On 16 November 2026, the manager of a cotton farm near Dillon, South Carolina (an example), drives two stakes into an 80-acre field due for deep tillage. Deep tillage means breaking a compacted layer with a chisel plow. One stake marks the worst strip, the other a fence row no machine crosses. At each, the manager pushes in a penetrometer, a rod with a gauge that shows how hard the ground is, and writes the date and the days since rain. The pass costs an assumed $1,200, and without these readings no one can tell afterward whether it paid. Judge deep tillage or a terrace only against a reading taken before the work, at a marked spot.
The tool is a soil sheet, one page per area (a compacted patch, a terrace or a waterway). It has a line for each reading, each piece of work and the date of the next reading. A reading is a visit to the marked spot that ends with numbers on paper.
How long does one pass last?
It depends on what you measure, so one reading after the work does not show whether the effect still holds. At Embrapa Wheat, a Brazilian public research center in Coxilha, in the south of the country, researchers chisel-plowed plots of a clay soil under no-till at dates from 0 to 36 months before the measurement. They measured all the plots in the same crop season. The effect lasted 6 months for soil density and porosity, 18 months for penetration resistance and 24 months for how fast water soaked in.
| What was measured | How long the effect lasted | Share of a 36-month schedule |
|---|---|---|
| Soil density and porosity | 6 months | One sixth (17%) |
| Penetration resistance | 18 months | Half (50%) |
| Water soaking in | 24 months | Two thirds (67%) |
That is one soil in one experiment, and your field may hold the change for longer or shorter. In cotton on the coastal soils of South Carolina in 2002 and 2003, deep tillage lowered penetration resistance and raised yield, but the differences were significant only half the time. The authors advise deciding case by case whether to till after a dry year. A schedule set in advance cannot follow an effect that changes with the soil, the traffic and the rain after the pass.
What does deep tillage on a schedule buy?
Sometimes very little, and only readings on your own field show which case you are in. A long trial at Embrapa Soybean in Londrina, Brazil, ran from 1988 to 2007 on a very clayey soil. Chisel plowing every three years under no-till raised soybean yield in 7 of 14 seasons under rotation, and in only 1 of 19 seasons under succession. Under succession, the same sequence of crops repeats every year. It lowered yield in 2 of the rotation seasons and in 5 of the succession seasons. The circular cites these results as unpublished data.
The authors add that the crop response seems to depend on the species, the soil, the weather and how compacted the soil was to begin with. The last of those is what a reading before the work records. On the example farm, the readings at the two stakes, set beside that season’s yield, are what place the pass in one of the three columns.
The crop sequence weighed on that result, so whether the field followed its rotation plan belongs in the same comparison as the tillage. A better crop the season after the pass does not settle it either. Comparing a field’s yield with the farm’s or the region’s average each season separates the effect of the year from the effect of the field.
What makes the reading after comparable with the reading before?
Start with the water in the soil, because a dry week can show compaction that is not there. Penn State Extension advises reading when the whole profile is at field capacity, approximately 24 hours after a soaking rain. Field capacity means soil that has taken up all the water it can hold and drained the rest. Too dry, and compaction is overestimated. Too wet, and it is underestimated. Embrapa Soybean adds that soil moisture at the time of the reading should be measured too, above all to compare fields, layers or dates.
Then the spot. Mark one spot in the area and one comparison spot, and read both on the same day. Penn State calls it extremely useful to compare the field with undisturbed ground such as a fence row, and where a field has subsoiled strips, it advises measuring in and out of the strip.
A Cornell Cooperative Extension educator in New York State used that pair on pasture. He divided the fence-line reading by the pasture reading, so harder ground gave a lower ratio, hoping to control for soil type and changing soil moisture. In August 2020 the ratios for both the pasture and a hay-mown test square dropped together, and he could not explain why. The one measurement he failed to track was soil moisture, and he writes that more questions need answers before the ratio is a reliable tool. What carries over is writing the moisture beside the number.
Two readings taken different ways do not compare, which is why the tool and the depth go on the same line as the number. A lab result follows the same rule: soil test reports copied onto one page per field compare only when someone writes the laboratory, the method and the month beside each number.
Is a terrace that got built a terrace that works?
Not necessarily. A terrace is a ridge and channel across a slope that holds runoff so it soaks in instead of running downhill. Brazilian researchers followed a terrace rebuild on one 80-acre (32.4-hectare) field in Bela Vista do Paraíso, in northern Paraná, in 2016. Of 15 broad-base terraces, every other one was rebuilt and the rest removed, leaving 7. The channel cross-section, which sets how much water a terrace can hold, ranged from 2.71 to 0.73 square meters (29 to 8 square feet) from one terrace to the next.
The team then simulated how much water the terraces had to store at three rates of soil intake: 30, 40 and 50 millimeters an hour (1.2, 1.6 and 2.0 inches). At 30, at least three terraces came out at about 49% of the efficiency they needed, which the authors call worrying. Raising the rate to 40 lifted the whole field from 76% to 97%, but the terraces that fell short stayed at about 63%, and only at 50 did they reach about 87%.
The authors conclude that the rebuild may not control erosion unless other practices raise the rate at which water soaks into the soil. They name excessive compaction among the conditions that make significant erosion likely. That ties the terrace to the compacted patch, because both answer to how fast water gets into the ground. On the example farm, a terrace already in the field gets its own line on the sheet, with the date someone last walked it, and “not known” is an honest first entry.
What do I write on the sheet, and when do I go back?
One sheet per area, with three kinds of line.
| Line | What goes in it | Why it is there |
|---|---|---|
| Reading | The date; the marked spot and the comparison spot; who went; the tool and how deep; days since the last rain; one sentence on what was found | Without the spot and the condition, the reading after has nothing to compare with |
| Work | What will be done; which reading it rests on; the deadline with month and year; the person responsible; done or not done, written at the deadline | Without the reading behind it, nobody can say later whether the work changed anything |
| Next date | The date of the next reading, or of the next walk along a terrace, and who will go | Without a date set in advance, the reading after does not happen |
A line with no reading above it cannot be judged, but it still dates the work, and the first reading after it becomes the starting point. One morning per area is enough, with a spade, a notebook and your phone.
For the first return, Coxilha’s shortest duration, six months, is a rough guide only. If the effect is gone at the marked spot, the next pass rests on that reading and not on the third year. Deciding when to read, who goes and what has to be found before more work is ordered is farm management applied to one patch of soil. The other field-by-field records are in the production library.