Relevant mathematical properties of the Richards function are described, along with references to its employment in previous work, and a brief description is given of some of the difficulties encountered in its use. Two introductory experiments were undertaken: firstly, applying the Richards function to leaf growth of Impatiens parviflora DC; and secondly, applying the function to artificial sets of data of differing variability and differing spreads of observation‐values along the growth curve. The purpose of the latter analysis was an attempt to elucidate some problems of fitting and interpretation arising in the first experiment. It was concluded that the prime aim in experimental work generally is the acquisition of data covering as large a range of sizes as possible; low variability of the replicate observations, while desirable, is not so important.
In another series of experiments, plants of Impatiens parviflora were grown at four different temperatures in the range 13–23°C inclusive. In terms of leaf dry weight, increase in temperature decreased the final size attained and increased the mean relative growth rate; the curvature of the growth curve and the mean absolute growth rate were, however, temperature independent. These results are brought together in a simple model which shows clearly how temperature affects the course of leaf dry weight increase in these plants. The results for leaf area are less clear, but both the mean absolute and relative growth rates increased with rise in temperature. The final area of a leaf and the curvature of the growth curve seemed to be little affected by temperatures in the range 16–23°C, but there was some evidence of differences in these features in leaves growing at 13°C. No model summary was attempted for leaf area increase.
SUMMARY
The question of which degree of polynomial should be fitted to plant data, for the purposes of growth analysis, has been considered in relation to the variability of the populations sampled. By means of actual and semi‐artificial data, it was shown that the variability of the samples taken at each harvest had a profound effect on the results of the tests of significance, when the relationship was fitted using individual observations. Significance tests performed on data of low variability resulted in over fitting, i.e. unrealistically high degrees were indicated as being most appropriate. Conversely, data of high variability were adequately fitted by lower order polynomials. In view of this, it is suggested that growth data should be fitted using harvest mean values, since in this case, the test of adequacy of fit is independent of the underlying population variability.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.