Basalt credits are counted in the topsoil. A Nature study says the carbon stops below it

The accounting gap sits between two measurement points. Assessments that infer removal from the disappearance of calcium, magnesium and other elements out of crushed basalt in the upper 10 to 30 centimetres of soil are measuring dissolution ^[6]. Louis Derry of Cornell, with Kate Maher at Stanford and Oliver Chadwick at UC Santa Barbara ^[2], argue that dissolution is the first step only, and that the reactive layer below decides how much of the resulting alkalinity ever leaves soils, travels down rivers and reaches the ocean, where it converts CO2 into forms that stay put ^[5]. Along that path, newly formed clays, oxides and carbonate minerals take calcium and magnesium back out of solution, or generate acidity that consumes the alkalinity weathering just made ^[7]. Earlier Icelandic work cited by the authors puts that retention at 30 to 67 percent of calcium and 45 to 81 percent of magnesium ^[8], which leaves a third to seven tenths of the calcium and as little as a fifth of the magnesium still moving ^[19].

The second constraint is arithmetic rather than chemistry. Carrying 2.5 tonnes of CO2 per hectare per year as dissolved alkalinity through the 0.25 metres of annual recharge available in the upper Mississippi basin ^[11] requires roughly 23 millimoles per litre in every litre that leaves the profile ^[15]: 250 grams of CO2 per square metre is about 5.7 moles, and there are only 250 litres a year to move it. The authors say concentrations in that range sit outside plausible values for soil water ^[12]. Drop the target to the 0.22 tonnes per hectare that median natural basaltic watersheds actually export ^[4] and the requirement falls to roughly 2 millimoles per litre ^[16], water nobody would look at twice. The proposed flux is about eleven times the natural median ^[17], and the wet tropical landscapes that produce the fastest natural basalt weathering shed eight to twelve times more water per year than that Mississippi cropland does ^[18].

Grinding was supposed to make up the difference. The paper reports only a weak relationship between particle size and commonly measured surface area except at very fine sizes, and no notable correlation at all among the basalt samples proposed or used for enhanced weathering ^[9]. If surface area does not scale with particle size outside the fine tail, milling energy is not buying a proportional reaction rate, and the rate assumed in a project's forward curve has to come from somewhere else.

None of this makes the practice pointless. The authors allow that spreading basalt can buffer soil acidity, supply nutrients and provide a modest offset ^[13]. Those are agronomic returns, and they are not what a carbon credit certifies. What the study puts under pressure is the assumption that carbon captured during initial dissolution in the soil can largely be booked as durable removal ^[14], which is the assumption the tonnage is priced against. Derry's own summary is that some CO2 can be buffered this way but much less than people have hoped, and that gigaton-scale reduction by this route is not going to happen ^[3].