On Rapid Periodic Variations of Terrestrial Magnetism. 23 



described in paragraph e). But the trains in question which are 

 most frequent during the night hours, are distributed quite 

 liapliazard with respect to the earthcjuakes, if we either take 

 shght and moderate earthquakes together, or conßne our attention 

 to the moderate ones only. In any case, a further study in tliis 

 direction seems scarcely promising, in so far as the characteristic 

 magnetic waves are of no local plienomena confined within a 

 very limited area of the earth's surface. 



On some occasions, slight irregular disturbances of very 

 abnormal type were observed in both components, but usually in 

 one component only, no corresponding simultaneous anomaly 

 being marked in the other. These irregularities are most probaljl}^ 

 due to some slight accidental shock acting as an impulse to 

 settling down a slight unstable position of tlie suspended system; 

 for example, a slipping of microscopic magnitude, or the in- 

 troduction of a minute dust particle at the point where the 

 suspension wire touches the adjusting screw, may cause a quite 

 sensible displacement in such an ultrasensitive state of suspension 

 as adopted in our instruments. To detect premonitory, if any, 

 magnetic disturbances, it will therefore be necessary to protect 

 the instrument more thoroughly from any mechanical shock or 

 other accidental disturbance. 



However, during the entire course of the observations, no 

 remarkable destructive earthquake occurred in the neighbouring 

 districts, so that we are not yet in a position to draw any definite 

 conclusion as to tlie value of magnetic observations as a means 

 for predicting strong earthquakes — which was the original purpose 

 of our present installation. 



PART III. 

 Special Statistical Investigations in Detail. 



6. In the following, some of the results of the detailed 

 statistical investigations with respect to some of the general results 

 above described, will be given, w^hich are on one hand a con- 

 firmation of .the qualitative remarks stated in the preceding 



