108 



Scientific Proceedings^ Royal Dublin Society. 



symmetrical round both axes ; tliis was no doubt due to the iron wire not 

 being quite free from torsion at the beginning of the cycle. 



The load on the wire was then increased by putting on extra lead discs, 

 and a complete cycle again obtained ; and this was repeated for eight different 

 values of the load. Tlie cyclic curves thus got were plotted on millimetre 

 paper, and the complete area of each determined in sq. cms. by the method 

 of counting the squares. These curves were plotted on a fairly large scale : 

 on the axis of abscissae 1 cm. represented 20° twist, and on the axis of 

 ordinates 4 cms. represented ten divisions on the galvanometer scale ; the 

 latter might have been made to represent coulombs, but for the purposes of 

 comparison scale-divisions were used. 



i-0 O 4-0 80 IfiO 76a 



Angle of Twist. 

 Fig. 3. 



In the sample half-curve given in fig. 3, however, on the axis of abscissae 

 1 cm. still represents 20° of twist ; but on the axis of ordinates 2 cms. represent 

 ten divisions on the galvanometer scale. This ratio was taken because it 

 shows out the proportions of the curve better than the other scale. 



The results obtained for the eight different loads are given in Table III., 

 and are also shown in the curve, fig. 4, in which the stress in grammes per 

 sq. cm. are abscissse, and the areas of the complete cyclic curves are ordiuates. 



