Mixing usually suggests smoothing. Put more of a substance in one compartment than another, allow it to move between them, and expect the difference to shrink. Now let two substances interact while they move. Can that combination make a small difference grow?
Yes, within appropriately specified mathematical models. The surprise is worth following through the rules, because it separates an explanation of how a pattern could arise from evidence that the same mechanism operates in a particular living system. An attractive pattern at the end of an animation cannot establish both.
In 1952, Alan Turing explored how reactions and diffusion could destabilize a uniform state. His paper also made the idealizations explicit, including simplified systems in which tissue growth and mechanical effects were set aside. Its two-cell example distinguished an illustrative reaction scheme from identifying actual substances and conditions. Turing, The Chemical Basis of Morphogenesis, pages 37–39 and 42–43.
We can investigate the central possibility with a small original model. It is not Turing's numerical example, a new chemical discovery or a simulation of an animal's coat. Its purpose is to make the difference between mathematical success and physical acceptance visible.
Begin with a state that settles down
Imagine two identical compartments. Each contains two quantities, U and V, near a chosen uniform equilibrium. Let u and v describe their small departures from the equilibrium values. Negative departures mean less than the equilibrium amount; they do not mean a negative concentration.
Stipulate these local reaction rules in dimensionless time:
u′ = u − 2v
v′ = 2u − 3v
A prime indicates rate of change. These are linearized rules for small departures, not a complete chemistry valid at every concentration. The first quantity reinforces its own departure and is opposed by the second. The second responds to the first and has its own restoring term. We have defined how the two changes couple, rather than simply naming one substance an activator and another an inhibitor.
Together the rules make small departures eventually decay in an isolated compartment. One way to check is to seek a mode proportional to e raised to λt, where λ is its growth rate and t is time. A negative rate makes the exponential decay. Substitute u = Ae^(λt) and v = Be^(λt), where A and B are the mode’s amplitudes. The rules become λA = A − 2B and λB = 2A − 3B. Eliminating the amplitudes gives (λ − 1)(λ + 3) + 4 = 0. Expanding yields λ² + 2λ + 1 = 0, or (λ + 1)² = 0. The repeated rate is −1.
The repeated root permits a factor proportional to time in some solutions, but that factor multiplied by e⁻ᵗ still tends to zero. A transient increase is possible; eventual decay is the claim. This distinction matters whenever a short observation is used to infer long-term stability.
Let the compartments exchange material
Add symmetric exchange. In compartment 1, the u rate receives an extra term dU × (u₂ − u₁); the v rate receives dV × (v₂ − v₁). Compartment 2 receives the opposite terms. The coefficients dU and dV are nonnegative exchange rates in our chosen units.
On its own, either exchange term smooths a difference. If u₁ exceeds u₂, the extra term in compartment 1 is negative and that in compartment 2 is positive. Exchange lowers the higher value and raises the lower one. It also conserves the sum across the compartments; it does not create material.
If both compartments have the same departure, exchange does nothing. Their mean follows the original local rules and eventually settles. To test whether a spatial difference grows, we need another description: Δu = u₁ − u₂ and Δv = v₁ − v₂.
Subtract the rates of the two compartments. The exchange contribution to Δu is −2dUΔu, since each compartment moves toward the other. The corresponding v contribution is −2dVΔv. Thus the difference rules are:
Δu′ = (1 − 2dU)Δu − 2Δv
Δv′ = 2Δu − (3 + 2dV)Δv
The factor of two is consequential. Looking only at one compartment's outgoing flow would give the wrong growth-rate calculation for the difference.
Equal rates and unequal rates give different answers
First choose dU = dV = 0.05. Both difference equations gain the same additional restoring rate, 0.1. The two growth rates shift from −1 to −1.1. These differences eventually decay. This is the smoothing comparison we expected.
Now keep dU = 0.05 and set dV = 1. The difference rules become:
Δu′ = 0.9Δu − 2Δv
Δv′ = 2Δu − 5Δv
For these equations the growth-rate condition is (λ − 0.9)(λ + 5) + 4 = 0. Expanding gives λ² + 4.1λ − 0.5 = 0. The two roots are approximately −4.2185 and +0.1185. A mode associated with the positive root grows rather than decays.
The calculation explains what changed. The locally reinforcing u departure is opposed by v. Faster exchange reduces the v difference much more strongly, changing that opposition's spatial response. The coupled system can amplify a difference even though each exchange term, considered alone, smooths it. The mean remains governed by the original stable local rules.
Unequal exchange is not sufficient in every case. Set dU = 0.1 and dV = 1 instead. The growth-rate equation becomes λ² + 4.2λ = 0. Its roots are −4.2 and zero. This case is marginal in the linear model: it has no positive growth rate. A claim that any unequal pair necessarily produces amplification has failed against a nearby counterexample.
An exactly uniform system with exactly zero disturbance also remains uniform under these deterministic rules. The growing mode describes what happens to a suitable small difference already present. It does not conjure a disturbance out of nothing.
What the calculation earns—and where it stops
The worked comparison establishes a mathematical possibility under stated assumptions. The same locally stable reaction rules can permit a growing spatial difference when exchange changes their coupling. That is a stronger achievement than pointing at a finished image and calling it a pattern.
It is still only an onset calculation. The linearized rules do not explain what stops the growth, what final concentrations become or whether a stable pattern remains. As departures become large, the assumptions can fail. A fuller model would need appropriate nonlinear behavior and physically meaningful concentrations. Neither a leopard's spots nor a leaf arrangement has been explained by the arithmetic performed here.
To turn a candidate mechanism into a physical account, a researcher would need identified quantities and independently supported reaction and transport behavior. A useful comparison would measure how a disturbance changes over time, rather than merely whether the end result resembles the preferred picture. Competing mechanisms should make distinguishable predictions under the chosen conditions.
Here is a proposed evaluation, not an experiment reported as completed. Define the system, measurement methods, admissible parameter range and what counts as a growing spatial difference before fitting the model. Compare the reaction-only behavior with the spatially coupled behavior. Reserve separate conditions for final prediction checks; do not use them repeatedly to select the parameter values and then call them independent confirmation. Include conditions predicted to stay stable, since a model that produces a pleasing pattern everywhere has lost an important opportunity to be wrong.
The fixed arithmetic examples in this article check the calculations only. The proposed evaluation would require physical evidence we have not collected. Keeping those two achievements separate leaves room for a clear judgment: the model demonstrates a possible mechanism, while its application to any particular natural pattern remains another task.
Four kinds of contribution
Geometry can specify a construction exactly, as the doubling square showed. Scale can change the demands on a boundary. Atomic arrangement can organize the interactions within matter. Spatial exchange can change the stability of a coupled process.
These are different contributions, and each needs the right support. A proof settles a consequence of premises. A model connects chosen quantities through rules. Measurements assess the connection to a system. A successful prediction under conditions kept separate from model fitting provides a further test.
The growing difference is therefore an invitation to ask a better question of a pattern. Instead of stopping at its resemblance to a familiar shape, follow the process that could produce it and look for a consequence that would distinguish that process from its rivals. An explanation becomes stronger when nature has a clear way to disagree.
Loading comments…