1 divisions, corresponded to 1000 inch, so that the change of length of the specimens was easily read to 100.000 inch. The length of the rod was about 4 feet, varying with different instruments. The angle turned through by the mirror in any test is so small that there is no appreciable error in using a straight scale for the readings. This was verified by turning the mirror in the Whitworth measuring machine through much greater angles than those through which it turns in the tests. It was also found that different strips (S) did not affect the calibration, so that a knifeedge could be used with different lengths of strip without recalibration. It was estimated that, under the conditions of test, the instrument reads accurately to 100.000 inch. 1 The kind of telescope used affects greatly the facility with which readings may be taken. The McGill Testing Laboratory telescopes are adjustable vertically and horizontally, besides moving bodily about a vertical axis. They are carried on speciallymade stands permitting any kind of adjustment to be made with ease (see Fig. 4). The extensometer must be carefully used in order to give correct results. The mirror should be, in its mean position, parallel to the scale and the telescope should be opposite to the mirror. The clip must be arranged so that the knife-edge is held quite firmly, otherwise it will not tilt correctly. The best clips are made from pieces of copper wire. If the direction of A B remains unchanged during test, the difference of the scale reading between two loads will be an accurate measure of the strain of AB for the given load difference, but if A B alters in direction this will not be the case. If, however, two readings are taken, one with the extensometer in the position shown, and the other with the knife-edge at A and the sharp edge of the strip at B, the mean of the two will be correct. When any doubt exists it is always better to do this so as to eliminate possible error. In the opinion of all who have worked with these instruments at McGill University they are the most simple, practicable, and accurate extensometers in use. It will be seen that they may be readily used in the most restricted positions, as, for instance, between the two angles of the double-angle members, where the width is only 3 inch. Fig. 4 shows the general arrangement of the apparatus. §2. The Specimens and the Method of Holding Them. Experiments were made on the three specimens, two singleangle members and one double-angle, shown in Fig. 5. All were ordinary shop products made by the Dominion Bridge Company of Montreal, and the angles were of uniform section 3 inches × 3 inches × 4 inch. Careful measurements showed that the section varied a little in the three specimens, and the actual areas, moments of inertia, etc., were computed and are given at the heads of Tables I, II, and III, pp. 153-155. The members were of a uniform length of 555% inches over all and 355% inches between the end plates, and were secured to the latter by means of four 3/4-inch rivets having a pitch of 221⁄2 inches and by lock angles riveted to the plate by three 3/4-inch rivets and to the outstanding leg of the angle by a like number. In the earlier experiments referred to above, similar angles were used, without lock angles, riveted to end plates of the same width, 3 inches, as the angles, secured directly in the grips of the testing machine, the distance from the end of the grips to the end of the angle being about 51⁄2 inches. With these end plates it was found that the restraining couple upon the angle was very small, and it was thought desirable, in the present tests, to substitute end connections more nearly similar to those met with in practice, i.e., wide plates firmly secured. To this end special grips were designed to fit on the jaws of the machine. These are shown in Fig. 6. They were steel castings. The end plates, which were I foot 2 inches wide and 3% inch thick, were held by steel pins 3 inches in diameter fitting tightly into bushes in the castings, and were restrained from turning by six set screws on each side, as indicated in Fig. 6. The distance from the end of the angle to the set screws was approximately 2 inches, so that the ends of the members were quite as effectively held as in most practical cases. The position of the pin with respect to the rivets was made different in the different specimens. In Specimens I and III it was in line with the back of the main angle, while in Specimen II it was in line with the centroid of the main angle. The object of this was to find the effect of a change in the line of pull on the gusset plate, i.e., of different eccentricity of end connection. Further changes were also made during the tests, as will be described later. The specimens were also tested with the lock angles removed in order to study the action of the latter. The machine used was the Emery testing machine in the Testing Laboratory of McGill University. This machine is of the vertical type and has a capacity of 150,000 pounds. It is eminently suited to this kind of work as the line of pull, suitable means being taken to steady the straining head, remains constant, and there is an entire absence of vibration. |