The goal of this dissertation is to advance the state of the art of research in constraint-based phonotactics. It takes a two-pronged approach: a technological contribution intended to facilitate future research, and experiments which seek to shed light on high-level questions about the properties of phonotactic models that can guide the development of theoretical work. The technological contribution is a software package called Speriment which allows experimenters to create and run experiments over the internet without advanced programming techniques. This software is particularly well suited to the kinds of experiments often run in phonotactic research, but can also be used for experiments in other domains of linguistics and the social sciences. It is hoped that this software will make it faster and easier to conduct phonotactic and other experiments as well as encourage experimenters to increase the reproducibility and transparency of their research. The experiments presented here address questions that assume constraint-based phonotactic frameworks, but that do not rely on particular theories of the content of the constraint set. That is, they apply to constraint-based frameworks for theories of phonotactics, with the first study seeking to distinguish between two such frameworks, a linear version of Harmonic Grammar and Maximum Entropy, while the second investigates whether phonotactic knowledge is independent of knowledge of phonological alternations. These coarse-grained questions about phonotactic knowledge on how pieces of phonotactic knowledge interact with each other and with another part of the grammar are intended to add to the groundwork on which phonotactic models and models of all phonological knowledge are built. Their findings have implications for which constraint-based frameworks should be used for future theories and how these theories can be reliably tested.