#test-results 2026-04-19

a single sample tells you about a single vial. that is genuinely all it tells you

which lab

unrelated but is quantity testing available as a separate thing or bundled

slightly off topic but i told QST about a result they would not have liked and they did not argue. credit where it is due

the widest spread in my data is about 2.6 percentage points across five lots from one supplier

test it again

does anyone log the negative results, the boring ones

a point or two between the certificate and an independent result is inside what two labs disagree by

anyone tested the same lot twice at different labs

a result without a lot number and a purchase date is an anecdote with a pdf attached

how much does sample handling on my end affect the number

sample handling on your end absolutely affects it. draw cleanly, ship fast, do not let it sit warm

weigh the vial before you reconstitute. it costs nothing and catches the common failure

selection bias

update from 6 months ago: retested the same supplier, result came back better, logged both

one bad test does not condemn a supplier. one good test does not clear one. both halves get ignored equally

certificate said 99.2 and the independent came back 98.6, is that a problem

both back. same vial, split into two, posted the same day in one envelope to two labs

so which one is right

both. and the table already explains itself if you read the bottom two rows

the threshold?

lab B reports down to 0.05% and lab A stops at 0.10%.
so lab B counted two extra small peaks that lab A folded into the baseline.
those two peaks came out of the main peak, which is why B reports lower purity.
nothing disagrees. they made different reporting choices and both documented them

this is the single best example we have of why 0.8 points is not a discrepancy

and note the content figures are 0.07mg apart, which is essentially perfect agreement. quantity travels between labs better than purity does

why is that

because quantity is measured against a standard and the answer is a mass, and mass does not care about your integration baseline. purity is a ratio of areas and areas are full of judgement calls

which is a good argument for buying content tests over purity tests, incidentally

that has been my position for a year and nobody listens because purity is the number vendors advertise

inter-lab diff — sample SWB-25-1104 (split): purity 99.1% vs 98.3% (Δ 0.8) · content 9.62 mg vs 9.55 mg (Δ 0.07) · flag: within expected inter-lab range

the bot has a range built in?

one to two points on purity, and it is a convention we picked, not a measured tolerance. it is there to stop people panicking

and if a split came back four points apart the flag would fire and we would go and read both chromatograms properly

has that ever happened

once, on a TFC sample in 2025. 99.4 and 94.9

and the resolution was?

one half had been decanted into a tube that had previously held something else. the submitter admitted it and withdrew the result

which was the right thing to do and we still lost the money

so a big disagreement is usually us and not the labs

in every case we have chased, yes. the labs have not been the problem once

*not once that we could demonstrate. i would not claim more than that

fair correction

so what do i file for this vial

file both numbers, both thresholds, both wavelengths, and do not average them

why not average

because averaging two different measurements of two different things creates a number that neither lab would sign. keep them side by side

the archive holds pairs, not means. that has been the rule since the first split

and the Saber split we have pending will be the fourth pair on file, so slowly we are building something

worth the extra postage honestly

it is the most informative money this channel spends