>Edward also made it clear to me that I am didn't describe that aspect clear enough. Rreading my text, I understand that.
>
>The resulting table should be a concatenation of the 4 tables, with all duplicates removed, but that should happen in an evenly distributed way, like the numerical examples displayed.
If a random pick with probabilities assigned in your proportion would do, then... something like this:
Let's say you have a set of four duplicates, one from each table, and the probabilities are 0.1, 0.2, 0.3, 0.4. Then just add them cumulatively, so the cumulative probability for the first one is 0.1, for 2nd 0.3, for 3rd 0.6 and 1.0 for the last. Then
locate for probability>rand() while key=lcKey
and use the record that you land on, and discard the others. This isn't perfect, and requires some legwork beforehand (got to fill those fields and got to total the probability, then divide it by total probability for each key), but I think it'd get you close to what you want.