We said American topsoil was washing away faster than ever. The federal record says it has fallen by two thirds.

On 2 June we published a piece on this page saying that topsoil across the world's grain regions is eroding 10 to 40 times faster than it forms, that yields are flattening because of it, and that the map of where food grows is about to be redrawn. We gave no source for the ratio. We have now gone and read the federal record, and the direction of the American part of that story is close to the opposite of what we told you.

What the measurements say

The US Department of Agriculture has been measuring soil loss on American cropland since the 1930s, first through the Soil Conservation Service and now through the National Resources Inventory. The series is one of the longest continuous environmental measurements the country has.

In 1930, on land in row crops, small grains and rotation meadow, the average rate of soil loss was 14.9 tons per acre per year, with a stated 95 percent confidence range of 13.9 to 15.9. By 1982 it was 7.8 tons — a 48 percent fall. By 1992 it was 6.3 tons — a 58 percent fall from 1930. The 2017 National Resources Inventory puts the all-cropland average at 4.63 tons per acre per year: 2.67 from water and 1.96 from wind, and a further 35 percent below the 1982 figure.

Four measurements, sixty years apart at the ends, all from the same programme, all pointing the same way. American cropland is losing roughly a third of the soil per acre that it lost when the Dust Bowl was in living memory.

What we conflated

The mistake in the original piece was to treat two different things as one. Erosion is soil physically leaving a field in water or wind. Depletion is what remains in the soil that stays — organic matter, and the trace elements a plant needs in milligrams. They are measured differently, they respond to different practices, and one of them can improve while the other does not.

Contour ploughing, no-till, cover cropping, terracing and the conservation compliance rules attached to federal farm payments are aimed squarely at the first. The record above is what that effort bought. None of it directly rebuilds what has already been mined out of the soil that stayed put, and our piece attributed to erosion a problem that mostly is not erosion.

What we are not claiming

We are not saying soil is fine. We are saying we published a number we could not source, in a direction the federal record does not support, and that the honest version of this story starts from a harder question than the one we asked: not whether soil is washing away faster than before, because measurably it is not, but what is left in the soil that stayed, and whether anyone is measuring that as carefully as they measured the erosion.

That is the piece we should have written, and it needs evidence we do not yet have in hand. We would rather say so than fill the gap with another unsourced ratio.

Why this matters here

The Central Valley grows a quarter of the food eaten in the United States, and Sacramento sits at the top of it. When a national soil story is told wrong in either direction — panic or complacency — the consequences arrive on the roads outside this city, in water allocations and in what a farm can plant. We would rather be corrected in public than be trusted for the wrong reason.