65-66/70
Depends if they take a mark off for last q probability: in one of the lines of working I wrote (1- r) instead of (1-r^2) for the limiting sum of the GP but that was only for one line. Got it out though.
Think I made up for my poorer than expected 4U performance.