San Luis Obispo, Calif. -- A handful of consultants in California is disputing the validity of a tannin assay developed at the University of California, Davis. Touted as a simple, quick and cost-effective means of determining tannin concentration in grape must, the Harbertson-Adams wine tannin assay is the subject of a paper published in the Journal of the Association of Official Analytical Chemists in September 2008 (read the it at Association of Official Analytical Chemists).

Citing the results of trials conducted in fall 2007 at five separate labs, the paper concludes that use of the Harbertson-Adams assay to track tannin concentration is "premature" in commercial wineries.

Developed by UC Davis professor Dr. Douglas Adams and then-student Dr. James Harbertson (now extension enologist at Washington State University), the Harbertson-Adams assay applies the 30-year-old Hagerman and Butler plant tannin assay to grapes. Showcased at the Trellis Alliance's Recent Advances in Viticulture and Enology (RAVE) forum in March 2007, it has been adopted by several wineries and at least one commercial lab in California. It was also key in a survey Harbertson published earlier this year of tannin concentrations in wines from select viticultural areas in Washington state (see Wines & Vines article).

But significant variation in results from trials at five labs led consulting winemaker Larry Brooks of LM Brooks Consulting in San Luis Obispo, Calif. and his co-authors--Leo McCloskey, principal of Enologix Inc., Enologix laboratory and customer relations manager Doug McKesson, and Dr. Marshall Sylvan, former head of the mathematics department at the University of California, Santa Cruz and a consultant to Enologix--to term the Harbertson-Adams assay invalid.

"I was surprised that the range of variation was as great as it was," Brooks told Wines & Vines. Tests on nine bottles of three varieties of wine, purchased from local grocers, exceeded statistical norms of ±5% variation.

While the industry is seeking a faster, more cost-effective method of assaying wine tannin, which is a significant element in qualitative evaluations of wines and hence ratings, Brooks believes precipitation assays such as the Harbertson-Adams method are less precise than the high-performance liquid chromatography (HPLC) method his own firm uses. "I've given up on this type of assay for tannin, both this protein-based one, (the) Adams method, and the methyl cellulose one as well. They have too much variation to have any validity," he said.

The trials were conducted by Windsor, Calif.-based Vinquiry Inc., a commercial lab that provides assay services for a charge, and by "trained analysts from four independent enological laboratories" serving four California wineries. (Enologix, which does analyses as part of its enological consulting work, was not one of the labs.)

Vinquiry declined comment on the paper and will let UC Davis lead the response to the findings.

But Dr. Steve Price, principal of Price Research Services Inc. in Corvallis, Ore., and consultant to ETS Laboratories in St. Helena, Calif., (which has its own proprietary HPLC assay method), said the variation Brooks encountered is common in production laboratories. It's particularly true of the Harbertson-Adams assay, which relies heavily on the assay operator.

"The actual physical procedure itself, is influenced by the operator and has to be carefully controlled," Price said. "If the laboratories aren't running a control, something that they can test, something that they know, they may have difficulty measuring and monitoring their own precision."

The trials Brooks commissioned merely highlights this point, he said. The degree of variation, Price said, "means the precision is poor on the assay, as run by the production laboratories."

In highlighting this, he said Brooks' paper does the industry a service.

"It points out that you'd better be aware of what you're doing in your laboratory and watch precision, and understand the assay within the context of your own laboratory," he said.

He doesn't believe production labs typically maintain controls, however. Indeed, this wasn't something Brooks did. "You cannot control the variation. That's the very issue," Brooks said, arguing that HPLC's precision makes it a de facto control. "We didn't make up model solutions where we knew the absolute quantities (of tannin) that were in each. I would say the tannin chemistry, the analytical chemistry, is not to that point, as it's practiced commercially or academically."

Price doesn't believe a comparison between two methods makes sense, however, and overshadows the variation in assay results the paper discusses. "I think that the bigger issue is that the paper is flawed in that it's comparing apples and oranges to begin with," he said, noting that the Harbertson-Adams assay measures tannin reactivity to proteins, while the HPLC method looks at molecular structure.

A draft version of the paper circulated in late 2007 garnered criticism from the Trellis Alliance, which backed the work of Adams and Harbertson and argued that assay results were indeed reproducible, citing the experience of the several speakers at RAVE 2007 who use the assay. Trellis Alliance president Kay Bogart referred questions concerning the conclusions of the published paper to Adams, who declined any immediate comment on Brooks' paper.