DIMENSIONS OP RINGS AND RODS OP IRON AND OTHER METALS. 213 



It may, perhaps, be well to state briefly here, though more will be said on this 

 point in an appendix to the paper, that each ring or rod has its own peculiar curve, 

 which is perfectly definite and is at any time obtainable by the method above 

 described without material variation. 



Being thus satisfied that these curious effects of magnetism are, except as to mere 

 details, independent of the form of the iron, and having regard to the fact that it is 

 much easier to obtain intense fields with straight than with circular coils, I thought 

 it worth while to continue the experiments upon rods only. The metals used, in 

 addition to iron, were cobalt, nickel, manganese-steel, and bismuth. 



The iron rod was that already described as No. 1. 



The cobalt rod was obtained as a rough casting from Messrs. JOHNSON and 

 MATTHEY. It was turned in the lathe to a cylindrical form, and, when finished, its 

 diameter was 7'1 mm., and its length 67 mm. A single scale division, therefore, 

 corresponds to 27 ten-raillionths of its length.* Brass rods were soldered to its ends 

 as usual. 



The nickel used was in the form of a strip 1 00 mm. long, 9 mm. wide, and 

 075 mm. thick. 



The manganese-steel rod was given to me by Dr. FLEMING, who obtained it from 

 Messrs. HADFIELD. It was said to contain 12 per cent, of manganese. Its length 

 was 100 mm., and diameter 4'5 mm. A scale division corresponds to 1'8 ten- 

 millionths of both the nickel and the manganese-steel rods. 



The bismuth was a cast rod procured from Messrs. JOHNSON and MATTUEV. Its 

 length was 132 mm., and diameter 7 mm. One scale division represents 1'4 ten- 

 inillionths of its length. 



The results of a series of experiments with the iron, nickel, and cobalt are given in 

 Table II. and plotted in fig. 4. The mode of operating was exactly the same as that 

 before described, each rod being demagnetised before every observation recorded in 

 the Table.t 



* For each scale division corresponds to O000018 mm., or, since the length of the cobalt is 67 mm., 

 to 0-000018,67 = 27 ten-milliontbs of length. 



t The experiment with nickel was made immediately after that with iron, and the number of battery 

 colls used was, after the first two currents, increased similarly in both cases. It was fonnd that, owing 

 to the constancy of the battery, the ammeter readings in the nickel experiment were, from the third to 

 the ninth cui-rents, practically identical with those obtained in the iron experiment. It was, therefore, 

 assumed that they would be the same with the last three currents, and these were not actually read. 

 Thus the metal was not unnecessarily heated by strong currents, which was a matter of great impor- 

 tance, the magnetic behaviour of nickel being far more sensitive to slight changes of temperature than 

 that of iron or cobalt. 



