Poole — On the Convection of Heat in Vertical Water Columns. 273 



I'ediiced. The length of the column has also a great effect on the convectivity 

 for a given gradient; in fact, the flow of heat up a tube of given diameter, 

 for a given temperature difference between the ends, was generally about 

 the same for short and long tubes, in spite of the fact that with the former 

 the gradient was much greater. 



Since the gradient in the earth's crust is about 3 X lO-* we have to 

 extrapolate the results down to a value for log (? of — 3-52. It is evident 

 tliat no reliance could be placed on the actual figures thus found, but they 

 shoixld furnish some idea of the magnitude of the convection effect in cavities 

 of various sizes in the earth's crust. If we take the curves for the three 

 long tubes, each about 23 cm. in length, which gave much more consistent results 

 than the shorter ones, and assume that their slopes remain unchanged for 

 small gradients, we obtain for the convectivity at earth gradient, 16 X 10-\ 

 1-5 X 10-^ and 4 X 10-l^ for single columns, 28, 20, and 11 cm. in 

 diameter, respectively. For a double column each 28 cm. in diameter the 

 corresponding figure is 2-4, and for a pair 11 cm. in diameter, 14 X 10^^. 

 The shorter columns would, in each case, yield lov/er results. 



The enonnous importance of area of cross-section, especially in the case 

 of single columns, is at once apparent. It would seem that for single columns 

 less than 2 cm. in diameter, and for double ones less than 1 cm., the effect 

 of convection at earth gradient should be negligible. Thus we should expect 

 that, in the upward flow of heat through a water-logged poi-ous rock, convection 

 would play a very small part. The presence of water in the pores would 

 doubtless greatly reduce their thermal resistance and prevent the rock from 

 behaving as a very bad conductor of heat like dry pumice, but there is no reason 

 to suppose that it would raise the effective thermal conductivity even as high 

 as that of an otherwise similar rock devoid of pores. 



On the other hand, the presence of water-filled fissures a couple of centi- 

 metres across, or more, should add very considerablj'- to the upward flow of 

 heat, especially if their shapes were such as to favour continuous circulation. 

 The results obtained with the largest pair of tubes give a value of the 

 convectivity some 400 times as great as the conductivity of common rocks. 

 We might expect that a similar circulation would occur in a single fissure 

 if its extension in both directions in its plane were considerable compared with 

 its width. 



The evidence of these experiments is, then, in favour of the view that 

 convection in stagnant water in the earth's crust is of importance only in 

 regions where fissures of appreciable size exist in the strata. 



In conclusion I wish to express my gratitude to Prof. W. E. Thrift, f.t.c.d., 

 for the laboratory facilities which he so kindly granted to me in Trinity College, 

 Dublin. 



SUMpAKY. 



Experiments are described on the convection of heat in single and also 

 in double vertical water columns. 



The results show that the flow of heat in most cases increases much more 

 rapidly than the temperature gradient. The smaller the column the more 

 rapid is the rise of heat flow with rise of gradient. 



It is concluded that, for the small gradient existing in the earth the effect 

 of convection in water-logged porous rocks would be negligible. Where, however, 

 water-filled fissures occur, we should expect an appreciable increase in the 

 vertical flow of heat. 



