## Monday, October 8, 2018

### The number of wineries per US state

A few weeks ago I wrote about The number of wineries in the USA, looking at the changing number through time. However, I did not look specifically at the current numbers of wineries in each of the states, which is what I will do here.

The data shown here come from the U.S. Tax and Trade Bureau (Bonded wine producers count by state: 1999–2018), and are current as at 30 June 2018. Note that: "Bonded winery premises include every licensed production facility of single firms or individuals, licensed warehouses, experimental wineries and wineries with no casegoods production or fermentation capacity." It is also important to note that the number of winery premises per state is not necessarily related to either wine production or vineyard area.

Note that the vertical axis in the graph, which counts the number of wineries, uses a logarithmic scale, while the states are simply listed horizontally in decreasing order of winery numbers. As expected, California is way out in front (with 4,481), followed by Washington (1,089) and then Oregon (713). Interestingly, Texas (602 wineries) is then just ahead of New York (593), indicating that not all "winery premises" are necessarily associated with either vineyards or large wine production.

One thing that I have emphasized in this blog is that so many potentially complex patterns can often be modeled with incredibly simple mathematics. In particular, patterns are often very well fitted by what are known as Power or Exponential models, which are among the simplest available. This is what makes mathematicians fascinated by the world, and leads to their long-held belief that the creator of the universe was obviously a mathematician. (Biologists, such as myself, think instead that the creator had "an inordinate fondness for beetles".)

It is, however, important to remember that we are interested in both the way in which the data fits the model and the (possibly many) ways in which it does not fit. Both aspects tell us something, although often very different things — the proverbial glass is both half full and half empty, and we would do well to be interested in both halves.

In this case, the next graph shows you that the winery count data fit an exponential model (drawn as the line) very well (97% fit), with two exceptions. Both California and Washington do not fit the model (highlighted in pink), although most of the other states do fit very well.

In this case, the excellent fit to the model is intriguing. There are obviously many factors that could possibly influence the number of wineries that exist in any given state of the USA, and yet these all seem to "cancel each other out", leaving us with a very simple pattern. The commonness of such patterns tells us that we should not be surprised by this — it is simply the way the world works, so that complexity leads to simplicity.

On the other hand, California and Washington stand out clearly — they both have many more winery premises than the model indicates can be expected for the top two states. This is presumably a by-product of history. For example, not only has California been the leading wine producer since the Spanish missionaries first started making wine there, but it weathered Prohibition better than any other state, and was thus in a position to take advantage of the post-World War II boom in wine appreciation. "Go west, young man" has a special meaning for viticulturists, as well as movie makers!