Short answer: A 1998 Japanese study detected THC, cannabinol, cannabidiol, and cannabichromene in commercially available cannabis seed products. The highest concentrations were associated with bracts mixed with or attached to the seeds. The paper shows why an analyst must identify exactly what portion of a seed product was tested.

What samples did the researchers examine?

The authors purchased cannabis seeds marketed principally as bird feed and separated seed material from bracts. They used thin-layer chromatography and gas chromatography to identify and quantify cannabinoids. Reported cannabinoid content in the seed samples ranged from 0.10 to 2.02 mg per 100 g, while the bracts contained far higher concentrations—82.3 to 441 mg per 100 g.

Why does the bract distinction matter?

A botanical seed and resin-bearing plant material are not analytically equivalent. If bracts or surface material remain in a commercial lot, a whole-sample result can reflect those components rather than cannabinoids produced inside the cleaned seed. Grinding, washing, sorting, and subsampling can therefore change the measured concentration.

What did the mouse experiment test?

The investigators prepared a benzene-wash extract from seeds and administered it intravenously to mice at an amount corresponding to 3 mg/kg of THC. They assessed hypothermia, catalepsy, pentobarbital-induced sleep prolongation, and locomotor activity. The mixture produced pharmacological effects that the authors reported were stronger than the matched amount of THC alone.

That comparison suggests other constituents or interactions affected the animal endpoints. It does not show that eating cleaned seeds produces the same exposure. Intravenous administration of a concentrated solvent extract bypasses digestion and first-pass metabolism.

What can a modern forensic analyst take from this?

The durable lesson is about sampling and attribution:

• Document whether bracts, hulls, dust, or other plant fragments were present.
• Describe washing, grinding, extraction, and subsampling procedures.
• Report individual cannabinoids rather than treating the sample as one undifferentiated “cannabis” signal.
• Do not translate an injected mouse extract directly into a human oral-exposure conclusion.

The study is more than 25 years old and reflects specific commercial products, analytical methods, and regulations of its time. Current products require current testing.

Source

Matsunaga T, Watanabe K, Yoshimura H, Yamamoto I. Quantitative analysis and pharmaco-toxicity of cannabinoids in commercially available cannabis seeds. Yakugaku Zasshi. 1998;118(9):408–414. PubMed record (PMID 9779001).