I created a spreadsheet in which I intentionally solved for the minimum groups of 5 for any number of participants. I found that, for group sizes 20-102, it was possible to have the percentage of people in groups of 5 decrease as the group size got larger.
I'm not sure if you and I are talking about different things, or if my math is wrong, or if I've seen something you didn't see. But I'll include the graph here, and I'll share the spreadsheet with you by email.
I'm very interested to hear your comments about this!
RE: An Analysis of the Genesis Fractal's Group Structures and Random Processes