270 REPORT— 1898. 



moving smoked-glass surfaces, -whilst others recorded the relative motion 

 of the pendulum with similar pointers attached to the end of light levers, 

 which multiplied the movements of the pendulums. 



Professor J. A. Ewing established in Japan a heavy pendulum 20 feet 

 in length, the movements of which relatively to the ground were recorded 

 as two components by means of short levers. 



The result of our experience with these instruments, which extended 

 over several years, was that, although they occasionally yielded useful 

 diagrams of local disturbances, it so oftei^ happened that an earthquake 

 took place which had a period approximately agreeing with the natural 

 period of the pendulums that these were set in violent motion. 



Inasmuch as the period of local earthcjuakes is for the most part less 

 than three seconds, whilst the period of the long pendulum at Catania is 

 ten seconds, the bob of this instrument becomes a practically steady point 

 for such disturbances. 



Directly, however, we turn to the pendulums one or two metres in 

 length which we tind so largely employed in the Italian Peninsula, it seems 

 impossible that the character of the seismograms of local shocks which 

 they furnish should not be largely affected by the free period of these 

 instruments. 



I — and I think I may add all observers who liave had experience with 

 the ordinary pendulum type of apparatus, and also with the long-period 

 duplex pendulums and the steady point bracket seismograph used in 

 Japan, not only in recording actual earthquakes, but also in testing such 

 instruments by subjecting them to artificiiiliy produced earthquake-like 

 movements, which movements were absolutely measurable — woul'd not 

 hesitate in adopting the two latter types of apparatus in preference to 

 those in which the principal feature is the bob of an ordinary free 

 pendulum. 



Directly we turn to a consideration of the form of apparatus best 

 adapted to record the movements due to earthquakes which have origi- 

 nated at great distances we are upon uncertain ground. 



What is chiefly required is to determine the time at which a movement 

 commences, the duration of its various phases, and to measure its varying 

 amplitudes, periods, and directions. 



A source of error, especially with regard to the measurement of 

 amplitude or angular motion, rests in the fact that the periods of various 

 phases of earthquake motion may vary between as much as five and sixty 

 seconds. The consequence of this is that at some time or other there is 

 synchronism between the natural period of the instrument and that of 

 its moving platform, with the result that records may be greatly distorted. 



A good illustration of the difference in records obtained from different 

 instruments is seen when we compare the records of horizontal pendulums 

 and those of the vasca sismica (see p. 2G'i). 



Dr. Cancani's observations on the great Indian earthquake of 189G 

 (p. 207) also show the amount of difference which may be expected in 

 seismograms obtained from pendulums varying in period. 



From what we know at present, seismograms which are trustworthy 

 in their main features are yielded by apparatus which for rapid preliminary 

 tremors records the same relatively to a steady point ; and if the pendulum 

 has a natural period of ten to fifteen seconds it apparently follows the 

 slow movement of the succeeding waves. 



