PART. II. POLAR MAGNETIC PHENOMENA AND TERRE1.LA EXPERIMENTS. CHAP. Ill 453 



THIRD SERIES. 



Tables giving the record of storminess for each live-day period. There will be one table for each 

 station, containing the positive, negative and absolute storminess for each component, and one column 

 containing the total storminess. The numbers will be expressed in absolute units. 



The method of calculating the storminess is very much the same as that employed for calculating 

 the perturbing force. The "normal line" is drawn on the magnetogram in the way described. During 

 the perturbations a number of areas are formed by the registred curve and the normal line. The areas 

 on both sides of the latter are taken out for each interval of two hours, and from them, knowing the 

 scale value and the length of the interval, we can find the positive and negative storminess. The relative 

 values given in the first series of tables are simply these areas given in centimetres and reduced to the 

 same sensitiveness for all four stations. 



In taking out average values, it is necessary, as we know, to have a value for every two-hour 

 interval throughout the period. It will unavoidably happen that in some records short intervals of time 

 may be missing, but the blank interval due to the change of paper on the cylinder will generally be so 

 short that it will practically introduce no error; for the intermediate values can be found by connecting 

 harmoniously the two ends of the curve. If during a perturbation, the curve is invisible for a short 

 interval of time, we have employed the same method of completing the curve by harmoniously connecting 

 the two parts. 



In the curves it has occasionally happened that records were wanting for several hours. If 

 considerable disturbances were occurring at the other stations during these intervals, we should have 

 to omit the whole five-day period; but as a rule we have been able to estimate thestorminess for the 

 blank intervals. 



Values which are not found directly from the curves, and consequently cannot claim great accuracy, 

 will be put in brackets. These values may be found in various ways e. g. by completing the curve 

 for a short blank interval or by estimating the value from the curves of the other components at the 

 same place or from the curves of the neighbouring stations. 



During such investigations it became clear that the great storms did not show the same properties as 

 the small ones with respect to distribution in space and time. It was therefore of interest to find the 

 average properties of the great storms separately. 



The classification of the storms into great and small is of course to a certain extent quite arbitrary. 

 We have decided on the following procedure : To find the storminess of great storms, we take 

 that of every two-hour period for which the positive or negative storminess is greater than 15 y in any 

 of the components. If the condition for a great storm is fulfilled for a certain two-hour period in one 

 component, the corresponding storminess is counted in the case of the other components even when it 

 is less than 157. 



Hirkeland. The Norwegian Aurora Polaris Expedition, 19021903. 58 



