Tracing down tobacco aromas in aged brandies

I’ve become so accustomed to fermented beverages having flavors completely unrelated to their starting ingredients that I all too easily forget that there’s anything strange about it. And then I introduce someone to wine tasting for the first time and find myself having to explain that, yes, the wine tastes like cherries and no, they don’t put cherries in the wine to get it to taste that way. (But also sometimes that, yes, the wine tastes like oak, and yes, they do put oak in the wine, or wine in the oak, to make that happen.)

Wine and aged brandies, like Cognac, sometimes taste like tobacco. No one adds tobacco to them.* What happens instead is an example of what you could liken to chemical convergent evolution. Brandies can indeed contain odiferous molecules also characteristic of tobacco flavors. But those molecules appear to come from the breakdown of compounds present in grapes, not from any exposure to tobacco (and, needless to say, the tabanones in tobacco don’t come from grape-derived compounds).

Megastigmatrienone is more conveniently called tabanone. Or, more properly, megastigmatrienones are are more conveniently called tabanones, because the designations apply to a group of similar molecules. They seem to show up in wine because carotenoids – the highly pigmented class of antioxidant molecules you’ll recognize from it’s best-known carrot-colored member, beta-carotene – can become megastigmatrienones under appropriately acidic conditions. The precise pathway from grape-derived carotenoids to tobacco aromas remains something of a mystery. It seems that brandies, distilled from grape wine, end up with concentrated amounts of tabanones or precursors which then have an extra chance to develop into tabanones during barrel aging.

That mystery makes all the more interesting a recent study to quantify tobacco aroma molecules in Cognac and Armagnac**. Most of the (many) examples they tested contained some tabanones. Concentrations varied widely, but the Armagnacs averaged higher than the Cognacs and – evidently to everyone’s surprise – the aged rums thrown in for comparison ranked even higher than the Armagnacs. Much higher, in fact, even though the molasses from which rum is distilled doesn’t contain significant amounts either of tabanones or of the identified precursor molecules present in grapes.

So, why is this interesting?

  1. Tobacco aromas in aged spirits are evidently not just coming from molecules originally present in grapes.
  2. The oak barrels used for aging rum, as well as various brandies, probably contributes substantially to tobacco aromas in those spirits by leaching precursor molecules out of the wood and into the spirit and providing favorable conditions for their development.
  3. The researchers behind this study seem to think that their new-and-improved means for quantifying particular variants of tabanones could be useful for distinguishing Cognac from Armagnac, presumably to defend against passing one kind of spirit off as another. Looking at the enormous variation in tabanones between one kind of Cognac and another, or one kind of Armagnac and another, this suggestion seems frankly ridiculous. Chemistry backs up tasting experience: individual bottlings vary dramatically in the presence or absence of tobacco aromas as a defining characteristic. Yes, Armagnacs beat out Cognacs on average, but averages don’t help distinguish individuals when individual variation is so high. A more interesting use of this research, I think, is the starting point it provides for thinking about how barrel-aging adds to tobacco aromas, in spirits and possibly in wines. I suspect that the people who produce these spirits already have a lot to say on that point. Joining their knowledge with some well-placed chemical analyses could improve everyone’s understanding of how tobacco aromas happen and how to manipulate them.


* Ewwww.

**Camper English has drawn up a helpful comparison between Cognac and Armagnac at

Waste not? Capturing every last bit of aroma potential from fermentation

The aroma of baking brownies arouses two categories of responses. Summary of Category One: “Wow, that smells delicious!” Summary of Category Two: “Oh no, just smell all of that chocolate flavor escaping into the air that won’t be in the finished brownies any more.” This new research is for people who empathize with category two.

As Saccharomyces cerevisiae ferment grape sugars into ethanol, they produce massive amounts of carbon dioxide. Most of that CO2 gas escapes from the top of the fermenting liquid (small amounts remain dissolved in the wine) and, with it, escape wine aroma molecules. One of the two defining features of an aroma molecule is that it’s volatile, which is to say that it evaporates readily; you can’t smell a molecule if that molecule can’t make its way into your nose. (The other defining feature of aroma molecules is their ability to activate sensory receptors; even after a molecule makes its way up your nostrils, it still has to trigger a sensorineural response.) Volatile molecules will escape from a stationary liquid to some degree, but they escape a whole lot faster when the liquid is moving (hence our habitual practice of swirl, then sniff), and even faster still when bubbles forming inside the liquid carry those volatile molecules up from the inside out.

In short, the logic of physics and chemistry have it that a lot of aromatic molecules are lost to the air during fermentation. In theory, those molecules could be captured and deposited back into the wine after fermentation to yield a more aromatic final product.

This idea occurred to a bunch of Italian food scientists, who gave it a try with two sangioveses and a syrah and have published the results in the American Journal of Viticulture and Enology. It seems to have worked pretty well. They attached a condenser to the top of their stainless steel fermentation tanks and collected the vapor rising from the tanks as a liquid that they could then dose back into the wine after fermentation was complete. Which is, of course, precisely what they did. A sensory panel was universally able to identify which wines had received more and less condensate and which had received none at all though, to the study’s detriment, only eight test-sniffers* were involved and they weren’t asked to report on what they thought about the wine’s aromas or on how the wines tasted. 

There’s a certain appealing efficiency to the idea of capturing a heretofore lost byproduct of fermentation – all of those lovely aromas of fermentation, plus a healthy dose of alcohol (the liquid condensates contained about 24% ethanol) – and doing something with them. The condenser itself is a simple apparatus consisting, essentially, of a heat exchanger. Adding the condensate back to the wine is an obvious use for the stuff, but it’s certainly not the only one, and so in effect this is an experiment in increasing our efficient, sustainable use of available resources, like turning vine prunings into biochar or using grapeseed flour as a nutritional supplement. Though, the folks with Category One brownie aroma responses might have a different perspective.


*Eight test-sniffers who were described only as “panelists,” and who were therefore most likely  happened to be around the lab when testing needed to happen. Since the panelists’ job here was limited to answering the questions, “which of these wines doesn’t smell like the others?” and “which smells most intense?” it’s not frightfully important whether they were trained sensory testing experts or wine professionals or ordinary-Jane wine consumers. If and when this idea is taken into “which of these wines is most appealing to you?” territory, someone will need to decide whether they care about the responses of wine experts, random occasional drinkers, or both.

Playing the “what’s that smell” game with aged red Bordeaux

One of wine chemistry’s best parlor tricks is the “what’s that smell” game. Why does your wine smell like raspberries, or turpentine, or mint? If there’s not yet a chemical explanation for how it got there, we can often at least point to a molecule. And as various plays on molecular gastronomy have made clear, pointing to the molecule can be a lot of fun. We can ask where else that molecule pops up and devise creative food-wine combinations, or just marvel at the envirochemical systems that make compounds produced by South African trees pop up in aged red Bordeaux.

Speaking of South African trees and aged red Bordeaux, recent wine chemistry research has identified a molecule best known from the former as the explanation for minty aromas in the latter. A group of (no surprise) Bordeaux wine chemists have concluded that the classic minty-ness some red Bordeaux acquire* can be attached to piperitone, a molecule previously undiscovered in wine but, the European Bioinformatics Institute tells me, is commonly extracted from South African eucalyptus trees.

Along the way, the authors make a second and perhaps more interesting point about how chemists go about pairing up molecules and aromas. A typical method involves gas chromatography-olfactometry: the gas chromatograph separates out the components in a (vaporized) sample, and the olfactometer lets you systematically sniff the components that come out of the chromatograph. The excellent thing about this method is that it lets an experimenter identify the “odor-active” components of a sample. The downside is that you’re smelling and studying each of those components individually rather than investigating how they interact or what difference they make to a wine’s total aroma. Even if the conventional tasting note gives individual aromas as though the OWP** is smelling down a list, smelling is more complex than that. The wine-sniffing nose receives a whole pile of odor-active molecules at once, the brain processes them together, and then our memory and language processing functions arrive at descriptors that probably aren’t one-to-one matches for the molecules that came in at the front.*** At least in theory, a molecule may smell minty without contributing to how we perceive mint in a wine, or vice-versa.

An alternative this article touts is “aromatic reconstitution:” a complex sample (vaporized wine, for example) is separated out into odor-active “fractions” using that gas chromatograph, and the fractions are recombined in their original proportions but with one missing. Meanwhile, other wines naturally lacking the smell-of-interest are spiked with the most interesting odor-active fractions. The recombinations and spiked samples are smelled, and results are tabulated for what smells appear or disappear in association with what fractions. Researchers can then go back to the fractions that matter the most to the smell-of-interest and work out what the heck, molecularly speaking, they’re dealing with. Much to their credit, in this study local Bordeaux wine professionals were pulled in to do the smelling.

All of that (and a lot more, in this very thorough study) let the researchers say that piperitone contributes to the minty aroma of aged red Bordeaux, a much stronger conclusion than saying that piperitone is found in aged red Bordeaux and piperitone smells like mint without directly connecting the two. The piperitone-mint link might have useful consequences of some kind for people with wine aging problems or chefs creating gently eucalyptus-scented lamb roasts, but it’s also one more step toward working out the moving parts of both smell and wine aging. And eventually, that’s going to be more than a parlor trick.

*Including, in this study, a 1998 Pomerol. I mean, it probably wasn’t Petrus, but this still sounds like the researchers enjoyed themselves.

**Obnoxious Wine Professional

***How the brain processes smell is a good deal more complex than I’m making it out to be here, and neuro-sensory-scientists are still working out a lot of the details. Modern science tends to be pretty good at teasing out how representational processes work – how we see, for example – but seems to make a lot less progress with affective processes involving emotion and non-linear, extra-logical processes.