The joy of statistics

I’ve been wrestling a lot with statistics recently, trying to get them to reveal to me essential truths about the past and present of winemaking in the Cape. I’m not a great one for numbers, I confess – I’d ideally like to belong to one of those cultures that counts “one, two, many….” and leaves it at that. When, for example, I dutifully try to understand Distell production or big wine business deals (even though my heart is really with winegrowers who have only a few dozen barrels max) I start getting a headache. If I think it important to understand, I phone Michael Fridjhon, whose number system really only begins with “many”, and beg him to patiently explain it to me (and if I don’t always understand his explanation it is entirely my own fault).

But sometimes statistics work for me, especially when they’re in a graphical form, or if I can somehow visualize the pattern. Although, crucially, of course, you have to know what to do with statistics, somehow provide a framework of meaning in which they can glow with life. Take this graph below, which is my favourite. (I’ve lovingly redrawn it, in my own way, from the dourly monochrome and hard-to-read original in KWV Statistic Survey No 9: 1985, part of the invaluable series of annual publications now continued by SAWIS, an organization for whom my admiration is nearly boundless – I’ll come to my reservation later.)

This graph shows “Cultivar trends, 1960-1983”, and if you know how to look at it properly it tells, or hints at, quite a good chunk of the story of South African wine in the 20th century. Look at the splendidly simple X shape marking the decline of cinsaut and the rise of chenin blanc: they cross each other in 1968 at a little more than 20% before proceeding on their separate paths to chenin glory (well, sort of) and cinsaut ignominy.

Cinsaut was going down partly because of the decline in fortified wines, for which it was much used, as well as an increasing preference for white wine. It was never going to recover, as smarter varieties started being used for reds – you can see cabernet creeping along the bottom of the graph, just starting to climb after 1972. Palomino plantings were declining rapidly because of the collapse in the “sherry” market.

Chenin Blanc was the main beneficiary of the introduction of cool fermentation technology in the post-war years. Lieberstein was introduced in 1959 and during the 1960s became the biggest selling brand in the world. Everyone was making off-dry, fruity white wines – and making them with chenin. Quite a bit was also going into the growing brandy business too – as was colombar, also shown in this graph, steadily rising.

One interesting point (for this statistic junkie anyway) is that all the percentages in this graph (made in 1985, remember) should really all be a little higher than they are shown. That’s because until a few years ago sultana was also included in the statistics; then it was dropped, as so little of it is used for wine. So it skewed all the reported proportions. (Hanepoot is still included, as more of it is used for wine, but it is much less statistically significant anyway.)

In fact the change in what was measured made quite a difference to the statistics, and still leads to some misleading information. When people talk about the decline in the proportion of chenin in the national vineyard, they usually speak of it having once been as high as 33% (it is now down to about 18%). In fact, if you take sultana out of the picture, chenin in 1988 would have occupied a remarkable 36.4% of the total surface area of vines, acording to my own (re-)calculation!

Another little tweak that SAWIS have made in recent years has been in their reporting of the distribution of vineyards (Stellenbosch has 17% of the total hectares under vines, Robertson 15%, etc). Unfortunately, for reasons that are utterly obscure to me and anyone else I’ve asked, they still don’t report on the Wine of Origin areas as such, but lump various ones together – so that Constantia’s vineyards are included under “Stellenbosch”, Worcester includes Elgin and Hemel-en-Aarde etc.

This makes absolute nonsense of those statistics and renders them useless to surely everyone. It leads to big confusion too – in the book The Essential Guide to South African Wines, by Swart and Smit, when they are talking about Worcester (the appellation) they claim that it “boasts the largest surface area of vineyards (nearly 21 000 hectares) in the country”. Now, this is simply not true – their figure comes from the fact that Worcester (SAWIS’s statistical area) used to include Breedekloof, as well as the areas mentioned above. Now SAWIS has started listing Breedekloof separately – and suddenly “Worcester” has shrunk to under 9 000 hectares!

If only SAWIS would abandon these amalgamated areas that they inherited from the KWV statisticians, and give the figures for the WO areas, I’d regard them as just about a perfect insitution (of which there are not many in the winelands – or elsewhere in the country).

As it is, I still look forward to their next annual booklet, in which many of the previous year’s statistics are presented. Meanwhile, if you want to explore some more fascinating stuff, you’ll find plenty on the SAWIS website for starters.

Leave a Reply

Your email address will not be published. Required fields are marked *

Are you human? *