# Dispensers: research notes

Date: 2026-10-25. Scope: manual and battery-powered devices that hold a liquid (or, rarely, a bar of soap) and release a measured portion of it on demand — pump and foaming-pump bottles for hand soap, lotion and dish soap; touchless/sensor dispensers for soap and hand sanitiser; multi-chamber shower dispensers for shampoo, conditioner and body wash; sink- and counter-mounted pump dispensers; and dry soap mills. In the Zamana taxonomy this is `supplies/household/dispensers`, a sibling leaf with no children. Excluded, with the sibling that owns it: the dispensed product itself — soap, lotion, sanitiser, toilet paper, paper towels, tissues (`toilet_paper_wipes`, `paper_tissue_towels`); static holders and racks with no dispensing action — toilet-paper holders, towel bars, toothbrush holders (`bathroom_storage`); trash-bag box dispensers, which are packaging for the bags (`garbage_bin_bags`); dry-goods gravity bins for cereal, flour or sugar (`kitchen_organizers`); and plumbed or compressor-driven water, ice and beverage dispensers/coolers, a different taxonomy branch entirely (`machines/appliances`, "Water Filters, Coolers & Dispensers"). Where a product could plausibly sit in either leaf (e.g. a wall-mounted toilet-paper holder with a cutting mechanism), it stays with the sibling that already owns the base product.

Quote labels: **verbatim** = I read the sentence myself, either because the fetch tool returned it in quotation marks, because I curled the page or a JSON API and read the raw text, or because pdftotext produced it from a downloaded PDF. **fetched excerpt** = the fetch tool summarised the page but returned the key numbers or phrases in quotation marks, which I treat as reliable for those quoted fragments only. **search excerpt** = from a search-engine summary, page unread; not used for any number in category.json. WebSearch reported its budget exhausted for this session after three queries, so research here is direct fetches (WebFetch, curl) and the EuropePMC REST API (which returns raw abstract JSON and counts as verbatim once curled).

## Seed

Seed run output (recorded verbatim):

```
=== Example entry (from an unrelated category) ===
category: olive-oil
what this shows: a facet that experts rank first, that almost no listing carries,
and that is expensive to backfill. ... Also shows a derived metric hanging off a
researched facet.
<example> harvest_date ... </example>
=== Persona ===
A competitive endurance athlete, who thinks in cost per unit of the active thing
and what is a waste of weight
```

1. **The dispensers equivalent of harvest date is `refill_system` — open/bulk-fill versus a sealed cartridge or pouch.** Almost no consumer listing states it outright (a plain pump bottle is open-fill by default and nobody labels the obvious), it needs a one-sentence judgement attached, and it is the single number three independent peer-reviewed studies actually measured: bulk-refillable soap dispensers run meaningfully dirtier than sealed ones, and open sanitiser dispensers lose alcohol strength to evaporation that sealed ones do not. Survived the data check and became split 3.
2. **The endurance-athlete persona (cost per unit of the active thing, what is wasted weight) would ask, for a hand-sanitiser dispenser specifically: how much of the 60%-plus alcohol you paid for is still alcohol by the time you use it?** That turned out to be a real, measured risk, not a stretch: a 2022 PLOS ONE study found open refillable dispensers let ethanol evaporate, producing subpotent sanitiser in a meaningful share of samples (see below). Survived and is now the strongest line in `refill_system.why_experts_care`, the ethanol-evaporation judgement, and one of the two safety judgements. Dropped: a literal "dead volume in the dip tube" facet (real, well known anecdotally, but I found no measured source for how much is unreachable, so it stays here as colour rather than a number in the JSON).

## Sources and what each supports

### Mechanism (primers)

**Wikipedia, Soap dispenser** (curled and read directly: `wiki_soap_dispenser.html`)
https://en.wikipedia.org/wiki/Soap_dispenser
- verbatim: "A soap dispenser is a device that, when manipulated or triggered appropriately, dispenses soap (usually in small, single-use quantities). Soap dispensers typically dispense liquid soap or foam soap. They can be automatic or manually operated by a handle."
- verbatim (liquid pump parts): "Actuator – This is the top of the pump from which is pressed down to get the liquid out... Dip tube – This is the visible tube that carries liquid from the bottom of the bottle up to the housing... Interior components – A spring, ball, piston or stem that helps move the liquid to the actuator."
- verbatim (operation): "When the user presses down the actuator, the piston compresses the spring and upward air pressure pulls the ball upward, along with the liquid product into the dip tube and then reaches the housing. When the user releases the actuator, the spring returns the piston and actuator to the normal position and the ball returns to its earlier position to stop the backflow of the liquid back to the bottle."
- verbatim (dry soap): "Some soap dispensers grate, plane or grind solid soap bars to flakes or powder as they are dispensed... About 40 grams (1.4 oz) fresh weight of soap is equivalent to 1 liter... providing soap for up to 400 handwashings. Soap mills are common in public washrooms in Germany."
- verbatim (foam): "Foam soap dispensers have dual foam pumps that when used, move both air and soap, injecting both together through small openings to create a lather. They can be found in both manual and automatic varieties."
- verbatim (automatic): "An automatic soap dispenser is specifically a hands-free dispenser of liquid or foam soap, and generally can be used for other liquids such as hand sanitizers, shampoos or hand lotions. They are often battery-powered... The touch-free design dispenses the liquid when a sensor detects motion under the nozzle."
- Supports: the mechanism section and its three families (manual pump, foam pump, touchless), the dry-soap-mill aside, and that automatic units generalise across soap/sanitiser/lotion/shampoo. Confidence high.

**Wikipedia, Automatic soap dispenser** (fetched excerpt, quoted fragments)
https://en.wikipedia.org/wiki/Automatic_soap_dispenser
- fetched excerpt (quoted): "Radar-based sensor: This kind of sensor sends out bursts of microwave or ultrasound energy and waits for the energy to reflect back." "Passive infrared sensor: Infrared sensors detect infrared energy that is emitted by one's body heat." "Without having a wide variety of individuals touching the pump, bacterial transmission will be eliminated." "Dispensers will only distribute a set amount of soap per motion activation." "If a photo sensor-based dispenser is not designed properly, it may only detect a narrow range of skin colors."
- Supports: the three sensor types (vocabulary) and the touch-free hygiene sales pitch, which judgement 3 below treats with caution because the article gives no citation for the "bacterial transmission will be eliminated" line and the one study that looked at the actual contamination route (next section) found the loophole in the pump mechanism, not in touch. Confidence medium (unsourced claim inside the article) for the hygiene line; high for the sensor-type definitions.

### Hygiene — bacterial contamination of refillable dispensers (academic, via EuropePMC REST API, raw JSON read directly)

**Zapka et al., "Bacterial hand contamination and transfer after use of contaminated bulk-soap-refillable dispensers," Applied and Environmental Microbiology, 2011** (PMID 21421792; curled from `https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=...` as raw JSON)
https://pubmed.ncbi.nlm.nih.gov/21421792/
- verbatim (from the abstract, read directly in the API's JSON response): "Bulk-soap-refillable dispensers are prone to extrinsic bacterial contamination, and recent studies demonstrated that approximately one in four dispensers in public restrooms are contaminated." "Gram-negative bacteria on the hands of students and staff increased by 1.42 log10 CFU per hand (26-fold) after washing with soap from contaminated bulk-soap-refillable dispensers. In contrast, washing with soap from dispensers with sealed refills significantly reduced bacteria on hands by 0.30 log10 CFU per hand (2-fold)." "the mean number of Gram-negative bacteria transferred to surfaces after washing with soap from dispensers with sealed-soap refills (0.06 log10 CFU) was significantly lower than... contaminated bulk-soap-refillable dispensers (0.74 log10 CFU; P < 0.01)."
- Supports: the one-in-four public-restroom contamination rate, and the 26-fold-up versus 2-fold-down direction of the refill_system split. Confidence high (peer-reviewed, controlled field and lab study).

**"A loophole in soap dispensers mediates contamination with Gram-negative bacteria," 2023** (PMID 37877653, same API method)
https://www.ncbi.nlm.nih.gov/pubmed/37877653
- verbatim (abstract): "a bacterial contamination rate of 25%, with values of up to 10^8 colony-forming units/mL (CFU/mL), was reported" [citing the earlier literature]. "Following the collection of 104 liquid soap dispensers from hotel rooms across Germany... 70.2% of the refillable pump dispensers (mean total bacterial count = 2.2 × 10^5 CFU/mL) but only 10.6% of the nonrefillable press dispensers, were contaminated (mean total bacterial count = 1.5 × 10^1 CFU/mL)." "we contaminated different dispensing systems with P. aeruginosa/P. gergoviae, to reveal the route of contamination and identied the pressure release of standard pump dispensers as the loophole for microbial contamination."
- Supports: a second, independent contamination comparison (70.2% vs 10.6%, a far wider gap than the "one in four" headline figure, because this sample is hotel-room units rather than public restrooms) and, critically, that the mechanism responsible is the pump's pressure-release valve itself, which exists on both touch and touch-free pump dispensers. Confidence high for the measurement, medium for generalising the exact percentages beyond hotel rooms (noted as a judgement below).

### Hygiene — alcohol evaporation in open hand-sanitiser dispensers (academic)

**"Presence of unsafe chemical impurities, accelerated evaporation of alcohol, and lack of key labeling requirements are risks and concerns for some alcohol-based hand sanitizers and dispenser practices during the COVID-19 pandemic," PLOS ONE, 2022** (PMID 35303018)
https://pubmed.ncbi.nlm.nih.gov/35303018/
- verbatim (abstract): "the use of refillable dispensers designed to accept ABHS from bulk containers allows for mixing and evaporation that may compromise ABHS integrity." "we performed laboratory-based experiments to determine the impact of dispenser design on alcohol evaporation rates." "Laboratory-based experiments show dispenser design significantly impacts the rate of ethanol evaporation of ABHS products, especially if stored in open refillable dispensers without an internal reservoir." "Subpotent ethanol concentrations were observed in 9.09% and 82.05% of samples from community settings and the school district, respectively, with the school district sample results being associated with dispenser misuse."
- Supports: the chemical half of the refill_system judgement — open, no-internal-reservoir dispensers let alcohol evaporate below an effective strength, measured in the lab and in the field; the 82% school-district figure is explicitly tied to "dispenser misuse," so the register treats it as an upper bound, not a typical outcome. Confidence high.

### Hand sanitiser: effective strength and child poisoning (regulator, primer, professional association)

**CDC, Clean Hands, "About Hand Hygiene"**
https://www.cdc.gov/clean-hands/about/index.html
- fetched excerpt (quoted): "use an alcohol-based hand sanitizer that contains at least 60% alcohol." "If soap and water are not readily available, you can use an alcohol-based hand sanitizer that contains at least 60% alcohol." "You can tell if the sanitizer contains at least 60% alcohol by looking at the product label."
- Supports: the 60% alcohol floor used in the vocabulary entry and as context for the evaporation judgement (evaporation is a problem specifically because it can push a sanitiser below this floor). Confidence high.

**Wikipedia, Hand sanitizer** (fetched excerpt, quoted fragments)
https://en.wikipedia.org/wiki/Hand_sanitizer
- fetched excerpt (quoted): "Alcohol-based hand sanitizers, which contain at least 60% alcohol (ethanol or isopropyl alcohol), are recommended by the United States Centers for Disease Control and Prevention (CDC)." "Products with 60% to 95% alcohol by volume are effective antiseptics." "Alcohol gel can catch fire, producing a translucent blue flame." "Hand sanitizer should be stored in temperatures below 105°F and should not be left in a car during hot weather due to risk of flammability." "the U.S. Food and Drug Administration (FDA) controls antimicrobial handsoaps and sanitizers as over-the-counter drugs (OTC)."
- Supports: second source for the 60% floor, the flammability note and the FDA OTC classification, all kept as vocabulary/judgement context rather than facets (they are properties of the liquid, not of the dispenser). Confidence high for the CDC-sourced figure; medium for the storage-temperature line (not independently re-sourced).

**America's Poison Centers (AAPCC), "Hand Sanitizer" tracker**
https://poisoncenters.org/track/hand-sanitizer
- fetched excerpt (quoted): "As of February 29, 2024, Poison Centers have managed 2,774 hand sanitizer exposure cases in patients that were 0-12 years of age." "Hand sanitizers should be kept out of reach of children at all times, and used only with adult supervision."
- Supports: a current, narrowly-scoped running count (0–12 years, a single tracker page, method and window unstated) and the keep-out-of-reach line. Confidence medium (no stated methodology on the page itself).

**"Pediatric hand sanitizer exposures reported to United States poison centers, 2017–2021," 2023** (PMID 37358035, peer-reviewed, National Poison Data System)
https://pubmed.ncbi.nlm.nih.gov/37358035/
- verbatim (abstract): "There were 95,718 alcohol-based hand sanitizer pediatric cases during the study period." "Most (n = 89,521; 94%) were unintentional, occurred by ingestion (n = 89,879; 93.9%), occurred at home." "Few children (n = 81; 1.4%) were admitted to the intensive care unit." "Population-adjusted rates, by state, of alcohol-based hand sanitizer cases ranged from 280 to 2,700 per million children."
- Supports: a far larger, methodologically stated figure (all ages ≤19, full National Poison Data System, 2017–2021) that disagrees in scale with the AAPCC tracker above because the two count different age ranges over different windows with different methods, not because they contradict each other. Confidence high. Both are kept; see judgement 4.

### Capacity, chambers and power (spec sheets — two independent brands)

**simplehuman, Rechargeable Liquid Soap Sensor Pump, 9 fl. oz.** (fetched excerpt, quoted fragments)
https://www.simplehuman.com/products/rechargeable-sensor-pump-9-fl-oz
- fetched excerpt (quoted): "Capacity: 9 fl. oz." "Built-In Lithium Ion Battery." "4 Months On 1 Charge." Charging time "3-4 hours." "engineered to last through 150,000 cycles." "2-Year Warranty."
- Supports: one brand-specific data point for capacity (9 fl oz / 266 ml) and for the rechargeable value of `power_source`; the 4-month/150,000-cycle figures are kept as this brand's own claim, not generalised. Confidence high for what the page states, about this product only.

**simplehuman, Sensor Pump collection page**
https://www.simplehuman.com/collections/sensor-pumps
- fetched excerpt (quoted sizes): "9 fl oz" (two models), "14 fl oz" (two models, with caddy). All four current sensor-pump models are rechargeable; none in this line uses disposable batteries.
- Supports: that this maker's whole touchless range is rechargeable, which is why the disposable-battery value of `power_source` needed a second brand (next entry). Confidence high.

**Better Living Products, Dispensers collection** and **TOUCHLESS XL Soap & Sanitizer Dispenser product page** (fetched excerpts, quoted fragments)
https://betterlivingproducts.ca/collections/dispensers ; https://betterlivingproducts.ca/products/touchless-soap-sanitizer-dispenser
- fetched excerpt (quoted): "FOAMA Touchless Foaming Soap Dispenser... 295 ml." "TOUCHLESS XL Soap & Sanitizer Dispenser... 18 oz (530 ml)... 4 AA batteries required (not included)... Sensor activated operation." "TOUCHLESS Soap Dispenser... 7.6 oz (225 ml)." Shower dispensers listed as "2 Chamber," "3 Chamber" and "4 Chamber" in the product names themselves (e.g. "AVIVA Shower Dispenser 3 Chamber," "CLASSIC Shower Dispenser 4 Chamber").
- Supports: the second brand's capacity points (225–530 ml), the disposable-AA-battery value of `power_source`, and that chamber count is stated directly in the product name for every multi-chamber shower dispenser on this page, which is why `chamber_count` is marked `source: listing` rather than `researched`. Confidence high for what the pages state.

### Could not reach

GP PRO, Kimberly-Clark Professional and Grainger/Uline dispenser category pages (403, or redirected to an unrelated page); Moen and Delta kitchen soap-dispenser spec pages (404/no spec table rendered — these are JS-rendered catalogues); Home Depot product pages (403); OXO's soap-dispenser page (403); the ADA.gov 2010 Standards pages for sections 308/309 (404 — the site has reorganised and the old chapter URLs no longer resolve; not pursued further because mounting height is a commercial-fixture concern, not a consumer buying decision, so it is noted and dropped rather than chased through the new site map); Aptar's hand-hygiene pump-OEM page (404, would have been the source for any alcohol/gasket-compatibility claim, which is accordingly left out of the register rather than guessed at). Amazon search and product pages returned 503/403, consistent with every earlier register's note that retail listings block fetches; no claim here rests on an Amazon page.

## Data points used for derived metrics

- `cost_per_use`: formula is `refill_price / capacity_ml` (or the listing's stated servings, where given). No sourced numeric constant is needed for the formula itself; it depends only on the listing's own price and the `capacity_ml` facet. The judgement that this number only means something when `refill_system` is open (so you are free to shop any brand's refill) rather than sealed (locked to one proprietary cartridge) is ours, reasoned from the refill_system facet itself, not a separate source.
- `capacity_ml` typical range: 225, 266, 295 and 530 ml across the two spec-sheet sources above (Better Living 225/295/530; simplehuman 266 (9 fl oz) and 397 (14 fl oz)). Stated in the register as "most countertop and touchless units run 225 to 300 ml; touchless stations and some shower caddies go up to about 530 ml," which is the span actually observed across the two brands, not an invented bound.

## Against the existing add-on schema

`zamana-expertise/data/schema-parts/supplies/household/add-ons/dispenser.jsonc` is currently empty (`{}`). `household-standard.jsonc` already supplies `materials` (a shared enum that covers plastic, acrylic, glass, ceramic, metal/steel and more — nothing dispenser-specific needs adding there) and `mounting_install`, whose enum already includes `freestanding`, `wall_mounted_screw`, `adhesive_stick_on`, `over_sink` and `suction` — which is every mounting style this register found evidence for (countertop pumps, sink-top pumps, and suction- or screw-mounted shower caddies). `supplies-standard.jsonc` supplies `sizing.capacity` (with a `ml` unit already in its enum) and `high_signal_markdown`. Because `materials` and `mounting_install` are generic across every household category and this register found nothing dispenser-specific to add to either, this register does not propose new fields for them; it only says which of their existing values actually fire here (see above), per AGENTS.md's instruction that a base-schema field must be a vertical anchor, quality currency, hard filter or sizing fact shared densely across the vertical, not a one-category elaboration.

| Facet in this register | In add-on today | Proposed kind under AGENTS.md | Note |
|---|---|---|---|
| `intended_liquid` | no | hard filter (categorical) | Decides the whole form factor. Mostly derivable from the listing's own type/title words (soap, sanitiser, dish, shower) rather than a stated field; proposal: `dispenser.jsonc` gets an enum `intended_liquid` [`hand_soap_or_lotion`, `dish_soap`, `hand_sanitizer`, `shower`]. |
| `dispensing_mechanism` | no | hard filter (categorical) | `manual_pump` / `foaming_pump` / `touchless_sensor`. Usually in the title ("foaming," "touch-free," "sensor"); cheap to extract. |
| `refill_system` | no | hard filter (categorical), but see note | `open_refill` / `sealed_cartridge`. Almost never stated on a listing (open is the unstated default for a plain pump bottle); this register's own research is the only source for why it matters, so mark `researched`, low backfill cost by rule (plain pump bottle with no cartridge slot in its own photos/description defaults to `open_refill`; a stated cartridge, pod or pouch name defaults to `sealed_cartridge`). |
| `capacity_ml` | no (but `sizing.capacity` already exists in supplies-standard) | sizing | Already covered by the shared `sizing.capacity` block (unit `ml` is in its enum); this register just confirms the `ml` reading applies and gives the observed typical span. No new field proposed. |
| `chamber_count` | no | hard filter (numeric) | Only fires for shower dispensers, and only there is it dense and listing-stated (in the product name). Proposal: `dispenser.jsonc` gets `chamber_count` (1–4), null for everything that is not a multi-chamber shower unit. |
| `power_source` | no | hard filter (categorical) | `none` / `disposable_batteries` / `rechargeable_battery`. Stated on touchless listings' titles or bullets ("4 AA batteries," "rechargeable"); null for manual/foaming pumps. |

Add-on fields this register considers noise for dispensers, if ever proposed by analogy with other add-ons: a `safety_certification` or `warranty` field (AGENTS.md rule 5 cuts certifications and warranty blocks outright; the one safety-relevant fact here, the contamination/evaporation risk of open refill, is a judgement about use, not a certification any dispenser carries); a `color` or `finish` field (visually derived, belongs to the image pipeline per AGENTS.md rule 1).

## Branches

- `intended_liquid` = "shower (shampoo, conditioner, body wash)" → then_ask `chamber_count`. Chain: Better Living's entire shower-dispenser line is sold and named by chamber count ("2 Chamber," "3 Chamber," "4 Chamber"), a question that simply does not arise for a single-liquid hand-soap or dish-soap pump. SUPPORTED.
- No branch for `intended_liquid` = "hand soap or lotion," "dish soap" or "hand sanitiser": each stays a single-chamber pump and nothing downstream changes because of which of the three it is; the safety-relevant difference (evaporation, poisoning risk) attaches to `hand sanitiser` directly inside that facet's and `refill_system`'s `why_experts_care`, not as a branch to a different facet.
- `dispensing_mechanism` = "touchless sensor" → then_ask `power_source`. Chain: Wikipedia's Automatic soap dispenser article states these are "often battery-powered," and both spec-sheet sources confirm it concretely (simplehuman: built-in rechargeable; Better Living TOUCHLESS XL: "4 AA batteries required"); a manual or foaming pump has no power source to ask about. SUPPORTED.
- `dispensing_mechanism` = "manual pump" or "foaming pump" → makes_irrelevant `power_source`. Trivial (no electronics in either mechanism per the Wikipedia mechanism description).
- No branch keyed to `refill_system`'s two values: the facet's own `why_experts_care` already carries the full consequence (contamination risk, evaporation risk, cost-per-use, brand lock-in) and no other facet's relevance changes depending on the answer.

## Trimmed detail

- `mechanism.plain`: the exact priming sequence ("the piston compresses the spring and upward air pressure pulls the ball upward... This process is called 'priming'") is in the Wikipedia source above verbatim; the register's 150-word cap only has room for the compressed version.
- `refill_system.why_experts_care`: the exact percentages (one-in-four / 70.2% vs 10.6% / 9.09% and 82.05%) are all in the Sources section above; the field itself states the direction and magnitude in rounder terms to stay under 80 words.
- Dry soap mills (soap graters): real, sourced ("Soap mills are common in public washrooms in Germany... About 40 grams... equivalent to 1 liter of liquid soap, providing soap for up to 400 handwashings"), kept as a mechanism family and a vocabulary entry, but not given its own facet: it is a single, rare product type with no second source on how common home models actually are in the catalog, and no listing field would distinguish it beyond the title.
- Judgement 4's full set of numbers (both the AAPCC tracker and the NPDS study's age/route/ICU breakdown) is longer than the 50-word note cap; the complete figures are in the Sources section above.

## Dropped

- A literal "dead reservoir volume" facet (how much of a pump bottle's contents the dip tube can never reach) — real and part of why the endurance-athlete seed answer led here, but no measured source was found for how much is typically unreachable, so nothing numeric went in the JSON.
- A gasket/seal material-compatibility claim for alcohol-based liquids (would a touchless sanitiser pump's internal rubber or silicone parts degrade faster than a soap pump's) — plausible, but the one manufacturer page that would have addressed it (Aptar, an OEM pump maker) 404'd, and no second source was found. Not in the register.
- ADA reach-range and operable-parts mounting heights for wall dispensers — a real commercial-fixture requirement, but the current ADA.gov standards pages for those sections would not resolve, and more to the point this is a facilities-compliance concern for the installer, not a fact a consumer buyer weighs when choosing a dispenser, so it fails the category's own "would a novice ever decide on this" test even before the sourcing problem.
- `intended_liquid` = "dish soap" was considered for its own branch (sink-mounted pumps sometimes have a longer tube to reach an under-sink reservoir bottle), but no spec sheet that would state a tube-length range could be reached (Moen and Delta's dispenser pages are JS-rendered catalogues with no spec table in the fetched HTML), so no range or branch was written for it.

## Corrections (2026-10-26)

- Added facet `dose_control` (fixed vs. adjustable dose) for touchless dispensers, confirmed on both betterliving_dispensers (TOUCHLESS XL: "a quick turn of a dial will adjust the volume of liquid dispensed") and simplehuman_sensorpump (hand-distance controls volume); re-fetched both pages to verify, now branched from `dispensing_mechanism` = touchless sensor.
- Lowered `capacity_ml.range.min` from 225 to 200 ml: re-fetched betterliving_dispensers and confirmed the CLARA Foaming Soap Dispenser Medium, a single-chamber, in-scope product on that same page, states "6.7 fl oz (200ml)".
- Reworded the `intended_liquid` split question and matching `cold_start_question` to "Hand soap, dish soap, lotion, sanitiser, or shower products?" so all four facet values are nameable from the question text.
- Softened `refill_system.why_experts_care`'s 26-fold/2-fold bacterial-transfer figures with "a figure no other cited study repeats" rather than adding an unverified second source for that exact figure.
- Added one sentence to `dispensing_mechanism.definition` stating soap mills are a separate, rare mechanism out of scope for that facet's enum, rather than adding a fourth value with no listing signal.
- Re-graded `chamber_count.source` from `listing` to `researched` (and `backfill_cost` from `none` to `low`), since the "stated in the product title" claim is confirmed only on Better Living's own site; independent retail listings were unreachable to confirm it.
- Added "Pressure-release valve" to `vocabulary`, since `dispensing_mechanism.why_experts_care` used the term without defining it.