130 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Applying to the mean values of E, found above, certain small correc- 

 tions which take account of errors in the graduation of the thermometers, 



we get, — 



T K 



2G°.6 1028 X 10" 8 volt 

 41 ..; 980 " 



54°.5 936 « « 



71°.l 870 « u 



These numbers plotted, with temperatures for abscissas and electro- 

 motive force per degree for ordinate*, indicate a curve which descends with 

 wry gradually increasing slope with rise of temperature. Indeed, this 

 curve is so nearly a straight line that its curvature cannot be satisfactorily 

 shown iu a small figure. It would be almost perfectly straight if the 

 numbers given under E were 1028, 975, 929. and 870. 



The minute examination which I was obliged to irive to the individual 

 small cylinders, during measurements of their electrical resistance, led me 

 to notice defects and possible distortions which might, I feared, have 

 aifected their thermo-electric quality. Accordingly, several months after 

 tests just described were made, I undertook a similar test with ten of tho 

 somewhat larger cylinders already mentioned, which had apparently suf- 

 fered much less in the process of milling. In this later test I found it 

 convenient to enclose the bars in a tube of glass instead of a tube of 

 wood. This test gave 1064 X lO" 8 for E at 15°. 4 C. 



The earlier tests, above described, did not run so low in temperature, 

 but by extrapolation they give, for 15°. 4 C, E = 1065 X 10~ 8 , or some- 

 thing very close to that, a satisfactory agreement. 



Determination of the Difference of Temperature of the In- 



GOIK D Ol rGOING WATEB at THE CHAMBER ABOVE THE 



Disk. 



It has been already stated that a differential platinum thermometer 

 •i ed for this purpose instead of the two copper I rerman-silver thermo- 

 electric junctions which had been employed in the preceding investiga- 

 . This change was the result of a conviction, the fruit of much 

 exp< and vexation, that no suitable permanent protection can be 



found for the thermo-electric junctions against the action of hot water. 

 A thin layer of shellac well dried on appears to be the best coating; but 

 this is liable to give way at a most inconvenient time and bring to caught 



