224 SCIENCE PROGRESS. 



there is a mass of rapidly accumulating data, it at present 

 seems largely to be only capable of being- used in a more 

 or less empirical fashion and still to be in need of some 

 rational explanation. The researches of the Committee of 

 the Institution of Mechanical Engineers on "Alloys" has no 

 doubt done something to shift the question into a clearer 

 light, but the field open for research is still very wide. 



Repetition of Stress. The questions of the value of the 

 natural limit and of safe working stresses are intimately 

 bound up with that of the effect of repeated loads on the 

 endurance of materials. Many engineers, taught by experi- 

 ence, often costly, had realised the importance of this 

 question before Wohler's classical experiments drew the 

 attention of every one to this vital point. It is, therefore, to 

 further research in this field, combined with the experimental 

 work that I have just alluded to, that we must look for a truer 

 insight into many obscure questions, and for the explanation 

 of many apparently inexplicable failures and break-downs. 

 It is, however, a most costly line of work, and one which 

 from its character can only be adequately carried out, when 

 funds are forthcoming either from the State or from the 

 Technical Institutions interested in the question. The 

 machinery needed is purely special, and must be continually 

 at work, and it involves the use of a not inconsiderable 

 amount of power constantly running, in addition to the 

 time and attention of observers who must be largely free 

 from other duties. 



There is one difficulty in connection with this subject 

 which is perhaps not sufficiently attended to in experiments 

 on the effect of repeated loads, namely, in the determination 

 of the actual range of stress in any given case. This can 

 in most cases only be determined by certain calculations, 

 which generally ignore, or consider as non-existent, effects 

 due to vibrations set up by the instantaneous application 

 and removal of the loads, or to inertia of moving parts of 

 the machine. These may no doubt with carefully de- 

 signed appliances be practically eliminated, but the first of 

 them at any rate may possibly quite unexpectedly creep in, 

 and in an experiment of this nature a very small error may 



