436 REPORT— 1887. 



l)ent laterally. Generally the bars were 1 inch wide by ^ inch thick, and 

 32 inches long between the bearings. The steel specimens were cut 

 from the tension member plates of the Forth Bridge, and had a tensile 

 strength of about 70,000 lbs. per square inch, and an elongation of 20 per 

 cent, in 8 inches. The iron specimens were rolled bai"s. 



The different effects produced on different materials by the frequent 

 repetition of stress is well shown by those experiments — thus comparing 

 Nos. 8 and 14 in Series No. 1, the stress applied being in each case about 

 40 per cent, of the ultimate strength, the hard steel failed with only 

 32,445 revolutions, while the soft steel withstood 14,876,432. Again, 

 comparing experiments Nos. 16 and 23, it will be seen that with about 

 the same number of revolutions the hard steel, though of more than 

 double the tensile strength of the iron, broke under a repeated stress only 

 10 per cent, greater, thus demonstrating that the ultimate tensile strength 

 of a metal as observed in a testing-machine is no adequate measure of its 

 value as a material of construction . 



Other points of interest may be referred to in connection with Series 2, 

 In general the bars were tested in pairs, so that when one bar broke, its 

 companion could be otherwise tested and examined. For example, the 

 companion to No. 28, after being subject to 18,140 bendings, was tested 

 for tension, and failed with 48,000 lbs. per square inch, and 2'6 per cent, 

 elongation; the original strength of the steel being 70,000 lbs. and 20 per 

 cent, elongation. Again, the companion to No. 32 was, on close exami- 

 nation, found to have a flaw like those found in crank-shafts. Nos. 33 

 and 36 were companion bars bent one way only, so that the stresses were 

 not alternating, hence the largely increased endurance. They were both 

 taken out before actual fracture, but with deep-set flaws, clearly illus- 

 trating that the cause of failure under repeated stresses is very frequently 

 not so much a gradual deterioration or crystallisation of the metal, as the 

 establishment of small but growing flaws. 



Another noteworthy fact illustrated by these experiments was, that a 

 structure or piece of mechanism may be subject to a repeated stress equal 

 to 90 per cent, of that which would break it, and yet specimens cut from 

 the metal may exhibit no signs whatever of deterioration. The broken 

 half of nearly every specimen in Series No. 2 was tested with that 

 result. Thus, as the stress was applied at the centre of the bars, it fol- 

 lowed that at a point distant 90 per cent, of the half-span from the 

 bearings, the stress would be 90 per cent, of that which broke the bar. 

 Although the bars broke short off at the centre, at the point referred to 

 they could invariably be bent double without fracture. Having reference 

 to this fact, and to the fact that the tensile strength was also little 

 affected, Mr. Baker considered that it was hopeless to expect to learn much 

 from testing specimens of raetal from structures or machines which have 

 been long in use, unless the experimenter happens to hit off the right 

 moment immediately preceding the commencement of failure. 



In order to ascertain whether alternating stresses were as prejudicial 

 to members, such as piston-rods, subject to direct pull and thrust, as to 

 shafts subject to transverse bending, a series of experiments (No. 3) was 

 carried out on specimens so designed as to give alternate direct tension 

 and compression on small pieces of metal. These specimens were of three 

 types, illustrated (not to scale) by figs. 1, 2, and 3. In the first, the 

 pieces of metal tested were sometimes of round and sometimes of flat 

 cross-section, and were bolted to a couple of spring bars, as shown on the 



