48 PROCEEDINGS OF THE AMERICAN ACADEMY. 



While it takes a long day to make an accurate determiuation with 

 our large apparatus of the absolute couductivity of a slab, two determi- 

 nations may easily be made in the same time of the relative con- 

 ductivities of the slabs which go to form a prism, since the gradient 

 on the axis of the slab does not sensibly change after four hours 

 of heating, and it is then only necessary to note the readings of the 

 thermopiles. With our smallest apparatus and thin slabs two hours are 

 often sufficient for a measurement. Our experience seems to show that 

 this method of comparison is susceptible of great accuracy. We have 

 made a very large number of direct determinations of the conductivities 

 of different slabs of stone, but, in view of the fact mentioned above that 

 the amount of moisture in the stone affects the conductivity very appre- 

 ciably, even if the les.s tedious method of comparison were not ecpially 

 accurate, we should think it wise in future to determine with great care 

 the absolute conductivity of a standard substance unaffected by moisture, 

 and then compare with it the conductivity of the stone slabs. The ac- 

 curacy with wliich the comparison can be made is greater of course than 

 that of a single absolute determination. 



The particular kind of glass which we have found useful as a compari- 

 son substance was selected some years ago from the stock of the Boston 

 Plate Glass Company. The faces of each plate are very nearly plane, 

 but the planes are not in every specimen quite parallel. The conduc- 

 tivities of different plates are somewhat different, but the conductivity of 

 each plate remains sensibly constant within large ranges of temperature. 

 Cut from this glass we have a number of slabs 60 centimeters square, a 

 number of slabs 30 centimeters square and some disks about 20 centi- 

 meters in diameter. 



We shall wish to discuss the properties of this kind of glass at higher 

 temperatures more particularly on another occasion. For our present 

 purposes, it is worth while to measure the temperatures to tenths of 

 degrees only and the thickness of a slab to the nearest twentieth of a 

 millimeter, and an account of a few experiments to this degree of ac- 

 curacy, chosen almost at random from the large number of which we have 

 records, will suffice. 



Slabs A, B, C, and D are cut from one particular large homogeneous 

 piece of this glass, the conductivity of which, according to our determi- 

 nations, is to that of Plate III. mentioned below as 187 to 175. We 

 shall assume the conductivities of these slabs to be 0.00277 at all ordinary 

 temperatures. We have not been able to detect any differences in their 

 conductivities. 



