36 N. N. ZIRBEL 
It has been found that the duration of time during which the wire is 
baked at a certain temperature is of little importance so far as the effect on 
the temperature coefficient between 20° and 50°C is concerned. In the results 
shown, this time varied from 3 minutes to 7 hours. It has also been found 
that after a wire has once been heated to a certain temperature, subsequent 
bakings at lower temperatures have no effect. That is, the curves shown can 
be traversed in one direction only, and after a wire has once been carried 
above the critical temperature no later heat treatment can diminish its tem- 
perature coefficient. Finally, measurements made on the two halves of the 
same wire show that the temperature coefficient is uniformly distributed 
along the wire and is not due to any local irregularity. 
We have no explanation for the peculiar behavior shown in the last three 
figures. However, incomplete results on a fourth sample from an entirely 
different source show the same sort of behavior, and we believe that the ef- 
fect, whatever its cause, is genuine. Suitable precautions have been taken to 
avoid the effect of convection currents and other disturbing factors. The fact 
that in all cases so far observed, involving wires from three sources, the tem- 
perature coefficient is initially in the same direction is also difficult to explain 
and may be merely a coincidence. 
In conclusion, the writer wishes to express his thanks to Professor H. A. 
Wilson, under whose direction this work was done. 
180 
