In the past decade, several arrays or lines of magnetometers have been used in order to help monitor the behaviour of external current systems. In most cases, the basic approach consists in fitting model currents with adjustable parameters to the data, a technique which has the severe disadvantage of restricting the number of parameters allowed to vary. Consequently, it is common practice to approximate the induced fields, which always contaminate ground-based magnetic measurements, by those of a perfect conductor located at some depth below the surface of the earth, this type of representation having the advantage of being solved extremely easily (even in a spherical earth) by image theory. Many careful semi-quantitative analyses of the region in terms of its electrical properties are usually carried out before the data recorded are investigated for their external contribution (e.g. KuPPERs et al., 1979), but it may happen that the depth to the equivalent perfect conductor is arbitrarily fixed without consideration of the source field. This latter procedure is never justifiable since the depth is dependent on the frequency and-to a lesser extent-on the wavenumber content of the source as well as on the average electrical properties of the region. It is also worth remembering that the definition of a single perfect conductor, fixed at a given depth during the model fitting procedure, implies that the source consists of a single dominant periodicity. A random event, such as a substorm, contains a broad spectrum of energy and may have to be analysed by initially passing the data through a narrow band filter before the calculation is done.The determination of the depth at which to locate the perfect conductor so that it will generate field amplitudes similar to those observed at the surface of the real earth is always easy to accomplish once the periodicity content of the source is determined (e.g. see Fig. 1, after MARESCHAL,1981). However, it is essential to remember that, since the image of a real source field by a perfect conductor is real, this representation loses all phase information. In order to evaluate the severity of the simplification, MARESCHAL (1976) compared synthetic electrojet fields induced in a flat earth either of finite conductivity varying with depth alone or of perfect conductivity. She showed that for an electrojet oriented along the E-W (Y) axis, phases (or time lags as she put it) over a typical section were totally negligible for the X -component but more important for the Z-component. Depending on the location of the observer, the