On Rapid Periodic Variations of Terrestrial Magnetism. 23 



required per day and night. To save the excessive use of paper, 

 it was replaced only twice during 24 hours, i.e. usually at G o'clock 

 in the morning and evening. Hence tlie record was run over 

 three times hy each component, which caused considerahle 

 confusion of the record, on quiet days by tlie superposition of the 

 trace of the same component, and on disturbed days, by the 

 mutual intersections of the different components either direct or 

 reflected b}" the traps. The confusion is, however, merely ap- 

 parent in most cases. The time marks, the difference in the 

 intensity of traces and also slight differences in the optical error of 

 the mirrors of different instruments, causing different shading of 

 the photographic traces, served together as convenient signs for 

 disentangling the intricate record. Ambiguous cases were ex- 

 tremely rare, even if two branches of traces were nearly super- 

 posed. After we became familiar in the course of the investiga- 

 tion, with the characteristic behaviours of short period disturbances 

 for different components, the distinction of the different com- 

 ponents became still easier. 



During the summer months, the pedestal P3 was covered 

 with large cases of zinc, which consisted of seven pieces and when 

 put together in series, formed a continuous channel extending along 

 the entire length of the pedestal. Five or six dishes with CaCl2 

 were placed inside the channel. These arrangements were, 

 however, far from being efficient for the j^revention of the ex- 

 traordinary dampness of the chamber during a certain period of 

 the warmest season. 



The clock for giving the time mark was occasionally compared 

 with that of the Post Office in Misaki and also checked by means 

 of the sundial placed in the yard of the Marine Laboratory, so that 

 the accuracy of the time was only of the order of a minute at the 

 most. A convenient control could be made by the very frequent 

 occurrence of slight local earthquakes which left distinct marks on 

 the photographic records and could be easily identified with the 

 seismographic record obtained in Tokyo. ^^ 



1) Ttie distance between the Observatory and the Tokyo Central Meteorological Ob- 

 servatory is about 60 km., so that the time taken by the principal phase of seismic waves 

 will never exceed 18 seconds. 



