On Lunar Periodicities in Earthquake Frequency .
457 at all upon either steam or carbon dioxide .
Magnetic oxide of iron is the final product of the action of steam or of carbon dioxide at a high temperature upon metallic iron :\#151 ; 3Fe 4- 4H20 = Fe304 + 4H2 .
3Fe+4C02 = Fe304+4C0 .
Now , metallic iron has been detected in basalts and some other rocks by Andrews ( ' Brit. Assoc. Rep. , ' 1852 , Sections , p. 34 ) , and by other observers ( e.g. , G. W. Hawes , ' Amer .
J. Sci. , ' Ser. 3 , vol. 13 , p. 33 ) , and I have verified this observation in the case of the gabbro of Locli Coruisk .
But it must be remembered that both the reactions indicated in the equations just given are reversible , and therefore the presence of metallic iron along with the magnetic oxide in such rocks cannot be taken by itself as final proof that the oxide and the associated gases , hydrogen and carbonic oxide , are the products of the action of steam and carbon dioxide upon metallic iron .
The presence of marsh gas in these rocks and the production of large quantities of hydrocarbonous gases , as well as liquid petroleum , in many parts of the earth 's surface , tend to support the view , which is apparently gaining ground , that in the interior of the earth 's crust there are large masses , not only of metal but of compounds of metals , such as iron and manganese , with carbon .
Assuming the existence of such material , it is easy to conceive how , by the action of water at an elevated temperature , it may give rise to metallic oxides and mixtures of hydrogen with paraffinoid and other hydrocarbons .
This view was put forward some years ago by Mendelejeff ( " Principles of Chemistry , " Translation by Kamensky and Greenaway , vol. 1 , 364\#151 ; 365 ) , and it has lately received further support from the results of the study of metallic carbides , which we owe especially to Moissan ( ' Roy .
Soc. Proc. , ' vol. 60 , 1896 , pp. 156\#151 ; 160 ) .
" On Lunar Periodicities in Earthquake Frequency .
" By C. G. Knott , D.Sc .
, Lecturer on Applied Mathematics , Edinburgh University ( formerly Professor of Physics , Imperial University , Japan ) .
Communicated by John Milne , F.li .
S. Received November 4 , 1896 , \#151 ; Read February 4 , 1897 .
( Abstract .
) 1 .
Introduction.\#151 ; The paper is a discussion of Professor Milne 's Catalogues of 8331 earthquakes , recorded as having occurred in Japan , during the eight years 1885 to 1892 inclusive .
These catalogues , forming vol. 4 of the ' Seismological Journal of Japan , ' Prof. C. G. Knott .
are unquestionably the most complete ever constructed for an earthquake-disturbed country .
The discussion is really a working out of certain lines suggested in a paper on " Earthquake Frequency , " communicated by me in May , 1885 , to the Seismological Society of Japan , and published in vol. 9 of the ' Transactions ' of that Society .
In that paper I pointed out the importance of subjecting earthquake statistics to some strict form of mathematical analysis , and gave a simple arithmetical process for separating the annual and semi-annual periods in earthquake frequency .
The results then obtained have been fully corroborated by Dr. C. Davison in his paper " On the Annual and Semiannual Seismic Periods " ( ' Phil. Trans. , ' vol. 184 , 1893 ) ; and my suggestion that the annual period is connected with barometric pressure is also strongly supported by Dr. Ferd .
Seidl in his pamphlet 'Die Beziehungen zwischen Erdbeben und Atmospharischen Bewegungen ' ( Laibach , 1895 ) .
The semi-annual period , which was first clearly brought into evidence in my earlier paper , does not admit of a very ready explanation .
In my paper of 1885 I also considered in some detail the various tidal actions which might reasonably be supposed to have a determining influence on earthquake frequency .
From lack of material it was not possible at that time to make a satisfactory search for lunar periodicities ; but the remarkable fulness of information contained in Professor Milne 's latest catalogues tempted me to undertake the labour involved in ( first ) tabulating the statistics in terms of lunar periods , and ( second ) analysing harmonically the tables so prepared .
2 .
The Lunar Daily and Half-daily ..\#151 ; In one of the catalogues the earthquakes are classed according to district .
Districts 1 to 6 lie on the N.E. and E. coasts of Japan , reckoning from the north ; districts 6 to 11 on the S. coast ; and 12 to 15 on the W. coast .
Districts 6 and 7 are the most important , the former being the region including Tokyo and Yokohama , and the latter the region including Nagoya , which was the scene of the destructive earthquake of October 28 , 1891 .
The investigation into a possible lunar daily period is conveniently based upon this classification into districts .
Had that not been done by Professor Milne the labour involved in taking into account differences in local time would have been enormous ; for , to compare the time of occurrence of a recorded earthquake with the immediately preceding meridian passage of the moon , it was necessary to apply corrections for longitude and local time .
The statistics for each district were , in the first instance , separated out and tabulated according to time of occurrence , estimated in hours after the immediately preceding passage of the moon .
The method On Lunar Periodicities in Earthquake Frequency .
is explained in full in the paper .
To lighten in some measure the labour of the harmonic analysis , certain districts were thrown together to form a district group .
Table I contains the number of earthquakes in each district or district group , which formed the material for discussion .
Table I. N umber of Description of District .
earthquakes .
district .
1 397 Nemura .
2\#151 ; 5 627 E. coast .
6 1432 S.E. corner .
7 3632 Nagoya , \amp ; c. 8 245 Kii Channel . .
9\#151 ; 10 335 E. and S. of Kyushu .
11 384 W. of Kyushu .
12 112 'i 118 V W. coast of Main 13 Island .
14\#151 ; 15 145 J Of the tabulated numbers for each district or district group , overlapping means of every successive five were taken , and these were divided by the mean of all .
The numbers so obtained represent relative frequencies throughout the lunar day , and are given in Table II , which also contains a like series for all the earthquakes taken in combination .
The most important are the frequencies for districts 6 and 7 , and also for all combined .
They are shown graphically in the figure ( p. 46i ) .
Each series of numbers was then discussed by harmonic analysis in accordance with the Fourier expansion n \#151 ; qo 1000 + 2 Cn sin n 71 = 1 where x is 1000 times the relative frequency at time t , estimated in hours after the mei'idian passage of the moon , and where the amplitude cn and the phase a.nare to be calculated .
The amplitudes and phases for the first four harmonics are given in Table IV .
There is a tendency for the second harmonic amplitude to be greater than the first , while in half the number it is the greatest of all .
As regards the times of occurrence of the maxima for the different harmonics , there is no regularity except perhaps in the case of the second harmonic .
In four ( 1 , 6 , 7 , 8 ) the maximum of the second harmonic falls within two hours of the half time between the upper and lower meridian passage of the moon .
In the others it falls within two hours of the times of upper and lower meridian passage .
Frequency of Earthquakes throughout the Lunar Day .
Prof. 0 .
G. Knott .
All Japan .
38383888 S !
33g8888SgS ; iSSSSgJ83 OOiOiOiOiOOOOOOOiOJOaOOOlQOiOOOOO ... ... ... # ... ' . .
... .\#171 ; ... . .
rHOOOOrHrHrHrHiHOOOOOOOOOOrHrHrHrHrH Districts .
14\#151 ; 15 HOO^CDO050iNOOC0HQ0N^l\gt ; WCl ; O ?
0Ol^H00 OOHOi^CO^OcpNOiOpoiNOHHasoOOQOOOJ^CO rHrHOOOOOOOOr\#151 ; IrHrHr-l-HrHr-lOOrHOrHrHrHT\#151 ; ( 00 COOIOl^aOjOONOiCOCCNWTfl^N^CONOlTHHOO 05N^^Q0OHO05Q0 ?
9O|0QH99l\gt ; l^aiOH0:^H OOOOOrHrHrHOOrHrHtHrHrHOOOOOrHrHrHrHrH COtOCO^OO\lt ; MCO*\gt ; .CSCOlOUOTj\lt ; l\gt ; t\gt ; \#187 ; HeOiOX^COCO'^\lt ; MeOC^ HHOQOrtpOWpGOOOOiCOyNO^O^HCOiO^iO rHrHrHOrHrHrHrHrHrHOOOOOOOOOOrHrHr-irHrH 3 OON^iOOi^iOOSOCOHWlOOi^NVOO^OOiOiOTfKMN 9909990iOiOOHHogiooGpoo99\#174 ; oooHH rHrHrHOOOOOrHrH r-^ rHrHOOOOOOO'HrHrHrHrH 0 1 0QNQOcqaiNaNHcooiN0ioooai\gt ; joQOO0^coco 90^00x059905909099^90599000999^ OOOOOOOOrHr\#151 ; 1 rl rIHHHHOHOOHi " ' H H rl 1 !
00 !
COOQOCO^OOOCOOO^COO^Of^COOCCOlM^QO^GO^ 00O5O5\#174 ; 01HHOOHOfl5rlO0i05OQ05OOfl5 05'0l0 ) . .
OOOOOrHrHr-fHrHrHOrHHOOrHOOHHOOOO i\gt ; t^O'ilOaOOOr^CiOOlO^t^CO^OO^^O^OOCOOOiMOO'^T^^ OOiaiQXOHHHOSNOGONOOQOOJOMHOOlMOOOW 005050 0:00009905050i059 ? ? ? ? ? ? ?
?0 r\#151 ; lOOrHOrHrHrHrHOOOOOOOOr\#151 ; 1 H H H rl H O rl j o l\gt ; a\lt ; O^0QOOO^O^t\gt ; NN0O^OlOQOOHHTfUOCO TflH( ?
q^COaiCOH^^I\gt ; NNH^lOHOQON(M(NNCiO aoo999oc ?
99 9aiOO ?
9 ?
C59o:9 9 0io rHrHrHOOOrHiHrHrHOOOrHrH'HrHrHOOO OOO rH T o* ^OOHQ^WOCOHGO^HvOHoOOOM(^CDCO\#187 ; OCOOOH 059ai090oooH009xai05 99999990o 0*H\#169 ; rH0rHrHrHrHrHrHrH00000rHrH00\#169 ; i"^rHi\#151 ; ( rH 1 1 NO(MOOKM(MCOCOHCO(NQOQOCOiO(M(M\#187 ; CCOCCiOiOiftC oooH090oo50iooooN(X)99HH09oo99H rH rH rH \#187 ; H rH rH rH rH O O O O O O \#169 ; rH rH rH rH rH rH rH rH rH rH I W H^CO^iOCCf\gt ; OOO5OH(^CO^iO0NQOCiOHWM^iO HrHHHHHMHHHCN\lt ; MOJCN\lt ; M'M !
Prof. C. G Knott .
Table IY.\#151 ; The Coefficients c and a , the amplitudes and phasecoefficients .
District .
Cl. c2 .
^3* c4 .
*1- a2 .
a3 .
( X4 .
i 94 *4 68-7 46 -9 16-6 7-85 9-95 2-08 3-43 2-5 29*9 36-5 32 -4 35-2 24 -2 4-33 0-88 2-95 6 18*4 20 -7 29 -9 14-6 21-8 11 -5 1-79 5-65 1 13 *0 16-4 3-17 8-98 6-7 7-9 1 -3 6-1 8 54 *2 45-8 10-0 6-17 20 -2 9-03 2-53 2-07 9\#151 ; 10 65 -0 56-9 53 -8 5-26 15-9 3-94 3-0 1 -92 11 42*8 91-5 32-5 26-0 3-55 4-19 6-91 1-53 12 245*0 233-0 111 -5 36-7 3-38 4-93 3-57 4-48 13 73-9 167-0 193 0 7-48 15-1 4-62 3-9 1-09 14\#151 ; 15 175-0 247-6 91-9 41 -5 11-2 2-78 2-79 1-75 All 10 3 17-9 10 -9 3-97 6-62 7-97 2-42 2-43 A comparison ' of these times with the times of high water in the various districts failed to establish any relation .
We are forced to the conclusion that if there be any lunar-diurnal periodicity imposed upon earthquake frequency , it is the result of tidal stresses acting directly on the approximately rigid crust of the earth , and not indirectly through the loading due to the ocean tides .
Because of the comparatively great number of earthquakes the results for districts 6 and 7 are the most important .
During the eight years under discussion , the shocks in district 6 occurred with normal frequency .
All were comparatively small ; none were disastrous .
On the other hand , the case of district 7 is altogether peculiar .
In general , this is a comparatively quiet district ; but the great disaster of October 28,1891 , was followed by a vast number of after-shocks .
These show distinct daily and half-daily periodicities , the latter having the greater amplitude .
Thus , from district 6 , with its 1432 earthquakes distributed with fair uniformity over eight years of normal activity , and from district 7 with its 3(332 earthquakes , almost wholly included in a short fierce interval of fourteen months , we obtain very similar evidence as to the existence of a lunar half-daily period in earthquake frequency .
The results for " All " depend , in the main , upon the statistics for districts 6 and 7 .
The curious way in which the comparatively prominent 1st harmonics of these two districts tend to cancel ope another , is a warning of the danger of lumping together statistics of different countries or different seismic areas in the search for possible periodicities .
3 .
The Lunar Monthly and Fortnightly Periodicities.\#151 ; There are five distinct kinds of months recognised by astronomers , namely:\#151 ; On Lunar Periodicities in Earthquake Frequency .
4(53 ( 1 ) The anomalistic month ( 27*545 days ) .
( 2 ) The tropical month ( 27*322 days ) .
( 3 ) The synodic month ( 29*531 days ) .
( 4 ) The sidereal month ( 27*3228 days ) .
( 5 ) The nodical month ( 27*212 days ) .
Of these , the last two cannot be regarded as having any influence bn earthquake frequency , for the only conceivable effect is a tidal one , and the sidereal and nodical months have no necessary tidal relations .
At the same time the periods of the sidereal and tropical months are so nearly the same that they can hardly be discriminated in the lapse of eight years .
On the other hand , the anomalistic month may show itself in earthquake frequency , since the moon in perigee has a greater tidal action than when it is in apogee .
Again , because of the moon 's variation in declination , being now north of the Equator , now south , we may reasonably search for a tropical monthly periodicity .
And , finally , the synodic or common month may make itself apparent , there being possibly a greater tidal stress when the moon is in syzygy ( as in ordinary spring tides ) than when the inoou is in quadrature ( as in neap tides ) .
The earthquakes were accordingly tabulated according to these four months , whose periods differ appreciably ; the nodical month being also included .
For , by analysing the statistics in terms of both the tropical and nodical months , we may be the better able to draw conclusions as to the real existence of one or other periodicity .
The relative daily frequencies , as finally reduced , are given in Table Yl , and the curves are shown in the figure .
As in the case of Table II , each of the tabulated numbers is the mean of five successive numbers , and is regarded as belonging to the time of the middle one of these five .
It should be mentioned\#151 ; and the remark applies also to the former cases\#151 ; that the number of earthquakes which really occurred during the last time interval was increased in the proper ratio : so that the frequency during this last interval was made comparable with the frequencies of the other intervals .
It was interesting to find how admirably the number so obtained harmonised with its neighbours of the first and penultimate interval .
In all cases the obvious aftershocks of any earthquake occurring on the same day were neglected .
The 3000 aftershocks of the great disaster of October 28 , 1891 , were also left out .
The earthquakes on which the discussion is based numbered from 4725 to 4741 , the number varying slightly for each monthly period , since , at the beginning and end of the eight years ' interval , there were always a few , differing for the different months , which did not make up a complete period , and were , consequently , neglected .
Each series of numbers was analysed harmonically as far as the 464 Prof. C. Gr .
Knott .
Table " VI.\#151 ; " Monthly " Frequencies .
Day .
Anomalistic , from apogee .
Tropical , from 0 ' decl .
N. to S. Nodical , from ascending node .
Synodic , from full moon .
1 0-919 1 -077 0-937 1 -064 2 0-945 1 -072 0-925 1 -081 3 0-976 1 -107 0-998 1 -029 4 0-980 1 -069 1 032 1 -ooo 5 0 -999 1 -052 1 -056 0-961 6 1 013 1 -040 1-068 0-963 7 1-061 1-006 1 -103 0-964 8 1-033 0 -902 1 -045 0 -984 9 1-058 0-928 1 -051 0 -980 10 1064 0 930 1 -025 1 -002 11 1-023 0-945 1 -050 0-999 12 1-002 0 -952 1-047 0-983 13 1 -005 1 -020 1 -037 1 -009 14 1 -012 1-000 1011 1-029 15 1 -021 0-978 0-964 1 -030 16 1-048 0-982 0 -901 1 -042 17 T -061 0-974 0-858 1 -032 18 1-022 0-936 0-896 1 -039 19 1 -oio 0-969 0-901 1 -039 20 1-004 0 -967 0 -939 1 -005 21 1 -006 0 -964 0-981 0 -985 22 1 -ooo 0-975 1-018 0 -965 23 1 -017 0-991 1-011 0-918 24 1 004 0-987 1 -044 0 -905 25 0 -952 1 -020 1 -028 0-939 26 0-955 1 -035 1 -028 0-945 27 0-920 1 -054 0 -992 0-973 28 0-906 1 -086 0 -994 1 -020 29 \#151 ; \#151 ; _ 1-045 30 \#151 ; \#151 ; \#151 ; 1-060 \#151 ; \gt ; \#151 ; , : 1 first four harmonics , according to a formula identical with that already given , due regard being paid to the different periods and the time unit involved .
The results are given in Table VII , the phase coefficients being given in days .
Table VII.\#151 ; Amplitudes and Phases .
" Month .
" Ci .
c2 .
c4 .
\#171 ; ! .
a2 .
1 Anomalistic . . .
46-2 47-8 12-9 16-5 21 -7 8-5 5-2 6-2 ; Tropical 54-7 40-7 23-1 17-2 6-0 1 -9 7-9 2-4 j Nodical 49-5 55-2 28-3 17 -6 1 -2 7-9 6-9 2-7 i I Synodic 11-0 52 -1 24-5 4 -7 13-7 2-7 7-7 0-6 !
A study of these tables discloses the presence of certain features which have no raison d'etre on any rational theory of tidal stress .
On Lunar Periodicities in Earthquake Frequency .
465 The most important of these is the fact that the nodical month , which has no direct connexion with tidal stress periodicity , is characterised by harmonic amplitudes greater , on the average , than those corresponding to the other months .
This is particularly evident in the graphs . .
* There are , however , other features which favour the hypothesis of seismic tidal stress , such as the occurrence in the vicinity of perigee of the Anomalistic 1st Harmonic amplitude ; the lagging , by one day , behind full and new moon of the Synodic 2nd Harmonic maxima ; the distinctly greater amplitude of the Synodic 2nd Harmonic as compared with those of the other harmonics\#151 ; a fact which is in accord with the fortnightly succession of spring tides .
It is , certainly , a striking fact that the same statistics which , when grouped according to an approximately twenty-eight days ' period , give a prominent 1st harmonic should , when grouped according to an approximately thirty days ' period , give a comparatively small 1st harmonic but a prominent 2nd harmonic .
4 .
General Conclusions.\#151 ; The conclusions are summarised under eight heads .
( a ) There is evidence that the earthquake frequency in Japan is subject to a periodicity associated with the lunar day .
( b ) The lunar half-daily period is particularly in evidence , both by reason of its relative prominence and the regularity with which , in each of two groups of the several seismic districts , its phase falls in relation to the time of meridian passage of the .
moon .
( c ) There is no certain evidence that the loading and unloading due to the flow and ebb of ocean tides have any effect on seismic frequency .
( d ) Hence we must look to the direct tidal stress of the moon , in its daily change , as the most probable cause of a range in frequency which does not exceed 6 per cent , of the average frequency .
( e ) There is distinct evidence , both as regards amplitude and phase , of a fortnightly periodicity associated with the times of conjunction and opposition of the Sun and moon .
{ / ) No definite conclusion can be drawn from the apparent monthly and fortnightly periodicities which seem to be associated with the periodic changes in the moon 's distance and declination , for the simple reason that fully as prominent harmonic components exist when the statistics are analysed according to the periodic change in the moon 's position relative to the ecliptic , and with this particular-period no tidal stresses can be directly associated .
VOL. lx .
2 N 466 Proceedings and List of Papers read .
( g ) Nevertheless , the value of the phase lends some support to the view that thei'e is a real connexion between the change in the moon 's distance and earthquake frequency , since the maximum frequency falls near the time of perigee .
( h ) These conclusions have , in comparison with previous similar investigations , a peculiar value , inasmuch as they are based upon accurate statistics of fully 7000 earthquakes occurring within eight years in a limited part of the earth 's crust , throughout which the seismic conditions may be assumed to be fairly similar from point to point .
February 11 , 1897 .
The LORD LISTER , F.R.C.S. , D.C.L. , President , in the Chair .
A List of the Presents received was laid on the table , and thanks ordered for them .
Communications from Professor Oliver Lodge , P.R.S. , and Dr. Larmor , F.R.S. , on the recent discovery by Dr. P. Zeeman of the effect of a magnetic field on the light emitted by a soda flame , were read by the Secretary .
The following Papers were read:\#151 ; X. " The Oviposition of Nautilus macromphahcs .
" By Arthur Willey , D.Sc .
, Balfour Student of the University of Cambridge .
Communicated by Alfred Newton , M.A. , F.R.S. , on behalf of the Managers of the Balfour Fund .
II .
" Report to the Committee of the Royal Society appointed to investigate the Structure of a Coral Reef by boring .
" By W. J. Sollas , D.Sc .
, F.R.S. , Professor of Geology in the University of Dublin .
HI .
" The artificial Insemination of Mammals and subsequent possible Fertilisation of their Ova .
" By Walter Heap , M.A. , Trinity College , Cambridge .
Communicated by Francis Galton , F.R.S. IV .
" On the .
Regeneration of Nerves .
" By Robert Kennedy , M.A. , B.Sc. , M.D. ( Glasgow ) .

