A sandcastle stands because water bridges hold the grains together. In the pendular regime — a small amount of water forming individual bridges between adjacent grains — each bridge exerts a capillary force proportional to the surface tension of water and the curvature of the meniscus. Remove the water and the castle collapses into a pile. Add too much and the castle dissolves into mud. The structural integrity lives in a narrow moisture window.
Recent experiments on cohesive granular column collapse (arXiv:2402.07285) reveal that this isn't a simple matter of “more water = more strength.” Two different sources of cohesion — capillary bridges (wet) and polymer coating (dry adhesion) — produce qualitatively different collapse dynamics. Wet cohesion creates heterogeneous, localized failure patterns because liquid bridges break and reform during shear. Polymer adhesion creates more homogeneous failure because the bonds are fixed.
Note (2026-11-10): I checked the abstract and searched the full text of arXiv 2402.07285. The essay's main contrast does not match the paper. The essay says the two sources of cohesion produce qualitatively different collapse dynamics, with wet cohesion giving heterogeneous, localized failure and polymer adhesion permanent bonds that stay broken. The paper compares the two sources in a common framework: a dimensionless number Co, the ratio of the macroscopic cohesive strength (from shear-cell yield strength measurements) to the particle weight, is shown to capture the effects of cohesion on the final morphology, for a cylindrical column that spreads axisymmetrically and a confined rectangular column. The paper says both sources of cohesion are capable of reforming after being broken (its polymer coating is a cross-linked polyborosiloxane). I did not find a comparison of heterogeneous and homogeneous failure. The paper's data are final deposit morphologies and scaling laws, and it notes that cohesion that is brittle would need further work.
The engineering insight: the behavior of partially saturated granular materials depends not just on how strongly grains are held together but on how the bonds respond to deformation. Capillary bridges are self-healing — break a bridge, and if the grains return to proximity, a new bridge forms. Polymer bonds are permanent — break them and they stay broken. The dynamic response differs even when the static strength is identical.
This matters for real geotechnical engineering. Slope stability, foundation settlement, and embankment failure in partially saturated soils all depend on how capillary cohesion redistributes during loading. The sandcastle isn't just a toy — it's the simplest case of a problem that governs dam stability, road construction, and earthquake-induced liquefaction. The physics of play is the physics of infrastructure, scaled up.