58 REPORTS ON THE STATE OF SCIENCE. 
registers are incomplete; Australia and New Zealand, because prior to 
1850 we were without knowledge; and China, because its accessible 
registers end about 1644. 
All materials prior to a.p. 1600 have been discarded on account of 
their fragmentary character, and it is often impossible to say whether 
certain entries refer to destructive earthquakes or only to comparatively 
small tremors. The only earthquakes considered are those which have 
been destructive, and these are divided into the following three classes : 
(I.) shocks which have cracked walls or damaged chimneys ; (II.) shocks 
which have destroyed a few buildings; (III.) shocks which have caused 
widespread disaster. For Italy and for Japan these three classes have 
been taken separately, in pairs, and en bloc. Which ever way we have 
plotted them one result is clear, viz., in each of these two widely 
separated districts during the last three hundred years there have been 
periods of activity and periods of comparative rest. When the zigzag lines 
which show frequency from year to year are smoothed to curves you 
obtain a series of undulations the crests of which are separated from 
each other by periods varying between five and twenty years. There is 
no indication of a recurrence of activity after regular or equal intervals of 
time. The following table gives dates for the crests of these waves. In 
comparing any two of these dates it must be remembered that either of 
them might be increased or decreased by a year. The reason for this is 
twofold. First, an earthquake or earthquakes which occurred at the 
end of a year might, for the purposes of this investigation, have been 
assigned to the year following. Similarly, those which occurred in 
January of a given year might have been referred to the previous year. 
Also, it is difficult to determine the exact position for the crest of a wave. 
An inspection of the table shows for Italy eighteen dates for wave-crests, 
and fourteen of these agree very closely with dates indicating periods of 
seismic activity in Japan. These coincidences suggest that a relief of 
seismic strain in one part of the world either brings about a relief in some 
other part, or that relief is governed by some general internal or external 
agency. 
Periods of Seismic Activity. 
Japan Italy | Differences | Japan Italy | Differences 
Year Year 
1613 1612 1 1751 1755 4 
— 1626 — 1765 1767 2 
1644 1642 2 1782 1784 2 
1663 1660 3 — 1798 — 
1697 1693 4 I 1803 1806 3 
1704 1703 1 | 1834 | 1833 1 
|. Samay 1717 o | 1856 1856 0 
| 1728 1728 0 i — 1873 —_ 
| = Slee | 1898 1896 2 
VII. The Time of Maximum Motion as indicated by Three differently 
installed Horizontal Pendulums. 
The three pendulums are the Milne type (see ‘ B.A. Report,’ 1902, 
p. 60). Pendulum A records east-west motion; it stands on a brick 
column, the cross-section of which is 18 inches by 18 inches. 
