ABSTRACT
In an experiment on the effect of nitrogen and potassium upon the growth of cauliflowers, four levels of nitrogen and two levels of potassium were tested: Nitrogen levels and Potassium levels. The response variable is a frequency distribution of cauliflowers of various sizes. After preliminary inspection of the data, the first step is to form one or more derived response variables in terms of which treatments can be compared. A technical point in the analysis concerns the design used and the estimation of contrasts and error. In the analysis the dependence on nitrogen level has been assumed, for simplicity, to be locally quadratic. It would be possible to fit a more complex functional relationship, such as one representing growth to a limit, which would be more plausible on general grounds. The nitrogen linear contrast NL is significant at the 0.1 per cent level.
