As I understand it, during the Roman Empire, the soltices and equinoxen were on the 25th of the month. How did they end up on the 21st in the modern era?
Here is an explanation of the Precession of the Equinoxes, complete with math.
Tris
I hadn’t thought of precession of the equinoxen, but something is wrong with it.
According to that page, precession relative to the sun is about 22,000 years. In 2,000 years, the equinoxen would move more than a month, not just 3 days. Or has it moved that far? Did the Romans have their winter solstice in January? I’m pretty sure it was Dec 25 for them.
The precession of the equinoxes has nothing to do with it because the calendar is based on the tropical year which is, by definition, the time between two successive equinoxes. But the length of the tropical year was not measured accurately until recently in history so that old definitions of the year contained errors and the equinoxes moved within the calendar. I am not sure where you get that the equinoxes fell on the 25th as I always heard the Julian calendar had them on the 21st but due to the error it introduced, the equinoxes started to drift back and by the 16th century the error accumulated was ten days so in 1582 Pope Gregory decreed the new calendar which took those ten days out in one go and redefined the length of the year. reference:
http://www.tondering.dk/claus/calendar.html
I believe Roman solstices / equinoxes were about on the 25th instead of about on the 21. “about” is an important word here. Because we only correct every 4 years, the physical equinox / solstice dates wander about over a calendar day from year to year
(see equinox / solstice dates here - http://www.erh.noaa.gov/er/box/equinox.html )
Part of the answer the OP is looking for is that when the Gregorian calendar reform sailor mentions was introduced, they corrected the calendar to realign the seasons with observations made in the early part of the 4th century, by which time nearly 4 centuries of error had accumulated in the Julian calendar since the time of Julius Caesar. The error in the Julian calendar is about 3/4 of a day per century, and people sometimes ask why Gregory didn’t remove 12 days, not 10, to get back to the “start”.
“wander about over a calendar day from year to year” was awkward. What I meant was that they drift by about a day over a 4 year period (at which point they are corrected by a leap year), and might be on different days from year to year in a given locale.
dtilque, I was starting to think you must be Dutch or German, because of the plural you keep using: “equinoxen.” Then I got the joke. 
yabob, thanks for the answer. For some reason I never thought about what the Gregorian correction actually did.
Jomo, one xerox, two xeroxen ==> one equinox, two equinoxen. English is real regular about that.
Ok then. Will you explain it to those of us who don’t get it and still think he’s Dutch or German?