450 



NATURE 



[January i, 1920 



axis and are not independent cylinders. He states that 

 no case is linown to liim in which the idiomorphic 

 forms of the crystals are segments of spherical shells 

 which have crystallised out of a liquid except in the 

 ternary alloys referred to. The reason why idio- 

 morphic shell crystals develop under such conditions 

 and the laws which govern their formation await 

 further research. Meantime, Prof. Bragg, to whom 

 some of the separated crystals have been sent, has 

 kindly promised to study them. H. C. H. C. 



FORECASTING FROSTS. 



JN most countries during the spring, and to a lesser 

 extent in the autumn; there are periods in which 

 the meteorological conditions result in a frost. 

 Leaving out of the question spells of cold weather, 

 the prediction of which is the concern of a Meteoro- 

 logical Service, there remains the possibility of local 

 frosts in isolated districts, occurring on clear, windless 

 nights and lasting for a portion of the night and 

 early morning. These frosts are capable of doing 

 great damage to fruit-trees, etc., and the possibility 

 of forecasting them in time for the fruit-growers to 

 take precautions is of interest and importance. 



It has long been recognised that local cooling of 

 the soil can be largely prevented by a smoke pall 

 produced by the burning of damp materials such as 

 straw. Boussingault (" Economie Rurale," Paris, 

 1844) discusses this, and records an observation on 

 the point by Pliny. In America definite systems of 

 frost prediction have been in operation for some 

 years, and practical methods have been evolved by 

 which the grower can economically combat the 

 danger to his crops. A study of these preventive 

 measures is instructive in showing that several causes 

 are concerned in producing a frost. The methods are 

 varied. Leaving out those which attempt to delay 

 the flowering-time until the danger period is 

 past, they fall into four main classes: — (i) In- 

 creasing the water-content of the area (spraying 

 or flooding); (2) "smudge" burning (damp smoke 

 from wet straw, etc.) ; (3) temporary roofing ; and 

 (4) dry heating. The last method supplies additional 

 heat mainly; the other three are largely indirect, and 

 aim at reducing the rate of temperature fall either 

 bv increasing the heat capacity of the soil by the added 

 water or by restricting the radiation from the soil. 



Up to the present no complete correlation has been 

 made of frost in any particular locality and its causes. 

 For this purpose an examination by statistical methods 

 of a series of continuous observations (of the auto- 

 matic recording type) of meteorological factors is 

 needed. The published papers deal usually with cne 

 factor, such as dew-point or air temperature, and the 

 number of dailv obser\'ations \made is small. This is, 

 no doubt, due to the necessity of keeping the cost of 

 apparatus and working as low as possible for the 

 sake of the growers. However, a general idea of the 

 factors concerned can be obtained from a broad survey 

 of the various papers. 



The effect of an overcast sky on air temperatures 

 near the ground has been studied by Hellman 

 (Preuss. Akad. Wiss., Berlin, 38, 1918, p. 806), who 

 on clear nights found an exponential decrease of 

 temperature with height, the average difference from 

 ground-level to a height of 50 cm. being 37° C. An 

 increase of cloudiness by 1° of scale (o = clear, 

 io = overcast) diminished this difference by at least 

 J° C. There was no temperature gradient when the 

 sky was overcast, while windy and rainy weather 

 resulted in a slight reversal of the gradient. Schubert 

 (^fet. Ztsch., 32, 1915. p. 343) considers that during 

 the last half of the night the fall of temperature is 



NO. 2618, VOL. 104] 



relatively greater with dry than with moist air, owing 

 to the formation of dew, and frost from the latter 

 resulting in latent heat being set free by the con- 

 densation The presence of water-vapoiir in the 

 atmosphere al.so retards the radiation cooling of the 

 soil. 



J. Warren Smith {U.S. Monthly Weather Review, 

 42, 1914, p. 573; 45, 1917. P- 402) has examined the 

 accuracy of various methods of temperature predic- 

 tion. The first, and simplest, is to subtract from 

 the maximum temperature of the day the known 

 average fall in temperature on clear, still afternoons 

 and evenings for the appropriate period of the year. 

 This temperature range varies in different months, 

 but is remarkably uniform under similar topographical 

 conditions and at similar seasons of the year. It has 

 been used by Church (Nevada Station Report, 1915, 

 p. 46). 



The second method is due to Smith, and mvolves 

 two temperature readings daily. Smith discovered that 

 the daily temperature curves showed marked similarity 

 in periods of calm, clear weather when a high-pressure 

 system was centred over the district and conditions 

 were favourable for strong insolation during the day 

 and free radiation at night. For these curves the 

 half-way point in the temperature fall from the 

 maximum of one day to the minimum of the next 

 morning (the "median") occurred at very nearly the 

 same time. Hence a forecast of the probable mini- 

 mum can be made by subtracting from the maximum 

 the temperature shown at the time previously ascer- 

 tained to be that of the median, and then subtracting 

 this difference in turn from the observed median tem- 

 perature. The values thus obtained agreed much 

 more closely with the observed minima than those 

 aiven by the original dew-point method, which is 

 mentioned immediately below. 



The third method, as developed by Smith, is an 

 elaboration of the dew-point determination. This, 

 as us^d bv Hazert {Minn. Expt. Sta. Bull. 12, 1890), 

 and bv O'Gara (U.S. Farmers' Bull. No. -}oi,_i9i5). 

 consists simpiv in determining the dew-point in thf 

 earlv evening (6-10 p.m.), and assumes the dew-poinl 

 temperature will be the probable minimum tempera, 

 ture reached. Smith found that the prediction could 

 be made much more accurately if the relative 

 humidity of the atmosphere was also determined, 

 and he' used the correlation method to show that 

 with high relative humidity the minimum tempera- 

 ture falls below the determined dew-point, while with 

 low relative humidity the reverse is true. \ satis- 

 factory equation expressing this relation was obtained, 



Y=i8-3i4-o-39R, 

 where R = relative humidity in the evening, and Y is 

 departure of minimum temperature of the following 

 morning from evening dew-point. .\ determination 

 of R gives the value of Y, which added to (or sub- 

 tracted from) the dew-point gives the probable n;i_ini- 

 mum temperature to be expected. The numerical 

 terms in the equation differ for different localities. 



Recently T. B. Franklin <Proc. Roy. Soc. Edm., 

 39, 1919,' p. 120) has published some observations 

 on the cooling of the soil at night, with special 

 reference to late spring frasts, and has arrived at a 

 number of important conclusions, which will help 

 considerably in developing methods for forecasting 

 the minimum surface-soil temperature in this country. 

 As a result of observations of temperatures in the air, 

 on the soil, and at a deoth of 4 in., Franklin concludes 

 that a prediction of frost depends on assessing the 

 value of : — (i) -Average relative humidity during the 

 night; (2) the temperature of a given depth (4 in.) at 

 the time of surface minimum temperature; (3) the 

 conductivity of the laver between the assigned depth 



