Manchester Memoirs, Vol. xhiii. (1904), No. T. 5 



If this were to go on the extension would increase until 

 ultimately the bar would fracture, its shape changing as 

 in a statical test. 



Now slip surfaces produced in a crystal — especially 

 since such surfaces are not quite plane and hence cannot 

 fit or gear perfectly together — must be sources of 

 weakness. If the load is removed there will be almost no 

 reduction of extension it being practically wholly 

 permanent, and hence there will be no — or very little — 

 relative motion of the slip surfaces. If however the load 

 is now reversed, that is applied in the opposite direction, 

 the surfaces will slip over one another, and it appears 

 that ultimately this grinding produces cracks which once 

 started extend rapidly by tearing at their edges, and hence 

 the special weakness of bars subjected to alternating 

 stresses. This is illustrated in Table I. On the other 

 hand the mere removal of the load should give rise to at 

 the most only a little creeping of the surfaces due to the 

 slight elastic extension always present, and hence we 

 should hardly expect the cracks to form and the extension 

 would proceed until finally the bar would break. 



Unfortunately the bulk of experiments have been 

 performed with alternating stresses yet even here con- 

 siderable distortion of the specimen has frequently taker, 

 place. In general however under such conditions the 

 specimen breaks along the cracks with a rough fracture. 



In support of this view that probably with a unidirec- 

 tional stress the cracks would not be formed or if formed 

 would not be of great importance, we have the fact that 

 for a given maximum stress the bar subject to alternations 

 is immensely weaker than the bar subject to a unidirec- 

 tional stress. (Table \.) 



The results are not strictly comparable but the 

 enormous difference in every case which is also borne out 



