384 REPORTS ON THE STATE OF SCIENCE, ETC. 
The results of the tests and calculated stresses are given in the table opposite. 
It is unfortunate that exact details and observations of this very remarkable 
fatigue test are not on record. Full details of the girder are not available. The 
plates attached to the Board of Trade Report have been assumed correct in pre- 
paring fig. 13. The total weight is given by Fairbairn as 7 ewt. 2 qrs., but the weight 
of the girder shown (assuming the stiffeners to be double 3” x 14” x 2” tees joggled 
on to the main angles) is below that figure. 
The rivets were }in. diameter and arranged as shownin elevation ; the pitch would 
be about 44 inches. The arrangement of riveting in the flanges is not shown. In 
estimating the net area of the bottom flange it has been assumed that there were 
two rivets in line breaking pitch with the rivets in the vertical legs of the angles. It 
is important to note, however, that in Fairbairn’s calculations of the stresses in the 
net area he has deducted 0-625 sq. in., which is the area of the three rivet holes. 
The stresses given in the table have been calculated in the following ways : 
Fairbairn’s, by the formula 
tress = 
stress =—— 
4 ad 
1 = span in inches, taken as 240, the clear span ; 
w = the weight at the centre in tons ; 
d = overall depth (16 in.) ; 
«a — area of the flange, the total sectional area of the plate and two angles (2-4 sq. 
in. gross, 1-775 sq. in. net). 
Method I. (a), by moment of inertia, ignoring the rivet holes and then increasing 
the stress in the ratio of the gross to net areas. Two rivets allowed for. The span 
of girder taken as 21 feet. 
Method II. (a), by moment of inertia calculated with the two rivet holes out. 
Span, 21 feet. 
The testing arrangements were intentionally made so that the load should be applied 
with a jarring effect, and under (6) of Methods I. and IJ. an increase of from 7 to 10 
per cent. is made for the velocity with which the load was probably applied. 
Illustrations of the testing arrangements indicate that the load was applied to the 
bottom flange, as shown in fig. 13. Severe secondary stresses must have been set 
up in the flange with every loading, and although the calculated stresses required to 
cause failure were comparatively high, they would probably have been much higher 
had the load been applied to the top flange. The distance from the centre of the 
point where the flange failed is not recorded. 
The endurance of some of the rivets is remarkable. In referring to the first test, 
Fairbairn writes: ‘It is satisfactory here to observe that during the whole of the 
1,005,175 changes none of the rivets were loosened or broke.’ At the centre of the 
beam the rivets holding the lower angles to the web must have been very severely 
loaded. The 4-ton load, if taken by the three nearest rivets, would give a bearing 
pressure of 21-3 tons per sq. in., but a shearing stress of only 3-4 tons per sq. in. 
A second experiment was made on the same beam. It was ‘ repaired by replacing 
the broken angle-irons on each side, and putting a patch over the broken plate equal 
in area to the plate itself.’ The results of the second test are given in the table. 
e 
