The forms, instead of being removed gradually, were quickly removed, and the sudden application of the load caused a strain

which cracked the cantilever at the shank on the left-hand side. The props were restored and the cantilever washed with neat

cement. The props were then gradually removed and the cantilever remained unchanged until tested with pig iron. The application of this load caused the cracks to open at left side of shank and the cantilever collapsed. As bars of the requisite size were not available a bundle of smaller rods was substituted; the shear members were insufficient to hold the cantilever and it pulled

apart at the left shank and failed as shown in Fig. 16 before the compressive resistance of the concrete in the right center section of the shank had been reached. The loading was as follows:

20607

Failed 4-11-16 in. extension at the center.

Besides these, tests were made of cement shingles, concrete sewer pipes, cement bricks and hollow blocks.

The Collective Portland Cement Exhibit and Model Testing Laboratory was assembled with a view of exploiting the American Portland Cement Industry and not with a view of advertising any particular process, plant or product. It was highly beneficial in disseminating a better knowledge of the proper methods of testing and of the nature, uses and properties of Portland cement; and in recognition of this fact it received a grand prize, the highest award of the Exposition.

It is to be regretted that the time available for the experimental work was not longer, so that much more data could have been obtained.

It is, however, a matter of gratification to all those connected with this exhibit that the work thus started will be continued under the direction of the United States Government and other interested persons, and it is to be hoped that the exhaustive series of investigations of structural materials which have been planned under the direction of the advisory board may be successfully carried out thereby supplying information of inestimable value to the engineering profession.

PROPER METHODS IN CONDUCTING PAINTING

TESTS.

BY GUSTAVE W. THOMPSON.

A number of years ago, the President of our Society, Dr. Dudley, wrote a series of papers treating of the characteristics of paint and, we believe, for the first time in the history of the art, attempting some scientific understanding of paint problems. Not merely were these papers the first, in our opinion, that treated of the subject of paints scientifically, but, it may be said, since that time practically no progress has been made along this line. It is to be deeply regretted that Dr. Dudley has not had time to enlighten the public further in this direction; for, we are confident, had he been able to do so, the science of painting and paint manufacture would have been very much simplified. The situation, as it is presented to-day, is this: New paint materials are offered to the consumer, from time to time, these paint materials having more or less value, but in regard to which the manufacturers, as a rule, claim a somewhat more universal merit than these paints in actual use would warrant. It is natural for the paint manufacturer to advocate the use of his product or products for every condition under the sun, because he hopes by this broad advocacy to increase their sale. If there is one principle, however, on which the members of this society are pretty well agreed, it is that there is no complete paint which is suitable for all conditions of exposure; that is, every paint must, at least, have its various components put together in different proportions according to the conditions of exposure to which it is to be subjected.

We believe that it is the business of the engineer to design the paint which he is going to use under any given conditions. We believe that the paint manufacturers are only too anxious to supply what the engineer may demand. If the engineer starts out to select a paint, his proper attitude is not to take any one of the prepared paints on the market and specify that, because the manufacturer tells him it is the best for the purpose in question; but

it is his duty to make such experiments as are practical for the purpose of finding out what paint is best for the conditions to which the paint surface is to be exposed. To examine a paint on a large scale is, of course, the most satisfactory way, provided the paint turns out well. We are unquestionably of the opinion that the surest proof of a paint is its use, just as we are confident that the best way to detect a poison is to take it internally. The unfortunate part of the matter is, however, that the use of paint in order to find its value may leave the user firmly convinced that this value is a negative quantity. Supposing that he is able to try a paint in a limited way on full-sized structures, and the result, while not a minus result, is really very near that-that is to say, suppose the result leaves him somewhat in doubt-how is he going to tell whether the fault lay in the paint or its method of application? how can he be sure that the relative failure of the paint may not have been due to inherent destructive conditions external to the paint itself but peculiar to the structure painted? A painting test of this kind, properly called a "field test," is of inestimable value when accompanied by tests conducted under conditions all of which are known. These are what may be called "laboratory tests" and, if conducted with care and parallel to field tests, they act as a check upon the results obtained by these field tests and give a better basis on which to rest scientific conclusions. These laboratory painting tests, made with the intent to exposure under service conditions, should be supplemented by other laboratory tests which should parallel information enabling the engineer or paint chemist to determine the relative value of a given paint in advance of the results which are obtained by field or laboratory tests under service conditions. We must not reverse this order, however, and attempt to substitute purely chemical tests for exposure tests under service conditions until the laboratory tests have demonstrated their value.

By the term "painting tests" in our title, we have intended to include those tests which can be conducted with scientific accuracy, but which, at the same time, are as close as possible to ordinary practice so long as this accuracy is not impaired. Unfortunately, tests supposedly of this kind are in most cases without value: So much is this the case that laboratory painting tests are often spoken of in words of derision, as though they could be made