What Tested Actually Means: Three Different Things, and We Reach the Third Once

Tested names three things: a building was inspected, a product was tested once, or this lot was tested. Almost every badge in the category is one of the first two. TARE holds three certificates and, on 6 September 2026, one names the lot inside a TARE tub — and that one was run by ViaCore, the manufacturer’s own laboratory.

One word is doing three jobs, and only one of the three is about the container in your hand.

What follows is the ladder, the four questions that tell you which rung a document is standing on, and then the uncomfortable part — those same four questions run against TARE’s own three documents, including the one that scores highest on a sample that was never a TARE product.

What does “tested” mean on a supplement label?

Three different things, and they are not close to equivalent. The building was inspected — a facility audit, on a date, against a written standard. A product was tested once — a certificate run on a representative sample of a formula or an ingredient. This lot was tested — a certificate naming the lot, with a method beside every row and a specification beside every result.

Only the third one touches a specific quantity of powder. The first is a statement about a building and the second is a statement about a formula. Both are real work, both cost real money, and neither is a measurement of the thing in front of you.

What lets all three pass for each other is a single phrase in common use that is true of every one of them and reads identically in each case — true of a facility audit, true of a four-year-old representative sample, true of a lot-numbered assay. It appears nowhere on this site as a badge, which is why a facility mark gets an article here rather than a logo.

What four questions separate a real test result from a decorative one?

Whose sample was it, whose laboratory ran it, which lot is named, and what was not run. Four questions, in that order. A document answering all four is a measurement of something a person can buy. A document answering none of them is a design asset with numbers on it.

1. Whose sample? A finished container off a real fill run, or a sack of raw material before it was made into anything. Both produce real numbers. Only one of them was ever a product.
2. Whose laboratory? An outside laboratory with no stake in the answer, or the laboratory belonging to the company that made the powder. A maker grading its own work is a weaker check, and a document almost never says so in those words.
3. Which lot, and is it named? A lot code is what separates a document about a specific quantity of material from a document about a formula in the abstract. No lot, no subject.
4. What was not run? The strongest documents print the rows they did not test. Where a test is simply missing, the absence reads as a blank rather than as a finding. It is a finding.

Those four are questions about the claim. The narrower business of reading the document itself — result column against limits column, method beside each row, a date, a signature — is six questions and a worked example, and it is the companion to this page rather than a repeat of it.

What do TARE’s own three certificates score?

One of the three names the lot inside a TARE tub, and the laboratory that ran it belongs to the company that makes the powder. The other two name raw-material lots bought by an ingredient supplier for a different company. No document TARE holds is an outside laboratory’s test of a finished TARE container.

Read the bottom row, then read the top row again. Base Work scores 4 of 4 — the highest mark on this site — on a document that never touched a TARE container. The sample photographed on it is a sealed poly bag of creatine monohydrate from Accobio USA, the customer named on it is Vox Nutrition, and the figure it reports is real: creatine monohydrate at 99.87% against a greater-than-99.5% specification, by HPLC. A real number, about a real sack of powder. The sack is not the tub.

One caution about the bottom row before it is quoted anywhere. The Lot Mark is our own count on our own documents, not an award. It is four questions — identity, potency, heavy metals, microbiology — and a tally of how many of them the certificate answers with a measured result. Nobody grants it, no body could withdraw it, and the denominator is printed every time precisely so that it reads as arithmetic rather than as a rosette.

That row is the argument of this page in one line. A score answers question four. It does not answer question one. Quoting the score without the provenance tells the true half, which is the most durable way to mislead somebody without saying anything false.

The reverse case is the hero product. Under Load’s document names lot 121125450 — a 350 g lot of the powder that goes in the tub — and reports protein at 9.8 g against a 9 g specification by ISO 14891:2002, with lead at 0.07 mcg against a 0.5 mcg Proposition 65 limit. It answers questions one and three better than anything else here. It answers question two worst: ViaCore Nutrition of El Reno, Oklahoma is the laboratory of the company that manufactures the powder. That is a weaker check than an outside one, and a reader should hear it here rather than work it out later. It is the hero product and, on documents alone, it is the one we would tell you to buy last. On the ladder at the top of this page, that document stands on the third rung and the other two stand on the second.

What does Not Tested mean on a certificate of analysis?

Night Work’s certificate reports microbiology and nothing else. Seven organisms, tested 9 March 2026, every row a pass. On the heavy-metal page the Result column reads Not Tested on all four rows — arsenic, cadmium, lead, mercury — and the Pass/Fail column reads a dash. There is no identity test on the document and no potency section on it at all. Four questions, one answer: Lot Mark 1 of 4.

There are numbers on those four rows and not one of them is a measurement. Beside arsenic sit 0.01, 0.001 and 15; beside lead, 0.01, 0.001 and 5. Those are the method’s limit of quantification, its limit of detection, and the acceptance limit — what the instrument can see and what the rule allows, never what was found. Read at speed, a row of numbers next to a metal name looks like a result. It is the most convincing empty cell in the category.

The same document prints Serving Size: N/A, which is the tell for question one: a finished capsule has a serving and a sack of powder does not. Night Work publishes at 1 of 4 with the three untested axes marked as untested, because the untested axes are the reason a count is worth printing at all. Every row is transcribed on Testing & Certification.

Does who funded a study change the result?

A certificate says what is in the powder. It is silent on whether the powder does anything, and that second question is settled in the peer-reviewed literature, where the equivalent of the provenance question is funding. It is not a small effect. In at least one meta-analysis, splitting the trials by who paid for them removed the result entirely.

Myung and Park, The American Journal of Medicine 2025;138(9):1264–1277, pooled 23 randomised trials, n = 1,474, of collagen supplementation on an outcome TARE makes no claims about. Pooled across all 23, collagen came out ahead. The reviewers had pre-specified a subgroup analysis by funding source, and ran it: in the trials not funded by supplement or pharmaceutical companies, there was no effect. The higher-quality studies showed no significant effect in any category either.

One paper, one outcome, and it should not be stretched into a law. What it is, precisely, is a measured demonstration that who paid is a variable — and no certificate has a column for it.

Four more, and two of them cost us something

The largest null in the tendon literature is the one with no industry funding recorded on it. Balshaw and colleagues, Medicine & Science in Sports & Exercise 2023;55(11):2083–2095, gave 15 g a day — our own protocol dose — to 39 men for 15 weeks of supervised lower-body resistance training, and measured the patellar tendon by MRI rather than ultrasound. No between-group difference on any tendinous measure. It is larger than most of the positive trials, longer than most, used the better imaging method, and it is not flagged as industry-funded. It appears wherever favourable tendon evidence appears on this site, including here.

The one positive result in the strongest group of muscle trials has the ingredient maker on its author list. Aussieker and colleagues, Medicine & Science in Sports & Exercise 2025;57(3):544–554, ran a tracer study in 28 men on a 25 g whey plus 5 g collagen blend. Muscle connective protein synthesis rose above placebo in the rested leg only, and not in the exercised leg (p = 0.11). A co-author is affiliated to GELITA AG of Eberbach, Germany — the collagen manufacturer. The paper reads positive in the title and is null in the leg that was trained.

The most-cited framing document in the field was funded by the industry it frames. Holwerda and van Loon, Nutrition Reviews 2022;80(6):1497–1514, sets out the hypothesis the whole category runs on: that connective tissue is short of glycine and proline, and that dietary collagen is rich in both. Its recorded funders are PB Leiner and the Tessenderlo Group — a gelatin and collagen-peptide manufacturer, and its parent. It is a narrative review, so no systematic search and no risk-of-bias assessment, and the same authors’ own later trials did not confirm the hypothesis it advances.

And the one that costs us most. Pajuelo and colleagues, Nutrients 2024;16(24):4367, compared four magnesium compounds head to head in 40 healthy adults. Oxide raised plasma magnesium at one hour and citrate at four hours; bisglycinate — the form in Night Work — produced no significant rise at any timepoint. Three things belong beside that sentence: the study was funded by Lubrizol Life Science, which sells a competing magnesium ingredient and whose own product won; a six-hour plasma curve is not a measure of magnesium status; and it was a single dose. All three are real caveats, and none of them is a reason to leave the result off this page.

Funding disclosure cuts both ways, and this section is the demonstration: two of those findings are unfavourable to something TARE sells, and one of them was produced by a company with its own interest in the answer. Naming the funder is not the same as dismissing the paper. It is question four — what was not run — asked of a study instead of a certificate.

Does a certificate of analysis show that a supplement works?

A certificate is a composition and contamination document. It answers what is in the material and how much of it, and it is silent on every question about outcome. A brand offering a certificate as evidence for a benefit is using the wrong document, confidently. Outcome evidence lives in trials, and trials have Not Tested rows of their own.

Two on our own ingredients, stated as the papers state them. The newest systematic review of collagen and tendon — Buchalski and colleagues, Journal of Functional Morphology and Kinesiology 2026;11(1):130, eight randomised trials, n = 257, of whom 246 were male — found tendon cross-sectional area and stiffness increased at 15 to 30 g a day with concurrent training, and graded that evidence strong. The same review graded the evidence strong against an effect on muscle strength. Both halves are one paper, and splitting them is the misuse.

The 2024 meta-analysis on the same question — Bischof and colleagues, Sports Medicine 2024;54(11):2865–2888, 19 studies, n = 768 — did find a statistically significant effect on maximal strength. It is SMD 0.19, and the authors graded the certainty of that evidence low. Printing the number beside the grading is the entire difference between a citation and a decoration, and it is the same difference as a result printed beside its specification.

The literature keeps its own empty columns too. An audit of 343 randomised creatine trials published between 1994 and 2023 — Bučević Popović and colleagues, Nutrients 2026;18(16):2697 — found that 80.2% used monohydrate, that 15.5% of reports did not state which form of creatine was used at all, and that bioavailability was mentioned in 11.1%, mostly in discussion sections. Five trials out of 343 compared one form against another. That is a Not Tested row in a different kind of document.

Run this on any brand, including this one

Ask for the document. Then ask the four questions in order: whose sample, whose laboratory, which lot, and what was not run. Two minutes, no expertise required, and it works on every certificate in the category, because all four questions are about the document rather than about the chemistry.

And if the document does not arrive, that is the answer. Whatever arrives instead is worth reading for which question it answers rather than for its tone. A specification sheet answers what a formula was meant to be. A facility letter answers where it was made. A page about a commitment to quality answers nothing. None of the three is a result on a named lot, and telling them apart takes no chemistry at all.

Run against TARE it returns this. One document on a finished TARE lot, analysed by the manufacturer’s own laboratory. Two documents on raw material bought by an ingredient supplier for another company. One product whose heavy-metal rows read Not Tested. And no outside certification of any kind — there is no badge on this site because there is no certificate behind one.

Every figure quoted on this page is transcribed row by row on Testing & Certification, and the document-reading half of the same discipline is at what a certificate of analysis proves, and what ours do not.

More in this cluster: specification against result, in one column, what a facility mark covers, the label format that makes the arithmetic impossible, and the sixty-second label procedure.