Moisture behaviour of wood is the most important property of wood to understand, to be able to understand how and why wood structures shall be built the way they shall be built.
Wood swells, because the walls of the cell pipes absorb water and swell.
 |
A good rule of thumb: When bone-dry wood gets soaking wet, it swells 8 % in the direction of the annular rings (tangentially to the trunk), 4 % radially to the trunk and not at all lengthwise. And shrinks with the same percentages when drying through the same moisture scale. |
These percentages depend on the species, and even the individual, but are good rules of thumb.
Swelling is different in different directions, because the walls of the cell pipes are not uniform, but have different thickness in different directions.
In practice wood is seldom bone-dry nor soaking wet, but something in between. Hence swelling and shrinking percentages are also something below the mentioned extremes.
 |
An old wooden kitchen chair seat, 40 cm / 16" wide, has left a 0.8 cm / 1/3" friction mark (red arrow) on the underside. The seat is glued to the frame only on one side (above in the picture). Other sides are fastened with clips that allow the seat to swell and shrink freely. The swelling/shrinking has been about 2 %. |
 |
The 60 cm / 24" tall panel of an old wooden panel door has swollen/shrunk by about 1 cm / 1/2", again about 2 %. The door has been painted blue in the end of the summer (high air humidity), when the panel has been at it's widest. The picture has been take in the end of the winter (low air humidity), when the panel has shrunk to its narrowest. Exposing the old red paint. |
 |
Measurement of slits in an old wooden floor gives a similar result: This is because of high air humidity in the summer, low in the winter. |
A boat is wetted not only by air humidity, but also by the "swimming element". No conclusions about the amount of swelling can be drawn from this, however. It's the change in moisture that counts, not the absolute amount. Wood that always stays soaking wet, stays dimensionally stable.
 |
A tangentially sawn plank can be expected to swell by a maximum of 8 %. |
 |
Whereas a radially sawn plank would swell only by 4 %. Traditionally boat planks were sawn like this. It is very unlikely to find radially sawn lumber in a lumberyard, however. Radial sawing typically produces less usable lumber out of a given trunk, so it's considered uneconomical and thus banned. |
 |
Typical planks are not tangential nor radial, but something in between. |
 |
In a plank like this there is more "annual ring direction" (more swelling) on one side than the other. So when the plank swells, one side swells more than the other, and the plank curves to a convex cross section. |
Because of this convexity, a plank that is to be fastened by the edges, should be fastened heart side facing in.
Similarly, a plank to be fastened only in the middle should be fastened heart side facing out.