I was using a balun wound on an Amidon FT114-43 torroid, on opposite sides.
Since then, someone posted impedence_matching_bryant.pdf or similar (was it
DxAce?). Anyways, I wound another one with the overlapped method, and I
must say, performance seems to be much better, especially between 5000 and
10000 khz. I was able to pull many stations in last night, despite
increased solar activity, including many that I had never received before.

So thanks, to whoever posted it. And I can say, from my impressions only,
with absolutely no evidence, that the overlapped winding method is better
than the opposite-sides method.