Gene transfer experiments using sublethally irradiated pollen have given inconsistent results, with the species used falling generally into two classes: (a), those in which the Ml plants, arising from the use of irradiated pollen, are highly variable and show a "maternalisation effect", many plants lacking one or more dominant markers from the irradiated parent and (b), those in which the Ml plants are generally alike and similar to the Fl plants obtained by the use of unirradiated pollen, and there is no maternalisation effect. It is suggested here that this difference relates to the absence (a) or presence (b) of certain chromosome and DNA repair enzymes in the pollen, for which, it is argued, plant species may be widely variant and "polymorphic". The proposal is in agreement with the theory of chromosome repair through somatic recombination and gene conversion previously invoked to explain the maternalisation effect and gene transfer through the use of sublethally irradiated pollen.