We know there is variance in d8 lab results but haven’t seen anyone quantify it. So we did. We sent a homogenized sample to 9 accredited 3rd party laboratories. This is 100% an ongoing experiment so if there are recommendations for additional labs please share. If any labs see this and are interested in an intra-lab study that be fun too.
d8 average: 89.8%
d8 standard deviation: 5.94%
d9 average: 2.05%
d9 standard deviation: 2.33%
Variables that might explain variations-
Processor sampling technique
Experience level of Third-Party Lab personnel
Use of stock vs. modified vs. custom Third-Party Lab sample prep and methods
Differences in lab certifications, for example, AOAC certified vs non-AOAC certified
Use of orthogonal testing (multiple testing methods) vs single testing methods
Overall - No enforced standard methods for testing cannabinoids
Conclusions
It is fairly easy to get ND-d9
d8 results vary significantly
The results from @kcalabs were perfectly within the average and standard deviation justifying their crown.
Proverde told us not to send any more d8 samples
For the sake of transparency, I have uploaded all of the COAs. We are not claiming/stating the material is x% d8/y% d9 so I have not included chromatograms (not all c-grams were shared with us anyway). This post is not about the accuracy of the reported results, its to emphasize the differences in COAs from different labs being released to describe the cannabinoid content of a homogenous sample.
edit: I owe @StableScott a thank you for his integral involvement in this project.
Some labs seem to not see the d9 at all and are not running any additional or confirmatory analysis. Other labs seem to be quantifying off the HPLC only.
Would be interesting to add a time domain to this. Make up a homogenized batch, pull samples into vials with enough for four tests per lab and pop them in a freezer. Send one a month for four months and see what the data looks like. Then you’d have n=4 per lab as well as values as a function of time to examine. But of course, time and money make this difficult…
Definitely. Sending the same sample multiple times to each lab would be interesting as well. We did this with green scientific but samples were taken at a different time so the data was not included in the study
Probably because their HPLC is the instrument within their scope of accreditation for cannabinoid potency testing. Not sure how it would rub ISO to know that values are being reported for a method within the scope of accreditation only after being confirmed by a method/instrument that is not within the scope of accreditation.
Great idea. You’ll be the go-to expert witness when the inevitable high-profile case involving D8 appears, and the defense revolves around testing ambiguities. If the experts can’t get it right (“Right” defined as consistent within a statistically reasonable level of variation, which these pretty clearly aren’t) then how can the government’s single-source testing evidence look persuasive?
FYI, only D8-THC, D9-THC, CBD and CBN% was included in the data set. If a cannabinoid was not reported on all the COAs the cannabinoid was excluded from the data set.