Investigating random homicides involves constructing models of an odd sort. While the differences between these models and scientific models are radical, calling them models is justified both by functional and structural similarities. Serial homicide investigations illustrate the marked difference between theoretical models in science and the models applied in these criminal investigations. This is further illustrated by considering Glymourian bootstrapping in attempts to solve such homicides. The solutions that result differ radically from explanations in science that are confirmed or disconfirmed by occurrences. Unlike the scientist, the flatfoot gumshoe is also barefoot: he is bereft of a general, determinative theoretical frame. This result shows that criminal investigations do not apply ‘science’ in the Galilean sense.