374 Dr. P. E. Shaw. The Magnetic [May 22, 



a metal would produce magnetic expansion on its own account which 

 would mask any small expansion due to the metal. From the analyses 

 above, we may test this hypothesis. 



Aluminium, silver, bronze, copper and brass all have more iron than 

 1 in 10,000 parts. Now in a field of 1900 c.g.s., the value of 81/1 for 

 iron is about 4 . 10~ 5 ; assuming simple superposition of the expan- 

 sions of the two ingredients, then 1 of iron in 10,000 of metal should 

 produce a value for 81 f I equal to - 4 x 10~ 9 . 



But, as shown above, this movement is readable on the electric 

 micrometer. In aluminium, and even in brass and copper, the 

 impurity should have produced quite a -large movement in my 

 apparatus. 



It would be interesting to examine the relation of the amount of 

 iron in a specimen to the magnetic expansion. My experiments 

 certainly do not confirm the simple superposition theory. 



(b) At one time it was not uncommon for writers on magnetic 

 expansion to deduct a contraction e = B 2 /87rM from the whole expan- 

 sion observed, this being supposed to be always caused by the Maxwell 

 stress in the magnetised substance. 



But this course has been shown by Dr. Chree to be unjustifiable.* 



In these experiments, putting B = H = 1900 c.g.s. and giving 

 M the value 10" c.g.s., we have e = 2 x 10~ 6 cm. for each unit length 

 of specimen, or for the rods used (19 cm. long) e = 4 x 10~ 5 cm., 

 roughly. 



Any such quantity would be easily detected by the apparatus. 

 Hence the conclusion is that the stress does not act wholly, if at all, 

 on the matter. If only one ten-thousandth of the stress acted directly 

 on the -matter the effect would have been perceptible. 



The negative results of these experiments, like those of Sir 0. Lodge,! 

 seem to show that the mechanical connection between matter and ether, 

 if existent, is inappreciable by our present methods. 



I have pleasure in acknowledging my obligation to the Royal Society 

 for the Government grant made to ine in 1901, by which the cost of 

 this research has been defrayed. 



* See letters by Chree and others, * Nature,' Jan., 1896. 

 t * Phil. Trans.,' A, 1897. 



