HETEROSIS IN A NEW POPULATION 431 



As pointed out before, in the first three varieties listed above, 3-6 j)er cent 

 of the lines were classified as A , 12-15 per cent as B, and the rest were classi- 

 fied as C. In Chalqueno, the Si families were discarded more heavily and the 

 67 Si families selected for further inbreeding were classified as follows: 16^, 

 35B, and 13C. It is clearly evident that by far the majority of S-> lines saved 

 for further study after testing in topcrosses came from the Si families origi- 

 nally classified as A . 



In the early stages of the program, about five hundred plants were inbred 

 in each variety. At harvest, about two hundred, or 40 per cent, were selected 

 for further work. The above data indicate that, as far as the varieties listed 

 were concerned, the same results might have been obtained with a more 

 drastic elimination of lines at the beginning of inbreeding. 



Selection of testers for use in the isolation of good inbred lines is always a 

 problem. In Mexico, chief concern in the early stages of the breeding program 

 was the isolation of lines with good general combining ability. For this pur- 

 pose the inbred lines in each zone were topcrossed on at least two different 

 testers, usually unrelated adapted open-pollinated varieties. The good com- 

 biners, when selected on average topcross performance, often were disap- 

 pointing when crossed inter se. In those zones where two varieties were avail- 

 able which, when crossed, produced a desirable hybrid agronomically, a re- 

 ciprocal method of testing was used. With this method, inbreds from variety 

 A were topcrossed on variety B, and inbreds of variety B were topcrossed on 

 variety A. Good combiners thus isolated from variety A were crossed with 

 good combiners from variety B to form single crosses and subsequently 

 double crosses. This method of double cross formation was more efficient 

 than the recombination of lines with so-called general combining ability from 

 the above method. Where a good single cross of first generation inbred lines 

 was available that could be used as a tester to isolate inbred lines which com- 

 bine well with it, this cross proved more efficient than either of the above 

 two methods in the formation of good double crosses. 



Utilization of Semi-inbred Lines in Synthetics and Hybrids 



Almost from the beginning of hybrid corn production, breeders have 

 sought to discover methods of utilizing superior inbred strains in more or 

 less permanent combinations. In a country such as Mexico, where the majori- 

 ty of the farmers will not readily adopt a practice of securing new hybrid seed 

 for each planting, superior synthetic varieties would have real advantage. 



As Sprague and Jenkins (1943) pointed out, four factors operate to deter- 

 mine the yield of advanced generations of hybrids: (1) the number of lines 

 involved, (2) the mean yield of these lines, (3) the mean yield of all of their 

 possible single crosses, and (4) the percentage of self pollination. Since maize 

 is almost wholly cross-pollinated, the last factor may be largely ignored. 



