Great analysis, Eoin. That really describes it well.
The problem is that (with smaller wheeled wagons) the wheel nearest the frog falls into the gap/well just before the point of the frog. When that wheel hits the frog point, it meets resistance, momentarily 'stopping' the wheel. A moment-arm through the centre of gravity of the wagon (generally in the center of the wagon on relatively straight track) causes the axle to rotate toward the 'stopped' wheel/frog which (if unchecked) will allow the opposite wheel (on the stock rail) to rotate toward the check rail and ride up onto it - derailed.
This also aggravated the wagon 'tilting' toward the frog.
If the check rail is the correct distance (close enough to) the stock rail, it provide a counterforce on the inside of the flange of the wheel on the stock rail, keeping it in line with the rail and forcing the 'fallen' wheel to remount the frog point and keep rolling.
If the space between the stock rail and check rail is too wide counter pressure cannot be applied to the inside of the flange on the stock wheel flange to prevent the axle rotating. Eoin's solution exploits that by narrowing the gap between the stock rail and check rail (on each road) allowing the counter pressure on the stock wheel to occur. This suggest that the distance between the stock and check rail on the Hornby points is too wide opposite the frog. Does anyone know if the Hornby point stems from an era when wheels were a cruder design with deeper flanges and the point design was never improved upon?
As Eoin points out, the deeper the well the greater the probability of the a problem. So this would be less of a problem with a Code 75 point than Code 100 (which is all that Hornby make, I think). This seems of be less of a problem with Peco points, so they obviously have the check in the correct position regardless.
The problem does not occur with larger wheeled rolling stock as the wheel crossing the frog is large enough to ride over the gap and frog point without falling and attempting to re-mount the frog point.
I suspect that this is also a problem for lighter wagons and maybe bogies more than fixed axles. Since I don't have a layout, this is somewhat speculative so please correct any points that are incorrect.