Manchester Memoirs, Vol. xlviii. ( 1 904), No. 1. 7 



thing like as great as for the first cycle of loading. On 

 the other hand it is not strictly fair to compare a range 

 between o and 8 tons with a range between + 8 and — 8 tons 

 as in the latter case the lag per cycle will probably be 

 something less than twice that in the former. Hence if 

 we doubled the repetitions of the alternating stresses 

 we should obtain a fairer basis of comparison. But when 

 all things have been considered it remains true that there 

 is some special weakness induced by alternating as 

 distinct from unidirectional stresses and hence the 

 probability that the formation of cracks is at least not so 

 serious in the latter case. 



It may be wondered how the process of extension 

 due to a lagging of the extension sketched in Fig. i is 

 applicable to alternating stresses. Figs. 3 and 4 illustrate 

 how it can occur. The change of stress from a positive 



+ 



Q 

 < 

 



-I 





f7 







^ 



/ 



r EXTENSION 



/ 



y / 



+ 



/^ 







I'lg- 3- 



