1921-22.] Gyroscope and “ Vertical” Problem on Aircraft. 305 
This simple form of erector is capable of providing an accuracy of 3 or 
4 minutes of angle when mounted on an aeroplane. The balls, it is to 
be remembered, rotate on a hard metal track forming part of a system 
which includes a high-speed gyroscope. The track is being tapped con- 
tinuously, as it were, by an automatic hammer which vibrates to and fro 
with a frequency of about 300. The balls, as a very important consequence, 
are extremely sensitive to tilt of the track. 
Consider now the arrangement of pushers and checks shown in fig. 26. 
If the device is used as shown hi that diagram, each rod acts as a pusher 
for one ball, and as a check for the ball in the following compartment. 
The balls are free to move, relatively to their pushers, through a distance 
on the track subtending at the centre an angle approximating to Itt. The 
value of a is great ; if the diameter of the track is 6 inches, the value of a 
will approximate to 5 inches. 
It will be clear, from what has been said, that the device, so used, has 
the following properties. When set up on a table, deflected through a 
large angle 6, and left to itself, it will move towards the vertical quickly so 
long as 6 is great. The angle /9 approximates to 90°, and the device moves 
up without precession in the ordinary sense of the term. When 6 is small 
the rate of recovery is relatively small, inasmuch as the balls do not move 
far from their pushers. 
This device should not be thus used on an aeroplane. If so used, then 
when the aeroplane is turning, the device moves rapidly at each instant 
towards the apparent vertical. The value of ^ is great for all the values 
of ijj usually met with on an aeroplane. This follows from the large 
freedom given to the balls relative to their pushers, and from the fact 
that the accelerating forces experienced by the balls in the track, when 
the aeroplane is turning, are usually great. Hence the arrangement, so 
set up, is of little use. 
It is important to note these facts. A novice, viewing the performances 
of two such instruments when set up on a table in a room, one provided 
with checks in front of the balls so as to limit their freedom, the other 
unprovided with checks, as shown in fig. 26, would be apt to conclude that 
the latter arrangement was preferable to the former. In point of fact, the 
latter arrangement results in a poor performance on an aeroplane, the 
former in an excellent one. We shall return to this form of erector. 
The complete action of the type of erector described above (figs. 12, 13, 
14) may now be explained. So long as the balls move radially in the slots 
(the first stage of the erecting process) each integral erecting couple 
results from a couple which comes into existence approximately at A 
VOL. XLII. 20 
