[famous]

A well-intentioned lie! Actually I'd call it more of an approximation. And the meteorological definition (Dec-Feb) is a better approximation of the coldest quarter than is starting the season on the solstice or on the cross-quarter day, in early November.

You cite predictability and definitiveness of starting the season on the solstice, but the meteorological season is better there, too.

[rquantz]

Not actually, because in reality the coldest dates are different every year, and may change with global warming to boot. Start talking about meteorological seasons, and we can start arguing about what data to use. Does every climate region get their own definition of winter? Or do we use a one size fits all definition for an entire hemisphere? Do we go with an average of surface temperatures since they've been recorded, or do we use a running average of the last ten or twenty years? Do we use only surface temperatures, or should we take ocean temperatures into account as well?

And then you say, well, we'll just do an approximation and say Dec 1, but then you've chosen a truly arbitrary date that, again, is just as approximate to the "actual" meteorological date as Dec 22, and is off by close to the same amount.

Edit: because this post sounds pretty vehement, I just want to leaven it by saying that I recognize that all of these things are essentially arbitrary and this is not, in the end, a topic that is actually extremely important to me. I do think it's worthwhile to point out that the meteorological seasons are not as clearly better as this article makes them out to be. There are reasons for using astronomical winter, and they aren't really any worse than the reasons for using meteorological winter, as far as I can tell.