G20 I'HESIDEXT's ADDHESS SECTIOX H. 



■w-ithout any Haw to serve as a nucleus, the first breach of continuity 

 being set up through repeated grinding on a plane of slip in perfectly 

 sound metal." It is obvious that in a bar l)uilt up of grains in the 

 way we have described some of the grains will have the direction 

 of tlieir natural planes of slip more favourably situated for movement 

 taking place than others; this is probably why the slip bands show 

 themselves in only a few grains, most likely in those in which the 

 direction of the shearing stress is most nearly parallel to the gliding 

 planes. Besides this, it is probable in such a bar, built up of ciystals 

 iriegular both in fomi and size, that there will be some variation 

 in the way in which tlie stresses are distributed among the crystals, 

 so that some Avill Ijegin to show signs of giving befoi'e others. The 

 microscopic examination of specimens broken under fatigue experi- 

 iiients at the National Physical Laboratory by Messrs. Stanton and 

 Bairstow. as well as the work of other experimenters, have borne 

 out these conclusions. In the case of moderately high carbon steels, 

 ir has been shown that there is a strong tendency for these cracks 

 to develop in the grains of ferrite, as thougli this were the weak 

 constituent of the steels and the cause of the ultimate failure. While 

 luuch still i-emains to be explained in connection with fatigue, these 

 investigations imdoubtedly nuirk a big step forward, inasn^uch as 

 they have shown us something of the mechanism by which these 

 remarkable effects are produced, and some of the features of fatigue 

 that we have discussed become readily intelligible. 



Testing a bar of mild steel in the ordinary way under tensioi. 

 we obtain the well-known tj^je of stress-strain diagram in Fig. 'j.. 

 From A to B, the proportional limit, the diagram is a straight line, 

 showing very little extension, but its amount is always proportional 

 to the stress. From B to C there is a slight bending of the line, 

 showing that Hooke's law no longer holds perfectly true, and at C, 

 the yield point, the steel l)egins to stretch rapidly ; the nuiterial 

 seems to become in a jJastic condition and the bar stretches quite a 

 considerable amount withdut any increase in the stretching force. At 

 this stage the behavioui- of the material seems to change from that 

 of an elastic solid to tliat of a viscous Huid. Now, from A to 1? 

 there is no time effect, that is to say the amoimt of stretching does 

 not depend on the length of time the load is left on the bar. But 

 somewKere near B a. time eft'ect begins to show itself, and at or 

 beyond C the time effect becomes considerable, and the nature of 

 the curve depends on the rate at wliicli tlie bar is stretched. Up to 

 the point B, or thereabouts, tlie elongation is practically a function 

 of the stretching force only, but ])eyond B it is a function 

 of both force and time. The fact that the time eft'ect begins to show 

 itself distinctly-jat B. and is still more nuirkecl afterwards, suggests 

 that at this point, since it is not to be expected in a gi'anular bar 

 that the stress is distributed with perfect uniformity, some of the 

 crystals haye already reached their yield point, which is only reached 

 for the bar .as a whole at C, and that the deviation of BC from the 

 straight line-is due to the bar becoming plastic in spots. That even 

 under the most careful conditions of experiment a mild steel bar 

 does begin to yield locally long before the plastic stage is reached 

 for the bar as a whole has been conclusively proved by Fremont* 



* See "Nature," Jan. 21, 1904. 



