Bacteria growing in microchannels have to compete for space. When channels branch into networks, the growing colonies push against each other at junctions. Pellicciotta, Angelani, and Di Leonardo (arXiv: 2603.23320) discovered that this competition produces long-range order that looks exactly like ferromagnetism.
Reader note (2026-11-17): Checked against the full text of arXiv:2603.23320 (Pellicciotta, Angelani and Di Leonardo; March 2026), which I had earlier held as matching on the abstract. The headline matches: E. coli in a network of single-file microchannels, an effective equilibrium spin description with ferromagnetic coupling, and simulations of growing elastic cells that trace the coupling to internal stress accumulated at the nodes. One detail differs: the essay says the effect vanishes when the channels are widened beyond the typical cell birth size; the paper's condition is on the length of the network links, which lose their correlations when they exceed the typical cell size at birth. In the paper's spin mapping each node is in one of two states, cells circulating clockwise or counterclockwise, and the experiments are analyzed through four node spins.
Each node in the channel network has a flow direction — bacteria grow preferentially one way or the other. These directions couple through the internal stress that accumulates when growing colonies collide at junctions. The coupling is ferromagnetic: neighboring nodes tend to align, producing coordinated growth patterns that span the entire network and persist across generations. When the channels are widened beyond the typical cell birth size, the effect vanishes — the confinement is essential.
The remarkable part: despite being wildly out of equilibrium (bacteria are growing, dividing, and pushing), the system's statistics are captured by an equilibrium Ising-like model. Each node's flow state acts as a spin. The ferromagnetic coupling comes from mechanical stress, not magnetic force, but the mathematics is identical.
The through-claim: ordering doesn't require equilibrium; it requires coupling. The specific mechanism — magnetic dipoles, elastic stress, chemical signaling — doesn't matter as long as the coupling between neighboring elements favors alignment. The equilibrium framework works not because the system is in equilibrium (it emphatically isn't) but because the relevant degrees of freedom — the node flow states — happen to obey the same statistical mechanics regardless of what drives them. The universality class doesn't care about its own mechanism.
Pellicciotta, Angelani & Di Leonardo, 2603.23320. Biophysics / active matter / statistical mechanics / bacterial growth.