Caption: The Earth orbiting the Sun with the zodiac constellations displayed.
Features:
From the Earth-at-rest (or geocentric) perspective that we often take in observations, the Sun also moves eastward at the same rate as the whole midnight night sky.
The objects in the midnight transit category for one day, get replaced in that category by objects farther east on the celestial sphere on the next day because of the Earth's eastward motion.
The former midnight transit objects are now west of the meridian at midnight. They now transit before midnight.
Thus, the relatively unmoving objects that rise with Sun one day, rise before the Sun on the next day.
One can be a bit more precise. The mean solar day ≅ 86400.002 s (J2000) (i.e., the mean time from solar noon to solar noon) and the mean sidereal day ≅ 86164.1 s. Thus, earlier by 86400.002 s - 86164.1 s = 235.9 s = 3m:55.9s ≅ 4 minutes.
Some more understanding---maybe---is given from a simple calculation.
Say we observe a particular rising at
solar time Y on
ordinal date X of some year.
One common year (365 days) later,
there must again be a rising at about
solar time Y on
ordinal date X.
But this must be the rising 365+1 since last year at
solar time Y on
ordinal date X
since the
mean sidereal day is shorter
than the
mean solar day
and
the relative positions of the
Earth
Sun
and the fixed stars
have NOT repeated until now.
So accumulated
time difference for 366 rising periods must be one day less than for 366 days so that
the 366th rising happens 365 days later.
The calculation for the accumulated time difference is
Why is this accumulated value time difference NOT exactly 24 h?
Several reasons.
The main reason is that the
sidereal year = 365.256363004 days (J2000),
NOT 365 days.
The sidereal year is nearly the exact
repeat time for relative positions of the
Earth
Sun,
and the fixed stars.
However, that ∼ 0.25 days
decimal fraction
of the sidereal year
means the rising times
etc.
CANNOT happen at the nearly exact repeat time, but
are offset from it by ∼ 0.25 days.
Now the sidereal year
is closely approximated by the
Julian year = 365.25
which is the average year of the
leap year cycle of 4 years.
If you average over a
leap year cycle
of 3 common years = 365 days
and 1 leap year = 366 days,
you get an average accumulated value time difference of 24.00 h which 24 h to 4 digits.
Highly accurate/precise
calculations would have to account for
the axial precession
and various other factors including
astronomical perturbations.