Current techniques for analysing growth are compared using data from young plants grown in artificially-lit growth cabinets. A statistical method of fitting curves to log dry weight and log leaf areas led to values which often diverged from biological expectation. It was more satisfactory to make the calculations assuming that a quadratic relation always applied. It appears that the statistical methods cannot be used unmodified for experiments involving comparisons between artificially-lit cabinets. Cabinet environments are seldom replicated and therefore the consequences of isolated anomalies in either the environment or in a particular harvest are not allowed for and are reproduced in the final curves. In addition the statistical method assumes that all the growth curves are entirely independent, whereas there is a tendency for the opposite to be true.