For a century astronomers had suspected that time based on the rotation of the earth was not uniform. In 1939 Spencer Jones demonstrated without a doubt that the rotation of the earth speeded up and slowed down. In 1948 Clemence suggested that two kinds of time were necessary, a new, uniform time consistent with the motion of celestial bodies, and the old time based on calendar days of the rotating earth. In 1950 that suggestion was endorsed by an international conference Constants fondamentales de l'astronomie held in Paris.
At the 1952 General Assembly in Rome the IAU resolved that the long history of subdividing calendar days of the rotating earth into Universal Time was no longer adequate for describing the motions of bodies in the solar system. They resolved that starting in 1960 the motions of planets and moons would be described using Ephemeris Time, a new, uniform time scale which was not affected by the variations of the length of a calendar day produced by the non-uniform rotation of the earth. They also resolved that no change would be made to Universal Time, so that the progression of calendar days and the setting of civil clocks would continue as they always had.
In 1954 Sadler wrote a monograph about the impending changes. He described the implications of two kinds of time. He explained how the existing Universal Time really worked, and he compared that with the new Ephemeris Time.
Sadler's 1954 description was a clearer exposition than any other publication about time for the next 60 years. Alas, Occasional Notices of the RAS remains obscure and not generally accesible. Here is a transcription of Sadler's 1954 monograph:
In 1955 the IAU resolved to redefine the second to be unrelated to the calendar day (defined by planetary motions) while at the same time keeping radio broadcast time signals using the old kind of second that was related to the calendar day. At the same meeting the IAU discussed the cesium atomic chronometers which were already being compared with time based on planetary motions.
By 1959 it was clear that cesium atomic chronometers provided a time scale more uniform than the rotation of the earth. The metrology agencies of the US and UK met in August and agreed to use cesium chronometers for their radio broadcast time signals starting at 1960. The metrology agencies agreed to coordinate the offset steps of broadcast time and frequency so as to keep calendar days based on earth rotation. In 1960 the general assembly of URSI recommended that other countries should also coordinate their radio broadcast time signals using cesium chronometers. In 1961 the IAU General Assembly resolved that the BIH should specify the offset steps of broadcast time and frequency starting at 1962.
In 1964 the IAU explained the need for two kinds of time and again described their characteristics. This was the last apparent point of of agreement on time between metrologists and astronomers. Over the next 5 years the rift grew, and astronomers vainly pointed out that an official time scale with steps was not a good idea.
In 1970 the CCIR ignored the advice that astronomers had been giving for 20 years, and excluded them from the discussion. The CCIR decided that a single time scale with leap seconds was the way to go. Over the next decade they sold UTC with leap seconds to various national and international agencies as the perfect solution.
But in 1970 the US Naval Observatory decided that despite the CCIR decision, the reality of the way the world works meant there must be two kinds of time used in radio broadcasts. The USNO radio broadcasts for the automated navigation systems LORAN-C, OMEGA, and TRANSIT switched from the old style of coordinated time to purely atomic time with no leap seconds. A decade later as the GPS satellites became operational they also implemented a purely atomic time scale with no leap seconds.
Machines need atomic time that proceeds uniformly with no regard for whether or not the earth has rotated far enough for today to become tomorrow. Alternatively, human history, culture, and law expect that the change from one calendar day to the next is measured by one rotation of the earth, not by watching cesium atoms. One kind of time cannot serve both purposes.