Screws in end grain

Why does a screw driven into the end of a board hold so much worse than the same screw in its face?

The same screw, the same board, the same driver, a difference of ninety degrees. Into the face it bites, tightens, and stops with a satisfying firmness. Into the sawn end it goes in suspiciously easily, then either spins uselessly at the last quarter turn or splits the board along its length.

So the difference is direction, and that is an odd thing for a material to care about. Steel does not care which way you drill it. What is it about wood that makes an orientation into a property?

Start with what a board actually is. A tree grows by adding long, thin, hollow cells running up the trunk — fibres and tracheids, a few millimetres long and a few hundredths of a millimetre across. They are stiff and strong along their length, because that is the direction the tree needs to resist. Sideways, they are stuck to their neighbours by a thin lignin-rich layer between the cell walls. That sideways bond is real, but it is the weakest thing in the assembly. Split a log and you are not cutting wood; you are unzipping that bond, which is why an axe works one way and a saw is needed the other.

Now ask what a screw's thread has to do to hold. Pulling a screw out does not slide it against friction. It has to destroy the wood sitting between successive turns of thread — a helical ribbon of material that must be sheared or crushed before the screw can move.

So look at where that ribbon lies in each orientation. Drive into the face, and the screw axis runs across the fibres; every turn of thread cuts across many of them, and the ribbon that must fail is fibres loaded across their length, severed. That is the strong direction — the failure surface has to cut material, not merely peel it.

Drive into the end, and the screw axis runs along the fibres. The thread now lies in the gaps between them, parallel, gripping each one along its side. To pull the screw out you need not cut a single fibre. You need only separate them from each other along that weak lignin bond — exactly the surface an axe exploits. The screw does not fail the wood; it slides the wood apart.

A second effect compounds the first. A tapered screw is a wedge, and in end grain that wedge drives directly along the natural cleavage plane, so it does what a wedge in a log end does: it splits. A split board has stopped gripping over much of the screw's length, weakening an already weak hold. Those two mechanisms together are why the difference is not a modest percentage but a change of category.

That shows up in design practice, not only workshop lore. Codes do not merely derate end-grain fastening; for nails in withdrawal they conventionally credit it with nothing at all, and for lag screws a reduction factor applies — I recall that factor as 0.75, but it is recalled and the treatment varies by code edition. I would not quote a single "end grain holds X per cent" figure: measured values swing hard with species, density, moisture and screw geometry.

Can we test it in a way that could fail? Drive two identical screws into one offcut, face and end, leave equal lengths proud, and lever both out against a block. Then look at what comes with them. The refuting observation: if the end-grain screw brings out a cylinder of sheared, crushed wood with fibres cut through, as the face-grain one does, the "separation along the weak bond" account is wrong. What you should see instead is fibres pulled apart lengthwise and intact, or a board that split before the screw yielded at all.

Here is the household lore worth taking apart, because the advice is right and the reason usually given for it is not. "Always drill a pilot hole in end grain" is good practice, and it is generally explained as easing the drive or preventing splitting — the second half true, the first beside the point. Removing the wedge stops the split. It does not touch the reason the joint was weak, which is the direction the failure surface runs. A pilot hole converts a joint that splits into a joint that merely pulls out.

So the real fix is never a better screw. It is to change the load path so some fastener is bearing across fibres rather than between them: a cross dowel threaded into a peg whose grain runs the other way, a glue block offering a face to screw into, or a pocket screw entering at an angle so it engages face grain. Flat-pack furniture uses cross dowels for this reason, not for ease of assembly.

Two honest qualifications. End-grain glue joints are also notoriously weak, which looks like the same story but is partly a different one — the open cell ends drink adhesive out of the joint line and starve it — so do not fold the two together. And engineered boards muddy the picture usefully: chipboard and MDF have no grain direction at all, yet edge-screwing them is still poor, because there the weakness is low density and short fibres. "Edges are weak" is a true rule of thumb reached by two unrelated routes.

• Why end-grain glue joints fail for a partly different reason, and how sizing the joint recovers some of it.
• How pocket-hole joinery angles a screw to convert an end-grain fixing into a face-grain one.

Every term the collection defines is gathered in the glossary.