22 



ART. 8— H. XAGAOKA AXD K. HONDA 



free from errors, wliich, though very small, were sufficient to disguise 

 the minute effect. One serious drawback in these experiments is 

 the use of an unproportionally large mass of iron, Avhich if the ratio 

 of dimensions be sufficiently great, would not have been altogether 

 objectionable, but as it generally happened to be, did not give the 

 expected result. Xon uniformity of the field is another source of 

 error, Avhich unfortunately has been too often neglected ; the result 



obtained in fields, which are not uniform, 

 will indeed be very difficult of interpre- 

 tation. If the material under test be not 

 j)laced axial ly in the direction of magne- 

 tization, and the mechanical force ursfing: 

 the magnetic substance in one or other 

 direction, comes into existence, the 

 change of shape of volumenometer will in 

 some cases be of such an amount, that it 

 not only diminishes the measured change, 

 but screens the desired etfect. These 

 various sources of error may, by proper 

 construction of the volumenometer, be 

 easily eliminated. 



The accompanying cut, taken from a 

 photograph, shows the arrangement of 

 the volumenometer, which we have used 

 in our observations, luio a o-lass tube 



o 

 ()(j, to which is hermetically sealed a 



(•;q»illary tube cc of about 0.4 mm. inter- 



^^^ liai diameter, tit two brass rings rr. 



'J'hese rings are partly open, and by the 



elastic force, fit the tube quite tightly. The extremities of the ovoid 



