FATIGUE STUDIES OF NON-FERROUS SHEET METALS 581 



natural frequency of vibration of the specimen and this may super- 

 impose additional stresses upon it. This is determined by observing 

 the operation of this specimen by means of a stroboscope and accu- 

 rately setting the speed of the machine to a point where the vibratory 

 motion of the specimen is uniform. 



The machine has a capacity of forty specimens. Twenty specimens 

 are tested on each end of the motor drive. The machine is statically 

 balanced and is smooth in operation. It is customary to test at least 

 five specimens of each material at each stress. The machine therefore, 

 has a capacity of four alloys of five specimens each at two deflections. 

 The practice of using five specimens for each deflection was adopted 

 because it was seen from the results of previous experimenters in 

 fatigue testing that a more accurate result might be provided by 

 doing so. The capacity of the machine is sufficiently large to permit 

 this being done. 



Materials 



The materials investigated consisted of five alloys of alpha brass, 

 three of which had been hardened with nickel silicide and one alloy 

 each of nickel silver, phosphor bronze and Everdur. The three alloys 

 of hardened brass have been described previously.^ 



The chemical composition of these alloys is given in Table I and 

 the tensile strength, proportional limit, modulus of elasticity, per cent 

 elongation in 2 inches, Rockwell hardness and endurance limit are 

 given in Table II. The heat treatments and amount of cold work 

 expressed by number of B. & S. gauge reductions from standard anneal 

 are shown also on Table II. 



Fatigue Endurance Test 

 Method of Determining Stress in Specimen 

 The method of determining the stress in the specimen consists in 

 clamping a specimen in the same manner as on the fatigue machine. 

 The clamped specimen is mounted on the table of a Societe Genevoise 

 Star Comparator so that its large surfaces are parallel with the vertical 

 plane, the axis being parallel to the table. By mounting the specimen 

 in this manner its weight has no appreciable effect on its deflection. 

 The stress in the specimen is determined from the load deflection 

 curve obtained by applying dead weights >2 inch from the end of the 

 specimen and measuring the amount of deflection for various units of 

 load by means of a microscope mounted on the comparator in such a 



* "Heat Treatment and Mechanical Properties of some Copper-Zinc and Copper- 

 Tin Alloys Containing Nickel and Silicon," by W. C. Ellis and Earle E. Schumacher. 

 Proc. A. I. M. M. E., 1929. 



