This research examines scientific productivity among authors in natural scientific reference groups. A broad literature review surveys models of knowledge production, including measures of scientific products at different levels of abstraction. Data is drawn from authors in ten specialty areas, elites identified by co-citation analysis. These co-citation clusters are analyzed in general, in disciplinary sets, and as specialty groups. Results show that variance in the productivity of elite authors is not predictable on the basis of stratification variables. Descriptive differences in disciplines and specialties reflect contextual diversity in the social production of scientific knowledge. Differences in average annual paper publication, citation and highly cited paper publication do not correspond to differences in career age, job sector or prestige of highest degree. In general, stratification by experience and affiliation is not reflected in the variation of bibliometric measures of scientific productivity. This suggests that co-citation clusters are partially comparable to general populations of science, since author productivity is not simply predicted on the basis of social stratification for either type of population. Co-citation cluster authors are heterogeneous, like scientists in general, and their bibliometric differences do not correspond to variation in experience or affiliation. / Master of Science