SABINE’S EXPERIMENTS. ot 
The next table contains the preceding results, collected and exhibited in one 
view ; the several reductions have been computed as already described in the 
similar table in the observations at New-York : the three last enumerated ex- 
periments, wherein the room was heated by charcoal, were designed to try 
the correctness of the allowance for expansion, by the comparison of results 
obtained under great differences of temperature ; the heat was kept up tonear- 
ly the same amount for several hours before the pendulum was set in motion, 5 
but in spite of endeavour, was too fluctuating during the period of the coinci- 
dences, to entitle the results to be considered as decisive in a matter of so much 
delicacy and minuteness ; the detail of these three experiments has been given 
at length, in order that the amount of this fluctuation may be seen, and the 
consequent diminution of. value i in the results appreciated : inasmuch, however, 
as they may still deserve estimation, they justify, by their agreement with the: 
nine preceding observations at the lower temperature, the allowance of 0, 435 
of a_second in 24 hours for each degree of Fahrenheit, which has been 
adopted on the assumption that the expansion of plate brass may be taken at 
a mean at 0.0218 of an inch per foot, for 180°, or 0000101 parts of the length 
of the pendulum for a change of temperature of one degree : it is desirable 
to establish the ratio of expansion of the individual pendulums by experiment, 
and it is purposed to be accomplished when they shall have been brought 
back from the Arctic Circle, for which they are on the point of embarkation ; 
the difference in the number of vibrations in 24 hours at the same spot in 
widely differing, but well ascertained temperatures, appears much preferable, 
in this determination, to the methods of instrumental measurement, and is pe- 
culiarly suitable to the purpose for which the rate of expansion is wanted ; 
but it will be an operation of much delicacy, requiring that the possible errors 
from every other source shall be reduced with certainty within very narrow 
limits, in order that any difference between the assumed and the real expan- 
Sion may be detected, and its effect, which can only be very oie attributed 
with confidence to its true cause. 
It is in this respect only that the account of the experiments now com- 
. } g ‘ = i / 
