I sorta figured that if I raised the signal I'd hear from you. My use of your name was entirely deliberate, Lee.

We're interpreting the same set of facts differently. When I talk about circular flows, I'm referring to the process as you describe it of populations going off and becoming isolated for a while, and then some representatives of that population traveling again and rejoining the original gene pool or merging with another one. The period of those cycles is a function of our average migration rate and mutation rate, and show up as harmonics in this cyclical flow.

A classic example is the branch of humans that settled in Australia 40,000 years ago. There's strong evidence that about 30,000 years later some of those people discovered boat-building, and returned east along the Indian Ocean over the course of a few hundred years, to arrive back at the east coast of Africa. Their genes, as well as a few characteristic cultural artifacts associated with latter-day humans (I think it was the sudden introduction of yams into African agriculture that was the marker), can be traced radiating from the coast into the interior again. The bottlenecks aren't as severe as you portray them because the invention of boat-building opened up many alternate routes. Starting about 10,000 years ago, migration by sea as well as land became possible. The Mediterranean was the first barrier to drop to the simplest earlier boats, and over time the Indian Ocean and then the Atlantic and Pacific ceased to be barriers to the migration of genes. The entire coastline of Africa eventually opened for genetic migration.

Another route to consider is the "accidental rafting" scenario. While it's true that the odds of a mating pair of hominids surviving an ocean journey clinging to debris washed out to sea in a storm are vanishingly small, consider how many years we have to work with and how often it may have happened. Such a pair would only have to survive the journey once to establish a line of hominids far from home.

If the flow of genes were truly fluid, then you'd be right about there being little variation possible. A contiguous population reinforces the main blood line rather than varies it. But because of the pattern of migration and return, the flow isn't fluid, it's cyclical, and that makes all the difference.

The cyclical model is more powerful than the simplistic unidirectional branching model because it creates muiltiple opportunities for recombination to produce stronger hybrids, which are then reradiated outward for another cycle of mutation and recombination. The view of Africa as not the headwaters but the center of the whirlpool I think better explains the rapid development of humans, especially the exponential character of our latter phases. You don't get syngergistic effects with unidirectional radiation, and the curve of human development shows that powerful synergistic effects were at work.

Then there's the whole notion of blood types and Rh factors keeping successful intermixing to low levels. That probably goes along way in explaining the extremely high mortality rates for both mother and child in the past. It's a problem that was of major concern up until very recently with the advent of modern medicine.

And isn't that direct evidence of the remixing process? An isolated population wouldn't have characteristic problems whose symptoms sound very much like a local gene pool adapting to the injection of fresh and competitive genes. Or maybe those could be interpreted as the symptoms of inbreeding, in which case we know the cure: Genetic mixing.

Mixing the genes trumps genetic isolationism every time. The faster we mix our genes, the faster we advance.