296 



NA TURE 



LAu».i.si 25, 1923 



internal magnetic system, I, an external system, E, and 

 a non-polLMiti.il system, N, or of three equivalent 

 systems, X, Y, Z. The I-system constitutes about 

 94 per cent, of the total magnetic field, and K and N, 

 eacn about 3 per cent. (There is a possibility that 

 relativity effects, R, may play a part in the exact 

 evaluation of the three systems.) 



2. As a resultant effect of all systems causing the 

 secular variation of the earth's magnetism, the north 

 end of the magnetic axis of the I-system during the 

 past eighty years has been moving slowly towards the 

 west, and apparently at the same time slowly towards 

 the equator. The indications from all available 

 data are that if the magnetic axis completely revolves 

 around the axis of rotation, regarding the possibility 

 of which there may be some doubt, the period would 

 not be some hundreds of years, but many thousands 

 of years. The magnetic secular variation results 

 from changes, with lapse of time, both in the direction 

 of magnetisation and in the intensity of magnetisation ; 

 the latter quantity has been steadily diminishing 

 during the past eighty years at the annual average 

 rate of about 1/1500 part. 



3. A suggestive effect, dependent apparently upon 

 the distribution of land and water, has been disclosed, 

 namely, that the average equivalent intensity of 

 magnetisation for corresponding parallels north and 

 south, is generally larger for the land-predominating 



?arallel than for the ocean-predominating parallel, 

 he secular changes, however, are on the average 

 larger per annum for the south, or ocean-predominat- 

 ing, hemisphere than for the north, or land-pre- 

 dominating, hemisphere. 



4. B^or the earth's internal uniform magnetic field, 

 the following data apply for 1922. The magnetic 

 moment, M, is 8-04 x 10" C.G.S. ; * the components of 

 M, respectively parallel and perpendicular to the 

 earth's axis of rotation, are, Mp=7-88 x 10" C.G.S. , 

 and M, = i-6o x 10" C.G.S. ; Mp=4-93 M,. Were the 

 earth's magnetism uniformly distributed throughout 

 the earth, which is not likely, the average intensity 

 of magnetisation would be 0-074 C.G.S. The mag- 

 netic axis intersects the North Hemisphere in latitude 

 78° 32' North and longitude 69° 08' West of Greenwich. 



There has recently appeared an account of an 

 analysis of the earth's magnetic field, also for 1922, 

 by Sir Frank Dyson and Prof. H. H. Turner.* These 

 authors reach conclusions which apparently are at 

 variance with mine given in (i) as to the definite 

 existence of the two systems, E and N. However, a 

 critical examination of the residuals obtained by 

 them when they endeavour to represent the rect- 

 angular components, X, Y, Z, on the hypothesis of a 

 magnetic potential due alone to systems below the 

 earth's surface, and assuming that a non-potential 

 system does not exist, is found, in fact, to strengthen 

 my conclusions.* 



Question (a) [Non-potential System, N). — The exist- 

 ence of the N-system implies the non-vanishing of the 

 line integral of the magnetic force taken around 

 a closed circuit on the earth's surface. Such line 

 integrals have been computed for large land areas, like 

 the United States, and for very large ocean areas, with 

 data from the cruises of the Carnegie, both in the 

 Northern and Southern Hemispheres. The results 

 are so consistent that they cannot be accounted for 

 wholly by observational errors. The analysis of 

 the earth's magnetic field shows that the coefficients 

 derived from the east- west component, Y, will not give 

 an entirely satisfactory representation of the south- 



' The value of the magnetic moment frequently found in text-books, as 

 dependent on Gauss's analysis for 1830, is 8-55 x 10" C.G.S. The average 

 annual rate of loss between 1830 and 1922 is about 1/1500 part, thus cor- 

 responding with the annual average rate as given in (2). 



* Mon. Not. Roy. Astr. Soc., Geophys. Sup., vol. i. No. 3, May 1933, 

 pp. 76-88. 



• Terr Mag. and Attn. EUcL for March-June 1923, pp. 24-28. 



NO. 2808, VOL. I 12] 



north component, X. A similar experience has been 

 encountered in recent analyses of the diurnal variation 

 of the earth's magnetic field, of magnetic disturbances, 

 and of eclipse effects. Thus the evidence is in favour 

 of the existence of non-potential magnetic systems. 

 The difficulty has been in the adequate physical 

 interpretation of the results. According to classical 

 theory, line-integral values are a measure of electric 

 currents passing perpendicularly through the area 

 enclosed l)y the circuit. The average strength of 

 such indicated currents for the earth's magnetic 

 field is found to be more than 10,000 times that of the 

 vertical conduction current of atmospheric electricity. 

 The average strength of vertical currents that may m 

 part be responsible for the magnetic diurnal variation 

 is about 2000 times that of the currents causing the 

 diurnal variation of atmospheric electricity. We are 

 then forced to conclude that the magnetic line- 

 integrals are a measure of something else than is 

 recorded by atmospheric-electric instruments. Various 

 suggestions are at present receiving careful considera- 

 tion. The very interesting point was recently raised 

 by Sir Arthur Schuster that no one, so far as he knew, 

 had experimentally verified the generally accepted 

 hypothesis that the magnetic force was accurately 

 at right angles to the current which produced it, and 

 he further remarked that he had very recently come 

 across the statement that according to Einstein's 

 theory the force and the current should not be exactly 

 at right angles. But there are at present difficulties 

 in trying to attribute the observed non-potential 

 effects wholly to such a possible relativity cause. 



The general system of vertical currents for the 

 earth's field is as follows : negative electricity flowing 

 into the earth in polar regions and flowing out in 

 lower latitudes ; for positive electricity these direc- 

 tions would of course be reversed. The system of 

 vertical currents is unsymmetrical both about the axis 

 of rotation and the equator. (A similar system of 

 vertical currents will explain the present facts of 

 the annual variation of atmospheric electricity.) 

 Enough has been said to show of what extreme interest 

 the final elucidation of the magnetic non-potenti*d 

 effect is likely to be. 



Question [b] [External Potential System, E). — This 

 system is disclosed by the fact that the coefficients 

 determined from the horizontal components, X and Y, 

 will not reproduce completely the vertical component, 

 Z, but will leave outstanding effects of a character 

 which, according to classical theory, can only be ex- 

 plained by an external system of electric or magnetic 

 forces. However, if any portion of the earth's total 

 magnetic field is to be' attributed to causes which 

 involve relativity effects, R, then E, in whole or in 

 part, may have to be regarded as resulting from R. 

 It is hoped that a special investigation now under way 

 will throw further light on this interesting question. 



As the result, apparently, of the extensive increase 

 in knowledge of the earth's magnetic field over that 

 at the command of previous analysts, the coefficient 

 of the first degree zonal harmonic is found to be three 

 times that resulting from Schmidt's careful analysis 

 for 1885. The magnitudes of Schmidt's coefficients 

 for the various zonal harmonics were such that he 

 did not deem it safe to draw a definite conclusion as 

 to the reality of an external system. The case is 

 different, however, for our 1922 analysis ; as stated 

 under conclusion (i), we can no longer ignore the exist- 

 ence of effects similar to those from an external system. 



Question [c) [Relativity Effects). — This question has 

 already received some attention in the consideration 

 of questions (a) and [b) ; it also enters into question [d). 



Question [d) [Is the Integral of dfi Zero?). — All 

 analysts beginning with Gauss have assumed that 

 the integral of df>. is zero. The gissumption enters not 

 only into the determination of the coefficients of the 



