Infinity Supercritical LLC

Publication Title | How Accurate is Potency Testing

Cannabis Search Engine Series

Cannabidiol | Cannabinoid | Cannabis Sativa | CBD - Marijuana | CO2 Extraction | Oil Separation | Supercritical CO2 Fluid Extraction search was updated real-time via Filemaker on:

Cannabidiol | Cannabinoid | Cannabis Sativa | CBD - Marijuana | CO2 Extraction | Oil Separation | Supercritical CO2 Fluid Extraction | Return to Search List

Search Completed | Title | How Accurate is Potency Testing
Original File Name Searched: RingTestOShaughnessys_Aut11.pdf | Google It | Yahoo | Bing

Page Number: 001
Previous Page View | Next Page View

Text | How Accurate is Potency Testing | 001

Samples A, C, and D: replicates

The results of all tests performed are shown in the table on the page at right. For the herbal cannabis samples (A- D), potency is expressed in percent, so that 1% equals 10 milligrams of a given cannabinoid per gram of plant mate- rial. For the tinctures (F-G), it is expressed in milligrams per milliliter (mg/ml).

Although CBN is not produced by the cannabis plant it- self, it is commonly formed by degradation of THC under the in uence of light, storage and heat. It is therefore not surprising that all samples contained at least some CBN. Levels were generally very low, on average around 0.6% or less in all samples. However, reported CBN values in Sample G varied from 0 to 1.44%. Such a wide variation in results is unacceptable for quality control.

Sample B: high CBD

While most labs measured the potencies of samples A-D in terms of percentage of total sample weight, labs #2, #5, #7, and #9 reported them in terms of dry weight instead. In order to convert their measurements to total weight, labs that report dry weight need to report water content of the sample. Measurements of water content in samples A-D ranged from 6% to 9%. For the sake of consistency, the data in the table are all reported in terms of total weight.

Because most labs were informed that they were partici- pating in a ring test, they were able to give these samples extra attention. Several labs revised or corrected results af- ter initially misreporting them. In a couple of cases, this appears to have been due to simply misreading or misre- porting the data. In one case, a lab initially misreported a sample as having no cannabinoids at all. After being ques- tioned about it, they re-tested the sample and came up with reasonable results. Two other labs, whose results were gen- erally reasonable, slightly misreported some initial data, but later corrected them on their own initiative.

Samples F and G: extracts

In principle, when the majority of labs re- port similar values, and one or a few labs have found signi cantly different results, it should be the outliers that need to take a closer look at their methodology.


An examination of the reported data provides some in- sight into technical issues that play a role in the laboratory. For example, fresh material might normally be expected to contain substantially higher amounts of acidic cannabi- noids (THCA, CBDA) than neutral ones (THC and CBD). However, a couple of labs reported the opposite in certain tests (e.g, lab 3 for CBD and lab 7 for CBD and THC in samples A, C and D). Based on reported separation issues in HPLC, this might be caused by an overlap of CBD with the cannabinoid CBG (Cannabigerol), causing the two to be confused in the detector. CBG is present to some degree in most cannabis samples, so it should be con rmed that the chromatographic peak for it is positively identi ed in the chromatogram, and does not interfere with any other cannabinoid analyzed. The same applies to the commonly found cannabinoids Cannabichromene (CBC), Tetrahydro- cannabivarin (THCV) and their acidic precursors.

O’Shaughnessy’s • Autumn 2011 —17—

How Accurate is Potency Testing?

RING TEST: A standard procedure in the analytical testing industry for external quality-control assurance, in which identical samples are sent to a variety of testing facilities in order to compare results.

other cannabinoids in the cannabis plant (e.g. CBD is pres- ent as CBDA). Only after heating are THCA and CBDA converted into their more potent, non-acid forms, THC and CBD. This process is known as “decarboxylation.”

labs using HPLC, as CBDA standards are not readily avail- able.

Want to know the potency of your medicine? How much, if any CBD, does it contain? Has it been sprayed with dangerous pesticides? Is it infested by molds or bac- teria? The only way to answer these questions for sure is to have it tested by an analytical lab.

When cannabis is smoked or cooked, the user mainly ingests activated, non-acid cannabinoids. However, if cannabis is consumed raw or in an unheated oil or alco- hol extract, the cannabinoids remain in their natural acid form. The properties of cannabinoid acids have not been well investigated. Although THCA is not psychoactive, it is thought to have some medicinal activity.

Chemical standards of known potency are crucial to ac- curately calibrate testing equipment. But because they con- tain controlled substances, cannabinoid standards are sub- ject to DEA regulations, making them dif cult to obtain. As an alternative, some labs resort to obtaining them from illicit, gray market sources; others manufacture their own standards without independent validation. The reliability of such standards is uncertain.

Many labs now offer testing services for the medical can- nabis industry. How accurate are the results they provide?

Because it is performed at room temperature, HPLC an- alyzes cannabinoids in the chemical form in which they are actually present in the sample. For fresh herbal material, that means lots of THCA and a little free THC (as a result of ‘spontaneous’ decarboxylation during harvest, process- ing and storage). Consequently, labs using HPLC reported two different entries: THC and THCA. From this it is pos- sible to calculate the effective “THC Total” by adding the measured THC and THCA. In order to adjust for differ- ent molecular weights, the THCA measurements need to be multiplied by a factor of 0.877 (one lab failed to do this correctly). The same applies to the measuring of CBD and CBDA.

The two liquid alcohol extracts (tinctures) were prepared by soaking cannabis in ethanol and ltering out the solid residues. After proper shaking, liquid extracts should be perfectly homogenized with respect to cannabinoid con- tent, regardless of the presence of some minor solids such as bers from the paper lter, or small particles of plant material. To our surprise, the tincture results proved very inconclusive and varied with the kind of testing technology employed. For unknown reasons, labs using GC generally reported higher readings for THC and CBD than labs using HPLC. For GC, the THC measurements for samples F and G averaged 42% to 46% higher than for HPLC, while CBD measurements were 20% to 33% higher. Theoretically, there should have been no discrepancies between GC and HPLC analysis.

Our investigation was launched in the winter of 2010/11. An identical set of samples was submitted to 10 labs. The labs were asked to measure THC, CBD, and CBN, the three major cannabinoids for which testing is generally available. To encourage participation, the identity of the labs was kept con dential; in this report they are identi ed by numbers only (Lab 1, 2 etc.).

In most cases, lab results were consistent to within 20% of each other. To some degree, the differences in results might be explained by natural variations in the consistency of the cannabis samples used; to some degree, by differ- ences in lab procedures. In certain cases, there were glaring discrepancies suggesting laboratory error. Three of the 10 labs performed poorly on half the tests. Particularly in the case of liquid alcohol extracts, test results were troublingly inconsistent. Nevertheless, the general results showed good agreement between most lab results involving herbal sam- ples. Lab performance can be expected to improve in the future as the industry responds with improved standards.

In contrast, when samples are analyzed by GC, the high temperature of the injector causes the cannabinoids in the sample to instantly decarboxylate, converting all THCA into THC before the sample enters the chromatographic column. Therefore, labs using GC have only a single corre- sponding entry, “THC Total.” In the same way only a value for “CBD Total” is reported for GC.

Alcohol tinctures are in principle easier to measure than raw plant samples, because tinctures (after proper dilu- tion of the sample) are immediately ready to be analyzed. This raises obvious questions about the accuracy of current methods for testing tinctures and other extracts.


Labs were asked to examine six different samples: four herbal cannabis samples (A - D), and two liquid (ethanol) extracts (F - G). The cannabis samples were taken from herbal material homogenized in a blender to minimize vari- ations in potency. Samples were stored in a cool, dark room in tightly closed containers until sent to the labs. All testing was blinded: labs did not receive any information about the samples that could have skewed their analytical results.

These samples were exactly identical, and intended to check the reproducibility of participating labs, including their extraction protocol, sample preparation, and analyti- cal methodology. Samples consisted of roughly one-gram packets of a THC-rich cannabis mixture that had been ho- mogenized in a kitchen blender, followed by manual stir- ring.

This sample consisted of one gram of a similarly pre- pared mixture of CBD-rich herbal cannabis. The sample was intended to check for labs’ capability of identifying and quantifying CBD.

Samples F and G were alcohol tinctures of about one milliliter each. Both were prepared from a single CBD-rich herbal mixture. Material in Sample F was decarboxylated by heating in a closed glass jar at 100°C for 90 minutes before soaking in alcohol. Sample G was prepared from the same material, but unheated. Samples were extracted in 99% pure ethanol for 12 minutes and ltered in cheese- cloth, then a coffee lter. These samples were intended to evaluate only the testing methodology of the participat- ing labs, by removing the need for extraction and sample preparation.

Different kinds of lab equipment were used to test the samples. Five labs employed gas chromatography (GC), in which the sample is rst vaporized under heat. The result- ing gases were subsequently analyzed either by means of a ame ionization detector (GC-FID, used by four labs), or a mass spectrometer (GC-MS, used by one).

As expected, labs using HPLC found high levels of THCA and CBDA and lower levels of active THC and CBD in green, unheated samples. Labs using GC generally reported comparable total levels of THC and CBD in the same samples.

One potential explanation could be that our sample ex- tracts were made in a different solvent (ethanol) than labs commonly use to analyze their own samples or to dissolve their calibration standards in. This could lead to signi - cantly different outcomes, as a result of different expansion of solvents in the GC injector. HPLC or TLC methodology, in contrast, would not be affected by this.

Four labs used a technology known as high-pressure liq- uid chromatography (HPLC), in which the sample is ana- lyzed by forcing it at high pressure through long columns to separate its components.

Although the samples A, C, and D consisted of the same homogenized cannabis material, the results ranged from 4.16% to 14.3% THC. Although lack of uniformity in the samples might account for some degree of the variance, the wide variation was mainly due to several results that were far out of line with the average reported by most labs.

GC and HPLC are by far the leading technologies for cannabis testing. One lab used a third method, Thin Layer Chromatography (TLC), which is commonly used for qual- itative analysis ( ngerprinting/pro ling) of cannabis sam- ples. As a tool for quanti cation (potency testing) it tends to be less accurate, because the results are scored inexactly by visually judging the size or density of a spot.

In a set of 40 THC measurements of cannabis samples, seven were more than 25% out of line with the average reported by other labs. In another set of 10 CBD measure- ments, three were more than 25% out of line. Ignoring these outlying results, the reported potency of the three identical samples ranged from 8.4% to 12.5% THC —a variation of about +/- 20% from the mean.

Even though care was taken to homogenize the herbal samples, there is always a chance that one or more showed some deviation in cannabinoid content. Therefore, liquid samples (F, G) were included because they eliminate the need for extraction and have already been fully homog- enized before distributing the samples. Analysis of these liquid samples therefore purely depends on the accuracy of the methodology applied. Liquid samples F and G were prepared from homogenized herbal material like the other

Fresh herbal cannabis does not contain signi cant amounts of free THC. Instead, the plant produces its bio- logical precursor, THC-acid (THCA), which lacks THC’s activity and is not psychoactive. The same is true for most

Three of the 10 participating labs reported anomalous results on at least half of the samples, raising obvious ques- tions about their accuracy. Two other labs reported a single anomalous result. One of these admitted to technical dif- culties in obtaining the necessary chemical standard for testing CBD-acid, which is indeed a common concern for

continued on next page

By Dale Gieringer and Arno Hazekamp


Another explanation could be problems in calibrating the equipment to the relatively weak potency of the tinctures, which was signi cantly lower than that of normal herbal samples.

Because we do not know what the true potency values were for our test samples (we would need a reliable lab to tell us, but evaluating reliability is the purpose of our study), we cannot say with certainty which lab gave more accurate results. We can, however, look at the reproducibil- ity of the same sample in a single lab and the consistency of results between labs. In principle, when the majority of labs report similar values, and one or a few labs have found signi cantly different results, it should be the outliers that need to take a closer look at their methodology.


Image | How Accurate is Potency Testing

Supercritical CO2 Fluid Extractor | Oil Extract Using Infinity Supercritical CO2 Extraction System - Botanical CO2 Extraction System - Go to website

Search Engine Contact: