208 MR W. GORDON AND MR G. H. GULLIVER ON 
the width increases, though in a somewhat irregular manner. The more striking method 
is to compare the actual shapes of the fractured bars. This has been done in 
fic. 10, where the mean outlines of the bars originally about }$, 1, 2, and 4 inches 
wide are drawn to the scale of their original dimensions, all reduced to the same size 
for convenience of comparison. It is evident that the proportional reduction in width 
1-0 $$ a ——s 
0-9 2"anol" 
Va" 
r06 
ae oh 
al> 
Spo % 
{jz ep) 
Z|s ; 07 Alt 
(oe) =< 
06z\5 
Se 
0-5 z|\7 
0:3 -}o 
p< 6 
I 
0-4 ae 
Zz|O 
0-2 iq w/o 
03 15 
0:2 
0-1 
0-1 
L I ae | ee + mee Aili | | Len} | 
CENTRE Ot 0:2 0-3 0-4 0:5 0-6 0:7 0:8 0-9 1-0 
FRACTURE Leh 
Fie, 10.—Mean profiles of width of flat steel test-pieces after fracture ; measurements taken from bars having widths of 
4, 1, 2, and 4 inches, and a constant thickness of 4 inch. 
Insct.—Mean profiles of thickness of same bars, measured on longitudinal centre line. 
is less the wider the bar, and that there is consequently less extension of the wider 
bars in the constricted region. The inset in fig. 10 shows in a similar manner the mea 
thickness of the same four sets of bars, as measured along the longitudinal centre line; 
the order of the curves is just the same as for the width. It is interesting to note that 
the final minimum thickness of all the bars measured at the broken surface is nearly 
constant. For the larger diagram the origin is at the centre of the gap, while for the 
