79 



oscillation in about 5 f^econds, and its decrement of aniplitiule should not exceed 

 about 1 mm. piT osfillation. 



Wlicn tile iTi'onl is dojiositcd un a plato or drum started by the eartlujuake 

 and >t<>(ijM.(l soon aitiT it, the iMjiiiiibriiini may properly lie made still less stable 

 tiuin this; fur in that easi' "ereeping" of tlic jiointcr licfore or after the distnr- 

 i>anec docs no harm. 



Methods [4J and [ö| have eadi eirtain merits and defects. E.xperienec 

 shows that when method [4J is used, and the etjuilibrium is nearly neutral, the 

 pointers creep about through small distances almost lumtinually, from causes 

 probalily the same as those whii'h ])revented the mirror in Messrs. Darwin's 

 e.\l>eriments (referred to in the Introduction) from remaining at rest. The result 

 is that the circles which the ])ointers trace out on account of the continuous 

 rotation of the plate become gradually widened ; and hence when an earthquake 

 ijceurs its smaller movements are indistinctly recorded, and its beginning is 

 sometimes untraceable. Instances of this will be .<ccn in .-several of the I'lates. 

 The |xisitiiin of the point<^'rs nn tlic glass jilate, :ind the plate itself, rciiiiire to be 

 changed from time to time, when the lines become too wide. On the other hand, 

 nu'thod [0] may allow a part of the earthquake to pass unrecorded ; and it is 

 <i|Mn til the fuitiier objection that there is no easy way of seeing that the .■-tarting 

 gear is in good order, whereas the ap[iaratiis used in iiietlmd [4] shows at a 

 glance whether it is in a fit state to receive a record. As regards difficulty and 

 (•lift of ctinstructiim, [4] requires a much more carefully made and costlv piece 

 of clockwork, but avoids the circuit-closer, magnet, and liattery which must l)e 

 ciirnliiu'd with thi' simpler clii<4< of methdd |.")|. i'"i)r use in an uliserNalm v where 

 the iM>trMments can receive frequent inspection and skilled attention (acdudition 

 ( >sen(ial to.-uieess in all seismic tibservationl, the writer greatly prefers method [4]. 



Where the normal position of the index depends upon the pidl I'xerted iiy 

 a spring, as in tlii- vertical M-ismometers of §J^ 4(; and 17, the crce|iing tendency 

 allud<'d to abdvc is m> great ;is to make [.')) tlie «mlv praetiealile inellmd nf 

 regi-tration. 



^ Xl. ConMfwtifu ililiiil-t. 



'I'lic principal puini.-. to lir- attended In in the de,-ign tif an absoluti' scisimi- 

 metcr are. iIicm- :- 



The point or axi> «if .«npiMirt (pnints m axes, if (hei-i- be :i group nf nias.'ies 

 instead nf only niie) nuist partake as exactly as pnssible nf the ninlinn nf the 

 earth at the place of <jliMM-vution. This I'equii'esa rigid post or stone table (whose 

 dimensions must be grt".itiy less than the shortest wave-lengths in the earth- 

 ipiakes to lie measured), firmly indx'dded in the earth ; a stx-ady clamp, giving 

 deliiiite ihiint.« of altachnienl ijj '2'-\\; and a rigid supporting fnime (o carry the 

 |Kiint or axis of support. 



'I'he distance from the axi.i of xupixirt to the steailv line must gn-atlv exceed 

 the gre!ite««t earth-motion lo be measured, except where the apparatii« is astatic, 



