468 



NATURE 



[January 8, 1920 



LETTERS TO THE EDITOR. 



[The Editor does not hold himself responsible for 

 opinions expressed by his correspondents. Neither 

 can he undertake to return, or to correspond with 

 the ■write.rs of, rejected manuscripts intended for 

 this or any other part of Nature. No notice is 

 taken of anonymous communications.] 



The Deflection of Light during a Solar Eclipse. 



In discussing the effects of atmospheric refraction 

 during- solar eclipses Prof. Anderson disregards the 

 shallowness of the effective layer of air as compared 

 with the diameter of the moon's shadow. Unless the 

 sun be \iery near the horizon, a line of sight drawn 

 from the centre of the umbra to a point in the corona 

 will remain within the umbra right through this laver. 

 This consideration vitiates the method of solution 

 adopted by Prof. Anderson, and consequentlv its 

 results. On reading his first letter (Nature, Decem- 

 ber 4, 1919) I was struck by the ingenuity of his 

 explanation, more especially as I believe lie under- 

 valued the amount of the angular deviation arrived at 

 on his theory through taking the sun's radius to be 

 half, instead of a quarter of, a degree. In view of 

 the importance of the subject, a fuller investigation 

 seemed to be required. I hope soon to publish a 

 note giving the complete solution of the problem, 

 and may therefore confine myself here to a statement 

 of the result, which is quite fatal to Prof. Anderson's 

 explanation. I take the altitude of the sun to be 45° 

 and the maximum fall of temperature 4° ; the figures 

 given may easily be modified to suit other condi- 

 tions. I further assume the most favourable distribu- 

 tion of temperature, which is that adopted by Prof. 

 Anderson, when the line of maximum fall of tempera- 

 ture is parallel to the edges of the moon's shadow 

 and independent of altitude. Two stars at a distance 

 of three solar, diameters from each other might then 

 show an increase in apparent distance owing to refrac- 

 tion amounting to the 240,000th part of a second of 

 arc. If the diminution of the temperature effect with 

 altitude be taken into account, this figure should be 

 divided by 4. Arthur Schuster. 



The Magnetic Storm of August 11-12, 1919. 



The principal question raised by Mr. Evershed in 

 N.4TURE of January i, viz. the simultaneity of S.C.s 

 (sudden commencements of magnetic storms) at 

 different parts of the globe, has alreadv a considerable 

 literature. It has been discussed by Prof. S. Chap- 

 man and myself in the Proceedings of the Phvsical 

 Society (vol. xxx., p. 205; vol. xxvi., p. 137; and 

 vol. xxiii., p. 49). It scarcely admits, perhaps, of a 

 precise answer. S.C.s vary from one part of theearth to 

 another, not merely in size, but also in tvpe. In India, 

 for instance, they are normally unidirectional, and 

 much larger in H (horizontal force) than in the other 

 elements. At Kew, and still more at Eskdalemuir, 

 they are often oscillatory, the main movement, a ri.se 

 in H, being preceded by a shorter and smaller fall. 

 In the Antarctic (.Scott's stations) they seem to be 

 always oscillatory and of similar magnitude in the 

 different elements. The time when a movement be- 

 comes visible depends on its size and the sensitiveness 

 of the magnetograph. Magnetographs differ widely in 

 sensitiveness and vary in type. An oscillatorv move- 

 ment that is very small or of very short period cannot 

 be recorded by an ordinary magnetograph. 



Whether the time of the S.C. is affected bv the 

 meridian position of the sun, i.e. by the local time, 

 has been discussed by Prof. Chapman. Whether the 



NO. 2619, VOL. 104] 



results he got implied any real difference is a matter 

 of opinion, but if any difference existed it was a 

 question, not of minutes, but of seconds. If any 

 difference existed in the times, one would expect it to 

 be at least as conspicuous in the amplitudes. As 1 

 have lately shown, the type of the S.C. recorded in 

 the Antarctic does seem to depend on the local time. 

 Eleven S.C.s which occurred between iih. 59m. and 

 lyh. 2om. G.M.T. agreed in type; while six which 

 occurred between 2ih. 3m. and 23h. 25m. also agreed 

 in type ; but the two types were fundamentally 

 different. The first class represent noon and the 

 earlier afternoon at Eskdalemuir, but midnight or 

 early morning in the Antarctic; while the second class 

 represent hours near noon in the Antarctic. Complete 

 measurements of the horizontal amplitudes of the 

 S.C. movements exist for eight of the first and three 

 of the second class. Dividing the sum of the Esk- 

 dalemuir movements by the corresponding Antarctic 

 sum, we get 0-43 for the first class and 0-42 for the 

 second, the mean from all the S.C.s of which I have 

 complete records at both stations during 1911-12 

 being 042. The values of the ratio vary greatly for 

 individual S.C.s, so that the coincidence in the above 

 figures must be largely accidental. But, at all events, 

 it seems incompatible with any conspicuous influence 

 of local time on the amplitude. 



.\s to the particular .S.C. of August 11-12, I9i(), 

 when first measuring the I\ew curves I made the 

 tiine slightly before yh. G.M.T., whether one or two 

 minutes before I now forget. Remeasuring it now, 

 with as little prejudice as possible, I make the time 

 6h. 58m., agreeing with the value got by Dr. Crichton 

 Mitchell for Eskdalemuir. The time-breaks at Esk- 

 dalemuir occur on the curve itself, so the estimate 

 there is free from the uncertaintv to which, I presume, 

 Mr. Evershed refers, which is usuallv known as 

 "parallax" between the curve and time lines. This 

 source of uncertainty is also practicallv non-existent at 

 Kew, but the train and tram disturbances now experi- 

 enced there make all measurements less certain than 

 thev used to be. The S.C. on .August 11 was, however, 

 so large, and the discontinuity in the H curve so con- 

 spicuous, that I think the uncertainty might fairlv be 

 put at +0-5 minute. The uncertainty of the ordinary 

 measurement at the average observatory, even for 

 S.C.s, is certainly not less than this, and is probablv a 

 good deal larger. C. Chree. 



Relativity and Radio-activity. 



With regard to some of the postulates of relativity, 

 it seems interesting to ask if radio-active instabilitv 

 might not be capable of providing a timekeeper which 

 would retain its uniformity independentlv of motion 

 relative to the aether. 



.\s to how such a clock might be made practical or 

 whether it must remain theoretical is beside the 

 present question. So also is the degree of accuracy 

 which might be attainable. Primarily, we might sup- 

 pose the radio-active clocks rated one with another by 

 a simple count of the a-rays emitted over a certain ; 



solid angle and during a certain time interval, the ] 



clocks being in the one locality. Thereafter these 

 clocks would serve to define simultaneity in widely 

 separated localities, the diminishing quantitv of the 

 radio-active substance notwithstanding. 

 • .'Vs I say, the primary question is not so much one 

 of practical application, but as to whether it would 

 be theoreticallv possible in this wav to observe motion 

 relative to the aether. 



Or is radio-activitv also "in the conspiracy"? 



.T. .TOLY. 



Trinity College, Dublin. 



