Where a restaurant chair actually fails
A dining room does not load a chair the way a home does. It drops 80 kg onto the seat in under a second, leans 30 kg of body weight into the backrest at a 15 degree rake, and drags the whole chair 2 m across a tiled floor with a full tray beside it. At 250 trading days and 8 turns a day, one chair sees roughly 2,000 sittings a year, and by year three it has taken more load cycles than a domestic chair sees in a decade.
Across the projects we have supplied over the last six years, about 70 percent of structural returns trace to three points: the joint between the back post and the seat rail, the leg-to-apron connection, and the corner block under the seat. Seat panels, foam and fabric make up most of what is left, and those are replaceable or refinishable. The frame joint is the part that decides whether a chair reaches year four still tight.
This is why the joint type matters more than the timber species or the finish. Four constructions are common in chair frames, and each one fails in a characteristic place, in a characteristic way, at a fairly predictable point in the service life.
Mortise and tenon: the strongest joint, and the easiest to get wrong
A blind mortise with a 10 to 12 mm tenon gives the largest glued surface of any timber joint, and when it is cut properly the shoulder of the rail sits hard against the post so the load travels timber to timber rather than through glue alone. A 12 mm tenon into a 35 mm post leaves 11.5 mm of wall on each side. That is the arithmetic behind a rule we do not bend: no back post under 32 mm stock takes a full tenon, so a slim 28 mm post on a drawing is a joint decision disguised as a style decision.
The failure mode is not the tenon snapping. It is the tenon working loose, because the mortise was cut 0.3 mm oversize, or because glue was spread on one face and starved on the other. A starved joint passes inspection in the factory and opens up after 6 to 9 months in service, which is usually after the container has been signed for and the defect is somebody else's problem. The tell is a fine dark line of compressed fibre around the shoulder, where the rail has been moving against the post.
Moisture content has to be 8 to 10 percent before machining. A tenon cut at 14 percent shrinks 0.4 to 0.6 mm across its width as it dries out in a heated room, and that is enough to turn an interference fit into a loose one. This is one of the checks we run on every batch of wood restaurant chairs before the frame goes to assembly, and it is the reason the same drawing can produce a tight chair from one supplier and a loose one from another.
Dowels and finger joints: cheaper, and where they give
Dowels of 8 mm diameter and 40 mm length are common in mid-range frames. Done properly, with a jig, two dowels per rail end and glue on both faces, a dowelled joint is serviceable and we would not argue against it in the right application. Done at speed, with a single dowel and a nail, it is not: the rail has almost no resistance to racking, and a guest leaning back on two legs puts the whole load through about 50 square millimetres of timber.
Finger joints belong in leg stock and glued-up components, not at the rail-to-post corner. A 10 to 15 mm finger profile gives good glue area along the length of a straight member, but the joint has no mechanical interlock in shear, so it has nothing to offer where the load reverses direction every time somebody sits down. A finger joint placed where a tenon belongs is a cost decision that presents itself at year two.
Dowelled and finger-jointed frames are legitimate products at the right price. The only question that matters is whether the quotation tells you which joint is in the chair. If the drawing shows a corner and no joint note, ask, because the answer moves the price by 8 to 15 percent and moves the service life by considerably more than that.
Corner blocks and the leg-to-apron connection
A corner block is a triangular or L-shaped timber screwed under the seat, tying two seat rails together. It stiffens the seat frame in the horizontal plane and it costs very little, but on its own it does nothing to stop the back post pulling away from the seat rail. Where a block is used together with dowelled rails the corner is adequately braced. Where it is used instead of a tenon it becomes the only thing holding the corner square, and two 4 mm screws into end grain will not do that for five years.
Leg-to-apron is the other weak point. A leg butted to the apron and screwed through from the inside will splay under side load; it needs either a tenon into the apron or a steel plate and bolt. Where the drawing shows a leg offset from the corner, ask how the offset is carried, because an offset leg puts the same load through a longer lever arm and the joint pays for it.
On metal restaurant chairs there is no glue line to inspect, so the equivalents are weld penetration and tube wall. A 25 mm square tube at 1.2 mm wall with a spot weld at the leg junction is not the same structure as the same tube at 1.5 mm with a fillet weld running the full section. The thin version passes a static load test and cracks at the heat-affected zone after 18 to 24 months of covers. None of this is visible in a photograph; it shows up in a section cut or a welding procedure specification.
How to check a joint without taking the chair apart
Put the chair on a flat floor, hold the seat down and push the top of the backrest sideways with roughly 15 kg of force. Movement over 5 mm at the top of the back means the back joint has opened, and it will keep opening. Then lift one front leg 10 mm off the floor and set it down: a click or a rattle is the leg joint moving inside its mortise or against its plate.
Repeat the test on three chairs from the same batch rather than one. Joint quality varies chair to chair far more than timber quality does, because it depends on the operator and the jig rather than on the material. If one chair in three moves, that is a process problem, not a sampling problem, and it will be repeated across the whole container.
If you want to see the joint itself, ask for a cut sample rather than a photograph: a 30 mm section through the back post and seat rail, sawn so the tenon and the glue line are exposed. Good glue coverage reads as a continuous film over more than 80 percent of the tenon face. A sample that arrives polished and assembled is a sample of the finish, not of the joint.
What to write on the drawing
A drawing that specifies overall dimensions, timber species and finish still leaves the joint to the factory, and the factory will choose whatever it runs fastest that week. Write the joint instead: back post to seat rail, mortise and tenon, tenon 12 mm, PVA D3 adhesive, no fasteners through the show face. On steel, write the tube section, the wall thickness and the weld type, and state that welds are dressed on visible faces only.
Then match the joint note to the model you are actually ordering. A JP-C004 wood dining chair with a mortise-and-tenon frame and a JP-C002 metal dining chair with a fully welded frame are two different structures that happen to sit in the same range, and a sample or a report from one tells you nothing about the other. Both are quoted against your drawing, so the joint note travels with the order rather than staying in the email thread.
Structural testing has to be run on the completed chair for the same reason. Our 150 kg whole-chair load test applies load to the seat and back assembly rather than to the seat alone, because a joint problem only appears when the back and the legs are loaded together. Reports are held per model, and restaurant furniture quality control at the workshop covers a moisture check on incoming timber and a weld inspection on steel frames before powder coating, so a joint that fails is caught before it is painted over.
Where the money goes
A mortise-and-tenon frame costs more than a dowelled frame, and it is worth saying plainly where that money lands: about 8 to 15 percent on the frame cost, which on a 60 USD chair is 5 to 9 USD. Spread across a five year life at 2,000 sittings a year, that works out at well under a cent per cover, which is a strange thing to trade away for a chair that goes loose in year two.
The calculation reverses for a short-lease site or a seasonal room where the furniture will be replaced inside three years. In that case a dowelled frame with a properly fitted corner block is a rational purchase, provided the drawing says so and the price reflects it, and provided nobody later compares it against a tenoned chair on price alone.
For a mixed room, ask for the joint note per model rather than per order: timber frames in the dining area and steel frames on a terrace will not have the same construction, and a single blanket note will be applied to whichever one the factory finds easier. On solid wood dining chairs the same drawing can be built three ways at three prices, and the only way to know which one you bought is to have asked before the deposit.