_ MI; .1 \irii; ON THK KF.COVKKY OF IKON KKOM OVERSTRAIN. 



square inch, ami immediately after this yield-point had been passed, Curve No. 2 was 

 obtained. 



h will } n..ti(v<l t'mm the diagram that in Curve No. 2, the removal of the load 

 was stopped midway and the stress of 20 tons per square inch allowed to act over a 

 niijlit. Had the load l>een entirely removed, then Curve No. 2 would have continued 

 in some such fashion as is illustrated hy the dotted line in the diagram. 



The continued action of the load of 20 tons was found in the morning to have 

 produced the slight extension shown. On then testing the specimen by first increasing 

 and then decreasing the load for a short range on either side, Curve No. 3 was first 

 obtained, and then Curves Nos. 4, 5 and 6. 



Curve No. 3 shows a short range of nearly, though not quite, perfect elasticity. 

 A slight discrepancy of sooooth of an inch was obtained on each side of the starting 

 !><>sition. 



( 'urves Nos. 4, 5 and 6 show how elastic behaviour is departed from, and greater 

 and greater indications of hysteresis obtained as the range of loading is increased. 



Curve No. 7, drawn on the bottom half of Diagram VI., shows the effect of applying 

 a load of 20 tons to this same specimen, and allowing it to act for over sixteen hours 

 liefore proceeding with the loading. That is, the effect produced by a prolonged pause 

 in the loading of an overstrained specimen is shown. After the pause, the curve 

 starts off at a much steeper gradient ; but shortly it falls back again to a rather less 

 inclination than if there had been no interruption in the loading. The dotted line in 

 Curve 7 shows the manner in which the curve would have continued had no pause 

 occurred in the loading. This continuation was accurately known ; for previously the 

 specimen had been loaded to 40 tons, three times in succession, and the last two 

 applications had given very accurately the same curve no two readings differing by 

 more than yoooo^bs of an inch. At the stress of 34 tons, in Curve No. 7, the effect of 

 a three minutes' pause is shown to be similar to, but of course much smaller than, the 

 effect of the long pause at 20 tons. This slight effect at the higher load may, however, 

 be explained by simply considering that if creeping be allowed to occur at any load, 

 then a small increase of load cannot be expected to produce so great an extension as 

 it otherwise would. 



After Curve No. 7 was obtained, the complete cycle represented by Curve No. 8 

 was gone through. The part of this cycle lettered a represents the partial removal 

 of the load from Curve No. 7. The stress was only reduced to 20 tons, and then one 

 and a-half hours were allowed to elapse. Slight back-creeping occurred instead of 

 forward-creeping, such as happened in Curve No. 2. The load was then increased 

 and Curve b obtained. This curve arrived very accurately at the same top point 

 as Curve a. The load was then entirely removed, and Curve c, which coincides 

 s. far with Curve a, was obtained. After a pause of five minutes under no load, 

 during which K-.r-k-pivi-ping showed itself, the load was re-applied to 20 tons 

 (<'urvi. </), decreased to /.,-ro (Curve r), and then increased to 40 tons (Curve/), 



