Ira Kreft asked the following very pointed question:
Q: In the filter/unfiltered debate, what compounds have you heard winemakers argue that get filtered out? I have heard the generality but not the specifics.
A: Winemakers not in favor of filtering argue that the process strips wine of color, aromas and flavors, resulting in an inferior wine. But Ira is right -- there have been mostly generalities and no specifics in these discussions. To answer the question, we have to consider the filter medium and size of particulates processed through that medium.
Typically, wine is clarified by successively filtering it through pads down to 1 micron or less, and then down to as low as 0.25 micro in preparation for sterile filtration. The latter is performed at bottling time using a membrane filter to remove latent microorganisms (yeasts and bacteria) that might otherwise cause spoilage during bottling aging. Specifically, spoilage yeasts and bacteria of concern in wine are in the order of 1.2 microns and 0.65 micron, respectively, or bigger. Water passes through these filter media as the size of a water molecule is in the order of 0.0003 micron.
Aromas and flavors belong to various classes of chemical compounds including alcohols, aldehydes, acids, esters, thiols, terpenes and terpenoids, saccharides, amino acids and proteins, and tannins (phenolics). These are all very small and pass through filter media unaltered.
Color belongs to the class of phenolic compounds known more specifically as anthocyanins. The simplest anthocyanin structure is one that exists in an unbound form and is thus very, very small. These are known more precisely as anthocyanidins. But anthocyanidins exist mainly in bound form in wine -- one or more sugar moieties are attached to the anthocyanidin molecule. These are still too small to be filtered out; however, during storage and aging, anthocyanins bind with other phenolic compounds, such as large, heavy tannins, and become suspended in solution (wine) and then precipitate. These are large molecules that get filtered out while in suspension if processed through filter media. If allowed to precipitate, they either form a sediment in the tank/barrel, which can be racked/filtered prior to bottling, or form a sediment in the bottle. So yes, filtering does impact color, but these are molecules that would normally bind and precipitate over the course of storage/aging. The wine would therefore normally "self-clarify" and become lighter on its own accord. Filtration achieves the same, but only faster.