EVENING DISCOURSES. 419 



the effect of pendulum gravitation which is proportional to T produces a pre- 

 cession proportional to T, and the earth's directive action which is proportional 

 to A produces a rate of decrease of T which is proportional to A. The result is 

 a vibratory motion. We find that T is usually very small compared with A. We 

 find that A is at the end of its swing when T is zero and vice versa. A is 

 alternately west and east of north. Now it ie necessary that the compass should 

 get to the north in diminishing swings and come to rest there ; that is, the swings 

 must be damped. We cannot directly damp the azimuthai motion, because as in 

 the magnetic compass of our Navy the support and damping medium are carried 

 round by the ship. Anschiitz, in his early form of compass, by the use of an air 

 blast gets rid of this connection with the ship. The air blast was arranged to 

 oppose the movement in azimuth when the wheel tilted, and thus he obtained 

 good damping. The strength of the air blast which varies proportionately to 

 the tilt should be nothing when the compass is at rest on the morth, that is when 

 the tilt is nothing, and this would be true with the compass at the equator. In 

 other latitudes, however, the compass sets itself with a. tilt stili remaining. The 

 wheel has a tilt because it is trying to set itself due north, and this leaves a 

 residual air blast producing a constant error in azimuth. There is aiways this 

 latitude error if we introduce forces in azimuth proportional to the tilt. It is 

 therefore preferable to damp the swings by acting on the tilt motion, because in 

 this case there will be no latitude error. How this is done widl be described later. 

 There are three forms of gyro-compass now in u.se. The Auschiitz (German), the 

 Sperry (American), the Brown (British). 



In the Anschiitz the instrument is supported by a bath of mercury (as 

 originally suggested by Lord Kelvin). There is a quasi-solid friction about the 

 vertical axis, however clean the mercury may be, and this introduces error. In 

 the ' Sperry ' the gyi'o is supported by a wire, the twist when it occurs being 

 taken out by a ' follow-up ' motor through an electric contact which switches 

 on the current to the motor. In the ' Brown ' the lower end of the vertical 

 spindle acts as the ram of a pump and stands upon a column of oil. The oil is 

 under great pressure (some 500 lb. to the square inch), and is kept pumping up 

 and down, and thus raising and lowering the vertical axis continually, eome 

 180 times per minute. The continual movement of the spindle results in a 

 vertical support which has an inconceivably small amount of solid friction. 

 A ship has an angular motion of one revolution in one day. If she sails north 

 at, say, 20 knots, she has another angular motion of one revolution in forty-five 

 days. A gyro-compass on the ship is sensible of both these angular motions, and 

 sets itself to make a compromise between them, and so points, not to the true 

 north, but one or more degrees west of the true north ; this deviation is the north 

 steaming error. Knowing the latitude and north speed of the ship we have tables 

 to allow for tliis error, and a special form of repeater was exhibited by 

 Bi'own in which the card can be set eccentrically so that the correction may be 

 automatically applied without further reference to the tables. Acceleration or 

 change of north speed will have another effect on the compass, as it acts on the 

 pendulous weight, and so starts an oscillation which may be called the ballistic 

 error. 



A little mathematics shows that if the period of the compas» is eighty-five 

 minutes, the compass takes its new position without oscillation, or, as we say, 

 as if it were dead beat, and there is no ballistic error. But in actual practice it is 

 found that when a ship turns in its course, and especially when it describes 

 a quadrant or semi-circle, an oscillation is set up in any gyro-compass. Mr. Brown 

 discovered this to be due to the methods of damping employed, and mathematics 

 shows that this is the case. This he calls the damping error. That this error 

 is due to damping and may therefore be got rid of in a simple way was now 

 published for the first time. On a merchant ship the damping error is of little 

 moment, but in a war vessel which is manoeuvring it may be serious, as it may 

 swing the compass off its correct reading by several degrees. 



Mr. Brown said that his difficulties had been endless. As soon as he had 

 what he thought to be a perfect compass an error would be discovered, and 

 months or years were spent in correcting it. The other inventors, who were more 

 or less satisfied with their own results, had no doubt met with the same diffi- 

 culties, but these errors were unknown not merely to new inventors but even to 



