370 REPORTS ON THE STATE OF SCIENCE, ETC. 
strengthened up to the standard adopted by the Board of Trade, on condition that 
the Treasury should grant £150 to defray the expense of experiments to ascertain 
the durability, and the measure of strength to be allowed, of wrought-iron bridges, 
subjected to changes, shocks, and vibrations of a continued and variable load.’ ! 
This fatigue test carried out on a riveted girder 20 ft. long is unique. Dr. W. C. 
Unwin, F.R.S., who assisted Fairbairn in making this remarkable experiment, which 
continued for two years, remembers the circumstances well and has kindly helped to 
supply information. The load was applied at the centre of the girder at the rate of 
about eight changes per minute and the shock was sufficient to make the girder 
vibrate vertically for some seconds at every application. 
The details of the experiment are rather scattered among Fairbairn’s books and 
Reports. Some particulars with additional figures are given in the appendix hereto. 
The experiment proved that the metal of the girder would withstand over three 
million repetitions of a stress, accompanied by shock, of 6°25 tons per sq. in. 
on the net area, and we know now that if three million repetitions fail to rupture, 
the stress can be little if any greater than that which may be applied an unlimited 
number of times. 
Notwithstanding the result of the experiment, the 5-ton limit remained un- 
altered, and this preliminary reference to the introduction of the stress system of 
design and the fixing of the 5-ton limit is given in order to emphasise the fact 
that the limiting stress fixed was not based on any very thorough investigation of 
the stresses in the successful or unsuccessful iron bridges built before the introduction 
of the new limit. Among the bridges still in daily use, therefore, we have, as stated 
in the extract from Fairbairn above, bridges designed intentionally with various 
factors of safety. The stress limit having been fixed, however, any disregard of it 
led to delay in the opening of new railway extensions,” consequently all new railway 
bridges had to be built to that limit. Although stipulated as the maximum, stresses 
in those days were only calculated approximately, and tke 5 tons per sq. in. can 
only be regarded as a nominal stress. 
Stresses Considered in Design: 
Changes of stress in bridge members can be determined with considerable accuracy 
by strain meters, as has been done in the recent investigations. An increase of stress 
measured in this way * while an engine or train passes over at speed would generally 
be greater than that measured with the engine standing. Both these measured stresses 
would be different from the figure arrived at by calculation in the usual way even 
for the exact axle-loads of the engine and train. Denoting the several stresses by 
symbols, we have: 
fo = stress increase measured under travelling load; 
LOSS os rE » standing load; 
jp = a = calculated for $ 33 
fa = stress due to dead load of structure. 
The difference between f, and f, may be due to speed combined with its concomitant 
vibrations of bridge and load ; that between f, and f, to the method of calculation 
not being sufficiently exact as to stress distribution or inclusion of secondary stresses. 
The difference between f, and f,, though not actually measured in the early days, 
was nevertheless realised from the beginning, and some system was required for the 
graduation of the nominal stresses so as to enable bridge members to withstand the 
expected increase without damage. The next development was the preparation by 
several engineers and various railway authorities of designing specifications which, 
in addition to fixing the stresses, gave instructions regarding details of design and 
formule for compression members. Two distinct methods have been followed in 
drawing up these specifications. 
1 Researches on the Application of Iron to Buildings. W. Fairbairn. 
2 See reference to Torksey Bridge at Hull Meeting, Engineering, Sept. 15, 1922, 
p. 350. 
3 A stress deduced thus from strain must of necessity be the increase due to the 
live load. The dead-load stress cannot be measured, as the girder cannot be relieved 
of the dead load. 
