I found this webpage while looking at some lateral thinking puzzles. It says:
Sky Map shows full moons at 20:29 (UT) Jan. 30, and 11:52 Mar. 1, 1866, so that checks out. But the Metonic Cycle (225 lunar months, which is almost exactly 19 years) should give us another February without a full moon, and yep Sky Map says 16:19 Jan. 30, and 04:00 Mar. 1, 1885 were full moons. Perhaps the question meant EST, or similar. Still, just the law of averages should have it more often than 2 1/2 million years. And now that I think about it, SeanF and I talked about Feb. 1999 here, full moons at 16:07 Jan. 31, and 06:59 Mar. 2, 1999.

Is this a misstatement of some other puzzle?

PS: Nevermind, I found that it was taken from one of Paul Sloane's books, and it appears on his errors page.

I would think that 10^6 years would produce more than one date (1999) since 1889 for a full moon in February.

Isn't there an algorithm to calculate such trivia?

Generally if an error is posted on most of the www BBS's there will be a lot of shots taken, and if it is an autour with books to sell he gets a lot of free advertising.

__________________
Do nothing simply when a way can be found to make it complex and worderful.