I will describe work with Jason Starr in which we show that the group of sections of a family of abelian varieties over a higher dimensional base is determined by the restriction of the family to a "very general conic curve".