Suspect corks & the 1998 Vintage

[n4sir]

As I've mentioned a couple of times, this year I seem to be having a rotten run with corks - the last couple of months it's been about every 2nd or 3rd bottle, and it's starting to really cheese me off.

Now I know on the law of averages I was probably due to start striking a few as I've had a remarkably good run over the years, but was I just imagining that the majority of the bad wines were from 1998?

I quickly checked my records and totalled the 1998 wines tried for the year - of 69 wines, 16 were definitely affected by TCA or random oxidization, a 23% failure rate (or every 4-5th bottle). There were a few other suspect bottles too, so the rate could be even worse.

There has been a fair amount of criticism of the 1998 vintage being over-hyped the last few years, but has anyone else noticed an increased rate of cork faults for this specific year?

Cheers,
Ian

[Red Bigot]

There has been a fair amount of criticism of the 1998 vintage being over-hyped the last few years, but has anyone else noticed an increased rate of cork faults for this specific year?

Cheers,
Ian

Ian, I haven't noticed any particular more extreme problems with cork in my 98 vintage reds, but Ric keeps pretty close tabs on faulty wines so may have some more accurate info from his experience.

A fault rate of 23% is awfully high, my cork-related faults hover around 7% I think, although it seems higher at times when the laws of randomness are not being kind. Any particular regions, red/white, size of producer, particular producers?

[n4sir]

Reds, mainly SA/big producers and quite a few Coonawarra - that's what I was buying at the time.

I've still got quite a few left - hopefully as you suggested Brian this is a bad patch and I don't strike one for a while, but it's still a worry.

Cheers,
Ian

[TORB]

Very interesting!

I just did the analysis and out of 413 bottles of 1998 wine opened, 38 or 9.2% had cork related problems, a further 2.9% had other problems, and .7% were suspect. That's a total of 12.8 %.

The overall average of cork related problems 2,708 tracked bottles show 7.1% cork problems, 2.0% other faults and 1% possible - total 10.1%

So 98 looks like a problem! But it may have to do with the sample size. Time will tell.

[Ian S]

If only I still recalled my uni degree (in probability and statistics ), I might be able to throw a little analysis into the pot.

At least the sample sizes are pretty big. Some issues re: biasing, in particular if the stats includes 24 Lovedale Semillons, but no HVD semillons, then cross producer variations might influence the results. The bigger and more varied the sample size, the safer the approximations become. Whilst this involves human judgement, it's at least experienced palates which helps. Variations between palate perception of faults would also be a factor, but as long as each taster has a similar spread of vintages, then the data could be pooled easily (and the bigger the pool, the safer the approximations and the more definitive the findings).

Potential for analysis if anyone knows any budding statisticians they could pass the info to?

regards

Ian

[Finney]

I agree with Ric here, we need larger sample sizes and well kept records before we can come to any conclusions here.

Regards


Finney (Craig)

[Roddy]

Very interesting!

I just did the analysis and out of 413 bottles of 1998 wine opened, 38 or 9.2% had cork related problems, a further 2.9% had other problems, and .7% were suspect. That's a total of 12.8 %.

The overall average of cork related problems 2,708 tracked bottles show 7.1% cork problems, 2.0% other faults and 1% possible - total 10.1%

So 98 looks like a problem! But it may have to do with the sample size. Time will tell.

The sample size of 413 at 95% confidence and a 10% fault rate gives a rough margin of error of +/- 3%. So the figures are within statistical variance. Just.

Still, there's lies and then there's statistics. I wager a larger sample of say 800 would give a definitive answer (+/-2%). If we still saw the same fault rates with that sized sample, it would probably indicate a 1998 problem.

[Davo]

Very interesting!

I just did the analysis and out of 413 bottles of 1998 wine opened, 38 or 9.2% had cork related problems, a further 2.9% had other problems, and .7% were suspect. That's a total of 12.8 %.

The overall average of cork related problems 2,708 tracked bottles show 7.1% cork problems, 2.0% other faults and 1% possible - total 10.1%

So 98 looks like a problem! But it may have to do with the sample size. Time will tell.

The sample size of 413 at 95% confidence and a 10% fault rate gives a rough margin of error of +/- 3%. So the figures are within statistical variance. Just.

Still, there's lies and then there's statistics. I wager a larger sample of say 800 would give a definitive answer (+/-2%). If we still saw the same fault rates with that sized sample, it would probably indicate a 1998 problem.

I'm a 1500 man myself.

[Ian S]

Very interesting!

I just did the analysis and out of 413 bottles of 1998 wine opened, 38 or 9.2% had cork related problems, a further 2.9% had other problems, and .7% were suspect. That's a total of 12.8 %.

The overall average of cork related problems 2,708 tracked bottles show 7.1% cork problems, 2.0% other faults and 1% possible - total 10.1%

So 98 looks like a problem! But it may have to do with the sample size. Time will tell.

The sample size of 413 at 95% confidence and a 10% fault rate gives a rough margin of error of +/- 3%. So the figures are within statistical variance. Just.

Still, there's lies and then there's statistics. I wager a larger sample of say 800 would give a definitive answer (+/-2%). If we still saw the same fault rates with that sized sample, it would probably indicate a 1998 problem.

Roddy
I'm presuming you've approximated to a normal distribution (which as the numbers increase, so does the validity of the approximation). It is important to recognise if there's any non-random variation that could influence the results. Perhaps TORB bought more Barossan wine that year, but more from Marg River in 1999. Maybe some years/styles even show taint more obviously than others - certainly if we were talking about aromatic whites then taint/cork faults are arguably more noticeable.

In the old statistical terminology we'd describe something around 5-10% as "some evidence" to support the hypothesis. Below 5% became strong evidence and around 1% became very strong evidence.

However one concern is that Ian's stats are slightly tainted (sorry ) by the fact that he's noticed the difference and then put the hypothesis forwards. In general it's best to run the test blind and interpret the results, rather than see the results and try and test conclusions (politicians would argue with me there!).

regards

Ian

[n4sir]

My original post was intended to be a bit of an open question rather than a hypothesis as such, although the numbers do seem to confirm I'm having a particularly horrible run with the 1998 vintage this year.

I'm not sure if TORB's 1998 bottles are just for this year or his total consumption, but for arguments sake if you add our figures together it comes to 54 definite cork related problems out of 482 bottles - an 11.2% failure rate which looks marginally more realistic, even if it's still on the very high side.

I was actually hoping to see if a winemaker/retailer had any definite statistics or comments of their own to add. They would have access to far wider/accurate statistical data and would have a much better idea of the actual failure rate. The 1998 vintage has been recently criticized by Max Allen among others as being over-hyped and inconsistent, and this may be just one contributing factor that's skewing this opinion if it just happens to have any substance.

It's presumptuous to draw any conclusion from these figures, but it's also a bit of a coincidence that not long after this vintage the push for alternative closures really took off because the failure rate was deemed to be too much.

Cheers,
Ian

[Roddy]

Very interesting!

I just did the analysis and out of 413 bottles of 1998 wine opened, 38 or 9.2% had cork related problems, a further 2.9% had other problems, and .7% were suspect. That's a total of 12.8 %.

The overall average of cork related problems 2,708 tracked bottles show 7.1% cork problems, 2.0% other faults and 1% possible - total 10.1%

So 98 looks like a problem! But it may have to do with the sample size. Time will tell.

The sample size of 413 at 95% confidence and a 10% fault rate gives a rough margin of error of +/- 3%. So the figures are within statistical variance. Just.

Still, there's lies and then there's statistics. I wager a larger sample of say 800 would give a definitive answer (+/-2%). If we still saw the same fault rates with that sized sample, it would probably indicate a 1998 problem.

Roddy
I'm presuming you've approximated to a normal distribution
regards

Ian

I did indeed. Anything else was just too difficult at that time of the evening.

[TORB]

Interesting topic!

Some years ago I had a suspicion that either I was getting much better at picking cork problems and/or the corks in Oz wine were getting worse. That was why I started to kee records of all bottles opened.

This is the first time I have done an analysis like this and I was surprised in some ways at the result.

My numbers go back to June 2001 when I started keeping track.

[Craig(NZ)]

You simpleton 2dp people...

According to the deluxe version release 56 of the kiwiwinefanclub cellar and wine experience database, I have calculated that 45.341% of wine freaks over analise their wine experience, 68.568% of those being australians, 42.875% of them are overweight, and 25.974 have way too many of those dodgy 1998s

But seriously now, ive had a pretty good run with my wines. In the last 10 years I honestly think ive only had 5 so wines that are worth over $30 that have been faulty - plenty of cheap supermarket wines but surprisingly few flash ones

[TORB]

Y In the last 10 years I honestly think ive only had 5 so wines that are worth over $30 that have been faulty - plenty of cheap supermarket wines but surprisingly few flash ones

I didn't know they sold cask wine over $30, even if they are K1W1 $$.

[Kieran]

It's a bit sad if 1998 was a great vintage for wine but a bad vintage for corks.

Bah. Bring on the stelvins.

Kieran

[Ian S]

Roddy
I'm presuming you've approximated to a normal distribution
regards

Ian

I did indeed. Anything else was just too difficult at that time of the evening.

Hey it was beyond me, so no complaints from here.

What it does demonstrate is that there is indeed potential that '98 did have worse corks - or perhaps that combined with '98 bottles not living up to expectations (as per Max Allens comments). With results that are based on judgement, the answers can be harder to interpret. A wider survey would be good, but I doubt many keep stats in such a way. Also not everyone has Ric's finely tuned palate.

[707]

Interesting question Ian and one I'm concerned with being the owner of a substantial number of wines from this vintage.

In my absence Blacktongues did four 1998 tastings this year and the number of TCA and oxidised wines was substantial.

I've just put together the 2007 tasting schedule for Blacktongues and there are another three 1998 tastings planned that advance the best wines tried this year into a different line up of rivals.

Will we get another batch of faulty wines in 2007?

Over summer I'll be drinking alot of 1998s to bring the number of bottles down to a more sensible cellar balance so I'll carefully watch the fault rate.