At www.currentscience.ac.in/Volumes/105/01/0061.pdf … computer modelling of eclipses doesn't fit with observation. The above link shows recorded eclipse data from India, which is regarded as reliable, but it does not match NASA models of eclipses. The Indian data is between 400 and 1800AD and the paper goes on to suggest the lunar orbit of the Earth must change more than predicted and that the tides on the Earth must be strong enough to change the length of day. What this is saying is that length of day is dependent on the Moon and its orbit (to a degree).
Even more interesting is the research into the changes in lunar orbit must also impact the tides, the distribution of water around the planet, and the shape of the Earth. These are small changes but the discrepency in the observations of eclipse arises primarily when the Moon is close to one of the standstills – and is possibly even larger during major standstill (see also http://chiefio.wordpress.com/2014/01/25/a-remarkable-lunar-paper-and-num…