If we use only a Sundial to measure time, the Sun keeps perfect time, wherever we are. But the moment we use any (other) sort of clock we find there are usually differences between clock time and sundial time.
The most obvious difference between sundial time and the time shown by a modern clock is in the length of the hours due to the seasonal variation in the length of the day. If we divide the day into twelve hours and the night into twelve hours, the length of a night hour will be different from a day hour, except at the equinoxes, and the day and night hours will also very in length according to the seasons. A modern clock avoids this difficulty by making all the hours the same length, but many of the early and mediaeval clockmakers, particularly the makers of Islamic water clocks, went to most elaborate lengths to ensure that their clocks kept what they considered to be the proper hours. (Reference 13).
When you are setting up a sundial you must align it correctly with the Earth's axis - for this reason sundials intended to be portable often included a compass for finding North. You also need to set the angle of the gnomon for your latitude. Once properly set up the sundial will always give the correct sundial time. If you carry a sundial around, and keep it correctly aligned as you do so, it will always show the same time as any other sundial in the same place. But a clock which shows the same time as a sundial in one place will show a time different from a sundial in a different place, due to the difference in the time at which the Sun reaches its zenith at different longitudes and the difference in the length of the day at different latitudes.
Sundials measure true solar time so when the Sun is at its zenith a sundial will say 1200. The Earth's orbit round the Sun is not circular but very slightly elliptical. This means that as it gets nearer the Sun it speeds up, so the time from zenith to zenith gets longer. But a clock gives all days as the same length, so there will be a seasonal difference between midday on a sundial and midday on a clock - this difference is the equation of time and can be as much as fifteen minutes. This was of great importance to the navigators of the sixteenth and seventeenth centuries in the measurement of their longitude and could make the difference between a safe landfall and disaster (Reference 4).
This is why we have to say that a true solar day is only approximately 24 hours: correctly a 24 hour period is called a mean solar day, the length of a solar day averaged over a tropical year.
A sundial and a clock will show the same time for noon only four times a year: on April 4th, June 6th, August 20th and December 14th (Reference 11).
This is all more fully discussed in Reference 2
The Earth's core is not rigid, but there are movements in it similar to convection currents in a liquid or gas. This means that although its mass and centre of gravity do not change the distribution of its mass (first moment of inertia) does, and this and other factors cause short-term variations in its speed of rotation. This short-term variation is of little practical concern to most people.
There is also a long-term variation in its speed of rotation due to the frictional effect of the tides, among other factors. This long-term variation must be taken into account, otherwise clock time would gradually drift away from solar time. From time to time therefore leap seconds are added at midnight at the end of December 31st.
There is much more on the
Earth's rotation on the International Earth Rotation Service web site -
please click here to visit it.
Water clocks were widely used in ancient and mediaeval times, and as described above the mediaeval Islamic water clock makers became very good at making their clocks keep what they regarded as the proper time, that is, with hours of different lengths according to the seasons. The first mechanical clockmakers were Europeans, and they made all the hours the same length because they lacked the skill to make their clocks reliable enough to work in any other way.
The history of time measurement is very fully discussed in David S Landes fascinating book Revolution in Time, published by Viking. This book was not used in the writing of these web pages but it is well worth reading.Why we believe that one method of measuring time is more accurate than another is really a philosophical rather than scientific question. However we normally assume that if we have ten different clocks and after a week nine show the same time and one shows a different time it is better to use the time shown on the nine clocks than that shown on the one clock.
As mechanical clocks got more accurate, that is, as different clocks began to keep the same time as each other for days and weeks or even months, the difference between twenty four hours on a clock and the length of the true solar day began to be significant. Later still clocks became even more accurate and the variation in the length of the solar year became apparent, and the use of the Earth as a timekeeper was finally abandoned.
Originally the second was defined as a certain fraction of a true solar day. Then it was defined as a certain fraction of a mean solar day, then as a certain fraction of the tropical year 1900. Today it is defined as the time taken for a certain number of vibrations of a caesium atom under specified conditions.
There is much more about clocks and timekeeping on the A Walk Through Time website, which you can link to from the Solar Year page.