130 SECTIOKAL ADDRESSES. 



the specification is that the curve shall fall above Q. But the test does 

 not require the curve to be determined. 



If we admit that the fatigue limit is the proper basis for engine- 

 strength calculations, there are a number of interesting modifications 

 required in the common theory of the strength of materials. It will 

 no longer be possible to neglect, as has been so general in the past, the 

 uneven distribution of stress in irregularly shaped parts of machines. 

 It has been generally recognised that sharp corners should be avoided 

 when possible, but no theory is available to enable the stresses at corners 

 to be calculated or to enable their effect on the strength of the member 

 to be estimated. If fatigue is the critical factor in failure under fluc- 

 tuating stresses such theory is most necessary. Even the roughest 

 guide would be of great value. The nature and magnitude of the con- 

 centrations of stress which occur in practice have been investigated 

 experimentally by Pi'ofessor Coker by his elegant optical method which 

 has given most valuable results, some of which are already being used 

 in designing offices. If the mathematical theory is too difficult, it may 

 be possible to lay down practical rules deduced from such experimental 

 results — but the method still has many limitations, perhaps the most 

 serious being that it can only be used on flat models. I believe Professor 

 Ooker expects to be able to extend the method to round models. 



As a simple example to show the importance of the subject let us 

 consider the effect of a groove round a straight round bar subject to 

 alternating tension and compression — such a groove as a screw thread. 

 There will be a concentration of stress at the bottom of the groove. 

 The ratio of the stress at the bottom of a groove to the mean stress in 

 the bar has been worked out mathematically by Mr. A. A. Griffith, and 

 his calculations have been confirmed experimentally by his elegant soap- 

 bubble method. The ratio depends on the relation between the depth 

 of the groove, the radius at the bottom, and slightly on the 

 angle between the sides. For a Whitworth form of thread the ratio 

 will be about 3. If the Fatigue Limit is exceeded at the bottom of the 

 groove the metal will fail and a minute crack will form there ; this crack 

 will soon spread right across the bar and total failure will result. Thus 

 we see that the safe mean stress in the bar will be reduced to one-third 

 what a plain bar will bear. The truth of this theory regarding the 

 importance of concentrations of stress has still to be proved experi- 

 mentally ; if true, it is of far-reaching importance, since it applies to all 

 concentrations of stress in machine parts subject to fluctuating loads. 



The theory does not apply to steadily loaded members ; in these the 

 local excess of stress is relieved by the stretching of the minute portion 

 which is overloaded, and no further consequences follow. 



The theory appears to apply to grooves however small, and has an 

 important bearing on the smoothness of the finish of machine parts. 

 The surface of any engine part finished by filing is certainly entirely 

 covered with scratches. Emery likewise leaves the surface scratched — 

 though the scratches are smaller. If, however, polishing be carried 

 further the surface may ultimately be freed from scratches and left in a 

 burnished condition. In this condition amorphous metal has been 

 smeared over the surface — the smooth appearance is not simply due 



