From this description we may see that the first mechanical clock was a transition from the water clock to the purely mechanical clock of Europe which used no water power anywhere. However, this does not mean the clock of I-Hsing was a water clock by any means. What it means is that the first mechanical escapement for the clock was worked not by a failing weight or by springs, but by water power. This is quite a different thing from having an actual water clock whose time indicator rises with the level of water (or mercury) in a tank. It is not unreasonable that in developing a mechanical clock, the Chinese turned to water as a power source, since all previous clocks had been water clocks, and the association of ideas was natural to them. The perpetual flow of water was likened to the perpetual turning of the heavens. As Su Sung wrote in 1092 in speaking of his own improved clock:

The principle of the use of water power for the driving mechanism has always been the same. The heavens move without ceasing and so also does water flow and fall. Thus if the water is made to pour with perfect evenness then the comparison of the rotary movements of the heavens and the machine will show no discrepancy or contradiction; for the unresting follows the unceasing.

  The actual workings of the clock of I-Hsing consisted of a vertical water-wheel which instead of paddles (such as are turned by a rushing stream) had cups at the ends of the blades. These cups were filled by water dripping from a water clock. I-Hsing's clock was therefore a mechanical clock dn'ven b_v a water clock. When one of the cups was sufficiently full, it would weigh enough to turn the great wheel by one notch, overcoming the resistance of a restraining tooth which had held the wheel still while the cup filled. Various gear arrangements would then transmit the movement to the time indicators, etc. As the ancient text tells us, shafts, hooks, pins, and interlocking rods were all part of the apparatus. Some idea of what these may have been like may be gathered from the fuller description of Su Sung's clock which we shall see below.
  This clock of I-Hsing would have been a poor time-keeper, moving jerkily and representing more the first realization of a wonderful idea than a superb piece of machinery. It was because of the enormous pronuse that this clock offered for future improvements that contemporaries would have had every reason to be excited. And its promise was more than borne out in succeeding centuries. I-Hsing himself died only two years later, and was unable to construct a 'second generation' model.
  I-Hsing's clock was, like water clocks, subject to the vicissitudes of the weather. In order to keep the water in them from freezing, torches generally burnt beside them. This was of course not necessary for the small mercury clocks, as mercury does not freeze at any temperature likely to occur in the Earth's weather. Therefore, in the next great clock of which we have accounts in China, mercury was substituted for water because of the freezing problem. This clock was built by Chang Ssu-HsiAn in 976 AD. It represented a considerable elaboration and improvement over I-Hsing's machine, though the latter no longer existed and had been lost even before the end of the T'ang Dynasty in 906.
  Chang Ssu-Hsiin's clock was apparently much larger than I-Hsing's. It was certainly far more complex. The dynastic history of the time describes it:

More...

John Combridge's model of the water-wheel escapement of Su Sung's great astronomical clock. The water or mercury poured from the tank on the right, regulated by means of a water clock, and turned the wheel clockwise. Each spoke bears a bucket which, when filled, turns the wheel a fraction. The motion was jerky, but it sufficed. (Science Museum, London.)