Microscopic Study of Strain in Metals. By F. Rogers. 15' 



elusion must not be accepted without careful examination, how- 

 ever, as there are several opposing circumstances. Thus, when 

 sul^mitted to strains of equal amount, it is probable tliat the stress 

 in ferrite is about 10 p.c. less than that in pearlite. Again, the 

 writer has elsewhere shown that on account of the difference 

 between the coefficients of expansion of ferrite and pearlite, it is 

 probable that there is an initial compression stress in the ferrite, 

 and tension in the pearlite, of steel, which may in some cases 

 amount to as much as 5 tons per square inch. A third considera- 

 tion is, that those portions of incipient cracks which are seen to 

 pass through ferrite probably meet pearlite grains beneath the 

 ferrite, and may even have formed partly on account of the 

 incipient cracking of the pearlite beneath. However, careful and 

 wide study shows that there is a quite decided tendency to selec- 

 tion of a course through ferrite, which is no more exclusive than 

 these reasons would lead one to expect. 



The nature of strain effects in pearlite deserves notice. Vary- 

 ing as it does from slightly transformed microscopically homo- 

 geneous sorbite to definite alternate laminae of ferrite and cementite,. 

 it is found that the incipient cracks in a pearlite of the former 

 variety take indefinite and irregular courses, whilst in the latter 

 the course is often along the plates — apparently through ferrite 

 — occasionally directly across them through a whole pearlite grain, 

 and sometimes step by step, first along ferrite, then across cementite, 

 and so on. 



The percentage compositions of the steels on which the main 

 series of heat treatment and fatigue experiments were carried out,, 

 are shown in the following table : — 



The alternate stresses M^ere applied in machines working upon 

 Wohler's cantilever principle, in which the specimen projects 

 axially from the end of a rotating shaft, and a load is applied by 

 means of a calibrated spring in such a manner as to bend the 

 specimen, whose rotation, therefore, causes the alternation of the 

 plane of bending in the specimen. The design of the machines, 

 and the method of preparation of specimens were such, that any 

 local application of stress due to the method of gripping, any con- 



