426 REPORT — 1882. 



30 tons, the corresponding stresses would be 50 per cent, greater — 

 that is to say, they -wonld range from 6f tons for an all live 

 load to 10 tons for an all dead load. If the stresses be limited as 

 above, the results of experiment and of actual practice show that a struc- 

 ture may be subject to an indefinite number of repetitions of the stress 

 without injury. But, in the case of hurricanes, the repetitions will 

 necessarily be few and far between, and higher stresses are therefore 

 admissible in the members of wind-bracing than in the piston-rod of a 

 steam engine, though both are subject to alternate tensile and compres- 

 sive stresses. When it is remembered that the dead weight between the 

 piers of the 1,700-feet span is upwards of 10,000 tons, whilst the live 

 load due to a couple of heavy coal trains would be less than one-tenth of 

 that amount, the relative lowness of the 7^ tons per square inch maximum 

 stress in the Forth Bridge girders, under the combined action of an im- 

 possible load and an improbable hurricane, will be conceded by all. 



So far as the author is aware — although it has been established 

 beyond all dispute that repeated application of a tensile stress amounting 

 to two-thirds of the ultimate strength of the material would in time 

 cause fracture — it has never been proved that the same conclusion applies 

 to metal in compression. In fact, some of the author's experiments lead 

 him to think that a contrary result might obtain. For example, one of 

 the consequences of heavy varying stresses and vibrations is that the 

 quality of the metal deteriorates, both iron and steel becoming more 

 crystalline and less ductile. These conditions are rather favourable than 

 otherwise to the resistance of a compression member. Thus the author 

 tested some columns, 30 diameters in leugth, of high-class ductile steel, 

 and of inferior crystalline steel, both having, however, the same tensile 

 strength. The inferior steel bore 40 per cent, more load than the high- 

 class steel, and it appears not improbable that, if the quality of the 

 latter had been deteriorated by vibration and heavy stresses the ultimate 

 resistance would have been increased as regards compressive stresses 

 almost as much as it would be diminished in respect of tensile stresses. 

 Similarly, in columns of the same proportion, the commonest class of pig 

 iron would beat the finest brands of wrought iron. 



Long struts were avoided at any expense by some engineers, but, the 

 author thought, without good reason. The tubular struts in the Forth 

 Bridge, if made of iron, would resist as high a compressive stress per 

 square inch as the top flange of any existing girder, and the same remark 

 would apply to steel. An interesting series of experiments was recently 

 made in America with full- sized wrought iron hollow columns, from 

 8 inches to 12 inches in diameter, and up to 28 feet in length. The influ- 

 ence of the length of the column on its resistance was singularly small. 

 For instance, the 8-inch column, when 15 diameters in length, failed with 

 a stress of 16'2 tons per square inch ; and when as much as 42 diameters 

 in length, with 15"6 tons. Now the model of the Britannia Bridge, 

 75 feet in span, failed with 14'8 tons per square inch compression ; 

 Brunei's 66-feet girder, having a 3-feet wide cellular top member, failed 

 with 126 tons ; and some girders tested by the author with 15 tous per 

 square inch as an average on the top flange. 



Taking the mean results of a large number of expei'iments, the influ- 

 ence of length between the practical limits of 15 to 25 diameters would 

 be just appreciable ; but, if only a few experiments were compared, the 

 deduction might be drawn that lengthening a column, or rounding its 



