@timhutton Hi Tim!

Cells reproduce normally, albeit a bit slower than in your original Squirm3 because I made the arena 125 times bigger by area (5000x3000 vs your 400x300), so they encounter soup less often. To counter, I added a water droplet feature that acts as a "soft jar." These circular droplets trap cells inside as well as soup atoms. The water also allows for more Brownian motion than in dry zones.

As for randomness, it's on. Your cases’ stochasticity is intact. In the save files that capture the run, the atom histogram is a/b/c/d/e/f only.

As for the bounding membrane, I observed that it has a single-cell origin.

One cell's membrane breaks open because of a failed division. The membrane opens, leaving the rest of the bonded atoms exposed but still intact. The membrane then closes again, but not around its "organelles," but rather around an empty space. Topologically, the cell turned itself inside out, like an inverted sock.

The gene strand is still alive, still bonded to its anchor atoms, except now those anchor atoms are on the outer surface of the new big ring. The strand divides normally. The pulling sequence has nowhere to go but the only empty space around, which happens to be the interior of the new ring. Daughters land inside, divide again, and populate the shell (the outer organelles of the bounding membrane can still produce more cells of their own, pumping into the space enclosed by the bounding membrane).

I've attached a video illustrating this process, you can observe it around the 0:15 mark.