Comments on: Harbertson-Adams assay debate.

By: Winemaker29

Winemaker29 — Tue, 25 Aug 2009 03:50:11 +0000

Naturally we assume that all science is good. The goal of science, like journalism, is about shedding light where there is darkness. Yet, science and journalism can be wrong quite often when no one is chekcing the scientists and journalists. Just because a scientist produces science does not guarantee the science works in the hands of trained scientists. Brooks et al, highly experienced winemakers, could not replicate the science. Thanks.

By: Brad Kitson

Brad Kitson — Tue, 10 Mar 2009 08:04:10 +0000

Larry,
The Intralab (Table 4) data:
(1) Indicates this study had operator problems and
(2) The problems were most likely related to sample recovery technique.

The authors of the paper missed these two very important points. You should have dug deeper when you saw the Table 4 data. There’s a source of error not explained by limitations of the assay. It’s in your table 1-3 data, too. These tables are full of the imprecise commercial lab data.

In tables 1-3, at a minimum, all data from the commercial lab should be ignored and all analysis using the commercial lab data should be ignored.

I question whether the authors are qualified to comment on what is a useful assay for a winemaker. More importantly, I question the conclusions of this paper. I’ve already shown operator issues. When operator issues are involved, its very hard to rationalize any other problem. When the authors don’t recognize it, it’s even harder to rationalize any other problem.

Your previous comments calling the assay garbage were rude, mean and unwarranted.

Brad

By: Larry Brooks

Larry Brooks — Mon, 09 Mar 2009 21:39:12 +0000

Dear Mr. Kitson,
I strongly suggest that you read the paper more carefully before attacking it’s conclusions. The tables 1,2, and 3 deal with interlaboratory results and as I said they show that the commercial lab was if anything less precise than the winery labs despite performing this assay many more times. Table four as is clearly stated on page 1092 of the paper addresses intralaboratory precision. To qoute, “The commercial laboratory served to measure intralaboratory precision by performing analyses on the same wines on 2 different occasions.” There were issues with all labs on yeild and some labs with precision. None of this is acceptable performance for an assay as far as I’m concerned.
I’d also like to remind you that using CAPITAL letters in communication of this sort is the text equivalent of yelling. In the interests of civility why don’t you tone it down?
Larry Brooks

By: Brad Kitson

Brad Kitson — Mon, 09 Mar 2009 20:00:17 +0000

Larry, Would it be possible to stick to the paper and, more specifically, Table 4? What exactly did I misunderstand? Are you saying the results for Pinot#2, Cab#2 and Merlot#2 aren’t precise, in terms of CV?

Are you saying Pinot noir #1 and Pinot noir #2 are not the same wine? (I assumed they must be because of the data was used to compare recovery and this comparison would be meaningless between different wines.)

Are you saying formula 2b doesn’t have first result divided by second result? That implied sequential to me.

The analysis put forth in the paper in regards to Table 4 missed the point completely. It compares recovery between precise data and imprecise data. IT”S PROBABLY A GOOD THING THAT PRECISE DATA DIDN’T COMPARE WELL WITH IMPRECISE DATA! Though, the paper says it’s a bad thing. The IMPORTANT analysis is why is the SAME lab on the SAME wine consistently imprecise then consistently very precise?

My conclusion was learning. How do you explain the 5x improved precision between Pinot noir #1 and Pinot noir #2? And, how do you explain similar improved precision for Merlot #2 and Cabernet #2? If not by learning, then how?

Brad

By: Larry Brooks

Larry Brooks — Mon, 09 Mar 2009 17:27:06 +0000

Brad, With all due respect I don’t believe that you’ve understood the data presented. If as you state that the cv data are the most important you will see in tables 1,2,&3 that indeed the commercial lab had higher cv than the winery labs involved in this study. Table 4 does not show sequential analyses and improvement in technique by the commercial lab. This accredited lab was already using this assay and selling it to its customers and had done hundreds if not thousands of these analyses. What it shows is unacceptable variation in recovery and other analytical aspects of two different wines sent in multiple samples.
You have a good point about the “standard”. There was no standard used in the original presentation of this assay by Adams and Harbertson. Our validity study was focused on variation between wines. This does not require a “standard”. The variation is unacceptable with or without a standard.

By: Brad Kitson

Brad Kitson — Fri, 06 Mar 2009 19:34:17 +0000

I appreciate that the study was done. I did a “quick” review of the paper itself.

The most useful data in the paper are the CV numbers. CV is a very good measure of precision. In Table 4, the pinot noir no.1 CV is 14.1%. That means there’s 14.1% variation in testing this wine. Not so good. When at a later date the lab tested this wine again (Pinot noir #2), they got CV numbers of 2.8%. Wow, that’s pretty good! If you look at the CVs for the Merlot and Cab, the same trend of better precision was observed. What we are seeing here is a CLEAR indication of the lab getting better at the analysis.

A further look at Table 4 shows a trend of higher recovery for all three of the wines. This a CLEAR indication that as the lab got more precise, the lab also got more accurate.

My conclusions from the DATA are that one commercial winery has learned to run this analysis with much better than AOAC-approved precision. It appears that the commercial lab also got more accurate. How accurate can’t be determined from the data. The study lacked a standard, as specified by AOAC guidelines, to measure accuracy. The conclusions contained in the paper are NOT supported by the DATA.

Larry’s last post doesn’t agree with the DATA. The commercial lab’s results were by far the best after they learned the analysis.

Brad Kitson
Assistant Winemaker
Farella-Park Vineyards

By: Larry Brooks

Larry Brooks — Sat, 28 Feb 2009 21:19:17 +0000

Linus,
Actually, Vinquiry which participated in this study is an accredited lab. They were sent the same blind samples as the winery labs. The precision of this accredited lab was no better than the winery labs so I think we can concluse that the winery labs performed as well as the accredited lab. It is also worth noting that this assay was released and promoted by UCD specifically as a winery lab analysis. We were testing it in the environment that it was supposedly designed for. I would stick with my original conclusion, garbage analysis, garbage results.
Larry

By: Linus

Linus — Sat, 28 Feb 2009 00:16:27 +0000

I know it late to add my two cents but I just stubbled on this thread and there is something I would like to say as a reader of the Brooks JAOAC 91(5) paper and a soon-to-be-former AOAC member (having declined to renew my subscription after this gross disappointment in their publishing standards): this paper, is not a validation study at all. The laboratories used were not all certified labs,which is fine but in the absence of that condition (non standard conditions) other mention of equipment calibration and uniform training should have been given (which it was not). I equate this excercise to someone exclaiming “Clocks found useless,” after asking multiple people the time and getting different answers. I’m sure the statistics were preformed to utter mathematical precision but its a case of: garbage in = garbage out, as the study design is embarassingly bad.

By: wine sooth » Blog Archive » Electrifying wine.

wine sooth » Blog Archive » Electrifying wine. — Fri, 19 Dec 2008 18:18:16 +0000

[...] or ionization. The polyphenols in the wine might then react with those metal ions. After all, the recently debated Harbertson_Adams tannins assay takes advantage of the fero-reactive tnature of some polyphenols to precipitate [...]

By: Leo McCloskey

Leo McCloskey — Fri, 19 Dec 2008 04:12:33 +0000

We scientists can thank or discredit winemakers for reporting on our work.

University of California’s chair of Department of Viticulture & Enology discredits Brooks et al’s publication in the December 18 letter.

I recommend thanking Brooks’, afterall his work was published in the prestigious J Assoc of Official Analytical Chemists, which required high precision research, and peer review processes. When was the last time any of us published in this journal?

Thanks Brooks, too, for using commercial laboratories accredited by the International Standards Organization (ISO). The chair reports the winery laboratories were not accredited. The fact is the chair does not know whether the laboratories are accredited or not. Two commercial laboratories accredited by ISO were used by Brooks, one for finding the error within the Adams-Harbertson and another for finding the error between HPLC and the Adams-Harbertson. When was the last time the American Society of Enology and Viticulture performed a validity study?

Winemakers such as Brooks et al could get the impression that there is a culture of the cover-up in California if we do not thank them whenever they want to report from the vineyard or and winery. Winemakers should be encouraged to report problems they find, whether they are as innocuous as co-pigmentation and tannin or as econmically important as black-goo, Primitivo, rootstocks, and TCA.

Thanks Larry.