Abstract: | This article describes an empirical investigation of the extent to which incorrect arithmetic algorithms persist over time. The use of subtraction procedures by children aged 9–11 was charted over time periods of 1–3 months. The ways in which several bugs might be related to a single misconception were considered and a set of “bug classes” was identified. The results indicated that the prevailing tendency to attribute stable procedural knowledge and systematic behavior to children of this age is misguided. They provide some insight into the development of mathematical knowledge under the influence of conventional instruction. |