382 REPORTS ON THE STATE OF SCIENCE, ETC. 
The error introduced by not applying the corrected modulus in this example would 
be as follows: 
Correct mean stress on gross area 0-0004 x 10,150 = 4-06 tons per sq. in. 
Approx. ,, 5s 0-:0004 x 12,150 = 4-86 ,, ,, 4, error 20%. 
Correct mean stress on net area 0-0004 x 10,150 + 0-741 = 5-48 tons per sq. in. 
Approx. ,, 4 BS 33 0-0004 x 12,150 + 0-741 = 6°56, ,, ,, 
error 20%, 
The discrepancies often found between calculated stresses and those deduced from 
measured strains are no doubt partly due to inaccuracy in the modulus, as well as to 
stiffness of end connections, etc. 
Conclusion. 
The intricacies of the problem suggest that useful progress can best be made by 
experiment and measurement undertaken in a systematic manner. 
The course by which the required knowledge can be most readily acquired would 
appear to be by treating existing bridges as full-size experiments. What can be done 
in this direction was demonstrated in the extremely valuable paper by Messrs. A. C. 
Cookson and J. S. Nicholas ! read at the Hull Meeting. 
A bridge’s history and the traffic it has carried is on record, and the strengths and 
stresses can be determined as accurately as required; also the loads they have carried 
can be related to the stresses calculated in the usual simple manner. 
With this information, by the use of similar calculations and similar nominal 
stresses, girders with the desired endurance could be designed. This course avoids 
the numerous difficulties of secondary and indeterminate stresses to which reference 
was made at the Hull Meeting by Mr. Conrad Gribble? in a paper full of valuable 
suggestions, 
The investigations outlined above, together with some experiments to determine 
the relative damaging effects of want of balance in old locomotives compared with 
new ones, would permit of allowances being made for changes of that nature. As 
suggested by Mr. Gribble, the testing to destruction of old bridges on their removal 
from service would provide valuable additional data, especially if fatigue tests were 
applied. 
The other course, which necessitates the establishment of accurate allowances for ~ 
impact and a standard stress, has, in addition to the impact difficulties, to overcome 
that of fixing the stress and details of the method of its application. 
Some of the difficulties are reflected in the following resolution® passed by the 
Indian Railway Bridge Committee after four years’ work: 
* After carefully studying the Fereday-Palmer extensometer records of 1500 tests 
on bridges of spans varying from 15 feet to 358 feet on the North-Western Railway, 
the Committee are of opinion that, although these tests do not afford sufficient in- 
formation to warrant their recommending any modification of the Pencoyd formula 
for impact at present, the records indicate that a single load concentrated at the 
main driving-axle of each engine passing over a bridge more accurately represents 
the true allowance for impact than a uniformly distributed load. The Committee 
are also of opinion that many more tests are required before a scientific formula 
covering all sources of impact can be evolved.’ 
APPENDIX. 
Faticur Test oF A RIVETED GIRDER 20 Fert LONG, MADE BY 
Sirk WinL1AM FArIRBATRN. 
Particulars of the experiments are given in the Report published by the Board of 
Trade *in 1864, and additional information will be found in various papers and in 
Fairbairn’s books. 
1 Engineering, September 8, 1922. 
2 Engineering, September 8, 1922. 
3 Indian Railway Bridge Committee, Fifth Report. - 
* Report of Mr, Fairbairn to the Board of Trade, Command Paper 1864. 
