286 



REPORTS ON THE STATE OF SCIENCE.— 1916. 



All the readings given in columns a, h, c, d, and e are for the tests 

 in which the torque was applied by weights. During the intermediate 

 runs of repeated stressing at the frequency of 200 per minute, readings 

 were taken of the range of strain which corresponded very fairly (see 

 Table) — up to Test No. 7 — with the ranges obtained in the dead-weight 

 tests at the same ranges of stress. The former readings, i.e. at 200 per 

 minute, are read to the nearest 05 scale division. It will be noticed that 

 there was a distinct increase of range of strain and of width of hysteresis 

 loop during the 36,000 cycles at +5 50 tons per square inch ; and a larger 

 increase in both of these during the 228,000 cycles at ±5-62 range. Also 

 at the change of speed, after the run of 228,000, from 200 to 8 cycles per 

 minute, the range of strain altered from 6-90 to 7-24 ; this is an example 

 of the speed effect already found by the author in previous work.^ 



It appears, then, that for the steel tested there is a limit to elastic 

 ranges of strain in the neighbourhood of ±5 50 tons per square-inch 

 range of stress. A torsion test, made with continuously increasing torque, 

 of another specimen (solid) cut next iu order to the specimen of these 

 tests from the same bar of steel, gave a }Tield point of 9 85 tons per square 

 inch, and a limit of lU'oportionality in the neighbourhood of 5 80 to 6 tons 

 per square inch. 



After Test No. 8 (see Table\ a succession of tests at smaller ranges 

 of stress showed the hysteresis loop to be wider even than at the higher 

 ranges of stress of the cycles imposed before the limit to the elastic ranges 



Test Specimen. 



■Square p^,^ii^,for2' 



Square 



Holes For bolts to nx 

 mirror holders. 



of stress had been passed. The apparent recovery of elasticit}^ with 

 rest in Test No. 15 is presumably the counterpart for alternating cycles 

 of the well-known phenomenon of recovery with rest after overstrain. 

 The foregoing experiments illustrate the following points : — 



At a range of stress — applied by equal direct and reverse torsion — 

 which may be determined with considerably more accuracy than the 

 elastic limit in a static test (i.e. with slowly increasing stress in one direc- 

 tion\ the hysteresis increases largely with continued application of the 

 cycles. At a smaller range of stress, the increase of hysteresis, if any, 

 is very small and may probably be regarded as an increment of elastic 

 hysteresis. Other of the author's tests have shown that 250,000 cycles 



■ ' On Speed Efiect and Recovery in Slow-speed Alternating Stress Tests ' — Froc. 

 Boy. Soc. A, vol. 92, 191fi. 



