174 E. W. Morley—Oxygen in the Air. 
degree, and a greater number waste time and possess no advan- 
tage whatever. 
The eudiometer was calibrated by filling it with air-free 
water, and weighing the quantity expelled as the mercury rose 
to each successive mark of the graduation. This was done 
most favorable conditions, and its tension determined ; it was 
then brought to the other division, and its tension again deter- 
mined. Two independent measurements thus obtained elimi- 
nated the chance of error in identifying divisions on the scale, 
and also affor the means of ascertaining the probable 
error of a measurement. In the analyses contained in this 
has been computed that the probable error of a single determ1- 
nation of volume is its 5800th part. Hence the probable error 
of a determination of oxygen in the air is the 7200th part, and 
the probable difference of two determinations on the same sam- 
ple is the 5100th part. A second analysis was always made 
when the first showed a deficiency of oxygen. A comparison 
of the results will show whether the accuracy indicated by 
computation was obtained.* The writer has in hand an entirely 
new construction of the pressure tube, and some modifications 
of the optical appliances for reading the level of the mercury 
in the eudiometer tube, by which he hopes considerably to 
lessen this probable error. 
The samples of air analyzed were collected in the open 
country in glass vessels with due care as to admixture with the 
air from the collector's lun 
_* The divergence of the second resule of February 26th from the first is due to 
the fact that in the second analysis the hydrogen used was not pure. —— 
_of the sample remained for a thi i e second It is given in con- 
firmation of the first. But this pair of results should not be used in computing 
ee eee 
