OF ARTS AND SCIENCES. 331 



TABLE FOR QUARTER-INCH IRON ROD. 

 F=24 27. 



(1 Cell.) (4 Cells.) 



c = ph = 13.0 mm. ph = IG.O mm. 



6 13.5 



10 14.0 



15 18.? 



20 17.5? 



80 



17.7 

 18.8 

 18.5 

 16.1 

 16.? 



The zero points were 10.5 mm., 29.5 mm., and 38.5 mm., nearly, as 

 before. 



No experiments have as yet been made upon nickel or cobalt, the 

 stress of the whole investigation having been thrown upon soft iron. 

 The results of a long series of experiments similar to those quoted 

 above, with slight modifications introduced by the method described 

 below, and corrected as far as possible by the ordinary processes of 

 differencing, are embodied in the large Table for Phase Retardation 

 at the end of this paper. A set of curves of constant retardation are 

 also given. 



Although these experiments were conducted with the greatest 

 exactness possible with the rough apparatus employed, they were open 

 to criticism in that they were separated by long intervals of time, and 

 possibly subject to wide variations of condition. To confirm their 

 general results, and in a manner to leave no doubt that the retarda- 

 tion of phase really depends upon the speed of the commutator (the 

 other conditions being constant), so as to gradually disappear and re- 

 appear with the diminution and increase, respectively, of its velocity, 

 an electric motor was employed, whose revolving wheel was heavy 

 enough to keep itself and the commutator in motion for a long time 

 after the electricity had been cut off. In coming thus gradually to 

 rest, all velocities between maximum and zero must have been passed 

 through, thus affording the best possible opportunity for a close com- 

 parison. 



Meantime, each swing of the galvanometer was noted (its period 

 being made sufliciently great), as well as the exact time of stopping 

 of the machine, obtained by a sharp signal from an assistant. The 

 mean between two successive points of turning gave the average de- 

 flection of the galvanometer for the interval of time between them ; 

 and knowing the length of time of each swing (a constant), the rate 

 of stopping of the machine (nearly constant), and the exact time of 



