The differences in the size of coefficients between the uncut and cut data sets were then 

 modeled as a function of time-since-last-cutting (TIME) , expressed as the number of 5-year 

 intervals since the last cutting. The uncut virgin data sets were given a value of 60 (or 300 

 years) . I hypothesized that the transition from "cut" regression coefficients to "uncut" 

 regression coefficients would be along a smooth path, and therefore three sigmoidal curve 

 forms were developed using MATCHACURVE techniques (Jensen and Homeyer 1970). All three curves 

 are zero when TIME is zero and one when TIME equals 60. The differences in the three curves 

 are their forms (fig. 9), and they were chosen more or less subjectively. Because the TIME 

 data fell at both extremes, I felt that a more objective approach was not possible. The three 

 equations are: 



- -0.244178 + 1.244178*EXP(-(1. 176471 - . 01 9607*TIME) 2) 

 A^ = -0.095336 + 1 . 095336*EXP (- (1 . 25 - . 020833*TIME) "+) 

 A^ - -0.000203 + 1.000203*EXP(-(1 .428571 - . 023809*TIME) ^) 



where 



TIME = time-since-last-cutting expressed in number of 5-year growth periods. 



Time Since Cutting in Five Year Growth Periods 



Figure 9. --Time-since-last-cutting sigmoidal curves. 



59 



