The ones you can read off a list, and on most farms there is no list. The sheet that answers the question has one line per field and four columns, every one of them copied from paper that already exists: the field name exactly as it appears on the farm map, the area, the date of the last sampling, and where that report is filed. At the top of the page goes the interval the farm has adopted, written in years. Every line older than that gets circled.
The farm does test soil, and nobody doubts it. There are laboratory envelopes in a drawer, an invoice for the sampling and somebody who remembers walking the zigzag. What does not exist is the list. Ask which fields were sampled last cycle and the answer comes out of memory, one field at a time, and one is always missing.
The ground that came in on a lease two years ago has never been sampled, because the tenant assumed the owner had a report and the owner assumed the tenant would pull the cores. The block split in two when the crop changed is still sampled as one, and both halves have been fed the same recommendation ever since. The field by the yard, which everybody drives past, has been sampled three times in four years. The one at the back, which nobody crosses, once.
What that costs is a liming or fertilizer decision taken on a four-year-old number with nobody in the room aware that it is four years old, and a piece of the farm sitting outside the checking system entirely without anyone ever having decided to leave it out. A farm with four fields loses count the same way a farm with four hundred does, because what goes missing is never the quantity. It is the date.
Why can nobody hand you the interval?
Because the institutions that write intervals down do not agree on the number, and never have. USDA Natural Resources Conservation Service Conservation Practice Standard 590 for nutrient management, in the version adopted in New York state, instructs planners to “Use soil tests no older than 2 years when developing new nutrient management plans” and, for revisions and maintenance, that “soil tests must be taken at least every 3 years”. The same federal standard as adopted in Wisconsin says something else entirely: “Soils shall be tested a minimum of once every four years by a DATCP-certified laboratory”. Extension guidance from the University of Massachusetts, written for growers in that state, tells them to “Take soil samples every 2-3 years”.
Two, three and four years, from three documents that all carry authority in the jurisdiction that issued them. Move outside the United States and the spread gets wider rather than narrower: Embrapa Amazonia Oriental, in its 2006 guide to soil and plant sampling for Brazilian conditions, tells the grower to resample perennial crop areas every 3 to 5 years and to sample intensively cropped areas once a year.
The disagreement is the argument, not a flaw in it. What all four documents share is the shape of the rule: an interval exists, it is written down before anybody needs it, and a report older than the interval does not count. That shape crosses every border. The number does not, which is why the interval on your page has to be yours, adopted deliberately with your agronomist, and written where a date can be measured against it.
What does the interval actually decide?
How long the farm goes on being wrong before it finds out. An interval fixes a floor on delay that no amount of attention can get under, and the arithmetic is the same one that governs detection delay with years in place of days. Sample every four years, assume a change is equally likely to begin at any moment inside the window, and the average change is two years old when the next report reveals it. Sample every two years and it is one year old. Nothing about diligence moves that figure. The interval moves it.
Which is why the interval belongs at the top of the page rather than in somebody’s head. A number held in the head cannot be compared to a column of dates. Written above one, it does the single piece of work no laboratory can do for you: it sorts the fields into those still inside the interval and those already past it, before anything gets spent. A declared sampling interval is the only entry on this sheet that is not copied from an existing document, and it is the one the rest of the farm’s soil paperwork keeps pointing back to.
Is testing the soil the same as having the record?
No, and the difference shows up the moment somebody goes looking for the paperwork. A review in Sustainability by Lisa Lobry de Bruyn of the University of New England and Susan Andrews of the USDA Natural Resources Conservation Service went through two decades of national reporting in Australia and the United States and found “participation in soil testing and farm planning has remained stable” across the whole period, and stable at a low level: a quarter of landholders testing soil and three in ten doing farm planning, in the national figures of each of the two countries.
The sentence that matters here is the one about the paperwork: “While farmers report consistent monitoring of soil conditions, few have consistent records of such”.
The same review explains why that is invisible from outside. Soil monitoring at farm level “does not usually occur in a formal, documented manner”, so national-level reporting can say whether somebody tested at all and “fails to indicate the frequency and intensity of soil testing”. That is a review of government surveys in two countries, not a count of farms anywhere else, and it does not say what any particular operation does.
What it does establish is that the two states exist and are easy to confuse. Believing the soil is being watched, and being able to show which field on what date, are not the same condition, and only the second one survives a question from a buyer, a lender or a new agronomist.
Who else reads a row per field?
Everyone who builds anything on soil data, and every one of them assumes the row is there. Ros and colleagues, publishing in Environmental Science and Technology, built an open framework for scoring soil health and ran it across “all agricultural fields in the Netherlands”. They are candid about what makes that possible. The framework needs many inputs per field, which is “easily attainable for countries like the Netherlands where the routine laboratory data are widely available, but not for others”.
That is one country with a national soil database behind it, and none of that machinery transfers to a farm anywhere else. The requirement does. A scoring tool, a buyer’s questionnaire, the soil test the recommendation rests on and a written soil conservation plan all read one row per field, and they have to be reading the same row. A farm with three fields missing a date has three rows nobody can fill, in every one of those documents at once. The sheet is what makes the row exist, and it is the cheapest paper in the whole chain.
Why does each of the four columns earn its place?
Because each one answers a question somebody will ask, and each one is copied rather than judged. Nothing in the table below requires interpreting a result, which is the competence of whoever signs the recommendation.
| Column | Copied from | What its absence costs |
|---|---|---|
| Field name or number | The map the farm already sows and harvests by | Two people describing the same ground with different names, and two lines that never meet |
| Area | The same map | No way to tell whether the missing fields are a corner of the farm or a third of it |
| Date of last sampling | The header of the report itself | The interval has nothing to be measured against, so nothing can ever be overdue |
| Where the report is filed | Wherever it actually sits today | A date that says a report exists and an afternoon lost proving it |
Where there is no date, the word none goes in the cell, spelled out. An empty cell reads as work not finished yet, and somebody will assume it will be filled later. The word none reads as a finding, and it is the most useful information the sheet produces: those are the fields that have been outside the system without anyone deciding to put them there.
The lines that do carry a date are the ones with somewhere to go next. Located once, each of those reports becomes the soil test report by field, a page carrying the depth and the laboratory copied off its own header and the method copied off the line of each determination, which is a job this sheet does not attempt and does not need to.
Where does the division of fields come from?
From the division the farm already uses to sow and to harvest, and never from a new one drawn for this sheet. The University of Massachusetts fact sheet gives both halves of the instruction in two consecutive sentences: “Divide each field into uniform soil and past cropping areas” and “Assign a permanent identification name/number for long-term record keeping”. Embrapa arrives at the same place and names the artifact it produces, telling the grower to identify those blocks definitively and draw a map, so soil fertility can be followed as the years pass, and, quoting Raij, to divide the property into uniform areas by management history, crop, soil color and texture, giving each of them a number for later identification.
Permanent is the word doing the work in both. A field renamed, split or merged between one sampling and the next has no series behind it, only two unrelated numbers, and this sheet is usually where somebody notices. The block that got split when the crop changed is exactly that case, and it stops being one line and becomes two, each with its own date. One of those dates will say none, which is the honest answer and was true before the sheet existed.
The record is also the reason the sheet lives on the farm rather than at the laboratory. Massachusetts extension puts the instruction in the imperative, “Keep a record of soil test results on each field to evaluate long term trends in nutrient levels”, and gives the reason: “Fertility trends over a period of years provide important information, indicating the adequacy of a fertilizer program”.
A laboratory holds analyses. Only the farm holds the sequence, and only if somebody writes it down. Stacked year on year, the same column of dates stops being a coverage check and turns into the field’s soil test series, which is a different sheet built on these same field names, with one row per year instead of a single date.
Whose report is it, once it is paid for?
The farm’s, and public soil laboratory networks say so in their own descriptions of the service. Argentina’s INTA runs a Red de Laboratorios de Suelo, Agua y Vegetal, and the ministry note on the Catamarca laboratory describes the analysis service as aimed at producers, researchers and institutions of the region, with growers, advised by specialists, using the results to understand the condition of their ground and to decide about amendments and management practices. Nothing in that description makes the report the property of whoever sold the fertilizer, and a farm that has to ask a supplier for a copy of its own analysis has already lost the argument about which fields were covered.
FAO’s Voluntary Guidelines for Sustainable Soil Management, endorsed by the FAO Council in December 2016 and written to apply in any member country, puts testing in the category of things a farm adopts rather than buys once. Soil and plant-tissue testing and field assessments “should be adopted and used”, the document says, providing the guidance that informed decisions and the monitoring of progress rest on. Monitoring progress needs two dates. The second one is worthless unless somebody wrote down the first, which is the entire job of the column.
Where to start, and what closing it looks like
One hour with the farm map and the folder of reports, plus one phone call to the laboratory for the samplings whose date nobody can find. Nothing here needs to be bought and nothing needs to be measured in the field.
The circled lines are the ones the farm will argue about, and that argument is worth having in front of the page instead of in a pickup. Some of them turn out to be fields nobody intends to sample again, because they are going into pasture, or coming out of a lease, or being sold, and writing that on the line converts a gap into a decision with a name attached.
What the sheet stops allowing is the third state, the one most farms have been living in, where a field is neither inside the interval nor deliberately outside it, and the only reason it went unsampled is that nobody’s list ever had a line for it. That state is the one the checking system in Production is built to remove, and closing it is what the four functions of management describe when they say a cycle closes.