March ii, 1920] 



NATURE 



47 



so as to set its axis of rotation as nearly as possible 

 parallel to the axis of the earth. It is only when 

 the two axes coincide that the wheel is free of any 

 further tilting* action — that is, when it is pointing 

 true north; deviate the axis, however slightly, from 

 this position of rest, and the action of the earth comes 

 in again to precess the wheel back again to the 

 north. 



Here is a simple form of gyrostat with three 

 degrees of freedom. If I hold it in my hand and 

 revolve on my axis, this does not move the wheel, 

 which still keeps pointing to the same part of the 

 i-oom. On the other hand, if I restrain or clamp 

 one of its degrees of freedom so that I am able to 

 tilt the axis of the wheel during my revolution, the 

 wheel is caused to precess and to set its axis parallel 

 to the axis on which I am revolving. Reversing the 

 rotation, the wheel also reverses. 



This is what takes place with the gyrostat on the 

 earth's surface provided it is frictionlessly mounted. 

 Such an instrument is before you, and I will try to 

 demonstrate by its means the rotation of the earth. 

 A wheel is rotating inside this case at 15,000 revolu- 

 tions per minute. The case is constrained to move 



about this vertical frictionless axis. Mere motion 

 of translation has no effect in changing the direction 

 of the axis of the wheel, but if this room rotates the 

 axis of the wheel tends to set itself parallel to the 

 axis about which the room is rotating. 



We all believe th^t this room is rotating about the 

 axis of the earth ; if so, the axis of the wheel must 

 set itself parallel to the axis of the earth, but it 

 must be kept horizontal, and, therefore, it will point 

 north and south. Here it is pointing in an east-and- 

 west direction ; it is held by a string. I will now 

 burn the string, and it will find for us the true north. 

 Observe that it is really the true north direction, 

 whereas that magnet points to the magnetic north. 

 I set it away from the" north, but on the other side, 

 and repeat the experiment. 



Such a simple form of gyro-compass could not be 

 of anv use on a moving ship, because the rolls of 

 the ship would react too violently on the spinning 

 wheel and cause considerable deviations in the read- 

 mgs of the compass. The use of a gyro-compass on 

 land is verv limited, and its great value at the present 

 time is on board ship. The spinning wheel is acted 

 NO. 2628, VOL. lOSi 



upon by forces which tilt the axis. Now, a rolling 

 and pitching ship is about the worst place to put a 

 gyrostat to act as a compass, because the ship's 

 movements all tend to tilt the axis. 



The problem, therefore, is to make the compass 

 insensible to the movements of the ship and respond 

 only to the slow angular rotation of the earth. To 

 indicate the severity of the ship's movements, 1 may 

 recall a recent trip of this gyro-compass on board a 

 fast destroyer. During a severe gale the ship was 

 recorded to roll more than 50° of total angle. Many 

 of the crew were forced to lie on the decks, the 

 lockers emptied their contents, and even some of the 

 oil-lamps suspended from the ceiling- were unseated 

 by the pitching of the vessel ; yet the gyro-compass 

 maintained its accuracy, and allowed the ship to be 

 steered safely into harbour, to which she had to run 

 for safety. In all this whirlwind of movement the 

 gyro-compass heard, and only responded to, the still, 

 small voice of the earth's rotation. 



For use on board ship the compass must be 

 mounted on a pendulum in gymbal rings, and its 

 period of oscillation is lengthened to something like 

 85', which is usual in practice, so that the rolls, 

 which are of the order of 7 to 15 seconds' period, 

 shall have but small effect on the compass. In this 

 case the rotation of the earth does not act directly 

 upon the gyro-wheel, but by means of the force of 

 gravity through the pendulous weight. Unfortunately, 

 this form of mounting introduces troubles of its own. 



Suppose we study our simple gyrostat and see what 

 happens when we attach a weight to the end of the 

 horizontal spindle ; this will give us some idea of 

 \vhat occurs when the force of gravity is acting 

 through the pendulum trying to tilt the gyro-wheel. 



We know from our law that the wheel will precess 

 under the tilting action, but the new direction of 

 rotation that we are trying to produce by means of 

 the weight, unlike that produced by the earth, which 

 is always in one direction, is in this case continually 

 carried round by the precessing wheel, and the pre- 

 cession is, therefore, permanently maintained. We 

 also find that if we hurry the precession the spindle 

 rises, lifting the weight; while, on the other hand, 

 if we delay the precession, the spindle drops and the 

 weight falls. The rate of precession is proportional 

 to the weight. Halving the weight, for instance, 

 halves the rate at which the wheel rotates round the 

 vertical support. 



Coming back again to our pendulous-mounted 

 gyro-compass (Fig. 2) ; suppose the spindle is pointing 

 west and is horizontal, then the earth as it rotates 

 will leave the wheel pointing in this one direction in 

 space, but the weight will try to follow the earth's 

 rotation, and will start precessing the gyro towards 

 the north. The rate at which the wheel comes to 

 the north depends upon the weight W attached 

 to the casing. All the* time the wheel is coming to 

 the north the earth is adding to the rate of the pre- 

 cession, and the spindle is, as a consequence, tilted, 

 and deflecting- the weight at the north position. 

 Under these conditions the effect of the weif^ht is^ to 

 continue the precession, and the gvro-wheel will swing 

 through the north position, and continue to move 

 until the effect of the earth arrests and reverses the 

 motion. 



The compass will therefore continue to swing 

 through the north position with constant amplitude 

 backwards rtnd forwards, undamped. To render the 

 compass of use, some method of damping the swing 

 mu<t be introduced so that the compass may finally 

 settle on the north. This damping can be carried out 

 bv means of friction, preferably fluid friction, between 

 thf vertical spindle and its support; but. although 

 this will damn the swings, it is inadmissible becnusr 



