6 ROYAL SOCIETY OF CANADA 
Working in this way, Mr. J. H. Pratt obtained the following results, 
employing ferrous sulphate : ' 
Time. FeO taken. FeO found. Error. 
lOMINULES EEE RENE ‘0543 “0825 ‘0018 — 
10 Ba OS A PRE VLC ‘0594 ‘0578 : ‘0016 — 
10 MA DR nr ‘0739 ‘0728 ‘0011 — 
When very accurate determinations are desired Pratt recommends 
the use of a platinum crucible with perforated lid and platinum tube, 
arranged as the apparatus of H. Rose for the ignition of precipitates in 
hydrogen. Here the air can readily be displaced by carbon dioxide and 
the crucible heated directly over a small flame. By heating in this way 
the operation is much more rapid than when the water-bath is used, and 
the decomposition of difficultly attackable silicates much more easily ac- 
complished. Pratt found in experiments with black tourmaline that 
nearly as much of the mineral was dissolved by boiling for ten minutes 
over a flame as by heating quietly for two hours over the water-bath.” 
In cases where the silicate is easily decomposed I have used a modification 
of the ordinary apparatus which is very convenient. The bottom is eut 
off from a three-necked Woulf’s bottle, which is then stood in a beaker of 
slightly greater diameter than the bottle and containing a little water. 
The beaker is supported on a water-bath. Two of the necks of the bottle 
—provided with corks and tubes—are utilised for the ingress and egress 
of a stream of pure carbon dioxide, while the third is for the introduction 
of a platinum stirring rod which moves freely through a cork in the 
central neck and reaches to the bottom of the platinum crucible contain- 
ing the mineral for analysis and supported by a small stand made of lead 
or other suitable material. With this apparatus there is, of course, no 
danger of oxidation and the contents of the crucible can be readily stirred 
from time to time—an advantage especially in cases where the mineral 

is liable to cake. ‘ 
The estimation of water in minerals (more particularly silicates) is a 
matter which deserves careful attention, and careful determinations of 
this constituent will no doubt do much to elucidate the constitution of 
many imperfectly understood species. The loss which takes place when 
a silicate is ignited may include not only water (possibly hydroxyl) in 
different states of combination, but also carbon dioxide and fluorine ; and 
in cases where much ferrous oxide is present any value which the deter- 
mination may have is further diminished, more or less of the ferrous oxide 
being converted into ferric, so that in some cases there is a gain instead of 
loss on ignition, even when water and other volatile constituents are present. 


' Am. Jour. Sci., Aug., 1894. 
* See also in this connection a paper by Chester and Cairns, Am. Jour. Sci., Aug. 
1887. 
