The Direct Current Potential Drop Method (DCPDM) is a frequently used technique for crack detection and crack length measurement in fatigue experiments. Recent investigations have shown that with multiple potential drop measurements the location of crack initiation can be determined in single edge notch specimens (Wiehler and Bar, 2020), (Bar and Nahbein, 2022) and in round bars (Hartweg and Bar, 2019). In the present work a more detailed investigation of crack initiation and propagation is undertaken on round bars to advance the geometrical model by Hartweg and Bar (2019). Three potential probes were equipped on notched steel bars - at 0 degrees (U-1), at 120 degrees (U-2) and at 240 degrees (U-3). The three potentials were measured simultaneously during the fatigue tests using amplifiers of the control electronics. Moreover, the introduction of overloads was used to mark the crack front on the fracture surface in defined intervals to compare crack and size calculated from the potential drop measurements with the real crack size and geometry. In the case of an initiated single crack, the experimental set-up delivers different potentials depending on the distance of the probes to the crack. The geometrical model can be used not only for the determination of the crack angle like performed by Hartweg and Bar (2019) but also to calculate the crack size. On the measured data a function is fitted that allows the calculation of the crack length independent of the location of the crack initiation site. The experimental results have shown that it is possible to determine a crack with a size of less than 1 % of the specimen cross-sectional area. (c) 2022 The Authors. Published by Elsevier B.V.