PACIFIC EXPLORATION: L. J. BRIGGS 
403 
Let h — hi = Ah represent the difference in the height of the column 
at the two stations. On substituting this value of hi in equation (2) we 
have 
g / g, = l-Ah/h (3) 
in which Ah represents simply the difference of the micrometer readings 
at the two stations, and does not involve the absolute height of the 
column. Since h is at least 200 times as large as Ah, the absolute 
height of the column does not, in this method, need to be determined 
with great precision. 
Discussion of errors. — The apparatus was designed with the object 
of obtaining, if possible, an accuracy of 1 part in 100,000 in the measure- 
ments. The discussion of the various sources of error will therefore 
be made on this basis. 
Temperature variation. — During the measurements the whole apparatus 
was surrounded with melting ice, so that the effect of slight variations 
in temperature need be considered only in connection with the tem- 
perature of the gas. Artificial ice was employed and all ice that was 
not perfectly clear and crystalline was discarded. The purity was 
systematically checked by measuring the electrical conductivity of the 
tap water. The impurities were never sufficient to depress the theoreti- 
cal freezing point more than 0?001C., whereas a variation of =tO?003C. 
would be necessary to produce a change of 1 in 100,000 in the gas pres- 
sure. The drip water was allowed to escape from the ice tank through 
a small trap near the bottom in order to insure the ice extending below 
the bottom of the gas chamber.^ 
Error in setting point in contact with mercury surface. — Lord Rayleigh^ 
found in his investigations with the micromanometer that the fixed 
ground glass point could be set in contact with the mercury surface 
with an error not exceeding ±0.0015 mm. An accuracy of 1 part in 
100,000 in gravity observations necessitates a probable error in the 
micrometer observations not greater than ±0.008 mm., which is readily 
attainable from a series of readings at sea under favorable conditions. 
The writer has found that in measurements at sea a metallic point is 
superior to a glass point, due to the fact that the latter becomes electri- 
fied through the motion of the mercury surface and this influences the 
readings. 
Error in the determination of the instrument constant. — Reference to 
equation (3) will show that since the height of the column is at least 
200 times the total range {Ah), the uncertainty in h may be 100 times 
that permissible in Ah. The constant of the instrument (i.e., the verti- 
cal distance from the lower mercury surface to the zero on the scale) 
