(^69 ) 



commencing at the odd hours for one half to the preceding and for 

 the other half to the following hour. 



Because a storm as a rule expires gradually, it is often'irapossible 

 to give the exact moment it is past. If doubtful I have always taken 

 the longest time for its duration ; hence many days following a great 

 storm are reckoned as being disturbed, which otherwise would have 

 passed as undisturbed. 



For the time of the maximum I have taken the moment of maxi- 

 mum agitation, which does not always correspond with the hour 

 of maximum postturbation. 



I believe the hour at which the meaji H-force reaches its lowest 

 level is a better time-measure for the storm-maximum, but to determine 

 it a large amount of measuring and calculating is required, the 

 change in level being often entirely hidden by the ordinary solar- 

 diurnal variation. 



The intensity of the storm has been given after a scale of four 

 degrees : 1 = small ; 2 = moderate ; 3 = active ; 4 = very active. 



It is not possible to give a definition of this scale of intensity in 

 words, the reproduction of typical cases would be required for this. 



Hourly distrihution of the heginning of storms. 



It is a known fact, that the starting impulse is felt simultaneously 

 all over the earth. The Greenwich and Batavia lists furnished me 

 with 53 cases of corresponding impulses, which, if the simultaneity 

 is perfect, must enable us to derive the difference in longitude of 

 the two observatories. 



I find in 6 cases 7^12™ 

 „19 „ 7 

 „28 „76 



Mean 7^ 7^^153 



True difference l^l^l^K 



It it very remarkable indeed to derive so large a difference of 

 longitude with an error of 4 seconds only, from 53 cases measured 

 roughly to 0.1 hour. 



The simultaneity should involve an equal hourly distribution if 

 every ,S-impulse were felt over the whole earth. As this is not the 

 case, which is proved by the lists of Greenwich and Batavia, it is 

 easy to understand that the Batavia-irapulses show indeed an unequal 

 hourly distribution. We find them more frequent at 6'^ and 10'' a. m. 

 and 7^ p. m. 



