"Anyone who's taught in a true comprehensive, with the full range of ability, as opposed to a secondary modern without the top 25-30%, will tell you that the presence of a sizeable number of able and interested pupils lifts the performance of the middle band."
I think this is certainly the case.
The other bit about the mix in schools that I find it upsetting is the social split it creates. This is not really an educational point, but my daughter's Grammar school friends are all Pizza Express and Gap clothes people, her secondary modern is completely McDonalds and Primark.
So my daughter hides her secret Pizza Express life from her friends, it's bizarre.
And while I hope my son goes to Grammar school I'd much rather he mixed with a range of people. He's pretty much going to assume everyone can afford foreign holidays and nice stuff. he may get good exam results but he's going to need tutoring in reality! 
"You've said that grammar's have the advantage of attracting better staff. I can well imagine that MIGHT be true, I don't know. I'm not aware of any huge problem of secondary moderns being plagued by substandard staff."
It might be true, and after half a bottle of wine on a dull Saturday night I made a FOI request to find out. I asked the difference between the number of applications for Grammar teaching posts vs Moderns, and also the number of supply teachers used in each.
I can say that in the two schools my daughter's been to there's been a huge problem with teachers leaving and supply teachers in lessons for months. This wouldn't be so bad but the supply teachers never bother to set homework.
Good and bad teachers exist in all schools, but you'd think if Grammar school's get more applications for each teaching job this will lead to some effects on the rest of the schools.
If anyone can think of any stats stuff to show Grammar Schools are a Good Thing I will quite happily look at that too.
It was a good point someone made about overall exam results being no better in Kent or other selective counties. But if there is some statistical proof that a selective education system is better I will be happy to see the other side.