A Simple Model of Technological Research 141 



Breeders had a well-defined set of breeding and testing pro- 

 cedures. For a twenty- to thirty-year period, these procedures re- 

 mained pretty well fixed. The genetic material utilized in the 

 crosses remained relatively constant as well. 



Stage 3 represented a distinct advance in the structure of dis- 

 covery in the form of interspecific hybridization breeding pro- 

 cedures. This advance in breeding methodology was developed in 

 Java and India and was based on increasing the variance of the 

 distribution of potential new varieties. Noncommercial cane 

 species with desirable traits were introduced into the breeding 

 programs. 



Stage 4 was based on an organizational change, in that experi- 

 ment-station breeding programs were expanded to many new 

 regions of the world after 1930 to enable the breeding of varieties 

 specifically tailored for the geoclimate conditions in particular 

 countries. 



The model set forth here describes the discovery of new 

 varieties as a process of random drawings from a distribution of 

 potential new varieties. Figure 8.1 portrays the basic features of 

 the distribution for stages 1, 2, and 3. It can be viewed as the dis- 

 tribution of all possible varieties given the genetic material and 

 breeding methodology. Each variety is measured in terms of eco- 

 nomic value (yield per acre in this example). The potential in stage 

 1 is very limited since the distribution of genetic combination 

 possibilities is limited to rare random natural combinations. The 

 stage 2 advances opened up tremendous new possibilities for 

 combination. These did not necessarily increase the mean of the 

 distribution, but a large increase in variance was realized. 



As the development in the following section shows, the in- 

 crease in variance also enables an increase in expected yield of 

 new varieties. With a constant mean and variance, diminishing 

 returns to research hold. The effect of the discovery of the in- 

 terspecific hybridization methodology (an increase in scientific 

 knowledge) is portrayed in figure 8.1 as a shift in the mean and 

 variance of the distribution searched. This shift in the distribu- 

 tion shifts the expected discovery pattern. 



The history of sugarcane varietal development in Barbados, 

 British West Indies, provides support for the model set forth here. 



