This is the second in a set of related reports on (1) nonstable behavior of widely used ionization gauges, (2) causes of nonstability and nonreproducibility in widely used Bayard-Alpert (BA) gauges, and (3) a stable and reproducible BA gauge design with approximately a tenfold improvement in both stability of gauge calibration and reproducibility gauge-to-gauge compared to older ionization gauge designs. Computer simulations of electron and ion trajectories are utilized to illustrate a variety of causes of changes in BA gauge sensitivity with time of operation and in reproducibility gauge-to-gauge. Both computer simulation and experiment confirm that the ion production and collection rates in a BA gauge are far more sensitive to small changes in electrode geometry and potentials than has been commonly realized. Any change in geometry or potential of any electrode or insulator in the vicinity of electron or ion trajectories will affect gauge sensitivity as will any change in electron or ion space charge. Changes in the distribution of emission along the cathode will affect sensitivity if large axial field components are present in the anode volume. If these physical changes occur from gauge to gauge, the sensitivity of the gauges will differ, and this will cause them to give different pressure indications when exposed to the same actual gas density. If the changes occur with time of operation in a given gauge, the gauge sensitivity will change with use and the pressure indication will be in error even if the gauge has been carefully calibrated. This new understanding has been used to develop a significantly more stable and reproducible BA gauge.