Jim is correct and Hoyle said they worked out the Eclipse year as 346.6 days x 19 years = 6585.4 days so the hit the 6585.333r slightly late but always on the correct day with a reset.
'Aside from that, however, the eclipses of August 21, 2017, and August 11, 1999, are essentially the same. For example, the U.S. city of Salem, Oregon, is in more or less the same place on the path of totality as the UK city of Truro, Cornwall. If we look at the animation of the eclipse for each city, we can see that, separated by 18 years, 11 days, and 8 hours, the two eclipses in Salem and Truro are almost identical.'
This is in the article Jim posted.
18 years 11 days and 8 hours is 6585.333r days
6585.333 - 11.333days = 6574 days = 18 years so one solar eclipse year is 365.22222r days.
Eventually this must have been worked out over quite literally thosands of years as Hoyle does state. The records are recorded in stone in the UK as professor Thom has proved beyond reasonable doubt.
the imperial system does appear to record the key unit as 5280 / 360 = 14.666r x 449 = 6585.333r and x 450 = 6600 or 5280 /8 x 10 so a base 10 system considered sacred by Thom.
Pi in this system is 22/7 or 14.666r/4.666r difference 10 units.
Jim has provided some good evidence that that ancients knew the Saros cycle using UK stone circles.So have i and so has Bill Wilkinson, all of us working independently.
It goes further as the Babylonians have this ancient unit 1/360 named the She.
The 1 is one metre.
1/360 x 5280 = 14.666r x 360 = 5280.
Jim has worked the metre into his equations and they were using it at Carnac in France as discovered by Howard crowhurst using the work of Thom.
variations on pi are a concincing way of explaininghow the ancients manipulated Pi to produce meaninful information for example
864/275 = fibonacci Pi as used by Jim in some of his calculations.
14.666/4.666r is anotehr vesion and is the same as 22/7.
A key in understanding the manipulation of pi is 22/7 x 2499/2500 = 3.1416 this places 22/7 as a base 10000 version of Pi.