A. M. Turing (1952), Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 237(641):37–72 — DOI-10.1098/rstb.1952.0012
Key takeaways
- A system of reacting and diffusing chemicals can start homogeneous and stable and become unstable as its parameters drift. Random disturbances then grow into a pattern, so symmetry is broken without any pre-existing asymmetry that matters.
- In the case that matters most, the pattern is a set of stationary waves whose spacing, the “chemical wave-length”, is fixed by reaction rates and diffusibilities, not by the size of the tissue. For two morphogens this needs them to diffuse at different rates.
- The patterns are steady only dynamically. They need “a continual supply of free energy”, taken from “fuel substances” degraded to “waste products”, because diffusion continually degrades energy (p. 66).
Standing in the record
Filed on 2026-10-02 at the owner’s request, as one of the canonical systems of the dissipative-structure programme. The Nobel lecture (LIT-516) names the stationary reaction–diffusion instability of the Brusselator the “Turing bifurcation”, after this paper. Nicolis and Prigogine’s monograph (LIT-524) works the Brusselator case out.
Its bearing on what the record holds:
- It is a dissipative structure in all but name, and the paper says why. Turing writes no thermodynamics, but his §10 makes the thermodynamic point plainly: the final patterns are “only dynamic equilibria” and are maintained by free energy degraded from fuel to waste. That is the Prigogine criterion (an open system, held away from equilibrium, sustained by flows) stated a quarter-century before the Nobel lecture, for a biological model.
- The demarcation question in LIT-192. Turing’s mechanism is offered as part of how an organism gets its form, yet it is the same mechanism as a chemical pattern in a gel. That is a concrete case of what Nahas and Sachs report: dissipative pattern formation is shared by organisms and non-organisms, so it cannot be what demarcates them.
- Primitive metabolic cycles (LIT-036). Its catalytic, non-equilibrium cycle of species whose homogeneous state goes unstable is an heir of this paper’s method. NOTE-079 records that its authors set their mechanism against Turing’s and say the difference is that “it is truly the catalysts and not the reactants that self-organize”.
- The Belousov–Zhabotinsky reaction (LIT-531) and Cross and Hohenberg (LIT-527). Turing’s case (e), travelling waves, needs three or more morphogens; BZ waves are the excitable-medium kind his linear analysis does not cover. Stationary Turing patterns in a chemical reactor came only in 1989–1991 (Ouyang et al. with an imposed gradient; Castets et al. 1990; Ouyang and Swinney 1991, without one), in open reactors fed continuously with reagents. Cross and Hohenberg (§X.B.5, §XIII) review them as “confirming Turing’s original intuition”.
No instruction for machine-learning practice. The anthology does not hold it.