36 



seismonieter should Iw. If its jwriod of i^winging i.* iiuicli greater than tlie ]ioriocls 

 tlio cirllniuaki' waves, tlicse will not generally eoinmiinieate any imnioilerate 

 motion to it, and even if it swings eon.siderably there will no difficulty, provided 

 the motions are recorded on a continnously moving plate or drum, in distin- 

 guishing the long-period wave.* due to the swing of the pendiduni from the 

 relativei\- mneii slmrter waves due to tile motion of tlie ground. The latter are 

 superposed on the former, and in measuring them the undulating path traced out 

 on account of the swing is to be taken as the datum line, or line of no displace- 

 ment, instead of the line which would be traced if the pendulum had remained 

 rigorously at rest. 



The tendency of a jieuduluni tii acquire a swing when its point «if .su])]>ort 

 is? shaken may be reduced and even comi)letely iili\iated by intnulucing frietioual 

 resistanci', which dissipates, moi'c ur less comjiletely, the energy couiinuni<-ated to 

 the hanging mass by the .-uccessivc Imrizuntal impulses to which it is subjected. 

 This action neces.sirih- (iccurs to a certain c.-ctent in all seismographs, a certain 

 amount of friction being inevitable, es](ccially where the eaith's motion is con- 

 sidcralilv nnilti|illed liv a I'ecording lever. The friction mav of cour.se be 

 ])urposeIy increased, and by doing this even a short jjendulum may l>e |)revented 

 from ac;|uiring a swing during a prolonged disturbance. Hut any introduction 

 of friction involves a .sacrifice of accuracv in a seisniometei', and the effect of 

 adding enough resistance to jpievinl a short pen<lnlum from swinging is that it 

 altogether fails to rec(ird niiniile eartliMiiakes, and gives unduly small indications 

 during more considei",d>lc shocks. 



Many obsci'vers iiave u.scd for the measureinent of eavthi|ual;e motion a 

 jienduluiii piovided with a tiacing point or other kind of index liv which its 

 greate.--t (lis])laci'ment irom the vertical is registered. Now it is clear that if the 

 liendnlnm is lialile to b<' .-it swinging during an carthijuake such indications are 

 Mholi\' witlioiil v.ihic. The\' (ii'|iend not onlv on tlie amplitude i>nt also on tile 

 iluration ot the disturbance, and, very dirctb', on tiie period of the waves. 

 They mav i>e, and oi'len will be, manv liine< ^leatcr than the motion of the 

 ground. 



< )n the olher lian<l, if the ]iendulum is so much ictarded by friction that 

 the energy which it I'ceeives by any one displacement of the ]ioint of .«upport is 

 dissipated bcfoi'c the next displacement occurs, no accunudatioii of small motions 

 can occur. Tlie ]iendnluin will not swing, and its gn'atest dis]>lacement cannot 

 exceed, but will actually be always less than the greatest motion of the gKiund 

 during the disturbance. 



11 we attempt to obtain, by means of a pendulum, a stufic record of earth- 

 quake motion — that is a record inscribed on a stationarv plate (or registered by 

 some other ei|uivalent contrivance) instead of on a continuously nio\ing |>late — 

 then, since we have no means of distinguishing (in the recortl ) earthi|uake 

 motions ])roi)er from the swing of the ])euduluni, we must resort to the arrange- 

 ment just mentioned. That is to .say, the ]!endnlnm nin<t iiave enough frictional 



