Three things
are called tested.
Almost every badge in this category is one of the first two. This page is the third one, transcribed: three certificates of analysis, the lot each was run on, the method beside every figure and the specification beside every result. It also carries the two facts that come out worst for us — one certificate was run by the company that makes the powder, and we hold no certification of any kind — and both of them are near the top rather than near the bottom.
The building. The product.
This lot.
Three different things are all called “tested,” and the industry lets you confuse them. They are not close to equivalent, and the difference is the entire reason this page is as long as it is.
- One — the building was inspected. Good Manufacturing Practice. NSF/ANSI 455-2. An auditor visited a factory, on a date, and assessed its quality systems against a standard. That is real work and it is worth doing. It is also a statement about a building. Nobody opened your tub, and nothing was measured in it.
- Two — a product was tested once. This is most certificates of analysis. A sample was pulled — often a “representative” one rather than the batch you bought, sometimes years ago, sometimes with the laboratory's name taken off the page before it was published. It tells you a formula was capable of passing. It does not tell you that the powder in front of you did.
- Three — this lot was tested. A certificate naming the batch, with a method printed beside every row and a specification printed beside every result. It is the only one of the three that touches a specific quantity of powder rather than a factory or a formula.
Almost every badge on almost every supplement is tier one or tier two. We publish tier three. Three certificates, three lots, every method and every specification, below.
The phrase that lets all three pass for each other is “third-party tested.” It is true of a facility audit, true of a four-year-old representative sample, and true of a lot-numbered assay run last December, and it reads identically in all three cases. That is why you will not find it used as a badge anywhere on this site. Where an outside laboratory ran one of our tests, this page names the laboratory, its city and its accreditation number. Where it did not, this page says that too, in the same size type.
Why the distinction is not pedantry
In March 2026 an independent testing programme started buying supplements anonymously at retail — off the shelf, as a customer, rather than accepting samples from brands — and sending them to an independent laboratory. More than half of the top sellers it tested failed label accuracy. In the creatine, NAD+, urolithin A and berberine rounds, 22 products came back at 0–3% of their claimed active ingredient.
Those were not obscure products. They were shelf-stocked, well-reviewed, and many of them carried badges. Every one of those badges was tier one or tier two, and not one of them was contradicted by the result, because none of them had ever claimed to measure the thing that failed.
That is the gap this page exists in. It is also the reason this page ends with a list of what our own documents cannot tell you — including the one thing that programme does and we do not, which is buy the jar anonymously off a shelf.
What is on this page today, and what is not
All three documents are here, reproduced whole. A handful of secondary rows are still being read.
We changed manufacturer in August 2026, so every document on this page is new and there are three of them rather than four. Two products did not survive that change and are no longer sold; the account of why is further down, in full, because it is the most useful thing on this page for judging the rest of it.
All three documents are now on this page, reproduced whole. Our manufacturer granted publication in writing on 17 August 2026, having first advised against it, and each certificate sits on the panel of the product it belongs to with a caption naming what it does and does not cover. What we still do not host is the original PDFs — what we hold are the images supplied through the manufacturer's dashboard. The account of that is at the foot of this page.
Some rows are named as unread rather than filled in. Every figure this page argues from is transcribed. Several secondary cells are not. Each one is marked where it sits, and all of them are listed in one place so you can total them up rather than hunt for them. Printing a number we cannot yet stand behind is the exact failure this page exists to prevent.
What a certificate of
analysis actually is.
Nobody is born knowing this. It takes about three minutes and it makes every supplement shelf in the country legible.
A certificate of analysis is a lab report on one production lot. It is not a certificate in the sense of an award — nothing is being awarded. A quantity of powder was made, a sample was pulled from it, tests were run, and the numbers came back. The certificate is the paper those numbers are written on. It names the product, it names the lot, and it lists one row per test.
Every row carries up to four things, and reading a certificate is reading those four in order.
- Analysis
- What was tested. Lead. Salmonella. Protein. One row each.
- Method
- How it was tested.
ISO 14891:2002,USP <233>andICP-MSare published methods — anyone can look up what they involve. A row with no method printed on it is a row you cannot check. - Specification
- The limit the formula has to meet.
0.5 mcgof lead per serving means the result has to come in under half a microgram. This is a target, not a finding. - Result
- What the test actually found.
0.07 mcg. This is the finding. - The Lot Mark
- Ours, not the document's. A count of how many of four questions this lot answers with a measured result — identity, metals, microbiology, potency — written
n / 4with the denominator always printed. It is not a grade, not a score out of a hundred and nobody awards it. All three marks are here.
The whole skill is holding the last two apart
A specification is what the formula is supposed to be. A result is what a test found. Read a certificate that prints only the specification column and you have learned what the manufacturer was aiming at. You have not learned whether they hit it. That is tier two of the ladder above, wearing tier three's clothes.
All three of ours report both columns. That is new. Until August 2026 we published four certificates and two of them reported no test result of any kind — composition and limits, nothing measured — and both panels said so at the top, in words, with a stamp on them you could read from across the room. Both of those products are now retired, and the reason is on this page rather than in a press release.
How to read a result once you have one
Read it against the specification beside it, never on its own. 0.07 mcg means nothing by itself. 0.07 mcg of lead against a limit of 0.5 mcg is a margin of seven to one, and now the number has told you something. That is why every table below keeps the two columns side by side instead of printing a tidy list of results.
Two conventions worth knowing before you meet them. “Record only” in a specification column means the method has no pass-or-fail limit set for this product: the value is recorded and reported, not judged against a ceiling. And a result written as less-than something, or as below the detectable limit — which is how three of the four metals read on our collagen certificate — is the smallest amount the method can report, not zero. It means the test could not find any, which is a different sentence from “there is none.” We are not going to round that up for you, and no margin is ever computed against one, because a detection floor is a property of the instrument rather than a quantity of anything.
And the fifth thing, which is ours
Four columns are the document's. The fifth is a count we do the reader's side of, and it is called the Lot Mark. A certificate can answer four questions: is this the substance the label names, how far under the metals ceiling did it come in, is it clean against a stated limit, and how much of the constituent is in it. Count how many of the four this lot answers with a measured result, and print the count. Base Work's lot CM20251225A is 4 / 4. Under Load's lot 121125450 is 3 / 4. Night Work's raw-material lot MGL260212 is 3 / 4 — the same count as Under Load's and not the same kind of three, which its panel says at length.
Each axis reads exactly one of three words, never a fourth and never a blend. MEASURED — a result column exists for this axis with a method printed beside it. STATED — a figure exists with no test behind it, a composition line or a limit. NOT RUN — no row of any kind exists, and the absence is the finding. Nobody awards this and there is no body that could take it away: it is our own arithmetic, on our own documents, and every operand is printed beside every operation so you can redo it in ten seconds.
Five conventions govern the margin column, and they are the line rather than house style. One, a ceiling like < 0.5 mcg reports as limit divided by result, rounded down, written n× under. Two, a floor like > 99.5% reports as a subtraction in the document's own units, written n above the floor — no ratio, because a ratio of two near-equal figures reads as a margin that is not there. Three, a censored result never becomes a ratio: below the detectable limit is the method's floor, not a quantity. Four, a cross-unit comparison never becomes a ratio — if results are printed in one unit and limits in another and the document shows no conversion, neither do we. Five, the mark reports the narrowest margin on the axis and never the widest.
Convention two costs us the number this site quotes most often, which is the reason to keep it. Base Work's creatine assayed at 99.87% against a specification of more than 99.5%. On the mark that reports as 0.37 above the floor — a third of a percentage point — and not as a ratio, because 99.87 divided by 99.5 is 1.004 and writing it that way would dress a third of a point up as a margin. Both readings are arithmetically true. Only one of them is honest about how much room there was. A system that always shows you its best row is not a system, and a reader who works out that it cherry-picks has learned the one thing about us that would cost the most. The three marks have a page each, with the chain, the arithmetic and what each one cannot tell you.
And what a table replaces
This category runs on short phrases printed on labels — that a product is tested, that a facility is registered, that a jar is clean. A phrase is a summary somebody wrote about a document, and the ladder at the top of this page is what a phrase is for: it lets tier one and tier three sound the same. We are not going to tell you anyone else's summary is wrong; we have not read anybody else's paperwork and we are not going to guess at it.
What we will do is name the difference and let you use it. A phrase has no lot number on it. A table does. Ours has the lot codes on it, and the specification is printed beside every result so the margin is visible rather than described. Take that to any other jar on your shelf and ask it for the same table. That is the only comparison worth making, and it takes about ten seconds.
The full policy on what we may and may not say about testing — including the phrases we decline to use and why — is on the policies page. The rule underneath all of it is the same one: we read the documents, and we do not write anything the documents do not say.
What we hold,
at a glance.
Three certificates of analysis, one per product. All three report measured results — a figure a test found, printed beside the specification it was measured against. Night Work's does so in the laboratory's own serving, on a raw-material lot, and its panel says what that does and does not cover. All three name the laboratory that ran them. Two of those laboratories are independent of the people who made the powder. One is not, and it is the hero product's.
| Product | Lot on the certificate | Work dated | What it reports | Laboratory that ran it |
|---|---|---|---|---|
| Under Load — collagen | 121125450 | 2025-12-22 | Measured results | ViaCore Nutrition, El Reno OK — the company that makes it |
| Base Work — creatine | CM20251225A | 2026-02-18 | Measured results | Ethos Analytics, Phoenix AZ — ISO/IEC 17025, Acc #117798 |
| Night Work — magnesium | MGL260212 raw material | 2026-02-24 | Measured results | Ethos Analytics, Phoenix AZ — for Vox Nutrition |
Under Load's date is the date of analysis, signed by Austin Snowden. Base Work's is the date the work was completed — the sample was received 2026-02-10 and the report signed by Noel Samsum, Laboratory Director, on 2026-02-18. Night Work's is the date the work was completed: the sample was received 19 February 2026, the three analyses are dated 22, 23 and 24 February, and the report was signed by Noel Samsum on 24 February 2026. Its lot number, MGL260212, is the lot of raw magnesium glycinate powder the ingredient supplier had tested — not a TARE fill lot, which is why the cell carries the qualifier rather than the code alone. This document replaced an earlier one on 15 September 2026 — a microbiology-only test on a different sublot of the same lot, which had itself replaced a November 2025 document for a different magnesium product on 2 September. None of the November figures survive anywhere on this site, and the September document's dates and sample number appear only in the dated history on the panel.
Two of the three were run by Ethos Analytics Laboratory, 3020 E Camelback Rd STE 397, Phoenix AZ 85016, which holds ISO/IEC 17025 accreditation number 117798. That is the international standard for the technical competence of testing laboratories: it is granted by an accreditation body against a defined scope, it carries a number, and the number is checkable. It is not a certification of our product and we are not presenting it as one — it is a fact about the laboratory, and it belongs to the laboratory rather than to us.
The third was run by ViaCore Nutrition of El Reno, Oklahoma — which is the company that manufactures the collagen. The certificates held for Creatine and Night Work were issued by an independent laboratory that had nothing to do with making them — but each covers raw material bought by the ingredient supplier, not a finished TARE container, and the magnesium document covers a raw-material powder lot: it carries a measured magnesium figure, a heavy-metal panel and a microbiological panel for that powder, and no identity test, so it still cannot vouch for what is in the capsule. The collagen was tested by the company that makes it, on the lot our own tub was filled from. Each of those is a different kind of gap, and you should hear about them from us rather than work them out later. It is set out at length on the Under Load panel, which is where it belongs, next to the figures it qualifies.
This is a change, and the direction it moved matters more than the position. The four certificates we published until August 2026 named no laboratory at all — one of them identified its lab only as Lab ID: 1 — and two of the four reported no test result of any kind. We said so on the page at the time, in the same size type as everything else. What happened to those four is further down.
Under Load — collagen peptides, Type I & III
Lot Mark 3 / 4 Lot 121125450 · analysed 2025-12-22 · ViaCore Nutrition
| Axis | State | Figure, with its opponent | Margin | Method |
|---|---|---|---|---|
| Identity | NOT RUN | No identity assay of any kind on this document — no infrared scan, no chromatographic fingerprint, nothing testing the powder against a collagen reference. The ingredient line reads collagen peptides, which is a statement rather than a test. The protein row below is a quantity: it says how much protein is in a serving, not that the protein is collagen. | — | — |
| Metals | MEASURED | Lead 0.07 mcg per serving against a 0.5 mcg Proposition 65 limit. Mercury, arsenic and cadmium all read below the detectable limit. | 7× under | ICP-MS |
| Microbiology | MEASURED | Total plate count, E. coli, salmonella, yeast and mould — five rows, every one a pass. | No ratio: the document reports verdicts against limits rather than counts | AOAC VIDAS |
| Potency | MEASURED | Protein 9.8 g against a 9 g specification, in the document’s 10 g serving. A quantitative result, not a pass. | 0.8 above the floor | ISO 14891:2002 |
- Chain
- Analysed by ViaCore Nutrition, 625 Jensen Road West, El Reno, Oklahoma — the laboratory of the company that manufactures this powder, not an independent one. Signed Austin Snowden, 22 December 2025. No accreditation is claimed on the document and we claim none for it. Three of the four metals read below the method’s detectable limit, which is a floor rather than a quantity.
- Arithmetic
- 7× and 0.8 are ours: 0.5 ÷ 0.07 rounded down, and 9.8 − 9. Both operands of each are printed above and in the table below. Everything else is transcription.
- Limit
- This is the lot on the document. It is not the lot in your jar.
The document is a quantitative analysis with a heavy-metals page against California Proposition 65 limits and a microbiological block. Row names below are as they appear on the certificate. The serving it reports against is 10 g — the tub’s printed serving. Our protocol is 15 g on training days, which is a site instruction rather than a label figure, and the arithmetic that runs both figures forward to it is directly under the table.
| Analysis | Method | Specification | Result |
|---|---|---|---|
| Protein | ISO 14891:2002 | 9 g per serving | 9.8 g per serving |
| Lead (Pb) | ICP-MS | 0.5 mcg per serving | 0.07 mcg per serving |
| Arsenic (As) | ICP-MS | Prop 65 limit | Below detectable limit |
| Cadmium (Cd) | ICP-MS | Prop 65 limit | Below detectable limit |
| Mercury (Hg) | ICP-MS | Prop 65 limit | Below detectable limit |
| Microbiological panel — five rows | AOAC / VIDAS | Limits not yet transcribed | All pass |
| Ingredients | — | — | Collagen peptides |
Two things on this table are honest gaps rather than transcription. The Proposition 65 limits for arsenic, cadmium and mercury are not yet transcribed — lead’s is, and it is the one with a numeric result beside it; the other three read below the method’s detectable limit, so no margin could be computed for them either way. And the microbiology block is five rows on the document — total plate count, E. coli, salmonella, yeast and mould — collapsed here into one line, because we have not yet transcribed the individual limits and method numbers and a row with an empty specification cell would be worse than a sentence saying so. They go in individually when they are read.
Running the two numbers forward, because the serving on the document is not our serving
The certificate reports against a 10 g serving. The protocol on this site is 15 g, three scoops, on training days. So the specification of 9 g of protein becomes 13.5 g at our serving, and the laboratory’s 9.8 g becomes 14.7 g. Both figures are ours: 9 × 1.5 and 9.8 × 1.5.
The measured figure is the larger of the two. We print the number we were promised and the number the laboratory found, in that order, and it is the second one that is bigger. That is the only direction that arithmetic is allowed to run on this site.
One thing it is not. Collagen is a poor protein — no tryptophan, very little leucine — and the 14.7 g should not be counted against a daily protein target. We do not sell this as a protein supplement and the reason is on the product section.
This certificate was run by the company that makes the powder.
ViaCore Nutrition of El Reno, Oklahoma manufactures this collagen, and ViaCore Nutrition analysed it. The document names its laboratory, which is more than any of the four certificates we published until August 2026 did. It does not name an accreditation, and we are not going to imply one for it.
Creatine and Night Work were tested by an independent lab that had nothing to do with making them. The collagen was tested by the company that makes it. That’s a weaker check, and you should hear it from us rather than work it out later.
How much weaker is a fair question and we are not going to pretend to answer it precisely. An in-house laboratory can be excellent; ISO/IEC 17025 accreditation is about demonstrated technical competence and impartiality, not about whether a machine works. What is true, and simple, is that nobody outside the building checked this one, and the incentive on the person running the assay is not the same as the incentive on a laboratory that has no stake in the answer. Every figure on this panel should be read with that sitting next to it.
What would change it: the same powder, the same lot, assayed by an outside laboratory. That costs a few hundred dollars per product and it is the cheapest upgrade available to this brand. When it happens, the result publishes here in full — including if it disagrees with the figures above, which is the only version of that promise worth making.
The hero product having the softest paperwork is an awkward sentence to write on the page whose job is paperwork. It is also the honest ranking, and it is in the product section too, in the same words: if you are buying one jar this month on documents alone, buy the creatine.
Base Work — creatine monohydrate
Lot Mark 4 / 4 Lot CM20251225A · completed 2026-02-18 · Ethos Analytics · ISO/IEC 17025 Acc #117798
| Axis | State | Figure, with its opponent | Margin | Method |
|---|---|---|---|---|
| Identity | MEASURED | The label claim profile assays creatine monohydrate at 99.87%. On a one-ingredient powder that row answers what the substance is as well as how much of it there is — see the note under the table, which argues the case against us. | No ratio: an identification is not a quantity with a ceiling | HPLC |
| Metals | MEASURED | Heavy metals by ICP-MS on a numeric USP <233> page — results printed as figures, every line a pass. | No margin printed: the individual figures are not yet transcribed | ICP-MS USP <233> |
| Microbiology | MEASURED | Microbiological panel, every line a pass. | No ratio: the document reports verdicts against limits | Petrifilm qPCR |
| Potency | MEASURED | 99.87% creatine monohydrate against a specification of more than 99.5%. Limit of quantitation 100 ppm. | 0.37 above the floor | HPLC |
- Chain
- Ethos Analytics Laboratory, 3020 E Camelback Rd STE 397, Phoenix AZ 85016. ISO/IEC 17025 accreditation #117798; licence 000026LRCND60176649. Signed Noel Samsum, Laboratory Director, 18 February 2026. The customer of record printed on the document is Vox Nutrition, 8224 Industry Circle, West Jordan UT 84088 — a private-label supplement manufacturer that bought the raw material; the certificate was issued to them, not to us. The powder itself was made by Accobio USA Inc., named on the bag in the sample photograph. Sample received 2026-02-10, work completed 2026-02-18.
- Arithmetic
- 0.37 is ours: 99.87 − 99.5. Both operands are printed above and in the table below. Everything else is transcription. Two of the four axes are scored from one row, which is the one place this mark could be argued down — the argument is directly beneath the table and we make it ourselves.
- Limit
- This is the lot on the document. It is not the lot in your jar.
This is the best-documented thing we sell, and it is the cheapest. It carries a named laboratory with a checkable accreditation number, a named signatory with a title and a date, a sample-received date and a completion date, a stated limit of quantitation, and a quantitative assay against a numeric specification. That is what a tier-three document looks like when everything goes right.
| Analysis | Method | Specification | Result |
|---|---|---|---|
| Creatine monohydrate — label claim profile | HPLC · LOQ 100 ppm | > 99.5% | 99.87% |
| Heavy metals | ICP-MS · USP <233> | USP <233> limits | Numeric — pass |
| Microbiological panel | Petrifilm / qPCR | Limits not yet transcribed | Pass |
The heavy-metals page on this document prints numeric results against USP <233> limits rather than verdicts alone, and the microbiological block is a panel of individual organisms. Neither set of figures is transcribed yet, which is why the result cells read as they do rather than carrying numbers. A verdict we have read is a verdict we will print; a figure we have not read is not going on this page in any form. Both blocks are on the list of what is still being transcribed.
The one place this mark can be argued down, made by us
Two of the four axes — identity and potency — are scored from a single row. An HPLC assay reporting 99.87% creatine monohydrate says what the substance is and how much of it is there, in one measurement, which on a one-ingredient powder is genuinely both answers. We count it as both.
Someone could reasonably say one row should score once, and call this 3 / 4. That reading is not wrong, it is a different convention, and we would rather print the objection than have a reader find it and wonder what else is arranged in our favour. Every other axis on every other mark is a separate row on the document.
Whose certificate this is
The customer of record on this document is Vox Nutrition of West Jordan, Utah — not TARE, and not the company that fulfils our orders. Certificates in this trade are issued to whoever commissioned the assay, which is normally the manufacturer of the raw material rather than the brand on the tub. Ours is passed down the chain to us.
We are printing that because it is on the paper and because it is the sort of detail a brand rounds off. What it means about the supply chain behind this lot is a question we have put to our manufacturer and have not yet had a full answer to. It is on the list of what these documents do not cover.
Night Work — magnesium glycinate
Lot Mark 3 / 4 Raw-material lot MGL260212 · completed 2026-02-24 · Ethos Analytics · ISO/IEC 17025 Acc #117798
| Axis | State | Figure, with its opponent | Margin | Method |
|---|---|---|---|---|
| Identity | NOT RUN | No identity section of any kind on this document, and no row anywhere on it that would answer the question. Nothing here confirms the powder is magnesium glycinate rather than oxide, citrate or a buffered blend — and glycinate is what the label names. The magnesium row below is a quantity: it counts magnesium by ICP-MS and would read the same for any of those salts. | — | — |
| Metals | MEASURED | Heavy metals by ICP-MS on a numeric USP <233> page: arsenic 0.034 ppm, lead 0.044 ppm, cadmium and mercury below the limit of quantitation (0.010 ppm; limit of detection 0.005 ppm), every line a pass. The limit column is printed in a different unit — 15 / 5 / 5 / 15 mcg per day — and the document shows no conversion. | No ratio: results in ppm, limits in mcg per day, and the document shows no conversion between them | ICP-MS USP <233> |
| Microbiology | MEASURED | Seven organisms, all pass. Total aerobic plate, coliform, yeast and mould each <20 CFU/g against limits of 100,000 / 100 / 1,000 / 1,000 CFU/g. E. coli, salmonella and staphylococcus aureus each negative per 10 g against a limit of negative per 10 g. | No margin: every result is a less-than bound rather than a quantity | Petrifilm / qPCRUSP <2021>, <2022> |
| Potency | MEASURED | Magnesium 102.54 mg against a 91.6 mg specification, in the document’s 926 mg serving — the laboratory’s serving, not the label’s three capsules. A quantitative result, not a pass. Limit of quantitation 0.1 ppm. The 275 mg on the label is a label figure; this row does not measure it and is not run forward to it as measured. | 10.94 above the floor | ICP-MS |
- Chain
- Ethos Analytics Laboratory, 3020 E Camelback Rd STE 397, Phoenix AZ 85016 — the same independent laboratory that ran the creatine. Signed Noel Samsum, Laboratory Director, 24 February 2026; sample received 2026-02-19, work completed 2026-02-24. The licence and accreditation numbers were read off this document directly rather than copied across from the creatine certificate, and they are the same pair: Lic # 000026LRCND60176649, ISO/IEC 17025 Acc # 117798. The customer of record is Vox Nutrition — not TARE and not Rocktomic.
- Count
- The numerator counts axes carrying a measured result with a method beside them. Three do. The magnesium row is an elemental assay: it scores potency and cannot score identity, because ICP-MS counts magnesium and would return the same figure for oxide, citrate or glycinate. That is the difference between this document’s three and Base Work’s four, where one HPLC row is counted as both and argued against on its own panel. That is 3 / 4, and it is 3 / 4 with every character on the document read.
- Arithmetic
- 10.94 is ours: 102.54 − 91.6, both operands printed above and in the table below. Nothing else on this mark is ours. The heavy-metal results are in ppm and their limits in mcg per day; the document shows no conversion, so neither do we. And nothing here is run forward to the label’s three capsules as a measured figure — the arithmetic that would do that is under the table, named as arithmetic.
- Limit
- This is the lot on the document, and the lot on the document is a sealed bag of raw powder — sublot MGL260212-01-0001, 505 kilograms, hand-marked “Lab.” MGL260212 is the magnesium glycinate lot Vox Nutrition had tested, before anything was encapsulated. It is not a TARE fill lot and it is not the lot in your bottle.
Three analyses over three days: the sample was received 19 February 2026, heavy metals ran on the 22nd, microbiology on the 23rd and the magnesium assay on the 24th, and the report was signed that day. Row names below are as they appear on the certificate. The serving it reports against is 926 mg — the laboratory’s serving, which the document does not relate to a capsule. Our label is three capsules supplying 275 mg of elemental magnesium from 2,500 mg of magnesium glycinate, 30 servings to the bottle, and every one of those is a label figure confirmed to us in writing rather than a measured one; the arithmetic that connects the two servings is directly under the table, named as arithmetic. Every magnesium figure this site carried before August 2026 belonged to a different bottle and none of them are on this page any more.
| Analysis | Method | Specification | Result |
|---|---|---|---|
| Magnesium — label claim profile | ICP-MS · LOQ 0.1 ppm | 91.6 mg per 926 mg serving | 102.54 mg per 926 mg serving |
| Arsenic (As) | ICP-MS · USP <233> | 15 mcg per day | 0.034 ppm — pass |
| Cadmium (Cd) | ICP-MS · USP <233> | 5 mcg per day | < LOQ (0.010 ppm) — pass |
| Lead (Pb) | ICP-MS · USP <233> | 5 mcg per day | 0.044 ppm — pass |
| Mercury (Hg) | ICP-MS · USP <233> | 15 mcg per day | < LOQ (0.010 ppm) — pass |
| Total aerobic plate | Petrifilm / qPCR · USP <2021>, <2022> | < 100,000 CFU/g | < 20 CFU/g — pass |
| Coliform | Petrifilm / qPCR | < 100 CFU/g | < 20 CFU/g — pass |
| Yeast | Petrifilm / qPCR | < 1,000 CFU/g | < 20 CFU/g — pass |
| Mould | Petrifilm / qPCR | < 1,000 CFU/g | < 20 CFU/g — pass |
| Escherichia coli | Petrifilm / qPCR | Negative per 10 g | Negative per 10 g — pass |
| Salmonella | Petrifilm / qPCR | Negative per 10 g | Negative per 10 g — pass |
| Staphylococcus aureus | Petrifilm / qPCR | Negative per 10 g | Negative per 10 g — pass |
Row names are the certificate’s; the mould row is printed “Mold” on the document. The heavy-metal specification column is the document’s Limit column, printed in mcg per day while the results are in ppm. The document shows no conversion between the two and we do not supply one, so no margin is computed on that axis. A result written as less than the limit of quantitation is the method’s floor, not zero; the limit of detection for all four metals is printed as 0.005 ppm. Every cell here is a transcription, and nothing on this table is ours.
Running the one number forward, because the serving on the document is not our serving
The certificate reports against a 926 mg serving, and the document does not say what that is a serving of. Our label is three capsules. The one piece of arithmetic that connects them is that the document’s specification, 91.6 mg, is one third of the 275 mg on our label — 275 ÷ 3 is 91.67 — and that arithmetic is ours; the document does not say its serving is a third of ours.
Run the measured figure forward on that assumption and 102.54 mg becomes 307.6 mg by arithmetic, against 274.8 mg for the specification. Both figures are ours: 102.54 × 3 and 91.6 × 3. Nothing was measured at three capsules, the 275 mg on the label stays a label figure, and the measured figure is the larger of the two — the only direction that arithmetic is allowed to run on this site.
Identity is still not run, and the lot is still not ours. Here is the certificate.
This panel has described three different documents in six weeks, and the dates are the account. Until 2 September 2026 our manufacturer carried a November 2025 report for a different magnesium product, with a potency figure that would not reconcile with the label; we would not print it, and we were right not to. On 2 September it was replaced by a microbiology-only test on the right ingredient, and this mark fell to one of four. On 15 September it was replaced again, by the document below — a different sublot of the same raw-material lot, with a magnesium assay, a heavy-metal panel and the same seven microbiological rows — and we read it on 15 October 2026, a month later. The lag is ours, and it is printed here rather than smoothed over.
What it still does not do. It runs no identity test, so nothing on it confirms the powder is the glycinate the label names. And it is a test of a sealed bag of raw powder for Vox Nutrition, in the laboratory’s own 926 mg serving: not a test of a TARE capsule, not a TARE fill lot, and its magnesium figure is not a measurement of what three capsules contain.
The permitted sentence, in full, is this one. The magnesium glycinate raw material was tested by Ethos Analytics, for Vox Nutrition, lot MGL260212, completed 24 February 2026, for magnesium, heavy metals and microbiology: 102.54 mg of magnesium per 926 mg laboratory serving against a 91.6 mg specification, heavy metals pass, microbiology pass. That certificate carries no identity test, its serving is the laboratory’s and not the label’s three capsules, and it does not test a TARE container. Anything shorter than that is an overstatement, including “tested” on its own.
What is still owed is a certificate for a finished TARE lot carrying identity, potency at the label’s three-capsule serving and heavy metals. Until one exists this mark stays at three of four, identity stays NOT RUN, and no magnesium figure on this site is called measured at the label’s serving. It is on the open ledger with a date on it. That has happened on this site before and the account of it is further down this page.
The one magnesium number that matters for stacking
Night Work is the only supplemental magnesium in the range. At 275 mg of elemental magnesium a day it sits 75 mg under the 350 mg tolerable upper intake level for supplemental magnesium — that arithmetic is ours, it is 350 − 275, and both operands are printed. The 275 is a label figure and it is named as one every time it appears.
That headroom is not spare capacity we are inviting you to use. It is the room a multivitamin, a second product or a fortified drink would take up, and it is your job to count those rather than ours to assume you have none. The upper limit applies to supplements, not to magnesium from food.
Four documents became
three. Here is what
happened to the rest.
We audited our supplier and two products did not survive it. A third survived in half. This is the long version, kept on the page rather than deleted with the products, because a testing page that quietly loses the entries it did badly on is not a testing page.
Until August 2026 this page carried four certificates. Two of them — the creatine-and-electrolyte powder and the shilajit capsule — reported no test result of any kind: composition and limits, nothing measured. Both scored 0 / 4 on the Lot Mark and both panels said so at the top in the loudest type on the page. Every product that scored nought is gone. That is what the count was for, and it is the only test of a scoring system that matters — whether it is allowed to cost you something.
First Food — colostrum. The number was on the front panel, the certificate did not measure it, and on 31 July 2026 we took it off.
The certificate for lot R506064 contained no immunoglobulin assay of any kind. Not a low result and not a failed one — the test was not run and the row did not exist.
What the certificate did establish was identity. An infrared scan matched the powder against a colostrum reference, and it cleared its specification. That confirmed the powder was colostrum. It said nothing about how much immunoglobulin was in it — an identity match is a spectral score rather than a purity figure, and standing next to a milligram claim it would have read as one.
The powder was bought to an immunoglobulin specification, and that specification was the supplier’s own. It was what we ordered and what we paid for, and it is why the figure was on the label to begin with. It was not a result, and the certificate did not corroborate it. We asked for an assay that would. We did not get one.
So on 31 July 2026 we took it off. The immunoglobulin percentage came off the front panel and off this site, along with the milligram figure that had been multiplied out of it, and nothing was put in their place — not the identity match either. None of those figures is reprinted here. They described a product we no longer sell, and carrying a retired number forward as history is exactly how it ends up quoted back as a current one.
The product is now retired. When we changed manufacturer we asked the same question again and got the same answer: no lot from either supplier carries a finished-product immunoglobulin assay. A product whose headline number cannot be measured is a product this brand cannot sell. First Food scored 3 / 4 and is gone anyway, which is the more interesting case of the three — the missing quarter was potency, and potency was the whole argument.
Mineral Base — shilajit. Retired: there is no certificate at all.
Its old certificate reported composition and specification limits and no measured result on any axis. Lot Mark 0 / 4. We said at the time that a measured lead figure was the single most useful number our manufacturer could send us, since heavy-metal content in a mineral pitch is a property of where it was dug.
No certificate of analysis exists for the equivalent product at our new manufacturer. Not a weaker one — none. So the product is gone rather than carried forward.
Training Water — retired as constituted. The creatine half survives.
Same shape of document as the shilajit: composition and limits, no results, Lot Mark 0 / 4. The formula was creatine plus an electrolyte load — sodium, potassium and magnesium in one scoop.
No certified electrolyte product is available to us. The creatine half survives on its own as Base Work, which went from the worst-documented thing we sold to the best-documented, and the electrolyte load is simply gone until a custom formula is funded. We are not going to replace a measured product with an unmeasured one to keep a slot filled.
Two things that went with it also leave this site: the sodium load that scoop carried, and the stacked-magnesium arithmetic that ran the old four-jar range against the 350 mg upper limit. Neither describes anything we sell now, so neither figure appears anywhere on it.
And the four old lot codes
R506064, 5120130, 6010323 and MGL250923 are not on this page as live documents any more, and no figure from any of them appears anywhere on this site, including in the account above. The old lot pages are kept rather than deleted — a link that used to reach a certificate now reaches an explanation of why the product is gone, which is the only version of a retirement a page like this can honestly ship.
We hold none.
Here is what we
hold instead.
This is the section where a supplement brand shows you its badges. We do not have any, and this is the section that says so rather than the section that changes the subject.
TARE holds no third-party certification of any kind. Not NSF Certified for Sport. Not Informed Sport. Not USP Verified. No facility mark of our own, and no product mark of any description.
That sentence is the whole of it, and it is printed at the top of the section rather than in a footnote, because a page that spends this long on the documents we have owes you a plain sentence about the ones we do not.
What came off this page, and when
Until August 2026 this section carried a table of facility attestation letters — an assessor, a standard, a scope and an expiry date, in two columns, because the products were made under two different ones. We were careful about them at the time: we wrote that neither body had tested a TARE product, that what was assessed was a building's quality system on a date, and that we held our manufacturer's letter rather than the certificate itself.
They were still somebody else's certificates, and on the day we changed manufacturer they stopped describing anything we sell. So they came off — the assessors, the standards, the dates, and the badge on the label that carried them. We did not carry them across, we did not soften them into "made in a certified facility," and we have not replaced them with anything, because a replacement has to be a document in hand and we do not have one.
Our current manufacturer may well hold facility certifications. Most contract manufacturers of this size do. We have not been sent the letter, so there is no badge and there is no sentence about it here. When the letter arrives, it goes in this section with its assessor, its standard, its scope and its expiry printed beside it, and you can hold us to the expiry the way you could hold us to the last two.
The certification we would most like to hold, described fairly
NSF Certified for Sport is the real thing and we are not going to run it down because we do not have it. It screens for banned substances per production lot rather than once, it verifies that what is on the label is in the tub, and it requires facility audits. If you compete under a testing body, that is the programme to look for, and you should look for it by name rather than for the word “certified.”
What it does not tell you — what no certification of any kind tells you — is whether the doses inside are meaningful. A product can be certified, screened, audited and honestly labelled, and still contain a quantity of an ingredient chosen because it looks good on a panel. That is a separate question, it is answered by publishing the dose and the evidence for it rather than by a mark, and it is answered on The Monograph.
Why we do not hold one: it costs money we do not have. These programmes are priced per product, per year, and the bill for three products is a meaningful fraction of everything this company has ever spent. That is the entire reason, and we are not going to dress a budget up as a philosophy. If it becomes affordable we will buy it, and it will not change a single figure on this page — which is rather the point.
What we have instead
Three certificates of analysis, transcribed row by row on this page, with the lot each was run on, the method beside every figure and the specification beside every result. Two of the three run by an independent laboratory named on this page with its city and its accreditation number — against raw material rather than a finished container of ours. One run by the company that makes the powder, on a lot we filled, said out loud at the top of its own panel. A count of how much of each lot was actually measured, with the arithmetic printed underneath so you can redo it. And a list, below, of everything these documents cannot tell you.
A certification is a summary somebody else wrote about documents you cannot see. We have skipped the summary and published the documents. That is not a superior position and we are not claiming it is — a lot-screening programme catches things a certificate of analysis does not, and we would take both. It is the position we can actually defend, every figure of it, today.
If you ever see a badge on a TARE surface, it is a mistake
Nothing on this site, this label, this packaging or any advertisement says TARE is certified, approved, registered or endorsed by anybody. If you find one that does, it is an error and we want to be told about it — support@taketare.com, and a person answers within one business day. That is a real request rather than a rhetorical one. The single fastest way to lose the only thing this brand has is to let a badge onto a surface nobody proofread.
What these documents
do not cover.
Publishing a document is undone by overstating it. Eight things these three certificates do not do, written out rather than left for you to notice.
- They are not your jar. All three name a lot —
121125450,CM20251225AandMGL260212— and naming a lot is one of the things that separates a real document from a marketing PDF. Only the first is a TARE fill lot; the other two are raw-material lots on certificates issued to another company. None of them is your lot. TARE is fulfilled by a contract manufacturer and never takes possession of inventory, so we cannot connect the lot on a published document to the jar in your hands. We do not offer per-lot traceability and we do not claim it. The day that sentence stops being true it is deleted, and not one day sooner. - One of the three was run by the company that makes the powder. Under Load's certificate is ViaCore Nutrition's, on collagen that ViaCore Nutrition manufactures. It is the hero product and it has the softest check of the three. The full account is on its panel, at the same length as everything else.
- We hold no certification. Not NSF Certified for Sport, not Informed Sport, not USP Verified, no facility mark of our own. There is no badge on this site because there is no certificate behind one. The section above says so at length, including what it would cost and why we do not have it.
- Nobody bought these jars anonymously off a shelf. The programme described at the top of this page buys at retail, as a customer, precisely so that the brand cannot choose the sample. Our documents were supplied to us by the people who make the powder, and two of the three were issued to a third company entirely rather than to us. That is normal in this trade and it is still a weaker chain than an anonymous purchase. It is the single most useful thing we could pay for next.
- Some rows are read and some are not. Every figure this page argues from has been transcribed. Several secondary rows have not: the numeric heavy-metal figures on two of the three, and the individual microbiological limits. They are named where they sit and listed in one place — the ledger below — rather than filled in from a label or left to look complete.
- Nothing here says what else the facility handles. A certificate of analysis is about a quantity of powder, not about the lines it ran on or what ran on them before. Until we are told which allergens the facility handles and whether lines are shared, “free from” and “-free” language stays off all three products rather than being assumed. None of the three contains a declared major allergen by formula; that is a statement about the formula, and it is not the same statement.
- These lots are not new, and certificates of this kind carry no expiry. The work described here was done between December 2025 and February 2026. Three things are open with our manufacturer: whether a certificate is generated for every production run, whether we can be told which lot is currently shipping, and whether we are notified when any of these documents is updated or superseded. Until those are answered, treat every date on this page as the date printed on it and nothing more.
- A certificate verifies contents. It does not verify effects. Nothing on this page is a claim about what any of this does for anyone. What we are permitted to say, what the research shows and where it is weakest — including the trial sizes, and the outcomes no trial has shown — is on The Monograph and in The Recovery Journal, and it carries the disclaimer at the foot of this page.
Three documents,
published whole.
Three certificates of analysis. Every measured figure in the tables above comes out of them, and as of 5 September 2026 all three are on this page as images so you can check the transcription against the paper. Where a figure on this page comes from a label instead — Night Work's 275 mg of elemental magnesium is the only one — it is named as a label figure in the same sentence, next to what the certificate does and does not corroborate.
Until today this block said we had not confirmed in writing that we may reproduce these documents, so we had not posted them. That was true when it was written. It stopped being true on 17 August 2026, when our manufacturer granted publication in writing — having first advised against it — and the sentence outlived the fact by three weeks. That is the kind of drift this page exists to catch, so it is corrected here rather than deleted quietly.
What is posted: each certificate reproduced whole, unmarked, uncropped and unredacted, on the panel of the product it belongs to — Under Load, Base Work, Night Work — each with a caption naming the laboratory, the customer of record, the sample, the lot, the dates and, in plain words, what the document does not cover. Every one of them carries the same clause in its footer: “Reports are prohibited from being reproduced with alterations of any kind.” That forbids alteration, not reproduction, and nothing has been altered.
What is still not posted: the original PDFs. What we hold are the images our manufacturer supplies through its dashboard; the source files are the laboratory's and the ingredient supplier's. If you want one, email support@taketare.com and we will ask for it and tell you what we are told. A person answers, within one business day.
Every cell on this page that reads as a blank rather than a figure is listed here. Each one is a cell we could have filled in from a label, a supplier's index table or a plausible guess. None of them is going to be.
Night Work — closed 5 September 2026, re-anchored 15 October 2026. Three rows left this ledger on 5 September and none of them by being transcribed: the manufacturer replaced the ROC824 document on 2 September, and that replacement carried no potency section, so the measured elemental magnesium figure did not exist rather than being unread. On 15 September the manufacturer replaced the document again, and the one it carries now — a different sublot of the same raw-material lot, MGL260212, completed 24 February 2026 — carries a magnesium assay, a heavy-metal panel with figures and the same seven microbiological rows. All of it is transcribed on the panel: 102.54 mg of magnesium per 926 mg serving against a 91.6 mg specification, four heavy-metal rows with their results, their limits of quantitation and detection, and the limit column as printed. We read it on 15 October, a month after it went up, and that lag is ours. The panel carries the account, and the document itself.
What is still a testing gap rather than a transcription gap: Night Work has no identity result, because none was run, and the document still describes a raw-material powder lot tested for the ingredient supplier in the laboratory’s own serving rather than a TARE capsule. That is on the mark as one NOT RUN axis and is owed by our manufacturer, not by us. It is on the list of what these documents do not cover.
Base Work — the numeric heavy-metal figures. That document prints numbers rather than a verdict alone, and this page currently reads “pass” because a verdict is what we have read. Night Work's heavy-metal rows are not on this list: all four are transcribed with their figures, on its panel.
Under Load — the Proposition 65 limits for arsenic, cadmium and mercury. Lead’s is transcribed and it is the one with a numeric result beside it; the other three results read below the method’s detectable limit.
Under Load and Base Work — the individual microbiological rows and their limits. Collapsed into one line per certificate today, with the organisms named where the document names them. Night Work's seven rows came off this list on 5 September 2026 and are now printed individually, each against its own limit.
Each one goes into the table above the day it is read, in the same edit that removes it from this list. If a figure arrives and disagrees with something already on this page, the correction is printed with a date on it rather than swapped in quietly.
One more thing, since this page is about checking
Everything above is transcribed by hand out of three PDFs. If you find a figure here that does not match the document when the documents go up, write to us and we will correct it on this page and say what changed. We edit in public rather than quietly. That is the only version of this that means anything, and there is a worked example of us doing it further up the page.
Nine questions,
answered in fifty words.
Each one is the compressed version of something set out at length above. Nothing new is claimed here — if an answer matters to you, the panel it came from has the whole of it.
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. A product was tested once — most certificates of analysis, often on a representative sample. This lot was tested — a certificate naming the batch. Almost every badge in this category is one of the first two.
The phrase that lets all three pass for each other is “third-party tested,” which is true of a facility audit, true of a four-year-old representative sample and true of a lot-numbered assay, and reads identically in every case. You will not find it used as a badge anywhere on this site. TARE publishes the third kind: three certificates, three lots, the method beside every figure and the specification beside every result. The ladder is at the top of this page.
Who ran TARE's tests, and is the laboratory named?
Yes, on all three. Base Work's creatine and Night Work's magnesium were run by Ethos Analytics of Phoenix, Arizona, which holds ISO/IEC 17025 accreditation number 117798. Under Load's collagen was run by ViaCore Nutrition of El Reno, Oklahoma — the company that manufactures it.
One of the three certificates names the lot our own tub was filled from — the collagen — and the laboratory that ran it belongs to the company that makes the powder. The other two were run by an independent laboratory, but against raw material bought by the ingredient supplier rather than against a finished TARE tub. Neither of those is a weaker version of the same thing; they answer a different question. You should hear that from us rather than work it out later. It is the hero product, and it is also the one we would tell you to buy last on documents alone. The full account is on its panel.
Is TARE certified — NSF, Informed Sport, USP?
No. TARE holds no third-party certification of any kind — not NSF Certified for Sport, not Informed Sport, not USP Verified, and no facility mark of its own. There is no badge on this site because there is no certificate behind one.
NSF Certified for Sport is the real thing and we are not going to run it down because we do not have it: it screens for banned substances per production lot, verifies label contents and requires facility audits. If you compete under a testing body, look for that programme by name rather than for the word certified. We do not hold it because it costs money we do not have, and we are not going to dress a budget up as a philosophy. What we publish instead is the documents themselves.
Which TARE certificates report measured results?
All three. Under Load's and Base Work's each print a result beside the specification it was measured against, and since 15 September 2026 Night Work's does too: magnesium at 102.54 mg against a 91.6 mg specification, in the laboratory's own 926 mg serving, on a raw-material lot — with a heavy-metal panel and microbiology beside it, and no identity test. That is a change: until August 2026 we published four certificates and two of them reported no test result of any kind, and both of those products are now retired.
A specification is what a formula is supposed to be. A result is what a test found. A document printing only the specification column tells you what the manufacturer was aiming at, not whether they hit it. Both of the documents that did that are gone — every product that scored nought on the Lot Mark was retired, which is what the count was for.
What are the strongest measured numbers on this page?
Base Work's creatine assayed at 99.87% against a specification of more than 99.5%, by HPLC, at Ethos Analytics. Under Load's protein measured 9.8 g against a 9 g specification, and its lead 0.07 mcg per 10 g serving. The 0.5 mcg Proposition 65 figure is a limit per day rather than per serving, so the two have to be put on the same basis: at our 15 g protocol — 1.5 servings — that is 0.105 mcg a day against 0.5 mcg a day, 4.8 times under. Night Work's magnesium measured 102.54 mg against a 91.6 mg specification in the laboratory's 926 mg serving — on a raw-material lot, not a TARE capsule.
The creatine figure is reported on the Lot Mark as 0.37 above the floor rather than as a ratio, because a ratio of two near-equal figures dresses a third of a percentage point up as a margin it is not. The protein figure runs forward, and the forward step is ours: 9.8 g measured per 10 g serving against a 9 g specification, and at 1.5 servings, our 15 g protocol, that is 14.7 g against 13.5 g by arithmetic. Nothing was measured at 15 g. The measured figure is the larger of the two. That is the only direction that arithmetic is allowed to run here.
What lot is on each certificate — and is it the lot in my jar?
Under Load's is 121125450 and Base Work's is CM20251225A. Night Work's is MGL260212, and it is a lot of raw magnesium glycinate powder the ingredient supplier had tested rather than a lot of finished capsules. None of them is your lot: TARE never takes possession of inventory, so nothing connects a published lot to the jar you receive.
Naming a lot is what separates a real document from a marketing PDF, and it is not the same thing as a promise about your jar. We do not offer per-lot traceability and we do not claim it. The day that sentence stops being true it is deleted, and not one day sooner. Whether we can be told which lot is currently shipping is an open question with our manufacturer.
How do I read a certificate of analysis?
Read four columns in order: what was tested, how it was tested, the specification, and the result. The specification is the limit the formula has to meet — a target. The result is what the test found. Hold those two apart and every supplement shelf in the country becomes legible.
Read a result against the specification beside it, never on its own. 0.07 mcg means nothing alone; 0.07 mcg of lead against a limit of 0.5 mcg is a margin of seven to one. Two conventions to know: record only means no pass-or-fail limit is set for that row, and a result written as less-than something, or as below the detectable limit, is the smallest amount the method can report rather than zero. The long version takes three minutes.
Why did TARE stop selling colostrum and shilajit?
Because their paperwork did not survive an audit of our supplier. The shilajit has no certificate of analysis at all at our new manufacturer, and its old one reported no measured result on any axis. The colostrum's certificate never assayed the immunoglobulin figure its label leaned on, and no lot from either supplier carries one.
The creatine-and-electrolyte powder went the same way in half: no certified electrolyte product is available to us, so the creatine survives alone as Base Work and the electrolyte load is gone until a custom formula is funded. The full record is on this page, including the day in July 2026 when a figure came off a label because the certificate did not measure it.
Where are the certificate PDFs?
The documents themselves are on this page. Each certificate is reproduced whole — unmarked, uncropped, unredacted — on the panel of the product it belongs to, under written permission our manufacturer gave on 17 August 2026. What is still not published is the original PDF: what we hold are the images the manufacturer supplies through its dashboard, and the source files belong to the laboratory and the ingredient supplier.
Until 5 September 2026 this answer said we had not confirmed in writing that we may reproduce them. The permission arrived three weeks before the sentence came down, and a page about documents going stale should not go stale about its own, so the correction is printed rather than swapped in quietly. If you want an original file, email support@taketare.com and we will ask for it and tell you what we are told, within one business day.