ACCURACY. 99 



their sensitiveness much sooner than others. If it is 

 noticed that a machine is becoming less sensitive, the knife 

 edges must be looked to, and, if necessary, taken out and 

 re-ground. 



The results of experiments made some years ago by 

 Prof. Kennedy, on a Greenwood machine which had been 

 used for over twelve thousand experiments, showed that the 

 increment of load required to produce a movement in the 

 last lever varied from T ^V o th to g-j-- th of the total load at 

 5,000 Ib. and 50,000 Ib. load respsctively. In some 

 machines, whose knife edges are quite new and in good 

 condition, the sensitiveness may be TO~.FO oth ; while in fluid 

 pressure machines, such as the Emery, still greater sensitive- 

 ness is attained. 



One ten-thousandth of 100 tons is 224 ^ 00 = 22-4 Ib. ; so 



that with this degree of sensitiveness, at the full load of 

 100 tons, the lever would be moved by 22 '4 Ib. ; this is 

 in excess of practical requirements. 



The sensitiveness and practical absence of friction of a 

 good knife edge have been illustrated by Mr. Wicksteed, 

 in a description of a simple experiment which he made on 

 one of his large single-lever machines. He first broke 

 twelve pieces of No. 16 cotton thread, the average breaking 

 load being 3'5 Ib. ; then, twelve No. 12 threads were broken 

 in the same way, at an average breaking load of 3'81b. 

 Pieces of the same threads were then broken by the appli- 

 cation of deadweights at precisely the same loads. 



In an Emery machine exhibited at Paris it is recorded 

 that a specimen was broken at a load of 90,000 Ib., and 

 immediately afterwards a horsehair was put in the machine 

 and broken at the very small load of 1 Ib. Mr. Adamson, 

 a few years ago, described how he had found that the 

 recording lever of his machine was sensibly affected by 

 differences in the temperature of a test bar in the machine 

 produced by simply placing his hand upon it. 



These few instances serve to show that a degree of 

 sensitiveness exists in most machines, especially when their 

 knife edges are new, which is safely in excess of the 

 requirements of even the most refined tests that are made. 



55. Accuracy. In all testing machines the load, as 

 indicated by the measuring appliances, should be truly and 

 accurately the real load on the specimen. Accuracy is only 

 a relative term no physical apparatus is absolutely accu- 

 rate ; that is to say, all experimental appliances are liable 



