Poisonous Plants in 1820 New England Sampler: Statistical Odds

Sampler depicts 15 plants; all have medicinal uses, only 1 labeled poisonous. Question: Is this random or intentional (medicinal catalog)?

Defining "Poisonous" Is Complex

• Obvious toxins (oleander, poison ivy) = small minority.
• Medicinal plants toxic at high doses.
• Edible plants with toxic parts (apple seeds, tomato/potato foliage, fig sap).
• Context-dependent: chocolate (dogs), holly berries (humans vs birds), pokeweed (age-dependent), tapioca (cyanide precursors).

Expert Estimates

• ~30% of plants "poisonous" in some context.
• 1/15 (6.6%) labeled poisonous = low but plausible.
• All 15 medicinal = unlikely random (~50% expected).
• 1/15 deadly to humans = typical (1–3 expected).

Sampler Context

Themes: culinary herbs, wildflowers, medicinal. Medicinal samplers rare. Even culinary/wildflower samplers ~50% medicinal overlap. Result suggests intentional medicinal selection.

Statistical Model (Follow-Up)

Operational definition: "poisonous" = online description says poisonous despite 19th c. medicinal use. Estimated odds for 1/15: 10–20%. Peak at 2–3/15 (40–50% each). No formal studies exist.

Expert insight provided by Melissa.