414 C. Barus — Torsional Magnetostriction in Strong 



phenomenon, otherwise quite consistant throughout, is spurious. 

 As an example I may give the series of results, figure 4 (1, 2, 

 3,) and figure 5 (1, 2, 3), for annealed iron wires -044 cm and 

 •024 cm in diameter, respectively, and 41 cm long each. The 

 twist, 0, imparted to either wire is given in degrees by the 

 abscissas, positive or negative as specified. The ordinates show 

 the deflections 80, in radians, obtained when the field is made, 

 so that *05T 80/0 = 8n/n is the increment of rigidity. 



Fi%. 5, (3)\tfoad,500g.. I 



1500* 



The consecutive curves given correspond to successively 

 increasing loads or vertical pulls on the system, as indicated. 

 Observations were made in the cyclical way shown by the 

 arrows. The same zero was retained throughout. In general 

 for positive twists the return curve is positive relatively to the 

 outgoing curve ; for negative twists the return curve is nega- 

 tive relatively to the outgoing curve. Definite consideration 

 will be given to this interesting result in my next paper so that 

 the present reference will suffice for this and succeeding cases. 

 The effect of loading (among many results of a similar kind) 

 is specially shown in figure 4 at A, B, giving evidence even of 

 sign reversal. The deflections are large for the thin wires as 

 compared with the thicker wires (different scales are needed in 

 the two cases). Special experiments showed that if a series is 

 repeated the curves obtained, however irregular, are identical. 



Notwithstanding the trustworthy character of the results 

 obtained, I became convinced that the excessive effects pro- 

 duced by loading, and hence the whole phenomena, were some- 

 how in error. 



5. In the experiments of §4, the magnetized wire was 

 enclosed in the fine glass tube mentioned above. In the pres- 

 ent experiments it is free (the tube being withdrawn) and visi- 

 bly suspended in the air gap. The latter was adjusted after 



