Puzzle-solving, like several other everyday activities, appears in a more sophisticated and ramified form in the realm of natural science. Improving on Thomas Kuhn's rudimentary account of puzzles in science, this paper formulates logical and functional criteria for the occurrence of scientific puzzles, and examines the two-fold nature of their solutions. Then, with the aid of erotetic logic, puzzle-posing questions are identified, their presuppositional relations to scientific theory and explanations are explored, and a new tool for history of science research (...) is developed. (shrink)