length of the pendulum vibrating seconds. 45 
probably take place in the adjustment of the sides of the dia- 
phragm, the end of the pendulum, and the disk, the results 
would be erroneous, as the interval would go on increasing 
till the pendulum came to a state of rest. 
The interval which best fulfilled these conditions was found 
to be about 530 seconds. This admitted five coincidences 
(affording four intervals) to be taken before the arc became 
too small for the observations to be continued with safety. 
With this interval an error of one second in the time of coin- 
cidence would occasion an error of only 0,63 in the number 
of vibrations in 24 hours. 
Here it must be evident that no sensible alteration could 
take place in the knife edge during the experiments without 
its becoming perceptible at every coincidence, since the num- 
ber of vibrations in 24 hours deduced from each interval, 
must vary with any change in the form of the knife edge. 
The following was the method pursued in making the 
observations. The small weight or slider being placed with 
its index at a certain distance (say one inch and a half) from 
the middle of the pendulum towards the great weight, and 
the second weight about five inches from the knife edge, the 
Ys of the support were elevated, the knife edge of the pen- 
dulum was placed in them, with the great weight above, and 
the frame gently lowered till the knife edge was left on the 
surface of the agate. The requisite adjustments of the tele- 
scope having been made, the pendulum was set in motion in 
an arc not exceeding one degree and four tenths, in order 
that its velocity might not be greater than that of the pendu- 
lum of the clock. 
