Industries / Food, farming and forestry / Agriculture and farming
Soil and planting
Getting a field ready to grow: loosening the soil or leaving it undisturbed, feeding it with manure or fertiliser, and putting seed in the ground.
7
Milestones achieved, from 1701 to 2018.
3
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1701
A drill sows in rows
Jethro Tull designed a horse-drawn drill that sowed seed in evenly spaced rows instead of scattering it by hand, so that a horse-drawn hoe could weed between the rows. London's Science Museum calls it a major step towards machines doing crop work.
Credit
Jethro Tull
1843
Broadbalk field trial begins
John Bennet Lawes and Joseph Henry Gilbert began growing winter wheat on Broadbalk field at Rothamsted, England, giving each strip a different mix of manure and mineral fertiliser. The trial still runs, so feeding plans can be compared across more than 180 years.
Credit
Lawes and Gilbert, Rothamsted
1913
Fertiliser made from air
Fritz Haber filed a patent in 1908 for making ammonia from the nitrogen in air, and a BASF team led by Carl Bosch opened the first plant at Oppau, Germany, in 1913. Erisman and colleagues (Nature Geoscience, 2008) estimate that nitrogen fertiliser fed about 48% of the world's people in 2008.
LessonMost of the nitrogen never reaches food: in 2005 farms used about 100 million tonnes of it and people ate about 17 million tonnes in their food (Erisman and colleagues).
Credit
Fritz Haber; Carl Bosch and BASF
1935
Dust Bowl prompts conservation
After drought caused widespread crop failures and dust storms on the US Great Plains, Congress passed the Soil Conservation Act on 27 April 1935 and set up the Soil Conservation Service, now the Natural Resources Conservation Service. It worked through local conservation districts; the first formed in 1937 and there are now more than 3,000.
Credit
Hugh Hammond Bennett, the service's first chief
1962
Planting without ploughing
Harry Young Jr. of Christian County, Kentucky, planted 0.7 acre (about 0.28 ha) of corn into unploughed ground with a modified planter, using the weedkillers 2,4-D and atrazine to do the plough's weeding. The University of Kentucky calls it the first commercial no-till crop in the US, and by 2018–19 conservation agriculture covered about 507.6 million acres (about 205 million ha) worldwide, by Kassam, Friedrich and Derpsch's count.
LessonNo-till holds soil and runoff in place, but it began by trading the plough for herbicides.
news.mgcafe.uky.eduno-tillfarmer.comno-tillfarmer.comparana.pr.gov.br
Credit
Harry Young Jr. with University of Kentucky agronomist Shirley Phillips; Herbert Bartz, who made Latin America's first known commercial no-till soybean sowing in Paraná, Brazil, in 1972
1992
Combines map their yields
Ag Leader's Yield Monitor 2000 measured grain yield inside the combine as it harvested, so yield could be recorded across every part of a field. By 2012, 51% of US soybean farms used a yield monitor, by the US Department of Agriculture's count.
Credit
Al Myers, Ag Leader Technology
2018
Less nitrogen, more grain
From 2005 to 2015 about 20.9 million smallholder farmers in 452 Chinese counties adopted better field practices. Their maize, rice and wheat yields rose 10.8–11.5% while nitrogen use fell 14.7–18.1%, saving 1.2 million tonnes of fertiliser (Cui and colleagues, Nature, 2018).
LessonMore nitrogen had stopped meaning more grain: the share of applied nitrogen that cereals take up fell from about 80% in 1960 to about 30% in 2000 (Erisman and colleagues).
Credit
Cui Z. and 45 co-authors
Watch for
Microbes in place of fertiliser
Pivot Bio says that in 134 field trials with 129 US growers in 2025, corn treated with its nitrogen-fixing microbe product PROVEN G3 replaced an average of 33 pounds of nitrogen per acre (about 37 kg/ha) and yielded 2.1 bushels per acre more than each grower's usual practice. The release names no outside party that ran or checked the trials.
Achieved whenreplicated trials by universities or public on-farm research networks, over more than one season, publish the same nitrogen cut without yield loss, with methods and data.
Claimed by
Pivot Bio, February 2026
Lab-grade soil carbon in field
Cense Analytics, a 2024 spin-off of Tampere University in Finland, says its laser-based method measures soil organic carbon in the field with laboratory-grade precision. The announcement gives no accuracy figures and cites no independent test.
Achieved whenan outside party publishes a blind comparison with laboratory results on the same samples.
Claimed by
Tampere University and Cense Analytics, March 2026
Driverless planting across a farm
Kentucky farmer Quint Pottinger planned to plant 90% of his 2,000 acres (about 810 ha) of 2026 corn and soybeans with a driverless tractor, using Sabanto's autonomy package and checking in about every six hours.
Achieved whensomeone other than the farmer or his suppliers reports the acres actually planted without a driver, and compares crop stand and yield with conventionally planted fields.
Claimed by
Quint Pottinger with Sabanto, April 2026 (reported by DTN)
Open
01
Score field soil-carbon readings
A field kit and an open scoring sheet for soil organic carbon. Scan soil in the field with a portable near-infrared spectrometer (a device that measures how soil reflects light just beyond what the eye sees) and predict carbon with an open model. Send the same samples to a laboratory and publish how close, how biased and how repeatable the field numbers are.
The briefCredit · unclaimed02
Map no-till fields statewide
A yearly map that labels each crop field in one Brazilian state, such as Paraná, as no-till (seed drilled straight into the last crop's straw) or tilled, checked against photos taken on the ground.
The briefCredit · unclaimed
Open 01 · the brief
Score field soil-carbon readings
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A field kit and an open scoring sheet for soil organic carbon. Scan soil in the field with a portable near-infrared spectrometer (a device that measures how soil reflects light just beyond what the eye sees) and predict carbon with an open model. Send the same samples to a laboratory and publish how close, how biased and how repeatable the field numbers are.
Why it is still open
An independent test of the Stenon FarmLab probe on 100 European arable soils found that "no in-field model fully matched laboratory standards" (Kohl and colleagues, Soil Systems, 2025). Newer claims of lab-grade field results, such as Cense Analytics' in March 2026, come with no published independent figures.
Built with what exists
The Open Soil Spectral Library v1.3 and its browser-based model app, which needs no coding; a portable near-infrared spectrometer such as the Büchi ProxiScout; a commercial soil laboratory for reference results; a hand soil probe.
First step
Take 20 soil cores from one field, split each, send half to a lab and scan the other half three times.
Done looks like
100 samples from at least five fields, each scanned three times in the field and tested once in the lab, scored against the same accuracy, agreement and precision tests Kohl and colleagues used, with the raw scans released. Their paper puts the cost gap at about €3–4 per field reading against about €44 per lab test.
Open 02 · the brief
Map no-till fields statewide
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A yearly map that labels each crop field in one Brazilian state, such as Paraná, as no-till (seed drilled straight into the last crop's straw) or tilled, checked against photos taken on the ground.
Why it is still open
Brazil counts no-till mainly through its farm census: more than 33 million ha, 62% of cropland, in 2017, with later estimates above 41 million ha (Paraná state government). The nearest satellite work found, by Embrapa Soja, estimated straw cover on 30,000 ha in four Paraná municipalities for one season (2019) and reports no ground check. The US satellite tillage system, OpTIS, publishes data only for the contiguous US, 2015–2021.
Built with what exists
Copernicus Sentinel-2 satellite images, whose short-wave infrared bands separate straw from bare soil (the basis of the tillage index Embrapa used); Google Earth Engine; MapBiomas crop layers to find fields; geotagged phone photos for training and testing.
First step
Pick one municipality, pull this season's Sentinel-2 images from the soybean sowing window, and photograph 50 fields whose farmers can confirm how they were prepared.
Done looks like
A map of the whole state for one summer season, with accuracy measured on at least 200 photographed fields not used in training, and the mapped no-till share set beside the census figure.