#test-results 2025-04-26

how much does sample handling on my end affect the number

Testing queue: 4 submissions open, 50 awaiting dispatch.

thats not great

log the boring ones

the honest limits: hobby chain of custody is weak and any result here should be read with that in mind

a supplier disputing a result is rare here and it has never gone well for the supplier

update from the thing i asked about in the spring. i have kept every result on one vendor for two years and here it is

that is the most reassuring table i have seen on this site

it is a good dataset by our standards. let me now take some air out of it

please, thats why i posted it

fifteen lots over two years from one buyer, mostly one lab, all sent by the same person with the same habits.
what it shows well: no drift, no bad quarter, no surprise.
what it cannot show: whether you are getting a different product line to a first-time buyer, and whether the lots you happened to buy are representative.
repeat customers sometimes get better boxes. that is not paranoia, it is a normal commercial behaviour

i have wondered about that. no way to test it without a second identity and i am not doing that

someone did do that in 2024 with two addresses and the results were indistinguishable, for one vendor, once

which is one data point against a real hypothesis. i will take it

aggregate lookup — WuXi TIDES: 31 community results on file · median purity 98.9% · median content 9.6 mg · 6 results flagged as handling-compromised · oldest 2024-06

the flagged ones, who flags them

we do, by hand, when the submitter tells us the sample got mishandled. it is voluntary and therefore incomplete

and that flag is the single most valuable column in the whole dataset because it is the only one that admits its own error

sorry, late. is anyone tracking Bachem the same way

hardly anyone tests Bachem because the paperwork is already good and the price already reflects it

which creates a weird hole. the vendor we trust most is the one we have least independent data on

yes. trust and evidence are inversely correlated in this market and that should bother people more than it does

same for GL Biochem in a different way. everyone treats them as upstream and real so nobody bothers testing the resold vials

and the resold vial is the one you actually receive

correct. the reputation belongs to the manufacturer, the vial belongs to whoever repacked it

that sentence should be pinned somewhere

*repackaged, but yes

so what would you want next from this dataset

content results from people who are not me, on lots i did not buy, sent by people who are not careful. that is the missing quadrant

the careless-buyer dataset. nobody wants to fund it and it is the one that reflects reality

grim but true