A seven-part exercise is described, based on published accounts of a school epidemic. Students have to complete each part before receMng the next: cumulative mistakes are therefore avoided. Analysis over four years shows consistent patterns of student responses. Details of statistical tests of the data are given.