THE FORTH BRIDGE. 427 



ends, increased rather than reduced its strength. Thus one steel tube, 

 20 diameters in length, tested by the author, bore 22 tons per square 

 inch, whilst a similar tube of half the length bore only 19-2 tons. 

 Again, a round-ended column, 20 diameters in length, bore lOS tons per 

 square inch, whilst a flat-ended one failed with 18 tons. No doubt 

 greater regularity would obtain with full-sized than with model tubes, 

 but still the pi-actical fact remained that, within the limits occurring in 

 the Forth Bridge, the compressive resistance of the tubular struts would 

 be as high as that of the top flange of any plate or box girder that could 

 be built, and that whatever stress was admissible in the girder would be 

 admissible in the struts. This was a reassuring result to arrive at, be- 

 cause the practical experience had been chiefly with girders, and not with 

 tubular struts. 



The author's experiments conclusively prove that steel is far superior 

 to iron as a material for struts, though the superiority is not so great as 

 when tension members are in question. Thus the Forth Bridge struts 

 will be from 30 to 40 per cent, stronger in steel than in iron, whilst the 

 tension members will be about 50 per cent, stronger. It does not follow 

 that the steel strut would not be 50 per cent, better also as regards actual 

 work, which is very different to what takes place in a testing machine. 

 A steel or an iron rail, tested for transverse strength in a machine, will, 

 as a rule, bend many inches, and fail by distortion of the head under the 

 compressive stress. In actual work hundreds of such rails break, but it 

 is the tensile and not the compressive stress which causes the failure, and 

 there is no distortion of the head as in the testing machine. Similarly, 

 when riveted girders break under traffic, it is not the top flanges with 

 a calculated stress on the average of about one-third of the ultimate re- 

 sistance that give way, but the bottom members, where the calculated 

 stress is only about one-fourth of the ultimate resistance. In short, the 

 universal experience is that fatigue is far more injurious to iron or steel 

 under tensile than under compressive stress, and it follows that the factor 

 of safety should not be the same in the two cases. This is quite con- 

 sistent with ordinary practice, for probably the majority of girder bridges 

 in this country have equal-sized top and bottom flanges, which, after 

 allowing for the riveting, would give a factor of about three for the com- 

 pression and about four for the tension members respectively. 



The peculiarities of steel are tolerably well undei-stood now, and 

 amongst other precautions it is especially desirable to so design the joints 

 in tension that no tearing action shall be set up along a line of rivet- 

 holes. Accidents of workmanship necessitate the provision of a good 

 factor of safety in steel joints in tension ; but, after watching the testing 

 of many steel tubes under compression, the author cannot conceive any 

 accident of workmanship which could bring about the failui'e of steel 

 tubes such as those in the Forth Bridge, even if the working stress was 

 raised to two-thirds, instead of about one-fourth, of the ultimate resist- 

 ance. ' In fine, he believes a steel tube to be the most trustworthy mem- 

 ber which could be introduced into a great and unprecedented work. 



As regards cost of manufacture, the difi'erence is comparatively 

 small between a circular and a rectangular member. The main tubes 

 will be made up circumfei-entially of ten bent plates, lap-jointed and 

 double-riveted through the flanges of ten rolled beams, which will run 

 longitudinally through the whole length of the tube. Stiffening rings 

 will also be introduced at suitable intervals. The junctions of the tubular 



