The Rupture of Steel bi/ Longitudinal Stress. 245 



strength of a metal, viz., by dividing the maximum load by the 

 original area, is a purely conventional method, and does not repre- 

 sent any real stress whatever; it simply shows what load would 

 be sustained by a given section before it broke, and though it is no 

 doubt a useful figure for engineers to know, it does not tell us 

 anything about the actual stress at fracture; this can only be arrived 

 at by dividing the load on the specimen at the point of rupture by 

 the contracted area measured after the specimen has broken. In 

 this paper, the breaking stress will always be measured in this way, 

 and will be referred to as the " true tensile strength" of the metal. 

 This is what M. Considere in his ' L'Emploi du Fer et de 1'Acier ' 

 (Paris, Dunod, 1885) calls " resistance de striction." 



It is well known that when a bar is subjected to tension the stress 

 not being uniformly distributed, as, for instance, when the pull is not 

 central the mean stress borne by the bar at rupture is less than it 

 would be if the stress had been uniformly distributed. 



It is stated in Thomson and Tait's ' Natural Philosophy,' Part II, 

 p. 258, that " a solid of any elastic substance, isotropic or seolotropic 



experiences infinite stress and strain in the neighbourhood 



of a re-entrant edge or angle, when influenced by any distribution of 

 force, exclusive of surface tractions infinitely near the angles or 

 edges in question." 



Three steel bars, numbered 829, 830, and 831, were taken and cut 

 in three pieces ; one piece was tested plain, and the second piece 

 with a \/-groove turned on it (see fig. 1). The tool cutting the 

 \/-groove was made with its cutting edges at about 90, and the 

 point as fine as possible. The results are given in Table I. It is 

 clear that the \/-groove is very prejudicial. For the same reason a 

 \/-groove with a rounded angle must be prejudicial, though not to such 

 an extent, since the distribution of stress is more uniform. Specimen 

 No. 834 was cut in three pieces and tested, one plain and one with a 

 groove of the same shape as Nos. 829, 830, 831, but with the point 

 of the cutting tool just rounded off. The strength of the grooved 

 bar is now 0'95 of the plain. Similarly, with specimens 822 and 50, 

 the grooved pieces have strengths of 84 and 89 respectively. These 

 experiments show that the mean stress at rupture diminishes as the 

 angle of the groove becomes more acute. 



When, however, we come to test specimens with a groove as in 

 fig. 2, we find that they are stronger than the plain specimens. 

 , Table I gives the results of such experiments on seven steel bars 

 -(including the four already mentioned). Each bar was cut in three 

 pieces. 



The plain bar was tested first, and the groove then cut in the 

 I second piece to the same diameter as the contracted area in the plain 

 I piece, so as to secure as much as possible similarity of conditions. 



