My Ph. D. dissertation concerns one of the major issues in the rhetoric of science, the function of metaphors in scientific discourse. It explores the relation between knowledge production and a metaphorical kind of language that posits, specifies, and guides research on what is as yet unknown. An appropriate term for this kind of language is quasi-metaphor; its special function is to project the characteristics of a known referent within an established epistemic category to a virtual referent in a known or unknown epistemic category. My goal is to unite two opposite views about scientific language: one held by positivists and scientists, that scientific language must be clear, brief, and trope-free; the other held by post-structuralists and epistemologists, that scientific language is ultimately metaphorical, and it thus never has had the precision claimed by the positivists. I argue that scientific language is and must be simultaneously precise and open-ended in order to articulate knowledge as well as accommodate the unceasing production of knowledge. My methodology is to establish a theoretical frame in negotiation with metaphor theories and then apply it to case studies.