DEPARTMENT OF TERRESTRIAL MAGNETISM. 333 



The above conclusion is not intended as a criticism of Schuster's theory, 

 however, since, as Schuster himself points out, the ultra-violet light is not 

 the only source of ionization in the upper atmosphere. Further, in the paper 

 here abstracted, it is shown that, if the calculation is not limited to a shell, 

 and account is taken of the infinite extent of the atmosphere, the magnetic 

 effects which result as the ultimate consequence of a feeble source of ioniza- 

 tion may be very much greater than those calculated on the basis of a shell 

 of finite thickness. 



Reply to E. H. Nichols's article (investigation with regard to the induced charge on the 

 Ebert electrometer at Kew Observatory). W. F. G. Swann. Terr. Mag., vol. 21, 

 pp. 99-102, 1916. 



In a previous paper (Terrestrial Magnetism, vol. 19, pp. 205-218, 1914), 

 the author developed a theory of the action of the potential-gradient in 

 modifying the results obtained in certain atmospheric-electric instruments. 

 One of the instruments under discussion was the Ebert electrometer. To 

 illustrate his theoiy the author applied it to the experimental data obtained 

 from one type of Ebert instrument, and found that in the case of this instru- 

 ment a potential-gradient of 70 volts per meter would result in an error of 

 26 per cent on the measured ionic density of the negative ions, if the specific 

 velocity of the ion were assumed to be 1.5 cm. per second per volt per centi- 

 meter. 



In the June number of Terrestrial Magnetism and Atmospheric Electricity 

 for 1916 (pages 87-98), E. H. Nichols quotes the results of his analysis of the 

 Kew data for several years, and also the results of special experiments made 

 by him for the pui-pose of investigating the error in the Ebert instrument. 

 He comes to the conclusion that when the instrument used at Kew is provided 

 with the cap supplied by the makers, the error in question is much smaller 

 than 26 per cent. He concludes that Swann's theory of the matter is of 

 limited application, that it omits to consider the effect of the wind and of 

 the cap, and that it applies only to a cylindrical opening. When the cap is 

 absent, however, he finds that errors as high as 40 per cent may result, and 

 admits that under these conditions his results appear to support Swann's 

 conclusions. 



In Swann's reply to the above criticisms he points out that his theory of 

 the matter does not neglect the effect of the wind, nor is it restricted to a 

 circular opening, or to a case where the instrument is unprovided with a cap, 

 and he cites references to his original paper showing that these matters have 

 been included in his theory. He points out, however, that his formula gives 

 the error in terms of a quantity which must be experimentally determined for 

 the apparatus used. His own experiments were made primarily for illustra- 

 tion; the instmment used was not provided with a cap, for a large amount 

 of published material has been obtained with such instruments, and even in 

 some modern forms of Ebert electrometer there is no cap. His theory, how- 

 ever, in no way denies the possibility of the existence of a much smaller error 

 in the type of apparatus used at Kew. In fact, his formula provides no 

 information whatever as to the magnitude of the error until the experimental 

 data appropriate to the instrument have been substituted in it. Further, 

 he calls attention to the fact that his theory shows the error to be proportional 

 to the specific ionic velocity, and that, as pointed out in his paper, he used 

 the value 1.5 cm. per second per volt per centimeter for this quantity. If he had 

 used the very low value 0.5 cm. per second per volt per centimeter, which E. H. 

 Nichols quotes as appropriate for Kew, he would have calculated an error of 

 only 8.7 per cent instead of 26 per cent. Quite apart from all other con- 



