Chris, I had similar unexpected results - where the F2 were all very very similar to my F1 and looked to be easy to stablilize that type, but then the F3 turned out to be a very mixed bag.
Another one was at F4 and seemed to be quite stable, but then the F5 suddenly saw a load of segregation for shape. These were hearts that segregated into many subtypes including paste shapes, nipples or not, etc.
My impression is that the shape genetics of pointy fruit are very very complex. There may be some epistasis involved which prevents things being expressed as expected. Then with another shuffle, bingo out they come.
Pretty baffling.
I was reading about it this spring, there are ??? seven I think different genes for 'pointy'. So crosses of two pointy fruit won't necessarily segregate as expected if there were only one gene for that.