To an outsider it must look as if the bridge engineers of this country were losing their heads!

The compression tests that started the present wave of apprehension did not have governing conditions corresponding properly to those of actual truss-members; hence I would suggest that, before the bridge engineers of America take the drastic step of assuming steel in compression to be only 60 or 70% as strong as the same material in tension, some really practical tests of struts be made under conditions corresponding to those in actual structures. Such a series of tests would cost considerable money; but, if the Bureau of Standards at Washington were to indorse the suggestion to make them, it ought not to be at all difficult to obtain from Congress an ample appropriation for the purpose.

The method of conducting these tests that I have in mind is this:

Let there be built a five-panel, riveted-truss bridge of about 100-ft. Span; let the middle panel-lengths of the top chords thereof be made decidedly weaker than all the other portions of the structure; and let a uniformly-distributed live-load be applied at the panel-points by hydraulic pistons. All portions of the structure, including the two weak members, should be scientifically detailed, so that failure will inevitably take place in the main portions of the weak struts and not in the details thereof, and so that the bridge can be used for a long series of tests to destruction of the said mid-panel lengths of the top chords. The weak members should be attached to the connecting plates with an ample number of rivets to develop the full strength of the test pieces; and these rivets should be removed carefully after each test to destruction is completed.

Of course it would be entirely practicable to vary the value of l/r in the different tests, provided that the attachment of the test-strut be not made eccentric.

There should be a supporting platform beneath the span to prevent its falling any material distance when failure occurs.

By adopting a weak vertical post instead of a weak panel-length of top chord, and by loading (at the elevation of the latter) three panel-points only, a series of tests could be made on vertical posts. A similar series could be carried out on the inclined end posts.

A cheaper method than that just described, but not quite as satisfactory, would be to build a single truss, instead of the complete span, and steady it laterally but not vertically, and then to apply the test loadings directly above the top chord at the panel-points. A great advantage of the span tests as compared with the truss tests is that duplicate tests could be made simultaneously on similar members. Moreover, the span tests would be in practically exact accord with actual conditions of loading, while the truss tests would not.

By removing occasionally the test loading, the elastic limits of the struts could be ascertained through noting the absence or otherwise of permanent set.

Experiments similar to those described could be made on a pin-connected span or a pin-connected truss so as to determine the strength of struts with hinged ends.

In view of the immense amount of bridge manufacture and construction that is likely to be done in the United States within the next five or ten years, the making of this proposed series of tests is worthy of being considered a matter of national importance.

In order to elaborate the preceding communication, the author had his associate engineer, F. H. Frankland, Member American Society Civil Engineers, write another letter to Engineering News-Record embodying certain subsequent thoughts upon the details of the proposed series of experiments, which letter was published in the issue of March 6, 1919. It reads as follows: