Last year the crew called one field “soy 3”. This year it is “the creek field”, and the 190 hectares (about 470 acres) written beside it came from nobody knows where. Its yield, cost and soil results no longer match last year’s, and its cost per hectare rests on an area nobody can vouch for.
Keep a register with one line per field: a code that never changes and an area you measured and dated, plus one history line per season (crop, sowing and harvest dates, operations). The boundary drawing may change; the code and the lines already written do not. With the register, every yield, cost and soil result stays tied to the same field from one year to the next. In this article a field is a piece of ground where the whole area gets the same crops on the same dates, so it has a single history.
What counts as one field in the register?
Two written definitions of what counts as a field (the unit of the register) give two answers, and the register has to pick one. The first comes from the FAO, the United Nations food and agriculture agency. Its manual for national farm censuses defines a field as a piece of land bounded by recognizable limits, such as paths or hedges. A field of that kind can hold more than one crop.
The second comes from Brazil’s agriculture ministry. It appears in a 2025 rule for its climate risk zoning (the official maps of the sowing dates with the lowest climate risk for each crop) that sorts areas by management level. There, a field is a delimited cultivation area inside a farm, defined by uniform conditions of management and cropping history, including sowing and harvest dates and operations. Operations here are the passes on the ground, such as tillage, spraying and fertilizing. It is the one written source here that ties the unit to its history, and history is what a register keeps.
The two disagree where it matters. Take a piece of land bounded by a road, with soybean on 80 hectares and corn on 50. By the FAO definition it is one field. By the ministry’s it is two, because the history is not uniform. From here on, “field” means the ministry’s unit: ground where the whole area gets the same crops on the same dates, so each field has one history. The register exists to hold history, and a unit that holds two histories answers the question “what happened on this ground?” for neither.
The price of this choice is that a paddock fenced as one, with two crops, becomes two fields in the register even if the crew still calls it one.
The sources give no rule for what comes next, so this one is the library’s own. Split a field the day two parts of it get different crops or dates. Retire the old code and never reuse it, create two new codes, and write the parent code and the date on each new line.
Can I trust the area I already have?
Only as far as you know how it was taken. The one study among the sources that measured this comes from West Africa, on small farms. The working paper is by two researchers of the International Food Policy Research Institute, a food policy research center. In it, farmers in Burkina Faso, Mali, Niger and Nigeria stated field areas that differed from the GPS measurement. The gap ran above the measure in Burkina Faso, Niger and Nigeria and below it in Mali, by the amounts in the figure. In Nigeria, 171 percent above means a stated area about 2.7 times the measured one.
The averages hide a split: small fields were overstated and large ones understated, and in all four countries larger farms tended to make larger errors.
Read this with care. These are household surveys of small farms, comparing what a farmer told an interviewer with a GPS measurement. A grain farm keeps its areas in a title, a map or the file of a GPS-guided tractor, and none of the sources tests how far such areas are off. One habit carries over to your farm: write the method and the date beside every area (walked, drawn on a map, taken from the title or remembered). Without them you cannot tell a measured area from a remembered one.
There is also a check that needs no equipment. The FAO manual states that the areas of a farm’s separate pieces of land, owned or leased, must add up to the area of the farm. On your farm, add up the fields and the land outside them that you have written down (reserve left uncultivated, roads and yards, buildings), and compare the total with the farm’s area on the title. A gap does not prove an error, and it does not say which field is wrong. So start with the field whose area you trust least, walk its boundary, and then go to the next one you trust least.
The example is a grain farm with six fields and a title of 1,200 hectares (a hectare is about 2.47 acres). F-03 is the creek field. The values are illustrative, and the farm is not real.
Why must the code outlive the crop and the boundary?
Because the records you keep by field all depend on the code, and they stop matching when it changes. The yield series of a field, described in how to read each field’s yield year by year, needs the same field in every season. A soil test report has one page for each field, as in reading the soil test report.
A lot of grain you sold, one delivery for example, is traced back to the ground where it grew by the field its record names, as in tracing a lot back to the field. If those records are filed under the crew’s name, “soy 3” this year and “the creek field” next, the old records and the new ones stop matching on the day the name changes. A code that never changes keeps them together.
Three rules keep the code stable. They are this library’s own, not findings of the sources. The code holds no crop, no year and no owner, so it is F-07 and never Soy-2026. The name the crew uses goes in its own column of the register line, and every record is filed under the code. A retired code stays in the register, marked retired, with its date and the codes that replaced it, and no code is ever given to another piece of ground.
How strong is the evidence for keeping a register?
Thin, and mostly indirect. The definitions above are written standards, and the area gap was measured, but in another setting. On whether keeping records pays, the nearest source is a 2023 review, in the scientific journal Heliyon, of 24 empirical studies on farmers’ decision-making and farm performance. It found studies that link record keeping to better efficiency (more output from the same inputs). But it rates the link between planning and monitoring with record-keeping tools and farm performance as not a conclusive finding.
Farms with different record habits also differ in other ways, so a comparison between them cannot show that the records caused the result, and this article claims no cause. None of the sources tests a field register on a large grain farm of the kind found in Brazil’s Cerrado, a savanna region that grows soybean and corn at scale. Keep the register for a plainer reason than a measured gain. A yield, cost or soil record can only be matched from one year to the next when the field has a fixed code and a dated area.
What goes in the register, and where do I start?
Start with the field whose area you trust least, before the next sowing. The register has two parts: one register line per field, and one history line per field and season. The ministry’s rule for sorting areas by management level asks for three things on record for each area it sorts. They are the identification of the property and of its owner or tenant, the geolocation of the area’s perimeter, and the history of use and management. That sorting is valid only for the crop and season assessed, which is why the history needs a line for each season.
| Column | What it holds | What goes wrong without it |
|---|---|---|
| Code | A short code that never changes and never repeats | The history is lost when a field is renamed |
| Crew name | What the crew calls the field | The crew and the register talk about different fields |
| Area, in hectares | The measured area, not the remembered one | Every per-hectare number moves with the guess |
| Method and date | Walked, GPS track of the tractor, drawn on a satellite image, from the title, or estimated, and when | A change of method looks like a change in the field |
| Owned or leased | Whether the land is owned or leased, and from whom | A lease ends and its fields leave the register with no trace |
| Boundary | A dated drawing or GPS file of the boundary | The field cannot be found again after the boundary moves |
| Parent code | The code it came from, if it was split or merged, with the date | The history stops on the day of the split |
| Retired | Yes or no, with the date and the codes that replaced it | A retired code looks like a field still in use |
| Column | What it holds | What goes wrong without it |
|---|---|---|
| Season | The crop year the line belongs to | The lines cannot be ordered |
| Crop | What grew on the field | A drop in yield cannot be tied to a crop |
| Sowing and harvest | The two dates | The seasons cannot be compared on timing |
| Operations | Each operation on the field, such as tillage, spraying or fertilizing, with its date | What was done is kept only in memory |
| Source | Who wrote the line, and whether it came from a record or from memory | A line from memory cannot be told from a line copied from a record |
Once the history lines exist, you can check a field against its planned crop rotation: see whether the field followed the rotation plan. The same history line can also hold the group of each product sprayed on the field: which spray group went on each field, and when.
A register like this is the base of the records kept under farm management. More on field records is in the production library. Related: Cost per bushel by field: your invoices over your harvest and Pest scouting sheet: write the action level first.