A PENDULUM FOR THE REDUCTION TO A VACUUM. 
435 
applied to the results from the experiments with Captain Kater’s convertible 
pendulum (knife edge A) will, on the assumption that the specific gravity as 
taken by him is correct, be 8*72 6 (See page 415). So that these two pendu- 
lums, which were considered to be nearly in accordance when the old cor- 
rection w r as applied for the reduction to a vacuum, will now differ 6*331 vibra- 
tions in a mean solar day, from each other : or T f 5 th of an inch in the length 
of the seconds pendulum. In each of these computations the pendulum is 
assumed as making exactly 86400 vibrations in a mean solar day. 
It appears, from the general Table of comparisons above given, that the long 
cylindrical copper rod (No. 21 ) is the most affected by this newly discovered 
principle ; even more so than any of the spheres or cylinders suspended by 
wires, or than the thick brass bar (No. 31) which presents a flat surface of 
| of an inch in width, to the line of motion. We find also that the small 
spheres are more sensibly affected than the larger ones ; which agrees with 
what M. Du Buat observed in the experiments made by him, to which I shall 
presently allude. But the relation between the results of the other pendu- 
lums, does not appear to me, at present to be satisfactorily accounted for, or 
to be referable to any known principle ; and, in order to determine the effect 
which is produced in the results, they must in all cases be made the subject of 
actual experiment. We may however draw this inference from the whole, 
that we cannot strictly compare the results of any invariable pendulums, that 
have been swung in various parts of the globe, without subjecting them (or 
their prototypes) to this rigid test. The English pendulums have generally 
been made after one fashion, which is that of No. 22 in the above enumera- 
tion: but I have seen some French ones, of a different form, where the bob 
has been much thicker, and suspended by a cylindrical rod ; and which would 
probably give a very different value of n, if subjected to actual experiment. 
The rods of the pendulums taken out by MM. Freycinet and Duperrey, were 
cylindrical and about an inch in diameter : and it may be a matter of doubt 
whether the results with those pendulums are strictly comparable with the 
results obtained by pendulums of Captain Kater’s construction. I have 
already shown, in the experiments with the pendulums No. 40 and 41, that a 
similar difference in the form of the rod only (the bob continuing the same) 
causes a difference in the result, amounting to upwards of 1*2 vibration in a 
day : and there may probably be other sources of discordancy which can be 
