Curlitron

[Creativity]

few observations then that may lead to the non conclusional experiment results:

Is metal loop connected to the box? It will then be a parallel current flow, one via wire and one via the box. If not it will produce a simple capacitor and won't return the lost charge. Obviously, if the box will catch the electrons that have escaped it won't stop current from decreasing, because the electron will "leak" on the way not reaching the destination. To have a good measurement in this experiment we have to keep that in mind (even if the effect works, experiment may show no positive effects just because we forgot the lost charges in equation..)

In flowing current the hole may be replenished by the electron flowing by and not the electron that almost escaped. If not so then I guess the electron on its way back will reach the surface a bit away from the hole so it will return back to the hole via the wire and just against the positive influence it has just made. OK i just gave it a fast thought so maybe I am next to it. But if 'm right we would get some current and voltage fluctuations that a diode may sort out to count in one way only. 

I think the best would be to put the wire next to the magnet and no current at all. Then if an electron gets thermally kicked off there the part of e-field (between the hole and the electron) will be perpendicular to the m-field and some flow of current will be observable. The drawback will be that only some directions of thermal knocking will be utilised, giving much less voltage and current to observe.