Columnar Basalt Jointing Lab

Why Cooling Rock Cracks: Contraction Against Constraint

Basalt lava erupts at temperatures around 1,100 to 1,200 degrees Celsius as a low-viscosity melt rich in iron and magnesium silicate minerals. As it loses heat to the air, to water, or to the ground beneath it, it first solidifies into a coherent solid rock mass, and only after that does the real jointing story begin. Solid basalt, like almost all solid materials, contracts in volume as its temperature drops further. If a small block of basalt could cool in total isolation, it would simply shrink uniformly and nothing more would happen. But a real lava flow or intrusive sheet is not isolated: the recently solidified layer is rigidly attached to adjacent rock that is still hotter and has not yet contracted by the same amount. That neighboring material physically prevents the cooling layer from shrinking freely. The result is a buildup of internal tensile stress, the same kind of stress you would feel pulling on a rubber band, except here it is generated internally by the mismatch between how much the rock wants to shrink and how much it is mechanically permitted to shrink. Tensile stress is the specific type that matters here because rock, like most brittle materials, is dramatically weaker in tension than in compression: it takes only a modest pulling force to snap it apart along a plane, whereas it can withstand far greater squeezing forces before failing. As cooling continues and the temperature drop accumulates, this tensile stress climbs steadily. When it finally exceeds the local fracture strength of the basalt, at some point of weakness or stress concentration, a crack initiates. This is a purely mechanical threshold event: nothing mystical, no special mineral trigger, just ordinary Griffith-type brittle fracture responding to accumulated strain energy. The same basic mechanism, contraction against constraint, is what causes drying mud, cooling ceramic glaze, and even some drying starches to crack into polygonal networks; basalt columns are simply the coarsest, most durable, and most famous natural expression of this universal shrinkage-cracking physics, preserved in solid rock for millions of years rather than washed away in the next rain.

Why Cracks Run Perpendicular to Heat Flow

The orientation of any individual crack is not random; it is dictated directly by the geometry of the temperature field inside the cooling rock. Heat always flows from hotter regions toward cooler regions, moving along the steepest temperature gradient, and in a cooling lava sheet this defines a family of surfaces called isotherms, meaning surfaces of equal temperature, which are roughly parallel to the cooling boundary at any given moment. Tensile stress from contraction accumulates along the isotherms, in the plane parallel to the cooling front, because that is the plane in which each thin layer is being restrained from shrinking by its neighbors within that same layer. A crack forms to relieve stress most efficiently when it opens perpendicular to the direction of maximum tension, which means the crack surface itself ends up oriented perpendicular to the isotherms and therefore parallel to the direction of heat flow. As cooling continues and the solidification front migrates inward, away from the original cooling surface, each crack tip is pulled along behind that advancing front, extending the crack incrementally in the direction the heat is currently traveling at that location. The practical consequence is that a basalt column's long axis is essentially a fossilized heat-flow vector: it records the direction heat was moving through the rock at every stage of solidification. When a lava flow cools uniformly from a flat, horizontal top surface downward into a flat base, heat flow is simply vertical everywhere, isotherms are horizontal, and the resulting columns stand up straight and parallel, like the classic vertical palisades seen at many flood basalt exposures. But where the cooling surface is irregular, for instance near a stream channel that chills the base of a flow unusually fast, or near a vertical intrusive contact, or where a flow pools against uneven topography, the isotherms bend to follow that irregular boundary, and the columns bend with them. This is precisely why some columnar basalt outcrops display dramatic fanned, radiating, or curved column arrangements: the columns are simply tracking a heat-flow direction that itself curved because the underlying cooling geometry was not simply planar.

Why Hexagons: Self-Organization Toward Minimum Energy

A single isolated crack tells you nothing about hexagons; the six-sided pattern only emerges because many cracks initiate at roughly the same time across an extended cooling surface and then grow while continuously interacting with one another. Picture the cooling surface as a two-dimensional sheet under roughly uniform biaxial tension, tension pulling equally in all directions within that plane. Because the stress is essentially the same everywhere at the moment of crack initiation, cracks do not start from one dominant point and radiate outward; instead, numerous small crack-initiation sites appear in a scattered, quasi-random arrangement, each nucleating independently wherever a local flaw or stress concentration first crosses the fracture threshold. As each crack begins to grow, it locally relieves the tensile stress in its immediate vicinity, which changes the stress field experienced by its neighbors. Growing cracks effectively compete for the surrounding stressed material, each trying to capture and relieve as much of the local strain energy as possible, and this competition is what forces the network toward a specific, predictable geometric outcome. It can be shown, using energy-minimization arguments closely related to those used for soap-film networks and Voronoi tessellations, that a crack network dividing a uniformly stressed plane into cells releases the most total strain energy, using the least total crack length, when the cells approach regular hexagons meeting at 120-degree angles. This is the identical mathematical result that governs why soap bubbles meet in threes at 120 degrees and why many biological cell-packing patterns favor hexagons: it is simply the most efficient way to partition a plane. Real basalt columns rarely achieve a perfect regular hexagon in every case, because natural rock is never perfectly homogeneous and stress is never perfectly uniform, so the observed columns commonly range from four to seven sides, with five, six, and seven being the most frequent, but the statistical average across a large columnar basalt outcrop clusters strongly around six sides, exactly as the energy-minimization theory predicts, and exactly as observed in classic desiccation mud-crack polygons and drying starch-slurry experiments used as laboratory analogs for this same physics.

Slow, Uniform Cooling and Column Quality

Not every basaltic lava flow produces textbook columns, and the difference comes down almost entirely to how slowly and how evenly the rock cooled. Slow cooling gives the rock time to relieve stress incrementally through many small, well-spaced crack-growth increments rather than through one sudden, chaotic fracture event. Each time a crack advances a short distance as the solidification front creeps inward, the local stress is partially relieved, and the system has time to re-equilibrate before the next increment of stress builds up; this typically leaves a visible record of these growth pauses as fine horizontal ridges called chisel marks on column faces, essentially a growth-ring record of the crack's incremental advance. Slow cooling also allows the crack-spacing self-organization process described above to fully play out, because the competing crack tips have time to interact and negotiate toward the mechanically optimal hexagonal spacing across a wide area, producing large, wide, remarkably uniform columns like the meter-scale pillars at Giant's Causeway or the Devils Postpile in California. Cooling rate also directly sets column width: faster heat loss drives a steeper thermal gradient and a faster-advancing solidification front, which forces more crack-initiation sites to appear closer together to keep pace with the more rapid stress accumulation, yielding narrower columns, while slower cooling allows fewer, more widely spaced cracks and correspondingly thicker columns. This is why the interior of a thick lava flow, insulated by the rock above and below it, cools far more slowly than the flow's chilled outer margins and develops noticeably wider columns than the finely jointed, sometimes almost fibrous rock right at the top and bottom contacts. Uneven or turbulent cooling, by contrast, for example where groundwater intermittently floods a cooling surface, or where a flow's thickness varies abruptly, or where a fast-quenched glassy margin forms against water or wet sediment, disrupts the orderly geometric negotiation between neighboring cracks. The stress field becomes locally chaotic rather than smoothly uniform, crack nucleation becomes irregular and asymmetric, and the resulting columns end up irregular, poorly formed, bent unpredictably, or simply absent, replaced instead by a jumbled, blocky, non-columnar jointing pattern often called entablature, in contrast to the neat, orderly colonnade produced by slow, undisturbed cooling.

Reading Real Outcrops: Colonnade, Entablature, and Famous Sites

Field geologists studying a thick columnar basalt flow commonly distinguish two structural zones stacked one atop the other, and both are direct, readable consequences of the cooling-rate physics described above. The lower zone, called the colonnade, typically consists of thick, relatively straight, well-formed columns that cooled slowly and steadily from the base of the flow upward, or inward from a stable, relatively flat top surface, giving the crack network ample time to self-organize into clean hexagonal geometry. Sitting above it, the entablature consists of thinner, more irregular, often curved or radiating columns, reflecting faster, less uniform cooling, frequently attributed to the late arrival of surface water, perhaps rain or floodwater, that infiltrated the still-hot upper flow and chilled it unevenly and rapidly from numerous scattered points rather than from one smooth, advancing front. Giant's Causeway in Northern Ireland, roughly sixty million years old, displays this two-tier structure clearly, and its lower colonnade contains an estimated forty thousand individual columns, most with five or six sides, formed as a thick basalt lava pond cooled slowly within a broad, gently sloped valley. Devils Tower in Wyoming presents a different setting: rather than a surface lava flow, it is widely interpreted as either a shallow intrusive body or the solidified feeder conduit of a volcanic system that never fully erupted, cooling slowly underground within enclosing sedimentary rock, which explains its unusually tall, strikingly vertical columns, some exceeding eighty meters in unbroken length, since the heat-flow direction there was consistently perpendicular to a relatively simple, gently curved cooling boundary over a long, slow cooling history. The Devils Postpile in California and the Columbia River flood basalts of Washington and Oregon offer further variations on the same theme, each shaped by the particular thickness, cooling geometry, and cooling duration of its own lava body. In every case, the underlying rule is the same: measure the column geometry carefully enough, and you are effectively reading a fossilized record of exactly how, and how fast, that particular body of molten rock lost its heat.

Frequently asked questions

Try it live

Everything above runs in your browser — open Columnar Basalt Jointing Lab and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.