Say you are a crazy scientist and grow many many plants, then everytime you cross path with a much better than average specimen, you self it to see if its homogenous and its not, only 10% of plants are like the mother, proving its non linear self compatibility/self combining ability, you set it aside and repeat the process. Once you have - for sake of argument- 11 such plants with heterozygous genotypes and start crossing them with each other, effectively testing specific combining ability, although each plants are hetero, on your journey you may find a combo giving rise to predominantly single phenotype.
Shit that sounded good. I is drunk and high as fook. Do the most with what you have. And if you have too much, set some impossible boundaries.there.
I hope my sausage finger did not mutilate what I was trying to say 
GL