684 PROCEEDINGS OF THE AMERICAN ACADEMY. 



variometer reading, subtracting each from the first reading, and plotting 

 this difference as ordinates and the time as abscissa. The two curves 

 thus obtained should be approximately the same if the difference in 

 temperature and galvanometer deflection vary at the same rate. 



12. Variation in difference of temperature during a reading. — That 

 this variation was less than .1 of 1 per cent is shown by the results. 



13. Change of zero in the galvanometer. — Double deflections were 

 always used, consequently the zero was not needed. 



14. Different junctions give the same electromotive force. — See Sum- 

 mary of Results. 



15. A variation in the difference of temperature gives the same result 

 for the electromotive force per degree. — See Summary of Results. 



16. The resultant electromotive force per degree is the same no matter 

 which tank is at the higher temperature. — See Summary of Results. 



17. The temperature of the oil not the same as that of the spheres. — 

 The thermometers were placed in the same relative position in both 

 tanks, and since the tanks were constructed the same, and the oil rotated 

 in the same direction, and the difference in temperature being small, 

 the thermometers would both be at the same relative amount above or 

 below the temperature of the spheres. Placing the thermometers in 

 different parts of the tank gave the same difference. 



18. Variation of rate of agitation of oil. — The rate and direction of 

 rotation of the oil was changed repeatedly, but no effect could be dis- 

 covered in the results. The direction of rotation was the same in 

 both tanks, but the rate was sometimes the same and sometimes quite 

 different. 



For errors in calibrating galvanometer see errors under conductivity. 



19. Error in Beckman thermometer No. 2. — When comparing the 

 two thermometers, if the tank has changed in temperature, then it 

 becomes necessary to use the difference of two readings of Beckman 

 No. 2, but as this difference never exceeded a few hundredths of a 

 degree it was assumed that there would be no error. 



20. Error in connecting the two hemispheres. — It was found in the 

 research on iron by this method that a materially different electromotive 

 force was obtained when the position of the iron connecting the hemis- 

 pheres was changed. A number of experiments were made to clear this 

 point. It was found that the electromotive force was independent of the 

 method of attaching and position of the connectiug wire on the hemis- 

 pheres as long as the electrical contact was good. This was a simple 

 matter with lead. 



