396 
PACIFIC EXPLORATION: J. F. HAYFORD 
rapidly at very widely scattered stations that the new observations 
made within a single year may furnish a more accurate value of the 
flattening of the earth than has yet been obtained from all the work 
of the past. Let us make this more definite and concrete. Assume 
that it becomes possible to take a series of observations on a moving 
ship which will determine the force of gravitation at the point of obser- 
vation with a probable error of ±0.008 dyne (about 4 times the probable 
error of a land determination). Past experience indicates that the 
probable error in one such result due to all causes, including the anomal- 
ous part of the distribution of densities beneath the surface, will be less 
than ±0.020 dyne. Six hundred such observations could be secured 
in a single year scattered from latitude 60°N. to latitude 60°S. From 
these observations alone the value of the polar flattening of the earth 
could be computed more accurately than it has yet been computed. 
Of course if the best observations that can be made at sea are of less 
accuracy than this their value will be less. 
I am making this statement on the assumption that all such obser- 
vations would be corrected for topography and isostatic compensation 
by the method now in use in the Coast and Geodetic Survey. Such 
corrections are essential to reliability and serve to increase greatly 
the accuracy of the computation of the flattening. I have just compared 
a very recent computation^ by F. R. Helmert of the flattening of the 
earth with an older computation^ of the flattening by William Bowie. 
Both used gravity determinations. Helmert used 700 widely scattered 
selected observations from among the 3000 available over the whole 
world. Bowie used 122 out of the 124 gravity observations in the 
United States alone. Helmert made no corrections for topography and 
isostatic compensation. Bowie applied such corrections. I am con- 
vinced from a study of the evidence, including the evidence of systematic 
errors, that Bowie's value of the flattening derived from 122 gravity 
observations in a smafl region is more accurate and reliable than 
Helmert's value from 700 carefully selected and widely scattered 
observations. 
Observed values of gravity in the United States after correction for 
topography and isostatic compensation show no relation to the topog- 
raphy. On the other hand without such corrections, as Helmert's 
recent investigation again shows clearly, observations along the coasts 
stand in a class by themselves, observations in low interior regions in 
another class, those in mountainous regions in another class, and those 
on small oceanic islands in still another class, and each class is subject 
to its own peculiar systematic errors which are large. 
