In this study we examined the 638 annotations in 286 threads made by 64 undergraduates in one chapter of a physics textbook. We developed a taxonomy of annotation types that permits us to reliably classify the individual annotations in each thread. We next looked for relationships between the annotations and the studentsâ prior knowledge and self-efficacy. Finally, we examine the amount of knowledge sharing in each thread. In our presentation we will show the interaction between students in an online annotation system can lead to collaborative knowledge construction.