This paper is concerned with the application of the ACPD method to the measurement of surface-breaking cracks at the interface between an austenitic and a ferritic steel in a transition weld. The ACPD method, shown schematically in figure 1, involves the establishment of a uniform thin-skin field which is used to interrogate a surface-breaking crack. Measurements of surface potential difference made before and across a crack are then used to infer its depth through a simple one-dimensional formula derived on the assumption that the material is homogeneous and that the field remains uniform on the metal surface and the crack faces [1]. The question of how to interpret readings from an ACPD gauge used on a transition weld arose after preliminary measurements had been made on specimens with spark-eroded rectangular notches of different depths and of aspect ratio 2 at a range of frequencies [2]. The one-dimensional interpretation of voltage readings was found to be inadequate, but sensible estimates of depth were nevertheless obtained on the basis of a very simple theoretical model of the effect of the discontinuity in material properties on the surface field measurements, coupled with the application of a single correction factor for the effect of the aspect ratio of the flaw. In reality the aspect ratio correction, which is known as a Multiplier [1], can be expected to depend on the size of the probe relative to the size of the flaw, but the value used in those tests was certainly of the right order of size. The encouraging results of these tests, which were performed at Rolls-Royce & Associates, Derby, U.K., indicated overall that the method may have important application in this study. In furthering this work a specimen with a weld-induced surface-breaking crack of higher aspect ratio was manufactured and the construction of a more detailed model of the surface field which is being explored in the measurements was put in hand in order that a rational procedure for the correct interpretation of instrument readings could be formulated. This paper compares test data obtained on the manufactured crack with predictions of crack depth obtained from both the simple theory and from the more elaborate models. Details of the theoretical developments will be published separately [3]. This initial work considers only defects of large aspect ratio