Collection: Field Notes - Regenerative Materials

Series: Natural Marine Adhesives & Sealants Hub

How to Make Casein Glue — The VAKA Formula, Mixing Sequence, and Tannic Acid Experiments

This is a field note about making a glue that has been in continuous use for at least four thousand years, and about an ongoing attempt to improve it that started with a failure and has since opened into something more interesting. The working formula comes first. The experiments come after, one already run and failed, two still to run, and the chemistry is worth understanding before you decide whether to repeat them.

If you want to know why casein is the structural bond I build with, rather than how to mix it, the overview post covers what the material is, where it comes from, and how it stacks up against PVA and epoxy. This post is the procedure. There are two ways in, and they arrive at the same activated glue. The first starts with dry casein powder, from cheesemaking suppliers, homebrew retailers, and food ingredient wholesalers. Look for acid casein rather than rennet casein, because the acid form activates more readily with hydroxide alkalis. The second starts with fresh skimmed milk and a mild acid to precipitate the curd. The powder route is more consistent and easier to measure, and it’s what I use for construction batches. A kilogram makes many batches, costs a few pounds, and keeps indefinitely in a sealed container away from moisture. The dairy route costs almost nothing if you have access to fresh milk, and gives a satisfying connection to what the material actually is.

The VAKA casein formula

This is based on US Forest Products Laboratory Formula 11, developed by S. Butterman and C. K. Cooperrider and dedicated to the public (US Patent No. 1,456,842, 1923). My one modification adds copper sulfate. The FPL document specifies cupric chloride but notes that cupric sulphate can be substituted, which is what I do. All quantities are by weight.

On the alkali. Potassium hydroxide is the recommendation. Sodium hydroxide, caustic soda, can substitute at roughly 8 parts rather than 11, being a stronger base per unit weight. KOH gives a slightly more flexible glue line and dissolves more readily in cool water, which makes it the better choice for structural boat joints. NaOH is more widely available, sold as drain cleaner in most hardware shops, and produces a faster-setting, harder compound that suits some woodworking. For marine structural work, use KOH if you can get it.

On the lime. This is slaked lime, calcium hydroxide, not quicklime and not garden lime, which may be calcium carbonate. Builders’ lime and horticultural lime are both calcium hydroxide and work well. The FPL notes that a high-calcium chemical lime gives the best results and that a lower grade needs proportionally more. Start at 20 parts and adjust within 20 to 30 as needed.

The dairy route, making casein from milk

If you’re starting from fresh dairy rather than powder, skim milk gives the best results, because fat from full-fat milk interferes with the bonding chemistry. Raw milk from a farm has higher casein content and is worth using if you can get it. Reconstituted milk powder is a practical alternative that lets you control concentration precisely.

Warm the skimmed milk to around 50°C, warm to the touch, nowhere near a simmer. Add white vinegar or dilute acetic acid gradually, stirring continuously. Curdling begins immediately and visibly as the vinegar meets the warm dairy. Keep adding until no further separation happens and the remaining fluid, the whey, runs clear. Too much vinegar is not a problem. Too little means incomplete precipitation and weaker glue. Strain through a cloth or fine mesh, pressing out as much whey as you can, then rinse the precipitate briefly with clean water to remove residual acidity, which would otherwise fight the alkaline activation. What you have is wet casein, soft, white, slightly rubbery, chemically identical to the powder, only wetter.

Let it reach room temperature, then weigh it. Fresh-precipitated casein carries a lot of moisture, so cut the soaking water in the main recipe by roughly a third when using it, and press out as much as the cloth will give before weighing. To dry it for storage, press it as dry as possible, spread it thin, dry it below 50°C until brittle, then grind it fine. It keeps for months sealed.

Mixing the formula, in the order that matters

The order of addition matters a great deal. The FPL is precise on this and experience backs it up. Get it wrong and you produce a weak or lumpy result that never develops proper working properties.

Start by soaking. Weigh 100 parts casein powder, add 150 parts water, stir, and leave it 30 to 60 minutes. The casein absorbs the water and swells into a thick soft paste. Prepare the other components while it soaks. Dissolve the potassium hydroxide in its 40 parts water separately, where it will generate warmth, which is normal. Slake the lime in its 40 parts water to a smooth slurry. Dissolve the copper sulfate in its 20 parts water. Let all three reach ambient temperature before use, and handle the KOH solution with care, because it is strongly alkaline and will irritate skin and eyes on contact. Gloves are sensible.

Now build it up in sequence. Stir the soaked casein and add the potassium hydroxide solution, mix well, and leave it two minutes. The paste changes texture and may lighten as the chemistry activates. Add the lime slurry and mix thoroughly. The compound thickens noticeably. Leave it five minutes, because the lime needs to finish reacting with the activated casein before the copper goes in. Then add the copper sulfate solution and stir well. The compound takes on a faint blue-green tint and you mix until it’s a thick, smooth cream. That’s the target.

To adjust, if it runs freely off the spatula add a little more lime slurry and wait five minutes, and if it’s still too thin add a little more pre-soaked casein. If it’s too stiff to spread, add water by the teaspoon and mix. The right consistency spreads like soft butter and holds its shape on a vertical surface. Before applying, wipe the timber with a 3% tannic acid solution in water, which opens the grain slightly and improves adhesion, particularly on dense hardwoods, then let the surface dry. Apply to both faces of the joint, assemble promptly, and clamp. The FPL recommends 150 to 200 psi. For workshop clamps, firm and consistent pressure across the joint. Leave it clamped at least four hours at ambient temperature, longer in the cold, with full strength developing over 24 hours, and don’t stress the joint in that window. Pot life is two to four hours before thickening makes spreading impractical, so mix same day, use promptly, and discard what’s left. The dry components keep indefinitely. Only the activated compound degrades quickly.

The tannic acid experiments, one failure and two still to run

The surface pre-wash with tannic acid has been in the formula for a while. The rationale is plain. Tannins cross-link with protein, so pre-treating the wood creates extra bonding sites for the caseinate. Established chemistry. What I wanted to test was whether putting tannic acid directly into the glue mix, rather than only on the surface, would improve elongation to break and reduce brittleness in the cured joint. The hypothesis runs like this. Tannic acid reacts with the caseinate protein chains to form additional cross-links, giving a glue line that yields a little before it fractures and absorbs more energy before failing. A wooden hull flexes continuously. A glue line that gives slightly is preferable to one that doesn’t. Whether that’s achievable, and whether the benefit survives the competing reactions in an alkaline copper-bearing mix, is what the experiments are for.

The first one failed, and instructively. I mixed the tannic acid solution into the copper sulfate solution and added both together at the copper stage. The glue went off far faster than normal and the batch was unusable before I could apply it. The mistake was assuming the two additives were independent and could be combined before going into the main mix. They are not. Copper ions and tannins react immediately to form insoluble copper tannate complexes, so adding them together simply pre-formed that reaction outside the protein matrix. The result looked like a thickened batch but had no adhesive properties, and working time had collapsed to nothing. The lesson is clear enough. Copper and tannic acid must never meet before both have been introduced separately into the casein matrix.

The second experiment is still to run, and it puts the tannic acid in after the lime and before the copper. The most predictable sequence I can see is potassium hydroxide, then lime with a five-minute wait, then the tannic acid solution with a two-minute wait, then copper sulfate. That puts the tannin into an already fully activated caseinate in an alkaline medium, where tannin-protein cross-linking proceeds reasonably well, before the copper goes in. The copper then meets a protein matrix that has already begun forming tannin cross-links rather than free tannin. Whether that makes a meaningfully different glue line in the cured joint, less brittle, better elongation, is what the batch will show.

The third is the one I’m most uncertain about and most interested in. The question is whether adding tannic acid to the soaked casein before any alkaline activation, tannin-treating the raw protein before you unfold it, produces a different outcome entirely. The sequence would be soak the casein, add tannic acid and wait, then KOH, lime, copper. The tannin would react with native casein in a near-neutral, slightly acidic environment, close to the conditions in natural tanning, pre-cross-linking the protein chains before the alkali unfolds them. The uncertainty is real. If the tannin locks the structure before the KOH can unfold it, you might get reduced activation rather than a better glue, the pre-cross-linking inhibiting the chemistry rather than adding to it. If the tannin-treated protein unfolds and activates normally, you might have something genuinely different in the joint. I’ll run the second experiment first because the chemistry is more predictable, and the third follows if the second shows anything useful. The expected outcome of either is a slightly longer set time and different elongation, but I have no reliable data yet. That is what field notes are for.

At VAKA I design and build boats that don't destroy the environment. Find the plans as they are finalised at VAKA Plans and the full field notes here. If you are looking for a launching spot, the Hithe Finder is a community register of slipways, hards, and beaches for small boats.

VAKA. Traditional craft and natural materials. Nottingham. 2026.

References

Forest Products Laboratory, U.S. Department of Agriculture (1961). Casein Glues: Their Manufacture, Preparation, and Application. Report No. 280. Available via Oregon State University library:

Sutermeister, E. and Browne, F.L. (1939). Casein and Its Industrial Applications. 433 pp. Reinhold Publishing Co., New York. Cited in FPL Report No. 280.

Truax, T.R. (1929). The Gluing of Wood. U.S. Department of Agriculture Bulletin 1500, 78 pp. Cited in FPL Report No. 280.

Kelly, A. Ashmun (1921). The Expert Wood Finisher. Available via Woodworkers UK

Henley, W.T. (ed.) (1914). Henley's Twentieth Century Formulas, Recipes and Processes. Available via Project Gutenberg