July 14, 1910] 



NATURE 



43 



of the international observations under the direction 

 of Prof. Hergesell. The author's primary object was 

 to deal with the annual variation of temperature, but 

 he has found room also for the consideration of many 

 associated questions. Altogether 3S0 ascents were 

 considered, all of which reached 8 km. and twenty- 

 nine of which reached 16 km. Doubtful observations 

 were rejected. 



The principal features in the annual variation of 

 temperature are as follows. From the surface up to 

 3 km. the date of minimum temperature gets later 

 and the annual range decreases by about 4° C. 

 From 3 to 10 km. the minimum temperature occurs 

 at the beginning of March, but at still greater heights 

 there is a comparatively sudden jump back to the 

 beginning of January. The annual range increases 

 from 3 to 7-8 km. by about 4° C, then decreases up 

 to 10 km. by about 6° C, and finally re- 

 mains nearly constant from 11 to 16 km. 

 The results agree, on the whole, with those 

 obtained by the present writer and Harwood from a 

 slightly dilTerent period of observation. Dr. Wagner 

 deduces, from a consideration of the first two terms 

 of the Fourier series expressing the variation, that 

 the difference betw-een the maximum and the mean 

 temperature exceeds that between the mean and ihe 

 minimum, and that this asymmetry increases with 

 height; it appears doubtful if it is justifiable to 

 neglect the third term, which increases with height 

 and tends to diminish the asymmetry mentioned. 



The effect of water vapour on the gradient of tem- 

 ' perature is shown in thf differences between winter 

 and summer. The following table gives the gradients 

 for summer (June, July, August) and winter (Decem- 

 ber, January, February!, (i) from the present paper; 

 (2) from the report of the present writer and Har- 

 wood ; (3) for ascending saturated air : — 



From 3 to 8 km. ihe gradient is less in summer 

 than in winter, while the difference between the 

 " saturated " adiabatic gradients is greatest from 2 to 

 8 km. The approximation to the adiabatic state is 

 closer in summer than in winter. 



Dr. Wagner attributes the annual variation to con- 

 vection and conduction from the earth's surface, to 

 condensation of water vapour, and to radiation, solar 

 and terrestrial. A further cause ought to be in- 

 cluded, viz. the transference of energy in a horizontal 

 direction. The effect of conduction might fairly be 

 neglected, since even at 100 m., if conduction alone 

 were active, the amplitude of the yearly variation 

 would be less than i 'loooth of the amplitude at the 

 surface. The decrease of the amplitude up to 3 km. 

 appears to be a result of the action typified by 

 v. Bezold's law. The increase above 3 km. is probably 

 rightly attributed to the effect of the increased water 

 vapour on the average gradient in the summer 

 months. Condensation of water vapour is, more- 

 over, held responsible for the relativelv slow cooling 

 of the middle layers from summer to autumn, but it is 

 probable that the above-mentioned horizontal trans- 

 ference of energy and the radiation also contribute to 

 NO. 2124, VOL. 84] 



this effect. The radiation tends to increase the tem- 

 perature of the earth and lower atmosphere when the 

 amount of water vapour is increased, if the effect is 

 not counterbalanced by increased reflection of solar 

 radiation from clouds. In this connection it may be 

 pointed out that there is no experimental evidence to 

 justify the assumption repeatedly made that the air 

 between 3-4 and 8 km. may be regarded as diather- 

 manous. At 5 km. the average vapour pressure is 

 not much below i mm., and the experiments of 

 Paschen and Rubens and Aschkinass show that for 

 a vapour pressure of i mm. half the radiation of a 

 full radiator between 12 and 20/i would be absorbed 

 by a path of about 400 m., and half that between 5 

 and 8/i by a path of 50 m., while the CO, absorption 

 would add slightly to the absorption in 'the former 

 region ; and these results are affected but little by the 

 later experiments of Scheiner and von Bahr. 



Dr. Wagner finds that the departures of the tem- 

 perature in different localities from the general mean 

 values are small except for south Europe, where the 

 temperature is considerably above the mean in the 

 convective region, and for east Europe, where the 

 converse is the case. The peculiarity in the latter 

 region is largely due to the influence of Pavlovsk, 

 lat. 60°, which is the only station in the region 

 besides Koutchino, lat. 56°. 



The mean value of H, , the height at which the 

 advective region is reached, for different regions is 

 as follows : — 



1054 



1018 



11-07 



West Europe 

 1062 km. 



Dr. Wagner deduces from these results that the 

 value of H^ decreases from ocean to continent, as 

 well as from equator to pole. It is true that radiation 

 effects alone would tend to make H^ less over a dry 

 continental area than over the ocean at the same or 

 a higher temperature, but it is doubtful if such an 

 effect can be traced in these figures, according to 

 which north Europe (Berlin and Hamburg) has a 

 lower value than east Europe (Pavlovsk and 

 Koutchino). 



In considering the variation of temperature with 

 the pressure distribution. Dr. Wagner wisely adopts 

 the plan of eliminating the annual variation, and as 

 he uses no ascents from east or south Europe, the 

 correction for the local variation of temperature is 

 inconsiderable. It ought, however, to be remembere,d 

 that, although the mean temperatures of the year are 

 not far different, say, for Paris and Vienna, the cor- 

 rections to be applied to ascents made in the same 

 month at those two places are not necessarily the 

 same. Dr. Wagner's results corroborate those pre- 

 viously found in proving that cyclones are in general 

 colder than anticyclones, but a consideration of special 

 cases led to the important conclusion that for rapidly 

 moving systems these conditions were reversed, a 

 result foreshadowed by the work of Hanzlik. 



The mean temperature in October at 2 km. over 

 Berlin on p. 05 is wrongly given as o'6° C, and this 

 error is mainly responsible for the peculiar change in 

 the half-yearly variation at that height. In differen- 

 tiating Ab on p. gq the variation of T is not 

 negligible. It is simpler to proceed from the funda- 

 mental equations, which lead to the result that the 

 height at which Ab is a maximum is given by 



/i = RT.,-/T,,= RT„ nearly, 

 where 



_i_ _ f/i dh 

 T,r~./„ '1^ 



and T,, is the temperature at the height h and T„ 

 that at the surface. 



