Fining and Filtering Wine: What Gets Removed

Filter a wine tightly enough to hold back yeast and bacteria and you take something out of the glass as well: "filtration at this level may lighten a wine's color and body". That is the trade, and everything else is detail.

Fining and filtering wine are two separate jobs done near the end of winemaking. Fining adds an agent that binds to tannin and to the microscopic particles that would otherwise cloud the wine, so they fall out as sediment. Filtering then strains the liquid mechanically, for one of two purposes: clarification, which takes out what you can see, or microbial stabilisation, which takes out the organisms that could restart a fermentation in bottle. Clarifying or polishing catches particles larger than 1 to 4 micrometres. Yeast retention needs at least 0.65 micrometres. Bacteria retention needs 0.45. The tighter the filter, the more of the wine leaves with the microbes.

What do fining and filtering actually do to wine?

Fining agents are used "to remove tannins, reduce astringency and remove microscopic particles that could cloud the wines". Which agent a winemaker reaches for is a per-batch decision that changes from product to product and even from one year's grapes to the next.

Astringency is the point worth holding onto. Tannins react with proteins, and most fining agents are proteins. In the mouth, "Tannins add a gritty texture and chalky, astringent taste." Pull some of that tannin out in the cellar and you have made the wine softer and less structured at the same time, which is why the decision is not a neutral one. Our guide to tannins in wine covers what that structure is doing for a bottle in the first place.

Filtration has exactly two objectives, and microbial stabilisation is the more misunderstood of them. It does not mean sterility. It means a significant amount of yeast and bacteria has been removed to a harmless level for the wine's stability. Sterile bottling is a stricter thing again, and it also relies on filtration to get there.

Both interventions sit at the end of a clean-up that happens on its own. Across the three to six months of secondary fermentation and bulk ageing, grape proteins break down, spent yeast cells and fine particles settle out, potassium bitartrate precipitates, and "the originally cloudy wine becomes clear". Racking the wine off its lees does much of the work. Held near 2 degrees Celsius, a wine will settle and run clear in about a month with no chemicals at all.

Which fining agents end up in the bottle?

Almost none of them, by design. Gelatin has been used for centuries and remains the most common agent for cutting tannin. "Generally no gelatin remains in the wine because it reacts with the wine components", clarifies, and forms a sediment that filtration removes before bottling. The agent is a vehicle, not an ingredient.

The list a producer can choose from is longer and stranger than most buyers expect. Animal-derived options include micronized potassium caseinate, egg whites, egg albumin, bone char, bull's blood, isinglass from sturgeon bladder, lysozyme and skim milk powder, alongside PVPP, a synthetic compound. Finely ground eggshell turns up too, though rarely. The non-animal filtering agents are bentonite, a volcanic clay, plus diatomaceous earth, cellulose pads, paper filters and membrane filters, which are thin films of plastic polymer with uniformly sized holes.

Copper is a separate case with its own risk. The Australian Wine Research Institute describes fining with copper sulfate as the standard treatment for the rotten-egg smell of hydrogen sulfide, because copper sulfate reacts with it to form copper sulfide, which is highly insoluble. The AWRI also advises that laboratory trials should precede any addition to bulk wine, since an instability can result once copper in the wine exceeds roughly 0.5 mg/L, and lower than that in some wines. Copper fining also has a limit: once those sulfur compounds have been oxidised to disulfides, they no longer react with copper and cannot be removed by copper fining until reducing conditions are created to turn them back. That is the mechanism behind several of the reductive and oxidative faults you meet in older bottles.

Why do some producers bottle unfined and unfiltered?

Because the filter fine enough to guarantee stability is the same filter that lightens colour and body, and some producers would rather carry the risk than pay that price. The risk is documented rather than theoretical. The AWRI reports a rise in mousy wines through the 2010s, when winemakers experimented with extended lees ageing, high pH, minimal sulfur dioxide, and "oxidative ageing and minimal clarification or filtration in white wines". Mousiness is the fault a buyer should care about most, because of the way it hides. "Mousy taint is rarely detected by sniffing because the compounds involved are not volatile at wine pH." It arrives late on the palate, after you have swallowed, and takes a few seconds to build. Move briskly through a line-up and you can miss it entirely, which is worth knowing before any blind tasting exercise.

And it cannot be fixed. "There is no satisfactory method to remove mousy off-flavour", which the AWRI links to low sulfur dioxide and low acidity. Its own recommendation for red wine production, where cellar sanitation is in any doubt, is to work at 50 to 80 mg/L total sulfur dioxide, and higher for high-pH wines. None of that makes low-intervention bottling wrong. It makes it a bet on the cellar that filled the bottle.

Does haze cost a wine points?

Yes, and the schedules put a number on it. The OIV standard for international wine competitions scores limpidity on its own five-step scale: excellent limpidity takes 5 points, ambiguous limpidity 3, moderate cloudiness 2, and the floor is 1. OIV medals begin at 80 points for bronze and 85 for gold, so the full swing of that one clarity descriptor is the same 5 points that separates the two awards. No more than 30% of samples entered may take a medal at all.

Professional tasting grids agree. The Court of Master Sommeliers lists "Clarity / Visible Sediment" as the opening line of the sight section in its Advanced and Master Sommelier deductive tasting format, with three options: clear, hazy, cloudy. In published scoring systems, appearance is weighted as high as 15% of the total against 35% for the nose and 50% for the palate. What the eye reports is folded into the number, which is worth remembering when you read a 100-point score on a shelf talker.

What this means for a bottle you already own

A crystalline deposit is not a fault. Tartrate crystals are harmless, and an entire industrial step exists to stop them ever reaching you.

That step is cold stabilisation, aimed at the potassium bitartrate that the trade calls wine crystals or wine diamonds. Drop the wine close to freezing for one to two weeks and they separate out onto the tank wall. The motive is commercial rather than sensory: the practice improves aesthetic clarity and appeal, "as consumers often mistake harmless tartrate crystals for glass shards or wine faults". They can also form in bottles that have been stored cold, so a crystalline deposit in a case you bought years ago tells you about the storage, not the winemaking.

Genuine sediment is a different signal, and an expected one in older reds. What sediment in an old bottle actually tells you and how to decant an old bottle against a young one are the two practical follow-ons. Sparkling wine has its own answer to the same problem: bottles spend six months on a riddling rack before disgorgement clears the sediment that has accrued.

None of this tells you when to pull a cork, which is the question you actually have. Our drink-now shortlist does: it tells you which bottles in your cellar are in their window today, so you judge a wine on where it is in its life rather than on how it looks in the glass.