I see that we have lots of "indentations" on this one, which is nice.
So here I go again with my little thoughts.
First, a confession: I have not read M. Bailey, and should.
Now, given that ignorance, here are a few ignorant comments.
a) The distinction between environments and niches might prove critical. We can define niche as that (sub)part of the broader environment which is selectively used by the organism. Thus, twins and other sibs might share an environment in the broad sense, yet occupy very different niches within that environment. I can indeed see how two individuals (sibs) would move to occupy their own private spaces. The critical issue, I suspect, is the resulting similarity or dissimilarity in experience. Different niches within an environment can certainly give different experiences, and lead to divergent outcomes. Thus, the term "environment" as I lazily used it is likely to be too broad for satisfactory analysis.
b) A related point is that organisms create their subenvironments as well as respond to them. I can build a house to my own choosing. We mold our worlds throughout development. Thus any models that suggest passive reception of environmental (niche) influences are at best incomplete. Niche building as well as niche responding deserves explicit attention.
c) If twins and sibs reared apart become more similar even when separated, it would be good to know the reference groups. I can see how by some measures nearly every one within a culture appears "more similar" with time. (They know about Michael Jordon, for example.) I suspect Bailey covers this.
d) A more subtle point, indeed favoring genetic influences, is that as time goes by diverse experiences have similar effects within genotypes, just as similar experiences can have diverse effects across genotypes. This would work if experiences "select" among pre-existing (but unrealized) capacities as well as "instruct" in the old blank-slate sense.
A fascinating line of animal experiments would be to do something like this:
i) Rear animals of different genotypes in similarly "deprived" environments. ii) Then enrich the environments (another group) by providing more of a common input. iii) Design tests to see if different genotypes now diverge (rather than converge) in comparison to deprived conditions.
That would pretty well clinche the story I think.
[I once thought of a fun experiment. Imprint an animal on a potential predator (e.g. chicken to hawk). See if one day the poor prey "wakes up" and starts given antipredator responses to its former companion.]
Maybe there are data out there I do not know about.
J.
{To the potential relief of many, I am "off line" for a bit, to visit friends, colleagues and relatives in Oregon! By the way, all Oregonians look alike...until you spend time there. Wonder if there is a lesson.............?}