All english-speaking countries, and plenty of other cultures as well, I presume, have a tradition of the wife taking the husband's name after marriage. Further, the children born to the marriage take the couple's surname as well, themselves going on to continue its propagation or lose it entirely, as their sexes indicate.
This looks immediately unfair and arbitrary, putting unnecessary emphasis on one of the partners' names, and putting pressure on families to "produce" male children to continue propagation of the family name. Most cultures are full of arbitrary traditions, but this one seems particularly insidious due to a combination of how important name continuation is to some families, and how much the whole deal goes unquestioned by most others. Of course, same-sex relationships (specifically, long-term partnerships) don't have to deal with it, though it's only very recently that children being named by these couples have become an issue. Also, for the past hundred years or so hyphenation of surnames has been accepted by society, so both partners take both names, and the children end up with both as well.
Still, I think there's a better solution than hyphenation, and one that doesn't give the children an absurdly long name that will be difficult to handle in their own partnership. It's easy: give the male children the male parent's surname, and the female children the female parent's surname. This is perfectly fair, lets both name lineages continue for as long as people keep breeding, and ends the need to force one partner to mess around with changing their name, ever. Cooler still, it's a viral tradition because it spreads outward across generations, and would eventually overtake other naming schema even without cultural propagation.
Just as with the traditional scheme, though, it doesn't give an obvious way to name children raised in a same-sex partnership. I would hope for the parents to play paper-rock-scissors for the name (or some equally random method of choice), but it would have to be an individual decision. Trans- and post-gendered folks would have to make one as well, though if trans-gendered folks kept whatever surname they were given, it would provide means for "male" namespace to cross into "female" namespace and vice-versa.
If anybody knows of cultures which have traditionally used a naming scheme similar to the one outlined here -- and I'm sure a few exist -- tell me and I will list them below.
- mkb says that in Clarke's 3001, all of humanity follows this scheme. I don't think science fiction counts as another culture :-)
From jungans:
Hi. As you say, this tradition can sound a bit arbitrary but there is a reason for it to exist. When a child is born, society knows who the mother is because she carried him for 9 months. But what about the father? Society needs a way to know who the father is. That's why the kid is named after the male parent.
That's a good point, but I'm not buying. Granted that the woman carying the baby will be known in the community to be the mother, but if the father is absent and the mother names the child, she could use any name besides that of the actual father. The name proves nothing. Further down the line, when there might be a cultural need to know who the parents are, there is no (pre chemical testing) way to determine either parent's paternity -- the mother's advantage of having carried the child is lost in time. Either way, positive identification of the father is problematic, regardless of the kid's name.