This dissertation is about differences in communication practices across scientific fields, and how to study those differences. It explores how differences in communicative behavior of scientists can be traced back to differences in the kind of research they are doing. The focus is on one aspect of scientific communication in particular: how openly do research teams within a research specialty share scientific knowledge? This question is of particular relevance vis-` -vis a the World Wide Web's innovative potential to connect people and information worldwide. For the sciences this translates into an increased immediacy with which scientists can access and exchange scientific knowledge, as well as new ways of (re)evaluating, combining, and mining data. The methodological approach developed in this study combines qualitative (ethnographic) and quantitative (network analytic) methods. This approach supports scaling-up nuanced local ethnographic field studies to the aggregate level of research specialties for comparison between fields. Behavioral patterns are captured and quantified through structural analyses of publication networks that are constructed from the accumulated 20-year publication output of a research specialty. In turn ethnographic observations provide validation and interpretation for the quantitative measures used and help further refine the network analysis. Making use of this methodology a comparative study of two scientific communities in the chemical and physical sciences is conducted that identify a broad range of relevant aspects of research culture that feed into the field specific propensity for openness and sharing in scientific communities. Based on these findings an analytic framework is derived to support future comparative studies of openness and sharing in the sciences.