habitat in repeated regression problems (table 4) . The final line represents the best 

 model of the series. Coefficients derived from this solution are given in table 5. 

 The ratio of the mean-square deviation from regression to the mean-square deviation 

 about the mean is 0.2722 for this model. That is, the model accounts for about 73 

 percent of the variance of logarithm of height increment among the 1,165 trees in the 

 sample. Converting the residual error to the scale of feet of height increment 

 (Furnival's index), the mean-square error of estimate would be ±1.97 feet. 



To illustrate the implications of this functional form for height increment in 

 the context of stand development, the successive height/diameter curves were plotted 

 for a typical even-aged stand of grand fir carried through four decades of simulated 

 growth (fig. 4) . The curves in figure 4 show that the prediction functions for height 

 increment derived in this report can generate successive curves that conform to the 

 general shape and level of typical height/diameter curves in even-aged stands shown 

 in figure 3. Conformation of these curves for simulated stand development also depends 

 on the way diameter increment and mortality change among diameter classes within the 

 stand. 



Table 4. --Effect of varying coefficients by species and habitat on the mean square 

 residual of the logarithmic form of the differential model 



Number of : 







Variable 







: Me an 



coefficients : 

 in model : 



1 : 



ln(AD+.05) 



: In (D) 



: In (H) 





: square 

 : residual 



22 



H 



* 



* 



H 





0.2045 



31 



S 



* 



H 



H 





.2029 



32 



S+H 



* 



* 



H 





.2008 



32 



S+H 



H 



* 



* 





.1958 



40 



S 



H 



H 



H 





.1875 



41 



S+H 



H 



H 



* 





.1853 



Key: * indicates 



a single 



coefficient for all species 



and habitats. 







H indicates 



a unique 



coefficient of 



the indicated 



variable for 



each 



habitat . 



S indicates 



a unique 



coefficient of 



the indicated 



variable for 



each 



species . 



13 



