Fig. 2

Growth and structure of the active–passive interface in the bacterial swarm. Active–passive domain boundaries are designed with different initial shapes by varying the geometry of the aperture, such as an octagon [O] (a–e) or half-space [H] (f–j). a The interface position changes shape as it moves over time. b The interface position is identified by order parameters based on fluctuations in image intensity, ϕ (Δt = 0.1 s) and the bacterial speed, ϕ_v (Methods 4). In a, the boundary correspond to ϕ^* = 0. In b the boundary is manually traced. c The averaged one-dimensional profiles of the ϕ-fields smoothly transitions between the active (ϕ = 1) and passive (ϕ = −1) phases. Fits of the data (dashed lines, Eqs. (6) and (7) Methods) yield the mean location d and width w of the interface. d, e Parameters d and w are determined for multiple experiments and averaged together as a function of time. Black lines in d, e and i, j are the averaged results of N = 4 experiments per condition; gray represents their min to max variation. For the [O] aperture, the mean interface position d(t) (black line, measured radially from the center of the domain) decreases over time and follows the scaling law \(d\sim \sqrt {\left( {t_0 - t} \right)}\) with t₀ ≈ 40 s. f–j Corresponding results for [H] aperture. For this case, the interface position d initially follows d ~ t (red line) and then transitions to faster than \(d\sim \sqrt t\) (blue line) at t ≈ 2 s. Width w (e, j) ranges from 4 and 10 μm and varies little over time—except for the case of the octagon aperture at long times (t > 40 s) as the passive domain dissolves entirely. Black lines in d, e, i, j are the averaged results of N = 4 experiments per condition; gray represents their min to max variation. Scale bar, 50 μm