Thought experiment:

1. Take a very large group of men
2. Rank them by physical attractiveness
3. Count their children of both genders
4. Compare the results of 3 for the top and bottom quartiles (deciles, whatever) established in 2.

My guess is that with a large enough sample, there would be a difference. I would expect better-looking men to have more male children than female. I wonder if someone else has already done this experiement?