In a quantitative examination made in 1885 of all the known methods of separating beryllium from aluminium and from iron, the various precipitates obtained were dissolved and diluted to a known volume corresponding with the amount of bases in solution.
The solutions were spectrographically examined, and the photo- graphs compared with others taken from solutions containing accu- rately -weighed quantities of pure beryllia. The coil used was capable of giving a 5-inch spark in air. In place of a Leyden jar a pane of glass coated on each side with a square foot of tinfoil was used. The electrodes* were Ceylon graphite as in other experiments, the sole impurity in which was a trace of magnesium.
The following tabulated statement gives the wave-lengths of the lines, together with a description of the spectra photographed from solutions of different strengths (see next page).
The actual length of the line 2478*1, as rendered by solutions of 0-00001 per cent, and O'OOOOOl per cent, strength, is, in the former, 0'07, and, in the latter, 0*05 of an inch. The normal length of the line at this part of the spectrum is 0'22 of an inch. The quantity of sub- stance yielding this spectrum is equivalent to one-millionth of a milli- gramme of beryllium. As I have pointed out in the case of magnesium,! so also is it with beryllium, that the sensitiveness of the spectrum reaction may be increased ten thousand-fold by using a larger coil and more powerful condenser, but leaving the striking distance between the electrodes unaltered. The coefficient of complete extinction was there- fore practically not attainable for all the lines, or, in other words, the sensitiveness of the reaction is almost without limit.
It will also be seen from my description of the spectra, which have been quite recently re-examined, that the coefficient of extinction of the two lines A 3130'3 and 2478'! had not been reached by the dilution specified.
A number of thin sections of the Dublin granite containing microscopic crystals of hexagonal form were examined some years ago. The crystals were supposed to be apatite, but a very carefully executed analysis disclosed the fact that the proportion of phosphoric acid contained in 20 grammes of the rock was almost infinitesimal, and that on warming thin sections, in which the hexagonal crystals were visible, with nitric acid and ammonium molybdate, no deposit of yellow
- ' Phil. Trans.,' 1884, Part I, p. 49.
t ' Phil. Trans.,' 1884, Part II, p. 325.
