Fast interpretation of potential field data (magnetic data are a typical example) often uses simple geometries to describe a complex geologic reality. Many of these techniques assume that the potential field arising from the source body is homogeneous. The degree of homogeneity of a source is characteristic of its geometry. However, very few source geometries are known to generate a homogeneous field. The contact, thin sheet, horizontal cylinder, pole, and dipole all cause a homogeneous magnetic field. More complex geometries such as the thick dike or rectangular prism do not. Therefore, a major problem is to check for the validity of the homogeneity hypothesis when these types of interpretation techniques are used. The local wavenumber of a potential field calculated at a series of increasing heights above the measurement datum can be used to directly compute the depth to a source and its degree of homogeneity. In addition, the vertical derivative of the local wavenumber can provide an estimate of the depth to sources without knowledge of their degree of homogeneity. The proposed technique also allows us to test if the source is homogeneous or not, and it applies to any type of potential field data. The technique breaks down on synthetic magnetic data when anomalous sources are closer than about four times their depths. This behavior is expected from interpretation techniques that use upward continuation. The technique can be applied to profile and gridded data. Its main advantage is that it allows testing the homogeneity hypothesis and therefore the validity of the interpretation.