"Cork is cleaner and more reliable today than ever before and we are making real progress in defeating TCA altogether as a source of wine taint," Dr Cabral told delegates attending a workshop at the recent Australian Wine Industry Technical Conference in Melbourne.

Chaired by Dr Mark Sefton of the Australian Wine Research Institute (AWRI), the workshop examined the latest developments in cork processing technology and included presentations by scientists from several cork producers.

Among the audience were senior representatives of some of the world's leading wine companies.

AWRI's manager industry services Peter Godden, who coordinated the conference's extensive workshop program, described the cork workshop as "a huge success".

"We have had very positive comments regarding the calibre of the presenters and the quality of the science presented," he said.

Dr Cabral said tens of millions of dollars invested by major cork producers had been driving change across the industry.

Industry-wide initiatives include a code of good practice and an accreditation system. Over 360 companies from major cork producing countries were accredited in 2003.

In quality control there has been a shift away from sensory analysis to standardised procedures using GC-MS (gas chromatography and mass spectrometry) analysis.

At the company level, new TCA extraction processes such as Amorim's ROSA treatment reduce releasable TCA in corks by 70 to 80 per cent.

New boiling processes, such as the Amorim CONVEX system, prevent cross-contamination during boiling and produce cleaner drier cork that is much less prone to TCA formation.

Dr Cabral said research attention was now turning to the origin of TCA in cork forests and cork's role as a wine seal.

Transcripts of the Amorim presentations and panel discussion from the workshop are available at www.corkfacts.com/presentn.htm.

The trial evaluated wines bottled in September 2002 under various closures and included two reference cork stoppers.

After 18 months, both the reference 2 and reference 3 corks were rated zero for TCA aroma on a scale of zero to nine.

"The overall incidence of TCA taint identified in wine sealed with (the cork-based) and the reference cork closures is considered to be very low up to eighteen months post bottling compared to that identified in the AWRI's previous closure trial which was bottled in May 1999," said the authors of a report in the August 2004 edition of the AWRI's Technical Review.

The cork sampling protocol, known as acceptance sampling, is based on a method first used by the US military to test bullets during World War II. Testing each bullet in advance was impractical as it would have left no bullets to use, but testing no bullets at all could have had disastrous consequences.

Similarly, testing every wine stopper would be logistically and economically impossible.

To deal with the issue, the Cork Quality Council and ETS Laboratories in California came up with a statistically valid and scientifically proven method based on testing 50 corks in 10,000.

Corks are typically delivered in large bags or bales of 10,000 corks of the same grade from the same wash. The corks in a bale show similar characteristics.

The testing method measures 'releasable TCA', which is the amount of TCA (2,4,6 trichloroanisole) that the corks release into a soak solution. Releasable TCA is a good predictor of TCA in bottled wine.

Each group of 50 corks is soaked for 24 hours in white wine with 10 per cent alcohol. The laboratory techniques used to chemically analyse the group soak solutions can reliably measure TCA at levels as low as one part in a trillion - well below sensory thresholds for almost all wine consumers.

Trials at ETS Laboratories found that the results from group soaks accurately reflect the average of results gained by soaking the same corks individually.

In cases where a few corks had very high TCA levels, the group results were higher than the average of the individual levels indicating that the group score is skewed in favour of detecting particularly bad corks.

Increasing the sample size to 100 corks would give a more detailed TCA picture, but it would not significantly improve the identification of unacceptable bales. ETS comparisons of 50-cork and 100-cork samples showed that the smaller sample accurately predicted the 'pass/fail' rate of the larger sample.