On Lunar and Solar Periodicities of Earthquakes. 461 



than Mr. Knott's number in two out of every five eases. It will, I 

 think, be admitted that, until further evidence is brought forward, 

 the lunar day cannot be considered to affect earthquakes. 



4. In the second part of his paper, Mr. Knott treats of the 

 monthly and fortnightly periodicities. He expresses himself very 

 guardedly as to the results of this part of the investigation, chiefly 

 on the ground that the amplitudes found for the period of the 

 "nodical" month are quite as great as those found for the "tropi- 

 cal" or "synodic" months. As there is no conceivable reason why 

 the nodical month should affect earthquakes, the conclusion will 

 reasonably be drawn from this, that any period chosen at random 

 would give similar results. I think we must take it, in default of 

 further evidence, that this argument is valid and outweighs the not 

 very conclusive considerations which seem to Mr. Knott to favour 

 the reality of these lunar periods, in spite of the weakness of the 

 evidence to which he himself draws attention. The number of 

 earthquakes taken into consideration are about 4730, and this would 

 give an expectancy for each of the coefficients equal to O0258. A 

 reducing factor has to be applied which differs for the different 

 coefficients as has been explained in the case of the periods depend- 

 ing on lunar day. I obtain in this way the numbers shown in 

 Table II. For the sake of completeness, I give the amplitudes found 

 by Mr. Knott for the various months, and also the mean amplitudes 

 of the different monthly periods. 



Table II. 



"Month." 





C 2 . 



o 3 . 



C 4 . 





46 -2 



47-8 



12 



■9 



16 5 





54-7 



40-7 



23 



1 



17-2 





49-5 



55-2 



28 



3 



17-6 





11-0 



52-1 



24 



5 



4-7 



Mean of (1), (2), (3), (4) 



40 -4 



49*0 



22 



2 



14 -0 



Expectancy by theory of probability . . . 



24 -5 



20 -9 



15 



5 



9-3 



The striking feature of this table consists in the first place in the 

 roughly equal amplitudes for the different kinds of months, and 

 secondly in the fact that the expectancy is in all cases smaller than 

 the mean of the coefficients found, and in nearly equal proportions 

 at any rate for the monthly, third-monthly, and quarter-monthly 

 periods. 



5. This points not so much to the reality of the period found, but 

 to some common cause which has led to too low an evaluation of the 

 expectancy. The latter quantity has been calculated under the 



