Science as an input to patented inventions is a fundamental of economic growth. However, our understanding of how science is transformed into patents is limited. In the present paper I seek to fill this gap by examining the micro-foundations of science-patent transformations. Using an inductive, grounded theory approach to study the transformation of 12 scientific discoveries into patents I recast the relationship between science and patents: I show it as a particular process that affects patent breadth. Exploiting surplus patent breadth depends on the processes of abstraction and cognitive variety, which can be mobilized by patenting experts. The theory is tested using a recently published algebraic interpretive method for examining causal relationships in small-N studies.