The development of inbred lines and the search for their best hybrid combinations are the main bases of corn improvement in the United States. The most commonly used method for inbred development is to enforce self fertilization for several generations while practicing visual selection for the more highly heritable characteristics. During these generations, surviving stocks maintained on an ear-to-row basis become highly homozygous and highly homogeneous. Because' evaluation for combining ability by using test-cross procedures is expensive, it is usually delayed until after three to five generations of visual selection. Visual selection for combining ability among inbred progenies is rarely emphasized because of the commonly held opinion that it is relatively ineffective. Regardless of the effectiveness of visual selection, total genetic variability will be reduced following each selection cycle, according to the principle that the variability of a sample is less than the variability of a population. If selection is effective, the reduction will be even greater because of the removal of undesired genes, but the mean of the selected lines will exceed the mean of the original population.

Early testing, as proposed by Jenkins ( 1935 ), takes advantage of evaluation for combining ability in the So generation of a maize population, or the F₂ of a hybrid, before genes have been eliminated by selection and, therefore, when the genetic variability among individuals is at a maximum. Superior germ plasm, identified by early testing, need not face the hazard of several generations of random sampling as in visual selection. However, the considerable expense of testing restricts the size and, therefore, the genetic base of the original population.