Recent advances in genetics controversially suggest that the model plant Arabidopsis thaliana performs genetic repair using genetic information that originates in the individual's grandparent generation. We apply this ancestral genetic repair strategy within an Evolutionary Algorithm (EA) to solve a constraint based optimisation problem. Re- sults indicate that the grandparent based genetic repair strategy out- performs the parent alternative. Within this framework, we investigate the impact of storing only the ttest ancestors for use as a repair tem- plate. The in uence of performing repair in a xed direction is compared to randomly varying the direction in which error detection proceeds. Fi- nally we explore the impact of varying the direction of repair on the results produced. All results seem to support the non-Mendelian inheri- tance process suggested by Lolle et al.