Sniffing Out Counterfeit Wines . . . A New Method

Yeah, I get my wines news from a Sci-Fi website . . . and Zuc.

Using Atomic Bombs To Sniff Out Counterfeit Wine" onclick=";return false;

OK, I can see how they can nail the vintage. Bonus for the consumer on that one. They still cannot tell me it is DRC in the bottle vs. Hearty Burgundy.

Carlo Rossi channels Elmer Fudd.

Actually, you can do that too. I used to be an isotope geochemist. Based on the abstracts I’ve read, there an analyzable amount of Nd and Sr in a bottle of wine. I couldn’t tell you vineyard and I might need a fair bit of wine to get a large enough sample, but I’m pretty sure that with a 750ml bottle I could tell you what region a wine was from. C14 can obviously give you year within very good precision. Stable isotopes are trickier; it wouldn’t surprise me if the carbon and hydrogen isotopic composition of the H20 component of wine changed as part of the ripening process. Now if you had benchmarks, it would be fairly easy: if you knew the: 143/144Nd, 87/86Sr, [C14], 13/12C and 18/16O ratios of wines of confirmed provenance, that would be an easily identifiable isotopic fingerprint that you could use to confirm the provenance of other questionable wines. Though perhaps if someone were slick enough to fake DRC Richebourg with some lesser Richebourg, that might sneak through.

All seems plausible to me, though I wonder how much regional variation there is in CO2 makeup around the globe. As for the stable isotopes, there has long been a program in France, in face called SNIF-NMR, which compares the deuterium concentrations and distribution in ethanol to a database of information from various regions and even vineyards. They can indeed determine if the wine in the bottle is what it says it is.

Ach, you’re right, I meant D/H and 18/16O ratios.

Though I would imagine carbon ratios would work too for fingerprinting as long as you had benchmarks to go against, b/c carbon isotopes of dissolved CO2 or organics in juice almost certainly shift around as a result of biochemical and wine-making processes, and a finished wine, assuming it is well-homogenized, should have a distinctive ratio.

An interesting test (man, I wish I hadn’t left the field, b/c if I still had access to a lab and a mass spec this would be so easy to do) would be to get a few bottles of a large-production wine which is alleged to have multiple cuvees and to see if the isotopic fingerprints differ.

EDIT: The more I think about it, the more I think that radiogenic isotopes (Sr, Nd, U/Th series) is a superior method to stable isotopes (D/H, 18/16O). You’ve got a shitton of biochemical processes in wine that could potentially overprint any regional stable isotopic “fingerprint”; in contrast, the elements for radiogenics should be pretty uncontaminated. While radiogenics could be altered by certain kinds of fining (eg, addition of bentonite), in general they should be pretty robust.

I saw one of the articles prompted by a recent talk at a conference, so I asked one of my accelerator mass spec colleagues about the work. This is a very well established technique with a number of applications, and people have been writing papers about applications to wine vintage dating for a number of years. According to my colleague, the technique will be applicable to vintages between 1950 and 2010, but not so well outside those dates. As mentioned, this particular technique won’t distinguish between old DRC and equally old plonk.



Thanks for posting. Come to think of it…when was the last time we saw a national wine commercial?

Bill, if you wanted to spend enough money, yes, they could tell you that. Even without opening the bottle (though you have to draw out a bit of wine through the cork with a syringe).

There’s a bit in the Billionaire’s Vinegar where they were trying to use carbon dating to authenticate one of the “Jefferson” bottles.
The margin of error was 35 years each way.

See Al’s post above, and the original URL. The method would not be so accurate for pre-bomb wines (atomic bomb, not fruit bomb [wink.gif] ). I’m no radiocarbon dating expert, but this seems reasonable.

Yes, the problem is that the normal yearly change in carbon-14 is slow enough that you wouldn’t be able to get a precision that’s meaningul for vintage-dating if it weren’t for the pulse injected by atmospheric tests. According to my colleague, another problem is that the natural production of C-14 went through some changes probably related to the onset of the industrial era (and before the atmospheric tests). Consequently, dating samples in the 1650-1950 era is much less precise because you can get multi-valued answers. Much post 2010, the problem is that the extra C-14 from atmospheric tests is mostly gone, so it no longer helps.

Even in the post-1950 era, there is some ambiguity because the pulse of C-14 from atmospheric tests peaked in 1963. So, the simplest application of the method would have trouble distinguishing between 1961 and, something like 1965 (not sure with the exact date, the point is that it’s multi-valued).

The method would definitely have trouble measuring a vintage date like 1947 to a precision that matters, although it could distinguish a post-1950 wine passed off as a 1947. It would also have trouble detecting a fraud with a wine that was a blend of an older vintage and a somewhat younger vintage (ie, a blend concocted to spoof the technique). You could improve the technique by dating carbon from more than one component of the wine, but I suspect that might start involving a somewhat uncertain chemical model depending on the components.


Well, the problem pre-bomb is not necessarily the wiggles, since AFAIK there have been very few major changes in solar activity or ocean circulation in this century (and those are the factors that affect the atmospheric concentration of C14).

Instead, the problem is that the measurements you need to make to calculate age in the pre-bomb era require much more precision and because of that, create a larger margin of error. And that’s setting aside the calibration curve issues, which I don’t think are that important on these timescales.

As I mentioned above, the key is to couple C14 dating with other forms of isotopic analysis. Mass spec time is pretty expensive, so this is probably only justified for the most expensive of wines. IIRC our lab used to charge a couple of hundred bucks a sample for Nd and Sr data, and C14 dating is around the same price.

Pre-1950, I think there are two problems with the method for vintage dating wine. As we both mentioned, without the bomb pulse the precision of the method is too low to identify a particular vintage because the carbon-14 abundance doesn’t change enough per year. But, between 1650 and 1950 there are issues with the calibration curve that make it difficult to get an approximate vintage (ie, issues with the calibration curve makes the precision even lower than it would otherwise be).


When you say b/w 1650 and 1950: sure, you’ll run into problems with the calibration curve if you’re dating wines from the Maunder Minimum. But other than the billionaire’s vinegar and a handful of Madeiras (which to my knowledge haven’t suffered from rampant counterfeiting), who is going to need a vintage dating pre-1850?

For the most part, I think we agree that once the precision is less than a year it’s not particularly useful for vintage dating. I was pointing out some of the idiosyncracies of carbon dating.

As far as who might need a vintage date farther back than 1850, the claimed Jefferson bottles were supposedly from 1780’s. I think that is one of the few examples where a person might go to the expense of using this technique to try to date wine (and actually did).

As it turned out, a Munich lab concluded that wine from the bottle they tested contained a mixure of vintages, with nearly half the wine from 1962 or later. I don’t know the methodology they used, but it would be more difficult to date a mixture of vintages.

A later test, commissioned by the seller, concluded that none of the wine dated to 1962 or later. But it came from a different bottle, so it was not necessarily a contradiction of the earlier result.

However, to quote from a New Yorker article written about the affair, “carbon dating can’t provide a reliable determination of the age of wines bottled during the eighteenth and nineteenth centuries, and an examination of Bonani’s lab report reveals that his findings reflected a considerable margin of error. While the test might have ruled out the presence of late-twentieth-century wine, it did not provide absolute proof that the wine dated to 1787. ‘The test says only that the wine is from somewhere between 1673 and 1945,’ Bonani wrote in a recent e-mail.” (Bonani was the scientist commissioned by the seller).

A later test looked for Cesium-137 in a number of bottles, an isotope that would only appear in wines from the post atmospheric testing era. None was found, so this test was also not particularly useful since it just meant the wine was from the 1940s or earlier.

So, this is an interesting application of carbon-dating. But, like any application of scientific or forensic methods, it’s all about understanding your error analysis. A method will generally give you an answer, the real work is often to understand the reliability.


Al, been too long since I took radiochem, but it seems to me that carbon dating is pretty cut and dried, no? In a wine I assume you’re looking pretty much entirely at ethanol, there’s some fraction of C14 in there, that puts it at some point on the decay curve, and you’re done, no? If you blend an old and new wine, don’t you just get somewhere in the middle of the two on the curve?

Yeah, so how would they distinguish between a single vintage or a blend of two (or more) by carbon dating? I imagine they combined carbon dating and another technique (or tried to carbon date two components, eg using ethanol and tannins).


Not sure. I think you can probably use C14 dating to get an approximate age, within a couple of decades (this assumes natural occurrence, not post testing). I’m having trouble finding information about accuracy of carbon dating for young samples. Surely precision for younger wines (say the 20th century) can’t be anywhere near 1 year.

Actually - the correct use for any technology like this is to use it to detect TCA - that alone (ie, auction bottles that are pre-TCA screened) would be a boon to winedom.