Generally the surface of the wood under the bark, and therefore also that of any layer in the interior, is not uniform and smooth, but is channelled and pitted by numerous depressions, which differ greatly in size and form. Usually, any one depression or elevation is restricted to one or few annual layers (_i.e._, seen only in one or few rings) and is then lost, being compensated (the surface at the particular spot evened up) by growth. In some woods, however, any depression or elevation once attained grows from year to year and reaches a maximum size, which is maintained for many years, sometimes throughout life. In maple, where this tendency to preserve any particular contour is very great, the depressions and elevations are usually small (commonly less than one-eighth inch) but very numerous. On tangent boards of such wood, the sections, pits, and prominences appear as circlets, and give rise to the beautiful "bird's eye" or "landscape" structure. Similiar structures in the burls of black ash, maple, etc., are frequently due to the presence of dormant buds, which cause the surface of all the layers through which they pass to be covered by small conical elevations, whose cross-sections on the sawed board appear as irregular circlets or islets, each with a dark speck, the section of the pith or "trace" of the dormant bud in the center. [Illustration: Fig. 17. Wavy Grain in Beech (_after Nordlinger_).] In the wood of many broad-leaved trees the wood fibres are much longer when full grown than when they are first formed in the cambium or growing zone. This causes the tips of each fibre to crowd in between the fibres above and below, and leads to an irregular interlacement of these fibres, which adds to the toughness, but reduces the cleavability of the wood. At the juncture of the limb and stem the fibres on the upper and lower sides of the limb behave differently. On the lower side they run from the stem into the limb, forming an uninterrupted strand or tissue and a perfect union. On the upper side the fibres bend aside, are not continuous into the limb, and hence the connection is not perfect (see Fig. 18). Owing to this arrangement of the fibres, the cleft made in splitting never runs into the knot if started on the side above the limb, but is apt to enter the knot if started below, a fact well understood in woodcraft. When limbs die, decay, and break off, the remaining stubs are surrounded, and may finally be covered by the growth of the trunk and thus give rise to the annoying "dead" or "loose" knots.