GENERAL CONSIDERATIONS. 125 



approximately accurate results, if simultaneous readings 

 be taken with two extensometers, one placed on the 

 concave and the other on the convex side of the bar, and 

 a mean of these two sets of readings taken. Such a plan 

 has been very successfully adopted by Professor Bauschinger, 

 who has used two roller and mirror instruments : with this 

 arrangement he has been able to take measurements of the 

 elastic extensions of test bars which have probably never 

 been exceeded in accuracy. Mr. Strohmeyer has, in a 

 similar way, made use of two roller extensometers working 

 with pointers instead of mirrors. 



If, instead of employing two extensometers, only one be 

 used, fixed by two pairs of set-screws midway between the 

 two sides or opposite the axis of the bar, readings will be 

 obtained which are not very far from the correct ones, and 

 the same result will be obtained as if two instruments 

 had been used. This plan has been adopted in the great 

 majority of the more recent extensometers. 



Up to three years ago nothing very definite had been done 

 in the way of comparing extensometers of different kinds. 

 This has now been done by a Committee of the British 

 Association, consisting of Professors Kennedy, Ewing, 

 Capper, Beare, and Unwin.* The conclusions arrived at 

 are contained in a report published in 1890. This com- 

 mittee prepared bars of steel, arranged in four sets of 

 three. Each set consisted of a flat bar of mild steel, 2 in. 

 by in. in section, and two round bars of special steel, 

 l^in. and fin. diameter respectively. Two sets of these 

 bars were circulated among seven observers who had 

 undertaken to make measurements for the committee. 

 The extensometers used included micrometer microscopes, 

 lever multiplying arrangements of different kinds, and one 



T 1 1 



mirror extensometer. In carrying out his observations 

 each experimenter was required to make certain definite 

 measurements at stated loads, and send in his results on 

 special test forms drawn up by the committee. 



The preliminary measurements of the areas of the bars 

 by the different observers compared very favourably, and 

 showed that the error of measurement very rarely exceeded 

 about one-fifth of 1 per cent from the mean value given by 

 several observers, while the difference for any two 

 observers in some cases reached one half of 1 per cent. 



In the measurements of extensions for increments of 

 load of one and a quarter tons in the smaller bars, and 

 * British Association Report, 18U(i. 



