54 
PUTNAM. 
indicating apparently a considerable advantage for the latter. 
It is probable that no particular significance attaches to the 
individual residuals remaining after this application of 
Faye’s reduction for several reasons. Other values would 
certainly result if a different area were considered in esti¬ 
mating the average surrounding elevation or if weight were 
given according to proximity to the station in making this 
estimation or allowance made for variations in density, and 
further, the average elevations as given can be considered 
only roughly approximate, as they were calculated from a 
small scale map. An inspection of the table shows that 
the residuals with Bouguer’s reduction are very nearly pro¬ 
portional to the average elevations surrounding each station, 
though bearing no relation to the actual station elevations. 
As furnishing further evidence on this subject, a prelimi¬ 
nary study has been made of several former series of gravity 
determinations based on Washington and carried out by 
Doctor Mendenhall, Mr. Preston, and Mr. Smith in various 
parts of the world. As these include a number of oceanic 
island and coast stations, it was necessary, in applying Faye’s 
idea of reduction, to take account of the difference in density 
between sea water and land. This was done by assuming the 
water as condensed down until its density equaled that of 
land and using the upper surface of this condensed mass in 
estimating the average elevation of the surrounding region. 
This is equivalent to replacing the water by an equal weight 
of land—a proceeding which would not alter the pressure on 
the crust below. In application, the depth of sea water at 
any point multiplied by ^ 1 — j'j , where \ is density of sea 
water and 8 is density of land beneath the station, gives the 
depression below sea level of the condensed surface. By 
taking these depressions as negative altitudes the average 
elevation may be conveniently estimated for all stations 
surrounded either wholly or in part by water, the principal 
difficult}^ being the lack of sufficient data as to ocean depths 
and land elevations and the uncertainty in regard to the 
area to be considered. A similar treatment could be applied 
