272 



Scientifie Proceedings, Royal Dublin Society. 



(311 the thermal conductivity of tlie earth's crust, it appeared to be most 

 convenient to state the results in terms of a quantity analogous to the conduc- 

 tivity of a solid. This quantity, which may be called the " convectivity, " is 

 denoted by the letter G in the iigures, and is defined as the heat flow per sq. cm. 

 of total horizontal area per see. divided by the gradient Q. As, however, 

 the extreme range of G in the plotted results is from 5 X 10'- to over 10, 

 and that of G from 5 X 10"^ to over 1, all the results could not conveniently 

 be plotted directly on one scale, so their logarithms are plotted instead. 



Fig. 3. 



In the ease of the results shown in fig. 2 no correction has been made for the 

 mean temperature of the water column, as the effect of variations of a few 

 degrees at about air temperature does not seem to be verj' imijortant for single 

 columns. The general average of the mean temperatures for these results may 

 be taken as about 15° C, the temperature generally increasing with increase of 

 gradient. 



The results for the double columns shown in fig. 3 have all been reduced 

 to a mean temperature of 17° C. by assuming that the convectivity is proportional 

 to the excess of the mean temperature above 4° C. As unavoidable small 

 differences occurred in the dimensions of the glass tubes, the figures given 

 are, in each case, the means for the pair. 



It will be observed that the convectivity generally rises rapidly with rise 

 ill gradient, the effect becoming more and more marked as the diameter is 



