162 Mr. Galbraith on the Figure of the Earth. 
would, in part at least, explain the irregularity in question ; 
since at Formentera the pendulum was not much above the 
level of the sea, in the interior of France it was considerably 
elevated, and again at Dunkirk it descended nearly to the level 
of the ocean. Now it is obvious, that if the experimental re- 
sults were not properly reduced to what they would have been 
at the level of the sea, the configuration of the country would 
have, generally speaking, produced the irregularity we have 
endeavoured to point out. May net this also have its effect 
on the compression derived from the measurement of arcs ? 
It is probable it would to a certain extent. Indeed it appears 
certain, that irregularities of this nature must be expected, un- 
less regularity of ground, and similarity of geological charac- 
ter, be selected for either of these series of experiments; and 
this, it must be granted, cannot easily be obtained, though it 
should be attended to as far as circumstances will permit. The 
truth of these remarks will be obvious from a comparison of 
Captain Sabine’s observations at St. Thomas and Ascension, 
with those at Maranham and Trinidad. 
Captain Sabine has combined all his own observations with 
some of those of Captain Kater and of the French, and allows 
. each to have its proper share in determining the coefficients 
of the theoretic formula, as well as the compression; and in 
general this method is to be recommended, where solid objec- 
tions cannot be made to some particular observations. Now 
in the present case, we think strong objections may be urged 
against some of them; particularly those on basaltic or vol- 
canic bases, as those at St. Thomas, Ascension, Galapagos, 
&c. being combined with others on alluvial soils. 
It is true we have three determinations of the length of the 
pendulum, when nearly on the equator: one at Galapagos, one 
at St. Thomas, and another at Java; though the temperature 
at which this last was determined is not mentioned in the source 
whence we obtained it. A mean of all these would give about 
39:02 inches for the length of the equatorial pendulum ; but, 
unfortunately, two of them at least were obtained on rocky 
bases, and may therefore be considerably more than when de- 
termined on a basis of an ordinary state of geological charac- 
ter. Under these circumstances it would perhaps be prudent 
to reject those which are obviously affected by such a cause, 
and by means of the usual formule to reduce a considerable 
number of observations near the point where we wish to ob- 
tain it with great precision, to that point exactly. The ob- 
servations are recommended to be near it, in order, as much 
as possible, to avoid an error arising from any small error in 
the coefficients of the general formula. Proceeding on these 
principles, 
