400 
PACIFIC EXPLORATION: L. J. BRIGGS 
ner and Dickinson's measurements is about =±=0?003C. The measure- 
ments were carried out with a refinement which is probably unattain- 
able at sea, and may be taken to represent the limit of accuracy attain- 
able in such determinations on board ship. We may now consider the 
effect of such an error in the determination of g. A probable error of 
=i=0?003C. in the true boiling point temperature would correspond in 
atmospheric pressure to ±0.083 mm. of mercury and to a probable 
error of =±=0.11 cm. per sec. per sec. in the value of g. The uncertain- 
ties due to the boiling-point measurements alone 
under the most favorable conditions would, there- 
fore, result in a probable error of 1 part in 10,000 
in the value of g. This is 13 times the limit of 
error set by Hayford for ocean gravity measure- 
ments. To this must be added the errors in the 
barometric observations. Errors in the boiling- 
point temperature alone would justify Bowie's 
statement that Hecker's gravity measurements at 
sea "are subject to uncertainties as large as the 
largest of the new-method anomalies of gravity 
in the United States, that is, between 0.05 dyne 
and 0.10 dyne." 
Description of the New Method. — The new method 
of measuring g at sea described in this paper is 
based upon the change in height of a barometric 
column sealed off from communication with the 
atmosphere.^ All boiling-point determinations are 
thus avoided. The apparatus is kept at constant 
temperature in a bath of melting ice. The deter- 
mination only of the position of the upper end of 
the barometric column is necessary. The design of 
the instrument is such that in making this setting 
the enclosed gas mass is automatically reduced to a 
constant volume; and since temperature is constant, 
the measurements are made at constant pressure. 
A sketch of the glass part of the apparatus is shown in figure 1. The 
mercurial column is contained in the capillary c (bore 0.6-0.7 mm.) the 
lower end of which opens beneath mercury in the bottom of the gas cham- 
ber d. This capillary is sealed to the wall of the gas-chamber where it 
passes through the upper end. The upper part of the capillary is bent 
into a flexible zig-zag, and expands into the spherical bulb h (diameter 
2 cm.) The bulb contains a fixed iron point p sealed to the inside of the 
