Richard, your citation much appreciated, but I can't find it either in WorldCat or in the article databases accessible to me. But I do see Weber and Depew's, Evolution and Learning: The Baldwin Effect Reconsidered (2007), which might serve me just as well because it includes the history and changes in the idea, including its revival and interaction with the modern synthesis.

However, the rather abstract and philosophical question I've been raising does not speak for or against the Baldwin Effect. Roughly, it is a methodology that assumes that a) in principle all things are processes, b) (more contentiously) explanation must be in terms of processes rather than, say, a description of a causal mechanism uniting entities. To represent things as processes I have argued that it requires a unification of three modal aspects: a) structure (the local and observable effect of the past), b) real exogenous (all levels are seen as constraints of a structure on the possibilities of a more universal level) possiblities that are constrained by structure as a probability distribution, and (Morowitz would insist) potency, which I define as a probability gradient to anchor it to structure.

For example, a gene can be described, but to use it in explanation requires that it be seen as part of a broader process that engages unobservables: to see it as a structure (the local and observable outcome of the past), coupled with a probability gradient (to drive change), coupled with possibilities (inherited from the more universal level on which the living organism represents a constraint). Now, all this is obviously obscure and may seem gratuitous, but the implication is not so unconventional: the gene is only a (local, observable) structure that is only one aspect of a broader process, and this broader process is the unit of analysis necessary for explanation.

For example, this suggests that whether "information" can pass both ways across the Weissman Boundary is a false question, for it reduces a process to the causal or functional relation of the (static, local, observable, closed) entities of genotype and phenotype.

I admit that I've been imposing some vague and general conceptual garbage on rather specific empirical questions, and this does tend to invite trouble. However, I beg everyone's indulgence, for I'm not trying to reconstruct microbiology, but merely exercising a methodology by trying it out in a field entirely unrelated to the field for which it was developed.

Haines