246 



Pn>f. Carus-Wilson. 

 Table I. 



O, V, U are the tensile strengths of the plain, y -grooved, and U'? roOTe d bar* 

 respectively in tons per square inch. O,, V,, Uj the same, taking that of O as 

 c is the percentage contraction of area of the plain bar. 



Fig. 1 shows the dimensions of the V'K r ove adopted in all cases, except in 



822] 

 Nos. 834 V , where the corner was just rounded off by the cutting tool. 



50j 

 Fig. 2 shows dimension of the U'g roOTe - 



It will be seen that in every case the U -grooved specimen 

 stronger than the plain, the average superiority being 16 

 cent. 



The effect of the \J-groove by itself in producing non-unifonni< 

 of stress as in the \/-groove would tend to make the (J -groove 

 bar break at a lower stress than the plain bar, where the stress m\ 

 be very nearly uniformly distributed, but, in spite of this prejudi< 

 action, the U -grooved bar is the stronger. 



This phenomenon is quite distinct from that mentioned by mi 

 writers, who have pointed out that a grooved specimen is stronger 

 than a plain specimen of the same material the stresses being 

 reckoned in the conventional manner, viz., maximum load divided by 

 original area (cf. Unwin, 'Testing of Materials of Constructor 

 p. 82, and Burr, 'Elasticity and Resistance of Materials of Engine 

 ing,' p. 230). 



The reason of this is that the so-called tensile strength depends 

 the amount of drawing out before local contraction begins, and since 

 in a plain bar the general contraction of area may be considerable, 

 the actual load on the specimen, at the maximum, is much smaller 

 than on a bar of the same metal in which the drawing out is suppressed 

 owing to the groove. 



In the experiments quoted above, I have discounted altogether the 

 contraction of area, and considered only the actual stress on the 

 section at rupture. 



It is possible that in some cases the metal at the groove is stronger 



