There is one very radical difference between the behavior of a concrete-steel structure and that of a structure composed entirely of steel, such as a truss bridge. A truss bridge may be overloaded with a load which momentarily passes the elastic limit, and yet the bridge will not necessarily fail, nor cause the truss to be so injured that it is useless and must be immediately replaced. The truss might sag a little, but no immediate failure is imminent. On this account, the factor of safety on truss bridges is usually computed on the basis of the ultimate strength.

A concrete-steel structure acts very differently. As has already been explained, the intimate union of the concrete and the steel at all points along the length of the bar (and not merely at the ends), is an absolute essential for stability. If the elastic limit of the steel has been exceeded owing to an overload, then the union between the concrete and the steel has unquestionably been destroyed, provided that union depends on mere adhesion. Even if that union is assisted by a mechanical bond, the distortion of the steel has broken that bond to some extent, although it will still require a very considerable force to pull the bar through the concrete. It is therefore necessary that the elastic limit of the steel should be considered the virtual ultimate so far as the strength of the steel is concerned. It is accordingly considered advisable, as already explained, to multiply all working loads by the desired factor of safety (usually taken as 4), and then to proportion the steel and concrete so that such an ultimate load will produce crushing in the upper fibre of the concrete, and at the same time will stress the steel to its elastic limit. On this basis, economy in the use of steel requires that the elastic limit should be made as high as possible.

The manufacture of steel of very high elastic limit requires the use of a comparatively large proportion of carbon, which may make the steel objectionably brittle. The steel for this purpose must therefore avoid the two extremes - on the one hand, of being brittle; and on the other, of being so soft that its elastic limit is very low.

Several years ago, bridge engineers thought that a great economy in bridge construction was possible by using very high carbon steel, which has not only a high elastic limit but also a correspondingly high ultimate tensile strength. But the construction of such bridges requires that the material shall be punched, forged, and otherwise handled in a way that will very severely test its strength and perhaps cause failure on account of its brittleness. The stresses in a concrete-steel structure are very different. The steel is never punched; the individual bars are never subjected to transverse bending after being placed in the concrete. The direct shearing stresses are insignificant. The main use, and almost the only use, of the steel, is to withstand a direct tension; and on this account a considerably harder steel may be used than is usually considered advisable for steel trusses.

If the structure is to be subject to excessive impact, a somewhat softer steel will be advisable; but even in such a case, it should be remembered that the mere weight of the structure will make the effect of the shock far less than it would be on a skeleton structure of plain steel. The steel ordinarily used in bridge work, generally has an elastic limit of from 30,000 to 35,000. If we use even 33,000 pounds as the value for s on the basis of ultimate loading, we shall find that the required percentage of steel is very high. On the other hand, if we use a grade of steel in which the carbon is somewhat higher, having an ultimate strength of about 90,000 to 100,000 pounds per square inch, and an elastic limit of 55,000 pounds per square inch, the required percentage of steel is much lower.