August i i, 1923] 



NA TURE 



20 



Solar Activity and Atmospheric Electricity. 



Dr. CHREE.in a recent paper.^ giving the results 

 of his investigation based on the Kew atmospheric- 

 electric data, reaches the following interesting 

 conclusion, as stated in his abstract : " The results 

 obtained are on the whole not incompatible with 

 Dr. Bauer's conclusion, but they indicate that if a 

 relationship of the kind supposed exists, the sun- 

 spot influence must be very much less in the case 

 of atmospheric electricity than in that of terrestrial 

 magnetism." The conclusion ^ referred to by Dr. 

 Chree is that the potential gradient of atmospheric 

 electricity apparently varies during the sun-spot 

 cycle, increasing with increased sun-spot activity, 

 and that the diurnal range of the potential gradient 

 of atmospheric electricity, like the diurnal range of 

 terrestrial magnetism, increases with increased sun- 

 spot activity. 



I fully agree with Dr. Chree that if there is a 

 relationship between solar activity and atmospheric 

 electricity, it ought to turn out to be a world-wide 

 phenomenon, just as in the case of the recognised 

 relationship between solar activity and terrestrial 

 magnetism. However, this fact is to be kept in 

 mind, that disturbances, because of local conditions, 

 play a far more predominant role in atmospheric 

 electricity than in terrestrial magnetism, and may 

 in fact be of such a character as to mask completely 

 any world-wide effect. Accordingly, in atmospheric 

 electricity a very careful selection of stations is 

 necessary for the study of world phenomena. It 

 thus may turn out that one very favourably located 

 station, such as the Ebro Observatory at Tortosa, 

 Spain, upon the excellent and consistent observations 

 of which my first conclusions were chiefly based, 

 may lead to a more certain result than a dozen 

 unfavourably located stations. 



One of the best criteria for judging the unfavour- 

 ableness of a station is the ratio c^\c^ of the amplitude 

 of the 1 2 -hour and 24-hour waves, respectively, of 

 the potential gradient ; the smaller the ratio the 

 more readily does the station show " universal " or 

 " terrestrial " results. This ratio at Kew is one of 

 the largest of any of the stations known to me ; it 

 varies between winter and summer from about i to 5, 

 and on the average for the year is 2 against o-8 for 

 Ebro, 07 for Eskdalemuir, and 0-25 for the oceans. 

 Furthermore, the absolute value of the potential 

 gradient at Kew is more than twice the average 

 normal value for the region of the earth from 

 60° N. lat. to 60° S. lat. However, in spite of the 

 disturbed character of the Kew station. Dr. Chree 

 has succeeded in getting results which he says are 

 not incompatible with mine, and I shall show how, 

 by the recognition of another variation to which 

 atmospheric electricity appears to be subject, the 

 validity of his results respecting the sun-spot effect 

 may be enhanced. 



Every series of carefully made and undisturbed 

 observations, extending over a period approximating 

 a sun-cycle or more, has received investigation and 

 definite results have, in general, been obtained. 

 A notable exception is Potsdam, where, because of 

 severe climatic disturbances, instrumental changes, 

 and changes in reduction factor, the extensive series 

 of observations at this station unfortunately is subject 

 to discontinuities and therefore cannot safely be 

 utilised for the detection of a sun-spot effect. For 

 the present sun-spot activity, as indicated by the 

 Wolf or Wolfer sun-spot number, is taken as a measure 



' "A Supposed Relationship between Sun-spot Frequency and the Potential 

 Gradient of Atmospheric Electricity," Proc. Phys. Soc, London, vol. 35, 

 Part 3, April 15, 1923, pp. 129-136. 



« Terr. Mag. ani Aim. Elec, vol. 27 (1922), p. 30 ; see conclusion /. 



NO. 2806, VOL. I 12] 



of solar activity, the possible influence of which 

 upon atmospheric electricity is to be investigated. 

 Other measures of solar activity — for example, 

 prominences — are likewise included in the complete 

 investigation. 



Suppose we have for a series of years the mean 

 annual values of the potential gradient P in volts 

 per metre, as resulting from continuous registrations 

 throughout the year and based only on electrically 

 undisturbed days, without negative potential. Let 

 P„ be the mean value of P for the entire series 

 corresponding to the mean date T„, and let S„ be 

 the mean annual sun-spot number ; then with fair 

 approximation a formula of the following type is 

 found to hold : 



P = P„ + s(S-S„) + /(T-T„). 



(I) 



The coefficient s represents the change in P corre- 

 sponding to one sun-spot number, and t represents 

 the time-effect or annual change in P dependent, 

 apparently, upon the average character of the 

 particular sun-spot cycle in which the series of 

 P-observations happens to occur. 



For nearly every series discussed thus far, when 

 the station is fairly free from pronounced local or 

 climatic disturbances and the annual values of P have 

 been derived from electrically-undisturbed days, the 

 coefficient s is found to be positive and averages 

 about 0-30 per cent, of P„ for the mean of the year ; 

 hence, if the sun-spot development from minimum 

 to maximum is loo numbers, the potential gradient 

 P suffers an increase from the year of minimum sun- 

 spot activity to the year of maximum sun-spot activity 

 of about 30 per cent. ofV^. (See Table i.) 



The value and sign of the coefficient t may depend 

 upon whether the sun-spot cycle in which the P-series 

 occurs is below or above average development. 

 Thus the present cycle, beginning with the year 191 3 

 of minimum sun-spot activity, is above average 

 development ; hence t turns out to be negative, as 

 shown in Table i. No such high mean annual 

 sun-spot number as 103-9 for 19 17 has occurred 

 since 1870, when S was I39"i. How / may vary 

 with cycle, and s with season of year, will be discussed 

 in the complete paper which is to appear in the 

 September 1923 issue of Terrestrial Magnetism and 

 Atmospheric Electricity. 



The values of s and t, expressed both in percentages 

 of P^ and in volts per metre, as determined by the 

 method of least squares, will be found given in 

 Table i for three observatories, from Spain to 

 Scotland, and the mean epoch 191 6. The third and 

 second columns from the end of Table i contain 

 the mean square errors, Mo and m^, obtained respect- 

 ively, first on the assumption that the departures 

 of P from the mean value P„ represent errors of 

 observation, and secondly that formula (i) applies. 

 It will be seen, from a comparison of the figures in 

 the two columns, that by the application of the 

 corrections (sun-spot effect and cycle or time effect) 

 a very much better representation of the observed 

 values of P is obtained than by the arithmetical 

 mean P„. 



The last column, r„ contains the correlation 

 coefficient between potential gradient and sun-spot 

 activity after the application of the corrective term 

 - t{T - T„) to the observed values of P. It will 

 be observed that, in general, the coefficient so nearly 

 approaches unity, especially for Ebro and Eskdale- 

 muir, as to leave scarcely any doubt of a definite 

 relationship between the potential gradient of 

 atmospheric electricity and sun-spottedness. By the 

 recognition of the /-change, which is similar in its 

 effect to the secular change in terrestrial magnetism, 



