Joints Without Nails: The Carpentry Holding Up a Thousand-Year-Old Temple

A timber joint is usually admired for the wrong reason. The photographs that circulate show impossible-looking interlocks and invite the conclusion that Japanese carpentry is a kind of puzzle-making. The tests tell a different story: what matters is not how cleverly the pieces interlock but how the joint behaves when it is pushed back and forth several hundred times.

Alistair Renshaw looks at the buildings at Horyu-ji as structures rather than as heritage, and asks what specifically has kept them up.

The answer begins with the material. Wood is not one material but three, depending on direction. It shrinks roughly twice as much around the growth rings as across them, and almost not at all along the grain — and it is that difference, not the total amount, that splits a badly cut joint when the seasons change. A carpenter cannot stop the movement. The craft consists of deciding where it goes.

From there the book follows the laboratory evidence. Joints tested under repeated loading rather than pulled once to failure, because a building in an earthquake is a cyclic problem. The finding that fit governs stiffness, so an over-tight tenon is not a better joint but a different one, liable to split the mortise it was driven into. Frames tested with and without their infill panels, revealing that the parts everyone calls non-structural are carrying load. Instrumented pagodas whose measured first mode sits near 1.8 hertz.

One chapter is devoted entirely to the famous central pillar, and it is deliberately cautious. The popular account — that the shinbashira absorbs the earthquake — is neater than what the published analyses actually report, which is an interaction between the pillar and the stacked storeys with a good deal of energy dissipated at the joints themselves. Where the sources support only a mechanism and not a number, this book gives the mechanism.

There is also the uncomfortable question underneath any thousand-year-old wooden building: how much of it is a thousand years old. Members are replaced. Joints designed to be fitted rather than fastened can be unfitted. The structure persists as a design, maintained by people who are still being trained to cut these joints, which turns out to be the least romantic and most important part of the story.

Twenty-four drawings run alongside the argument, including a comparison of joint families with the load path through each drawn in.