382 Reports on Special Researches 



It is, of course, necessary to determine the reduction factor by which the indications 

 of the instrument must be multiplied in order to reduce them to the corresponding values 

 of the potential-gradient over a flat surface. In order to simplify matters, it is always 

 arranged that, as for as the essentials which affect the potential-gradient observations are 

 concerned, the configuration of the sails during observations is one of three specified types. 

 One of the chief requirements for the determination of the reduction factor is that of obtain- 

 ing a flat stretch of land, and in this matter careful judgment is necessary. It is essential 

 to bear in mind that flatness of the ground in the immediate vicinity of the apparatus is not 

 suflacient to prevent distortion of the field by distant topographical irregularities of large 

 size. Thus, to give an example which will illustrate the general nature of the considerations 

 involved, suppose the apparatus is situated between two parallel mountain ridges of semi- 

 circular cross-section. It can readily be shown that, if the mountain ridges are suflficiently 

 far from each other, each acts as a linear doublet of moment M = ^Xa- per unit length, 

 where A' is the electric field and a the radius of the semicircle. At a point midway between 

 the ridges the alteration in the field due to the ridges would amount to 2{2M/r^)=2Xa'^/r'^, 

 where r is the distance from the center of one ridge to the point of observation. Hence, to 

 consider a somewhat drastic case, if the tangent of the angle subtended by the top of the 

 ridge were 0.2, there would be an error of 8 per cent in the measured potential-gradient. 

 The error would obviously be the same for two ridges, each 1 mile high and 5 miles distant, 

 as for two ridges 1 foot high and 5 feet distant. 



For the actual shore observations it is desirable to choose a piece of ground which is 

 practically on a level with the surface of the sea, and is free from trees. In the method' 

 which has been used for the shore observations, a wire several meters long is suspended 

 horizontally from two posts by suitable insulators, and a collector is attached to its center. 

 The wire is connected to an electroscope at one end and simultaneous readings are then 

 taken with this apparatus and with the apparatus on the ship. Since the wire which 

 supports the collector lies in a horizontal plane, it does not acquire any charge, and so 

 does not disturb the field ; and for the same reason it does not contribute to leakage through 

 atmospheric dispersion. 



One of the chief sources of uncertainty in the determinations of the reduction factors 

 arises from the fact that it is usually impossible to get the ship nearer to the shore station 

 than half a mile; and it appears that if clouds are at all prominent, the ratio of the potential- 

 gradients on ship and on shore is by no means constant. For this reason it is felt that the 

 reduction factors should be measured as often as possible when favorable opportunities 

 arise, so that by taking the mean factors obtained under different conditions, and in 

 different localities, the importance of unknown irregularities will be reduced. 



The daily observations of the potential-gradient are taken over periods of about half 

 an hour, and, as far as possible, are arranged to extend over the middle period of the 

 determination of the ionic content and conductivity. The example on page 397 illustrates 

 the method of recording the observations. 



CONDUCTIVITY. 



In the measurement of each of the elements, conductivity, ionic content, penetrating 

 radiation, and radioactive content, there is involved, in some part of the work, a determina- 

 tion of the rate of loss of charge by some insulated system connected to an electroscope, and 

 usually several determinations of the quantity have to be made. It is customary in such 

 cases to read the indications of the electroscope at equal intervals of time, and then deduce 

 the desired results by makmg use of a caUbration curve. This method has certain dis- 

 advantages; one has eit her to rely on the constancy of the caUbration curve, a procedure 



■See G. C. Simpson and C. S. Wright, Proc. R. Soc. A., vol. 85, p. 1.S2, 1911. 



