Prehistory of the second known to us:

With the introduction of the French revolutionary calendar in 1792, an attempt was made to change the division of a day to the decimal system. A day was divided into ten hours, an hour into one hundred minutes and a minute into one hundred seconds. Accordingly, a day had 100 000 seconds compared to 86 400 seconds of conventional determination. For the smallest unit of time, this decimal time had an imperceptible effect, as it was only about 15 % shorter. The impact on the next two units of time was much greater. The decimal minute was almost one and a half times as long, the decimal hour almost two and a half times as long. A second can be divided hexagesimally into sixty small units of time, the tertians. However, this is completely uncommon.

Modern time:

As long as one assumed a uniform rotation of the earth, the second was the sixtieth part of a minute of the day divided into 24 hours of 60 minutes.

Around 1885, Karl Friedrich Küstner (Bonn Observatory) established that the Earth's axis of rotation performs a polar movement of 5 to 10 metres. This was analysed in more detail by Richard Schumann (Vienna University of Technology) and Seth Carlo Chandler (Harvard) and was the first indication that the rotation period itself also changes. However, it could not yet be proven with the best clocks of the time (deviation 0.05 seconds per day).

It was not until 1934 that Adolf Scheibe and Udo Adelsberger succeeded in doing so at the Physikalisch-Technische Reichsanstalt. After eliminating their own doubts, they published the results in 1935 and made them more precise with the quartz clock they had developed. From about 1950 onwards, it became clear that the continuously improved quartz clocks would be a better time standard than the Earth's rotation. The astronomical length of day not only gradually increases because of tidal friction, but also shows irregular changes due to magma flows between the Earth's mantle and core. Due to the slowing down of the Earth's rotation, the solar day shifts from a completely uniform measure of time. To compensate, a leap second must be inserted every two to five years to synchronise all clocks with the solar day, which becomes a few fractions of a second longer.

Until 1967, the second was based on astronomical measurements:

Solar second (until the 1950s): The fraction 1⁄86 400 of the mean solar day. This definition was introduced so that an average solar day is 24 x 60 x 60 seconds long. This corresponds to the time after which a fictitious mean sun is again in the same position. (The solar day is about 4 minutes longer than the Earth's rotation time because the Earth moves around the Sun during the day and it therefore takes a little longer for a point on the Earth to face the Sun again). Ephemeris second: The fraction 1⁄31 556 925.9747 of the tropical year on 0 January 1900 (= 31 December 1899) at 12 noon UT. It refers to the relationship between the duration of the year and the rotation of the Earth at that time. The ephimerides second was defined as a second by the International Committee on Weights and Measures (CIPM) in 1956.

Since 2018:

The precision with which the second can be realised (as of 2018: approx. 10^-16) limits the achievable accuracy of time measurements in the unit "second". In the meantime, clocks have been developed that are not based on cesium atoms and are up to two orders of magnitude more precise. With such clocks as the primary reference, even more accurate time measurements could be achieved. The prerequisite for this would be a redefinition of the unit of measurement. This could happen in 2030.

Hope this answered it. If there is any clarification needed or a follow up question just leave them in the comments.