After doing it I decided not to pursue it further until I had devised
a much quicker way of gathering and processing couples' data. I've
since done that, but due to a general lack of interest in the topic I
haven't followed up on it.

I was also a bit slow to do anything more after reading that G. Sachs
had found a similar result in a huge study years earlier and that the
general consensus amongst the prominent skeptics was that it was
caused by people erroneously writing down their partners' birth dates
as their own on their marriage applications (or something like that).
My data came from an entirely different source - from genealogy files
covering 3 centuries I think, so the same sort of 'error' should not
have occurred.

Rudolph Smit's website has an article on Sachs' work (and book), but
parts of it are extremely biased and unfair, although there was some
valid criticism. I don't have time at the moment to look it up. One of
the criticisms was that the percentage of variation was too small, but
that seems to be quite normal when we use an immense amount of data.