More policing reduces crime but little is known about the mechanism. Does policing deter crime by reducing its attractiveness, or because it leads to additional arrests of recurrent criminals? This paper provides evidence of a direct link between policing and arrests. During shift changes a peculiar redeployment of police patrols belonging to separate police forces disrupts policing and lowers the likelihood of clearing robberies with an arrest by 30 percent. There is no evidence that criminals exploit these dips in police performance. A back of the envelope calculation suggests that incapacitation explains 2/3 of the elasticity between robberies and policing.