THE STRENGTH AND DUCTILITY OF FLAT STEEL BARS. 201 
tion to the actual maximum stress is obtained by dividing the actual load supported at 
the instant before fracture by the minimum cross-sectional area of the broken bar; it 
may be called the mean breaking stress. The actual load carried just before rupture 
is difficult to determine with accuracy on account of the localisation of the distortion 
and the relative rapidity with which the metal extends in this region, but much care 
was taken to secure correct results. In order to determine the minimum cross-sectional 
area of the broken bar the average thickness of the metal was obtained by a number of 
measurements at equidistant points along the fracture, and this was multiplied by the 
minimum width of the bar; the difficulties of measurement do not admit of extreme 
accuracy here. ‘The values of the mean breaking stress are given in Table III., and are 
60 
TONS PER SQ.IN. 
a 
(2) 
> 
ro) 
STRESS. 
WIDTH 
THICKNESS 
Fic. 3.—Variation of mean breaking stress of flat test-pieces with change in the ratio of width to thickness ; measurements 
obtained from bars of soft steel having widths from 4 inch to 4 inches, and a constant thickness of 4 inch. 
plotted in fie. 3; the amount of variation is not great, hardly more than may be due to 
‘uncertainties of measurement. 
5. EXTENSION. 
The measurements of extension were taken chiefly from that side of each bar which 
as divided originally into inch lengths, and were always made symmetrically on each 
side of the inch within which fracture occurred. This procedure is necessary in order 
fo nullify the variations due to the position of the fracture, and is employed in most 
portant testing laboratories. The measurements were made along the longitudinal 
centre line of each bar, and the breadth of the gap left between the two broken surfaces 
vas deducted from the total length. The gap is usually broadest at about the middle 
of the width of the bar, owing to the fact that fracture generally begins there, and that 
“the neighbouring unbroken parts continue to stretch until they are sundered in turn. 
Fig. 4 is an outline of one of the bars after fracture; the thick irregular line shows the 
gap left when the two broken halves are pressed tightly together. The variation in 
the breadth of this gap with change in the size of test-bar can be obtained from Table 
Y. and figs. 8 and 12. 
In Table IV. are given the mean extensions of original lengths Of plends Onto wae 
