A plant breeding technique that is used to create entirely new varieties of plants that combine the best qualities of selected existing varieties. The method is limited to self-pollinating species. Parents are selected and artificially crossed. A few (say six to ten) of the resulting F
1 hybrid seeds are sown widely spaced to encourage prolific seed set. Assuming the parents were homozygous, the F
1 plants should all be identical. The seed from each plant is harvested separately and grown, again widely spaced, in separate 'family' plots. The genotypic variability becomes apparent in the F
2 and the characteristics of each plant are noted. The seed from a small number (perhaps only 1%) of promising plants is harvested to grow on to the F
3. The seeds from each plant are again sown in separate plots and in this way the 'pedigree' of each plant can be recorded. By noting the variability within each of the F
3 plots it may be determined which plants of the previous generation depended excessively on heterozygosity for vigour. At this stage whole families of plants may be discarded. As heterozygosity is approximately halved each generation, increasing emphasis is placed in succeeding generations on selection between rather than within families. If breeding a crop plant, then at about the F
6 generation seeds from selected plants are grown closely spaced as would occur in normal field conditions. Yield may then be determined. From this point selection is entirely between families. By the F
8 generation the remaining selected lines may be assumed to be over 99% homozygous and they are bulked up for variety trials. Compare
backcrossing.
