Sources of error in the experimental objects themselves. 



For the calculation of' the specific magnetisation, we need to know 

 the mass per unit of hnxjtli of" the substance with vvhicii the ex- 

 perimental tube is filled The total mass which the tube contains 

 can be ascertained as accurately as we wish, but the length over 

 which it is spread in the tube, owing to the irregularity of the 

 extremities cannot be determined more accurately than within 

 0.2 to 0.3 m.m. 



Fnither, it is assumed in tlie calculation that it is evenly distri- 

 buted over the wiiole tube which cannot be strictly accurate, owing 

 to sliglit differences in diameter and slight differences in the degree 

 of closeness which is attained in filling to different heights, but this 

 error is certainly small, and only influences the absolute value of 

 the calculated susceptibility ; if the same experimental tube is used 

 in the same position at the various temperatures, this error has no 

 effect upon the relative results, which are the principle object of 

 our research. 



The relative results may however become inaccurate, if between 

 two expei'imenfs at different temperatures, some anisotropic grains, 

 which together form an isotropic mass, each take a different direction, 

 e. g. so that the line of greatest susceptibility in them approaches 

 the direction of the lines of force, since I he mean susceptibility of 

 the group will thereby be changed. We can avoid that this happens 

 unnoticed, by taking the observations at low temperature between 

 two observations at ordinary temperature. To answer the more 

 general question in how far an apparent isotropy can be obtained 

 with substances which are in reality anisotropic, by pressing them 

 in a more or less finely granulated condition into tubes of 5 to 8 m.m. 

 we can, without changing the' conditions in any other respect, repeat 

 the experiments after having brought the experimental tube into 

 another direction relatively to the lines of force, by turning it on 

 its axis. In doubtful cases we applied this method. 



Numerical data: The coils described, can carry currents of 

 6 amp. {if) and 1,5 amp. {im) respectively, for a quarter of an hour ; 

 the force which they then exert upon each other is about 25 grams. 

 Tn ihe given arrangement this may therefore be regarded as the 

 limit of the force which can be measured with the apparatus. Under 

 favourable circumstances (little vibration) a change of 0,001 gram 

 in the force acting upon the carrier can be observed. Generally 

 speaking, the para-magnetic substances examined caused attractions 

 of a few grams, sometimes even more. The accuracy of the results 



