This paper proposes a new tool for understanding how the physical sciences have contributed to the population density, economic productivity and military power characteristic of industrialized societies. I describe a set of criteria historians can use to identify “infrastructural sciences:” sets of highly routinized, rarely celebrated technical practices that are central to industrial life. To reliably deliver water, food, energy, goods, and information, technical workers must continually adjust large technological systems to the changing external environment; operating the infrastructures that environmental historians call “second nature” requires monitoring “first nature.” This surveillance turns geographically-distributed, standardized observations into formulaic reports and forecasts. Because of the discipline and expense required to produce these public goods, most modern states pursue environmental surveillance as a basic aspect of governance — like national meteorological services, hydrologic monitoring, or geological mapping agencies. Though the infrastructural sciences are culturally modest (they make few claims about what it is to be human or our place in the universe), they do have a complex relationship to theoretical science. Academic researchers guide what and how networks monitor, and they often make theoretical claims using data repurposed from industrial use. This paper draws on examples from my own research in aeronautical meteorology, as well as Simon Schaffer’s and Bruce Hunt’s studies of 19th century electrical metrology and Michael Reidy’s work on tidal science. By drawing our attention to routine observation in the earth and physical sciences, infrastructural science reveals the importance of institutions, scientists, and work practices rarely noticed in existing historiography.