Timber splicing is the end-to-end joining of two timbers so that axial, bending, and shear loads transfer across the joint as if the member were continuous. Unlike a simple butt joint backed by metal hangers, a properly cut splice carries load through wood-to-wood bearing surfaces with mechanical fasteners or steel plates only resisting separation. The purpose is to extend a beam, post, or chord beyond available log length without losing structural continuity. Done right, a stop-splayed tabled scarf with hardwood keys can develop 70-90% of the parent timber's bending capacity.
Timber Splicing Interactive Calculator
Vary bolt size, hole clearance, scarf face gap, key moisture, and scarf slope to see drilling tolerance and splice-fit indicators.
Equation Used
The main drilling check uses the article rule that a timber splice bolt hole should be the bolt or drift-pin diameter plus no more than 1.5 mm. The calculator also checks the article fit limits for scarf face gap, hardwood key moisture, and stop-splayed scarf slope.
- Bolt or drift-pin hole tolerance is checked against the article limit of bolt diameter plus 1.5 mm.
- Scarf face seating is acceptable when the measured gap is 1.5 mm or less.
- Hardwood key moisture content is best near 12 percent.
- Stop-splayed scarf slope is treated as acceptable from 1:6 to 1:8.
Inside the Timber Splicing
A timber splice transfers load through three mechanisms working together — direct end-grain bearing, interlocking shear faces, and tension-resisting fasteners or fish plates. The geometry of the cut determines which of those carries the bulk of the load. A simple half-lap splice (also called a halved scarf) handles compression well but bends almost entirely on the bolts. A stop-splayed scarf with hardwood keys redirects the load path so the inclined bearing faces carry both compression and bending, while the keys lock the joint against axial pull-out.
The cut tolerances matter more than most people expect. On a tabled scarf the bearing faces must seat across the full contact area — gaps over 1.5 mm anywhere on the scarf face mean the load concentrates on the high spots and crushes the wood fibres there first. You will see the joint roll open under load. The hardwood key (usually white oak or hop hornbeam at around 12% MC) must be a tight drive fit; if it rattles in the mortise the splice loses its tension capacity and starts working as two separate cantilevers fighting through the bolts.
Common failure modes are predictable. Under-cut bearing faces crush at the toe. Over-tightened bolts split the cheek along the grain. Mismatched moisture content between the spliced timbers and the keys causes shrinkage gaps that open up months after assembly — this is why a post-to-post splice in a green oak frame must use kiln-dried hardwood keys, never green-on-green. The bolted timber splice with steel fish plates is more forgiving but transfers all bending through the bolt group, so the bolt-hole tolerance must be tight — typically the bore is the bolt diameter plus 1.5 mm, no more.
Key Components
- Bearing Faces (Scarf Cheeks): The angled or stepped wood-to-wood contact surfaces that transfer compression and a portion of bending. On a stop-splayed scarf the slope is typically 1:6 to 1:8. Surface flatness should hold within 1.5 mm across the full face — anything looser concentrates stress on high points and crushes fibres.
- Hardwood Key (or Folding Wedges): A hardwood block driven into a mortise across the scarf to lock the joint in tension and force the cheeks tight. White oak or hop hornbeam at 10-12% MC, usually 30-50 mm thick. Must be a drive fit, not a slip fit — a loose key kills the splice.
- Tabled Step (Shear Key): A square step machined into the centre of the scarf cheek that resists longitudinal shear. The table prevents the two halves from sliding past each other under bending. Depth is typically ¼ to ⅓ of the timber depth.
- Through-Bolts or Drift Pins: Steel fasteners — typically 16-25 mm diameter through-bolts or 19 mm drift pins — that resist joint separation and provide a secondary load path. Bolt holes should be drilled to bolt diameter +1.5 mm to keep the joint from rocking.
- Steel Fish Plates (for bolted splices): Mild steel side plates, typically 8-12 mm thick, sandwiching the timber on both faces of a butt-cut splice. The bolt group transfers both shear and moment through the plates. Plate length is usually 6-8× the timber depth to keep bolt-row stress within allowable bearing values.
Where the Timber Splicing Is Used
Timber splicing shows up anywhere a structural member needs to be longer than the available log length, or where an existing damaged section must be cut out and a new piece scarfed in. Heavy timber framing, traditional barn restoration, ship building, glulam tied-arches, marine pilings, and railway sleepers all rely on splice joints — the choice between a tabled scarf, a halved scarf, or a bolted fish-plate splice depends on whether the member primarily carries tension, compression, or bending.
- Heritage Barn Restoration: The Hancock Shaker Village 1826 round barn restoration in Pittsfield, Massachusetts used stop-splayed tabled scarfs with white oak keys to splice rotted sections out of the radial roof rafters.
- Timber Bridge Construction: The Powerscourt Covered Bridge in Quebec — North America's only surviving McCallum inflexible arch truss — uses bolted fish-plate splices at the lower-chord joints, with mild steel side plates and 19 mm drift pins.
- Wooden Ship Building: The Charles W. Morgan whaleship restoration at Mystic Seaport spliced new live-oak keelson sections to the original 1841 keel using hook-and-butt scarfs through-bolted with bronze.
- Glulam Mass Timber: Nordic Structures and StructureCraft routinely splice glulam beams over 30 m long using factory-cut finger joints, but on-site they use bolted steel side-plate splices for long-span tied-arches like the curling rink roof systems in Ontario.
- Utility Pole Repair: Bell Canada and Hydro One field crews splice rotted Douglas-fir distribution poles at ground line using C-channel steel splice sections — the Osmose C-Truss system — bolted through the sound timber above the rot.
- Railway Engineering: Canadian Pacific historically spliced bridge stringers with halved-and-tabled scarfs through-bolted with 25 mm carriage bolts, particularly on prairie timber trestles where ponderosa pine stringers exceeded 12 m.
The Formula Behind the Timber Splicing
For a stop-splayed tabled scarf splice in bending, the moment capacity depends on the bearing area of the inclined cheeks, the depth of the tabled step, and the tension capacity of the key or bolt group. At the low end of typical scarf slopes (1:4) the cheek face is short and the joint is dominated by the key in tension — the splice develops maybe 50% of parent-timber capacity. At the nominal 1:6 slope the cheeks are long enough to carry most of the bending in bearing, and the key only handles tension reversal — capacity climbs to roughly 70-80%. Push the slope to 1:10 and the cheek length is huge but the timber loses too much net section at the throat, and capacity drops back below 70%. The sweet spot is 1:6 to 1:8.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Msplice | Moment capacity of the spliced section | N·m | lb·ft |
| η | Splice efficiency factor — fraction of parent capacity developed by the joint geometry | dimensionless | dimensionless |
| fb | Allowable bending stress of the parent timber | MPa | psi |
| Snet | Net section modulus at the splice throat (parent S minus material removed for the scarf) | mm3 | in3 |
Worked Example: Timber Splicing in a heritage gristmill summer beam splice
A heritage millwright crew in Lanark County, Ontario is splicing a rotted section out of an 8 m white oak summer beam in an 1840s gristmill. The original beam is 250 mm × 350 mm. They are cutting in a 1.2 m sister section using a stop-splayed tabled scarf with white oak keys. Allowable bending stress for the salvaged white oak is fb = 12 MPa, and they need to know what percentage of parent-timber moment capacity the splice will develop at scarf slopes of 1:4, 1:6, and 1:10.
Given
- b = 250 mm
- d = 350 mm
- fb = 12 MPa
- Scarf length Ls at 1:6 slope = 2100 mm
- Table depth = 90 mm (≈ d/4)
Solution
Step 1 — compute the parent-timber section modulus and full-section moment capacity:
Step 2 — at the nominal 1:6 scarf slope, the cheek length is 6 × 350 = 2100 mm. This is long enough that the inclined bearing faces carry most of the bending in compression, and the white oak key only resists tension reversal. Empirical data from heritage timber-frame engineering (Sobon, Brungraber) puts efficiency η at roughly 0.75 for a well-cut tabled scarf at this slope.
Step 3 — at the low end, a 1:4 slope gives a cheek of only 1400 mm. The bearing area drops, the key takes more of the moment in tension, and η falls to about 0.55:
That is barely better than relying on a steel saddle hanger and a butt joint — you have done a lot of joinery for not much gain. Step 4 — at the high end, a 1:10 slope gives a 3500 mm cheek, but the scarf throat (where the table is cut) loses so much net section that Snet drops by roughly 30%, pulling η back to about 0.65:
So you spend an extra 1.4 m of timber on cheek length and end up worse than the 1:6 case. The geometry punishes you on both sides.
Result
The splice at the nominal 1:6 slope develops 45. 9 kN·m, or roughly 75% of the 61.2 kN·m parent-beam capacity. That is what a competent timber framer expects from a properly cut tabled scarf with tight white oak keys — the joint behaves almost like continuous timber under service loads, with a small amount of visible take-up the first time the floor is loaded. Compared to the 1:4 case at 33.7 kN·m and the 1:10 case at 39.8 kN·m, the 1:6 sweet spot is a real engineering optimum, not a tradition. If the as-built splice measures noticeably less stiffness than predicted — say the floor sags more than 8 mm under a known test load — the most likely causes are: (1) a loose key that was driven once and never re-driven after the green oak shrank, letting the cheeks separate under tension reversal, (2) cheek faces cut with a chainsaw and never trued with a slick, leaving 3-5 mm gaps that crush at the toe under first load, or (3) under-sized bolt holes that split the cheek along the grain when the bolts were torqued, dumping the secondary load path entirely.
When to Use a Timber Splicing and When Not To
The choice between a traditional cut scarf, a bolted fish-plate splice, and a modern factory finger joint comes down to what loads the member carries, how visible the joint must be, and how much shop time you can spend. Here is how the three stack up on the dimensions that actually matter on a job.
| Property | Stop-Splayed Tabled Scarf | Bolted Fish-Plate Splice | Factory Finger Joint (Glulam) |
|---|---|---|---|
| Moment capacity (% of parent) | 70-80% | 60-75% | 85-100% |
| Tension capacity (% of parent) | 40-60% | 70-85% | 90-100% |
| Shop/site labour per joint | 6-12 hours skilled joinery | 1-2 hours layout and drilling | Factory only — not field |
| Visible hardware | None or hidden pegs | Steel plates and bolts visible | Invisible glue line |
| Tolerance to moisture movement | Poor — needs re-driving of keys | Good — bolts can be re-torqued | Excellent — bonded joint |
| Typical material cost premium | Low (wood only) | Moderate (steel + bolts) | High (plant-controlled bonding) |
| Field repairability | High — recut and re-key | High — replace plates | None — must replace whole member |
Frequently Asked Questions About Timber Splicing
You are watching differential shrinkage. If the parent timber and the new sister section had different moisture contents at install — say 22% MC green oak butted to 14% MC kiln-dried oak — the drier piece holds its dimension while the green piece shrinks tangentially as it equilibrates. The cheek face rotates a fraction of a degree, and a 350 mm-deep scarf will open 1-2 mm at one edge.
Drive the key tighter. If the key bottoms in the mortise before the cheeks close, pull it, plane 2 mm off the bearing face of the key, and re-drive. Heritage timber framers expect to re-key a green-oak splice twice in the first year.
Bolted fish plates, every time. A scarf joint develops at most 60% of parent capacity in pure tension because the only thing fighting separation is the key and the through-bolts — the bearing faces do nothing. A fish-plate splice with a properly designed bolt group routinely develops 75-85% in tension because the bolts work in double shear through steel that is much stiffer than the wood key.
Save the tabled scarf for bending-dominated members like floor beams and rafters where the inclined cheeks actually earn their keep.
Rule of thumb: plate length ≥ 6 × timber depth, with the bolt group symmetric about the butt joint. A shorter plate concentrates moment into too few bolts, and the end bolts crush the wood in bearing — you will see the bolt holes elongate into ovals within the first heating season.
Two rows of bolts spaced 4-5 bolt diameters apart along the length and 4 diameters from the timber edge keeps bearing stresses inside the allowable values for most softwoods. If you cannot fit that geometry, the timber is too small for the moment you are trying to splice.
The creak is the bolts rocking in oversized holes. If you drilled the bolt holes at bolt diameter +3 mm or more (common when the crew used an auger meant for lag bolts), the joint takes load, slips a fraction of a millimetre until the bolt bears on the hole, and stops with a crack you can hear.
Pull a bolt and check the hole. The fix is either to drive a tight steel sleeve into the hole or to re-bore one size up and use a larger bolt. Aim for hole = bolt diameter +1.5 mm and no more.
You can do it but you must accept that the splice will work loose. Tangential shrinkage in white oak from 25% MC to 12% MC is roughly 6%, which on a 250 mm-wide cheek is 15 mm of dimensional change across the joint. The cheek faces will not stay in contact, the key will rattle, and the bolts will slacken.
If the schedule forces a green-to-dry splice, plan to come back at 6 months and 18 months to re-drive keys and re-torque bolts. Better practice is to dry the green timber to within 4% MC of the reclaimed piece before cutting any joinery — at minimum 6 months air-dried under cover for a 250 mm timber.
For pure compression, slope barely matters — a flat butt joint in solid end-bearing carries compression as well as any scarf, provided the end grain is square within 1° and the timber doesn't buckle laterally. The reason to cut a scarf at all on a post is to resist incidental tension, wind uplift, and lateral kick.
A 1:4 to 1:6 halved scarf with two 19 mm drift pins is plenty for a post-to-post splice in a wall plate. Don't waste shop hours on a tabled scarf in a column unless the column also carries bending — in which case it isn't really a column anymore.
The key was too thick for the mortise, or the table was cut with a square shoulder on a piece of timber with a checking grain pattern. Driving a tight wedge into oak with a radial check running through the cheek will follow that check and split the timber clean.
Two preventions: (1) inspect each timber for end-grain checks before laying out the scarf, and shift the joint by 100 mm if a check runs into the proposed key mortise; (2) bevel the leading edge of the key 5° on both faces so it eases into the mortise rather than wedging on first contact. Drive in stages, check the cheeks, and stop the moment you hear any cracking.
References & Further Reading
- Wikipedia contributors. Scarf joint. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
