Wine: A nose by any other name

Anthony Rose An electronic device exists to analyse wine by its aroma, but it is no match for a human's olfactory skills
We so-called wine experts might one day be rendered redundant by a machine that can spew out an extravagant Food And Drink-style tasting note. In fact, there is already a device that can "nose" a wine. At pounds 30,000 a throw, there are only four in the world, one of which sits in a laboratory in the Whitbread Technical Centre's research facility at Luton.

The Aromascan, as it's called, is the pride and joy of Dr Pete Channon and Dr Ian Ormrod, the white-coated boffins who head the facility. The two spend their days playing with gas chromatographs, mass spectrometers and other appliances designed to keep the quality of Whitbread's various products, from beer to coffee to Thresher wines, under close scrutiny.

The Aromascan looks like a cross between a coffee machine and a photocopier. Capable of analysing any liquid with an aroma, it tests wines for safety and consistency. It can sniff out aromas and grape varieties, so it can distinguish between a Macon from southern Burgundy, for instance, and a Chablis from the north. What it can't do yet is tell you exactly what a wine is or whether it's enjoying it or not.

How does it work? Take it away, Doctor Channon: "Clean air is bubbled through a wine, driving the odour over 32 semi-conducting polymers, or sensors, and a silicon chip passes a current through each of them. The polymers are made of various materials, so different aroma compounds bind to them. An electric signal comes out which gives a print-out called a neural network map." Still with me?

Batches of a wine can be measured against an original "fingerprint". If the print-out shows less than 80 per cent accuracy, the wine is referred back to the buyer. In two years of looking at Thresher's 15 best-selling wines, it's happened six times. "It's been useful, say, with the Chileans," say the doctors, "who we've come across selling one thing and sending another."

How does its performance compare with the human nose and descriptive ability? Its articulacy is limited to producing a colourful print-out of 14 or so taller or shorter lines each of which represents a flavour compound.

We tasted three wines, all from Errazuriz' Don Maximiano vineyard in Chile, then fed them to the Aromascan. A 1998 Merlot, juicy and blackcurranty with a hint of mint, appeared with tall brown and purple lines, showing it was dominant in isobutanol and iso amyl alcohols. A 1988 Carmenere, which tastes of grass and green bean, had smaller brown and purple lines but a tall orange (acetaldehyde) one. The distinguishing feature of the 1995 Sena, an oaky blend rich in cassis and spice, was a very tall orange line. So far so uninformative.

The good news for anxious wine writers is that the machine amounts to little more as yet than a fraction of a nostril. According to the doctors, who have already isolated 5,000 aromas in beer, the human nose has 50 million sensors. For the electronic wine nose to become more sophisticated, the number of sensors will have to rise dramatically, as will its analytical capacity. "I can't see the machine replacing the experts," says Dr Ormrod. "Not yet, at least"

Comments