148 
was the lack of uniform darkening and lighting of the entire field projected 
upon the screen as the analyzer was turned; while in others, especially 
in those that were whole, bright streaks of violet, purple and brown were 
seen, and found to rotate as the analyzer was rotated. In some cases 
these colored streaks were radial, and in still others they formed con- 
centric rings about the knob of the insulator as a center. In order 
to prove more conélusively that these phenomena were caused by internal 
stresses, which were in turn produced by poor annealing, a portion of in- 
sulator No. 1, which showed no initial color rotation, was poorly annealed, 
and when tested again in the polariscope the color rotation effect was 
found to have been introduced. Conversely when a portion of insulator 
No. 2 was properly annealed the color rotation initially present was found 
to have disappeared. 
The insulators which were first tested were those which had actually 
broken while in service upon the line, the parts of which were found on 
the ground near the poles where they were formerly installed. When 
later whole insulators of the same lot were tested it was found that in 
the latter the stresses were much more marked than in the broken parts. 
This fact caused the writer to suspect that some of the internal stresses 
produced by poor annealing were relieved by the breaking of the insulator, 
and to test this belief a whole insulator showing very marked color rota- 
tion was broken and the various parts placed in the polariscope for in- 
spection. It was found that in spite of the fact that the same portion of 
the insulator which showed the most marked stresses was used when 
broken out, practicaliy all the color rotation had been eliminated, although 
the stresses were still present to a less degree in the remainder of ,the 
insulator. In turn each quidrant of the umbrella of the insulator was 
broken out, and in each case the stresses were found to have been either 
reduced to a minimum or entirely eliminated. A further proof of the poor 
annealing was found in the fact that in insulators where the greatest 
stresses were present the umbrella shivered to bits when broken; while 
from insulators showing lesser stresses a whole quadrant could be broken 
out in a single piece. 
Although it is very probable that insulators which are improperly 
annealed fail in service because of sudden temperature changes due to the 
weather and leakage of current over their surface, it seemed advisable to 
show, if possible, what effect, if any, the internal stresses had upon the 
mechanical strength of the insulator in order to determine whether the 
