For a given surface area the loss of water from wood is always greater from the ends than from the sides, due to the fact that the vessels and other water-carriers are cut across, allowing ready entrance of drying air and outlet for the water vapor. Water does not flow out of boards and timbers of its own accord, but must be evaporated, though it may be forced out of very sappy specimens by heat. In drying a log or pole with the bark on, most of the water must be evaporated through the ends, but in the case of peeled timbers and sawn boards the loss is greatest from the surface because the area exposed is so much greater. The more rapid drying of the ends causes local shrinkage, and were the material sufficiently plastic the ends would become bluntly tapering. The rigidity of the wood substance prevents this and the fibres are split apart. Later, as the remainder of the stick dries many of the checks will come together, though some of the largest will remain and even increase in size as the drying proceeds. (See Fig. 27.) [Illustration: FIG. 27.--Excessive season checking. _Photo by U. S. Forest Service._] A wood cell shrinks very little lengthwise. A dry wood cell is, therefore, practically of the same length as it was in a green or saturated condition, but is smaller in cross section, has thinner walls, and a larger cavity. It is at once evident that this fact makes shrinkage more irregular, for wherever cells cross each other at a decided angle they will tend to pull apart upon drying. This occurs wherever pith rays and wood fibres meet. A considerable portion of every wood is made up of these rays, which for the most part have their cells lying in a radial direction instead of longitudinally. (See Frontispiece.) In pine, over 15,000 of these occur on a square inch of a tangential section, and even in oak the very large rays which are readily visible to the eye as flakes on quarter-sawed material represent scarcely one per cent of the number which the microscope reveals.