Visual appraisal of all departures shown in figure 8, one-third retention, sug- 

 gests that the form of the original model is especially appropriate in the midrange 

 of actual retention. At the lower extreme, average negative bias in estimated reten- 

 tion maximizes at perhaps 1 percent and average positive bias at 1.5 percent at the 

 upper extreme. So the original model, scaled to the data set from each of two new 

 areas, performed reasonably well when evaluated jointly for these applications. 



This limited check, at least, shows some promise of the model form being appli- 

 cable for Batholith soils in general. Additional support for this thesis is present 

 in table 4 where the Sy (standard errors of estimate) are fairly consistent. Note 

 that Sy,x - 2.1 for the original model is apt to be an underestimate since only two 

 degrees of freedom were charged against the fitting process. Actually an unknown 

 number of degrees of freedom were sacrificed in graphically exhausting the original 

 data set of information. 



Applying the same evaluation as above to the 15-bar retention model, an average 

 positive bias of about 1 percent in estim.ated retention exists at the lower extreme of 

 actual retention, an average of a negative 1 percent at the upper extreme. Although 

 the magnitudes of bias do not seem critical to conventional field applications, the 

 form of the original model seems consistently different for the two new data sets. 



12 



