Yiscosity and Rigidity of Iron and of Steel. 177 



i A 



(g2 /c 2 5) per Ampere. Current was obtained from five flat Grove 

 cells, in mean intensity of about one Ampere. Hence the iron 

 wires were probably not far from saturation, supposing that a 

 field of 140 c. g. s. units of intensity is sufficient to magnetically 

 saturate soft iron. 



Without elaborate precautions it is impossible to produce a 

 strong field for great lengths of time in this way, without per- 

 ceptibly heating the helix. This introduces a serious error, 

 for the viscous cletorsion actually observed is due to a heat 

 effect superimposed on a magnetic effect, both of which are of 

 the same order of minuteness. By using a helix like the one 

 above, in which current may be passed through the two partial 

 coils in the same or in opposite directions, the full heat effect 

 may be observed either with or without the magnetic effect. 

 For when the coils are joined differentially the field produced 

 is zero, whereas the heat generated in the helix is (cast, par.) 

 not changed in amount. This is the way in which I endeavored 

 to eliminate the temperature discrepancy. 



It is difficult to find two wires which are absolutely identi- 

 cal ; after softening some parts of a wire yield more easily to 

 stress than other parts. It is not until the less rigid parts have 

 been stiffened by receiving permanent set that the rate of twist 

 temporarily stored is the same throughout the length of the 

 wire. All viscous motion which is due to differences in the 

 mechanical or chemical properties of the metals may be detected 

 by allowing series of observations for open circuit to alternate 

 with similar series for closed circuit. 



Results. — Table 1 contains results as obtained with steel 

 annealed about midway between soft and hard. Here t is the 

 temperature of the helix and of the upper wire, t' the tempera- 

 ture of the lower wire, I and p the length and radius of each. 

 <p denotes the amount of viscous detorsion per centimeter per 

 wire (i. e., per two centimeters of the system), in radians at the 

 time A in hours; t denotes the rate of twist in degrees im- 

 parted to the system at the unmagnetic end. If <p increases or 

 decreases according as the sign of the twist is positive or 

 negative, then the unmagnetic wire is of greater viscosity than 

 the magnetic wire ; and vice versa. Intervals of observation 

 corresponding to open and closed circuits are appropriately in- 

 dicated. "Whenever the circuit is closed differentially in the 

 helix so as to produce a zero magnetic field, this is also stated. 

 ip is correct to 3 or 4 units of the last place. 



The three parts of this table show that the viscous differences 

 in question are invariably minute. In part first the originally 

 less viscous lower wire becomes perceptibly more viscous when 

 the circuit is closed around the upper wire. The apparent 



Am. Jour. Scl— Third Series, Vol. XXXIV, No. 201.— Sept., 1887. 

 12 



