1032 THE BELL SYSTEM TECHNICAL JOURNAL, SEPTEMBER 1954 



which minimized the magnetic energy of the void. As a result of this 

 state of affairs, the wall was not in equihbrium in the middle of the 

 sample, but always shrank around the void so as to magnetize the sample 

 in one direction or the other. It was impossible therefore to demagnetize 

 the sample so that the wall was at the center of the legs and then have 

 the wall stay there to be observed. The wall could be brought to the 

 center, however, and held there using the technique mentioned by 

 Williams and Shockley^ (see Fig. 8 in their paper). The legs of the sample 

 were so small that it was difficult to do this, but the wall was found on 

 three legs of the sample at different times in this way. The wall curved 

 a good deal in traveling along the legs. This sample was etched repeatedly 

 so that data could be obtained as a fuction of sample dimensions. The 

 variations observed led to a viscous domain wall damping independent 

 of dimensions if it was assumed that the domain wall was perpendicular 

 to the major (110) face. Therefore it was to this (112) plane that the 

 wall was brought for observation. 



Sample 2. Stationary walls at the corners w^ere rather patchy but 

 visible. The movable wall was traced along the outside faces of two legs, 

 which indicates that it lay in the (110) plane. The wall curved a good 

 deal. There were also other walls which enclosed patches of surface. It 

 is suspected that these patches were the bases of spike domains extending 

 into the sample from strain patterns on the surface. The sample was 

 therefore etched again. Unfortunately, the bath apparently became 

 locally overheated, and this etch took off rather more material than 

 expected. It also left a matte surface on which domain walls could not 

 be observed. The data taken on this sample, however, check those on 

 other samples if we assume a pattern in which the movable wall is in 

 the (110) plane, as our observations lead us to suspect. 



Sample S. The stationary walls at the corners were seen, but only 

 with difficulty. They were patchy. There was a good deal of structure 

 all along the legs on the major (110) face of this sample, but no wall 

 which ran around the sample could be seen on this face. 



On the outside of two of the legs, which are (112) faces, pitting and 

 extraneous walls were so bad that the main wall could not be discerned. 

 On a. third, the wall could be traced most of the way. On the fourth, 

 however, there were two walls, one of which could be traced along the 

 whole leg, the other of which went only three-fourths of the way along 

 the leg. Both walls on this leg showed a good deal of curvature. It there- 

 fore appears that the movable wall lies in the (110) plane, but that 

 there is another wall big enough so that it may move and affect our data. 

 This picture of a domain pattern with two movable walls was confirmed 

 by checking the data obtained on this sample with those from others. 



