inaugurate a series of further experiments that will determine it.

Meanwhile the author has adopted temporarily the formula given in, Chapter XIV., viz,

in which I is the percentage for impact to be added to the live load, and L is the length in feet of span or portion, of span that is covered by the said load.

This formula was established to suit the average practice of half a dozen of the leading bridge engineers of the United States, as given in their standard specifications, and not because the author considers that it will give truly correct percentages for impact.

In spite of all that has been said to the contrary in the past or that may be said in the future, the impact method of proportioning bridges is the only rational and scientifically practical method of designing, even if the amounts of impact assumed be not absolutely correct; for the method carries the effect of impact into every detail and group of rivets, instead of merely affecting the sections of the main members, as do the other methods in common use.

The assumption made in some specifications that the live load is always twice as important and destructive as the dead load, irrespective of whether the member considered be a panel suspender or a bottom chord-bar in a five-hundred-foot span, is absurd, and involves far greater errors than those that would be caused by any incorrectness in the assumed impact formula.

The author acknowledges that he anticipates finding the values given by the formula somewhat high; but it must be remembered that the said formula is intended to cover in a general way, also, the effects of small variations from correctness in shop-work, or to provide for what the noted bridge engineer, the late C. Shaler Smith, used to term the factor of ignorance.