Changing the seasons

In the northern hemisphere, the official seasons are as follows: winter goes from the winter solstice (December 21) to March 19 (the day before the spring equinox on March 20), spring from March 20 until June 19 (the day before the summer solstice on June 20), summer from June 20 until September 21 (the day before the fall equinox on September 22), and then it is fall from September 22 until December 20. (The dates given for the equinoxes and solstices are for 2012 and can vary by a day or so from year to year.)

The problem is that this labeling of seasons is somewhat arbitrary and does not have a solid empirical basis for it and neither does it correspond to our intuitive sense of winter as the coldest time of the year and summer as the warmest.

It would surely make more sense to define winter as ranging over the 91 coldest days of the year, which would make it from December 6 to March 6, with the mid-point being January 20, the coldest day.

Similarly, the hottest day is July 22, so summer should be 91 days centered on that day, i.e., from June 7 to September 5.

Then spring would be from March 7 to June 6 and fall would be from September 6 to December 5.

I think this would make a lot more sense. After all, people feel that winter has arrived long before December 21. It would also make the beginning of the fall season align more closely with our sense that the Labor day holiday in the US (the first Monday in September) and going back to school marks the end of summer.

But I think that it is unlikely that this system will be adopted because the power of tradition is just too strong. It will be ignored like my other ideas to improve the world, such as that shoes should be sold singly and not in pairs and that the American Thanksgiving holiday should be shifted to the last Thursday in October, closer to the one in Canada.


  1. niftyatheist says

    Thank you!! I have long thought exactly the same thing about the seasons -- and although I hadn’t thought about it long enough to come up with new dates, the ones you propose here are precisely the ones I would have chosen if I had. The points about Labor Day and the onset of winter coinciding better with your new dates, in particular, are things I have definitely thought on more than one occasion.

    Mano, you would have my support if only your idea could gain traction in the world! 😀

    Love your blog, by the way. It is fast becoming my first go-to on FTB.

  2. Desert Son, OM says

    I heartily endorse this initiative. As a rider, I propose we also change measurement standards in the U.S. officially to the metric system, and join basically the entire rest of the planet. Road signs go from miles to kilometers. Food labels go from ounces/pounds/fluid ounces to grams/kilograms/liters, etc. I’m 185.5 cm tall and have approximately 88.5 kg of mass.

    It would probably be very frustrating for many people (and I’m included in that number, as I grew up with the imperial system) for a while, until about a generation or two later when the issue would be moot, and the U.S. would go about life more readily able to interact with much of the rest of the world on some issues related to things like trade and travel.

    Cue RealAmericanPatriot(tm) cries of “socialist!” in 5 … 4 … 3 …

    Still learning,


  3. F says

    Going by astronomically defined seasons does not make much sense, really, as far as how we feel about it and what outdoors plants and animals do. But I would suggest ignoring also equal quarters for the seasons. Spring and autumn should be shorter things, transitional periods between between summer and winter.

  4. Trebuchet says

    Phil Plait, the Bad Astronomer, has said he’d like to see the seasons centered on the solstices and equinoxes, instead of beginning on them. That doesn’t work for me because there’s too much hysteresis in the temperature change. He’d have winter starting in October.

    Your system is better, but too hard to remember. How about just starting the seasons on the firsts of December, March, June, and September?

  5. Mano Singham says

    Or maybe the 6th of each of those months. That would be easy to remember and be close enough to the more precise dates.

  6. Alan(UK) says

    On Saturday afternoon we had sunshine and blue skies. We also had cloud, cold, strong winds, rain and hail. Some of these more than once.

    In England we do not really care what you call the seasons. We just have weather, every day, guaranteed.

    Now, if you want to do something really useful. Get rid of your (US) crazy system of time zones. On the Internet, ‘now’ is ‘now’ for everyone. UTC is the same everywhere -- why not use it?

  7. JustKat says

    I’m in southeast Texas and we only have two seasons here -- summer and August. The hottest day is always in August. We’re just getting started in July…

  8. barbrykost says

    Here in Southern California, the seasons are fire, flood, wind, and drought. The dates and duration are variable, but you never have any doubt as to which season you are in.

  9. Mano Singham says

    Are you sure the hottest day is in August? I would have thought the hottest day would be the same everywhere in the northern hemisphere.

  10. SallyStrange: bottom-feeding, work-shy peasant says

    Seconded! Also, no more daylight savings time! That always fucks my shit up.

  11. says

    Excellent. Many years ago I did a Fourier Analysis of the temperatures and came to the same conclusions as you did. January 21 is, on average, the coldest day of the year, and July 21 is, on average, the hottest. It’s the same sort of environmental lag that leads to around 3 pm being the hottest part of the day.

    When I’ve run the data for other parts of the country (not just Ohio), they hold up fairly well. Regarding Texas, the peak is extremely broad, though maybe skewed a bit towards August, but it doesn’t really affect things to still stick with July 21.

  12. Heidi Nemeth says

    In 2000, I read a suggestion we change the calendar to make the first day of every month a Monday. The first two months of each quarter would have 28 days, the third month of each quarter would have 35 days or 5 weeks. That would leave one extra day per year, Year Day, with no day of the week designation. Year day would be tacked on as a holiday at the end of the year. Leap years would have another day, Leap Day, tacked on after Year Day, again with no day of the week designation. Jan. 1, 2001 was a Monday -- an appropriate start date for such a calendar. The calendar has obvious advantages for knowing the day of the week and for evening out business months and quarters.

    Combine this with your suggestion about changing the timing of the seasons. Spring starts March 10th. Summer starts June 10th. Fall starts September 10th. Winter starts December 10th. Easy to remember! However, this makes winter longer than summer. Maybe Year Day and Leap Day should fall in the summertime, though that would be awkward for the Southern Hemisphere.

  13. says

    The concept of seasons is purely arbitrary, isn’t it? If it snows in June, it’s cold, and that’s all that matters, not what label you give that time of year. Plus, the worst snowstorms always seem to arrive in March. I will admit, though, that it is possible that we just perceive March snowstorms as the worst simply because we’re expecting the weather to improve, and when it snows, it’s obviously not improving.

  14. Mano Singham says

    Can you share a little more detail of the results of your Fourier analysis if you still have them? Did it converge rapidly? What were the dominant frequencies and how did amplitudes of the second and third largest contributions compare to the primary one?


  15. says

    It really wasn’t all that exciting. Basically, at the solstices, the average temperatures stay flat for quite a while, and daily averages are only reported to the nearest degree. So finding the maximum or minimum is not obvious. I kept 5 terms in the Fourier analysis, and then used that to find the extrema.

    I also tested 6 terms but found that didn’t make much difference at all as to where the extrema were.

  16. LeftyGomez says

    With the traditional nomenclature, the days get longer during the winter and shorter during the summer!

    The problem with defining the seasons to correspond with the temperatures is that there is no astronomical point of interest in when the temperature extremes occur. The astronomically interesting points are the solstices and equinoxes, which are specific instants of time, and so the quarter years between them are the most natural ones to name.

    The alternative mentioned in Comment #4 would make the summer be the quarter-year of longest days, winter the shortest days, etc. That would put the solstices and equinoxes in the middle of the seasons, but then temperature would line up even worse than they do now (Nov 5 --> Feb 5 -->May 5 --> Aug 5 approximately).

    Unfortunately, the length of days seems to be about a month out ahead of temperature, so we can’t make both notions them correspond to a single notion of season.

  17. P Smith says

    It sounds like a good idea to match the seasont to the weather, but since weather varies place to place, so would the seasons. I currently live in Taiwan which has two seasons, summer and not-summer, with “cold” (between 10 and 20C) from December to February. And I once lived in a part of Canada that had winter and not-winter, with snow from September to May and Januaries of -40C (-40F) for weeks.

    Myself, I’d like to shift the calendar year to start with the winter solstice, not eight days later. And make the first five months 31 days long, the rest 30, with December 31 in a leap year.

  18. JustKat says

    Just a quick Google of southeast Texas temp data (Beaumont) doesn’t really show much difference between July and August temps so the hottest day could actually be in either month.

    August probably stands out more for me (and others who live here) because by that point it has been so hot for so long that it seems almost unbearable.

  19. Keljopy says

    Why not use it: Because firstly, the whole world has time zones, not just the US, even UTC has zones, and secondly that’s easy for you to say or anyone within a couple zones of you, but it would be really confusing for scheduling things in real life if a new day/date started, for example, in the early evening (even worse in places where it happened in the middle of the work day). Unless you are saying use UTC on the internet only or label the zones by UTC offset (instead of our Eastern/Central/Mountain/Pacific, etc.), in which case, that would make sense, and shouldn’t be too hard for everybody to learn.

Leave a Reply

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