BTW, in case some readers don't know how the linear fitting procedure (I described earlier) for the logistic function works, I probably should mention some more detail. It's fairly easy to see. Note if we start with
y_n = y_∞/(1 + (1 + g)^(n* - n))
then some simple algebra yields
y_n/y_(n-1) -1 = g*(1 - y_n/y_∞) .
Note the RHS has the form of a line whose independent variable is the set of y_n values and which has a vertical intercept of g and a horizontal intercept of y_∞. So plotting y_n/y_(n-1) - 1 vs y_n yields a line with the aforementioned intercepts whenever the logistic model actually does represent the data reasonably well. That is all there is to it.