How this library is built

Six rules

1. 1

   The attribute is measured. The effect is inferred.

   A noise reduction coefficient is a laboratory number produced by a defined method. A guest sleeping better is an inference several steps away from it. Printing them in the same weight, on the same line, is how a test result becomes a health claim.

   appliedMeasured attributes and proposed responses never share a typographic register on this site. Every response carries its own evidence tier and its own exposure conditions.

2. 2

   A certification is not an outcome.

   GREENGUARD Gold means a sample passed an emissions test under specified chamber conditions. It does not mean an occupant of a finished room experiences better air, and it says nothing at all about acoustics, tactility or sleep.

   appliedEvery certification record carries an explicit field for what it does not certify, shown with the same prominence as what it does.

3. 3

   Exposure travels with the claim.

   Surface area, dwell time, viewing distance and contact mode decide whether a property reaches a person at all. A 2% accent of a highly absorptive material does not change a room's acoustics. Wood underfoot and wood on a distant soffit are not the same exposure.

   appliedA response link without an exposure condition is incomplete and is shown as incomplete.

4. 4

   Evidence attaches to the material class, never to the brand.

   No study has tested a named commercial carpet. Studies test wool, or textile surfaces, or absorptive treatment. A product can only inherit that evidence through its own measured attributes, and only as far as those attributes match what was tested.

   appliedCorpus records link to material classes. Products, when they exist, will link to evidence only by way of attribute values, with the inheritance shown explicitly.

5. 5

   Declare the unmeasured.

   An absent noise reduction coefficient is not a low one. A missing emissions test is not a clean result. Blank fields silently become zeros the moment anyone sorts a table.

   appliedUnmeasured attributes render as not measured, are excluded from comparison, and are counted on the coverage page.

6. 6

   Version the method, or the number does not publish.

   Emissions results are not comparable across test protocols. Absorption coefficients depend on mounting. Rating systems revise their thresholds. A number without its method and its date is decoration.

   appliedEvery attribute in the taxonomy names its measurement standard. Every standard record carries a version and an as-of date.

Evidence tiers

Applied to the claimed relationship, never to the material. A material is not strong or weak evidence. The proposition that this measured property reaches this outcome under these conditions is what gets rated, and the last tier is rated so that it can be answered rather than repeated.

• Strong

  Multiple peer-reviewed human studies, reasonably consistent, with at least one conducted in or close to a built environment.

  licenceMay be stated as an established relationship, with its boundary conditions.

• Moderate

  Several human studies, but with methodological limits, small samples, or a meaningful jump from what was tested to how a material is actually used.

  licenceMay be stated as a supported relationship. Never as a design rule.

• Emerging

  Limited direct human research, or evidence drawn mostly from adjacent exposures.

  licenceMay be stated as an open question with the relevant work named.

• Mechanistic hypothesis

  A biologically plausible route with no adequate built-environment evidence behind it.

  licenceMay be described as a hypothesis, labelled as one, and never used to justify a specification.

• Unsubstantiated

  A claim in circulation, usually a marketing claim, without independent evidence.

  licenceRecorded so that it can be answered. Never repeated as fact.

Exposure

Surface area, dwell time, distance and contact mode decide whether a property reaches a person at all. A two percent accent of a highly absorptive material does not change a room. Wood underfoot and wood on a distant soffit are not the same exposure, and a finding that omits which one it tested has not reported its conditions.

• UnderfootContinuous plantar contact during standing and walking. Thermal and acoustic properties dominate.
• Hand contactBrief, deliberate, high-acuity. Handles, rails, switches, hardware.
• Seated or reclining skin contactLong dwell, large contact area. Upholstery, bedding, headboards.
• Direct field of viewFoveal attention. Pattern, contrast and gloss are resolved in detail.
• Peripheral field of viewLarge-area surfaces read as luminance and colour rather than as pattern.
• OverheadCeilings, soffits. Rarely looked at directly, acoustically decisive.
• Large-area enclosureA material covering enough of a room's surface to change its acoustic or luminous behaviour.
• Ambient acoustic fieldContribution to reverberation, intelligibility and background level, whether or not the material is visible.
• Breathing zoneEmissions and odour reaching the occupant's air, governed by loading ratio and ventilation.
• Sleep durationSix to nine hours of unbroken exposure with no behavioural adaptation available. The defining hospitality case.