1904.] Length in Annealed Rods of Cobalt and Nickel. 



61 



the one published by myself 16 years ago* (curve (a), fig. 1); that 

 for "annealed cobalt" is an almost straight line, lying, however, not 

 above, but below the axis of H ; up to the field of 800 units, at which 

 the experiment was stopped, there is no evidence that the contraction 

 was tending to a limit. The authors make no comment whatever 

 upon this remarkable effect, which may perhaps be well known in 

 Japan, though I have never seen any reference to it in European 

 publications. 



The experiments described in the present paper were made with two 

 different samples of cobalt, the one a cast rod, 9 cm. long and 0'56 cm. 

 in diameter, the other a rolled strip, 10 cm. in length, 0'6 cm. in width 

 and 0'08 cm. in thickness. These were enclosed in porcelain tubes, 

 and placed side by side in the middle of a hot fire, not being removed 

 until the fire had died out some 5 or 6 hours later. 



In fig. 1, curve (b) shows the result of an experiment with the 



FIG. 1. 



IOC 



"Field. ' 



Cast cobalt, (a) unannealed, (b) annealed. 



annealed cast cobalt. It will be noticed that the points of observation 

 appear to indicate a slight sinuosity, but hardly greater than might be 

 accounted for by experimental errors a straight line drawn from the 

 origin to the final point seems to pass evenly enough through the 

 others. The curve agrees well with that given by Honda and Shimizu, 

 and though the field was carried to a much higher value than was 

 reached by them, there is still no sign that the contraction was 

 approaching a limit. Curve (a), fig. 1, shows the changes of length exhi- 

 bited by another piece of the same cast cobalt when in the unannealed 

 state. 



The effect of annealing upon the rolled cobalt was altogether 

 * ' Phil. Trans.,' loo. cit. 



