1) Many opinions about the seasons are linked to location,
so they are hard to discuss on this cosmopolitan list.
2) There's no law that requires having four seasons, or
locking them to temperature. For example, in Costa Rica
there are only two seasons:
-- Summer (verano) lasts four months.
-- Winter (invierno) lasts eight months.
Those have more to do with rainfall than with temperature.
Kurt Vonnegut famously argued that upstate New York has
six seasons, not four.
3) Day length is directly locked to the declination of the sun.
For most of history, at most latitudes, this controlled the
amount of work (of certain types) you could get done in a day.
This is a pretty hard lock, although it is subject to minor
local orographic vagaries.
In many parts of the world, there have been cultural events
associated with the summer and winter solstices for a long,
long time. Equinoxes not quite so much.
4) Let's now focus on the case where there are four seasons,
and focus on temperate latitudes. The question then arises,
why is the temperature not a simple function of the length
of the day?
There is an interesting physics answer to that question.
It turns out that the system is overdamped. To a decent
approximation, it can be modeled as a one-pole low-pass
filter. So it is closely analogous to an RC circuit.
("The same equations have the same solutions.")
So the response (i.e. temperature) will lag the input (i.e.
day length) by 90 degrees of phase.
It's not quiiiite that simple; the coldest day is more
likely to be 75 or 80 days after the solstice (early February
in the northern hemisphere) rather than 90 days. Similarly
the hottest day is likely to be less than 90 days after the
solstice. So we are somewhat splitting the difference between
direct proportionality and 90° lag.
So there are physics reasons why winter and summer (by the
astronomical definition) will /contain/ the coldest and
Since temperature is not firmly locked to the solstices and
equinoxes, you could fudge the definitions of the seasons by
a week or two without much changing this consideration.
5) In places like Vermont, peak foliage color will occur in
the middle of autumn. There are good physics reasons why trees
want to get rid of their leaves and move as much energy as
possible into the roots.
Similarly, annual grasses have good reason to set seed and
store as much energy as they can in the seeds. This has
treeemendous ramification for humans, who find it advantageous
to /store/ the seeds.
Again, this is not firmly locked to the solstices and equinoxes.
So you could fudge the definitions by a week or two without
much change to this consideration. Also it depends on latitude;
for example, in Canada the "thanksgiving" harvest festival
comes a month earlier than US thanksgiving.
6) The named months have no causal relationship to the weather
or anything else. This is particularly obvious in the large
parts of the world that use a lunar calendar; for example,
Ramadan might fall in the middle of winter one year and the
middle of summer a few years later.
People are free to define things however they choose, so long
as other people are free to choose differently. Speaking for
myself, if we are going to define seasons in terms of months,
I see no good reason why there should be four of them. If we
are not going to lock the seasons to the sun, we might as well
lock to the local monsoons or other weather, which usually
won't come in tidy 3-month chunks. Any such definitions will
make sense at most locally, not globally.