1903.] Expansion of some of the less Magnetic Metals. 373 



Various strains, other than those for which we are looking, will be 

 liable to occur, and to show themselves in the measuring apparatus, 

 e.g., due to (1) solenoidal suction of the coil on the end ; (2) directive 

 action of the earth's field on the magnetised coil ; (3) attractive action 

 on the coil of any iron in its neighbourhood. 



It is very difficult to eliminate all such false strains. At first I 

 worked with the coil attached to the lower end of the rod, counterpoising 

 the coil to remove all strain from the rod. This method was employed 

 by S. Bidwell,* it obviates solenoidal suction and may on that account 

 be useful with a ferro-magnetic core, but in my experience it was 

 troublesome. I then placed the coil on a stand separate from the core 

 and its supports, taking care to centre the rod in the coil. The coil 

 and core must not come into contact or the small movements of the- 

 former will displace the latter. 



A great deal of time was spent in finding out the best conditions of 

 Avorking ; thousands of readings often apparently quite consistent for 

 a long time were taken down. These readings varied from 1 x 10~ cm. 

 to 1 x 10~ 8 cm. But gradually, as the elimination of disturbing- 

 causes proceeded, the result in the case of every metal tried was found 

 to be nil. 



Bismuth was tested with special interest on account of the work 

 already done on it. 



S. Bidwellf recorded a small expansion, but later J he reported that 

 although the field used was 1500 c.g.s., and the unit of measurement 

 1/70,000,000 of the length of the rod, yet no movement could be 

 found. 



No other attempt seems to have been nearly as searching as this 

 last, although Knott, Van Aubel,|| and lately A. P. WillslI have also 

 taken up the matter and record the same negative result. 



After working with one specimen of bismuth, I had the material 

 recast of somewhat less diameter, taking care to remove the surface, 

 which might obtain impurities from the loam used in the casting. In 

 both field and length measurement, the test which I have applied to 

 bismuth appears to go further than any previous one, for, as shown 

 above, the maximum field is 1900 c.g.s., and the smallest movement 

 per unit length readable is 6/// = 4 x 10~ 9 cm. 



The metals used were obtained specially pure from Messrs. Johnson 

 and Matthey ; they were analysed for traces of iron by the sulpho- 

 cyanide colour test. The following list shows the amount of iron in 

 10,000 parts of metal : bismuth, 0'88 ; silver, 1-0; tin, O'l ; aluminium, 

 72; zinc, 0'68 ; lead, 0'26 ; copper, 4*8; bronze, 7'0; brass, 4'0. 



4. ( a ) It has generally been supposed that a small trace of iron in 



* Loc. cit. ' Nature,' June, 1899. 



t ' Phil. Trans.,' A, 1888. || ' Nature,' Aug., 1899. 



' Nature,' July, 1899. ^ * Phys- Rev ->' Jul 7> 1902 - 



