July 8, 1920] 



NATURE 



581 



the actual data when they come to hand. A com- 

 parison of the two would soon show that weather 

 prediction for more than a few hours ahead was im- 

 possible in present conditions. 



While, however, long forecasts are, for the most 

 ])art, mere untrustworthy guesswork, there are many 

 meteorological subjects now neglected which might be 

 investigated with success, but concerning which at 

 present practically nothing is known. Such are the 

 origin of the variation of electric potential in the air, 

 the origin of thunderstorms and lightning, the 

 coalescence or non-coalescence of cloud particles, the 

 origin of hail and the causes which determine the 

 -liape and size of snow cr\'stals or the volume of rain- 

 Irops, the forms of clouds, and many others. Also 

 I here are more general questions still to be answered 

 concerning trade winds and the circulation in equa- 

 torial regions. .Ml investigations on these subjects 

 should include the proper scales of comparison and 

 attempts to produce corresponding phenomena on a 

 ^mall scale. .\. Mallock. 



New I'niversity Club, June 28. 



The Rate of Ascent of Piiot°Balloons. 



In Natl'RE for June 17 Dr. van Bemmelen directs 

 attention to the excess rate of rising which pilot- 

 balloons often show in the first few minutes of their 

 ascent, and refers to two explanations of this pheno- 

 menon which have been put forward. These are that 

 the rapid rising may be due (i) to turbulence in the 

 lower layers of air or (2) to the tendency of balloons 

 to be drawn into rising columns of air and thus to 

 partake of their upward motion. The curves repro- 

 duced by Dr. van Bemmelen, which indicate the rela- 

 tion between rising velocity and height under different 

 conditions, are of great interest, and show that the 

 < ffect is not found when working on a small island 

 in the Java Sea. 



.As double-theodolite observations over a sea exposure 

 are not numerous, it mav be of interest to refer to 



the risuli> obtained by Capt. C". J. P. Cave and the 

 present writer in some ascents made from the Scilly 

 Islands during two winter months, November and 

 December, some years ago. Particulars have recentiv 

 been published by the Meteorological Office in 

 (ieophysical Memoir No. 14. The mean rate of 

 ascent of the balloons used was found to be 160 metres 

 j>er minute. Mean departures from this value for 

 each minute of the ascent measured from the stArt 

 are shown in Fig. i. Dr. van Bemmelen 's 

 diagrams for the Thousand Islands and Batavia 

 NO. 2645, VOL. 105] 



(0-3 p.m.) are also reproduced for comparison. 

 It will be seen at once that the rate of ascent 

 at Scilly, like that at the Thousand Islands, shows 

 no e.xcess above the normal in the first kilometre of 

 height; if anything, the effect is slightly the other 

 way. 



.Ascents in the Scilly Islands are of particular interest 

 in this connection. The area of the islands is so 

 small that no convection effects due to solar heating 

 would be expected, at .any rate in the winter. On 

 the other hand, tTie group contains a great nuinber of 

 small islands of a rocky and hilly nature, and these 

 are spread over an area of some ten miles by five. 

 Thev might naturally be expected to produce some 

 turbulence in the air passing over them, and such 

 turbulence is, in fact, shown by the records of 

 the pressure-tube anemometer on St. Mary's, the 

 largest of the islands. If the excess rate of rising so 

 frequently noticed in the first kilometre over land is 

 due to turbulence, as suggested by Wenger, we should 

 expect to find it in the Scilly ascents ; if it is 

 due to convection currents caused by solar heating, 

 we should not expect to find it. The evidence afforded 

 by this example seems clear. J. S. Dines. 



66 Sydney Street, S.W.3, June 22. 



Diamagnetism and the Structure of the Hydrogen 

 Molecule. 



In a letter to Naturk of June 24 (p. 516) Dr. J. R. 

 .\shworth has pointed out a possible origin of the 

 diamagnetism of hydrogen by assuming oscillations or 

 rotations of Bohr's paramagnetic hydrogen atbm or 

 molecule. Granted that such motions tend towards 

 a diamagnetic effect, it is important to exafnine the 

 plausibility of such a view in the light of recent 

 experimental data. We know that : 



(i) The specific susceptibility (xh) of gaseous 

 hydrogen at 16° C. is — 19-8(2) x io-'±o-i5 x 10-', 

 with a mean error of 076 per cent. (Take Son^, 

 Science Reports, Tohoku, vol. viii., p. 115, 1919). 

 No variation of this, unthin the li^nits of experiment, 

 could be detected oi^er a pressure range of i to 68 

 atmospheres. 



(2) The value of Xh ^O'' liquid hydrogen at a tem- 

 perature less than -253° C. is -27 x 10-' (Onnes and 

 Perrier, Proc. Amsterdam Acad., vol. xiv., p. 121, 

 1911). 



(3) The value of Xh for atomic hydrogen in various 

 types of chemical combination, as deduced from the 

 additive law of atomic diamagnetism for the hydro- 

 carbons, is -30-5x10-' (Pascal, Ann. de Chini. et de 

 Phys., vol. xix., p. 5, 1910). 



(4) There is no definite evidence that the diamag- 

 netic susceptibility varies simply with temperature 

 over a range - 180° C. to 20° C. Such small varia- 

 tions as do occur never change the sign of x (except 

 in the cas<' of tin), and are attributable to changes of 

 inolecular grouping, e.g. crystallisation or aggregation 

 (Ishiwara, Science Reports, Tohoku, vol. iii., p. 303, 

 1914; .\. E. Oxley, Phil. Trans. Roy. Soc, vol. ccxiv., 

 .\, p. 109, 1914). 



(5) The theory of molecular rotation developed bv 

 Honda and Okubo (Science Reports, T6hoku, vol. vii., 

 p. 141, 1918), which is similar to that proposed by 

 Dr. .Ashworth, accounts for the diamagnetism of 

 hydrogen and helium only if we suppose molecular 

 rotations of angular velocity 6-54 x 10'* sec -' and 

 3-80 X 10'" sec -' respectively. In the case of the para- 

 magnetic oxygen molecule it is necessary to suppose 

 (hat there is no rotation whatsoever in order to obtain 



