams of beryllium are so great, that both sets of experiments have been objected to on the ground, amongst others, that the metal employed was in all cases impure. My specimen admittedly contained a minute quantity of platinum, and the Proportion of known impurity in one of MM. Wilson and Pettersson's specimens reached 13 per cent. Unfortunately, Professor Humpidge's metal though claimed to be the purest yet prepared, is shown by analysis to be rather less pure than one of the specimens employed by Nilson and Pettersson, hence the experiments lately made known to the Society do not carry the inquiry beyond the point previously reached, save in one noteworthy particular, namely, that there appears to be a considerable, though irregular, rise in specific heat of the element as the proportion of impurity diminishes; but the value is still much below that required for the atomic weight 9·2. Thus for a specimen of beryllium which contained 13 per cent. of
known
of impurity Wilson and Pettersson obtained the specific heat 0·4084 between 0° and 100° C., and for a less impure specimen 0·425; while Professor Humpidge, in one of his experiments with a material that contained 6 per cent, of impurity, found the specific heat to be nearly 0·45 (0·4497). In all these cases corrections were applied which were believed to eliminate the effects due to the impurities known to be present—in part mechanically mixed with the metal and partly alloyed with it.
III. Preliminary note on some aluminium compounds