the Heat-conductivity of Stone. 



255 





1st. 



2nd. 3rd. 



4th. 



K 



0-00583 

 000236 



000624 1 000609 

 000237 0-00226 



00590 

 000252 



E 





Now, as the present investigation was designed to show the 

 feasibility of the experiments, and the possibility of deducing 

 from them the numerical values of K and E, rather than to 

 obtain with the greatest possible accuracy the values of K and 

 E for one particular Japanese stone, we have not thought it 

 necessary before the publication of this paper to repeat the 

 experiments, using greater precautions to ensure perfect uni- 

 formity of the outside temperature by making the water-bath 

 larger and the stream of cold water more rapid. And although, 

 had we done so, any one of the above four methods would 

 have given results differing extremely little from the truth, still 

 it would have been none the less interesting to have examined, 

 as we have done, what method of reduction is calculated to 

 give the most accurate results when there is a slight variation 

 in the external temperature during the course of an experiment. 



Where considerable accuracy is required we should advise 

 tho employment of the fourth method ; if still greater accuracy 

 is desired, then, several sets of observations having been made, 

 the values of K and E should first be calculated by the fourth 

 method; then, the mean of all the values of K having been 

 taken as well as the mean of all the values of E, these results 

 should be employed in drawing a number of exisothermal 

 curves, when, lastly, the way in which the curves obtained 

 from experiment lie among these exisothermals will enable us 

 to calculate, again by the fourth method, the final values of K 

 and E with any desired amount of accuracy. To illustrate 

 this ; we have drawn a number of exisothermals on the suppo- 

 sition that 



K = 0-00590, 



E = 0-00252; 



but the size of fig. 2, as engraved, only enables us to show 

 two, B B B, C C C, the exisothermals for 16°'4 C. and 19°-3 C. 



From an examination of the way in which our curve AAA, 

 obtained from the experiments, lies among these exisothermals, 

 it will be seen that our first approximation is quite accurate 

 enough for practical purposes, so that further approximations 

 are unnecessary in this case unless very great accuracy be 

 desired. 



VII. We made another series of experiments at the end 



