VARIATIONS IN THE DIRECTION OF THE VERTICAL. 107 



used for the screw, and it became evident that with proper 

 precautions as to stability, contact would be certain within 

 TT5i7!TrrnT ^^ ^^ inch, and since a pendulum of 17 ft. 26 "6 in. 

 would change its position by xoVu ^^ ^^^ inch for one second of 

 arc change in the vertical, it is obvious that this method is 

 capable of testing and recording on paper, the position of the 

 pendulum at regular intervals of say one minute or five minutes, 

 and shewing changes of ;/,y certainly, and perhaps -^}^ of a second 

 of arc. 



The details of the machine are not yet complete, but the design 

 is simple enough. Suppose that below the pendulum a very 

 massive piece of cast iron be placed to serve as a foundation for the 

 recording parts and tirmly attached to this cast iron an ordinary 

 micrometer, with screw of fifty to the inch ; this must be so 

 situated that when the screw is turned the sliding piece will come 

 into contact with the disc at the bottom of the pendulum, and 

 thus close an electric circuit (See diagram.) Suppose this is 

 situated due north of the pendulum, any change in the latter 

 to north or south will be indicated by a difference in the turn of 

 the screw required to make the contact. Now, the screw is not to 

 be turned by hand, but the cylinder which is to receive the 

 record will work on a vertical axis, and have on the top of its 

 rim a segment of a wheel, which as it passes the micrometer will 

 gear into the wheel on the head of it, and give it one turn, so 

 bringing the piece which is worked by it into contact with the 

 end of the pendulum, this contact brings into play an electrically 

 controlled pen which instantly marks the moving cylinder. Now, 

 it is obvious that if the contact is made a little earlier or later 

 • next time the cylinder comes round, that difference will be shewn 

 on the paper by the two pen marks ; in this way motions of the 

 pendulum to north or south will be recorded regularly once a 

 minute. Now, if we place another micrometer due east of the 

 pendulum, any motions east or west will be recorded on the 

 cylinder in the same way ; and by combining the curves any 

 other motion could be discovered, but since it will add very 

 little to the first cost of the machine to put in two other screws, 

 it is proposed to put one on the north-east side, and the other on 

 the south-east. In this way any motion north or south — east or 

 west — north-east or south-west — and south-east or north-west 

 would be recorded, and by combination of curves any other 

 direction. To any one not accustomed to mark with an electric 

 contact of this sort, it will no doubt appear that the means pro- 

 posed are inadequate to record such minute changes as may be 

 anticipated in the end of the pendulum from changes in the 

 vertical, but eighteen years experience of this method of 

 recording, that is by electrical contact, has given me abundant 

 proof that it is a quite sufficiently accurate result entirely 

 confirmed by the special observations made to test it. In 



