Edge-to-Edge Joinery

Edge-to-Edge Joinery:
Rob Porcaro explores the ubiquitous edge joint and looks at practical, small-shop techniques, rationales for various approaches and some controversies.

From tabletops to drawer bottoms, edge-to-edge joinery is found throughout furniture construction, making this a fundamental skill for every woodworker. Though we should be able to count on producing successful joints, some of these joints fail, mechanically and aesthetically, as we all have seen.
So, it is worthwhile to delve into the theory and practice of this joint. Executing it successfully is largely based on two factors: judicious wood selection and attention to key tolerances.

We can do better than this
Before exploring the good, let’s consider some bad and ugly.
One of the large drawer fronts in the factory-made red oak (Quercus rubra) desk at which I write has a small failure in an edge joint. A cross grain conflict with the particleboard sides is stressing the solid wood front, but if the glue line is supposed to be as strong as the wood itself, why is the split there?
A door panel in the same desk has intact edge joints, but the mishmash of flatsawn and riftsawn pieces, and random edge matching betray a lack of human attention. As craftspeople, we can pay a lot of attention to the wood, elevating our work above factory items. Good edge joints are an opportunity for that.
At the outset, I acknowledge that a little split here and there, though best avoided, is often not a big deal, and probably will not affect the function of the piece. My workbench, which I’ve used for more than 30 years, has several splits at the edge joints in its top that do not bother me at all. In fact, they seem to function as built-in stress relievers that probably help maintain the remarkably consistent flatness of the top throughout the seasons. 

Start with good boards
As a rule, the best appearance results from joining boards that are alike near their edges. Flatsawn boards with cathedral figure work best where there is rift figure beyond the width of the arches, ideally where the figure lines are nearly straight. Where this is not possible, a good appearance can still be had if the figure lines seem to flow into each other across the joint line. Boards with substantial figure ‘runout’ at the edge make it difficult to create a pleasing visual match at the joint. 

In Match 1 below, the edges are dissimilar in their annual ring patterns, and the board faces at the joint are a poor match. In pronounced cases, the awkward appearance may be accompanied by mechanical problems as the boards move differently with seasonal moisture content changes. This may create stress at the joint line, and though the elasticity of the glue will probably keep it intact, a slight variable step may develop on the surface of the panel.  

As for quartered and consistently patterned rift boards, these are easy to match, as long as the surface figure is fairly parallel to the edge of the board (see Match 2). Do not necessarily accept the original edge from the mill, but consider sawing a new one. For quartered boards, I ideally look for fairly vertical annual lines on the end grain throughout the width of the board, but avoid boards with one edge taken from close to the pith of the tree where the annual end grain lines are significantly arced.

Then make choices
Let’s look at what can be done with a simple panel composed of three red oak boards. It should offer no comfort that with only these three boards, there are 192 possible arrangements for the front face of the panel. Fortunately, matches that work well mechanically also tend to work well aesthetically, which makes it fairly easy to arrive at a good panel.
In the first arrangement (panel 1), which has all the pith sides up, there is no significant figure runout at the joints, and rift is joined to rift. The grain of the three boards runs in the same direction. This is an advantage if the board will be hand planed, though it does not matter structurally.
In panel 2, the pith and bark faces are alternated. Again, the grain direction is consistent among the three boards. The arrangements in the next two panels also look good, one with all pith sides up (panel 3) the other with alternating pith and bark faces (panel 4), but both with alternating grain directions between adjacent boards. Remember, if you alternate pith and bark faces and you want the grain in the same direction for all the boards, you must alternate the directions of the cathedral points.
You have plenty of options, but it pays to be aware of just what is going on with the figure and grain of the wood. Of course, longer boards are likely to have grain reversals and more variations in figure, but the concepts shown here will help in managing those.
A great aid in detecting the grain direction of any portion of a flatsawn board is the mnemonic from Dr Bruce Hoadley, ‘Pith side, plane with the points, bark side backwards.’

The great alternating question
Should the growth ring orientation alternate or be consistent when joining flatsawn boards to make a panel? There is no single answer here – experience tells us that both options can work – but let’s consider the issue in a way that will help us make decisions in the shop.
The reasoning usually given for orienting all the boards alike (with all the pith faces upward, for example) is that the seasonal changes will want to occur as one large arc, which is relatively easy to restrain, as in the top of a leg-and-apron table. However, each board actually wants to form an arc of its own. Three initially flat boards with their pith sides up, if they were not glued, would create three hills when they get drier. With the boards glued, to actually form a continuous curve, or, more practically, to be restrained flat, stress is induced particularly at the joint line.
On the other side of the argument, alternating pith and bark faces is said to confine the full depth of cupping changes within each board, and is supposed to be helpful in unrestrained panels. If the presumed slightly undulating surface forms, or if the panel is restrained from doing so, there is geometrically less stress at the glue lines than with the all-faces-alike board arrangement.
The reality is that elasticity at the glue line, and of the wood itself, bails us out of most problems, and we can usually just choose the configuration that looks best. Still, I avoid pushing my luck. I consider these issues, along with the width of the boards, the species’ tangential/radial movement ratio, the attachment of the panel in the piece and the environment where the piece will live, when deciding how to assemble a panel.

Considerations for book matched panels
Visually, the striking effect of the symmetry overrides most problems with figure runout. For planing the glued panel, there will necessarily be a grain direction conflict at the joint line, unless the grain is unusually parallel with the surface.
Similarly, for some ring-porous woods, such as figured walnut (Juglans spp.), directional light will strike the pores differently in each half of the panel. In a planed, unfinished panel, this can make the figure definition in the halves look disappointingly different. However, this resolves when merely a thin finish fills the pores.

The great camber question
In preparing an edge to be joined, I aim for the slightest camber (concavity) along the length – just enough to ensure there is no convexity whatsoever, and that there will be no loss of clamping pressure across the joint anywhere along its length. Thus, the reason for the camber lies in building the joint. It is basically a matter of a one-sided tolerance – ‘make a little bit of this to ensure that there is none of that’. This minimal camber is better felt than seen (see sidebar below). It should fully close by hand or with very gentle clamp pressure.
But wait, don’t we need to use a more substantial camber – perhaps a 1mm gap in a 1 metre-long joint, or 1/32in in a 3-foot joint – to build a degree of enduring elasticity into the joint? That is necessary, so the story goes, to prevent the joint from opening at the ends, where end grain exposure causes more rapid moisture exchange, which makes the ends of the boards shrink and expand differently from the interior of the boards.
True, most of the cracks that we see in panels occur at a glue line. Yet, if the joint was truly as strong as the wood itself, why did it crack there and not somewhere else instead? I think the answer is almost always that the joint was, in fact, not made well. Though I cannot cite a scientific analysis, I am sceptical that the effect of the supposed elasticity created by a large camber would truly focus at the joint line to specifically relieve stress there, and that it would actually be retained over the years. Also, making such a substantial camber on the assumption that insufficiently equilibrated wood will dry faster at the ends is simply starting off on the wrong foot and guessing at what might happen later.

Minimal camber
With any method of planing the edges, I want to finish with a continuous shaving to help ensure there are no localised bumps or troughs in the edge. Then, the best test is to clamp one board in the vice, and set the jointed edges against each other. Move the upper board by its end. It should pivot at the opposite end, proving there is an uninterrupted gap along the length.
If the upper board pivots somewhere other than at the end, that demonstrates that the edges are kissing, which is no good.
Check for twist within the joint by gently trying to rock the top board across diagonally opposite corners of the joint. Significant twist will produce a subtle but surprisingly detectable rocking.
Check for flatness in a few places by very gently holding a straightedge against the surface, again taking note of any rocking. This is a much more sensitive predictor of a flat glue-up than only checking the edge of the board with a square.