How this is calculated

Four things stand between a jar of flower and a dosed cookie, and each one loses you some of what you started with:

• The molecule changes. Raw cannabis contains THCA, not THC. Heat converts it, and the conversion sheds mass.
• Decarb is not complete. Some THCA never converts, and if you overshoot, some THC converts onward into CBN.
• Infusion does not capture everything. A good deal stays in the plant material and on the cheesecloth.
• The batch divides unevenly. Not a number you can calculate — but the largest real-world source of a surprise.

Three of those four are estimates with genuine spread. That is why every result here is a range rather than a single figure, and why the ranges are wider than the confident numbers published elsewhere.

THCA (C₂₂H₃₀O₄, 358.47 g/mol) loses one molecule of carbon dioxide (44.01 g/mol) to become THC (C₂₁H₃₀O₂, 314.46 g/mol).

314.46 ÷ 358.47 = 0.877

1 g of pure THCA  →  877 mg of THC (the rest leaves as CO₂)

This is the ceiling. No method, machine or technique beats it, because it is arithmetic rather than chemistry you can optimize.

The trap: a label reading “Total THC” has already applied this conversion for you. Applying it a second time under-reports by about 12%. A label reading “THCA” has not. The calculators ask which one you have for exactly this reason.

Efficiency is the share of available THCA actually converted. 240 °F for 30 to 45 minutes is the consensus sweet spot — hot enough to convert, cool enough that a useful share of the monoterpenes survive. Past the peak, THC begins converting into CBN immediately, so longer is not safer.

The time and temperature schedule behind the decarb calculator is a lookup table of discrete steps rather than a smooth curve. The reaction genuinely is first-order and could be modeled continuously, but a home oven that swings 25 °F, an unknown starting moisture content and an uneven bed of flower do not support that precision. A table of “at this temperature, this long” is the honest shape of the advice.

Of the activated cannabinoid, how much ends up in the fat rather than in the spent plant material. This is the loosest number in the whole chain, and the one most recipe sites quietly paper over with a single fudge factor. It moves with time, temperature, agitation, plant-to-fat ratio and how hard you squeeze the cheesecloth.

Sealed beats open, mostly because open systems lose volatiles and run hotter at the surface than the cook believes. The bands above are deliberately wider than figures published elsewhere.

Nothing goes above 200 °F. Past roughly that point terpenes boil off and THC begins heading toward CBN, so every ceiling in the table sits under it — a fat infusion is a long warm soak, not a simmer. Alcohol extraction has no temperature at all because heat drives the alcohol off; it is done cold and slow instead. A test enforces that ceiling, so raising one has to be an argument rather than an edit.

The usual starting point for a fat infusion is 1 oz flower per cup of fat. That is a convention rather than a rule — the calculator works from whatever you actually use, and there is nothing wrong with going weaker.

Carrier choice does two jobs: it supplies a density for converting between weight and volume, and it applies a modest efficiency modifier. Saturated fats hold cannabinoids somewhat better than unsaturated ones, but the effect is small next to method — a sealed jar of olive oil beats an open pot of butter every time. The modifiers stay near 1.0 on purpose; inflating them would imply a precision the evidence does not support. No modifier can push transfer above 100%.

labeled mg   = grams × percent × 10

activated mg  = labeled mg × 0.877 × decarb efficiency      (THCA labels)
              = labeled mg × decarb efficiency              ("Total THC" labels)
              = labeled mg                                  (distillate, RSO)

infused mg    = activated mg × transfer efficiency × carrier modifier

mg per ml     = infused mg ÷ carrier volume in ml

mg per serving = mg per ml × infusion used in ml ÷ servings

The low end of every band multiplies the low factors together, and the high end the high factors. The true value is very unlikely to sit at either extreme — which is precisely why a band is shown rather than a midpoint dressed up as an answer.

When a calculator works backwards — “I want 5 mg per serving, how much do I need?” — it sizes against the low end of your range, never the midpoint.

That is deliberate and it is not a rounding convention. If the batch turns out stronger than planned you get a pleasant surprise; if it were sized against the strong end and came out weak, you would have wasted the flower. Planning for the weaker outcome is the version of this mistake you can recover from. The same asymmetry governs how recipes split infused and plain fat.

All volumes are US customary, converted through milliliters:

Nothing is ever printed to more than one decimal, and nothing above 10 mg carries a decimal at all. A tenth of a milligram in a cookie is noise, and printing it implies a measurement nobody took. Intermediate values are never rounded — only the final display is.

These labels are descriptive only. Nothing on this site maps a dose to an effect, a condition or a person, and the tiers are not a recommendation for anybody.

• What your flower actually contains. Label percentages vary between labs and between batches, and potency drops in storage.
• How evenly it mixed. Cannabinoids ride in fat, and if that fat is not worked right through the batter some pieces carry far more than others.
• How it will affect you. Absorption depends on your body, what else you have eaten, and your tolerance. Two people eating identical cookies do not have identical evenings.
• Anything about bioavailability. These numbers describe what is in the food, not what reaches your bloodstream.

A home potency tester will get you closer than any calculator. Nothing but a lab will get you certainty.

These constants are our best reading of the available evidence, not received truth. If you have data that contradicts one of them — particularly the transfer efficiency bands, which are the shakiest numbers here — we want to see it.

Email hello@cannachef.app. This page exists to be argued with, and every figure on it lives in one file so a correction is a small change rather than an archaeology project.