A common format for assessment of learning is pretesting and post-testing. In this study, we collect student test data several times per week throughout a course, allowing for the measurement of the changes in student knowledge with a time resolution on the order of a few days. To avoid the possibility of test-retest effects, separate and quasi-random sub-populations of students are tested on a variety of tasks. We report on data taken in a calculus-based introductory E&M class populated primarily by engineering majors. Unsurprisingly for a traditional introductory course, there is little change in many conceptual questions. However, the data suggest that some student performance peaks and decays rapidly during a quarter, a pattern consistent with memory research yet unmeasurable by pretesting and post-testing. In addition, it appears that some course topics can interfere with prior knowledge, decreasing performance on questions related to earlier topics in the course.