﻿Manchester Memoirs, Vol. xlix. (1905), No. 7. 3 



measured by means of the vernier calipers both before 

 and after the application of the load. The ratio of the 

 decrease in length to the original length was taken as the 

 strain. 



In compression experiments this method of deter- 

 mining permanent strains from end measurements alone 

 has the disadvantage of giving rise to errors caused by 

 the departure of the bar from its original cylindrical 

 form under loads beyond the yield point. 



In the case of a tension bar originally parallel, the 

 tendency is for the bar to retain its parallel form when 

 loaded far beyond the yield point, but in short compres- 

 sion specimens of plastic material which are being 

 compressed between dies having plane parallel surfaces, 

 it is found that the effect of the immediate contact 

 between the specimen and the dies is to retard the radial 

 dilatation in the immediate vicinity of the dies. This is 

 probably to some extent due to the great friction set up 

 between the surfaces in contact. It is found that at loads 

 not greatly exceeding the yield point the specimen 

 retains its parallel form in its middle portion for about 

 one half its entire length, and tapers off to a smaller 

 •diameter towards the ends. This is shewn in Fig. 2. At 

 higher loads the parallel portion disappears, and the 

 specimen becomes barrel-shaped. In the present tests 

 care was taken not to carry the loading far enough to 

 destroy the parallel state for about one half the length. 



When a bar begins to taper at the ends the permanent 

 strain, instead of being uniform throughout the length of 

 the bar is found to be greater towards the middle than at 

 its ends. The consequence of this variation is that 

 measurements of strain taken between the extreme ends 

 are likely to be incorrect and misleading. 



In order to overcome the difficulty arising from this 



