MECHANICS AND USEFUL ARTS. 41 



and plate bridges, and also the lattice and trellis bridges. Fifteen 

 years ago experiments were made which led to the construction of 

 the Conway and Britannia tubular bridges on the Chester and Holy- 

 head Railway, and determined the form in which such structures 

 should be designed. Since that time some thousands of bridges had 

 been built entirely of iron. The requirement of five tons per square 

 inch on the part of the British Board of Trade appeared to be 

 founded on no fixed principle. It was well known that the power of 

 resistance to strain of wrought iron depends very much upon the 

 form in which it is combined, and unless the proportions of the parts 

 were permanently established, the five-ton tensile strain might lead 

 to error. For the purpose of making experiments upon the influ- 

 ence of vibration in causing the rupture of beams and bridges, he 

 had constructed a small iron-plate beam of twenty feet clear span, 

 and sixteen feet deep, representing the proportion of one of the 

 girders of the Spey Bridge, and exposed it to conditions similar to 

 those of a bridge subject to changes of load as produced by the pas- 

 sage of trains, and in proportion to the heaviest rolling load. The 

 beam was first loaded to one-fourth of its breaking weight, and it sus- 

 tained a million changes of load without injury. The load was then 

 increased to nearly one-half the breaking weight. With this weight 

 the beam gave way after 5,175 changes. It appeared, therefore, it 

 was not safe to build bridges in which the rolling load would bear 

 this proportion to the breaking weight. The beam was taken down 

 and repaired, and the experiments were then renewed. The load 

 was then reduced to two-fifths the breaking weight, and 25,900 

 changes of load were sustained. Lastly, the load was reduced to 

 one-third, and the experiments were still proceeding, the beam being 

 uninjured after 2,727,754 changes. In calculating the strain upon 

 the area of the metal after deducting the rivet-holes, which, it must 

 be remembered, were larger in proportion in this small beam than in 

 bridsres, he found that the beam would sustain no deterioration with 



^^ ' 



strains of nearly seven and a half tons to the square inch. With ten 

 tons to the square inch the beam broke after 5,172 changes. Now, as 

 the limit of elasticity was reached at about nine tons per square inch 

 in ordinary boiler-plates and bridge-plates, it would appear that it was 

 unsafe to load structures subject to a continually varying load beyond 

 that point. Within those limits, however, there was no evidence that 

 a deterioration of structure took place. For the present, he would 

 advise that in all beams and girders, tubular or plain, the permanent 

 load or weight of the girder and its platform should not, in any case, 

 exceed one-fourth of the breaking weight ; and that the remaining 

 three-fourths should be reserved to resist the roDing load in the pro- 

 portion of six to one. He earnestly directed attention to the laws 

 which governed the resisting powers of girders exposed to trans- 

 verse strains, to the best principles of uniting the joints, and, above 

 all, to the selection of the best material, which, in the parts of the gird- 

 ers subject to a tensile strain, ought always to sustain a test of from 

 twenty-two to twenty-four tons per square inch. The use of superior 

 metal for the bottom of the girders would give an increase of from 

 one-fifth to one-sixth in the strength. There was no economy and 

 he wished particularly to impress this on the Section in the use of 



