ETA-DOC-0001·Grade mark system·Rev 1·Methodology v0.1Provisional — reviewer not yet appointed
Specimen document. Simulated third-party surfaces are marked Illustrative. Grade records C-01, C-03, C-08, C-11, C-16, C-18 and T-01 are real and are issued under methodology v0.1; reviewer not yet appointed, so all grades are provisional.
Grade mark system · specification & specimen ETA-DOC-0001 · rev 1

The Etalyn grade mark

A notation for two independent measurements — how strong the evidence is, and how good the data underneath it is — designed to be readable at 16 pixels in someone else's footnote, to survive monochrome print in a journal, and to be worthless to anyone who copies it without a verification record behind it.

The mark is the primary asset. The website is where the mark lives and where it is verified; everything else in this document is derived downward from the geometry of the mark itself.

Document
ETA-DOC-0001
Title
Grade mark system
Methodology
v0.1
Issued
2026-08-07
Supersedes
None — first issue
Reviewer
Not yet appointed
Adjudicator
Not yet appointed
Status
Current
Identity
Approved 2026-08-07 — ETA-DOC-0002 rev 1
§01

Why the mark is built this way

Written for a sceptical reader who thinks rating marks are decoration.

Moody's and S&P are not taken seriously because of their websites. They are taken seriously because three things hold together: a compact letter notation that fits inside a sentence someone else is writing, a published methodology that can be argued with, and a named human who signed the opinion. The notation travels; the institution follows it. Etalyn is built on the same structure, for evidence about human performance rather than credit.

That fixes the design problem precisely. The mark's job is not to look impressive on etalyn.com. Its job is to survive being pasted into a footnote by a journal sub-editor, printed at 8 point on a supplement carton, screenshotted into a slide deck by a sales team, and reproduced in one colour by a print shop that has never heard of us — and, in every one of those places, to remain checkable.

Two measurements, never merged

Most rating systems collapse into a single score, and the single score is where the dishonesty enters. A claim can rest on excellent trials and terrible measurement, or on a superb wearable dataset that has never been tested against an outcome. Etalyn therefore reports an evidence tier (A–E) and a provenance grade (P0–P4) as two fields that are always shown together and never averaged. and are different animals, and the notation refuses to pretend otherwise.

Six decisions, and the reason for each

  • The mark is a ruled cartouche, not a rounded pill. Pills read as marketing badges. A ruled box with hairline field dividers reads as an instrument label, and it is what survives when a printer drops the colour.
  • Colour is the third channel, never the first. Every value is carried by a glyph (A–E, P0–P4) and by a position on a rail. Colour is redundant confirmation for sighted readers on screen, so the mark is unchanged for colour-blind readers and in monochrome.
  • Every letterform inside a mark is drawn as geometry, not set in a font. A badge that depends on a webfont breaks in email clients, PDF pipelines and print RIPs. Our glyphs are stroked stencils with counters wide enough to stay open at 16 pixels.
  • There is a version slot. A grade is an opinion issued under a specific methodology on a specific date. A mark without a methodology version is an unfalsifiable claim, which is the thing we exist to be against.
  • The mark carries a verification URL, and the graphic is deliberately not the proof. Anyone can redraw a rectangle. Nobody can produce a record in the public register. The mark is a pointer; the register is the authority.
  • A stale grade must look stale. Grades change. The superseded variant desaturates, dashes its frame and takes a cancellation bar across its foot, so an out-of-date mark is legible as out-of-date before anyone reads the small print.

What would change this answer

  • Field testing showing the 16 px composite is misread by more than 5% of readers in a forced-choice tier task.
  • Evidence that third parties strip the provenance field when embedding, which would force the two fields into a single inseparable glyph.
  • A registry standard (ISO, CEN) for evidence notation that Etalyn should defer to rather than duplicate.
§02

Anatomy of the composite grade stamp

Three fields, one silhouette, two geometric profiles.

COMPOSITE GRADE STAMP · 24-UNIT PROFILE · SHOWN AT 96 PX
Table 2.1 — Fields of the composite stamp
FieldCarriesRedundant channels
1 · TierEvidence tier A–ELetterform; 5-slot rail filled to rank; field colour
2 · ProvenanceData provenance P0–P4P-prefixed numeral; 4-slot rail; neutral value ramp
3 · VersionMethodology versionInk on paper; always the lightest field, so it reads as apparatus not verdict
Index notchAuthenticity silhouetteCut from the top edge at the field 2/3 boundary; survives as a bite in the outline at any size

The two profiles

The mark is not one drawing scaled up and down. It is two drawings with a switch-over at 20 px, because features that are correct at 24 units fall below the rendering floor at 16.

Table 2.2 — Small-size geometry, what changes and why
Feature24-unit profile16-unit profileReason
Tier scale5-slot rail, cumulative fillSingle position tick, 5 stopsSlot gaps fall below 1 device pixel and grey out into a smear
Provenance scale4-slot rail, cumulative fillSingle position tick, 5 stopsAs above
Provenance glyph (standalone)"P3""3" plus the chamferTwo glyphs in a 16 px square close up the counters; the chamfered container already names the scale
Stroke weight1.7–1.9 units1.1–1.3 units, integer-alignedHairlines must land on whole pixels or they render at 40% grey
Index notch3.2 × 3 units3 × 2 unitsKeeps the bite visible without eating the version field
Superseded overlay45° hatch + dashed frame + cancellation barDashed frame + cancellation bar onlyHatch at 16 px destroys the letterforms it is meant to annotate

Switch-over is automatic: any render below 20 px uses the 16-unit profile. Below 14 px the composite stamp is prohibited; use a standalone tier mark or the text notation.

§03

Mark builder

Every mark on this page is drawn by the same geometry engine that produces the asset files. Select values and read the SVG out.

Evidence tier
Provenance grade
Methodology version
Render size
24 px24-unit profile
Rendering mode
Grade state
notation A/P3 (v0.1) viewBox bytes state current
16 px
24 px
48 px
tier only
prov only

The builder writes no state to storage. Values are held in memory only.

§04

The two scales, complete

Tier marks are square. Provenance marks are chamfered at the top-left. The container shape alone tells you which scale you are reading, before any glyph is resolved.

4.1 Evidence tier A–E · standalone marks, 24-unit profile

Table 4.1 — Evidence tier definitions
MarkTierDefinitionRail

4.2 Provenance grade P0–P4 · standalone marks, 24-unit profile

Table 4.2 — Provenance grade definitions
MarkGradeDefinitionRail

4.3 Composite matrix · all 25 tier × provenance combinations at 24 px

A grade is always a pair. Publishing a tier without its provenance grade is a breach of the display licence (§14.3).

§05

Size ladder

Rendered at true pixel size — no scaling, no zoom. The profile switch happens between 16 and 24.

5.1 Minimum sizes

ApplicationMinimum heightProfile
Composite stamp, screen16 px16-unit
Composite stamp, print4.0 mm16-unit, mono
Standalone tier / provenance mark, screen14 px16-unit
Standalone tier / provenance mark, print3.5 mm16-unit, mono
Third-party badge44 pxfixed 300 × 76
Wordmark (secondary)11 px

5.2 Clear space

Clear space on every side equals the height of one tier field — that is, the full height of the mark. Nothing may enter it: no rules, no type, no product photography, no other certification marks. Inside running text the rule relaxes to 0.5× on the left and right so the mark can sit inside a sentence, but never above or below.

Hatched area = protected clear space, 24 units on all four sides of a 24-unit mark.

§06

Monochrome, single colour, inverted

The mark was drawn in monochrome first. Colour was added afterwards and can be removed without loss.

6.1 Monochrome (print)

Ink on paper. For journal reproduction, 1200 dpi bitmap or vector; never a screenshot.

6.2 Single colour (currentColor)

Inherits the host surface's text colour. Permitted only where the host cannot render two inks — packaging foil, embossing, single-plate print.

6.3 Inverted (ink ground)

Paper strokes, transparent fields. For dark interfaces and ink-ground documents.

6.4 Colour-blind and print safety

Table 6.1 — Channel redundancy audit
Reader conditionChannels still availableResult
Full colour visionGlyph, rail position, colour3 of 3
Deuteranopia / protanopiaGlyph, rail position, luminance3 of 3
AchromatopsiaGlyph, rail position, luminance3 of 3
Monochrome printGlyph, rail position2 of 3
Fax / 1-bit thresholdGlyph, rail position2 of 3
Screen readerSVG title + aria-labelFull notation read aloud

All five tier fields carry white text at a contrast ratio of at least 4.5:1, and all five differ in luminance as well as hue, so the ramp still reads as a ramp when hue is removed. The provenance ramp is achromatic by construction — it is a value scale from ink to paper, with the two lightest grades taking ink glyphs instead of paper ones.

The tier ramp deliberately avoids a red-to-green gradient. Green-to-ochre-to-oxblood keeps adjacent steps separable under the two common dichromacies, and the rail makes the question moot in any case: you can count the filled slots.

Never recolour a mark to match a brand palette. The colours are semantic, not decorative, and the licence does not permit it.

§07

Specimen in context

The mark's real habitat is surfaces Etalyn does not control and did not design. These five specimens are deliberately drawn in other people's visual languages.

7.1 Journal reference footnoteIllustrative
Ergogenic claims in consumer sports nutrition: a structured audit of label statements
J. Sports Sci. Nutr. 2026;48(3):211–229

Of 412 label statements audited, 61 (14.8%) were supported by evidence graded B or above by an independent rating body, and 208 (50.5%) rested on provenance grade P1 or lower.14 Grade records were retrieved on 21 September 2026; three had been superseded between audit and submission.15

  1. Etalyn. Creatine monohydrate: lower-limb strength output in trained adults 18–45 . Grade record EG-2026-0417, methodology v0.1, issued 7 August 2026. Reviewer: ⟨name⟩. Available from: https://etalyn.com/v/EG-2026-0417 [accessed 21 September 2026].
  2. Etalyn. Beta-alanine: time-to-exhaustion, recreationally active adults . Grade record EG-2026-0188, methodology v0.1, issued 7 August 2026, superseded 18 September 2026 by EG-2026-0512. Available from: https://etalyn.com/v/EG-2026-0188/v0.1 [accessed 21 September 2026].
The mark sits inline at 16 px on a 11.5 px serif reference line. Note that reference 15 is visibly superseded — dashed frame, grey fields, cancellation bar — without the reader having to parse the date.
7.2 Retailer product pageIllustrative
Micronised Creatine Monohydrate 500 g
★★★★☆ 1,284 reviews
£18.99
Independent evidence rating
Graded claim: lower-limb strength output in trained adults 18–45. The rating applies to that claim only, not to this product or brand.
The badge is served from etalyn.com/badge/EG-2026-0417/v0.1.svg, so it restyles itself the moment the grade is superseded. The retailer never holds a copy.
7.3 Presentation slideIllustrative
Evidence position, Q3 portfolio
  • Two of our six actives now hold a tier B grade or better
  • Provenance remains the binding constraint: four claims sit at P2
  • Reformulation target: move the endurance claim from D/P1 to C/P3
Grades as at 2026-08-07 · verify at etalyn.com/v/EG-2026-0417 · illustrative
On a slide the stamp is used at 30 px so the version field survives projection. Screenshotted marks go stale silently, which is why the slide furniture carries an "as at" date.
7.4 Academic citation, four formsIllustrative
Canonical
Etalyn Grade B/P2 (v0.1) — creatine monohydrate → lower-limb strength output, trained adults 18–45; issued 2026-08-07; reviewer ⟨name⟩. https://etalyn.com/v/EG-2026-0417
Footnote short form
Etalyn B/P2 (v0.1), EG-2026-0417.
Vancouver
Etalyn. Creatine monohydrate: lower-limb strength output in trained adults (Grade B/P2). Grade record EG-2026-0417, methodology v0.1. London: Etalyn; 2026 Aug 7. Available from: https://etalyn.com/v/EG-2026-0417
In-text
… strength output (Etalyn B/P2, v0.1) …
Tier and provenance are separated by a solidus and never reordered. The version is always parenthesised. The record ID is sufficient on its own to resolve the grade.
7.5 Printed monochrome pageIllustrative
Appendix C · Evidence positions, ingredient panel
EntityGraded claimGrade
Creatine monohydrateLower-limb strength output
Caffeine anhydrousTime-to-exhaustion, cycling
Beta-alanineTime-to-exhaustion
Tribulus terrestrisSerum testosterone
Grades verified at etalyn.com/v/ on 2026-08-07. Superseded grades are shown with a cancellation bar and remain citable as historical positions.
Simulated single-plate black reproduction at reduced size. Every value is still recoverable: letter, numeral, rail position. No colour is doing any work.
7.6 Current vs superseded, side by sideIllustrative

Current

Superseded

Three independent staleness channels: chroma removed, frame dashed, cancellation bar across the foot. At 16 px the hatch is dropped and the bar carries the signal alone.
§08

Retail reduction layer

The professional cartouche in §01–§07 is built for readers who are paid to read it. A shelf is not that. This section defines a second, separate instrument: one band, six states, derived from the professional record by fixed rule. Both layers exist. Neither replaces the other.

Identity approved 2026-08-07, reference ETA-DOC-0002 rev 1. The six band wordings below are frozen and exact. Methodology remains v0.1; reviewer and adjudicator remain Not yet appointed, so all grades shown are provisional.

8.1 Why one band, and not the cartouche shrunk down

Multi-scale labels roughly halve consumer comprehension against a single summary band. In a study of 12,391 participants across 12 countries, a single summary letter produced an odds ratio of 3.15 for correct comprehension (CI 2.68–3.71), against 1.66 (CI 1.41–1.95) for multiple separate scales shown together (PMC7678195).

The Swiss arm, 1,088 participants, is sharper still: 4.02 (CI 2.81–5.75) for the single summary form against 2.09 (CI 1.46–2.99) for the multi-scale form (PMC7046267).

This does not make the cartouche wrong. It makes the cartouche wrong for a shelf. A formulator needs to see that a tier is high while its provenance is not. A shopper holding a tub in one hand needs one thing they can act on. The reduction is a designed loss, applied by rule, and the rule is printed in §09 so anyone can check what was lost.

Two instruments, one record

Every retail band is derived from a professional grade record that exists in full, with a public verification URL printed on the band itself. The band is never the source of truth; it is a pointer to one. A reader who wants the two-field detail follows the URL and gets the cartouche.

Only one band may ever appear on a surface. Two bands on one carton reproduce exactly the multi-scale condition the evidence above measures.

8.2 The six states

Six states, no seventh, no in-between. State numbers are fixed and are part of the notation: the rail step and the numeral both encode position, so a band read in the dark, in monochrome, or by someone who cannot read the word still carries the position.

8.3 Four channels, in order of primacy

1The wordThe primary identifier. Set in the IBM Plex Sans stack as live <text>, at 19 units for single-line states and 15.5 units for two-line states. This is the only font-dependent element of the band, along with the mono furniture strip. If the font fails to load, the stack falls back to the system UI sans and the word still renders; nothing in the band's meaning is carried by the typeface itself.
2Rail positionA fixed six-step rail across the foot of the band. The active step is taller than the inactive steps and reads as a filled slot, not as a hairline. Drawn geometry, no font, no colour dependency.
3State motifA drawn mark in the state cell: solid bars, a diagonal cut, opposed triangles, a strike, a stub. No stroke in a motif is thinner than 4 units, so nothing disappears at print minimum. This is the channel that separates state 3 from state 6 at a glance.
4ColourLast, and removable. Every band renders identically in single-plate monochrome with zero information loss — the word, the rail step, the numeral and the motif all survive. Colour is a speed aid for a sighted reader in good light, and nothing else.

The word block is drawn in ink at all times, never in the state colour. Removing colour removes only the state-cell fill and the rail fill; it never removes text.

8.4 State 3 against state 6 — the distinction the layer exists for

“Widely studied, poorly measured” and “Too little evidence yet” describe opposite situations that every existing certification collapses into the same tick or the same silence. They are given the most separated geometry in the set: three full bars struck by a diagonal cut, at rail step 3, against a single stub bar at rail step 6.

8.5 Monochrome and minimum size

The band must survive 8 mm of carton height in single-plate print and 44 px on a 375 px viewport. Both are shown below at true 1× pixel size, not scaled up for the page.

Inactive rail steps in monochrome are drawn as 2.2-unit outlines, never as hairlines. Hairline elements below one device pixel vanish in print and on low-density screens; no element of this band relies on a stroke thinner than 1.8 units.

8.6 Wording that may never appear on a retail surface

These words are permanently prohibited on any surface carrying a retail band, in any language, including in surrounding brand copy within the band's furniture block: works · effective · proven · clinically proven · boosts · improves · enhances · supports · helps · guaranteed. They describe an effect on a person. A grade describes the state of a literature. The six state words are the complete permitted vocabulary.

§09

Derivation

The band is not chosen. It is derived, by nine ordered rules, from three inputs already on the professional record: tier A–E, direction, and provenance P0–P4. First match wins; evaluation stops there.

9.1 The nine rules, in order

    Order is the whole specification. Rule 5 placed above rules 6–8 is what makes a tier-A record with P1 provenance read “Widely studied, poorly measured” rather than “Strong evidence”. Moving it would change the meaning of the layer, not merely its output.

    9.2 Three caps

    Direction precedence (rules 1–3)

    Direction is read before tier and before provenance. A claim that was tested and failed cannot be rescued by a high tier, and a high tier cannot be reported as a positive finding when the finding is negative.

    Provenance cap (rule 5)

    Provenance P1 or P0 caps the band at state 3 regardless of tier. Volume of literature is not measurement quality. This is the cap that fires on C-16.

    Contested cap (rule 2)

    Where credible sources reach opposite conclusions, the band says so and stops. Etalyn does not break a live scientific tie on a shelf-edge label.

    9.3 Live derivation

    Pick a tier, a direction and a provenance grade. The rule that fires is highlighted in the table above; the resulting retail band and the professional cartouche for the same record are shown together.

    Rule fired
    Retail band — one summary
    Professional cartouche — same record

    9.4 Worked reductions from real records

    Every row is a published grade record, not a constructed example.

    RecordClaim as gradedTier / prov.DirectionRuleRetail band

    Record identifiers, tiers, directions and provenance grades are taken from the published creatine and Tribulus grade records as issued under methodology v0.1, including where a record's provenance differs from an earlier worked draft.

    9.5 Exhaustive matrix — all 100 combinations

    Every tier × direction × provenance combination with the state it produces and the rule number that produced it. Nothing in the derivation is undefined and nothing falls through to a default.

    The numbered square is an audit chip, not a retail band: it carries no word and no rail, and may never appear on a retail surface. It exists so 100 outcomes fit on one screen.

    §10

    Retail specimens in context

    The band on surfaces Etalyn does not control. Each simulated surface is drawn in its own visual language, not Etalyn's, and is marked Illustrative. Every specimen carries the full six-item furniture block required by the spec.

    10.1 Supplement carton, back panel — true print scaleIllustrative
    Set in millimetres. The band is 46 mm wide and 10 mm tall on a carton of this size — above the 8 mm minimum. Record C-01, the strongest real record in the set, so the carton case is not built out of an unrepresentative worst case.
    10.2 Retailer product page — 375 px mobileIllustrative
    Record T-01. The band reads Evidence says no directly above the buy button, on the page of a product whose category exists because of this claim. A rating layer that only appears where it flatters is not a rating layer.
    10.3 Shelf-edge label — single plate, no colourIllustrative
    Record T-01 again, printed the way shelf-edge labels are actually printed: one plate, black on white, no spot colour. The word, the rail step, the numeral and the motif all survive; the colour channel is simply absent.
    10.4 Professional cartouche against retail band — same recordEtalyn surface
    Record C-16. Left is what a journal, a formulator or a compliance reader receives. Right is what a shopper receives. They are the same finding, reduced by rule 5.

    Mandatory furniture — all six items, every retail surface

    A band without its furniture is not a band. Required on every surface: (1) the graded claim in words with its scope; (2) the sentence “Applies to this claim, not to this product or brand.”; (3) the methodology version and issue date; (4) a verification URL resolving to the full professional record; (5) a common-ownership disclosure at equal visual weight where the surface is under common ownership with Etalyn, per architecture §17. The scope line is load-bearing on T-01: the only positive androgen signal in the literature was found in hypogonadal men, and printing the claim without its scope would transfer that signal to a population it was never observed in.

    §11

    Two records, two failure modes — C-16 and T-01

    The two records below are the argument for the whole layer. Each breaks a single-band reduction in the opposite direction. C-16 is a strong tier that must not print a strong band. T-01 is a weak tier whose band is decided before tier is ever read. Judge them together.

    11.1   C-16  ·  the provenance capA / P1 (v0.1)

    Claim: creatine monohydrate is tolerated without excess adverse events, in adults 18–65 at 3–5 g/day.

    685 trials. 12,839 exposed participants against 13,452 on placebo. Adverse events 4.60% against 4.21%, p = 0.828, with no significant difference across any of 35 side-effect categories (PMC12291206). By any tier rule this is tier A, and it is graded tier A.

    But the harm data was captured by unstructured spontaneous reporting — participants mentioning what they happened to remember, not a structured instrument administered to everyone. That is provenance P1. Rule 5 matches before rule 6 is ever reached.

    Every existing supplement certification would have displayed a tick here. A tick certifies that a box was ticked. Nothing in a tick can express that the tier is high and the measurement underneath it is not, because a tick has one state. The band has six, and one of them exists for exactly this case.

    11.2   T-01  ·  direction precedenceD / P1 (v0.1)

    Claim: Tribulus terrestris raises serum total testosterone in healthy men. Scope: healthy adult men — not hypogonadal or androgen-deficient men.

    Ten studies, 483 participants. Eight of the ten reported no significant change in androgen profile. The two that did were in hypogonadal or androgen-deficient men, with an increase of roughly 60–70 ng/dL — a population the claim explicitly excludes, and a signal that must not be transferred across populations (PMC11990417).

    No pooled estimate exists because the review authors declined to produce one, stating verbatim: “Given a sharp selection bias and unreliable data description in the original trials, performing a meta-analysis was avoided.” Half the studies scored low methodological quality on the Jadad scale.

    Provenance is P1 rather than P0 because serum androgens were measured directly — but the regulatory assessment repeatedly records trials reporting significance with no numerical value, no effect size, no confidence interval and no exact p-value (EMA assessment report). Measurement occurred; measurement quality cannot be verified.

    Tier D would have produced state 6 under rule 4. Provenance P1 would have produced state 3 under rule 5. Neither rule was reached, because direction matched at rule 1. That ordering is deliberate: a claim that was tested and failed must not be reported as a claim that was never properly tested.

    11.3 Tested-and-failed against barely-tested

    This is the distinction that a tick, a star rating and a traffic light all destroy. T-01 and C-18 are both records nobody wants on a carton, and they mean entirely different things. Shown adjacent so the visual separation can be judged rather than asserted.

    T-01 sits at rail step 5 with a struck bar; C-18 sits at rail step 6 with a single stub. In monochrome, at 8 mm, the two remain separable by position and motif alone. A reader who confuses them has been told the literature is silent when in fact it has spoken — which is the specific error this layer refuses to make.

    What the pair establishes

    A tier can be high and still not earn a strong band, because a band summarises the whole record and not its best field. A tier can be low and still not produce “Too little evidence yet”, because direction is read first. Both outcomes are unpopular with the party paying for the surface, both are produced mechanically from the published record, and both are reproducible by anyone with the rule table in §09.1.

    §12

    Citation and verification structure

    The graphic is a pointer. This is the thing it points at, and the reason a forged mark is worthless.

    12.1 Canonical citation string

    Etalyn Grade <TIER>/<PROV> (<METHODOLOGY>) — <ENTITY> → <OUTCOME>, <POPULATION>; issued <ISO-8601 DATE>; reviewer <SURNAME, INITIAL>. <VERIFICATION URL>

    Worked examples

    ContextString
    Journal, fullEtalyn Grade B/P2 (v0.1) — creatine monohydrate → lower-limb strength output, trained adults 18–45; issued 2026-08-07; reviewer ⟨surname, initial⟩ https://etalyn.com/v/EG-2026-0417
    Footnote shortEtalyn B/P2 (v0.1), EG-2026-0417.
    In-text parenthetical(Etalyn B/P2, v0.1)
    Pack / label, ≥6 ptETALYN GRADE B/P2 · v0.1 · etalyn.com/v/EG-2026-0417
    Superseded, cited historicallyEtalyn Grade C/P2 (v0.1) — beta-alanine → time-to-exhaustion, recreationally active adults; issued 2026-08-07; superseded 2026-09-18 by EG-2026-0512. https://etalyn.com/v/EG-2026-0188/v0.1
    No grade issuedNot graded by Etalyn.

    Prohibited: "Etalyn approved", "Etalyn certified", "Etalyn recommends", any string that omits the provenance grade, any string that omits the methodology version, and any use of the word "score".

    12.2 URL pattern

    PurposePatternBehaviour
    Canonical record/v/{record}Always resolves; always shows the current state of that claim
    Human alias/v/{entity}/{outcome}302 to the canonical record
    Version-pinned/v/{record}/{version}The grade as issued under that methodology, with a current/superseded banner
    Immutable snapshot/v/{record}/{version}/{issued}Byte-stable archive copy for citation
    Machine record/v/{record}.jsonSigned JSON; includes status, changelog, reviewer, sample
    Attestation/v/{record}/attestation.txtDetached Ed25519 signature over the record hash
    Badge image/badge/{record}/{version}.svgLive render; automatically returns the superseded variant once superseded
    Registry index/registerAppend-only public list of every record ever issued

    {record} = EG-YYYY-NNNN. {version} = v<major>.<minor>. Record IDs are never reused and never deleted; a withdrawn grade resolves to a withdrawal notice, not a 404.

    12.3 The verification page

    https://etalyn.com/v/EG-2026-0417
    Current grade Issued 2026-08-07 · methodology v0.1 · no change since issue

    Claim as graded — verbatim

    Creatine monohydrate supplementation increases lower-limb maximal strength output in resistance-trained adults aged 18–45 over 4–12 weeks.

    Grade
    B / P2  
    Entity
    Creatine monohydrate (CAS 57-00-1)
    Outcome
    Lower-limb maximal strength output (1RM equivalent)
    Population
    Resistance-trained adults, 18–45, both sexes; not assessed in adolescents, over-65s, or clinical populations
    Sample
    17 randomised controlled trials, pooled n = 1,046; 3 registered protocols; 2 trials excluded for undisclosed funding
    Provenance
    P2 — structured self-reported training logs in 11 of 17 trials; device-measured force plates in 4
    Methodology
    v0.1 (2026-08-07) · read the methodology
    Issued
    2026-08-07
    Reviewer
    Not yet appointed
    Adjudicator
    Not yet appointed
    Record
    EG-2026-0417 · SHA-256 4f9c…a17e · attestation

    What would change this answer

    • Two or more adequately powered RCTs with force-plate outcomes reporting null effects would move this to C.
    • A registered replication with device-measured loading in ≥ 60% of participants would move provenance to P3, not the tier.
    • Publication of the three outstanding registered protocols, currently unreported, could move the tier in either direction.

    Changelog for this claim

    DateGradeReason
    2026-08-07Current. First issue under methodology v0.1.

    This claim has one entry because methodology v0.1 is the first issued methodology. Every prior grade remains publicly resolvable; Etalyn does not delete positions it has changed.

    Cite this grade

    Etalyn Grade B/P2 (v0.1) — creatine monohydrate → lower-limb strength output, trained adults 18–45; issued 2026-08-07; reviewer ⟨surname, initial⟩ https://etalyn.com/v/EG-2026-0417

    Right of reply

    The marketing-authorisation holder or manufacturer may contest this grade. Submissions and outcomes are published in the changelog above, whatever the outcome.

    Illustrative record. A real verification page differs only in its data.

    12.4 The superseded case

    https://etalyn.com/v/EG-2026-0188/v0.1
    Superseded Replaced 2026-09-18

    This grade is no longer current. It was issued on 2026-08-07 under methodology v0.1 and was superseded on 2026-09-18 by record EG-2026-0512, which grades the same claim .

    Any badge, label or citation still displaying C/P2 (v0.1) for this claim is out of date. Under the display licence a printed carton may continue to show it for twelve months from that date; a digital surface may not.

    12.5 How a stale or fabricated mark is detected

    • No record, no grade. Every legitimate mark carries a record ID. If /v/{record} does not resolve, the page returns NO SUCH GRADE — not issued by Etalyn, and the register search returns nothing. A fabricated mark fails at the first click.
    • Live badges cannot go stale. Digital badges must be served from /badge/…. When a grade changes, every embedded badge worldwide restyles to the superseded variant within one cache TTL (600 s).
    • Version mismatch is self-announcing. A pinned URL whose version is not the current one always renders the red superseded banner, even if the printed graphic looks pristine.
    • Signed records. Each record is hashed and signed (Ed25519); the public key fingerprint is published at /keys and the daily Merkle root of the register is mirrored to two third-party transparency logs, so Etalyn itself cannot silently rewrite history.
    • Content mismatch. The verification page prints the graded claim verbatim. A brand that attaches a real grade to a different claim is contradicted by its own verification link.
    • Geometric tells, used last. Index notch, 5-slot and 4-slot rails, stencil counters and field proportions are specified to the unit. They help spot a redraw, but they are never the proof — the register is.
    §13

    The derived document register

    The mark comes first. Type, palette, tables and furniture are chosen to make the mark unambiguous and to make the surrounding document read as a document. Reference class: The Lancet, Cochrane, ClinicalTrials.gov, EMA and EFSA registers, UK Biobank, FT and Economist data journalism.

    13.1 Type

    NEWSREADER 300/400/500
    Prose, headings
    17 / 25.5 px
    Optical sizing on

    Seventeen randomised controlled trials, pooled n = 1,046, report a positive effect on lower-limb maximal strength output in resistance-trained adults aged 18–45. Eleven of the seventeen relied on structured self-reported training logs, which is why the provenance grade is P2 and not P3.

    IBM PLEX SANS 400/600
    Tabular data, labels
    13.5 px / 11 px caps
    Unambiguous 1 l I 0 O
    Dense tables are the primary content type of this institution. Plex Sans is compact at 13.5 px, has genuinely distinct 1/l/I and 0/O, and carries the language coverage a European register needs. It is never used above 15 px and never for prose.
    IBM PLEX MONO 400/500
    Identifiers, furniture, grades
    11–12 px
    Tabular figures, always
    EG-2026-0417 · v0.1 · 2026-08-07 · SHA-256 4f9c…a17e
    ETA-DOC-0001 · rev 1 · REVIEWER NOT APPOINTED · CURRENT

    Five sizes on the page, no more: 46 / 29 / 19 / 13.5 / 11. Numerals are tabular everywhere without exception, because columns of grades must align down the page. Italic is for graded claim wording only — it marks "these are their words, not ours".

    13.2 Palette — one job only

    The document is achromatic. The only saturated colour permitted on any Etalyn surface belongs to a grade mark or to an interactive affordance in viridian. There are no brand colours in the ordinary sense, because a brand colour competing with the tier ramp would make the tier ramp ambiguous — and the tier ramp is the product.

    13.3 Tables and footnotes

    Table 13.1 — Graded claims, illustrative panel
    EntityGraded claimPopulationnGrade
    Creatine monohydrateLower-limb strength outputaTrained adults 18–451,046
    Caffeine anhydrousTime-to-exhaustion, cyclingTrained adults 18–402,318
    Beta-alanineTime-to-exhaustionRec. active adults744
    Ashwagandha (KSM-66)Perceived stress, PSS-10Adults 18–65412
    Tribulus terrestrisSerum testosteroneAdult males188
    Red-light panel, 660 nmRecovery, DOMS 48 hTrained adults 18–4596
    1. Claim wording is reproduced as graded; it is not Etalyn's paraphrase of the manufacturer's marketing.
    2. Grades are illustrative throughout this specimen and assert nothing about the named entities.

    Rules for tables

    • Horizontal rules only. A vertical rule in a data table is a confession that the columns are too tight.
    • Header rule 1.5 px ink; body rules 1 px at 22% ink. Nothing heavier, nothing zebra-striped.
    • Numbers right-aligned, tabular, with the unit in the header — never repeated in the cell.
    • The grade column is always last and always holds a real mark at 16 px, not the text notation.
    • Every table has a numbered caption above and a source or scope note below, set in mono at 11 px.
    • Footnote markers are superscript mono; footnotes sit under a full-width rule at the foot of the block they belong to, not at the foot of the page.

    13.4 Institutional document furniture

    Document
    ETA-DOC-0001
    Version
    v0.1
    Issued
    2026-08-07
    Reviewer
    Not yet appointed
    Status
    Current
    Changelog
    3 revisions

    The furniture block. Mandatory on every published surface, top-right on desktop, directly under the title on mobile. Six fields, never fewer.

    What would change this answer

    • A named, falsifiable condition.
    • A second, in a different direction.
    • What is currently unmeasurable, and by what.

    Standing block on every graded claim. Two to four bullets, each stating a specific finding that would move the grade, and in which direction. "More research is needed" is prohibited.

    Reviewer line
    Reviewer not yet appointed. Grades issued under methodology v0.1 are provisional until a named reviewer is in post. The post is open: terms of the external review.

    Named humans, with roles and a date. No "the Etalyn team". A rating nobody signed is a rating nobody is accountable for.

    13.5 Prohibited

    Do

    • Set the mark on paper or ink, at full opacity, at or above minimum size.
    • Keep tier and provenance adjacent, in that order, separated by a solidus.
    • Print the methodology version and the verification URL wherever the mark appears.
    • Use the served badge endpoint on any digital surface.
    • Let the mark be the only saturated colour in its clear space.

    Do not

    • Recolour, rotate, outline, emboss, add shadow, gradient, glow or rounded corners.
    • Stretch, condense, or alter field proportions; the version field is never removed.
    • Show a tier without its provenance grade, or an average of the two.
    • Place the mark on photography, gradients, or inside another badge or ribbon.
    • Animate the mark, or use it as a loading indicator, cursor, or favicon on a third-party site.
    • Say "Etalyn approved", "Etalyn certified", "A-rated by Etalyn" or use the word "score".
    • Use the wordmark larger than the grade mark in the same field of view.

    13.6 Secondary identity assets

    The wordmark and lockup are subordinate. They identify the issuer; they do not carry the verdict. In any composition containing both, the grade mark is set larger and reads first.

    §14

    Display licence, in outline

    Grade Display Licence GDL-1. Royalty-free, non-exclusive, revocable. Distribution is the asset; charging for the mark would defeat it.

    14.1 Who may display the mark

    • Any party accurately reporting a grade that Etalyn has issued: the manufacturer or authorisation holder of the graded entity, retailers, journalists, researchers, clinicians and critics alike. No permission request, no application, no fee.
    • Critical use is expressly permitted. A journalist may display an E/P0 mark next to a product whose maker would rather they did not. The licence cannot be revoked for unflattering but accurate use.

    14.2 Conditions

    • Tier, provenance and methodology version appear together, unaltered.
    • The mark sits adjacent to the exact claim wording it grades, or to a link resolving to it.
    • The verification URL is printed or linked wherever the mark appears.
    • Digital surfaces use the served badge endpoint, not a copied file, except where technically impossible.
    • No implication of endorsement, approval, certification, safety, or efficacy beyond the graded claim.
    • No comparative use implying an ungraded competitor would score lower.

    14.3 When a grade changes after a brand has printed it

    Table 14.1 — Transition obligations
    SurfaceObligationDeadline
    Served badgeNone — restyles automatically600 s
    Copied digital imageReplace or remove10 business days
    Owned-site copy, claim textUpdate to current grade10 business days
    Paid media in flightWithdraw at next scheduled change30 days
    Printed stock already producedMay be exhausted; verification URL must resolve to the superseded notice12 months
    New print runsCurrent grade onlyimmediate
    Withdrawn grade (integrity failure)Cease all display, all surfaces5 business days

    14.4 Breach and revocation

    Breach is defined as: displaying a tier without its provenance; removing the version field; altering colour or proportion; attaching a real grade to a different claim; exceeding the print exhaustion window; or any endorsement framing. Etalyn publishes notices of breach and revocation in the public register alongside the grade concerned, so the enforcement record is as public as the grades.

    The mark is filed as a certification mark. The register, not the trade mark, is the primary enforcement instrument: a party displaying a mark Etalyn never issued is not merely infringing, it is verifiably contradicted at a URL anyone can click.