The saros, the famous rhythm of similar eclipses a little over 18 years apart, is a fascinating subject with almost endless ramifications, explored over many pages of my Under-Standing of Eclipses.
For instance, in the section on the saros history of the great eclipse that’s coming in August, we see that it belongs to solar saros series 145; this began in the year 1639; its eclipses changed gradually from partial to annular and then to total; this eclipse is the 22nd member of the 77 in the series, and the 5th of the 44 total ones.
It may seem almost sacrilegious to complain that the saros has shortcomings. The first – actually a long-coming – is what looks like the least significant thing about it: the fractional part of the number representing its period. The period is basically 223 synodic months (cycles of the Moon), which works out at 18 years plus 10 or 11 or 12 days (depending on leap years) plus about a third of a day. It’s that third of a day that spoils it! Successive saros sisters may be similar, but they’re seen in far different parts of the Earth.
Relative to the situation on, say, 1999 August 11, when an eclipse that you may remember passed over Cornwall and Europe and Turkey, the spinning Earth has on 2017 August 21 made a third of a turn, and therefore the track passes over the U.S.A.
So a more convincing interval between similar eclipses is 3 saros jumps: 54 years plus a month (669 synodic months). Here is the totality-track of this August’s eclipse, compared with its sisters a tri-saros before and a tri-saros after. (And with one other eclipse.)
(The Moon’s shadow brushes from west to east. The little ellipses are the footprint of the total shadow at 10-minute intervals.)
You can see another change that happens as a saros series progresses: the track shifts in latitude. Series 145 began with partial eclipses near Earth’s north pole and is migrating southward. As we are showing eclipses three saros-jumps apart, the differences in latitude are quite wide.
(The tracks slope southeastward geographically because the season is approaching the September equinox, when Earth’s north pole leans forward. I tried rotating the picture counterclockwise – north-pole-forward – but I like it better as it is.)
The discovery of the saros periodicity in chronicles that mentioned eclipses probably did help ancient star-clerks to predict eclipses. If you were in Alaska or Maine in 1963, you had the means of knowing that the sky there would darken again – though only partially – 54 years and a month later.
I wasn’t around to see the eclipse of July 1963. I saw my first eclipse in July of 1972, from not far north of Maine, camping at Cap Chat on the shore of the St. Lawrence. That was an eclipse of a different saros series, 126. My mental arithmetic is pathetically slow, and when I thought of drawing this Tri-Saros picture I thought, Was that Canadian eclipse the precursor for this year’s? No, of course not, but when I plotted the 1963 track it took me aback by crossing the Gulf of St. Lawrence. So I realized that adding the 1972 track to the picture serves to show another debility of the sacred saros. Some eclipses, even if not far apart in time, can be pretty similar without being saros sisters.