December 8, 19 lo] 



NATURE 



171 



Potsdam curves. Finally, I have also given the difference 

 {dy) between the greatest and the smallest time given in 

 iris's Table I. for the commencement in H at the 

 lerican stations (Greenwich mean time is employed). 



Mean of numerical values 



0-3S 



307 



It will be observed that all the differences are consider- 

 rably below the error-limit which, according to the above, 

 must be reckoned upon, and the difference is as frequently 

 one way as the other. ■" 



These figures seem to'me to show clearly that in these 

 cases the magnetic impulse occurs, at any rate, very nearly 

 simultaneously ; in any case there cannot be time-differ- 

 ences of such a magnitude as in Dr. Faris's opinion there 

 are— for July lo, 1907, he even assumes that the storm 

 would take ii-6 minutes to encircle the earth. Further, 

 we see that the greatest difference between Potsdam and 

 the mean of the American stations, o-88m., is only about 

 two-thirds of the smallest difference, oV, between the times 

 ■at the American stations, i-3m. This circumstance, and 

 the fact that the relation between the numerical means 

 of these time-differences is as 035 : 3-07, would seem dis- 

 tinctly enough to show that the great time-differences 

 observed by Dr. Faris can only be due to inaccuracy in 

 the determination of the time, and that the error-limit 

 must be considerable. 



Further, if we consider the foundation that Dr. Bauer 

 has employed for the determination of the rate of propa- 

 gation in the case of the storm of January 26, 1903, it 

 must, I think, be deemed as weak and uncertain as the 

 above-mentioned, which I was able to control. Birkeland, 

 5n speaking of the table employed (loc. cit., p. 63), says : — • 



" The table shows that the time varies so little with 

 the geographical position that it would be premature to 

 draw conclusions from it. The slight differences may be 

 ascribed to inaccuracies in the determinations of time on 

 the magnetograms ; for we see that if a difference in time 

 for a certain point appears between two places, this differ- 

 ence is maintained for all the points, a circumstance which 

 seems best to be explained by an inaccuracy in the state- 

 I nient of the time. We may conclude from this that the 

 serrations appear simultaneously, or rather, the differences 

 in time are less than the amount that can be detected by 

 1 these registerings. . . . The above question, which is of 

 :oat importance, cannot be definitely decided until we are 

 possession of rapid registerings." 



Bauer holds, however, that by taking groups of means 

 can demonstrate, clearly and surely, time-differences 



at would prove that the cause of the perturbation was 

 transmitted eastwards at a rate of 6400 miles per minute. 



I also last summer determined the commencement in 



H of this perturbation in Potsdam, and found the time 



I to be 8h. 53m. Greenwich mean time. I moreover had 



i;he opportunity of going through the curves upon which 



rkeland's table was based. From these it appeared that 

 times for the comparative correctness of which there 

 ■vas some guarantee were from the five following places : — 

 Toronto, Kaafjord, Potsdam, Dehra Dun, and Bombay. 

 As regards the other stations, it may be remarked that 

 , ;from Honolulu, Baldwin, and Cheltenham there were only 

 received Indian-ink copies without hourly or two-hourlv 

 automatic time-marks. The parallax there could not be 

 ci':'termined accurately, and the uncertaintj- in the time- 

 termination must be considered to be relativelv very 



--at. 



In the copy of the curve for San Fernando the base- 

 le was a little curved. In that for Batavia the curve 

 id the base-line were very faint ; the parallax could not 

 determined with sufficient precision, and the time-marks 

 ^re also rather indistinct. A new determination of the 



NO. 2145, VOL. 85] 



time of beginning which I have just made gives as the 

 result 8h. 52-4m. for San Fernando and 8h. 52-Sm. for 

 Batavia. In the table these times are given as 8h. 54-3m. 

 and 8h. 54-9m. respectively, a fact that demonstrates the 

 uncertainty which attaches to these hours. At Christ- 

 church it seems from the D and Z magnetograms as if 

 the clock on that day was about i-5 minutes too fast, so 

 that the value 8h. 54-8m. given in the table probably should 

 be reduced to about 8h. 53-3m. Further, the beginning of 

 the base-line and the time-marks for the H curve were 

 rather unsharp. 



In addition, it may be remarked that the thickness of 

 the curve at Bombay was considerable, about 09 mm., 

 thus causing the commencement of the storm to be some- 

 what less clear ; but, on the other hand, there were two- 

 hourly automatic time-marks upon the curve itself, a 

 circumstance which is of great importance in exact deter- 

 minations of time. 



If we now omit those that we already know to be very 

 uncertain, we find the following times of beginning, 

 putting Dehra Dun and Bombay together : — 



Toronto 



8 52-6 



Kaafjord 

 52 6 



Potsdam 



S3 



Diflf. 



Dehra Dun 

 and Bombay 



• 53-3 •■• 07 



Thus the greatest difference is considerably lower than 

 the error-limit, and this would be still less if, as would 

 indeed be best, we attach more weight to Dehra Dun, 

 where the curve is exceedingly clear, than to Bombay. If 

 we attach double the importance to the former, we find 

 53" im. instead of 53-3m., and the difference will then be 

 reduced to o-5m. 



It seems to me, also, that this last method, where the 

 conditions are as they are here, must give a far more 

 certain result than that which Bauer has employed. 



The remaining characteristic points on the curve seem 

 to me to be too indistinctly defined to allow of being 

 employed in cases where the differences are as small as 

 they are here. 



Of the storm of May 8, 1902, I have no special observa- 

 tions that could serve to control Bauer's result. As regards 

 Potsdam, however, I have a determination of its beginning 

 in H, which I also made last summer before reading 

 Bauer's paper. I found the time to be iih. 58m. Green- 

 wich mean time. Bauer, however, in his table gives it as 

 i2h. om. It seems to me that this difference of two 

 minutes is characteristic of the uncertainty that attaches 

 to these statements. When Bauer finds that the weighted 

 mean of all European stations is iih. 58-24m., it looks 

 as if my determination were the best. When such great 

 differences can be found in the measurement of the same 

 curve, and the Potsdam curves must, I supp>ose, be con- 

 sidered to be among the most trustworthy of all. how 

 great must be the uncertainty that attaches to the others? 



There seems from this, at any rate, to be by no means 

 sufficient data to justify the conclusion that the magnetic 

 storms are generally propagated round the earth in from 

 about 35 to 4 minutes, and the theory that Bauer mainly 

 bases upon this we must be allowed to regard with corre- 

 sponding scepticism. 



But even if there are no such great time-displacements 

 in these " abruptly beginning storms " as Bauer thinks, 

 there is, of course, a possibility that small time-displace- 

 ments might exist. This question, which is of such great 

 importance for a full comprehension of the nature of the 

 magnetic storms, can only, however, in my opinion, be 

 solved, as Birkeland has suggested, by rapid registerings. 

 It would be comparatively easy, moreover, to carry some 

 such arrangement into effect by means of a number of 

 stations — at least three — where a short or long period was 

 registered continuously with very sensitive apparatus and 

 with frequent and exact automatic time-marks upon the 

 curve itself. This was the more easy of accomplishment 

 from the fact that, for the solution of the present ques- 

 tion, it was only necessary to register H in this manner. 

 It would then be possible to obtain a sure foundation for 

 reflections of the kind that Bauer makes in his last paper, 

 reflections that, however interesting they may be, must, 

 from what I can understand, be said to be in no small 

 degree premature. O. Krogness. 



Universitetets fysiske Institut, Kristiania. 



