Wisdom, Knowledge and Understanding LLC

10/19/2026

The h**p ban's 0.4 milligram total THC cap applies to the whole container. A 10 pack of 5 mg gummies carries 50 mg, 125 times the line, and the serving split on the label never enters the math. One 5 mg seltzer can is 12 times over on its own. The formulations that clear it are non-detect isolate products where the full potency panel on the entire package reads under 0.4 mg. A COA that reports per serving answers a question the statute does not ask.Full table of what leaves the h**p definition and on which date:

Intoxicating h**p ban 2026: plant-derived THC products lose h**p status December 11, synthetics November 12. Total THC math, the 0.4 mg cap, and what to do now.

10/16/2026

Three of the cylinders sold as 100 pound propane bottles fit on one extraction manifold. The fourth does not, and nobody tells you that until the fire marshal counts. The fire code limits connected cylinders in an industrial building to 735 pounds of water capacity per manifold, and a 100 pound propane cylinder is 239 pounds of water capacity: three total 717, four total 956. The same 735 applies to the cage of spares inside the building, and a second manifold needs 20 feet of separation from the first. Most labs learn this on the day throughput doubles, when the solvent order goes from three bottles to six and the extra three go wherever there is floor. The siting table, the cage distances and the list of changes after your certificate of occupancy that put you back in front of the inspector are in the new NFPA 58 guide:

10/12/2026

Four spare butane cylinders in a storage room turned a sprinklered F-1 lab into Group H-2 on paper. Every liquid line passed: 165 gallons of ethanol in listed cabinets under a 480 gallon limit, 10.6 gallons open in the winterization vessel under a 30 gallon open-use cap. The gas line did not. A 100 lb cylinder is about 650 cubic feet of v***r at normal temperature and pressure, so four in reserve is 2,600 cubic feet against a sprinklered maximum of 2,000 per control area. The reviewer runs that arithmetic by control area, not by building, and the hazardous materials inventory statement exists so you run it before he does. The fix was an exterior LP-gas enclosure for two of the cylinders. Wrote the full HMIS and HMMP build, row by row, here:

How to build the HMIS and HMMP a cannabis extraction fire permit requires: MAQ math per control area (120 gal Class IB, 1,000 ft3 flammable gas), a worked two-column lab, and the plan-review failures.

10/07/2026

Every liquid diamond cart that turned to sugar on the shelf was a melt. Warm THCA diamonds into sauce at 80 or 90C and they dissolve, but nothing changed chemically: cool the cart back to room temperature and the THCA comes back out, first on the glass, then through the whole mass, while the terpenes leaving through the mouthpiece make the solvent worse every day. A diamond only stays liquid when the THCA has been decarboxylated into THC, and that is a COA number, not a clock: THCA under 1% of total cannabinoids after a 115C melt that loses 12.3% of its mass as CO2 and stops bubbling. Pull it early at 3% THCA and the crystals seed in the cart within weeks. The label settles it before the shelf does: THC with THCA under 1% stays liquid, THCA with single-digit THC is a dab-jar product wearing a cart label. Wrote the full process here:

Liquid diamonds are decarbed THCA diamonds recombined with high-terpene extract. The decarb endpoint, blend ratios, viscosity by temperature, and why carts sugar.

10/05/2026

Euphorinol was built from an industrial ketone in seven chemical steps, and the THC-free certificate on the bottle cannot see it.The molecule behind the EPN label carries a carbon-nitrogen double bond on a cannabinoid skeleton. No cannabis plant has ever made that bond, so no reference standard exists, and a standard potency panel reports what it was calibrated for: C*D, CBG, no THC. The active ingredient shows up on third-party reports as an unidentified peak at mass 368, when it shows up at all. Several certificates leave it off the table while the packaging names it.A non-detect from a method that was never calibrated for the compound is not a finding. Wrote the full breakdown of how it is actually made, step by step, here:

Euphorinol is not extracted. It is built in seven steps from a methoxy acetophenone. The real EPN chemistry, the hazards, and why labs report an unknown peak.

10/02/2026

Edwards says 3,000 hours between oil changes on an RV series pump. Welch says 500 to 1,000, or sooner if the oil darkens. Both are right, because the manual interval assumes clean air and a pump backing a purge oven breathes butane, ethanol v***r, and water from every tray. That oil degrades on exposure, not on hours, which is why the lab that waits for the hour meter gets a vacuum oven that stalls at 28 inHg and a slab that will not finish. Pull the dipstick: milky or dark oil is the whole diagnosis. Cap the inlet and read ultimate vacuum; under 29.5 inHg with a capped inlet is the pump, not the oven. Change by color, run the gas ballast 20 to 30 minutes after every solvent-heavy job, and put the hour meter reading on the whiteboard next to every pump. The full interval table, the gasket material chart by solvent, and the downtime math are here:

Service intervals for vacuum pumps, closed loops, chillers, rotovaps and wiped film systems, a gasket material table by solvent, the downtime math, and five failures that look like process problems.

09/30/2026

A pesticide and mold screen caught the methylene chloride that put 49 v**e products on a Colorado recall, after those carts had sat on 405 shelves for 18 months with passing residual solvent results. The state panel is built around the solvents an extractor is allowed to run, butane, propane, ethanol, so a lab running the compliance method sees what the rule tells it to look for. Methylene chloride has no extraction or flavoring role in a cart; it is the workup solvent for a BF3 or mineral-acid conversion of C*D to delta-9, so finding it is a route fingerprint. The same batch carries what the panel cannot run at all: delta-8 above the 1 percent a plant ever makes, delta-10 and iso-THC the plant never makes, and olivetol from side-chain cleavage. An intake COA that passes on the state list has confirmed one thing, that the supplier did not leave butane behind. The three tells that separate converted distillate from botanical, with thresholds, are here:

How to detect converted THC: the residual solvents that flag each C*D-to-THC route, USP 467 limits, the delta-8 / delta-10 / iso-THC thresholds, the zinc tell, and what the Colorado methylene chloride recall exposed.

09/28/2026

Three hours at 240F does not bind THC to your coconut oil. THC is an oil-soluble molecule dissolving into another oil, and a cup of coconut oil holds grams of it before it comes anywhere near saturation, so a 4 gram infusion is finished dissolving about the time the flower warms through. Decarb is the only step that needs real heat: 240F for 40 minutes converts THCA to THC, and after that the temperature is doing one thing, cooking terpenes out of the jar. At 180F for 90 minutes with the lid on, the oil carries the same milligrams as the 3 hour version and a lot more of the plant's nose. The stronger long infusion people swear by is usually the decarb they never finished in step one. Milligrams come from the flower's THCA and your decarb, not from time in the pot. The carrier oil side, why MCT hits in an hour and olive oil takes two, is written up here:

MCT oil delivers 2.5x the bioavailability of olive oil for THC. Sublingual onset in 15-30 min vs 45-90 min oral. Titration protocol with specific mg ramps, carrier oil comparison table, and 5 dosing failure modes.

09/25/2026

Your THC non-detect and your 85 percent C*D recovery come from the same decision: where you drew the cut on the C*D band. Take the front 85 percent of the band and the delta-9 front stays in the waste, non-detect at a 0.01 percent LOQ. Push the cut to 97 percent of the band to chase recovery and the collected fraction carries 0.2 percent THC or more, which passes a 0.3 label and fails every market that tests at 0.1. The overlap zone between those two cuts goes in its own vessel and back on the column as feed on the next run, which is how a campaign past 10 runs claws back about half of what the conservative cut gave up. Column choice, mobile phase, loading, and cost per gram are in the full breakdown here:

Preparative HPLC cannabinoids guide: C18 elution order, ethanol gradients, 1 to 10 percent loading, cut width vs recovery, cost per gram, and 8 failure modes.

09/23/2026

A 5 percent LEL reading at rest with no run going is a slow leak or a drifted sensor, and the panel cannot tell you which. Most rooms alarm weekly and the crew learns to hit reset, which is how a real leak gets ignored on the one day it matters. Walk the portable at floor level along every sight glass gasket, every valve packing, and the spent biomass bin, because butane v***r is twice as dense as air and that is where the number lives. If the portable finds it at a gasket, replace the gasket and re-torque. If the portable reads zero everywhere, the fixed sensor drifted; re-zero it and bump it on 50 percent LEL span gas before anyone trusts the panel again. The reverse failure is worse: a catalytic bead poisoned by silicone spray or RTV sealant reads low, so a panel that never alarms is not evidence of a clean room.

Wrote the full false-high table and the reading bands by solvent here:

Complete safety equipment extraction lab checklist covering C1D1/C1D2 classification, LEL gas detection, ventilation calculations, PPE by method, and fire suppression. The compliance guide inspectors actually use.