Mortise and Tenon on a Sofa Frame: What the Joint Holds, and What the Shoulder Decides

What is a mortise and tenon joint, and what are its four parts called?

A mortise and tenon joint is one connection made from two cuts: a cavity, called the mortise, cut into one member, and a projecting tongue, called the tenon, cut on the mating member. The tenon is glued into the mortise, so the two members lock along a wide wood-to-wood surface without depending on a metal fastener. On a sofa, the joints that matter sit where the seat rails meet the posts, where a sitter's weight becomes a rotating load pulling the rail away from the post.

This page is about the geometry of that joint. What the frame is made from is a separate question. So is what holds the tenon in the mortise.

A product page can print the phrase "mortise-and-tenon joinery" and stop there. The line does not say whether the tenon is short or long, or whether the shoulder actually touches the post. Two very different joints carry the identical name.

Four parts carry names. The mortise is the cavity, cut into the post. The tenon is the tongue, cut on the rail, that fits inside it. The shoulder is the flat face of the rail that sits flush against the outside of the post once the tenon is seated. The cheek is the flat side of the tenon, the surface that meets the mortise wall inside the cavity.

A tenon is "fully inserted" when the shoulder bears directly against the post, with no gap. The researchers built the opposite condition on purpose: in half the specimens they left a 12.7-millimeter space between the tenon shoulder and the face of the post. That gap models a joint whose glue line has already given ground.

What does the shoulder do, and why is it worth 1.58 times?

Erkan Likos, Eva Haviarova, Carl A. Eckelman, Yusuf Ziya Erdil and Ayhan Ozcifci tested 270 mortise and tenon joints for "Effect of Tenon Geometry, Grain Orientation, and Shoulder on Bending Moment Capacity and Moment Rotation Characteristics of Mortise and Tenon Joints," in Wood and Fiber Science, volume 44, number 4, 2012, pages 462 to 469, published by the Society of Wood Science and Technology. Their central finding is one number: "Joints with shoulders had substantially greater moment capacity (1.58 times) than joints without shoulders." The same paper reports that bending moment capacity of all joints in which tenons were fully inserted in mortises was 54 percent greater than for joints in which tenons were not fully inserted.

The result, the authors write, "clearly illustrates the contribution of shoulders to the bending moment capacities of the joints and also indicates the loss in capacity incurred if glue joints loosen in service." A bearing shoulder keeps carrying load after a glue line ages. That is the durability half of the finding.

The paper also measured how far a joint twists under load, in radians of rotation per newton meter of applied moment. The average for specimens with shoulders was 9.2 x 10⁻⁴ radians per newton meter. Without shoulders, 12.8 x 10⁻⁴. The joint with a bearing shoulder flexes less for the same push.

Those 270 specimens were Loblolly pine conditioned to 8 percent moisture content, with rails 38 by 38 millimeters and tenons coated in a polyvinyl acetate adhesive at 40 percent solids content. That is a furniture-scale laboratory test, not a sofa. What carries forward is the direction and the size of the effect.

How much does tenon length actually buy?

Length is the spec a buyer already knows to ask about.

With tight shoulders, the averaged bending moment capacities were 124.4, 132.7 and 178.8 newton meters for tenons 25.4, 38.1 and 50.8 millimeters long. With loose shoulders, the same three lengths averaged 90.0, 95.9 and 97.4 newton meters. With a bearing shoulder, the 50.8-millimeter tenon carried 34.7 percent more capacity than the 38.1-millimeter tenon and 43.7 percent more than the 25.4-millimeter tenon. Without a bearing shoulder, length barely moved the number. A longer tenon earns its keep only when the shoulder behind it bears.

The failure-mode reading explains why. The low value for the short 25.4-millimeter rectangular tenons "probably indicates adhesive failure in the joint," the authors write, while the value for the 50.8-millimeter tenons "probably indicates primarily tenon fracture." A well-proportioned joint fails by breaking wood; a poorly proportioned one gives up a glue bond first.

Shape follows the same logic. The authors note that rectangular tenons are probably best suited to structural members with substantial width-to-thickness ratios, seat rails among them, while round or square tenons may suit smaller square members such as stretchers. A round tenon is not automatically the cheap shortcut.

Grain orientation was measured too. With tight shoulders, tenons cut at 45 degrees carried 8.0 and 11 percent more than radial and tangential ones, differences the authors themselves call "presumably too small to justify sorting wood members for grain orientation." A small effect, reported as small.

What happens to a joint that works loose in service?

The loose-shoulder specimens stand in for something specific: a mortise and tenon whose glue bond behind the shoulder has failed or shrunk away from the post, through seasonal humidity swings or through an adhesive that never fully bonded the cheek to the mortise wall. The 12.7-millimeter gap is the laboratory version of a joint that has aged badly. Those specimens were loaded once to failure in a testing jig, and the study does not model an upholstered seat frame with corner blocks.

A joint's name on a spec sheet describes a shape at the moment of assembly. What that shape does across years of use depends on whether the shoulder was cut to bear and whether the glue holding it there was applied correctly. The geometry is the part a buyer can ask about before delivery.

How does a stapled or nailed joint behave over years instead of minutes?

Douglas R. Rammer opens chapter 8, "Fastenings," of the Wood Handbook, General Technical Report FPL-GTR-190 from the USDA Forest Service Forest Products Laboratory, with a plain framing: each type of fastening "requires joint designs adapted to the strength properties of wood along and across the grain and to dimensional changes that may occur with changes in moisture content." Wood moves. A fastening has to survive the movement, not only the load on the day it goes in.

A nail holds well on the day it is driven. But if common smooth-shank nails are driven into green wood that is allowed to season, or into seasoned wood that is subjected to cycles of wetting and drying before the nails are pulled, the chapter reports, they lose a major part of their initial withdrawal resistance, and that resistance may fall as low as 25 percent of the values measured on nails tested soon after driving. A fastener's grip on the day of assembly and its grip years later are two different numbers.

Most factors that affect nails similarly affect staples, per the same chapter. Withdrawal resistance varies almost directly with the circumference and depth of penetration. For a comparable load, the diameter of each leg of a two-legged staple must be about two-thirds the diameter of a nail. The chapter closes its staple section with a caution worth quoting: "corrosion, sustained performance under service conditions, and durability in various uses should be considered in evaluating the relative usefulness of a stapled connection."

None of this makes a staple the wrong fastener. Staples have correct uses in a frame, and attaching upholstery fabric to one is among them. A staple holding fabric against a frame does a different job from a staple or a nail carrying a structural seat-rail connection, the joint that takes a sitter's full rotating weight. A glued tenon and a driven fastener answer the same question, how the connection behaves years after assembly, with different physics.

Dowels, loose tenons, and corner blocks: what is each one for?

Three more terms show up in the same conversation, and each means something different.

A dowel joint uses round wooden pins, glued into matching holes drilled in the rail and the post, to align and hold the two pieces. It relies on the glued surface of the pins themselves.

A loose tenon, sometimes called a floating tenon, is a separate piece of wood glued into a mortise cut in both the rail and the post, rather than cut as one piece with the rail. In load terms it is a real mortise and tenon: it still presents a large glued wood-to-wood surface, and it can still be built with or without a bearing shoulder. The question to ask is the same. Does the shoulder bear.

A corner block is a brace fastened across the inside of a corner where a seat rail meets a post. It supplements a joint. It cannot make up for a tenon cut short or a shoulder that never touched the post.

None of this is a protocol for inspecting furniture you already own. Whether a frame needs repair or replacement is a separate question, answered here.

What Covelle publishes about its frame joinery

The Covelle frame is kiln-dried hardwood from FSC-certified suppliers, joined with mortise-and-tenon joinery. Chain of Custody is held at the supplier level, by Martco, license FSC-C022036, and by Boss Wood Products, license FSC-C190932. Both license numbers belong to the suppliers who process the wood, not to Covelle and not to the finished sofa.

The assembled sofa uses water-based adhesives and finishes throughout, with no solvents, no lacquers and no added formaldehyde. That is a composition statement scoped to the adhesive and finish layer.

Covelle holds no joint test of its own. The bending moment and rotation figures on this page belong to Likos, Haviarova, Eckelman, Erdil and Ozcifci and to the journal that published their work. Every material is documented, and every active certification is verifiable through its issuing body. See the full layer-by-layer disclosure.

Six questions to send a maker about the joint

Naming a joint is not the same as specifying one. Before you commit to a frame described as mortise-and-tenon, ask the maker to answer these in writing:

1. Which joint type connects the seat rails to the posts?
2. Is the tenon cut as one piece with the rail, or is it a loose tenon fitted into two mortises?
3. How long is the tenon?
4. Does the shoulder bear fully against the post, with no gap?
5. Are the corner blocks glued and screwed, or only stapled?
6. Does any structural joint in the frame rely on a staple or a nail as its primary connection?

A maker who knows the answer will answer in millimeters, not adjectives. A page that says "mortise and tenon" and stops has named the geometry, not the execution.