845 



effect of 2N was balanced liv means of K, 13.4 scale <li\isioiis were 

 necessary while Nr. 8 took — J4.2 scale divisions. The resi<1nal effect 

 of the combined action was estimated at —0.5. The eciiiaiion 

 14.2 = 13.4-1-0.5 is satisfied to about 2 7.. 



e. Ejf'f.ct of hnigth of ixinonagtietic sample. To lest this four iron 

 wires of different lengths were cut from one piece. They had lengths : 

 6.73 cm., 7.80 cm., 10.00 cm. Two glass tubes were fastened to 

 the tube of CrCI, in the warm condition. Tests with these were 

 uiade on May 12 and on July 13, 1922. In the first set of tests 

 the paramagnetic action of the iron wires used in one of the tubes 

 at the axis was balanced by the diauuignetic action of suitably 

 chosen combinations of copper wires put into the suspended tube 

 and besides a coppei' wire was put into the othei- tube at the axis 

 so as to take up the bulk of the |)ai'anuignetic action of the iron 

 wire. The measurements were then repealed with the ii'on wires 

 turned upside down. No difference of using one end up railier than 

 the other was found thus showing that the iron wires are homo- 

 geneous. Both sets of measurements agreed in showing that the 

 effects of the wires having lengths 7.80 cm. and 8.80 cm. are nearly 

 equal and slightly greater than the effects of wires having lengllis 

 6.73 cm., 10.00 cm. The effect of the wire having JO cm. length 

 was only slightly smaller than the maximum. The copper and iron 

 wires used at the axis were made to change places and the ob.ser- 

 vations were confirmed the effect of the interchange being very 

 small. On July 13 further confirmation of the observed effect of 

 length was obtained the effect of the iron wires being this lime 

 compensated by changes in K. The number of scale divisions of S 

 required for compensation was tor the same wires in increasing 

 order of length 18.4, 22,6, 22.5, 21.1. 



The reason for the decrease in the paramagnetic effect at 10 cm. 



is not clear. The increase in the region of 7 cm. — 8 cm. must be 



due to the simple increase in the length in a fairly ■ homogeneous 



99 fi 7 80 



field because — — =rl.l7 while —::r^ = 1.16. It may be that the 

 18.4 b./o 



shield causes a peculiar distribution of magnetic field resulting in 

 the upper portion of the 10 cm. wire being in a stronger field in 

 the "up" position than in the "down". 



ƒ. Effect of criterion of comjtensatioii. When the beat frequency of 

 circuits Nr. 1 and Nr. 2 becomes nearly equal to the fiequency of Nr. 3, 

 it becomes at times difficult to distinguish them. Also in theory one 

 cannot deny a possible action of Nr. 3 on the combined system Nr. 1 

 and Nr. 2 and thus an effect on their beat frequency, if these effects 



