844 



Ijecaiise balance in tlie end |)08ition.s was found not to mean alwaj'S 

 a balance in tlie inteiinediate ones. 



^ 3. Preliminary Tests. 



a. Necessity of shield /or coil. Tiie fact that tlie coil had to be 

 shielded on the inside was ascertained by first trying the anangement 

 without the shield. A glass bottle lowered inside the coil by a 

 thread produced an appreciable effect on the beats. This effect was 

 absent when the inside of the coil was covered with thin tinfoil 

 strips. 



I). Bi'uiliiig of box lid. Many blank tests have been made to see 

 whether the strains in the lid due to the up and down motion of 

 the rod system Q affected the beats. No such effect was observed. 



c. fnteraction between circuits has been observed to be generally 

 very small. Thus by adjusting the beat frequency between Nr. 1 and 

 Nr. 2 to a low value like 100 no tendency of pulling ,into one has 

 been noted. Putting a paramagnetic sample in and then out again 

 under these circumstances did not alter the beat frequency noticeably. 

 The beat frequency employed in the measurements was of the order 

 of 5UÜ or 1000 per second and the interaction must have been slill 

 smaller. A further test of this was made by first compensating a 

 sample by changing K, say with set Nr. 1 going at a higher frequency 

 than Nr. 2 ; then Nr. 2 was set to the higher frequency and the 

 measurement repeated. The following table gives the results of the 

 changes of the scale .S'. 



The iron wire Nr. 8 was here put in a tube fastened to the para- 

 magnetic sample in the warm condition with thread and sealing 

 wax. This tube was appro-vimately at the axis of the coil while Nr. 7 

 was in the suspended tube almost at the shield. There appears to 

 be no true systematic difference between measurements made with 

 the two frequencies reversed. 



d. Direct test of balance. It so happened that a copper wire (2 N) 

 when used in the suspended tube balanced by its diaraagnetic eddy 

 current effect the paramagnetic effect of iroji wire (Nr. 8). When the 



