A PENDULUM FOR THE REDUCTION TO A VACUUM. 
441 
gave a less value for the factor n than the latter which is little more than 
4 tenths of an inch in diameter. Conceiving therefore that I might be enabled 
to determine the law by which such values were governed, I was induced to 
try other cylindrical rods, supported in the same manner as the copper one 
above mentioned, and of nearly the same length, but much smaller in diameter. 
I accordingly procured a brass rod, or wire, only O' 185 inch in diameter: in 
fact, it was a piece of the same kind of wire as that which was used with the 
solid brass cylinder No. 1 1, mentioned in the preceding part of this paper, page 
410. I also caused one to be made about the same length, still smaller in dia- 
meter : but, as brass was not exactly suitable to such purpose, when so small, 
I procured one of steel, only 0'072 inch in diameter *. The length of the brass 
rod was 56'4 inches, it weighed 3106 grains and its specific gravity I found to 
be 8'444. The length of the steel rod was also 56*4 inches, its weight (including 
a small brass screw attached to the end) was 433 grains, and its specific gra- 
vity I found to be 7'687. Each of these, when in use, was screwed into the 
shank of the knife edge apparatus, which was T55 inch long, as already 
described in page 409. The results are contained in the following short Table : 
where I have continued the numbers of the pendulums from the preceding 
Table in page 433, for the sake of a convenient reference : No. 42 being No. 21 
in the former list. 
Thirteenth set . — Results with plain cylindrical rods. 
Pendulum rods. 
No. 
n 
Weight of 
adhesive air. 
Copper, 58'8 inches long, 0410 inch diameter .... 
42 
2-932 
4-904 
Brass, 56‘4 inches long, 0'185 inch diameter .... 
43 
4-083 f 
1-484 
Steel, 56'4 inches long, 0-072 inch diameter .... 
44 
7-530 
0-479 
Now, here we find a regular increase in the value of n, as the diameter of 
the rod is diminished : and the inference is that, with a much smaller wire 
(such as is generally used in experiments with the pendulum,) the value of n 
* This was just 5 times the thickness of the iron wire, used in the preceding experiments, with the 
pendulums No. 1 — 20. 
t I ought not to omit stating that this is the mean of eight different experiments, made with two 
different rods, at two different periods. Four of them (viz. two double ones,) were made with the 
wire here alluded to, on June 14th, and the others on August 2nd, with another piece of exactly the 
