Everything You Know About Tannin Is Wrong
In the world of theoretical physics, there is something called the Standard Model, which pulls together several key building blocks of how scientists understand the basic particles and forces that make up the universe. In winemaking, there is another kind of Standard Model for describing the role and evolution of tannins, which goes something like this:
Skin tannins are better (softer, friendlier) than seed tannins, so fermentations should aim to get more of the former and less of the latter. Once in the wine, tannins form longer and longer chains, in the process becoming smoother and less astringent, until finally after long aging they become so large they precipitate out.
Based on this model, many people have made a ton of very good wine. Unfortunately, each of the phrases seems to be either vastly over-simplified or flat wrong. Like the legendary (imaginary?) hard, soft and green tannins, these parts of conventional tannin wisdom don't fare well under rigorous scrutiny.
Challenges to popular views of tannin behavior were a major thread running through the several days of presentations and discussions at the June meetings of the American Society for Enology and Viticulture (ASEV) in Napa, highlighted in the talks delivered by phenolics researcher Jim Harbertson of Washington State University, Prosser, and tannin chemist Paul Smith of the Australian Wine Research Institute (AWRI). On the heels of other counter-revelations in recent years, it seemed like time to try and separate the truths from the tales by following up with Harbertson and Smith and tracking down phenolicsologist Jim Kennedy, who is still settling in after his move from Oregon State University to a new position with the AWRI.
Grapes, seeds and skins
Of all the ASEV speakers taking shots at tannin dogma, Harbertson was clearly having the most fun. In a presentation during the High-Brix Winemaking Symposium, which kicked off the meetings (and which he organized), Harbertson touched several times on the qualities of seed and skin tannins, drawing upon his own research and other literature. His overall point was that seed tannins aren't the menace they are often seen as, and that perhaps they just suffer from "bad mojo" rather than bad chemical makeup.
Seed tannins are bad-mouthed and avoided by a great number of winemakers. They are alleged to be harsher, meaner, greener, more astringent and downright unpleasant. Minimizing them is part of the motivation for techniques such as delestage (rack and return) and pressing before dryness, and a reason many winemakers shy away from extended maceration. (Full disclosure: I have passed along these views in print on more than one occasion.) Legions of growers and winemakers make a practice of chewing seeds to make harvest decisions, waiting for the mean qualities to subside in hopes of keeping them out of their tanks. It's not clear whether all this avoidance is worth it.
Tannin is a collective term for a wide variety of complex compounds, all with multiple sub-units. By themselves, in splendid isolation (or in solution), the flavon-3-ols that are the building blocks of tannin -- catechin, epicatechin and various gallate forms -- are just monomeric tannin wannabes; to earn the title tannin, they have to form polymers at least two units long -- and possibly a whole lot longer -- joined by carbon bonds. During grape maturation, polymerization proceeds in both skins and seeds, chaining up some of the monomeric pieces.
Seed tannin can be extra-astringent when its terminal unit is a gallic acid ester, which doesn't happen with skin tannins. But Harbertson observed that gallic acid is easily hydrolyzed and removed in winemaking, making that temporary astringency moot. More importantly, Harbertson kept pointing out that seed tannins average 10 units in length, while skin tannins average a little north of 30 units, so they're not anywhere near the same size. And (jumping ahead a bit), longer tannin polymers--that is, the ones found in skins--have been shown to be more astringent, which is not what we're generally looking for.
The shorter seed tannins are more plentiful--3.5mg to 5 mg worth per grape berry, says Harbertson, as opposed to 0.5mg to 0.9mg of skin tannin--but more difficult to extract. Skin tannins come out earlier and easier, thanks to warm fermentation temperatures and rising ethanol levels, and then tend to plateau after a few days; seed tannins come out slowly and steadily and for longer duration.
This means that fermentation strategies aiming to maximize skin tannin and minimize seed are going after the least abundant and most astringent tannins. What's boosting the tannin count in prolonged maceration -- the factor that may account for the smoothing/rounding effect many practitioners report -- is probably an accumulation of less astringent seed tannin.
Makes you wonder.
Grape vs. wine tannin
Once the raw tannin material from the berries gets into a wine solution, all hell breaks loose. The commonly held view that the exclusive or dominant trend is for tannins to clump together in longer and longer polymeric chains seems to be far wide of the mark -- yes, that ha ppens, but it's only one of a hundred fates.
Tannin structures form and re-form, and combine with all manner of other molecules. They go through oxidative reactions (some involving our friends the aldehydes) and acid-catalyzed reactions; they glom onto anthocyanins in forms that preserve color and perhaps influence mouthfeel; they are "sticky" enough to hook up with many other compounds. Some of the conglomerate forms become locked in, no longer easily broken apart; others come and go.
Researchers in this area are fond of quoting Doug Adams from the University of California, Davis, who has called the wine tannin dance a "chemical train wreck." In an e-mail, Paul Smith of AWRI called it a "freaking nightmare."
And those are just descriptions of the parts we know something about. While methods for the quantitative analysis of grape and wine tannin are improving steadily, everyone agrees that techniques for qualitative analysis of wine tannin composition are woefully inadequate.
"There is no question that our understanding of what tannins are is at a high point prior to veraison," Jim Kennedy said. "From there it is all downhill. By the time a wine is 3 years old, we can chemically define less than 50% of the isolated material that we call tannins."
Size matters for astringency
While much remains mysterious, we do at least know that the notion of a linear progression toward longer polymers isn't the defining story. And it's a good thing, too, because if that were the story, wine would get more and more "tannic" as it aged.
The feeling of astringency is produced when tannins grab the proteins from saliva, lessening the lubrication in the mouth and making tissue scratch against tissue. Research in France in 2003, conducted by Stéphane Vidal, Véronique Cheynier, Ann Noble and others, using model wines in which the tannin composition could be controlled, demonstrated through tasting panels that longer tannin chains register as more astringent than shorter ones.
This makes intuitive sense: Bigger molecules have more ways to snatch proteins than smaller ones. Or, imagine the relative mouthfeel of fine beach sand and chunky gravel. (Fans of minerality in their wines can do the actual trials; for others, this is simply a thought experiment.)
Which brings us back to seed and skin tannin. One reason that seed tannins can be more astringent than skin tannins of the same length is that, while in the grapes, seed tannin chains are sometimes terminated with gallic acid, which increases astringency. But, Harbertson said, this effect is outweighed by the substantially shorter (and thus less astringent) length of the seed tannins; and in any case, the gallic acid -- which is not itself astringent -- can easily be cleaved under wine conditions.
Where do the tannins go?
In the Standard Tannin Model, at least some of the tannin chains get so long and heavy that they fall out over time, so that older wines are not only less tannic but contain less tannin. Wrong again, apparently.
The AWRI was fortunate enough to receive a more than 50-year vertical of Coonawarra Cabernet Sauvignon a while back, and reported the results of phenolic analysis of the wines in the proceedings of a seminar of the Australian Society of Viticulture and Oenology in 2005. Turned out, as Paul Smith explained in a seminar on wine aging at the ASEV meetings, that the 50-year-old wines had tannin levels comparable to, and sometimes higher than, much more recent vintages. Whatever hoops those tannins are jumping through, they're not going away.
The sensory phenomena are undeniable: Most red wines get mellower with age. But if the tannins don't just get longer and fall out, and they wouldn't be less astringent if they were all longer, what the heck is going on? Sorry, I'm out of space.
That's one of the main things phenolics researchers are working on, sorting through the myriad changes tannins go through to parse out what they might mean in the glass. It is equally possible that the explanation of lower tannic impression in older wines has to do with something else altogether, in the same way that elevated alcohol and perhaps a pinch of residual sugar mask the massive tannin levels in many high-end California Cabernets.
While all this is getting worked out, if you're a winemaker and like the way your tannins behave, your best course of action may be to keep doing what you're dong. And if you're a tannin researcher, take comfort that the "freaking nightmare" of tannin chemistry comes with job security. "One thing that is good about being a tannin scientist," Paul Smith conceded, "is you know you've got a job for life -- if the funding keeps coming in."
Tim Patterson writes about wine and makes his own in Berkeley, Calif. Years of experience as a journalist, combined with a contrarian streak, make him interested in getting to the bottom of wine stories, casting a critical eye on conventional wisdom in the process. Contact him through firstname.lastname@example.org.