Ask us anything.

Why the lactone difference is so large. Beta-methyl-gamma-octalactone exists in wood as non-volatile precursors, mainly galloyl esters, released by hydrolysis during seasoning and toasting. Quercus alba holds far more of the cis isomer, which is the aromatically potent one, with a perception threshold in wine an order of magnitude below the trans form. That single compound accounts for most of what people mean when they say a wine smells American.

Why the ellagitannin difference goes the other way. Quercus petraea is the richest of the three in vescalagin, castalagin, grandinin and the roburins, with robur close behind and alba well below. This is a genuine species difference, not a cooperage effect, although grain and seasoning modulate it.

Grain and seasoning modulate both. Tight grain, from slow-grown trees, gives lower extractable phenolics per unit area and a slower release. Long open-air seasoning, two to three years, hydrolyses part of the ellagitannin fraction and softens the green, bitter notes that raw wood carries. A tight grain, long seasoned French oak and a wide grain, short seasoned one are further apart than French and American at the same specification.

And the practical consequence for a substitution. If you are replacing American barrels with a tannin, match the aroma with the American profiles and then decide separately how much ellagitannin the wine needs, because the barrel was giving you less of it than a French barrel would have. If you are replacing French barrels, the reverse: the aroma target is easier and the structural contribution is the one you must not lose.

Why a single toast level never reads as barrel. A barrel is toasted by radiant heat on its inner face, and the heat penetrates as a gradient. Your wine reaches only the first 2 to 3 mm (0.08 to 0.12 in), but within those few millimeters there is a continuum, strongly toasted at the surface and progressively lighter with depth. What a taster registers as barrel is a blend of profiles arriving together, not one profile. Reproduce it with a single toast and something is always missing, even when the intensity is right.

How the blends are built. Not by mixing finished products at the end, but by composing at the wood stage, with several toast curves and, where the target calls for it, both Quercus petraea and Quercus alba. Because the composition happens upstream, the same recipe can be delivered as tannin or as stave without redesigning it.

Where they came from. These are not marketing assemblies. They were developed with consulting oenologists over years, against repeated targets we kept being asked for. That is why they are named by objective rather than by intensity.

And where a single profile is still the better tool. If your wine has one specific gap, a mid-palate that collapses, a finish that lacks tension, a single Primary Colors profile targets it more precisely and more cheaply. La Recette is for building a complete oak signature, not for patching one axis.

Objective 1: protection. Add early, on the must, and let the tannin take the first oxygen.

A tannin added at crush consumes dissolved oxygen before your wine's own phenolics do, and gallotannins and ellagitannins bind and precipitate the oxidative enzymes, tyrosinase in healthy fruit and above all laccase when Botrytis is present. That enzyme effect is the strongest reason to add at crush rather than later.

Use VBX at crush, 10 to 30 g/hL (0.8 to 2.5 lb per 1,000 gal) on reds and up to 50 g/hL (4.2 lb per 1,000 gal) on botrytised fruit. Use Anti-Ox whenever the wine needs protection later in its life, on any color, and where sensory neutrality matters. Harvest

On whites and rosés, the single most important instruction is timing: put the addition on the juice, at the press or in the must, before oxidation has started. Added afterwards it can only mop up what is left, and the product will look ineffective when in fact it arrived late.

Objective 2: structure. Two moments, two tools, and a narrow window on whites.

Use EBX 810, untoasted French oak, at the pre-fermentation stage, where it structures and protects at the same time. Then #10, also untoasted, during maturation, which is the same contribution a barrel makes through its untoasted heads. Bring in Pure Grape Seed case by case, when the mid-palate is short and needs condensed tannin.

White juice and wine are protein-rich, and condensed tannins bind protein: hardness, a drying finish, haze risk. That does not put whites off limits, it means the window is narrow and sits below the color threshold. In practice 0.1 to 0.2 g/hL (1 to 2 ppm) is where whites respond well: enough to give the palate something to hold on to, comfortably under the 0.5 g/hL (5 ppm) where the color starts to move. Above that you are trading texture for color, and on most white styles that is a bad trade.

Objective 3: flavor complexity. Lower doses, and always a bench trial.

Use the Primary Colors range and the La Recette blends: Amplitude, Fruitissimo and Légèreté on reds, Minéralité and Rondeur on whites, plus the Vinification Fraîcheur versions for whites and rosés. Doses sit an order of magnitude below the protective ones, and lower again on whites, because the margin between interesting and marked is narrow. Trial before every production addition, without exception. Decision making kits

Setting the dose on whites and rosés, which is where trials go wrong.

On a white, the addition becomes visible in the color from 0.5 g/hL (5 ppm). That is far lower than most people expect, and it is the number to have in mind before any trial. Whether it is a problem depends on your target style. On a buttery, oak-driven Chardonnay a golden tint reinforces exactly what the drinker expects, so the addition works with the style. On a pale, fresh, unoaked white, the same 0.5 g/hL is a fault. So decide the style first, then the dose.

On a rosé, the effect is on the shade itself, not the intensity, and rosé palettes differ enormously between markets and house styles. No rule of thumb survives that. Run the bench trial with a visual assessment against your target color, every time, before any production addition.

On reds, this ceiling does not exist.

Two chemical families do two different jobs, and confusing them is the most common reason an addition disappoints.

Condensed tannins (proanthocyanidins, from grape seed and quebracho) are polymers of flavan-3-ols. They build astringency, grip and mid-palate volume, and they react directly with anthocyanins. They work on texture.

Hydrolysable tannins (ellagitannins from oak and chestnut, gallotannins from gall nut) are esters of a sugar core with gallic or ellagic acid. They are far more oxygen-reactive. They work on redox balance, color stabilization and protection rather than on texture.

Flavor is a third axis, independent of both. Within the oak family, an untoasted tannin can be strongly structuring and sensorially discreet, while a toasted one builds and enhances flavor at a much lower phenolic contribution.

Why the extraction matters as much as the raw material. We extract at low temperature, in reverse osmosis water, with no solvent, and we steer that extraction on the resonance behavior of the molecules.

So the first question in front of a wine is not which tannin, but whether you are correcting texture, managing oxygen, or building flavor. Three different problems, three different tools.

Why single profiles rather than ready-made blends. A blend is somebody else's decision about your wine. Working from single profiles lets you add exactly the axis that is missing, and it makes trials interpretable: if you dose three profiles separately you learn what each does on your wine, and that knowledge stays with your cellar. A blend that works tells you only that it worked.

Why the number order is not arbitrary. Moving up the range moves you along the pyrolysis sequence. Low numbers sit on the untoasted and lightly toasted end, where vanillin and fresh oak dominate. High numbers sit where furfural, guaiacol and syringol dominate, which read as dry, roasted and vertical rather than sweet. That is why heavy toast is the right answer on cooked fruit and the wrong one on a wine that is already flat and sweet.

Why #10 is a different animal. Untoasted oak retains its ellagitannin load, because toasting degrades it. So #10 is the profile that carries the redox and structure work, and it is the analogue of a barrel's untoasted heads. If you want both structure and flavor, you need a low number and a high number, not a compromise in the middle.

Why the same profile across four formats is not obvious. It only works if the wood is the same wood, matured the same way and toasted on the same curve, and then extracted without distorting the profile. Because we do the sourcing, the maturation, the toasting and the extraction ourselves, the #14 in a bottle and the #14 on a stave are the same #14. That continuity is the whole point of being vertically integrated. How we work

Why we can answer this at all: we are both scientists and producers. We select the oak, we make the staves, we toast them in our own kilns, and we extract our own tannins from the same wood. Almost nobody sits on both sides of that line. Tannin suppliers generally buy an extract and formulate it. Coopers make wood and stop there. Because we do both, we can compare a tannin and a stave on the same oak, the same grain and the same toasting curve, and see what each actually delivers. That is where the tool comes from. It is not a marketing model, it is what we measured on our own material, and it is one of the few places in this category where the work was done rather than improvised.

The four contributions of a barrel. An ellagitannin contribution from the wood. A flavor contribution from the toasting. A slow, continuous oxygen supply, roughly 20 to 30 ppm over the first year for a new 225 L (59 gal) barrel, about 2 ppm per month. And time, twelve to eighteen months of it.

Why our tannins arrive finished. A fast hot extraction pulls out a young, superficial fraction, which is why so many tannins taste raw, drying and green on addition. Our extraction matches a twelve month infusion of 18 mm (0.7 in) staves: your wine receives the phenolic and flavor outcome of a year of wood contact, immediately. What remains missing is the oxygen, and with it the condensation reactions oxygen drives over time. That is the honest boundary.

A barrel is not homogeneous. The body staves are toasted, the heads are not. A wine in barrel is in contact with two woods at once: the untoasted heads contribute structure and ellagitannins, the toasted body builds flavor. Reproducing a barrel with a single toasted product misses half of it, which is why an untoasted component belongs in the protocol, #10 on the tannin side. One note on our range: EBX 810 is a pre-fermentation tool, not an aging one, so it does not belong here despite being untoasted French oak.

A barrel is toasted by radiant heat, and this is what most people miss. The fire works on the inner face and the heat penetrates as a gradient. Your wine reaches only the first 2 or 3 mm (0.08 to 0.12 in), but within those few millimeters there is no single toast level. There is a continuum, strongly toasted at the surface, progressively lighter with depth. What you taste is a blend of profiles, not one profile. A single toast level product therefore never quite reads as barrel.

This is exactly what La Recette was built for. It is not one toast, it is a composed blend reproducing that gradient, and it exists in both formats, as tannins and as 18 mm (0.7 in) staves. The profiles are Amplitude, Fruitissimo and Légèreté for reds, Minéralité and Rondeur for whites, plus Vinification Fraîcheur versions for whites and rosés. We also build La Recette sur-mesure in 18 mm staves when you have a specific target to hit.

Why thickness is not a detail. It is often assumed that a barrel oxygenates through its staves and joints. In fact a large share of the early oxygen comes from air held inside the wood itself, in the porosity of the vessels and lumens, released once the wood is wetted. Oak is a reservoir before it is a membrane. A 7 mm (0.28 in) stave holds very little air and releases what it has almost at once, which makes it a finishing tool. An 18 mm (0.7 in) stave carries a real internal oxygen charge and gives it up progressively, alongside an extractive diffusion that runs over months. Thickness governs both curves at the same time, phenolic release and oxygen release, and that pair is what defines barrel aging.

On the tank side, our other decision making tool takes aging duration, alcohol, temperature, pH, free and total SO2 and volume, and returns wood dosage, format, tannin additions, SO2 strategy and micro-oxygenation parameters. Where your tanks are fitted for cliquage we can size the regime precisely rather than approximately.

Where the oxygen in a stave comes from. It is usually assumed that oak oxygenates by diffusion through the wood. A large share of the early contribution is actually air held inside the wood itself, in the porosity of the vessels and lumens, released once the wood is wetted. Oak is a reservoir before it is a membrane, and reservoir capacity scales with volume, which is why thickness dominates.

Why the two curves move together. Thickness governs both phenolic release and oxygen release at the same time, and it is that pairing which defines barrel aging. A thin format decouples them: you get the wood without the oxygen, which is exactly the chip problem in a milder form.

What that means for the trial. A stave program cannot be trialled honestly at glass scale, because a stave does not fit in a bottle and the oxygen contribution is missing from the sample. A tannin bench trial still gives you a reliable direction and the profile decision, and the equivalence table converts it. But the final call on thickness is a decision about your vessel and your timeline, not about taste.

And the practical constraint. Staves need to be held in the tank and removed. Stainless leaches and infusion bags exist for that, and it is worth planning the handling before the order rather than after delivery.

A barrel does three things at once: it releases ellagitannins and volatile oak compounds, it lets oxygen in at a controlled rate, and it does both through a heated surface layer only 2 to 3 mm deep (0.08 to 0.12 in, roughly a tenth of an inch), which is why a barrel gives several aromatic profiles at once rather than a single note. Any alternative has to be judged on all three, not just the first.

Oak thickness and oxygen. The oxygen a piece of oak delivers comes partly from ingress and partly from the air held in the wood itself, in the porosity of the vessels and lumens, released once the wood is wetted. Oak is a reservoir before it is a membrane. An 18 mm (0.7 in) stave has enough mass to hold a meaningful reserve and to release it over months, on the order of 20 to 30 ppm of oxygen per year in a classic 225 L (59 gal) barrel, close to 2 ppm per month. A chip, at 2 to 4 mm (0.08 to 0.16 in), is saturated and spent almost immediately. A 7 mm (0.28 in) stave sits in between and behaves as a finishing tool. This is not a quality difference between the two woods, it is geometry. That is why the same oak, in two formats, gives two different wines: one where the tannin polymerizes with anthocyanins into stable polymeric pigment, and one where the wood aroma stays free and evanescent.

Why our tannins reach the stave result. Ellagitannins (vescalagin, castalagin, grandinin, the roburins) are the fraction that drives structure, oxygen buffering and the condensation reactions that build polymeric pigment. Our low temperature extraction with reverse osmosis water, steered on the resonance behavior of the molecules to select what we take, pulls that fraction and the volatile fraction in the proportions a barrel gives naturally, and nothing beyond it. Aggressive extraction technologies pull everything present, including compounds formed during toasting that have no business migrating into wine. Selecting is harder than extracting, and it is the part we have actually worked on.

Digestion, in chemical terms. Removing the wood stops the transfer, it does not stop the reaction. The ellagitannins already in solution continue to react with anthocyanins and with your wine's own phenolics for one to two weeks, and the volatile compounds continue to redistribute between the colloidal and free phases. Perceived oak keeps climbing after the wood is gone. With a slow release format the curve is flat enough that it hardly matters. With chips, where the transfer curve is steep, it matters a great deal. As an order of magnitude, a 2 g/hL step (0.17 lb per 1,000 gal, 20 ppm) is easy to add and impossible to take back.

Where the oxygen goes. Oxygen consumed by wine is not lost, it drives the Fenton chemistry that turns ethanol into acetaldehyde, and acetaldehyde is the bridge that links tannin to anthocyanin. No oxygen, no bridge, no fusion of the oak into the wine, and no color stabilization either. Micro-oxygenation alongside chips is not an accessory, it is the missing third of the barrel.

A practical note on barrels. A barrel is a beautiful maturation tool, capable of real elegance and precision, but only if you monitor it closely and give it the energy it demands. It loses roughly 60% of its oenological capital per year, both its ellagic capital and its volatile capital, so what you bought in year one is not what you are working with in year two, and a used barrel also carries a microbiological history and a sulfur demand of its own. If you want to stay with barrels or inserts, our decision making tool will estimate what your cooperage still has left and what to add on top of it. Barrel maintenance

If you want to see the three routes side by side on your own wine, the decision making kits let you run the trial at bench scale first.

Why circulation matters more than people expect. Extraction is diffusion-limited at the wood surface. A static boundary layer of saturated wine forms around each piece and slows transfer sharply. Movement replaces that layer, which is why the same weight of chips gives a different result in a tank that is worked and one that is not. It is also why a tank result can fail to match a bench trial that was inverted daily.

Why the bag matters beyond convenience. Free chips settle, compact, and the pieces at the centre of the mass extract far less than those at the edge. You end up dosing a fraction of what you paid for while believing you dosed all of it.

On the digestion window. Removing the wood stops the transfer, not the reaction. Ellagitannins already in solution keep reacting with anthocyanins and the wine's own phenolics for one to two weeks, and the volatile compounds keep redistributing between the free and colloidal phases. Perceived oak keeps climbing after the wood is out. On a slow format the curve is flat enough not to matter; on chips it matters a great deal.

And the toasting question when reusing. Do not reuse chips or staves between wines. The extractable fraction is largely spent after one use, and what remains is a wet, nutrient-rich surface that is a microbiological liability.

Why 22 mm rather than 18. Inside a barrel the wine is already getting oxygen through the staves and the head joints. The insert's job is to restore the ellagic and volatile capital rather than to supply oxygen, and a thicker section gives a slower, longer release across a full élevage without over-delivering early.

Why the arithmetic surprises people. A cellar describing a wine as thirty percent new oak, the rest in third and fourth fill, is receiving close to thirty percent of one barrel's worth of chemistry, not a broad oak contribution spread across the whole volume. Once you compute it that way, the gap an insert fills is usually larger than expected.

What an insert cannot restore. Oxygen transfer falls as a barrel ages, because the pores clog with tartrates and polymers. An insert does not reopen them. If your third fill wines taste tired, that is often two problems at once, no ellagitannin and no oxygen, and the second may need cliquage or micro-oxygenation alongside.

And the microbiological caveat. Wood is porous, and Brettanomyces colonizes the first millimeters in biofilms that cleaning does not reach. Restoring the oak of a barrel with a Brett history restores the flavor and not the safety. Barrel maintenance

Why the tannin trial is comparable and the others are not. What breaks a bench trial is usually the mismatch in oxygen, temperature and time. A liquid tannin addition carries none of its own oxygen, so a sealed sample stored beside the tank reproduces the tank. A chip or a stave brings oxygen with the wood, and a stave cannot be sampled at glass scale at all. That is a geometry problem, not a product problem, and no supplier can get around it.

Integration is the variable people forget. Oak intensity falls by roughly a factor of four over the first two weeks. What you taste on the day of the addition is not the answer, and judging at day one will always lead to over-dosing. The tool models that curve so that the reading at day 14 is the one that counts.

Reference conditions matter. The model runs on pH 3.5 at 68 °F (20 °C). A wine at a different pH or held at a different temperature moves along the curve at a different rate, which is why the parameters are adjustable rather than fixed.

The protocol. Draw the control and the trial samples from the same tank at the same moment. Work from a stock solution at 10 g/L and dose with a micropipette rather than weighing milligrams: at 1 g/hL on a 100 mL (3.4 fl oz) sample you would be weighing 1 mg. Use 250 mL (8.5 fl oz) minimum, fill and seal with minimal headspace, store the whole set together, control included, and read at 14 days, blind and in random order.

And the scale-up rule. Once the two-week reading is in, the equivalence table gives the production dose directly, in the format your cellar actually uses. Where you are working without the tool, go to the lower end of what the bench told you: the tank will keep working for months, you can always add more, and you can never take it out.

Why total polyphenol content is not the right axis. The index most datasheets quote measures reducing power against a reagent, not reactivity in wine. A heat-degraded extract can post a high number while having lost the catechol and galloyl configurations that do the work. Two products with the same index can behave completely differently in a tank, which is why we would rather be judged on effect than on a number.

Where the differences actually show. Three places. The raw and drying character on addition, which reveals a young, superficially extracted fraction. The rate of integration, which reveals the degree of polymerization. And the residual metal load, which you will not taste but which shows up months later as accelerated oxidation, since iron and copper catalyse the whole cascade.

A test worth running once. Take both products, dose to equal sensory intensity, and hold both samples with a deliberate small oxygen exposure over several weeks. The one extracted in mineral-rich water will age visibly faster. It is the clearest demonstration of why the water used in extraction is not a detail.

And one caution about sensory panels. Palate saturation across a dose series is real and it is fast. Randomize the order, limit the number of samples in a sitting, and re-taste the first sample last as a check.

Why the three clocks differ. Volatile compounds simply dissolve and equilibrate, which is fast. Integration is a redistribution between the free phase and the colloidal phase of the wine, tannins associating with polysaccharides, mannoproteins and other phenolics, which takes days to weeks. Condensation between tannin and anthocyanin, and the polymerization that softens texture, are slow reactions that depend on oxygen availability and temperature, and they run over months.

Temperature moves all three. Reaction rates roughly double for every 18 °F (10 °C). A trial held at 68 °F (20 °C) and a cellar at 54 °F (12 °C) are not on the same clock, which is why a bench trial should be stored beside the tank rather than on a bench.

Why intensity falls rather than rises. On the first day a large share of the addition is still free in solution and fully available to your palate. As it associates with the wine's colloids, the perceived intensity drops even though nothing has left the wine. The reference conditions we model are pH 3.5 at 68 °F (20 °C), and both parameters are adjustable in the tool because both move the curve.

Why the raw material answers most of the question. A tannin is an extract of a plant tissue, and what ends up in the drum is that tissue's soluble phenolic fraction plus water. There is no carrier, no fining agent and no processing aid to declare. That is a structural advantage of water extraction over solvent or carrier-based processes, and it is why the answers to vegan, allergen and residue questions are usually short.

Where the real allergen question sits in a cellar. In wine, the allergen declarations that matter are usually about fining agents, egg albumin, casein, isinglass, rather than about tannins. A tannin addition can in some cases reduce the fining requirement, which changes your declaration in the right direction. That has to be verified on a fresh heat test rather than assumed.

And on organic specifically, the distinction most schemes draw is between the substance and the process: the same tannin can be authorised or not depending on how it was extracted. Ours is water extracted at low temperature with no solvent, which is the configuration schemes are most comfortable with, but the confirmation still has to come from your certifier.

Why the lumps form. A polyphenol powder wets on its outer surface first, and the hydrated layer becomes a gel that keeps water from reaching the core. The lump then survives the pump-over intact, and the tannin inside it never reaches the wine. Sprinkling onto a large water surface avoids this because each particle wets independently.

Why water and not wine. Ethanol lowers the solubility of the more polymerized fractions, and the low pH of wine slows hydration. Water at moderate temperature dissolves faster and more completely. Avoid boiling water, since high temperature degrades ellagitannins into simpler phenolic acids and costs you exactly the reactivity you paid for.

Why homogeneity is not a laboratory nicety. Poor dispersion is one of the most common reasons a tank result fails to match a bench trial, and it looks exactly like a product failure. If a trial worked and the tank did not, suspect the addition before you suspect the product.

What the specification actually covers. Beyond the botanical origin, the Codex sets limits on heavy metals, on residual solvents, on moisture, on ash, and on the identity of the material. Those limits are the reason the extraction method is not just a quality question but a compliance one: extraction in mineral-rich water raises the metal load, and solvent extraction raises a residue question that water extraction does not have.

Why our process sits comfortably inside it. We extract in reverse osmosis water at low temperature, with no solvent. There is no solvent residue to declare, and the water contributes no iron, copper or calcium.

And why the purpose statement is not bureaucratic. The declared purposes are what tie the product to a measurable function. That is also why we prefer to be judged on function rather than on total polyphenol content: the index measures reducing power against a reagent, not what the product does in your tank.

Why oxygen is the first enemy. The catechol and galloyl configurations that give a tannin its oxygen-handling capacity are exactly the ones that oxidize. A partly full container has a headspace, and that headspace slowly spends the product's capacity before it ever reaches your wine. This is not visible: the powder still looks the same and still weighs the same. What has fallen is reactivity per gram, which is precisely the property you bought.

Why heat matters next. The same degradation that a hot extraction causes happens slowly at storage temperature: hydrolysis of ellagitannins into simpler phenolic acids, and loss of the higher polymerized fractions. Every 18 °F (10 °C) roughly doubles the rate, so a store at 86 °F (30 °C) ages a product several times faster than one at 59 °F (15 °C).

Why moisture matters on powders specifically. Polyphenol powders are hygroscopic. Absorbed water lets hydrolysis proceed and causes the caking that then makes the product hard to dissolve, which is usually the first symptom anyone notices.

And a practical check. If an opened pack has been sitting for a season, do not simply dose it at the usual rate. Run a small bench trial against the dose you used last time. It costs one afternoon and it tells you whether you are dosing the product or dosing its history.

Why we hold stock in California rather than shipping from France. Harvest does not wait, and the decisions this shop supports are mostly made within a few days of a specific operation: a press, a crush, a racking. A product that arrives after the window has passed is not a product. Holding inventory close to the customer is the only way that works, and it is also lighter in transport terms per unit delivered than shipping small orders individually across an ocean.

Why packaging sizes are what they are. They follow the dose logic rather than the other way round. A 5 kg (11 lb) bag of protection tannin covers a meaningful crush volume at 10 to 30 g/hL (0.8 to 2.5 lb per 1,000 gal). A 200 gm (7 oz) pack of finishing tannin covers a meaningful volume at 1 to 10 g/hL (0.08 to 0.84 lb per 1,000 gal). Buying a size that means an opened pack sits for two vintages costs you more than the saving, because reactivity falls in an opened container.

And on packaging itself. We keep it minimal and recyclable deliberately. It is a small thing next to the resource question, but it is one of the parts we control completely.

Why the decline is not linear in the way people assume. The extractable ellagitannin sits in the first few millimeters the wine reaches, and it is depleted by extraction rather than worn away. So the first wine takes a large share, the second considerably less, and the curve flattens near zero. The volatile capital follows a similar shape but from a different starting point, which is why an old barrel can still smell faintly of oak while giving almost no ellagitannin at all. Those two curves are worth tracking separately, and that separation is one of the things the tool tries to make visible.

Why oxygen falls too. Pores clog progressively with tartrates and polymers, so transfer drops with age as well. A wine in third fill barrels that tastes tired is often starved on both axes at once, no ellagitannin and less oxygen, over a long élevage.

Why the microbiological question sits underneath. Wood is porous, and Brettanomyces establishes biofilms in the first millimeters that high pressure washing and hot water do not reach. And as the ellagic capital falls, the wine also loses part of the redox buffer that helped keep it clean, so the two curves work against each other. Covering the empty barrel between wines helps on the side you can control.

And a caution about our own figures. The 60% per year is a good working estimate across the parks we have measured, but a barrel's history, its grain, its toast, the wines it held and how it was stored all move it. Treat it as a starting hypothesis for your cellar rather than a constant, and correct it with what you observe.

The molecules, since that is what the question asks. The ellagic capital is a family of C-glycosidic hydrolysable tannins: vescalagin and castalagin as the two principal monomers, epimers at C1, alongside grandinin and roburin E, and the dimers roburins A to D. On hydrolysis they release ellagic acid, which is where the family name comes from. These are the molecules that consume oxygen, mediate its transfer and drive condensation. Vescalagin in particular reacts with flavan-3-ols to form flavano-ellagitannins, the acutissimins, which is the same reactivity that makes it a partner for anthocyanins.

Our extract carries the same molecules from the same species, because it is made from the same wood. There is no synthesis and no substitution.

The volatile capital is a different chemistry entirely. It comes from thermal degradation during toasting: furfural, 5-methylfurfural and hydroxymethylfurfural from the sugars; vanillin, syringaldehyde, coniferaldehyde and sinapaldehyde from lignin; guaiacol, 4-methylguaiacol, eugenol and syringol as volatile phenols; and the whisky lactones, cis and trans beta-methyl-gamma-octalactone. Toast level moves the balance: light toast is vanillin-dominated, high toast shifts to furfural, guaiacol and syringol, which is why heavy toast reads dry and roasted rather than sweet.

And this is where toasting trades one capital for the other. The same heat that generates those volatiles degrades the ellagitannins. More volatile capital means less ellagic, so any program that wants both needs two separate inputs, an untoasted one and a toasted one. In our range that means #10 or EBX 810 on one side, the Primary Colors range on the other. Finishing tannins

The pyrolysis by-products, and why a barrel does not transfer them. Thermal degradation also produces polycyclic aromatic hydrocarbons. In a barrel they stay in the wood, for a straightforward chemical reason: they are strongly lipophilic and effectively insoluble in a hydro-alcoholic medium at 13% ABV, so solubility and contact time act as a natural filter. That filter disappears the moment an extraction uses temperature, pressure or solvent to maximize yield. Our extraction is designed to stay inside what a barrel transfers, which means accepting a lower yield by design.

Species, in chemical terms. Quercus petraea and robur are rich in ellagitannins and poor in whisky lactone. Quercus alba is the reverse, with several times the lactone content and a lower ellagitannin load. Moving from American barrels to a French tannin changes both axes at once, and the difference you notice is usually the lactone, not the tannin.

Kinetics, which is the real difference. Same molecules, different delivery. A barrel releases them over months alongside oxygen. A tannin delivers them at once, and our extraction is built so that the profile matches a twelve month infusion of 18 mm (0.7 in) staves rather than the young superficial fraction a fast hot extraction gives.

And the oxygen falls as the barrel ages too. Pores clog with tartrates and polymers, so transfer drops. A wine in third fill barrels is often called tired when it is simply starved: no ellagitannin, no oxygen, and a long élevage regardless. Three ways to restore it, and the tool arbitrates: micro-oxygenation or cliquage where you have the equipment, 18 mm inserts which carry oxygen inside the wood itself, or renewing part of the park.

And the microbiological side. Wood is porous, and Brettanomyces colonizes the first millimeters in biofilms no cleaning reaches. As the ellagic capital falls, the wine also loses the redox buffer that helped keep it clean, so the two curves work against each other. EBX Barrel Protect covers the empty barrel between wines and removes the uncontrolled SO2 a sulfur wick hands to the next one. Barrel maintenance

Why the per-hectoliter figure is misleading on its own. A barrel delivers wood, oxygen and time simultaneously. A tannin delivers wood chemistry only. Comparing prices without pricing the oxygen contribution understates the barrel; comparing without pricing the maintenance and the depreciation of the chemistry overstates it. The honest comparison prices the outcome, which is what the tool tries to do: what dose of what format reproduces what your barrel currently gives.

Where the economics change most sharply. On a young barrel park, a tannin replaces less than people expect, because the barrels are genuinely working. On an old park, it replaces almost all of the chemistry, and the barrel is then functioning as a container with a slow oxygen leak. That second case is where the arithmetic becomes very one-sided, and it is also the case most cellars are in without having measured it.

And the resource argument, which is not marketing. We buy oak. We are already sourcing it 200 km (125 miles) further north than forty years ago because the forests are under heat and drought stress. A tree yields the same wood whether it becomes barrels, staves or extract; what changes is how much wine it can serve. Using less oak per hectoliter is not a compromise on quality if the chemistry delivered is equivalent, and that equivalence is exactly what we are willing to be measured on.

Why the empty barrel is the weak point. Between wines the barrel is warm, damp, and rich in residual sugars, ethanol and nutrients trapped in the wood. That is close to ideal for Brettanomyces and for acetic bacteria. Wood is porous to a depth of several millimeters, and biofilms established there survive high pressure washing and hot water, which reach the surface but not the porosity.

Why the sulfur wick is a partial answer. Burning sulfur produces gaseous SO2 that fills the void and dissolves into the damp wood, giving a real antimicrobial effect at the surface. But the dose depends on wick size, burn completeness and barrel humidity, none of which you measure, and part of it converts to sulfuric acid in the wood, which is not something you want to hand to your next wine. The SO2 that transfers on filling is typically in the tens of ppm and is essentially unquantified.

Where the two curves meet. As a barrel ages, its ellagic capital falls, and with it the redox buffer that helped keep the wine microbiologically stable. So the barrels most at risk of Brett are precisely the ones giving the least protection to the wine inside them. Treating the empty barrel addresses the side of that equation you can actually control.

What it does not do. It does not restore the oak chemistry of a spent barrel. If your barrels are old, keeping them clean is necessary but it is a separate question from what they still deliver to the wine, and that second question is what our decision making tool is for.

What it is. Volatile phenols absorbed by the fruit: guaiacol, 4-methylguaiacol, cresols, syringol. A little is free, most is bound as glycosides that release over months in the wine and in the mouth under saliva enzymes. That is why the defect arrives as a retro-nasal ash after you swallow rather than as smoke on the nose.

Why a tannin cannot fix it. Nothing in a tannin cleaves those glycosides. The tools that lower the phenol load are upstream: short maceration, gentle pressing, then carbon or protein fining and reverse osmosis with resin.

Where it earns its place. Once the load is managed, what remains is a hollow, drying finish. Rebuilding mid-palate volume pushes the ash back and lengthens the wine over it.

How to run the trial, and this matters most. Taste after swallowing and wait, because the ash arrives late. Then re-taste after several weeks: the bound precursors keep releasing, and a blend that reads clean on day one can fail three months later.

Why the budget is not a simple sum. What goes in at crush is largely consumed: oxidized, bound to proteins, precipitated during fermentation. What goes in at finishing is largely perceived. They do not add up in the sensory sense, and treating them as one running total tends to produce the opposite of what was intended: under-protection early, on the grounds that more can be added later, then over-dosing at finishing because the wine is short.

The split we would suggest is protection early and without hesitation, since it is spent doing its job; structure in the middle, sized against what the fruit already gives; flavor last and sparingly, since it is the part the drinker meets directly.

Getting an anthocyanin figure without a laboratory. An extraction on crushed berries at pH 1 and at pH 3.2 gives total potential and extractable fraction. The gap between them says more about how the color will behave than either number alone. It takes twenty minutes at reception and it is the single measurement we would most encourage adding.

Why high Brix changes the objective more than the dose. Alcohol is a solvent, and at 15% the wine extracts seed tannin that would have stayed in the fruit at 13%. The higher pH that usually accompanies it also makes the color less stable. So the work tends to become protective and color-directed rather than structural, and the wine often needs tension at finishing rather than weight: heavier toasts, #14 and #15 on French oak or #23 and #24 on American, with Fruitissimo on reds or Minéralité on whites. That is a tendency and not a rule: EBX 810 also protects and stays reactive, so on hot fruit heading for a fast fermentation it can still earn its place. Judge it on the phenolic ripeness and on how fast the ferment will run, not on the Brix alone.

Why phenolic ripeness is a tasting judgement. Seed color and crunch, skin tannin that dissolves rather than grips. No instrument replaces walking the tanks with a handful of berries. On green fruit the budget shifts almost entirely to finishing: Pure Grape Seed for mid-palate, Amplitude on reds or Rondeur on whites, and lighter toasts, #12 and #13.

The YAN point, because it is easily misread. Low YAN gives sluggish fermentations and H2S. A tannin neither helps nor harms much there, beyond binding some of the medium-chain fatty acids that inhibit yeast late in fermentation. It is not a reason to reduce nutrition.

And where our tools can help rather than decide. Once the wine exists, the fruit numbers stop driving. Our decision making tool works from the wine's own pH, alcohol, SO2, temperature and aging target. The drop test tool turns one glass trial into a dose across liquid tannin, powder, chips and staves with the integration time for each. Both give proposals, not verdicts, and they are meant to be adjusted with your own oenologist.

Why early in a warm year. High pH pushes anthocyanins toward their colorless forms and makes them less stable, so free color disappears faster. More extraction will not save it. What saves it is giving the anthocyanins reactive partners at crush, so they condense into stable pigment before they degrade. The window matters more than the dose in a hot year.

Why the tannin carries more of the load. SO2 activity collapses as pH rises, so a warm-vintage wine is less protected than its numbers suggest. Adding more SO2 is poor value; the protection has to come from the tannin and from oxygen control.

Why heavy toast on cooked fruit. This one runs against instinct. Faced with a flat, jammy wine, the reflex is to soften it with a sweet vanilla toast, and that makes it worse: vanillin and the light toast register reinforce exactly the sweet impression you are trying to break. Strong toasting moves the profile toward furfural, smoke and roast, which read as dry and vertical rather than sweet. That is the tension a cooked-fruit wine is missing. And because strong toasting degrades much of the raw wood tannin, you gain that tension without adding astringency to a wine that already carries plenty of rustic tannin of its own.

Why less extraction in a cool year. Phenolic ripeness lags sugar ripeness, so the seed tannin is still green and bitter and extraction is the enemy. You end up with a wine that is correct but hollow, which is the right problem to have: a hollow mid-palate can be rebuilt at finishing, a green over-extracted one cannot.

Why the color argument matters more than the protection argument. At pH 3.4 roughly a quarter of the anthocyanin sits in its colored flavylium form. By pH 3.8 that share has fallen sharply, and the rest is present but invisible. Polymerized pigment does not behave that way: its color is far less pH-dependent, and it also resists SO2 bleaching. A high pH wine that has converted its anthocyanins looks quite different at bottling from one that has not, even when both started with the same analysis.

Which is why timing is even less forgiving than usual. The conversion needs anthocyanins, tannin and oxygen at the same time, and the anthocyanin pool is at its largest in the first days. On a high pH wine an addition made three weeks into maceration is working on a pool that has already faded, and the visible result will be disappointing.

Oxygen, carefully. Oxidation runs faster at high pH, so the same oxygen rate does more damage. But the condensation you need also runs on oxygen. The answer is a lower and better controlled rate rather than none, which is exactly the case where our decision making tool is worth using rather than applying a habit.

And the microbiological side, which often decides the vintage. High pH favors Brettanomyces and lactic bacteria, and the SO2 that would normally hold them back is weakened. Nothing in a tannin addresses that. What helps is hygiene: barrels covered between wines with Barrel Protect rather than a sulfur wick, and SO2 reserved for where it still works.

One measurement worth insisting on. Track molecular SO2 rather than free SO2. Two wines with the same free SO2 at pH 3.4 and pH 3.8 are not protected to the same degree, and a cellar that only tracks free SO2 will not see the difference until something goes wrong.

The two routes, and why both are in play. Direct condensation between anthocyanin and tannin gives the A-T and T-A adducts. The second route runs through an acetaldehyde bridge, and acetaldehyde comes from the oxidation of ethanol. In other words, the reaction that stabilizes your color runs on oxygen, which is not intuitive if you have been trained to keep oxygen out.

Why ellagitannins are the right partner rather than a competitor. They act as mediators: they take up oxygen and release it into the system as reactive intermediates instead of simply consuming it. That is why an ellagitannin plus a controlled oxygen regime outperforms either on its own, and why color work and micro-oxygenation belong in the same conversation.

Why timing beats dose here. Anthocyanin concentration peaks in the first days of maceration and falls from then on. Condensation and degradation are competing for the same molecules. An addition at crush works on a full pool; the same addition three weeks later works on what is left. This is the most common reason a color trial disappoints, and it is a scheduling problem, not a product problem.

Why the oxygen regime is not optional. Under inert conditions the ellagitannin sits unreacted and the product looks ineffective. Where you have cliquage or micro-oxygenation, our decision making tool returns the regime rather than a guess. Where you do not, the oxygen has to come from the container or from the wood itself.

And the high pH case. Less of the anthocyanin sits in its colored form and the whole system is less stable. The answer is not more extraction, which brings bitterness with it. It is earlier condensation, which is the same conclusion as the warm vintage strategy.

Why the SO2 bleaching test tells you the answer. Bisulfite adds to the C4 position of the flavylium cation and forms a colorless adduct, so free anthocyanin loses its color. Polymeric pigment does not have that position available and keeps its color. The gap between the two measurements is a direct read on how much of your color is still convertible.

Why time closes the window. Free anthocyanins degrade by hydration to the colorless hemiketal, then by chalcone formation and cleavage. Those losses are irreversible. Meanwhile condensation is competing for the same molecules. So every week you wait, a larger share of the pool has gone down the degradation route rather than the condensation route.

What actually helps late. Copigmentation effects fade with time and are not a durable answer. Adding condensed tannin to an anthocyanin-poor wine produces astringency without color. Adding ellagitannin plus oxygen to an anthocyanin-poor wine consumes oxygen without color. Neither is harmful, but neither is what you were hoping for.

And the prevention point, since it is the real answer. The reason color work belongs at crush is arithmetic: the anthocyanin pool is at its maximum on day one and falls from then on. Every color program that disappoints was scheduled late.

Why fast extraction tastes young. Astringency comes from tannin binding salivary proteins, and the binding depends on the degree of polymerization and on the configuration of the molecule. A short, aggressive extraction yields a population rich in small, highly reactive units and in free phenolic acids. Those bind salivary protein readily and give the drying, gripping sensation, and they lack the larger structures that read as volume rather than grip.

Why time does not always fix it. In a barrel, the same molecules arrive slowly and immediately meet the wine's polysaccharides, mannoproteins and anthocyanins, which associate with them and blunt the astringency as they arrive. A large addition delivered at once saturates that buffering capacity, and part of it stays free and aggressive. Slower delivery, or a lower dose split into two, gives the wine a chance to do what a barrel would have let it do.

Why heat also costs you function. The same degradation that makes an extract taste raw destroys the catechol and galloyl configurations that give oxygen-handling capacity. So a raw-tasting tannin is usually also a less effective one, which is why the two problems are worth treating as a single question about extraction.

And a practical check. Taste the product in water at the dose you intend to use, then in the wine, then again at 14 days. If it is aggressive in water and stays aggressive at 14 days, it is the extraction. If it is aggressive in wine on day one and comfortable at 14 days, it was the reading.

Why there is no toxicity concern. Condensed tannins show antimicrobial activity, but at concentrations far above anything used in winemaking. At 10 to 50 g/hL (0.8 to 4.2 lb per 1,000 gal) there is no measurable effect on viability, and trials comparing fermentation curves with and without tannin show no delay attributable to the addition.

Where the real benefit sits. Octanoic and decanoic acids accumulate in the last third of fermentation and inhibit yeast membrane transport. That is a classic cause of sluggish finishes, and it is the mechanism yeast hulls address. Polyphenols adsorb the same acids, so the tannin contributes to the same effect without being a solution to it.

The oxygen question, since it will come up. Yeast need oxygen early, for sterol synthesis. A tannin added at crush consumes oxygen too. In practice there is no conflict: the yeast oxygen demand is met by an aeration during the first third of fermentation, and the yeast take that oxygen up far faster than the tannin does. Do not change your aeration practice because a tannin is present.

What not to expect. No effect on alcohol tolerance, no effect on nitrogen availability, no rescue of a fermentation already stuck. If that is what you need, the honest answer is a nutrition and temperature review.

Why the imbalance peaks here. Extraction is sequential: anthocyanins early and aqueous, tannins later and alcohol-dependent. During cold soak the ratio of anthocyanin to tannin is the worst it will be all vintage, and free anthocyanin is exactly the form that degrades. Adding tannin at that moment corrects the ratio rather than adding structure for its own sake.

Why the tank is vulnerable. No fermentation means no CO2 cover and no yeast oxygen uptake. Every pump-over, every délestage and every probe opening puts oxygen into a must that has nothing to absorb it. And oxygen solubility rises as temperature falls, so a must at 50 °F (10 °C) holds appreciably more dissolved oxygen than the same must at 68 °F (20 °C).

Where SO2 does not save you. It temporarily bleaches the color, which masks what is happening, and on a high pH must its molecular fraction is too low to do much. This is a case where the tannin does the work SO2 cannot.

The practical protocol. Add at filling, before cooling. Keep the cap covered and work under CO2 or inert gas during transfers. And resist judging the trial on the color of the juice: compare tanks at the end of fermentation, after SO2 bleaching, which is the only number that predicts what will be in the bottle.

Where H2S comes from. Yeast reduce sulfate and sulfite through the sulfate assimilation pathway and produce H2S when nitrogen is short. Elemental sulfur residues on the fruit and the breakdown of cysteine and glutathione add to it. Nitrogen management upstream does more than any corrective addition, and that is worth saying before reaching for a fix.

Why it gets worse if left. H2S combines into methanethiol and ethanethiol, then oxidizes into disulfides. Disulfides are outside the reach of both copper and tannin, and they can revert to mercaptans later in bottle. The window for an easy correction is early.

Why the tannin works, and why copper is not the only option. Quinones generated from tannin oxidation add nucleophilic thiols onto themselves, removing them from the wine. A small oxygen input is therefore part of the treatment rather than a risk to it. Copper remains effective but leaves residual copper behind, and copper catalyses the oxidation cascade described above, so today's fix becomes tomorrow's oxidation. Keep the residual within your local limit, which in the United States is 0.5 ppm.

How to read the trial. Evaluate at 24 and 48 hours, not immediately, because the reaction is not instant. Run a copper reference at 0.2 ppm alongside: if copper clears the wine and the tannin does not, you are dealing with something the tannin will not reach.

Why acetaldehyde is the tell. Hydroxyl radicals react with whatever is most abundant, and in wine that is ethanol. So a rising acetaldehyde means the oxidation chain ran past the phenolics. It is a dosing indicator as much as a fault, and it should be read that way: slow the oxygen, or raise the tannin.

Why reds and whites are not the same problem. In a red, acetaldehyde is consumed as fast as it forms provided anthocyanins and tannin are both present, which is why a correctly run micro-oxygenation does not smell aldehydic. In a white there is no anthocyanin to consume it, so it accumulates and the only route left is SO2, which binds it. That hides the aroma but inflates total SO2 and leaves you with less free SO2 than before.

Which brings the two subjects together. A collapse in free SO2 after an oxygen event is usually acetaldehyde binding, not SO2 disappearing. Adding more SO2 treats the symptom. Getting the tannin in first treats the cause.

What to watch in the cellar. Aldehydic, bruised-apple notes, a palate that hardens instead of softening, and free SO2 falling faster than usual. Any of the three means the oxygen rate is above what the wine can absorb.

The cascade, briefly. Molecular oxygen is not directly reactive with most wine components. It is activated by iron and copper, reduced stepwise to superoxide, then hydrogen peroxide, then through Fenton chemistry to the hydroxyl radical. In parallel, catechol and galloyl phenolics are oxidized to quinones. Two different threats, and a tannin addresses them differently.

Against quinones the tannin is genuinely effective, because it is oxidized in place of your wine's own phenolics and can reduce quinones back. Against hydroxyl radicals no molecule is selective, they react at diffusion rate with whatever is nearest, so protection is purely statistical. That is the whole argument for adding early and in a wine that has not already started oxidizing.

Why reactivity, not just quantity. A degraded extract carries phenolic mass without the catechol and galloyl configurations that do the work. Our low temperature extraction preserves them, which is why a comparison with a commodity tannin should be made on effect, not on total polyphenol content.

And do not import the catalyst. Iron and copper drive the whole sequence. A tannin extracted in mineral-rich water brings some with it. Ours is extracted in reverse osmosis water and brings none.

One boundary worth stating. You do not want to quench everything. Quinones also drive the condensation that stabilizes color and softens tannin. The objective is to keep the oxidation on the phenolic pool and away from the ethanol and the thiols, not to stop it.

The reaction sequence, and why the tannin has to be there first. Oxygen oxidizes the catechol and gallol phenolics into quinones, releasing hydrogen peroxide. In the presence of iron, that peroxide generates hydroxyl radicals, which are indiscriminate and will attack whatever is most abundant. In wine, that is ethanol, and the product is acetaldehyde. A tannin addition enlarges the pool of reactive phenolics, so the oxygen is consumed higher up the chain and less of it reaches the ethanol.

What the wine tells you. If acetaldehyde is accumulating, the oxygen rate is above what the phenolic pool can absorb. The signs are an aldehydic, dried-fruit character and a palate that hardens rather than softens. That is not a product failure, it is a dosing error, and the fix is to slow the oxygen or raise the tannin, not to stop the program.

Temperature is the forgotten variable. Reaction rates roughly double for every 18 °F (10 °C). The same rate at 68 °F (20 °C) and at 54 °F (12 °C) does not produce the same wine, and a cellar that does not control its temperature cannot control its micro-oxygenation.

And two things not to do. Do not run structural micro-oxygenation during alcoholic fermentation, because the yeast take the oxygen and you learn nothing. And do not continue a maintenance regime out of habit once the wine has stopped responding: past that point you are simply spending oxygen on ethanol.

Sacrificial oxidation. Tannins oxidize ahead of your wine's own phenolics and ahead of the SO2. What you observe is a flatter free SO2 curve after an oxygen pick-up, not a step change.

Enzyme inhibition, the argument to lead with. Tyrosinase from healthy fruit is handled by SO2 anyway. Laccase from Botrytis is not: it resists SO2 at any dose you would use. So on botrytised fruit, adding more SO2 does not solve the problem and a tannin at crush does.

Where SO2 is poor value, and what to use instead. If you want the lowest possible SO2 in the finished wine, the place to start is not the total, it is the additions that were never effective in the first place.

Before and during fermentation. SO2 added to the must is largely bound by the acetaldehyde, pyruvate and ketoglutarate your yeast produce. It inflates total SO2 without raising the molecular fraction that actually protects. Protect the must with VBX instead and keep the SO2 for after fermentation, where it works.

Against laccase. Covered above: SO2 does not control it at any usable dose. Spending SO2 there is spending it for nothing.

On high pH wines. SO2 activity depends on the molecular fraction, and that fraction collapses as pH rises. At pH 3.8 you need roughly three times the total SO2 of a wine at pH 3.4 to reach the same protection. Chasing it with additions is expensive, it saturates the label, and it still underperforms. On those wines the return comes from tannin plus oxygen control, not from more SO2.

And remove the input you do not control: sulfur wicks in empty barrels. Barrel Protect gives none, so the SO2 in your wine is the SO2 you decided to put there.

What a tannin cannot do. SO2 is also antimicrobial: Brettanomyces, acetic and lactic bacteria. A tannin does nothing there. Cutting SO2 on that basis trades an oxidative problem for a microbiological one.

How to run it. On botrytised fruit go to 50 g/hL (4.2 lb per 1,000 gal) and do not compensate by raising SO2, because the enzyme is what you are fighting. Then follow free SO2 across the first week rather than at a single point.

On low and no-SO2 programs, tannins are central but only inside a protocol: strict oxygen management, clean fruit, inert gas at every transfer, and no sulfur wicks. Used alone in such a project they will disappoint.

One caution. There is no conversion between grams of tannin and ppm of SO2, and there cannot be, since the two act on different reactions. If anyone offers you a conversion rate, it is a sales figure, not a chemical one.

What actually precipitates, and why the percentage cannot be predicted. Three mechanisms run at once: reaction with proteins, self-association of the more polymerized fractions, and growth of anthocyanin and tannin polymers until they exceed solubility. The proportions depend on your wine's protein content, its temperature history, its pH and the tannin's own degree of polymerization. Two wines given the same dose will not lose the same fraction.

Where the tannin type matters. Highly polymerized and heat-degraded extracts come out of solution faster, because they arrive already close to their solubility limit. A tannin extracted at low temperature stays in solution longer and integrates rather than settling, which is one of the practical differences between our products and a commodity extract, and it is measurable in your cellar rather than only on a datasheet.

Cold stabilization is a filter you did not plan. Holding a wine near freezing precipitates tartrates, and it takes color and tannin down with them. So tannin work belongs before cold stabilization, with enough time to integrate, not after it.

And a note on bentonite. Tannins precipitate proteins, so a tannin addition can reduce the bentonite requirement on some whites. That is a real saving, but it has to be measured on a fresh heat test rather than assumed, and it never justifies skipping the test.

Why the closure matters more than the tannin at this stage. Oxygen ingress after bottling is measured in fractions of a ppm per year and varies by an order of magnitude across closure types. That flux drives the same chemistry as micro-oxygenation, only far slower: condensation, polymerisation, and eventually oxidation of the ethanol once the phenolic pool is depleted. A wine with a healthy phenolic reserve tolerates a more permeable closure; a wine without one does not.

Why reductive character can appear late. Under a very tight closure with no oxygen, disulfides can revert to mercaptans, and a wine that smelled clean at bottling can turn reductive in the following year. A tannin addition before bottling raises the redox buffer and reduces that risk, which is a real argument for #10 on wines destined for screwcap.

Why late additions are the wrong answer. An addition made after stability testing invalidates the tests, and an addition made immediately before filtration is partly removed by the filter and has had no time to integrate. The correct sequence is fining, then tannin, then integration, then cold stabilisation, then testing, then filtration.

And the part we would not overstate. Predicting a specific wine's trajectory in bottle from its analysis at filling is not something we can do reliably. What we can do is make sure it enters the bottle with the reserve it needs, and be honest that the rest is time.

Why alcohol changes everything. Ethanol is a co-solvent. It raises the solubility of the more polymerized and more lipophilic fractions, and it disrupts the hydrogen bonding between tannin and salivary protein that produces astringency. Remove the alcohol and two things happen at once: part of the tannin becomes less soluble, and what remains in solution reads as more astringent than it would in wine. So the useful doses are lower and the choice of fraction matters more.

Why a de-alcoholised drink feels hollow. Ethanol contributes viscosity, a sweet impression, and it carries volatile compounds to the headspace. Taking it out removes body, sweetness and aromatic release simultaneously. A tannin can only address the first of the three, but body is the one people notice most, and it is what makes the difference between a drink that finishes abruptly and one that has length.

Why the choice of botanical widens here. In wine the question is mostly oak and grape. Outside wine, acacia, wild cherry and other sources bring different tannin structures with different astringency and different color contributions, which is why these are distinct formulations rather than the wine range relabelled.

And on spirits. At 40% ABV and above the extraction is fast and the solubility limits are different again. The equivalence between a tannin dose and a wood program has to be recalculated for the strength, which is exactly the kind of thing our drop test tool is built to handle.

Why strength dominates in spirits. Ethanol content governs the solubility of the lignin-derived and lipophilic fractions. At 40% ABV and above, compounds that would never leave the wood in wine come out readily, including the whisky lactones and a broader slice of the toasting products. That is why a wine dose transposed at the same g/hL overshoots, and why the extraction curve is steeper and shorter.

Why beer is the more demanding medium sensorially. Lower alcohol, lower phenolic background and, in most styles, far less complexity to absorb an error. Astringency that would pass unnoticed in a Cabernet is a fault in a saison. The margin between interesting and marked is the narrowest of any medium we work in.

Why the maturation of the wood still matters here. Thirty-six months of open-air maturation degrades the harshest ellagic fractions and lets aromatic precursors form. In wine that shows as smoothness; in beer it shows as the difference between oak and raw wood. It is the same reason our toasting equipment was designed from pastry industry principles rather than cooperage fire: reproducible thermal curves matter more when the medium is unforgiving.

And what we do not yet fully explain. Colloidal behaviour of oak tannins in beer, particularly the interaction with proteins and polyphenols already present, is not something we would claim to model. We have empirical results, and we are still building the understanding behind them.

Why alcohol removal changes the chemistry and not just the taste. Ethanol is a co-solvent. It raises the solubility of the more polymerized and more lipophilic fractions, and it disrupts the hydrogen bonding between tannin and salivary protein that produces astringency. Remove it and two things happen together: part of the tannin becomes less soluble, and what stays in solution reads as more astringent than the same dose would in wine. So the useful window is lower and narrower, and the choice of fraction matters more than the dose.

Why the botanical range widens here. In wine the question is mostly oak and grape. Outside wine, acacia, wild cherry and other sources give different tannin architectures, with different astringency, different color contribution and different interaction with sugars and acids. These are distinct formulations rather than the wine range relabelled, which is also why some of them are marked not Codex: they are built for a framework where the Codex does not apply.

What is still open. Mouthfeel in a zero alcohol matrix is not a solved problem. The interaction between tannins, the polysaccharides used for body, residual sugars and the acid profile is complex, and the sensory literature is thin compared with wine. We have working hypotheses and a growing set of trials, and we will say plainly when a result is empirical rather than explained. If you are working on this, we would rather learn alongside you than sell you a certainty.

Why the two-week reading is the one that matters. Oak intensity falls by roughly a factor of four over the first two weeks as the addition redistributes between the free and colloidal phases. Products do not fall at the same rate, so a tannin that wins on day one often loses at day fourteen. Judging early systematically leads to over-dosing.

Why the control has to sit beside the trials. The three variables that break a bench trial are oxygen, temperature and time. Storing the control elsewhere, or in a partly empty bottle, changes all three. A control drawn from the same tank at the same moment and stored alongside is what makes the comparison mean anything.

Where the correlation is strong and where it is not. With liquid tannins the correlation with the tank is essentially one to one, because the addition carries no oxygen of its own and a sealed sample reproduces the tank. It is never quite true for chips, which bring oxygen with the wood, and it is impossible for staves, which do not fit in a glass. For those, the trial gives a reliable direction and the profile decision, and the equivalence table does the rest. That is a geometry limit, not a product limit, and no supplier gets around it.

And a limit worth stating. Every model behind these tools is a simplification of a wine that is more complicated than the model. We would rather you use it as a starting point and correct it with what you taste.

Why we built calculators rather than a dosing chart. A chart assumes the wine is average. The variables that actually move a result, pH, temperature, oxygen availability, what the wood already contributes, are exactly the ones a chart cannot hold. Each of the three tools exists because a question kept being answered by opinion in this industry, and we happened to have the measurements to answer it by calculation.

Where the numbers come from. We select the oak, mature it, toast it in our own kilns and extract our own tannins from the same wood. That means we can compare a tannin and a stave on the same oak, the same grain and the same toast curve, which is what the equivalence tables rest on. Almost nobody else sits on both sides of that line: tannin suppliers generally buy an extract, and coopers make wood and stop there.

The reference conditions, so you can judge the output. The drop test model runs on pH 3.5 at 68 °F (20 °C), and both are adjustable, because a wine at a different pH or a colder cellar moves along the integration curve at a different rate. Oak intensity falls by roughly a factor of four over the first two weeks, which is why the reading that counts is at day 14 and not on the day of the addition.

And the honest limit. No calculation replaces a bench trial on your own wine. What the tools do is stop you starting from zero, and stop you dosing on habit.

Why the ratio is so large. A barrel uses the oak as a container as well as a source of chemistry, so most of the wood's mass is doing structural work rather than oenological work, and the wine only ever touches the first 2 to 3 mm (0.08 to 0.12 in). Staves remove the container function. Extraction removes the geometry entirely and delivers only the soluble fraction. Each step recovers wood that was previously unused.

Where the argument is weaker, and we would rather say so. A barrel gives oxygen and time along with the wood, and those are real contributions a tannin does not supply on its own. So the comparison is only fair once the oxygen is accounted for, through the vessel, through micro-oxygenation, or through a thick enough wood format. Comparing on wood alone flatters us.

The rest of the footprint. Low temperature extraction is far less energy-intensive than hot or pressurised processes, we use decarbonised energy, and packaging is kept minimal and recyclable. None of that is dramatic on its own. Together with the resource ratio it is the part of our operation we can actually measure.

And the part we cannot yet measure well. The full life cycle comparison, including transport, cooperage energy and end of life, is not something we have published to a standard we would defend. We have the resource arithmetic and we are honest that it is one component of a bigger picture.

Why the system resists modelling. A wine contains thousands of phenolic species in equilibrium, distributed between free and colloidal phases, reacting through pathways whose rates depend on pH, temperature, oxygen availability, metal catalysts and the concentration of every other participant. Most published work isolates one or two variables. Applying those results to a real tank requires assumptions that are rarely stated.

Where the uncertainty is largest, in our experience. Anything involving perception rather than concentration. We can measure a polymeric pigment fraction with confidence; predicting how a taster will describe the resulting wine is much harder, and the relationship between the two is not linear. The same is true of astringency, which depends on salivary protein composition and therefore varies between people.

What we do about it. We measure on our own material, across formats, under controlled conditions, and we build the tools on those measurements rather than on literature values. Where we extrapolate, we would rather the tool say so. And we keep running trials with winemakers who are willing to share results, which is worth more than any internal programme.

And a practical consequence for you. Treat every dose on this page as a hypothesis for your wine, run the control alongside, and read at fourteen days. If our estimate is wrong for your cellar, that is information, and it is worth more to both of us than a number that was right on average.

Why sourcing sits inside the chemistry. Quercus petraea and robur are rich in ellagitannins and low in whisky lactone; Quercus alba is the reverse. Grain, driven by growth rate, changes extractable phenolics per unit area and the speed of release. Those are decisions made in the forest, and they set the ceiling on everything downstream.

Why controlling the extraction water matters. Reverse osmosis water carries no iron, copper or calcium. Iron and copper are pro-oxidant catalysts that drive the whole oxidation cascade in wine, so a tannin extracted in mineral-rich water arrives carrying the very catalysts it was added to fight. This is invisible on a datasheet and visible in a wine six months later.

Why low temperature is not a preference. Hot or pressurised extraction degrades ellagitannins into simpler phenolic acids and destroys the catechol and galloyl configurations that give oxygen-handling capacity. What survives is phenolic mass without function, which is exactly why we would rather be judged on effect than on a total polyphenol index. The index measures reducing power against a reagent, not what happens in your tank.

And the honest cost of it. Selecting rather than extracting everything means accepting a lower yield by design. That is a deliberate choice, and it is also why we do not compete on price per kilo.

What the cooperage inheritance actually contributes technically. Three things that are hard to acquire any other way. Knowing which tree to buy, which is a judgement made standing in a forest and not from a specification sheet. Knowing what open-air maturation does to a stave over three years, which you only learn by owning the wood for three years. And knowing what a toast curve does to the gradient inside the wood, which coopers learn at the fire.

Why that translates into extraction. Our process selects fractions rather than maximising yield, and knowing which fractions to select is the cooperage question restated in chemical terms. What a barrel gives a wine is not a mystery to someone who has made barrels; the work was to reproduce it deliberately rather than by tradition.

And what we are careful not to claim. Heritage is not evidence. It gives us better questions and a better starting point, not a guarantee about your wine. Every figure on this page is an estimate from measurements we have made on our own material, and each of them has a range around it that a real cellar can move.

What is actually happening in the stack. Enzymes present in the wood, activated by humidity, oxygen and time, progressively break down macromolecules: tannins, hemicelluloses and lignins. That degradation releases the more stable and harmonious aromatic compounds and removes the harshest phenolic fractions. Fungal colonisation of the surface plays a part too. It is a slow biological process, not simply drying, which is why kiln drying and three years of air are not interchangeable.

Why the ellagitannin loss is a gain here. Raw oak carries a high load of aggressive ellagitannins that read as bitter and drying. Maturation lowers it and shifts the balance toward the fractions that structure without gripping. So the wood arrives at the toasting stage with less to hide.

Why controlling the conditions changes the outcome. Enzyme activity has an optimum range for humidity and temperature. Left entirely to the weather, a stack spends part of the year outside that range and the work stalls. Keeping the wood in the working zone lengthens the effective maturation without lengthening the calendar, and it narrows the variation between batches, which is where reproducibility comes from.

And what remains empirical. The relationship between maturation conditions and the final sensory profile is something we manage well and do not fully model. We know what to do; we would not claim to be able to predict it from first principles. That gap is one of the things we are working on.

Why the oxygen question dominates. The reactions that turn an addition into a lasting result, condensation between anthocyanin and tannin, polymerisation that softens texture, are oxygen-dependent. Acetaldehyde, formed by oxidation of ethanol, is the bridge in the indirect route. No oxygen means the ellagitannin sits unreacted, the colour does not convert, and the oak stays beside the wine rather than melting into it.

Why too much is as bad as none. Above what the phenolic pool can absorb, the oxygen runs past the phenolics and oxidises ethanol, and acetaldehyde accumulates. The wine hardens and turns aldehydic. So the target is a rate matched to the tannin content, not a rate matched to the volume, which is the most common error in micro-oxygenation.

Why thermal inertia matters more than people credit. Reaction rates roughly double for every 18 °F (10 °C). A concrete vessel that holds a stable temperature runs a predictable reaction; a stainless tank swinging with the cellar does not. That is part of why identical additions in identical wines can end up different.

And where our knowledge is thinner. Oxygen transfer rates through unglazed terracotta and unlined concrete are poorly characterised in the literature and vary with wall thickness, porosity, age, and what the vessel has held before. We would treat any published figure as an order of magnitude, and we would rather work from what your wine tells you across a season.

Why colour is the binding constraint. Free anthocyanins are absent in a blanc de blancs and low in most bases, so nothing converts the addition into stable pigment. What you add stays as free phenolics, and phenolic yellowing shows directly. In a still white a light golden cast can suit the style; in a sparkling wine it usually does not.

Why pressing is the high-value moment. Quinones formed within minutes of pressing destroy varietal thiols and initiate the oxidative cascade. A sacrificial tannin on the juice is oxidised in place of the wine's own phenolics and can reduce quinones already formed back to phenols. Added later it can only mop up what is left.

What changes through second fermentation. Yeast autolysis releases mannoproteins and polysaccharides that associate with phenolics and modify perceived texture, and the pressure environment changes the redox picture. Both plausibly change how an earlier addition reads at disgorgement. We can describe that directionally; we would not claim to predict it.

And on foam. Protein and polysaccharide interactions govern foam quality, and tannins bind protein. There is a theoretical route by which a large addition could affect foam, and we have not seen it at the doses discussed here. We would rather flag it as unresolved than assert there is no effect.

Where the SO2 actually goes in a low-intervention protocol. SO2 added to must is largely bound by acetaldehyde, pyruvate and alpha-ketoglutarate produced by the yeast. It inflates total SO2 without raising the molecular fraction that protects. Moving that protection to a tannin at crush and keeping the SO2 for after fermentation is usually the single largest reduction available, and it costs nothing sensorially.

Why high pH compounds the problem. SO2 activity depends on the molecular fraction, which collapses as pH rises. To hold 0.5 ppm molecular you need roughly 20 ppm free at pH 3.4 and roughly 50 ppm at pH 3.8. Chasing that with additions is expensive and still underperforms. On those wines the return comes from tannin plus oxygen control.

Why oxygen discipline is the real work. Without SO2 as a buffer, every transfer, every topping and every sample is an oxygen event, and the wine has only its phenolic pool to absorb it. Enlarging that pool early is the strategy; the tannin is the tool, not the strategy.

And a limit we would rather state. There is no conversion between grams of tannin and ppm of SO2, and there cannot be, because the two act on different reactions. If anyone offers you one, it is a commercial figure rather than a chemical one. What we can help with is placing the additions so that the SO2 you do use is doing work.

Why composing at the wood stage matters. Blending finished extracts averages what you already have. Composing across toast curves before extraction lets us target the gradient a real barrel produces, strongly toasted at the surface and progressively lighter with depth, which is what a taster registers as barrel rather than as oak. It also means the same recipe can be delivered as tannin or as stave without redesign, because the composition lives in the wood rather than in the mixing.

Why we work from a control wine rather than a description. Sensory vocabulary is not shared reliably between cellars. Two winemakers using the same three words often want different wines. A control sample removes most of that ambiguity in one step.

Why a season rather than a trial. Integration takes two weeks, structure takes months, and bottle evolution takes longer again. A judgement made at forty-eight hours answers a different question from the one usually being asked.

And an honest limit on custom work. We can reproduce a target reliably when the target is achievable with oak chemistry. We cannot fix a wine whose problem is elsewhere, and we would rather say so at the start than deliver something that disappoints in the spring.

Why we build tools rather than only give advice. Advice does not scale, and worse, it is not reproducible. A calculation you can run yourself, on your own numbers, at two in the morning during harvest, is worth more than an opinion you have to wait for. It is also auditable: you can see what changes when you change an input, which is how a cellar builds its own judgement rather than borrowing ours.

Why the trial protocol gets so much of our attention. Three variables break bench trials: oxygen, temperature and time. A control drawn at the same moment from the same tank, stored beside it, sealed with minimal headspace, read at fourteen days, blind and randomised. Almost every product rejection we investigate turns out to be one of those four missing.

What we would like back. Results, including negative ones. Oenology advances on shared observation, and a cellar that tells us a dose did not behave as our model predicted is contributing more than one that quietly moves on. Our figures improve because people tell us when they are wrong.