48 



PROCEEDINGS OF THE AMERICAN ACADEMY 



lower than that of the steam in the cylinder at this instant, as 

 calculated from the indicator diagram. We cannot suppose this 

 possible, for during the admission of steam this inner surface must 

 have become very nearly as hot as the steam itself, else the interior 

 strata of the iron could not have become heated to the extent ob- 

 served, and the inner surface could not, after being thus heated, 

 cool faster than the steam in front of it and the iron behind it. 

 The same objection cannot be made to curves E and K, yet the 

 broken parts of these curves are open to much doubt, and it is not 

 certain that they should cross each other. A much more accurate 

 determination of temperatures than I have made is needed to draw 



i*o'C. 



FifO 



any one of the three curves with confidence. Moreover, observa- 

 tions at a depth of 3 mm. are desirable, though under the condi- 

 tions of my experiments the fluctuations of temperature at that 

 depth are probably not more than one or two degrees. 



Disregarding inaccuracies of the diagram, I shall now undertake 

 a rough estimate of the amount of heat contained by the cylinder 

 wall at the instant of complete cut-off, as indicated by the curve C, 

 in excess of the amount which it contained just before admission, 

 as indicated by curve K. The planimeter will show that the 

 average temperature along that part of curve C shown in the dia- 

 gram exceeds the average temperature along that part of K there 

 shown by nearly 12° Centigrade. This is equivalent to a uniform 

 elevation of about 2°. 4 extended to a depth of 1 cm. Taking the 



