62 A. A. Michelson— Rate of Tuning-forks. 
determined. The whole operation may be performed in less 
than ten minutes. 
To determine the rate of EUt, we will suppose that this is 
within a small fraction of a whole number of vibrations per 
second, and that this whole number is known; say 128 v.8. 
A mirror is attached to one prong of EUt, and in front of this, 
and at a distance of two or three feet, is placed a Geissler tube. 
This last is illuminated once every second. The tube itself is 
continuously illuminated and its image in the mirror presents 
the appearance of a broad band with well-defined edges. 
Against this the narrow flash is projected. 
Evidently if EUt, makes an exact whole number of vibra 
tions per second, then this flash will always find the fork in 
the same phase of vibration, and consequently its image will 
always appear at the same part of the band. If, however, the 
fork makes, say 128-1 v.s., then it will occupy successively 
the positions 
$686.8 $455.8: 718 9 10 
That is, there would be ten flashes between the positions (0) 
and (10). If, therefore, ten flashes occur in one ‘ period” then 
the whole number 128 is to be increased (or diminished) by 
one-tenth ; and in general if a flashes occur in one period then 
1 
EUt, makes 128+ — vibrations per second. 
Since EUt, vibrates continuously the number of periods 
which may be counted is unlimited, hence a can be found with 
any desired degree of accuracy. 
It was found that the chief difficulty in executing this plan 
lay in the imperfections of the break-cireuit. If thereby the 
seconds intervals differed as much as 0°002 seconds, then the 
operation would be impracticable, on account of the irregulari- 
ties in the position of the flashes. A great many break-circuits 
were tried, and the one which was finally adopted as being the 
most reliable and giving the least trouble was the “ mercury 
globule.” Even this, however, in its usual form was unsatis- 
factory. A great improvement was effected by narrowing the 
globule in the direction of the swing of the pendulum, by 
placing it in a small tube whose upper end was flattened. 1 
the current from a “bi-chromate” cell is interrupted by this 
break-cireuit the mercury oxidizes rapidly, and the flashes 
(given by a small induction coil through the Geissler tube) 
become irregular. 
To obviate this difficulty, the circuit of a * gravity’ battery 
was interrupted, and this worked a relay, which in turn inter- 
