R help - Jul 2004 - prior weights in binomial gam

Hi,

I'm adjusting a binomial gam and I use the inverse variance of each
proportion as weights in the gam function (it's a model for the sex
ratio of hake).

My doubt is what are these weights doing ? I get a negative UBRE and if
I use GCV the scale parameter is close to 0 ...

Regards

EJ