680 PROCEEDINGS OF THE AMERICAN ACADEMY. 



27430 

 Means of Readings on the Thermo-Electromotiye Force. 



Date. No. of Junction. 2D. Diff. of Temp. Temp. E. M. F. per degree. 



Feb. 7 1 and 4 outside 41.51 4°.318 44° .4 20.56 M. V. 



" 7 land 3 outside 43.14 4°.366 44°.4 20.57 " 



" 20 All 29.85 2°.650 50°.0 20.58 " 



" 22 All 35.09 3°.066 76°.5 21.43 " 



" 22 land 3 outside 32.50 3°.061 76°.5 21.39 " 



" 23 All 55.43 4°.814 87.°6 21.81 " 



These results are expressed graphically in the following curve (Fig- 

 ure 5). 



Sources of Ekror in Calibrating Thermo-Electromotive 



Force. 



1. Difference of temperature. — The difference of temperature was 

 determined by a Beckman thermometer standardized between 0° and 

 100° by the Reichstanstalt, to an accuracy of xoVo° ^ or a difference of 

 temperature not exceeding 5°. All corrections were applied to the 

 standard. 



2. True temperature. — An error in this reading of 0°.5 would 

 cause an error of less than ~^ of 1 per cent in the value of the electro- 

 motive force per degree. The error in the thermometer used for this 

 purpose was much less than this. 



3. Resistance of nickel lead wires. — This error was large, due to the 

 fact that the galvanometer was connected in series with this resistance 

 when finding the deflection, but when the galvanometer was calibrated 

 the deflection was produced by the difference in potential at the termi- 



