780
ATMOSPHERIC
tached to figures showing the diurnal inequality of thunderstorms is open to a little uncertainty. The data in Tables VIII. and IX. are all quoted by Thunder- Arrhenius {Met. Zeit. 1888, p. 348), except those storms. for Edinburg]lj which are from a paper by Moss-
ELECTRICITY Table XL Total deaths by Lightning.
man {Trans. R. S. E. vol. xxxix. p. 63). The data for Belgium are due to Lancaster, those for Italy to I errari; England and Wales 10 46 127 114 93 30 23 443 in the latter case only storms lasting less than three hours (24 years) are included. The data for the Russian empire are due to 116 337 641 683 484 153 82 2496 United States (8 years) Klossousky j they are based on ten years’ observations, 1870-79. In all cases the figures represent percentages of 29. If we compare Tables VIII., IX., and X., with the total number, afternoon hours are dashed. Tables L, II., and III., we see that to a certain Table VIII. extent high thunderstorm frequency and low potential gradient vary in parallel lines. Thus, Relative Diurnal frequency of Thunderstorms. thunderstorm frequency and intensity are greatest 10'—12' 8—10 10—12 O'—2' 2'—4' 4'—6'i 6'—8' 8'—10' Interval. 0—2 2—4 1—6 about midsummer, when potential gradient is least, 2-0 and there is at most stations a minimum of potential 9-2 5T 4-7 14-2 22-4 23-7 11-9 15-8 Edinburgh 1-7 4-1 8-4 19-4 21-6 6-4 12-9 2-0 Belgium . 3-0 1-5 gradient in the afternoon at about the same hour as 3-0 8-5 19-5 26-5 16-6 9-8 8-3 Italy 1-3 the thunderstorm maximum. The connexion is, however, far from close. Thus we have the thunderstorm Table IX. minimum occurring alongside of the early morning potenRelative Diurnal frequency of Thunderstorms. tial gradient minimum; and the diurnal and annual I 9'—12' variations in thunderstorms appear much more accentuated O'—3' 3'—6' 0—3 Interval. than the corresponding variations in potential gradient. 6 3- 11- 28-3 22-6 24-5 7- 9 8 13 North. Russia § 30. When we consider the influence of geographical position, 3 3 4- 14-1 23- 29-5 14-9 8- 5 1 Baltic Provinces, 3-3 2we also find thunderstorm frequency more variable than potential Finland gradient, at least so far as is at present known. If we accept 10-1 3-1 4-6 3-4 6T 21-0 29*9 21-8 West Russia Exner’s observations of potential as representative, the mean 5 3-3 8-7 26-6 27- 20-8 6-9 1 3-1 3Central ,, potential gradient does indeed diminish as we pass from temperate 98 2 7 3- 2-6 2-2 8-3 24- 28- 20-6 6 Ural . to tropical climates, but we do not encounter such notable dis5 2-6 9'2 25- 28-6 18-3 10- 0 1 Siberia 27 314-6 crepancies as are presented in Table XII. The data are from the 4- 1-2 0-9 4-1 14-6 31-8 28-3 5 Caucasus . sources already mentioned. 11- 9 0 9 5- 4- 11-6 12- 25-8 16-6 11-7 5 South Russia Table XII. 3-5 3-0 3-1 8'5 23-6 28-3 20-4 9-6 Mean. Average Number of Thunderstorms in a Year. Edinburgh, London, Paris, Batavia, 1770-1896. 1763-1896. 1873-93. 1867-95. § 28. Table X. shows how the number of thunderstorms 6-4 10-7 27-3 120 varies throughout the year. The figures are percentages It must be allowed that the drawing of safe conclusions from of the total number recorded. The epochs covered and thunderstorm data is particularly difficult, owing to the differencesthe sources of the figures are as follows: Edinburgh, between the records from stations in the same latitude. Instructive 1770-1896, Mossman {Trans. R. S. E. vol. xxxix. p. 63); illustrations are afforded in Plate I. of the memoir by A. J. Henry, on United States thunderstorms mentioned above. Table X. In some parts of the Pacific states the average annual number is put at from 1 to 5, as against 30 to Relative frequency of Thunderstorms throughout the Year. 45 in some of the central and south-eastern states. April. May. June. July. Aug. Sept. Oct. Nov. Dec. Jan. Feb. Mar. Even in some quite small areas, especially islands or districts containing high mountains, the differEdinburgh . j 3-8 12-6 20-8 28-5 18-8 7'2 2-2 0-9 0-9 1-8 1-2 1- coast ences in thunderstorm frequency are great. There 0-5 1-6 5 London . I 6-6 12-7 18- 25- 19-2 9-3 3-1 1-7 0-9 0-6 3 Paris . . j 7‘5 14-9 21-6 22-0 17-0 9’9 3-5 0-4 0-4 0-2 0-4 2- is the further consideration that a great deal may Hungary . 5’7 16-2 25-8 23-3 17-0 6-9 2-4 0-6 0-3 O'l 0-4 1-3 depend on the observer. 12-2 10-9 10-4 9-2 11-1 Batavia . j 10'5 7-9 5-5 4-3 3-8 5-4 § 31. Other thunderstorm data are more independent of the imagination, but all seem attended by 0-2 West Siberia 0’8 11-5 28-0 36-4 20'1 3-0 sources of uncertainty. An illustration is afforded by 2-1 North Russia | 0'7 7-5 27-6 32-2 24-4 5’5 South ,, i 3'3 14-2 27-3 26- 15- 8-3 5-3 2 4 the following statistics from the paper by Lawson 18-2 1 Italy . . I 7'4 7-2 19- 17-2 16- 14-6 5 already mentioned, giving the average number of persons killed annually per ten million inhabitants London, 1763-1896, Mossman {Quarterly Journal of R. in various districts of England and Wales. The original treats the epochs 1852-61, and three following decades. Met. Socy. vol. xxiv. 1898, p. 31); Paris (Parc St Maur), separately Table XIII. gives only the final means. The letter M denotes 1873-93, Renou {Met. Zeit. 1894, p. 277); Hungary, “midland” ; N, north ; S, south, &c. mean from numerous stations, 1871-95, Hejas {Met. Zeit. Table XIII. 1899, p. 219); Batavia, 1867-95 {Mag. and Met. ObservaPersons hilled annually yer Ten Million Inhabitants. tions, vol. xviii. 1895); Russian provinces, 1870-79, and S.E. S.M. E. S.W. W.M. N.M. N.W. York. N. Wales. Italy (1882-83) from Arrhenius {Met. Zeit. 1888, p. 348). London. 1-3 8-0 12-3 12-7 5-6 10'7 18'0 6-6 11'4 10'4 9H In the last-mentioned cases the data for winter are lumped The north - midland district (Leicester, Rutland, Lincoln together, so for convenience the year starts with April. Nottingham, and Derby) showed the largest proportion of persons Alongside this last table it is instructive to place the killed during each decade, and London (metropolitan area) the following particulars as to the number of persons killed least. From what is known otherwise, there can hardly be any by lightning in different months in England and Wales doubt that one would be seriously in error if one supposed the in Table XIII. to be an exact measure of relative thunderduring the twenty-four years 1857-80 (R. Lawson, Quart. figures storm intensity. The figures are largely influenced by the fact Journal R. Met. Soc. vol. xv. 1889, p. 140), with corre- that persons employed indoors have much less chance of being sponding data in the United States during the eight years struck by lightning than those out of doors. As emphasizing this 1891-98 (Henry, Bull. No. 26, U.S. Dept, of Agriculture, fact, Lawson mentions that between 1852 and 1880, only 104 were killed by lightning, as against 442 males. 1899). The six winter months October-March are grouped females § 32. Table XIV. gives the mean annual number of thundertogether. storms recorded per decade at Edinburgh, London, and Tilsit*
