A/A' WILLIAM SIEMENS, F.R.S. 459 



precise purpose for which the material was required. If it was 



wanti'd to make a chain-Krid-v, sici-1 of so tons Invakinj: strain 

 per square inch was the proper material ; if, on the other hand, 

 a suspension link bridge were constructed, the links could hardly 

 be made of steel of such hardness, and the engineer would 

 have to be content with a tensile strength of from 35 to 40 tons. 

 The reason why it would not be safe to take steel of a higher 

 ultimate strength was that its elongation would be insufficient 

 to give a sufficient factor of safety ; in case of unequal strains 

 between the parts, accidents might arise, whereas in dealing with 

 steel of from 35 to 40 tons per square inch, there would be an 

 elongation of something like 10 per cent, before breakage could 

 occur, and within that range the unequal strains could adjust 

 themselves. Again, if it were intended to put steel under a 

 compressive strain, a high limit of elasticity was the chief 

 desideratum, and it was possible in that material to vary the 

 limit of elasticity without interfering with the ultimate strength. 

 Two bars of steel, both with an ultimate strength of 35 tons to 

 the square inch, might have different elastic limits, but within those 

 elastic limits the ductility would be influenced. It might appear 

 that, owing to the great variety of conditions that were attainable, 

 steel was an uncertain material, but that was not the case ; 

 whereas, in the manufacture of iron a good deal depended upon 

 the personal skill of the worker, in steel it was rather a chemical 

 test which determined the qualities ; and it was quite possible to 

 ensure a uniformity in steel such as could not be approached in 

 iron, only it was necessary that the engineer should understand 

 clearly what was the quality of steel required for certain purposes, 

 and then it was for the manufacturer to produce that quality 

 uniformly and regularly. If, then, structures like links of a chain 

 should have a tensile strain of, say 38 tons per square inch, to 

 what extent might such a material be trusted ? In the case of 

 wrought iron, engineers thought it generally safe to trust a struc- 

 ture to one-fourth of the breaking strain, and the same rule might 

 still hold good for steel of that description, because the elongation 

 before rupture would be very much the same as in the case of iron. 

 But in considering the case of very mild steel the same rule should 

 no longer hold. If they constructed a bridge of cast iron they 

 would consider it unsafe to load it to more than one-sixth or even 



