44 JOURNAL OF THE 
SOURCES OF ERROR. 
Five main sources of error have occurred to me apart 
from any question as to the purity of the material. 
1. The deliquescence of the chloride. 
2. The loss of the finely divided zirconia. 
3. The corrosion of the platinum crucibles. | 
4. The attack of the glass flasks by the gaseous and 
aqueous hydrochloricacid. Itis very evident that the glass 
vessels would suffer when subjected to the action of hy- 
drochloric acid during such prolonged periods. The 
weighings easily revealed the extent of this action. It 
was found to vary. with different flasks. For instance, 
flask I lost 0.00116, 0.00031, 0.00023, and 0.00078 gram. 
On the other hand, flask IV weighed after the first ex- 
periment only 0.00011 gram less, and after the next two 
experiments the loss was 0.00013, and the total loss during 
three experiments was 0.00024. This is probably trans- 
ferred mainly as alkaline chlorides to the crucibles and is 
there volatilized. Sodium could readily be detected upon 
the lid of the crucible after partial ignition. Of course 
if all is volatilized then no error is introdued into the ex- 
periment, but there may be some non-volatile material 
also transferred from the flasks. The total amount is so 
small that the error cannot be a large one. 
5. Substances carried into. the drying flasks by the 
stream of gas. Much care was taken to avoid error from 
this source. ‘The main danger lay in the necessity for 
the use of some rubber connections. Vulcanized rubber 
was used. It was freed from excess of sulphur and was 
replaced by a fresh piece when sttacked by the gas. The 
dry gas does. not attack rubber very rapidly. Gaseous 
sulphur compounds, coming from the sulphuric acid, used 
in preparing and drying the hydrogen chloride, were 
probably carried through the drying flasks, but there 
seems to be no probability of their causing a decomposi- 
tion of the oxychloride. 7 
