Long-term vegetation dynamics based on paleo-pollen data display transient behaviour, often alternating in phase between predominant determinism and predominant 'turbulence', when viewed as a trajectory in a multivariate phase space. Given this, the metaphor of vegetation dynamics as a 'flowing stream', first introduced by Cooper in his classic 1926 paper entitled "The fundamentals of vegetation change", is re-examined and revealed to be not only useful, but strikingly realistic. Vegetation dynamic theory is reviewed and classic theories are found to reflect reality (...) poorly. It is suggested that vegetation dynamics is a far from equilibrium system, and that the application of nonequilibrium thermodynamic theory is appropriate. (shrink)