82 



a more accurate method tliau that often used, Involving the breaking of 

 a half dozen steel liars from one rod at different laboratories and compar- 

 ing the average breaking load. 



In the work of examining the accuracy of the Purdue University test- 

 ing machines, three nickel steel bars were used; two with a length be- 

 tween shoulders of 12 inches, and one with a similar length of 30 inches. 

 In the case of the latter bar, it was possilile to attach two extensometers 

 to the bar in tandem, and by exchanging the position of the extensometers 

 to compare the latter. 'J'he modulus of elasticity was measured -in case 

 of each bar on the machines of the three laboratories using the extenso- 

 meters possessed by the three laboratories. One of these extensometers 

 was sent from one laboratory to the other. If the extensometers are alike 

 in their graduation and the modulus of elasticity of the bars is found to 

 be equal on the various machines, the latter may be judged to have no 

 relative errors. The observations at the University of Illinois were taken 

 under the direction of Professor A. N. Tall tot. 



Taking the average of three bars tested at the three laboratories it 

 appears (Table I) that the value of Young's Mudulus at the Purdue labo- 

 ratory is (in 1(X),0(X) pounds per square inch unitsi 20.22: at Illinois 

 laboratory. 20.3.3: at Watertown laboratory. 2S.GG. Between the Purdue 

 lab(irntoi-y and the Illinois lalioi-atoi'v there is tlius a relative differ- 

 ence of only about i-j of one per cent., an accuracy much in excess of 

 that needed in any work for which these machines are used. 



By intcrclianging tlie ixisitidns of two extensometers in case of the long 

 bar. an opjiortuniiy existed of comparing the indications of two extenso- 

 meters ^)f different type— the Uiehle extensometer (a screw micrometer) 

 and the Johnson extensometer (a roller type). In Table II it is seen that 

 the two extensometers yield identical results. 



It is thus assuring to know the reliability of the ordinaiy type of 

 testing macliiiie and extensometer. If tlie WatertoAvn machine is con-ect, 

 the other machines yield result.s about two per cent. high. The Water- 

 town extensometer. however, was not compared with the othei* extenso- 

 meters. 



ELASTIC CHANGES. 



After the work of calibrating apparatus was complete, two of the 

 bars were used in the study of the varinbility of the proportional and yield 

 limit. 



