The most common definition of the tropical year is the interval between two successive passages of the Sun through the vernal equinox and lasts 365.242199 days UT. Note that any start and end point could be used as long as the sun reappears in the same position one year later.
The trouble is that using different start points for this measurement results in different year lengths. I will now attempt to explain why.
The orbit of the earth around the sun was defined by Kepler's Laws which tell us that the speed of the earth varies as it follows its elliptical path around the sun. Even so one complete circuit should include the slow parts and the quick parts so one would expect the time taken to remain constant. But the earth is also precessing, like a gyroscope, on its own axis. Put more simply it is " wobbelling " which varies its position in the sky. This means that when it has completed a circuit of the sun it appears that it has overshot by a time equal to 24 to 26 minutes. The exact value being dependent on when the circuit started.
In 2000 AD the following year lengths resulted from measurements starting and ending at the season stated :-
Start time | Year length in days |
---|---|
Vernal Equinox | 365.2424 |
Summer Solstice | 365.2416 |
Autumn Equinox | 365.2421 |
Winter Solstice | 365.2427 |
Average | 365.242199 |
So what is happening is that most astronomers and calendar students are defining the Tropical Year as starting with the vernal equinox but using the average figure for their calculations.
You can learn more about this from Simon Cassidy's web pages on Error in statement of the Tropical Year.
Simon quotes two other sites. They are the Royal Greenwich Observatory pamphlet No. 48 and Leroy Doggett's chapter on Calendars in the Explanatory Supplement to the Astronomical Almanac.