We review measured and calculated voltages and cathode-fall thicknesses against current and electrode separation for parallel-plane Ar glow discharges in the above-normal, or abnormal, current range. We consider dc and pulsed discharges, but not radio-frequency discharges. Reasons for the very high degree of variability among experiments and models are reviewed and suggestions are made for improving the reliability of both. Variations of a factor of two in experimental cathode-fall voltages at a fixed normalized current density are attributed to differences in the cathode emission properties, varying electrode separation, and effects of discharge tube walls. We explore a wide range of normalized current densities, models of the ionization coefficient against E / n, and the effective electron yields per positive ion reaching the cathode using simple scaling laws. Ionization source terms for local-field and non-local-field models are compared and used as the basis for a simplified non-local model. We discuss the present author's recommendation that, because of current large uncertainties in the electron yield per ion for practical cathodes, the yield used in cathode-fall models should be treated as an unknown. Recent models successfully predict the behaviour of transient abnormal discharges on the microsecond time scale.