ON STRESS BISTRIBUTIONS IN ENGINEERING MATERIALS. 169 



discrepancy is greatest with moderate loads and disappears when the 

 magnitude of the steady stress is increased. At this stage, however, 

 the metal is liable to yield in a ductile manner owing to the intensity of 

 the steady stress, so that it is hardly practicable (or profitable) to 

 continue the investigation with higher stresses. 



Fig. 4 shows also, by the curve D, a limiting range of stress beyond 

 which ductile failure occurs instead of the characteristic fatigue failure. 

 The exact position of the curve is difficult of determination, as the range 

 of stress necessary to produce ductile extension depends on the rate of 

 increase of stress in somewhat the same manner as the form of the 

 stress-strain diagram depends on the rate of application of load. Thus 

 fatigue appears to play a part in ductile extension as well as in brittle 

 fracture, accelerating the development of strain. 



The two specimens shown in Fig. 5 clearly illustrate the two 

 manners of failure. Thus specimen A fractured after 672,000 repeti- 

 tions of a cycle of stress in which an alternating stress having a range 

 of 23 '7 tons per square inch was combined with a steady stress of 

 9'87 tons per square inch. Specimen B, loaded with the same component 

 of steady stress, extended immediately with reduction of cross sectional 

 area, when the range of alternating stress reached the value of 29 tons 

 per square inch. Ductile failures occurred at lower ranges than this when 

 time was given for the phenomenon to develop, thus the curve D is 

 drawn through the ordinate at 252 tons per square inch range, no cases 

 of ductile extension having been met with below this Umit. On the com- 

 pression side of the diagram the curve D falls so low that it intersects 

 the curve F. Thus in the series of experiments made with a compres- 

 sive stress of 12"2 tons per square inch, all the specimens failed in a 

 ductile manner, some showing signs of cracks wliile others simplv settled 

 down suddenly without visible cracks. Tt is remarkable that the zone 

 between the two curves F and D is no wider than is shown in the 

 diagram. 



In Fig. 6 the maximum and minimum values of the stresses 

 producing fatigue fracture are plotted on a base representing the 

 steady component of the stress-producing- fatigue. The elastic limit, 

 yield stress, and maximum strength of the metal are represented by 

 the points E, Y. and M on the line OM. passing through the origin 

 and inclined at 45° to the axes. The full lines show the loci of the 

 maximum and minimum stresses for the B.A. mild steel, while the 

 dotted lines indicate, for the sake of comparison only, the corresponding 

 results obtained with a sample of Naval Brass. (A dotted line in 

 Fig. 4 likewise indicates the same results for Nnval Brass.) The strik- 

 ing difference between the forms of the loci indicates that a great deal 

 of experimental work with different metals is still required before any 

 general theory can be evolved. 



As indicated in the table which follows, the ratio between the 

 semi-range of the limiting fatigue stress (with eoual intensities of 

 tension and compression) and the maximum strength of the metal is 

 approximately O'Sl. In other mild steels tested by the writer, the 

 value of this ratio has varied between 0'5 and 0-6, the lower value being 

 met with in annealed metal and the higher in cold-strained specimens. 



