26 



a propoitioiial iiificasc of {fi}an(\(y ). Henoo in any i;iven c-aso (other things 

 remaining unchanged) there will ho a eortaiu value of the .suspended weight 

 whieh will give a iiKi.cinnDii of .steadiness. To attempt to calculate this maxi- 

 uuun theoretically is out of the question, and no more than a very rough 

 experimental determination is practicable. Tlie desirable amount of \veight will 

 be small if («) is a small part of the whole resLstancc. The writer's exi)erience 

 points to the conclusion that in practice the greatest steadiness is to be got by using 

 a ma.<s of no more than 1 to 2 kilos, as tlie bob of a horizontal iiendulum seismo- 

 gra|)h. A nia.ss greatly heavier than this, while it has mon> jiower to move its 

 tracing index, tends so considerably to crush the hard steel points by which the 

 .su|iporting frame is pivotted, that the frietional error introduced by the jivots 

 more than counterbalances the advantage gained, 



§ 27. Horhdutfil Pt'iiil iiltiii) ailli ficxih/r in (ind pimtti-il nlriif. 



Instea«! of using a ligid frame to carry the massivi' bob, we mav suspend it 

 by a .«ingle tie ami strut, and by making tiie tie flexible we may avoid having a 

 joint at its place oi' connection to the fixed su]p])ort. Thus, in the sketch, i\[ is 



the heavy mass hung from a fixed j'o-t P by a 

 wire 7^ which tics it to the p<i.st at a, and lield out 

 so as to fVirin a liorizimtal ];endu]um liy the rigid 

 strut <S pivotted in a conical socket h. The ma.ss 

 M may of course citiier be fixed to S, in whicli 

 ca.se the .steady line is at the axis of in.stantaneous 

 rotation of the system, or it may be pivotted at 

 the axis of in.stantaueitus rotation of .S' alone (the 

 mass of T l)eing negligible). A prolongation of 

 S beyond M forms a convenient writing pointer. 

 \ |iair of similar instruments serve to register 

 the two comjionents of the horizontal motion. 

 The necessary freedom of rotation about the axis 

 of support nh is afforded by combined twisting 

 and liending of the tie at n, whose elasticity gives 

 a small amount of stability to the system. If 

 that is too inconsiderable for jiraetice it may easily !)e increased by advancing the 

 ]ioint a slightly in front of tlw vertical line through l>. Except for imperfect 

 elasticity, there is no frietional resistance at 'i. The strut need not of course be 

 horizontal, but by placing it so we reduce the stress on it, and consequentlv the 

 friction at its pivot, to a minimum. 



This modification of the horiznntal ])endulum is due to Mr. T. Crav, who 

 has sugi,rested it in his japer " On Steady Points f(ir Earthipiake Measuiements '", 

 (■ommunie;ited to the Seismological Society of .Ta]ian shortly after the instrument 

 described in § S', had been exhibited. Figs. 11 ami 12, Plate VII, show an 

 elevation and ]ilan of the instiiiment as described bv Mr. ( Jrav in the Philosophi- 



^wyA'-A 



