146 



PROCEEDINGS OF THE AMERICAIT ACADEMY. 



rest, a current sent through the conductor should give a perfectly 

 straight record accurately perpendicular to the base line, and the 

 length of this record should be proportional to the strength of the 

 current. It sometimes happens that an oscillograph which records 

 accurately the march of a moderate current lags in its indications a 

 very little behind the strength of a comparatively feeble current owing 

 to the viscosity of the oil used for damping, which only then becomes 

 troublesome. I have myself had sad experience in drawing from the 

 records of an instrument of this sort, which I thought I had carefully 

 calibrated, elaborate inferences which were contrary to fact. If, however, 



Figure 40. 



Theoretical forms of direct and reverse current curves for a coil of 1394 

 turns belonging to the magnet Q when the resistance of the circuit is 8 ohms 

 and the applied voltage is 10.4. 



one has at hand, first, a well-constructed and mounted ballistic gal- 

 vanometer with a period of from eight to ten minutes, and means of 

 damping the swings of the suspended system (electromagnetically or 

 otherwise) without touching it, and secondly, some kind of chrono- 

 graph designed to close and after a given interval to open again any 

 circuit to which it may be attached, it is easy to test almost any 

 supposed fact about the growth of the flux through the core of an 

 electromagnet. 



The toroids I used had cores made of extremely fine, varnished iron 

 wire, costing about four dollars per kilogram. For some of these I deter- 

 mined by ballistic methods, as carefully as I well could, the hysteresis 

 diagrams for several excitations, and then compared with these other 

 diagrams obtained from the oscillograph records of current curves for 



