Abstract
The analysis of Krieger and Laufer is shown not to apply to semiconductors. States in the gap may appear in the tight-binding model for vacancies and surfaces. Moreover, states will always appear for nonpathological band structures.