When discussing winemaking, I try to be very careful about distinguishing science from religion. What do I mean by that? The fact that yeast converts sugar to alcohol and CO2 is definitely science. The fact that we prefer to use Assmanshausen yeast at our winery is religion, especially since we’ve never done trials to prove to ourselves that we really like it best. It’s important for me to distinguish what I know as fact from what I just believe.
One problem with making wine is that you only get one shot at it each year. That means it takes a long time to get useful data. And even then, the data you get is often questionable. I hear winemakers talk about doing trials to decide what works best—yeast, barrels, whole-cluster fermentation, whatever. I’m not sure that many of these decisions are based on truly valid data. Not to say that the decisions are wrong, just that their basis is often more religion than science.
Take barrels for instance. How would you decide which cooper’s barrels work best for aging the wines from each vineyard? First, you have to believe that there really is an answer to that question. You can’t try every barrel made, so you limit the choice to some subset. Let’s say we choose four different coopers. Since you’re a small producer, you don’t have very big lots of any one wine, which makes experimenting difficult. Maybe you get enough fruit from one vineyard to make 1,000 cases of wine, so you'll use that since it’s probably the only lot big enough to get you reasonable data. One-thousand cases is 40 barrels, so if you prefer to age 50 percent of your wine in new oak barrels, your trial will consist of 20 barrels--five from each cooper. Not too bad a start, but there's trouble up ahead.
Most of us get a mix of different vine clones from a vineyard, and we like to keep those separate to evaluate the different characteristics they bring to a wine. But you probably can’t do that this year because you’ll want a consistent blend of wine in all of the new barrels. Oh well. What if one clone didn’t fare as well under this year's weather conditions and its yield was lower than normal? Let’s forget that and go on.
Just before bottling, you taste through all the barrels and decide Tonnellerie Awesome is the best. Of course, if you used five different barrels from each of the producers, the results for each barrel may have been totally different. Why? Because the truth is that no two barrels are ever completely alike—different trees, different individuals making the barrels, different fire temperatures and times during toasting, etc. But let’s forget all that as well and assume all the barrels from a producer were identical.
So what have you figured out? You now know that you preferred that set of five barrels from one cooper for that vineyard in that vintage. That’s about it. You can try to extrapolate, but is that valid? You should probably try again next year. But how many years before you’re sure? I’m sure a trend will evolve over time, but what if you get mixed results? How do you interpret that? And even if a clear winner emerges, other factors may change that would affect future results: A new clone may be added to the mix. As the vines age, the quality of the fruit will change. Arrrgggh!
And what about all the other variables you neglected? Like issues of which forest the wood came from, the level of toast on the oak, the length of the toast, whether the barrels have toasted heads or not. You also made an arbitrary decision about which barrel producers to try. Maybe one you neglected would have been best.
And what about those lots of wine that are only 10 barrels in size? How do you test different barrels on them? There are just so many variables and not enough chances to test them all.
So what’s a poor winemaker to do? You find religion. You basically go on faith.