CORRESPONDENCE AND NOTES 403

REFERENCES -

Brooks, C. E. P. 1946 Quart. J . R. Met Soc., London, 72, p. 280. 1946 Quart. J. R. Met. Soc., London, 72, p. 279.

Walker, Sir Gilbert 1946 Met. Z., Braunschweig, 58, Heft 4. Stumpff, K. 1941 Quart. J. R. Met. SOC., London, 72, p. 266.

J. \/VADS\Z omi . 120 Cromwell Road, London, S.LV.7.

August 19, 1948.

5 5 1 .5 3 5 -4 Note on the direct measuremint of the thermal conductivity of soil

Hourly readings of soil temperature a t the levels I , 2 , 4 and 8 inches below the surface have been taken during many clear. nights over the past few years a t Cardington in Bedfordshire. In such conditions (clear nights) the temperature gradient is directed upwards a t all these levels, and the temperature change at eight inches is very slow. I t therefore occurred to me that the data could be used for making a direct measure- ment of the thermal conductivity of the soil in the following way : first, vertical temperature profiles were draxn a t twc-hourly intervals through- out each night. The area contained between two consecutive profiles and an arbitrary level will be a measure of the heat passing upwards through that level during- the two-hourly period, provided that the specific heat of the soil and its density do not change, which over such a short period seems likely. The mean temperature gradient through the level is easily measured from the profiles. Suppose now figures for the heat thus transferred and the temperature gradient are computed for several periods during the night, and plotted on a diagram which has heat transfer for ordinate and temperature gradient for abscissa. Then it might be expected that the points would be distributed, doubtless with some scatter, about a straight line through the origin whose slope would be a measure of the thermal conductivity of the soil. In fact, the distribution of points proved invariably to be quite haphazard, so that it was not possible to determine the thermal conductivity in this way.

The explanation for this result is not to be found in any marked variation in the soil properties during the night. Measurement has shown that the extreme range of the heat capacity per unit volume of the soil a t Cardington, from summer dryness to winter wetness, is only from 0.4 to 0.7 calories per cc., and the change during a clear night is probably negligible.

The most likely cause that suggests itself is the transfer of water vapour in the soil, probably in the direction of the temperature gradient, resulting in condensation within the soil of very small amounts of water which would yet liberate sufficient latent heat to mask completely the transfer of sensible heat due to the temperature gradient.

Analyses of mean temperature variations in the soil by Johnson and Davies (Quart. I . R . M e t . Sor., 53, 1927, p. 45) and Wright ( M e w . R. Met. S O L , 4, No. 3 1 ) have provided values of the thermometric conductivity, or diffusivity, of the soil. These estimates, derived from the combined effects of transfer of sensible and latent heat, cannot be used to provide a measure of the transfer of sensible heat alone In such problems as that of the heat balance we are concerned with the total transfer of heat,

410 CORRESPONDENCE AND NOTES

no matter how comprised, and there would be no objection to the use, on individual occasions, of a thermometric conductivity so determined could we 'be sure of its constancy even over short periods. Whatever their explanation, the results outlined here show that this canriot be assumed. Here is yet another example of the difficulty of conducting experiments in what L. F. Richardson has called " outdoor physics."

W. C. SWINBANK. Council for Scientific and Industrial Research, Melbourne.

May 24, 7948.

551.5" Comments on Professor 0. G. Sutton's paper, " Convection in the atmosphere near the ground "

With reference to Professor Sutton's paper on convection, the results of an analysis of some of the Leafield temperature gradient observations may be of interest. Values of the following potential temperature difference ratio (D)

= x).5 - 8 12.4

812.4 - e 1.2 (sufixes are heights in metres) were obtained for occasions when the lapse between 12.4 m. and 1.2 in. was from z to 2.5" F. during the summer months of 1925-26. The means obtained when the 177 values of I ) are grouped for 13 m. wind velocity are as follows :-

Wind velocity, m./sec. . . 1'5-3 3-4 4-6 6-8 8+ MeanD . , . . . 0'155 0'177 0218 0.229 0'255

Time of day and season 10.00 to 15.00 G.M.T. May to August (incl.)

No. of observations . . . 38 31 61 35 I 2 Index s . . . . . 1'53 1.46 1.33 1.30 1'23

The positive correlation between D and wind velocity is statistically significant. The last line of the table gives the values of s in

which accord with the respective I1 values. All the above values of s are appreciably smaller than the value 1.75

found by Professor Sutton from the data for 7 clear June days given in Geophysical Memoir No. 77. H e uses the observations over the whole height range 1 . 2 to 87 m. but the potential temperature differences for the two upper height intervals (30.5 to 57 m. and 57 to 8; m.) were only about 0 . 1 ' F.4ifferences too small to be accurately measured in view of the difficulties experienced in calibrating the installation. The 3 complete calibrations obtained in the 5-year period had to be supplemented by comparison of the records on overcast windy nights with those for the Eiffel Tower on similar occasions. Above 30 m., therefore, the corrections used were considerably dependent on the Eiffel Tower data which can hardly be regarded, in the light of present linowledge, as characteristic of open grassland.

I t is shown in the Memoir that on occasions of strong insolation, errors of the order of o.05' F. can occur in the recorded temperature ctitTerences owing to the various thermometer elements being unequally