A PENDULUM FOR THE REDUCTION TO A VACUUM. 
467 
might also take place in other pendulums ; which, from repeated trials, as 
above stated, I found to be the case. And this furnishes us with another proof 
of the propriety of varying such experiments in all manner of ways, in order 
to guard against the effect of any unsuspected source of error. 
Results by turning the face of the Iron 'pendulum (No. 30). 
Exp. 
Knife edge B. 
Knife edge b. 
Exp. 
205 
86220-190 
86220-999 
206 
207 
20-346 
23-499 
208 
209 
20-433 
21-818 
210 
213 
21-002 
21-976 
211 
215 
20-524 
23-309 
212 
216 
20-129 
22-967 
214 
218 
20-574 
21-791 
217 
219 
20-302 
22-338 
222 
220 
20-247 
22-450 
223 
221 
20-473 
22-401 
224 
227 
20-504 
22-881 
225 
228 
20-362 
23-077 
226 
229 
20-962 
23-033 
230 
Mean = 
86220-465 
86222-503 
It may here be stated that the knife edges of all the convertible pendulums 
in my possession are marked on both sides of the pendulum : on one side with 
the capital letters A, B, and on the reverse side with the small letters a, b. 
Therefore the column, in the above Table, designated as the knife edge B, 
denotes the results obtained when the side of the pendulum, marked B, is next 
to the observer: and the other column denotes the results when the pendulum 
is turned half round in azimuth, and consequently the side marked b is next 
to the observer. The mean difference in the results will be found, as already 
stated, equal to 2 , 038 vibrations in a mean solar day. If we compare the 
several results we shall find the partial differences somewhat greater than what 
generally occur in a regular series of experiments: but these have arisen from 
designedly varying the position of the knife edge on the agate plane, with a 
view to the discovery of the cause of the principal discordance; and which I 
can attribute to no other source than inequalities in the knife edge, or agate 
plane, or both ; but which are not immediately perceptible to the eye. From 
