should not exceed one-third of the ultimate strength. This restriction applies mainly to heat-treated eye-bars.

Some engineers deem it good practice on the score of economy to use high-carbon steel for reinforcing-bars; but the author is opposed to this for two reasons; first, such metal is liable to crack when being bent cold in the field; and, second, a high intensity of working stress on the bars tends to permit the formation of cracks in the concrete in their vicinity.

Very infrequent loads will permit overstresses, or, in other words, encroachment on the factor of safety. A future increase of live loads will affect differently different portions of a bridge. This is sometimes allowed for by reducing the unit stresses on certain members; but it is better to design the structure so that an increase of loading of fifty per cent will not overstress any member more than fifty per cent.

Some designers specify abnormally heavy loadings and correspondingly greater unit stresses. This gives well balanced structures; but the method is illogical, and it has a tendency to deceive. If such a practice were to become established, there would be a liability on the part of inexperienced engineers to employ high unit stresses with normal loads, which would be dangerous or at least ultimately uneconomic.

In very-long-span bridges the selection of live loads and unit stresses is of extreme importance; because a small legitimate increase in unit stresses or a small reduction of loading may result in a comparatively large saving in cost, due to decrease of dead load.

If, for the sake of either a real or an imaginary economy, an increase in unit stresses must be adopted, it is better to make it frankly in the main members, where the danger from overstress is least; and then the fact of the existence of this condition will be apparent. But if skimping were done in the detailing, there might easily be developed weaknesses of such a serious nature as eventually to cause disaster. Again, more money can be saved by reducing the sectional areas of main members than by trimming the details.

In the designing of ordinary truss bridges, the computer has practically no choice as to how the various stresses which he figures are to be combined, because the standard specifications by which he is governed indicate the method very clearly; but in structures unusual as to either type or magnitude and in trestles the designer should have some option in making the combinations. No hard-and-fast rule can well be given to cover all cases of spans of unusual size and character; except to state that the engineer should employ his best judgment in relation to possibilities and probabilities, stressing the standard amount for combinations that are likely to occur and increasing the intensities of working stresses as the combinations considered become more and more improbable of realization, up to the limit of an excess of fifty (50) per cent.

In bridges proper, with the exception of arches having less than three hinges, the only unusual combination is that of the ordinary stresses and the