198 MR W. GORDON AND MR G. H. GULLIVER ON 
one proportional to the square root of the cross-sectional area were employed. By 
means of tests of round, square, flat, angle, and channel bars, BarBa has shown that 
this expectation is nearly fulfilled (2). The longer standard round test-bar in use on the 
Continent is 20 mm. diameter and 200 mm. long, the length being thus 11°3./area. 
bars of other shapes and sizes are therefore given a gauge-length of 11°3,/area when 
comparable measurements are required. ‘The longer British standard round test-bar 
has a length eight times its diameter, equivalent to 9,/area; the shorter round 
standard has a length of 4,/area (1), but for bars of rectangular section the length is 
fixed at 8 inches. 
So long as the section is of compact form—that is, circular, square, or rectangular 
with a width not greater than about three thicknesses—the extension of any bar 
conforms closely with a simple linear equation due to Unwin (4) :— 
e per cent. = 100(a+0: = ; 
in which a and 6 are constants, A is the cross-sectional area, and L is the datum-— 
length. In the case of rectangular bars having a breadth considerably exceeding the 
thickness, this simple equation is not fulfilled. 
3. EXPERIMENTAL DETAILS. 
The primary object of the present experiments is to investigate the variation in _ 
the extension of flat bars of mild steel of constant thickness as the width is gradually 
increased. The advantage of varying the width instead of the thickness of the test- 
bars is twofold :— 
(1) The work of preparing the bars is less. 
(2) The material is of more uniform quality. 
The material employed was soft boiler plate, 4 inch thick. No chemical analysis 
was made, but a section of the metal, magnified 100 diameters, is given in fig. 2 
This shows the usual features of soft steel—the ground mass of ferrite or a-iron, the 
dark areas of pearlite or Fe-Fe,C eutectoid, and a few small slag inclusions of elongated 
form,—and it indicates the presence of 0°12 to 0°15 per cent. of carbon. 
The plate was cut, in the direction of rolling, into strips, approximately 3, 1, 13, 
2, 25, 3, 34, and 4 inches wide, and these were machined carefully on the edges, and 
lightly ground on the wide faces to remove the mill scale. There were three bars of 
each width, with the exception of those 3} inches wide, of which there were only two, — 
Hach bar had a total length of 18 srahies of which 3 inches at each end was held 
within the grips of the testing machine. A centre line was scribed on both wide 
faces of each bar ; that on one face was divided carefully into inch lengths, and on the_ 
other a length equal to 11°3 times the square root of the area of the bar was set out 
and subdivided into a number of equal parts. On both faces the divisions were 
