under endlong compression. 
193 
about 50 per cent, greater since the moment of the “ rectangular 
forces ” is greater than that of the “ triangular ” in the ratio §. 
As an example of this effect the case of a square sectioned strut 
may be cited ; for any given elastic deformation the required 
bending moment will be the same whether bending occurs about 
an axis parallel to one side of the square or about a diagonal 
or about any intermediate axis. When the elastic limit is 
exceeded, this state of affairs no longer holds — thus if bending 
occurs about the axis AB in Fig. 2 the limiting bending 
moment, as previously explained, is about 50 per cent, greater 
than the maximum elastic one. In Fig. 2 a however, where 
bending occurs about an axis such as CD, the limiting bending 
moment is only about 40 per cent, greater than the maximum 
elastic one. 
In other w T ords, the strut is weaker when bent as in Fig. 2 a 
than as in Fig. 2, therefore when such a strut collapses it might 
be expected to do so by bending about a diagonal axis. The 
author tested this theory by taking a f-inch square-sectioned 
strut and loading it till instability set in and collapse occurred — 
it was found that the bending was about a diagonal axis, and 
seven repetitions of this experiment gave the same result in every 
case. 
2. This helps to explain the experimental results obtained by 
the author in the following experiments. The material used was 
mild steel and was in the form of circular rods of diameter 
= 0256 inch, the ends of the rods being coned to receive steel balls, 
and these balls being used to transmit the endlong forces and at 
the same time to allow bending to occur with as little friction at 
the ends as possible. Preliminary experiments showed that the 
elastic limit of this material was about 33 tons per square inch 
and that the breaking load in direct tension was about 41 tons per 
square inch. The rods, each 24 inches long, were placed upright 
in the apparatus shown in Fig. 3 and loaded with a force P at 
the top whilst being pulled out sideways by a weight W. The 
lower block LB and the screw above were both recessed conically to 
receive steel balls (0'312 inch in diameter) and between these balls 
was placed the rod to be tested. The block LB rested upon the 
cast-iron platform of a steelyard so that any downward force 
exerted upon LB was measured by reading the steelyard lever — 
this steelyard read up to about 1 ton. The downward force P was 
exerted by screwing down the wheel SW until the strut-beam 
being tested was subjected to the required degree of endlong com- 
pression, this compression being easily measured as stated above. 
In addition to this endlong loading a lateral one was introduced 
by suspending a weight W from the cord shown on the right of the 
diagram. Both P and W were next increased — one at a time — 
