Abstract
Using an 85 x 25 matrix of annual maize yields in which the rows are magisterial districts the columns are years, principal components analysis has isolated a set of uncorrelated maize patterns over South Africa. Three of these patterns have been examined over time and it has been possible to define years in which annual rainfall anomalies have critically affected maize yields. In addition, the analysis has shown that in certain regions of the country, maize yields are more sensitive to variations in rainfall than others.