Granite Coping Joint Detail: Width Is Rarely the Problem, What Sits in the Joint Is
A granite coping joint detail has to do three jobs at every break in the run: keep water out, leave room for the slab and the sealant to move, and carry the drip groove across the joint. At 600 to 1500mm slab lengths the summer-to-winter movement is only 0.19 to 0.48mm, all inside the 3mm site floor, so joint width is rarely the argument; what goes in the joint and whether the groove reaches the slab edge is.
What three jobs does a granite coping joint have to do at once?
Every coping joint has three jobs at once: keep water out, leave room for the slab and the sealant to move, and carry the drip groove across the break. The joint is the only interruption in the whole coping run, water travelling along the top face will reach it, and the only thing stopping it is the bead of sealant inside. The slab expands and contracts with temperature, and with no room in the joint the two slab ends push against each other, squeezing the sealant out or peeling it off the slab edge. The drip groove sits under the slab near the outer edge, and once the groove stops short of the joint, water halts at the break and works back along the underside to the inner face of the wall. The three hold each other up: the sealant needs space to deform freely or it debonds, so it needs width; the joint has to shed water, so it cannot be packed with mortar or left hollow; the groove has to be continuous, so it must be machined to the slab end. Lose one and the other two go with it.
So how wide does the joint actually have to be?
Take the largest of three lower limits: what the chosen sealant can absorb within its movement capacity, the 3mm site floor (an empirical figure: the nozzle has to fit in and the backer rod has to wedge), and the installation tolerance at the two slab ends, which comes straight out of the finished width if you leave it out. Start with the movement: ΔL = slab length × 8×10⁻⁶ × 40. The 8×10⁻⁶/°C is the linear expansion coefficient taken for granite (real values run 6 to 9) and the 40°C is the summer-to-winter face swing; both are empirical, not code values. A joint has half a slab on either side, so the movement one joint must absorb is about one slab's movement. Put the four standard lengths through it and you get the table below. Of the eight cells in the table, only two clear 3mm: 1200mm with a ±12.5% sealant at 3.07mm, and 1500mm with the same sealant at 3.84mm. Switch either slab to a ±25% sealant and the figure drops to 1.54mm and 1.92mm; the other six cells are under 3mm anyway. The suggested value takes the largest of the three limits, uses 3mm wherever the figure falls short, and rounds anything above up to the next millimetre. Narrower than 3mm is where it stops, and 3mm is already 6 to 16 times the slab's own movement.
| Slab length | Movement of one slab | Width floor (±25% movement sealant) | Width floor (±12.5% movement sealant) | Suggested value |
|---|---|---|---|---|
| 600mm | 0.19mm | 0.77mm | 1.54mm | 3mm |
| 900mm | 0.29mm | 1.15mm | 2.30mm | 3mm |
| 1200mm | 0.38mm | 1.54mm | 3.07mm | 4mm |
| 1500mm | 0.48mm | 1.92mm | 3.84mm | 4mm |
Why can't the joint simply be pointed up with cement mortar?
Mortar turns two slabs into one continuous element, and that removes the break the coping needs. Where the run is held at both ends by walls or returns, that expansion has no outlet: a 1m slab grows 0.32mm over a 40°C swing, a 10m run accumulates 3.2mm, and the movement every joint should have absorbed ends up in the mortar and crushes it. Once the joints can give no more, the run can only push at its two ends and lift the end slabs. Crushed mortar does not disappear; it turns to powder inside the joint, the joint becomes a trough that holds water, water sits in it and seeps through the mortar's pores to the underside of the slab, and when it dries there are white salt marks on the wall face. So a pointed joint is not filled solid and therefore stronger; it ties together two things that have to move separately. Bedding is the opposite case: lay the mortar full and compact it with no void underneath, because a void takes in water and keeps it against the slab underside. Strip or dab bedding does not let the slab follow temperature either: mortar grips wherever it touches, and the space you leave only gives water somewhere to sit.
Why put a backer rod in before the sealant?
Sealant should bond to two faces, not three. The two slab edges are the faces to bond; the bottom of the joint is not. Once the sealant grips the backer rod as well it is working on three sides, and when the slab moves it peels at whichever interface takes shear first, starting at the bottom of the joint where nobody can see it. The backer rod closes that bottom and sets the depth. Rod diameter runs 20 to 40% larger than the joint width (an empirical figure), so it wedges in and stays; sealant depth is 0.5 to 1 times the joint width, an empirical range and not a code value — too thin and it will not hold, too thick and curing shrinkage cracks it. The slab must be dry before tooling: sealant on a wet slab bonds to the water film and wipes off after it cures. Tool the surface concave, never flat; a flat joint leaves only a thin lip at the mouth, and that lip cracks first in the sun.
Why does the drip groove break down at the joint?
Because of where the factory stops the cutter. The slab end is the most fragile part of the piece, so the grooving head is lifted a few centimetres short of the end to avoid chipping the corner; those few centimetres sit exactly on both sides of the joint, so the groove dies before it gets there. Water runs along the groove, stops at the dead end and can only turn back, and the sealant in that stretch sits in water and is the first to debond. So write two lines into the order: the drip groove runs through to both slab ends, and the end chamfer must not close over the groove mouth. Mitred corners are where the arithmetic goes wrong: on a 45° mitre the joint face sits at 45° to both slab edges, so the gap runs diagonally in plan and 3mm measured on the face is only 3 × 0.707 ≈ 2.1mm measured square to the joint. To get 3mm square to the joint, open the face measurement to 4.2mm. Always measure square to the joint face, not on the slab face. The simpler route is to order an L-shaped corner piece: the corner is one piece with no diagonal joint, and the only joints left are the two straight butt joints to the running lengths, sized as in section 2.
What goes on the work order and the acceptance sheet?
Six items, one after another: the joint width and what it was derived from (which slab length, which sealant); the sealant's movement class; backer rod diameter and sealant depth; drip groove carried to the slab ends; corner as an L-shaped piece or a mitre, with mitre width measured square to the joint; 1 to 2mm chamfer on the slab ends. Three acceptance checks: turn one piece over and sight along the underside to see the groove reach the end; press the sealant with a finger, it should be elastic and not tacky; pour a bucket of water slowly over the coping and look for damp marks on the inner face of the wall top. Our stock Sesame White G603 is 30mm polished slab, 100 m² minimum order, 5 days for stock, 346 m² per 20ft container.