Plant growth-analysis: further applications of a recent curve-fitting program.
Abstract
The authors had earlier [see HbA 44, 3551] described a computer program in which plant growth-analysis quantities of the general form (1/Y)(dY/dX), Z/Y and (1/Z)(dY/dX) were derived from simple curves. These were fitted to the logarithms of sequential estimates of measured variates Y and Z, where X is the independent variate (normally time). The dangers of over-fitting were stressed and it was concluded that, where values of the above quantities were needed, the best approach was to fit the lowest order of polynomial possible. The exclusion of trends more complex than lines or simple curves was acknowledged. This paper demonstrates how these methods can be extended to deal with more complex cases without recourse to the fitting of more complex curves.