33^ 



NATURE 



[June 26, 19 19 



Royal Meteorological Society, June i<S. — Sir Napier 

 Shaw, president, in the chair. — Sir Charles Close : 

 Note on the rainfall at Southampton and London 

 during a period of fifty-seven years ( 1862-19 18). The 

 variations in rainfall in England are so great that 

 any seasonal period can be detected only by the study 

 of many years' statistics. Even when the statistics 

 are available for a long period, the form in which they 

 are usually published does not readily lend itself to a 

 clear appreciation of the existence of <a simple seasonal 

 period. Thus the monthly means are usually uncor- 

 rected for variation in the lengths of the months, and 

 the custom of treating the months separately produces 

 an effect of discontinuity. If, however, after cor- 

 recting for monthly inequalities, the accumulation of 

 rainfall, reckoning from any fixed date, is tabulated 

 and plotted, the rainfall assumes a more regular 

 aspect. If, further, from these monthly figures of 

 accumulation we deduct the average precipitation, the 

 remaining figures approximate to a simple sine-curve 

 with an annual period. The irregularities left over 

 occur chiefly in September and October. The fifty- 

 seven years' rainfall at Southampton, from 1862-1918, 

 have been examined in this way, and the London 

 rainfall for the same period. For Southampton, 

 counting from April i (but any date jvill do), the 

 accumulation, in inches at n months, as represented 

 by the expression 



2-63 X n-o-95-i-35 sin (n x 3o°-45°). 

 For London by the expression 



2-i3n — 07 sin (n X 30°). 



The maximum irregularities left over" amount to 

 0-30 in. and 020 in. respectively on October i. It 

 would appear, then, that at the places in question the 

 rainfall can be considered to result from uniform 

 precipitation throughout the year, rnodified by a 

 simple annual harmonic term, further modified by 

 small irregularities in September and October. — 

 Lieut. J. Logic : Note on tornadoes. The paper aimed 

 at showing that no convection currents are capable of 

 producing tornadoes of the intensity claimed for some 

 of these storms. Working from the equation 

 dpldh= —gD (which is shown to be sufficiently 

 accurate for the purpose in hand, even in a tornado- 

 centre), and assuming that at some height the pres- 

 sure above the tornado is equal to that at the same 

 level outside, the author computes the difference of 

 temperature between the air in the centre of the 

 tornado and that outside. For a tornado having a 

 pressure reduction of 50 millibars at the surface the 

 mean temperature difference is found to be 23° A. if 

 the tornado extends only to 5 km. (16,000 ft.), 10° A. 

 if it extends to 10 km., and 5° A. if it extends to 

 15 km. From the known values of the lapse-rate of 

 saturated air, it follows that under conditions of 

 maximum instability a saturated ascending current 

 not less than 8 km. high might produce a tornado 

 of this intensity. Since such instability rarely occurs, 

 and, in addition, ascending currents of saturated air 

 are usually everywhere penetrated by descending 

 masses of cooler air, even a tornado of this intensity 

 is unlikely to be so produced in natural conditions. 

 The case of a 2:;o-millibar reduction is also con- 

 sidered as being at times actually achieved. In this 

 case the temperature difference, even if the tornado 

 reaches 15 km., is shown to exceed 35° A., a differ- 

 ence not capable of being produced by the release 

 of latent heat due to condensation of cloud, and still 

 Jess likely to be caused by simple heating of the 

 ground surface. It is suggested that the required 

 rise of temperature mav be due to the lightning 

 which IS usuallv described as a characteristic of the 

 funnel-cloud. — Capt. D. Brunt : A periodogram 



NO. 2591, VOL. 103] 



analysis of the Greenwich temperature records. The I 

 monthly mean temperatures at Greenwich for the I 

 years 1841-90 were taken and represented by a * 

 Fourier series up to 100 terms, so as to permit of 

 the detection of any periods of length greater than 

 one year. Periods of 9-5 years, 5 years, 4 years, 

 23 months, and 20 months were shown to exist, all : 

 having amplitudes of the order of 05° F. Some of 

 these correspond with periods found in other meteoro- 

 logical records, e.g. the 20-month period • has been 

 found by Prof. Turner in the rainfall records of 

 Greenwich and Padua. The interval covered by the 

 observations was insufficient to permit a detailed dis- 

 cussion of periods of length greater than about ten 

 years. Many of the periods found were not con- 

 tinuous during the whole interval covered by the 

 observations, e.g. the 20-month period died awav 

 about 1894, being replaced by a period of about 

 23 months. The general result of the investigation 

 was to show that periods in astronomical sense do 

 not exist in these temperature records. It was shown 

 that the effect of correcting the observations for the a 

 effect of the periods found was to produce an almost I 

 inappreciable diminution of the standard deviation of 9 

 the observations, tending to show that the variations 

 of the monthlv mean temperatures from year to year 

 are to be regarded either as purely chance variations 

 or as due to periods of length less than a year.— 

 Lieut. G. Green : The propagation of sound in the 

 atmosphere. Sound-waves emitted by a source 

 situated on the earth's surface generally undergo re- 

 fraction as they advance owing to the changing 

 conditions of wind velocity and of temperature as 

 the waves pass from layer to layer of the atmosphere. 

 In the paper a mathematical discussion is given of 

 the paths of sound-rays issuing in all directions from 

 a source, under certain conditions of wind velocity 

 and temperature closely resembling the conditions 

 generally to be observed in the atmosphere. The 

 mathematical results obtained would make it possible 

 to calculate the mean speed of sound in any chosen 

 direction from a source to a recording instrument at 

 a given distance from it, proyided observations of the 

 horizontal velocity of the w-ind and of the mean tem- 

 perature of the air have been taken at the earth's 

 surface and at various heights above the surface. 

 The mean horizontal velocity of a sound element in 

 tracing out a rav is equal to the sum of the velocity of 

 sound and the horizontal component of wind velocity 

 in the plane of the ray, taken at a height above the 

 earth's surface equal to one-third the total height 

 reached by the rav. Numerical results are given for 

 special cases to illustrate the effect of a given wind 

 gradient on the propagation of sound in all directions 

 from a source ; and the effect of the same wind 

 gradi'^nt combined with a favourable, and also an 

 unfavourable, temperature gradient. The case of a 

 wind increasing? in velocitv and veering as we ascend 

 from the earth's surface is illustrated in the closing 

 section of the paper. 



Institution of Mining and Metallurgy, June 19. — Mr. 

 H. K. Picard, president, in the chair. — W. H. Good- 

 child : The genesis of igneous-ore deposits. The 

 primary object of this paper is to provoke discussion. 

 Its scope is limited to the synoptic presentation of a • 

 few of the more important principles and processes 

 concerned in the formation of ore deposits from rock 

 magmas, together with an outline of sundrv' more or 

 less novel methods for elucidating the nature of those 

 processes. Starting out with the fundamental prin- 

 ciple that " the meaning of a vast number of the struc- 

 tures with which the geologist is confronted — either in 

 the field on the grand scale, or in the laboratory with 

 his small specimens — cannot, as a rule, be correctly 



