Determining the Elastic Constants of Glass. 555 



the surface of the beam or the cover-plate which increased 

 or diminished the observed values in the first case will have 

 an equal and opposite effect in the second case. The mean 

 values of the curvatures so obtained should then give a value 

 for the elastic constants independent of any initial small 

 curvatures of the beam or the cover-plate. The method 

 does not eliminate errors due to the bending of the cover- 

 glass under its own weight, but in the case of the cover- 

 glass used in these experiments (0*5 cm. x 6 cm. x 7 cm.), 

 the radius of maximum curvature so produced was of the 

 order of 10 7 cm., and its effect was therefore negligible. 



In order to avoid the mechanical difficulties of reversing 

 the bending-moment with the apparatus employed, the beam 

 with the wooden arms attached was inverted on the knife- 

 edges and the load applied as before. The cover-glass was 

 then placed beneath the beam, and by means of three 

 levelling-screws of small pitch was adjusted into contact 

 with the under surface of the beam. 



The method is illustrated by two sets of observations 

 given in Table V. 



& 



Influence of the Distance Apait of the Knife-edges. 



Some preliminary experiments were made to determine 

 the distance apart it was necessary to have the knife-edges 

 for a beam of given width in order to eliminate end-effects. 

 This w r as done by observing the effect produced on the cross- 

 bending by the insertion of rubber pads between the glass 

 and the knife-edges. 



It was found that when the distance between the knife- 

 edges exceeded three times the horizontal width of the beam, 

 no perceptible change was made in the bending by the 

 insertion or removal of the rubber, whatever load was 

 employed. When the distance was between 2*5 and 3 times 

 the width of the beam, however, the presence of the rubber 

 increased the cross-bending by an appreciable amount for 

 large curvature-, though its effect was still negligible for 

 small curvatures. 



It seems probable, therefore, that Straubel's values for 

 beams 2 cm. wide with the knife-edges 5 cm. apart are 

 affected to some extent by this source of error. The effect 

 of the knife-edges would be to reduce the transverse curvature, 

 and Straubel's values given in his summary on p. 553 show 

 on the whole a decrease in the value of Poisson's Ratio as 

 the distance between the knife-edges is reduced even for 

 beams of only 2 cm. width. 



2P2 



