In fact each of these properties is dependent upon the other, because if a structure be amply proportioned in its main members for the assumed loads, but improperly swaybraced, the actual dynamic stresses will be greatly in excess of the live-load stresses provided for, and the metal will be overstrained in consequence; while, on the other hand, if rigidity be provided for by ample sway-bracing, but at the same time the main members of the structure be not adequately proportioned, the overstrained metal of the latter will cause vibration to be set up in spite of the sufficiency of swaybracing. Both of these faults are to be found in existing structures. The effect of the first fault is usually the gradual wearing out of the structure by impact and rack, and that of the second, the sudden collapse of the bridge without previous warning.

Principle VIII.

The strength of a structure is measured by the strength of its weakest part.

This statement is as old as the hills, but is just as valid today as it ever was. The ignoring of its prime importance is constantly the source of waste of metal in structures, fundamentally weak in certain portions, by increasing the weights of other portions, and thus adding to the total load that the weak parts have to carry.

Principle IX.

In bridge-designing provision must always be made for the effect of impact, either by increasing the calculated total stresses by a varying percentage of the live-load stresses, or by decreasing the intensities of working stresses below those allowed for statically applied loads.

Different specifications accomplish this result differently. The former method is undoubtedly the more scientific and rational one, but the latter is the more common. The reason for this is that engineers, as a rule, dislike to  specify  various  percentages  to  add  to live loads for impact, when such  percentages  are  established  entirely  by  guesswork.   An  elabo-