PHYSICS, CHEMISTRY AND ENGINEERING 185 



tests the general form of equation, 



S=AN~ m (1) 



seems to fit test results fairly well/ giving stresses for large 

 numbers of repetitions which are somewhat lower than test 

 results, and hence being on the safe side. In the above equa- 

 tion S is the fiber stress in pounds per square inch, N, the 

 number of repetitions of stress necessary to cause failure, and 

 A and m experimentally determined constants. 



If the repeated stress on a metal is completely reversed there 

 is much more danger of fatigue failure than if the stress varies 

 from zero to a maximum in one direction. An examination 

 by Mr. F. B. Seely and the writer, of the available published 

 data on repeated stress tests, led to the proposed modification of 

 equation ( 1 ) by the separation of the factor A into two parts : 

 one denoted by B, an experimentally determined constant for 

 a material, and the other denoted by tttq , dependent on the 

 range of stress to which the material is subjected, Q being 

 the ratio of the minimum stress to the maximum. For com- 

 pletely reversed stresses Q is equal to — 1, for stress varying 

 from zero to a maximum, Q is equal to zero. 



From the examination of available test data, including data 

 for tests by various experimenters, tests with various kinds 

 of testing machines, and tests of various sizes and shapes of 

 test piece Mr. F. B. Seely and the writer have proposed for 

 metals under repeated stress the general formula' 



q _ ( o ) 



(1 — Q)TR 



For very high values of N this formula seems to give stresses 

 somewhat lower than shown by test results ; however, the test 

 data for high values of N are so meager that, as the formula 

 is on the safe side, no modification is recommended for general 

 use. 



A more convenient form of equation (2) for general use is 

 logS=logB—log(l—Q)— 0.125 logN (3) 



The accompanying table gives values of the constant B de- 

 termined from a study of test data. In using the table, equa- 

 tion (2), or equation (3) a word of caution is necessary. In 

 no case should the stress be taken higher than the safe stress 



: So far as the writer is aware this form of equation for repeated stresses was 

 first proposed by Professor Basguin of Northwestern University in 1910. 



3 See proceedings of American Society for Testing Materials for 1 9 1 S and for 

 1916, Moore and Seely on Repeated Stress. Also "Text-book of Materials of En- 

 gineering," by H. P. Moore, p. 169. 



