One page and six fields, signed before the first machine enters that ground. What goes on it: what will be done and in which field, who does each part and with what, when it starts and when it ends, what comes back to the farm and on what date, who may use the data that ground produces, and a name with a phone number on each side. An hour of talking and twenty minutes of writing. Two copies, one for each side, and the farm’s copy goes into the folder where the contracts already live.
The handshake version costs the farm four things nobody counted. A strip the operator cannot treat like the rest of the field, and the first application goes straight over it because his map does not know the trial is there. Machine hours, fuel, and the written record of everything done on that piece. Eighteen months later the result has not come back, not because anyone refused to send it, but because nobody ever agreed that it came back, to whom, and in what form. Then the buyer asks what happened in that field last year and the farm has a story instead of a document.
What is the farm actually putting in?
Ground, machine time, fuel, labor, and the records, and the guide that recommends the practice says so plainly. SARE’s bulletin on farm research, written by David Chaney, lists cost as one of the five challenges to weigh before starting, because on-farm research “may involve extra expenses for planting and managing field plots, collecting data, and any specialized tests that you run on soil or plant samples”. The same bulletin sets the balance of the relationship: professional researchers “may provide leadership, guidance and assistance, but you are directly involved in the design and management of the project”.
Directly involved is a commitment of hours, and hours from a person who has other work that week. The records are the input that outlives the trial. A strip held outside the standard the rest of the field is managed under turns up years later in the field’s soil test series as a reading that does not compare with the ones around it, so the term is also where the reason for that break gets written down while somebody still remembers what it was.
Is the farm named in the write-up at all?
Often it is not, and this is the part worth knowing before you say yes. Quentin Toffolini and Marie-Hélène Jeuffroy went through the published record of experiments run on working farms and grouped the practices into seven types. One of the seven they named “Considering farm fields as the locus of experiments without mentioning farmers”.
In that group the ground was chosen because the condition under study was there and hard to reproduce elsewhere, and the authors record that farmers’ “constraints and difficulties to implement the experimental treatments or technologies were never mentioned”. What that review reads is how researchers wrote about their own trials, not what the farms would have said. It is still enough to make the point: being the site of a trial and being a party to it are two different positions, and only one of them is written down anywhere.
What do you ask before you sign?
Four questions, and they already exist. The Utah State University Extension on-farm research guide gives ten questions for judging research information before acting on it, among them whether the research is “conducted or promoted by independent or commercial interests”, whether the researchers “explain how (methods) the research was conducted”, whether it was “replicated or repeated in time and location”, and whether the research information is “readily available”. The guide points those at published information rather than at the person standing in your field with a proposal. Aiming them at the person is Rurivia’s reading, and it holds: whoever cannot answer the four in a conversation will not answer them in a report either.
When does it start, and when does it end?
Two dates, and the second one is the one that gets left off. Chaney’s bulletin tells you to plan the experiment “for at least two growing seasons to increase the reliability of your results”, and where it turns to pasture and livestock it puts that same span in years: “most studies are conducted for at least two years”. The review by Toffolini and Jeuffroy records, for the more collaborative group, that “The trials usually lasted several years”.
Anything running that long outlives the student who was collecting the data and sometimes the project that funded it. The end date is what returns the strip to the standard the rest of the field is managed under, and without it the exception quietly becomes permanent. A field committed for three years is a cell somebody outside the farm now decides, which is why it belongs on the multi-year production plan with the same name and date as every other cell.
Who does which part, and with what?
Both lists, written side by side, and one farmer organization publishes exactly that. The Cooperators’ Program of Practical Farmers of Iowa states what the farmer is expected to do, including “Complete research project following a written protocol”, “Collect and record data as specified” and “Keep in contact with Practical Farmers staff with updates and questions”. It then states what the staff owe back, including “Help design project and write a detailed project description”, “Monitor progress of the project and provide support when needed” and “Summarize and publish results”. Two columns on one page. Copy the shape and fill it with the names of the people who will actually be in the field.
Whose calendar does the work sit on?
The farm’s, and that is the collision the term exists to prevent. Chaney’s bulletin puts the instruction before any field work starts: create a management plan and calendar for the project, be specific about how and when treatments are to be applied and what data will be collected, and then “make sure you review this plan with everyone who will be involved in the project”. His reason is the one that decides whether the trial survives its first busy week: good planning and communication help ensure “that the work is done on time, and that you have the equipment and labor available when you need it”.
Everyone who will be involved includes the operator whose map has to show the strip. The Utah State guide, writing about the notes and records a trial needs, gives the reason none of it is left to memory: “Regardless of how good your memory may be, important details will be forgotten if not written down.” Where the counterpart of the deal is training somebody, that line belongs on the written training plan, which is where a promise becomes something you can check.
What comes back to the farm, and on what date?
Whatever you wrote down, and nothing else. Chris Maughan and Colin Anderson followed three participatory research projects in England and Wales between 2018 and 2022 and asked the people who took part how it had gone. In the first of the three, what came back had a shape you could hold: after the field walks, “a farmer field guide to plant bioindicators was developed and distributed”, the project having set out to test whether the plants that come up on their own can be read as a sign of what the soil is doing. A guide, a report, a training day, a payment for time.
The form is what the page has to name, because nobody can check delivery against a word as loose as results. Those are three projects watched closely, and what they show is what happened there rather than a measure of what happens anywhere. Money is the other half of the same field: in the third project, a participant told them that being able to reimburse the host farmer for that time was what made taking part possible at all. Practical Farmers handles this by writing the number first: where there is extra money for a bigger project, staff and cooperators “pre-determine those payments before initiating a project”.
Who may use what that ground produces?
Whoever the page says, and this is the field that gets skipped most. Ashley Ellixson, at University of Maryland Extension, and Terry Griffin, at Kansas State University, set out the questions a farm should settle before its data leaves the gate: “What control will the farmer have over the data”, whether the data are “portable between computer media or transferable to other entities”, whether they can “later be deleted after use”, and whether the contract requires “affirmative consent before data can be shared” with a third party.
They are writing about United States law, and their own conclusion is what makes the field universal: “Until the legal framework catches up to agricultural technology, ensuring protection of farm data in a well-crafted contract is the best management practice to follow”. A yield map from one strip is a thin thing on its own. Beside the application records, the input invoices and three years of history for the same ground, it stops being thin.
Who answers when the other side calls?
One name and one phone number on each side, and the institution’s, not the student’s. Maughan and Anderson list facilitation among the shortcomings of their second case, and one participant there put the gap in a single line: the post-event community facilitation needed to keep that energy going. A trial that ends when one person changes jobs was never a partnership with an institution, it was a favor from a person who has now left.
Write the department, the role and the phone, and write the same three on the farm side, because the person who answers on the farm is often not the person who signed. That line then goes into a co-op membership record as one more row, with the start and end date where the annual contribution would be.
Where to start
One page, two signatures, twenty minutes of writing after an hour of talking.
The second term is the useful one
Nothing on that page proves anything the first year. It starts working on the day the end date arrives and somebody has to decide whether to renew, because that is the first time the farm compares what it wrote down against what it got. The trial that returned nothing, and the one where a name and a phone number were enough to get the results back, are the same length on paper and tell you different things about who to say yes to next.
Institutions ask for a written history before they commit to a farm. The sustainability axis is largely made of records the farm keeps about relationships it already has, and asking the neighbors and buyers what they think of the place will reach the partner institution too. What all of them sit inside is the same four functions of management.