Skip to content
DIY Agriculture

Sesame — Yield

Metadata research draftEvidence limits

⚠️ Status: metadata draft. Every DOI below is a candidate metadata record from the survey research/sesame-seeds/report.md — title and identifier only, not screened evidence. No yield number is accepted or quoted on this page — not even a range. The survey’s rule: “All field-yield values require cultivar/capsule type, environment, density, water/fertility, maturity, harvest method, shattering loss, and clean-seed definition. Seed yield and extracted-oil yield remain separate” (§3.10). Substantive values remain Unknown until S-#### source notes exist (docs/methodology/evidence.md).

Seed yield vs extracted-oil yield — never combine

Section titled “Seed yield vs extracted-oil yield — never combine”
  • Clean-seed yield (what the plot returns as harvested, cleaned seed) and extracted-oil yield (oil recovered per unit of that seed) are different quantities with different determinants. Seed yield belongs to the field; oil yield multiplies it by a processing recovery that is itself Evidence: Unknown. yield.seed_kg_ha and yield.oil_g_per_kg_seed are separate fields in data/crops/sesame.yaml, and both are null. Evidence: Known (project rule; survey §2, §3.10).
  • Reporting “kg/ha of sesame” without saying seed or oil, or quoting an oil figure as a field yield, is the primary error this page exists to prevent.

Every yield value must carry its context

Section titled “Every yield value must carry its context”
  • Before any number can be published it needs: cultivar and capsule type · environment/location · plant density · water and fertility regime · maturity at harvest · harvest method · shattering/seed loss · definition of “clean seed” (survey §3.10). A yield without these is uninterpretable and will not be accepted into data/crops/sesame.yaml.
  • Loss interacts with yield directly: dehiscence-driven shatter loss, pest loss, and threshing/cleaning loss all sit between standing crop and clean seed (harvesting.md). A “yield” that ignores them is a different, larger quantity.

Candidate evidence (coverage only)

Section titled “Candidate evidence (coverage only)”
  • Tillage, sesame seed yield, energy budget, and economic return (10.3906/tar-1011-1505, 2012) — a systems record; no result claimed.
  • Production challenges/opportunities review for Myanmar (10.3390/su12093515, 2020); Uganda production/export trends (10.1051/ocl/2020073, 2021) — country-level reviews/trends, not plot yields, and not transferable across environments.
  • Yield/quality research review (10.3389/fpls.2022.863521, 2022) — a review candidate; what it aggregates is unread.
  • Smallholder technical efficiency (10.1007/s43546-023-00604-3, 2023) — a farm-economics framing of yield gaps, not a yield value.
  • Mexico: genotype × environment interaction in sesame germplasm (10.1002/agj2.70077, 2025; metadata-level candidate — survey §4). No farm budget or household recovery outcome is established.
  • These titles establish that yield topics are covered by the corpus. None of them yields a usable number at this stage, and national statistics from other continents say nothing about a Mexican household plot.

“Clean seed” must be defined before any comparison

Section titled ““Clean seed” must be defined before any comparison”
  • A yield number is meaningless until “clean seed” is defined — winnowed only? screened to what? at what moisture? The survey requires the clean-seed definition to travel with every field-yield value (§3.10). A household measurement and a research measurement with different definitions are not comparable, however similar the plots.

Small-scale recovery gap

Section titled “Small-scale recovery gap”
  • What a household actually obtains — clean seed after cutting, drying, threshing, winnowing, and cleaning, with labor and losses attached — is Evidence: Unknown/Needs experiment (open questions 10–11; experiments E2–E3). A field-trial yield cannot be assumed to equal household recovery unless product definition, harvest method, and loss accounting match.
  • Oil yield at household scale is a second multiplication by an unknown: small-press recovery per kg of clean seed is Evidence: Unknown (open question 13, marked “Needs experiment”).

Units and recording

Section titled “Units and recording”
  • Per docs/methodology/evidence.md, any future value is recorded with level, type, conditions, and source id (S-####); seed yields in g/plant, g/m², kg/ha; oil as g oil/kg clean seed — never blended. data/crops/sesame.yaml yield note already fixes: “never combine seed and extracted-oil yields.”
  • Yield scales follow docs/methodology/units.md: choose the scale that matches the claim (g/plant and g/m² for a household bed; kg/ha only for field comparisons) and always state the area a number refers to.

Corpus share and transfer risk

Section titled “Corpus share and transfer risk”
  • The analyzer found 107 yield/economics metadata hits, but this broad unscreened count includes noisy or inapplicable records. The selected Myanmar, Uganda, smallholder, and Mexico candidates describe different systems; none may be transferred to a Mexican household plot without matching context.
  • Every record that survives screening enters with its full context list (above) or not at all — this is how the project avoids manufacturing a fake “typical” sesame yield by averaging unlike conditions.

Unknowns and experiments

Section titled “Unknowns and experiments”
  • Seed yield for any cultivar in the intended environment — Evidence: Unknown.
  • Household clean-seed recovery after all loss routes, with labor-hours — Evidence: Unknown; Evidence: Needs experiment (E2, E3).
  • Small-press oil recovery per kg of clean seed — Evidence: Unknown; Evidence: Needs experiment (open question 13).
  • Mexico-specific plot yields beyond the Guerrero G×E metadata lead — Evidence: Unknown.
Section titled “Related pages”

Provenance

Section titled “Provenance”