The failure of struts under end load is due to two causes, crushing, due to direct compression, and excessive bending, due to elastic instability. In a general way we may say that the end portions of the strut tend to fail by crushing and the middle portion tends to fail by bending, but we cannot make a rule to decide where the middle portion ends and the end portions begin, because both causes are really operating simultaneously all along the strut, and we can only form a vague impression of their relative importance at various sections of the strut. A parallel strut, for instance, unless it is very short, will fail by bending at or near the middle, and the end portions will show no signs of crushing; so we may fairly infer that weight can be saved in a solid wooden strut without much loss of strength by tapering the ends; and that on the whole the efficiency of a strut, as measured by the ratio of crippling end load to weight or (resistance + m weight) where m* is a fraction, can be improved by tapering.