150 Pottox—On the Extraction of Glucinum from Beryl. 
as a sudden rush carries over specks of carbon and unaltered oxide, thus contami- 
nating the chloride ; it is difficult absolutely to avoid this source of error. 
The main difficulty was to rake out the chloride sufficiently expeditiously into 
perfectly dry, wide-mouthed bottles, as the anhydrous chloride is exceedingly 
hygroscopic ; a little care is also necessary to avoid raking out any of the carbon 
from the tube. In a tube 2 feet long and 12 in. internal diameter, I was able to 
operate on forty grams of oxide at a time, though, as a rule, about ten grams of 
ignited oxide were taken. For operating on small quantities, a Hoffman furnace 
and hard combustion-tube were used. 
Chlorides were prepared from a variety of samples of basic carbonate of 
glucinum that had been fractioned with ammonium sesquicarbonate, and from 
various samples of hydrate purified from the last traces of alumina by solution in 
hydrofluoric acid, and treatment with acid potassium fluoride. The actual yield 
of anhydrous chloride was taken by weighing the dry bottle before and after 
placing the charge in it, but a great part was often lost by absorption of moisture 
before it could be got out of the tube. In each case the carbonaceous residue was. 
removed by the other end of the tube, and ignited to find how much oxide remained 
unaltered. The following are the particulars of a few of the results obtained :— 
In one experiment 10 grams of glucinum oxide ignited with 20 grams of sugar 
gave 25 grams of anhydrous chloride, and 1 gram of unaltered residue. 
In another experiment 10 grams of pure oxide of glucinum mixed with 40 grams 
of sugar, and charred; then chlorinated in the porcelain tube, at a bright red heat, 
in a gas furnace, with a good stream of well-dried chlorine, gave 15-5 grams of 
anhydrous chloride. The residue gave, on ignition, 1:6 grams of oxide left 
unacted on—time, three hours. The chloride was in pure white crystals, with a 
grey powder intermixed, undoubtedly due to a little fine carbon, and possibly some 
of the oxide being carried over by the current of chlorine. A good part was 
unfortunately left in the tube and lost, or rather washed into the residue bottle. 
Of course, any chloride left in the tube was not allowed to mix with the car- 
bonaceous residue before ignition to determine the proportion of unaltered oxide. 
In a third experiment 13°5 grams of pure oxide, ignited with four times its 
weight of sugar, and, chlorinated as above, gave 23°75 grams of anhydrous 
chloride ; and the residue, on ignition, gave 3:6 grams of unacted-on oxide. The 
chloride in the centre of the tube was in beautiful tufts of needle-like crystals ; 
and round the tube were found crystalline crusts of the anhydrous chloride. 
Other experiments gave similar results. Theoretically 10 grams of anhydrous 
oxide of glucinum should yield about 32 grams of anhydrous chloride. 
A very careful analysis was made of the chloride from the last experiment. 
The principal difficulty in the analysis of this substance was to weigh accurately 
