How to Make Casein Glue

Collection: Field Notes - Regenerative Materials

Series: Natural Marine Adhesives & Sealants Hub

How to Make Casein Milk Glue: The VAKA Formula and Three Experiments with Tannic Acid

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. I'll cover the working formula 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 replicate them.

The overview post on casein glue covers what the material is, where it comes from, and how it compares to PVA and epoxy. This post is the procedure. If you want to know why it works, go there first.

Starting Point: Powder or Milk

There are two routes in. The first starts with dry casein powder — available from cheesemaking suppliers, homebrew retailers, and food ingredient wholesalers. Look for acid casein rather than rennet casein; 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. The dairy route costs almost nothing if you have access to fresh milk, and gives a satisfying practical connection to what this material actually is. Both routes arrive at the same activated glue.

Casein powder from a food ingredient supplier is the most convenient starting point and 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 VAKA Casein Formula

This is based on the 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). The modification I use 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.

• Casein — 100 parts
• Water (soaking) — 150 parts
• Potassium hydroxide — 11 parts dissolved in 40 parts water
• Calcium hydroxide (slaked lime) — 20 parts slaked in 40 parts water
• Copper sulfate — 3 parts dissolved in 20 parts water
• Tannic acid solution (3%) — for pre-washing surfaces before application (optional)

On the alkali. Potassium hydroxide (KOH) is the recommendation. Sodium hydroxide (NaOH, caustic soda) can substitute at approximately 8 parts rather than 11 — it is a stronger base per unit weight. KOH produces 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, which suits some woodworking applications. For the VAKA formula and marine structural work, use KOH if you can source it.

On the lime. The calcium hydroxide here is slaked lime — not quicklime, not garden lime (which may be calcium carbonate). Builders' lime and horticultural lime are both calcium hydroxide and work well. The FPL document notes that a high-calcium chemical lime gives best results; a lower-grade lime requires a proportionally larger amount. Start with 20 parts, adjust within the range 20–30 as needed.

The Dairy Route: Making Casein from Milk

If you are starting from fresh dairy rather than powder, skim milk gives the best results. Fat from full-fat milk interferes with the bonding chemistry. Raw milk from a farm gives higher casein content and is worth using if you have access to it. Milk powder reconstituted in water is a practical alternative that allows precise concentration control.

Warm the skimmed milk to around 50°C — warm to the touch, not approaching a simmer. Add white vinegar or dilute acetic acid gradually, stirring continuously. As the vinegar contacts the warm dairy, curdling begins immediately and visibly. Continue adding vinegar until no further separation occurs and the remaining fluid — the whey — runs clear. The whey protein has separated from the casein precipitate. Too much vinegar is not a problem; too little means incomplete precipitation and weaker final glue.

Strain through a cloth or fine mesh, pressing out as much whey as possible. Rinse the precipitate briefly with clean water to remove residual acidity, which would otherwise interfere with the alkaline activation. What you have at this point is wet casein — soft, white, slightly rubbery. Chemically identical to the powder form, just wetter.

Allow it to reach room temperature, then weigh it. Fresh-precipitated casein contains significant moisture, so reduce the soaking water in the main recipe by roughly a third when using this form. Press out as much moisture as you can through the cloth before weighing.

To dry it for storage: press the precipitate as dry as possible, spread thinly, and dry at low temperature — under 50°C — until brittle, then grind fine. Keeps for months in a sealed container.

Mixing the Formula: Sequence Matters

The order of addition matters significantly. The FPL document is precise on this, and experience confirms it. Getting it wrong produces a weak or lumpy result that does not develop proper working properties.

Step 1 — Soak. Weigh out 100 parts casein powder. Add 150 parts water, stir to combine, and leave to soak for 30 to 60 minutes. The casein absorbs the water and swells into a thick, soft paste. Prepare the other components during this time.

Step 2 — Prepare the three solutions. Dissolve the potassium hydroxide in its 40 parts water separately — it generates warmth on dissolving, which is normal. Slake the lime in its 40 parts water to produce a smooth slurry. Dissolve the copper sulfate in its 20 parts water. Allow all three to reach ambient temperature before use. Handle the KOH solution with care — it is strongly alkaline and will irritate skin and eyes on contact. Gloves are sensible.

Step 3 — Add the KOH. Stir the soaked casein and add the potassium hydroxide solution. Mix well. The paste will begin to change texture and may lighten slightly as the chemistry activates. Leave for two minutes.

Step 4 — Add the lime. Add the lime slurry and mix thoroughly. The compound will thicken noticeably. Leave for five minutes. This waiting period matters — the lime needs to complete its reaction with the activated casein before the copper is introduced.

Step 5 — Add the copper sulfate. Add the copper sulfate solution and stir well. The compound takes on a faint blue-green tint from the copper. Continue mixing until fully incorporated and the batch is a thick, smooth cream. This is the target consistency.

Adjusting. If the compound is too thin — running freely off the spatula — add a small additional amount of lime slurry and wait five minutes. If still too thin, add a little more pre-soaked casein. If 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.

Surface preparation. Before applying, wipe timber surfaces with a 3% tannic acid solution in water. This opens the grain slightly and improves adhesion, particularly on dense hardwoods. Allow the surface to dry before applying the glue.

Application. Apply to both faces of the joint. Assemble promptly and apply clamping pressure. The FPL recommends 150–200 psi; for workshop clamps, firm and consistent pressure across the joint. Leave clamped for a minimum of four hours at ambient temperature — longer in cold conditions. Full strength develops over 24 hours. Don't stress the joint during this period.

Pot life. Usable for two to four hours at ambient temperature before thickening makes spreading impractical. Prepare same day, use promptly, discard what remains. The dry components keep indefinitely; only the activated compound degrades quickly.

The Tannic Acid Experiments: One Failure, Two Still to Run

The surface pre-wash with tannic acid is in the formula above, and has been for a while. The rationale is straightforward: tannins cross-link with protein, and pre-treating the wood surface creates additional bonding sites for the caseinate. This is established chemistry. What I wanted to investigate was whether incorporating tannic acid directly into the glue mix — rather than only as a surface treatment — would improve elongation to break and reduce brittleness in the cured joint.

The hypothesis is this: tannic acid reacts with the caseinate protein chains to form additional cross-links, producing a glue line that is slightly less brittle and absorbs more energy before failing. A wooden boat hull flexes continuously in use. A glue line that yields a little before fracturing is preferable to one that does not. Whether this is actually achievable, and whether the benefit survives the competing reactions in an alkaline copper-containing formula, is what the experiments are intended to find out.

Experiment 1 — What failed.

I mixed the tannic acid solution with the copper sulfate solution before adding both together at step 5. The glue went off very quickly — 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 introduction to the main mix. They are not. Copper ions and tannins react immediately to form insoluble copper tannate complexes. Adding them together before introducing either to the glue simply pre-formed that reaction outside the protein matrix. The result looked superficially like a thickened batch but had no adhesive properties. Working time had collapsed to effectively nothing.

The lesson is clear enough: copper and tannic acid must never meet before both have been introduced separately into the casein matrix.

Experiment 2 — Still to run: tannic acid after lime, before copper.

The most chemically predictable sequence I can see is: KOH → lime (wait 5 min) → tannic acid solution (wait 2 min) → copper sulfate. This puts the tannin into an already fully activated caseinate in an alkaline medium, where tannin-protein cross-linking proceeds reasonably well, before introducing the copper. The copper then interacts with a protein matrix that has already begun forming tannin cross-links, rather than with free tannin. Whether this produces a meaningfully different glue line in the cured joint — less brittle, better elongation — is what the batch will show.

Experiment 3 — Still to run: tannic acid before any activation.

The more interesting question is whether adding the tannic acid to the soaked casein before any alkaline activation — essentially tannin-treating the raw protein before you unfold it — produces a different outcome entirely. The sequence would be: soak casein → add tannic acid (wait) → KOH → lime → copper. The tannin would be reacting with native casein protein in a near-neutral, slightly acidic environment, which is actually close to conditions in natural tanning. You'd be pre-cross-linking the protein chains before the alkaline activation unfolds them.

The uncertainty here is real. If the tannin locks the protein structure before the KOH can unfold it properly, you might get reduced activation rather than improved glue — the pre-cross-linking inhibiting the subsequent chemistry rather than adding to it. On the other hand, if the tannin-treated protein unfolds and activates normally, you might have something genuinely different in the final joint.

This is the experiment I'm most uncertain about and most interested in. I'll run Experiment 2 first because the chemistry is more predictable. Experiment 3 follows if 2 shows anything useful. Results will appear in a follow-up field note when I have them. The expected outcome of either is a slightly longer set time and different elongation characteristics, 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