54 Agriciiliural Research and Prochictivity 



changes in percent of acreage planted to varieties that have in- 

 creased their share of planted acreage. If the percentage of total 

 acreage planted to a new variety increases from one year to the 

 next, a concurrent decrease in the percentage planted to other 

 varieties will occur. The coefficient estimated for the variety 

 change variable measures the yield advantage of the new varieties 

 over the replaced varieties. The results of the statistical analysis 

 are presented in table 3.7. Fertilizer and rainfall were not available 

 in all cases. The variety turnover variable has the expected posi- 

 tive sign, and the intercept term has a negative sign. A negative 

 intercept term reflects two phenomena, a natural deterioration of 

 yields due to disease and other factors, and changing age distribu- 

 tion. As the South African data show, the younger the cane, the 

 higher the yields. The varietal change measure may be correlated 

 with sugarcane age in these data. A consequence of such a corre- 

 lation would be that the intercept term is biased downward and 

 the varietal change coefficient has an upward bias. Although ma- 

 jor changes in average age do not appear to have occurred over 

 the period, we do not have cane-age distribution data that would 

 allow checking this point. 



The data in tables 3.6 and 3.7 allow two calculations. First is the 

 economic value of the experiment station activity in speeding up 

 stage 3 varietal diffusion. From table 3.6 we can compute the 

 speed-up or increase in turnover achieved from an investment of 

 $7,000 (approximately the cost of adding a senior researcher to 

 the staff in 1936). This increase would be approximately 0.1 1 per- 

 cent per year. The value of an increase in variety turnover of 0.11 

 percent computed from table 3.7 is approximately 0.08 tons per 

 acre in Australia and 0.02 tons per acre in the Caribbean and 

 South Africa. The total value of this speed-up, resulting from a 

 $7,000 investment, would be about $50,000 in Australia and from 

 $12,000 to $15,000 in South Africa and the Caribbean area. A rea- 

 sonable downward adjustment for bias resulting from changing 

 age distribution would reduce these estimates by one-third to 

 one-half.'^ It appears that the "extension-type'' side benefits 

 from accelerating stage 3 varietal diffusion justified much of the 



12. The South African yield by age data (n. 10), indicates that a 1 percent in- 

 crease in plant cane over and above the normal increase needed to maintain the 



