First Farmers: Hemp in the Neolithic

The best genetic estimate says cannabis became a farm crop in East Asia in the early Neolithic, around 12,000 years ago, give or take thousands of years. For a long time after that, people grew it for fiber and food. Some of the oldest dated proof is a set of fruits from Okinoshima, Japan, about 10,000 years old. No find we read from this era shows anyone getting high. Popular retellings say more. The Taiwan “10,000 years” story, the cloth marks on Neolithic Chinese pots, and the tidy “five grains” list all change shape when you open the sources. This chapter opens them.

A note on words. Neolithic means the age of farming villages and pottery. “Hemp” in this chapter means the crop: a plant grown for fiber, oil and seed. By one genetic reading, hemp types and drug types had not split yet. We use “hemp” as a short word for that early multipurpose plant.

Where did the first hemp farmers live?

In 2021, Guangpeng Ren and colleagues analyzed the whole genomes of 110 cannabis plants, 82 of them newly sequenced, from around the world. The set included feral plants (escapees from farms), old farm varieties called landraces, and modern crops. The team sorted the plants by how closely their DNA matched.

Four groups came out. The oldest branch held mostly feral plants and landraces collected in China, plus two feral plants from the United States. Both hemp types and drug types branch from that Chinese gene pool. The authors conclude that cannabis was first domesticated in East Asia in early Neolithic times.

They also dated the split of the hemp and drug ancestors from that basal group: about 12,000 years ago. The margin of error is huge. The 95% interval runs from 6,458 to 15,728 years ago. A genetic clock is a model, not a stopwatch.

c. 12,000 years ago (estimate; 95% interval 6,458 to 15,728 years) · , East Asia (exact place unknown)

Genomes point to East Asia as the first farm

Whole genomes of 110 plants trace hemp and drug types to one early gene pool, still found in Chinese feral plants and landraces. The authors date the split to about 12,000 years ago, with a very wide margin of error.

Evidence Genetic Confidence Medium (2 of 3)

Sources (2)

• Ren et al. 2021, Science Advances 7(29): eabg2286 (open access copy on PubMed Central)
• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518

Other researchers urge care. Dal Martello and colleagues write that the limited sampling of wild populations and the wide error bars call for “ground truthing” with seeds and other finds from the ground. In rice, a similar method gave divergence dates thousands of years older than the first farmed grains anyone dug up, up to 10,000 years older. When that team wrote, no ancient genomic data existed to calibrate the clock.

There is a stranger wrinkle. Ren’s team found no truly wild plants in their sample. Every feral plant they tested looks like an escape from a crop. They suggest that pure wild ancestors of cultivated cannabis may be extinct, and they say more sampling is needed. If they are right, the plant the first farmers picked up is gone.

A second team reads the same continent differently. Long and colleagues propose a multiregional origin of human use, because cannabis records appear at about the same time in Europe and East Asia. This is a hypothesis, like the 12,000-year date. Long’s team itself asks for more well-dated finds.

Ren’s team adds one more point that matters here. They propose that early cannabis was a multipurpose crop until about 4,000 years ago. Only then did a small number of early strains expand into hemp types and drug types. For most of the story in this chapter, one plant did all the jobs.

What did the Jōmon leave in the mud?

Japan holds some of the oldest pottery on Earth. The Jōmon people made it. The Metropolitan Museum of Art notes that the name Jōmon comes from the “cord markings” on their ceramics. It dates the earliest phase, Incipient Jōmon, to about 10,500 to 8000 BCE. Its deep cooking pots with pointed bottoms rank among the oldest known.

c. 10,500 to 8000 BCE (Incipient Jomon) · , Japanese archipelago

Jomon potters press cord into wet clay

Earliest Jomon pots carry cord markings and rank among the oldest pottery known. The museum essay does not name the cord's plant, and a cord mark cannot name it.

Evidence Archaeological Confidence High (3 of 3)

Sources (1)

• The Metropolitan Museum of Art, Jomon Culture (ca. 10,500 to ca. 300 B.C.)

So which plant made the cord? The museum essay does not say, and a cord mark cannot say. A mark in clay records the twist of a cord. It does not record the species. Archaeobotanists have said it plainly for Chinese Neolithic pots: at least 14 plants made cord impressions there. Hemp is one candidate for a Jōmon cord. It is not a proven one.

The better evidence is smaller and lumpier. Botanists sorting plant remains from the earliest Jōmon sediment at the Okinoshima site, in Tateyama City, Chiba Prefecture, found small, hard fruits. (What people call a cannabis “seed” is botanically a fruit.) Under the microscope, the fruits matched cannabis and not hops. They have an egg-shaped side view, a round knob at the top and a shallow hollow at the base. Hop fruits look rounder from the side and lack the hollow at the base.

A follow-up study dated the fruits themselves with radiocarbon. The result: about 10,000 calibrated years ago, in the earliest Jōmon pottery phase. When the team published in 2009, they called these the oldest record of cannabis fruit in the world, “at this moment.”

c. 8000 BCE (about 10,000 calibrated years ago) · , Okinoshima site, Tateyama, Chiba, Japan

Cannabis fruits in earliest Jomon mud

Botanists identified cannabis fruits in earliest Jomon sediment, and a follow-up study dated the fruits themselves to about 10,000 calibrated years ago. Fruit is harder to import than fiber, so the plant likely grew nearby.

Evidence Plant remains Confidence Medium (2 of 3)

Sources (3)

• Kobayashi et al. 2008, Japanese Journal of Historical Botany 16(1): 11-18
• Kudo et al. 2009, Japanese Journal of Historical Botany 17(1): 27-31
• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518

Why do fruits beat fibers as proof? The Okinoshima authors give the logic. Fiber could arrive by trade from far away. A fruit points to a living plant nearby. They conclude that cannabis grew in Japan at that time, and that people may have grown and used it around the site.

Other readers push further. Dal Martello and colleagues note that the Okinoshima fruits are wider than wild cannabis fruits and lack the caruncle that marks wild-type fruit. (The round knob described earlier is a different feature. It is the one that separates cannabis fruit from hop fruit.) They read the fruits as likely already a domesticated form. They add that crops did not pass between Japan and China at that time, so farming of the plant plausibly began independently in at least two places. Ren’s team sees one East Asian origin. The two readings do not fit together neatly.

Fibers turn up too. The Okinoshima authors note cannabis fibers from earliest Jōmon sediment at the Torihama shell midden in Fukui Prefecture. Fiber is a weaker witness. Specialists find it hard to tell hemp fiber from flax under the microscope. Torihama seeds are dated to about 5000 BCE in one source and to 4200 to 3400 BCE in another, so the sources disagree on their date.

Inspect the pot. Each hotspot says what the evidence shows and where it stops.

123456

Select a numbered point on the artifact.

Can a mark in clay name a plant?

Turn to the Yellow River basin, 5000 to 3000 BCE. Britannica describes the Yangshao culture as millet farmers who made painted pottery and produced silk. Village sites such as Banpo, near Xi’an, were dug in 1954 to 1957. The people of Banpo farmed millet as their main grain.

The earliest cannabis grains reported from China are older than Yangshao. Dal Martello’s review lists grains from Zhuzhai, in the middle Yellow River region, dated to about 5900 to 5800 BCE. They are larger than known wild seeds, suggesting “possible cultivation.” We have not opened the original site report, so treat that date as reported, not confirmed.

Now a famous claim about Neolithic Chinese pottery. Hui-Lin Li reports in a 1974 account of cannabis in China that the Swedish geologist Johan Gunnar Andersson believed some cloth impressions on pots were hemp. The full paper is paywalled, so we cannot say which site or culture it names.

The trouble is the kind of evidence. Archaeobotanists now rank it:

This ladder comes from McPartland and Hegman, who built it for European evidence, and from McPartland’s later Asian review. Dal Martello’s team applies the same caution to China: early claims of hemp textiles that rest on fabric impressions on ceramics have been disputed.

Neolithic (Yangshao culture, c. 5000 to 3000 BCE, is the usual setting; not confirmed here) · , Yellow River basin, China (approximate)

Cloth marks on Neolithic Chinese pots

Hui-Lin Li reports that Andersson believed some cloth impressions on Chinese pots were hemp. A 2023 review calls textile claims based on impressions disputed, and an impression records a weave, not the plant.

Evidence Archaeological Confidence Low (1 of 3)

Sources (3)

• Li 1974, An Archaeological and Historical Account of Cannabis in China, Economic Botany 28(4)
• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518
• McPartland, Hegman and Long 2019, Vegetation History and Archaeobotany 28: 691-702

Myth check Disputed

The claim: “Cloth impressions on Neolithic Chinese pots prove early hemp weaving.”

Hui-Lin Li (1974) reports that Andersson believed some cloth impressions on Chinese pots were hemp. A 2023 review calls hemp-textile claims based on fabric impressions on ceramics disputed. A securely identified woven roll comes from Gaocheng Taixi, about 1600 to 1046 BCE.

Sources (3)

• Li 1974, An Archaeological and Historical Account of Cannabis in China, Economic Botany 28(4)
• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518
• McPartland and Hegman 2018, Vegetation History and Archaeobotany 27: 627-634 (ranks pottery impressions as less robust)

Taiwan gets the same treatment, with a bigger number attached. Many websites say Taiwan holds hemp-cord pottery about 10,000 years old. Cord-marked pottery is real in Taiwan. Britannica dates the Dapenkeng culture to the 4th and 3rd millennia BCE, roughly 4,000 to 6,000 years ago. Ren’s 2021 paper, which this chapter cites, reports older cord-impressed pottery from South China and Taiwan at about 12,000 years before present. We have not checked that date, and no source we read ties it to hemp. Cord marks do not name a plant. That is why the claim stays disputed.

4th and 3rd millennia BCE (Dapenkeng culture) · , Taiwan

Taiwan's cord-marked pots are real; a hemp link is not shown

Dapenkeng cord-marked pottery dates to the 4th and 3rd millennia BCE. Ren et al. 2021 report older cord-impressed pottery from South China and Taiwan, about 12,000 years before present, unchecked here. No source we read ties any date to hemp.

Evidence Archaeological Confidence Low (1 of 3)

Sources (3)

• Encyclopaedia Britannica, Dapenkeng culture
• Taiwan Panorama, Prehistoric Pottery in Taiwan
• Ren et al. 2021, Science Advances 7(29): eabg2286 (open access copy on PubMed Central)

Myth check Disputed

The claim: “Cord-marked pottery from Taiwan proves people used hemp 10,000 years ago.”

Cord marks do not name a plant, so they cannot show hemp use at any date. Dapenkeng cord-marked pottery dates to the 4th and 3rd millennia BCE. Ren et al. 2021 cite cord-impressed pottery from South China and Taiwan at about 12,000 years before present, unchecked here. We found no source tying any Taiwan pottery date to hemp.

Sources (3)

• Encyclopaedia Britannica, Dapenkeng culture
• Taiwan Panorama, Prehistoric Pottery in Taiwan
• Ren et al. 2021, Science Advances 7(29): eabg2286 (open access copy on PubMed Central)

Did hemp cross the steppe on a wagon?

Move west. By about 4,000 years ago, Ren’s team says, fiber-grown cannabis was spreading west into Europe and the Middle East, and Bronze Age finds support this. Europe gives a clear test of the ladder above.

McPartland and Hegman collected 136 studies of European hemp finds and rated each by robustness. Their verdict: no robust evidence supports Neolithic hemp use in Europe. One Copper Age site in southeastern Europe, from the Gumelniţa-Varna culture, gives solid evidence. More robust evidence appears in the Bronze Age, in the Yamnaya and Catacomb cultures. The Yamnaya were mobile herders of the western steppe, and their Early Bronze Age horizon dates from about 3300 to 2500 BCE.

Bronze Age (site dates vary; Yamnaya horizon c. 3300 to 2500 BCE) · , Southeastern Europe and the Pontic-Caspian steppe

Europe's first robust hemp finds

A review of 136 studies finds no robust evidence of Neolithic hemp in Europe. One Copper Age site and Bronze Age sites of the Yamnaya and Catacomb cultures give the first solid records.

Evidence Plant remains Confidence Medium (2 of 3)

Sources (3)

• McPartland and Hegman 2018, Vegetation History and Archaeobotany 27: 627-634
• McPartland, Guy and Hegman 2018, Vegetation History and Archaeobotany 27: 635-648
• Wilkin et al. 2021, Dairying enabled Early Bronze Age Yamnaya steppe expansions, Nature

Here the story forks. Three papers read the same landscape in three ways. Label every one of them a hypothesis.

Long’s team sees a jump in East Asian fruit finds between 5,000 and 4,000 years ago. They say it might be associated with an exchange or migration network across the steppe. They name the Hexi Corridor, a route through today’s Gansu Province, as a hub for spreading farmed plants and animals from the west. They also ask for better dated finds from southern Russia and Central Asia. McPartland’s pollen team links early cultivated hemp pollen to the Pontic–Caspian steppe, and to the Copper Age Varna/Gumelniţa culture and the Bronze Age Yamnaya and Terramara cultures.

c. 5,000 to 4,000 years ago (hypothesis) · , Eurasian steppe zone and the Hexi Corridor

Hypothesis: a steppe exchange network

Long and colleagues note more East Asian cannabis fruit finds from 5,000 to 4,000 years ago. They say it might reflect an exchange or migration network across the steppe, and they ask for better dated finds.

Evidence Plant remains Confidence Low (1 of 3)

Sources (2)

• Long et al. 2017, Vegetation History and Archaeobotany 26: 245-258
• McPartland, Guy and Hegman 2018, Vegetation History and Archaeobotany 27: 635-648

A pattern on a map is not a trail of seeds. The papers link hemp finds to Yamnaya territory. None of the abstracts we read traces one seed from one place to another.

What did early farmers do with the stalk and the seed?

Put aside the pottery marks. Three kinds of evidence show what farmers did with hemp: cloth that survived, words that name it, and seeds in storage.

Cloth. At Gaocheng Taixi, in Hebei, archaeologists found a complete roll of woven hemp from the Shang dynasty, about 1600 to 1046 BCE. Dal Martello’s review counts it among the securely identified hemp fiber finds. Hemp cloth also served as tax payment during the Zhou dynasty, along with grain.

c. 1600 to 1046 BCE (Shang dynasty) · , Gaocheng Taixi, Hebei, China

A complete roll of woven hemp

At Gaocheng Taixi in Hebei, archaeologists found a complete roll of woven hemp from the Shang dynasty. A 2023 review counts it among the securely identified hemp fiber finds in China.

Evidence Archaeological Confidence Medium (2 of 3)

Sources (1)

• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518

Words. Early Chinese texts call the plant ma (麻). Careful here: ma later became a generic word for bast fibers and some oil plants, so not every ma is cannabis. Still, texts from the first millennium BCE describe hemp cloth and hemp farming. The Rites of Zhou groups hemp with the cereals, which points to its use as food as well as fiber.

That is where the famous “five grains” come in. Hemp sits in some lists of the five grains and not in others. Two Han commentaries show the split. In the Rites of Zhou, a note on the physician’s passage glosses “five grains” as hemp, two kinds of millet, wheat and beans. In Mencius, Zhao Qi’s note on “the five grains” lists rice, not hemp. Dal Martello’s team uses a third version, with hemp beside millet, rice, barley and soybean.

“The five grains” was never one fixed set. Hemp was a grain in some lists and absent from others.

Zhou to Han texts (c. 250 BCE to 200 CE) · , Han China

Hemp is a grain in some lists, not all

Han commentaries give different five-grain lists. A note on the Rites of Zhou lists hemp. Zhao Qi's note on Mencius lists rice instead. Hemp counted as a grain in some lists only.

Evidence Written record Confidence Medium (2 of 3)

Sources (3)

• Zhou li zhushu, Zheng Xuan's note and Jia Gongyan's subcommentary on the physician's passage (zdic Classics)
• Mencius zhushu, Zhao Qi's note on Teng Wen Gong I (Wikisource)
• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518

Myth check Exaggerated

The claim: “Hemp was always one of China's five grains.”

Some classical lists name hemp. A note on the Rites of Zhou does, and Dal Martello's team counts hemp among the five. Zhao Qi's note on Mencius lists rice instead. The five grains was never one fixed list.

Sources (3)

• Zhou li zhushu, Zheng Xuan's note and Jia Gongyan's subcommentary on the physician's passage (zdic Classics)
• Mencius zhushu, Zhao Qi's note on Teng Wen Gong I (Wikisource)
• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518

Hemp also needed work before it became thread. The stalk holds fiber glued to a woody core. Farmers soaked cut stalks in water in a process called retting. Retting is a microbial process that frees the fiber from the stem. In water retting, bacteria ferment the pectin, the plant’s own glue, and the fiber comes free.

A poem in the Book of Songs mentions it. James Legge’s English version, as hosted by the University of Virginia, reads: “The moat at the east gate / Is fit to steep hemp in.” The poem repeats the line for two other plants.

c. 1000 to 700 BCE (dating of the Shijing as given by Dal Martello et al. 2023; dating confidence medium) · , Central Plain, China (poem; site unknown)

A poem soaks hemp in the east-gate pond

A short poem in the Book of Songs says the pond at the east gate is fit for steeping hemp. Steeping, called retting, is a microbial process that frees the fiber from the stalk.

Evidence Written record Confidence High (3 of 3)

Sources (3)

• Shijing, Dong men zhi chi, James Legge's English version (University of Virginia Chinese Text Initiative; same stanza on ctext.org)
• Quantity and Quality of Flax and Hemp Fibers Obtained Using the Osmotic, Water-, and Dew-Retting Processes. Materials 16(23): 7436 (2023)
• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518

Seeds. The best storage evidence comes from Haimenkou, a pile-dwelling village in Jianchuan County, Yunnan, 2,190 meters above sea level. In 2008, excavators recovered more than 800 charred cannabis grains there, dated between 1650 and 400 BCE. One hand-collected sample held more than 700 seeds in one lump. The same site yielded clean samples of millet and rice. The authors read the cannabis lump as a stored food grain, a cluster burned out of a storage unit.

1650 to 400 BCE · , Haimenkou, Jianchuan County, Yunnan, China

800 grains show a crop starting to split

More than 800 charred cannabis grains from a Yunnan pile-dwelling village fall between fiber-type and small-seeded sizes. The authors read a young crop starting to diversify. Seed size is a clue to use, not proof.

Evidence Plant remains Confidence Medium (2 of 3)

Sources (1)

• Dal Martello et al. 2023, Morphometric approaches to Cannabis evolution and differentiation from archaeological sites: interpreting the archaeobotanical evidence from bronze age Haimenkou, Yunnan. Vegetation History and Archaeobotany 33(4): 503-518

Seed size is a clue to use. Modern fiber and oil types have larger seeds, and modern drug types have smaller ones. The Haimenkou seeds averaged 3.39 millimeters long and 2.2 wide, and they land in the overlap between the two. The authors call the assemblage “indicative of a crop beginning to undergo evolution” toward specialized uses. In their broader sample, wider fiber-and-oil-type seeds appear from about 3000 BCE. A second line of smaller seeds begins around 2000 BCE. The authors suggest it may include plants chosen for more THC. That is the first hint, in seed shape, of a split that Chapter 3 will meet again.

Could anyone get high yet?

This is the question every reader asks, so here is the plain answer. The evidence cannot say.

Seeds and fiber cannot show psychoactive use. A seed does not hold a dose. A cord does not hold a smell. Dal Martello’s team says that seeds from sites dated about 8000 to 3000 BCE fall in the gray zone of undifferentiated cultivars. Those plants could indeed have been used for psychoactive purposes. The next sentence adds that food uses are more likely to have left traces in the ground. “Could” is a possibility. It is not a find.

What would count as evidence? Burnt plant parts in a burial or a brazier. Chemical residue from a vessel. A written description of someone using the plant. No find from 12,000 to 2000 BCE in the sources we read has any of those. Ren’s 2021 summary says ritual and inebriant use is documented in western China from at least 2,500 years ago. That belongs to the next chapter.

So the honest summary of this era is modest. People grew a hardy plant. They spun it, wove it, soaked it, stored it and, by the first millennium BCE, wrote about it. They may have done more with it. The record is silent.

Check your understanding

3 questions. Pick an answer in your head, then open the question to check it.

1 Which kind of evidence carries the most weight for ancient hemp?

1. AA cord mark on a pot
2. BFruits or seeds checked under a microscope and dated
3. CA popular website timeline
4. DA cloth impression in a pot's base

Answer: B. Fruits or seeds checked under a microscope and dated

Fruits and seeds can be matched to the plant and dated directly. Marks on pots cannot name the plant.

2 Where did the oldest directly dated cannabis fruits in this chapter come from?

1. ATaiwan
2. BYangshao villages on the Yellow River
3. COkinoshima, Japan
4. DHaimenkou, Yunnan

Answer: C. Okinoshima, Japan

Okinoshima's fruits date to about 10,000 calibrated years ago. Taiwan's Dapenkeng pottery dates to the 4th and 3rd millennia BCE, and no Taiwan cannabis fruit is dated in this chapter.

3 What can seeds and fiber from 12,000 to 2000 BCE show about getting high?

1. ANothing certain: they cannot show psychoactive use
2. BThat every early farmer smoked it
3. CThat the plant was too weak to affect anyone
4. DThat hemp and drug types had already split

Answer: A. Nothing certain: they cannot show psychoactive use

Seeds and fiber give no dose. Authors say psychoactive use was possible, but food and fiber uses are likelier to leave traces.

1. 1Which kind of evidence carries the most weight for ancient hemp?

   AA cord mark on a potBFruits or seeds checked under a microscope and datedCA popular website timelineDA cloth impression in a pot's base
2. 2Where did the oldest directly dated cannabis fruits in this chapter come from?

   ATaiwanBYangshao villages on the Yellow RiverCOkinoshima, JapanDHaimenkou, Yunnan
3. 3What can seeds and fiber from 12,000 to 2000 BCE show about getting high?

   ANothing certain: they cannot show psychoactive useBThat every early farmer smoked itCThat the plant was too weak to affect anyoneDThat hemp and drug types had already split

0 of 3 answered

Next, the record stops being silent. In Chapter 3, burials, braziers, a Greek historian and a shrine in Judah give direct signs of people using cannabis for more than fiber and food. The plant gets a second life.