the late wood is gradual--the two kinds of failure, namely,
buckling and bending, grade into each other. In woods with very
decided contrast between early and late wood the two forms are
usually distinct. Except in the case of complete failure the
cavity of the deformed cells remains open, and in hardwoods this
is true not only of the wood fibres but also of the tube-like
vessels. In many cases longitudinal splits occur which isolate
bundles of elements by greater or less intervals. The splitting
occurs by a tearing of the fibres or rays and not by the
separation of the rays from the adjacent elements.

[Illustration: FIG. 8.--Failures of short columns of green
spruce.]

[Illustration: FIG. 9.--Failures of short columns of dry
chestnut.]

Moisture in wood decreases the stiffness of the fibre walls and
enlarges the region of failure. The curve which the fibre walls
make in the region of failure is more gradual and also more
irregular than in dry wood, and the fibres are more likely to be
separated.

In examining the lines of rupture in compression parallel to the
grain it appears that there does not exist any specific type,
that is, one that is characteristic of all woods. Test blocks
taken from different parts of the same log may show very decided
differences in the manner of failure, while blocks that are much
alike in the size, number, and distribution of the elements of
unequal resistance may behave very similarly. The direction of
rupture is, according to Jaccard, not influenced by the
distribution of the medullary rays.[7] These are curved with the
bundles of fibres to which they are attached. In any case the
failure starts at the weakest points and follows the lines of
least resistance. The plane of failure, as visible on radial
surfaces, is horizontal, and on the tangential surface it is
diagonal.

[Footnote 7: This does not correspond exactly with the
conclusions of A. Thil, who says ("Constitution anatomique du
bois," pp. 140-141): "The sides of the medullary rays sometimes
produce planes of least resistance varying in size with the
height of the rays. The medullary rays assume a direction more
or less parallel to the lumen of the cells on which they border;
the latter curve to the right or left to make room for the ray
and then close again beyond it. If the force acts parallel to
the axis of growth, the tracheids are more likely to be
displaced if the marginal cells of the medullary rays are
provid