A central goal of science education reform is for students to participate in scientific sense making rather than to merely acquire science facts. However, even in classrooms utilizing reform-based pedagogies, students are typically allowed to construct knowledge only insofar as they construct expected knowledge. In this report and reflection, we use activity theory to demonstrate how this pervasive tension between learning correct ideas and constructing one's own ideas often results in unacknowledged slippage between competing activity systems within reform efforts. We use an analogy to the domain of spelling to introduce invented science--a framework for describing the activity of science learning that reduces this slippage by giving knowledge construction true priority over the canon. We describe the origins and purposes of invented spelling to theorize the nature of learning in invented science. We conclude by articulating the theoretical and practical implications of this analogy for science teaching and learning.