Main

If water dominates carbon dioxide (CO2) as the vapour phase1,2, our argument3 about the pressure distribution driving a kimberlite dyke to the surface is reinforced. The key factor allowing the initial rapid ascent is the large difference between the high mantle source pressure and the low dyke tip pressure, the latter being buffered by the saturation pressure of the least soluble volatile phase4,5. The dyke tip pressure required for water to exsolve will be even lower than the pressure we inferred for CO2, thus increasing the pressure difference driving the magma upward through the opening dyke.

Any vapour-filled region at the tip of a dyke, breaking away to propagate faster as an independent crack, can incorporate6 some of the magmatic foam implied by our model3. Cracks longer than ∼20 m travel at ∼1 km s-1, which is ∼40% of the sound speed in rock7. Chilling of magma in the closing crack base ‘heals’ the fracture, restoring the country rock mechanical properties. When the dyke tip subsequently arrives, it encounters essentially the same conditions as if crack separation had never occurred. Only seconds are needed to chill a 1–2-mm-thick film of magma left behind by a ∼20-m-long crack; during this time the dyke tip, rising3 at ∼20 m s-1, travels ∼100 m—a tiny fraction of the dyke’s vertical extent. A new low-pressure region starts to grow below the dyke tip immediately after crack separation; we infer that the stress and pressure conditions we proposed will be present over most of the path of the rising dyke tip.

Our dyke geometries are only slightly larger than those of Sparks et al.1,2, and their minimum estimates of total magma volumes imply eruption durations only a few times longer than the time to establish the dyke pathway3; larger volumes will imply more prolonged events. Our calculations3 of adiabatic cooling refer to magma reaching the surface during the opening phase of an eruption; in a long-lived eruption, most of the magma finally emplaced in the sub-surface diatreme will indeed suffer less cooling.

We suggested3 a violent change from overpressure to underpressure as a dyke reached the surface, with rapid physical development of the near-surface pipe and diatreme system. A longer-lived eruption1,2 will indeed allow a range of additional failure mechanisms. Although ‘fluidization’ commonly relates to the near-steady passage of gas through unconsolidated granular materials, as in the waning phases of kimberlite eruptions8, the basic physics is the same as that in our violent opening phase.

Regarding preservation of diamonds during transit to the surface, we stress3 that rapid transport will maximise the survival of diamonds as they pass through potentially unstable combinations of ambient pressure and temperature conditions, irrespective of the chemical environment that they encounter9.