# Astronomy toys: research notes

Date: 2026-10-25. Scope: the taxonomy leaf `supplies > play > educational-science > astronomy_toys`. This covers two different kinds of product that the leaf's own add-on enum already bundles together: (1) optical instruments sold to beginners and children — toy telescopes and beginner telescope kits — and (2) non-optical representations of the sky and solar system — planetarium/star projectors, solar-system and orrery models, constellation cards, moon-phase learning toys, astronomy puzzle models, and space experiment kits. One register, one set of splits; split 1 (`has_optics`) does the dividing the brief asked for instead of writing two registers. Real expertise and the one serious safety risk in this category both live almost entirely on the optical side, so the register is weighted there and the representational side is kept deliberately short rather than padded with invented facets.

Not in scope: astronomy binoculars (not in the add-on's `astronomy_toy_type` enum and not listed as a sibling leaf — they'd fall under a general binoculars/optics category if one exists). Overlap with siblings under the same `educational-science` parent (from `catalog.json`: `abacuses`, `flash-cards`, `science-toys` — "Science & Nature Toys"): a chemistry set, microscope kit or generic nature-exploration kit with one astronomy chapter belongs to `science-toys`; a kit whose primary selling point is space (rocket/balloon launchers themed as space missions, "build a planet" kits) is `space_experiment_kit` here. Where a listing could plausibly sit in either, this register does not try to adjudicate it — that is the sibling register's call to make when it is written.

Quote labels: **verbatim** = the sentence in quotation marks was in text I read myself, via `curl` plus a Python tag-stripping pass, or `pdftotext`. **fetched excerpt** = a WebFetch summary I did not verify against raw text (none used for any JSON number). **search excerpt** = a search-engine snippet (none used; WebSearch reported its budget exhausted at the start of this run — see "Could not reach" below — so every source here was reached by direct `curl`/WebFetch of a known URL, a brand sitemap, or a Wayback Machine snapshot).

## Seed

Seed run (`seed.py --category astronomy-toys`), recorded verbatim from the script's output:

- Example: car-tyres `treadwear_grade` — "a number that is on every listing and still needs a judgement attached, because experts know it only compares within one manufacturer's range."
- Persona: "A pastry chef, who thinks in ratios, temperatures and why a recipe fails."

1. **The astronomy-toys equivalent of treadwear grade is the magnification number on the box.** Every cheap telescope's packaging leads with a power figure ("675x!!!"), and like treadwear it needs a judgement attached: a telescope's useful magnification is capped by its aperture (60x per inch at best, per Celestron's own knowledgebase, and "twice its aperture in millimeters" per Sky & Telescope — the same rule, two independent phrasings), so the box number is reachable only by stacking a cheap eyepiece with a Barlow lens, producing an image Sky & Telescope calls "blowing smoke." Celestron's own PowerSeeker 114EQ spec sheet makes the point without meaning to: its 4mm eyepiece alone already gives 225x against a 269x useful ceiling for that aperture, and the 3x Barlow it ships with would nominally push that to 675x. **Kept** as `advertised_magnification_x` (an annotation, not a sort key, with a strong judgement attached) alongside a derived `useful_magnification_ceiling_x`, exactly mirroring the treadwear pattern of "real number, needs a cap drawn next to it."
2. **The pastry chef's ratio is the focal ratio (f-number); the thing that fails like a wrong oven temperature is the mount.** A telescope's f-number (focal length ÷ aperture) is a fixed ratio for a given scope, exactly like a dough hydration ratio, and it trades off the same way: a low f-number ("fast") gathers more light for a wider, dimmer field good for star clusters and nebulae, a high f-number ("slow") suits planets and the Moon at higher magnification (Celestron, "F-Numbers and F-Stops Explained"). What a pastry chef would notice that a typical gift-buyer would not: the thing that ruins the result isn't the ingredients (optics) but the technique (mount). Sky & Telescope's "Hobby Killers" piece makes exactly this point — a telescope with good optics on a shaky, hard-to-aim mount "can still become a hobby killer," and a wobbling tube that takes more than 3-4 seconds to settle is the tell. **Kept**: focal ratio as an annotation facet, mount type and the alt-az-vs-equatorial judgement as the register's strongest trade-off.

## Sources and what each supports

### Standards and safety bodies

**American Astronomical Society (AAS) Solar Eclipse Task Force, "Solar Filters for Optics: Telescopes, Binoculars & Cameras"** (verbatim, curl + tag-strip)
https://eclipse.aas.org/eye-safety/optics-filters
- verbatim: "Solar filters provided with inexpensive telescopes, usually designed to thread into an eyepiece at the back end of the telescope, are dangerous. If the filter is attached to the spot where you place your eye, sunlight concentrated by your optics will burn right though it."
- verbatim: "In every case, the solar filter must be attached to the front of your telescope, binoculars, or camera lens."
- Supports: the `solar_filter_position` facet, its governing-standard note, and the single highest-confidence judgement in the register. Confidence high.

**AAS Solar Eclipse Task Force, "About the ISO 12312-2 Standard for Solar Viewers"** (verbatim)
https://eclipse.aas.org/eye-safety/iso12312-2
- verbatim: "The relevant international standard for this activity is ISO 12312-2:2015, 'Eye and face protection — Sunglasses and related eyewear — Part 2: Filters for direct observation of the Sun.'"
- verbatim: ISO 12312-2 "does not apply to solar filters meant to fit over the aperture (front) of optical devices such as camera lenses, binoculars, or telescopes. But any solar filter made of material that meets the transmittance, uniformity, and quality requirements ... of ISO 12312-2 should work as an aperture filter too."
- verbatim: acceptable luminous transmittance "between 0.00004% and 0.0032%" (AAS Task Force's working range, bridging the 2015 standard and Chou et al.'s 2021 proposed revision).
- Supports: the governing standard named in `solar_filter_position`. Confidence high.

**AAS Solar Eclipse Task Force, "How Can You Tell If Your Eclipse Glasses or Handheld Solar Viewers Are Safe?"** (verbatim)
https://eclipse.aas.org/eye-safety/how-to-tell-if-viewers-are-safe
- verbatim: "There's no way to tell just by looking at them whether eclipse glasses or handheld solar viewers are genuinely safe, but it's easy to tell if they're not safe." Indoor/outdoor/glance test described.
- verbatim: "you must never look through magnifying optics while wearing eclipse glasses."
- Supports: the note that a buyer cannot self-certify a filter and should look for ISO 12312-2 conformance from the supplier, not the product's own claim. Confidence high.

### Brand knowledge base and spec sheets (Celestron — read as the leading brand's own technical material, not a listing)

**Celestron knowledgebase, "What is a Solar Filter, and How Do You Use It?"** (verbatim, curl + tag-strip)
https://www.celestron.com/blogs/knowledgebase/what-is-a-solar-filter-and-how-do-you-use-it
- verbatim: "Solar filters provided with... eyepiece solar filters... has been discontinued because of danger that the telescope's concentrated solar heat may crack the filter, leading to permanent eye damage. Under NO circumstances use one of these obsolete over-the-eyepiece solar filters. For safety's sake, destroy any examples found."
- verbatim: "All Celestron EclipSmart solar filter products feature Solar Safe filter technology and conform to the ISO safety standard, ISO 12312-2:2015(E)."
- verbatim: filters block "more than 99.999% of sunlight"; full-aperture white-light filters block "more than 99%."
- Supports: second source (independent of AAS) for the eyepiece-filter danger, naming it discontinued/obsolete; the ISO 12312-2 conformance claim as something a filter maker states; filter material types (metal-on-glass, aluminized polyester/"Mylar," black polymer). Confidence high.

**Celestron knowledgebase, "Solar Safety Warning"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/solar-safety-warning
- verbatim, telescopes: "Never look directly at the Sun with the naked eye or with an optic (unless you have the proper solar filter)... Never leave your optic unsupervised. Make sure an adult who is familiar with the correct operating procedures is with your optic at all times, especially when children are present."
- Supports: the adult-supervision note attached to the safety judgement; this is the exact warning text Celestron prints for every scope regardless of price tier, including toy-grade ones. Confidence high.

**Celestron knowledgebase, "What is magnification/power as it pertains to telescopes?"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/what-is-magnification-power-as-it-pertains-to-telescopes
- verbatim: "the maximum usable power is equal to 50-60 times the aperture of the telescope (in inches) under ideal conditions... a 60-mm... telescope scope labeled" [continues in the "how much magnification" article below]. "Most of your observing will be done with lower powers (6 to 25 times the aperture... in inches)."
- verbatim: magnification = telescope focal length ÷ eyepiece focal length; a 2x Barlow doubles whatever eyepiece it's used with.
- Supports: `advertised_magnification_x`, the derived `useful_magnification_ceiling_x` formula (upper bound of the 50-60x/inch range), and the Barlow-lens vocabulary entry. Confidence high.

**Celestron knowledgebase, "How much magnification can I use with my CPC scope, and how much is too much?"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/how-much-magnification-can-i-use-with-my-cpc-scope-and-how-much-is-too-much
- verbatim: restates the 50-60x/inch rule and the 6-25x/inch "most observing" band with a worked 279mm (11") example (550x-660x ceiling).
- Supports: second internal Celestron confirmation of the same ceiling rule, worked at a different aperture. Confidence high.

**Celestron knowledgebase, "How to Determine Which Eyepieces to Use with Your Telescope"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/how-to-determine-which-eyepieces-to-use-with-your-telescope
- verbatim: "generally defined as 60x per inch (25.4mm) of aperture" for the highest useful magnification, and "3.6x per inch (25.4mm) of aperture" for the lowest (below which the exit pupil vignettes). Worked example: AstroMaster 130EQ, 650mm focal length, 130mm aperture → highest useful 307x, lowest ≈19x.
- verbatim: assumes "a scope that has been allowed to cool to ambient air temperature, and well collimated optics" — the thermal-equilibrium and collimation caveats behind the "ideal conditions" language elsewhere.
- Supports: the lowest-useful-magnification half of the derived ceiling, and the cooldown/collimation caveat in the mechanism text. Confidence high.

**Celestron knowledgebase, "Which scope aperture is best for different kinds of objects?"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/which-scope-aperture-is-best-for-different-kinds-of-objects
- verbatim: Sun — "Apertures of 6 inches (150 mm) or less and focal lengths of 1000 to 1500 mm." Planets — "at least 5 inches (125 mm) aperture and 1250 mm focal length or more." Bright nebulae — "apertures of 5-8 inches (125 to 200 mm) with focal lengths of 1000 mm or less." Faint nebulae/galaxies — "at least 8 inches (200 to 280 mm)." Globular clusters needing resolved cores — "11 inches (280 mm) or larger."
- Supports: the `primary_interest` split's branch guidance (object type → aperture floor). Confidence high.

**Celestron knowledgebase, "Benefits of a Bigger Aperture"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/benefits-of-a-bigger-aperture
- verbatim: "The larger the telescope's aperture, the more light the telescope can gather making the image brighter, sharper, and able to produce more detail... Larger scopes also have longer focal lengths."
- Supports: the `aperture_mm` facet's `why_experts_care` and the mechanism text. Confidence high.

**Celestron knowledgebase, "F-Numbers and F-Stops Explained"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/f-numbers-and-f-stops-explained-mastering-focal-ratios-for-better-astronomy-and-astrophotography
- verbatim: "To calculate focal ratio, divide an optic's focal length by its aperture... A lower f-number means more light enters the system... Low f-number systems are often called 'fast'... high f-number systems... 'slow.'"
- verbatim: "the aperture of the human eye typically ranges from 6 to 7 millimeters in young individuals" when fully dark-adapted, "4 to 6 millimeters" with age.
- Supports: `focal_ratio` facet and the 7mm pupil assumption behind the light-gathering-power derived metric (cross-checked below). Confidence high.

**Celestron knowledgebase, "Finderscopes 101"** (verbatim)
https://www.celestron.com/blogs/knowledgebase/finderscopes-101-how-to-choose-and-care-for-your-telescopes-key-tool
- verbatim: describes red-dot (StarPointer), straight-through optical, and Right-Angle Correct Image (RACI) finderscopes, and that larger telescopes (6"+) find a straight-through 50mm finder "challenging to use" near the zenith.
- Supports: `finderscope_type` facet values and note. Confidence high.

**Celestron product page, FirstScope Telescope, spec table** (verbatim, curl + tag-strip of the page's own spec table)
https://www.celestron.com/products/firstscope-telescope
- verbatim table: Optical Design "Newtonian Reflector | Spherical Primary Mirror"; Aperture "76mm (2.99")"; Focal Length "300mm (12")"; Focal Ratio "f/3.95"; Highest Useful Magnification "180x"; Lowest Useful Magnification "11x"; Limiting Stellar Magnitude "11.9"; Light Gathering Power (Compared to human eye) "118x".
- Cross-check: 180x ÷ 2.99 in = 60.2x/inch, at the top of Celestron's own 50-60x/inch rule. (76mm/7mm)² = 117.98 ≈ "118x" printed — confirms the 7mm dark-adapted pupil assumption behind Light Gathering Power to within rounding.
- Page tags (Shopify product metadata, not prose): `filter-aperture::3_inches (76-100 mm)`, `filter-mount::Manual_Alt-Azimuth`, `filter-Optical Design::Newtonian Reflector`, `young-beginner` — this is Celestron's own facet vocabulary for the product line, which is what this register's `aperture_mm` buckets, `mount_type` and `optical_design` values are built to match.
- Supports: the spec-sheet field list used for `listing_field_hint` on every telescope facet, and the light-gathering-power cross-check. Confidence high.

**Celestron product page, PowerSeeker 114EQ Telescope, spec table + description** (verbatim)
https://www.celestron.com/products/powerseeker-114eq-telescope
- verbatim table: Aperture "114mm (4.49")"; Focal Length "900mm (35")"; Focal Ratio "f/7.89"; Magnification of Eyepiece 2 (4mm) "225x"; Barlow Lens "3x"; Highest Useful Magnification "269x"; Light Gathering Power "265x".
- Description (verbatim): "Manual German Equatorial telescope... two slow motion control knobs... in both Right Ascension and Declination axes."
- Cross-check: 269x ÷ 4.49 in = 59.9x/inch (same ceiling rule). (114/7)² = 265.2 ≈ "265x" printed — second independent confirmation of the light-gathering formula. The 4mm eyepiece alone (225x) already sits within 16% of the useful ceiling (269x); the included 3x Barlow would nominally multiply the 4mm eyepiece's 225x to 675x, 2.5 times the scope's own stated ceiling — Celestron's spec sheet does not draw this comparison itself; it is this register's reading of two numbers on the same page.
- Supports: the worked magnification-inflation example in the Seed section and the `advertised_magnification_x` judgement; `mount_type` value "manual_equatorial" and its listing wording. Confidence high for the printed numbers; the inflation comparison is this register's own arithmetic on Celestron's numbers, not a claim Celestron makes.

### Specialist reviewers and enthusiast community

**Sky & Telescope, "Hobby Killers: What Telescopes Not to Buy," Jerry Oltion, Nov 26 2019** (verbatim, via Wayback Machine snapshot — live site blocks non-browser fetches)
https://web.archive.org/web/20250830031752/https://skyandtelescope.org/astronomy-equipment/choosing-astronomy-equipment/hobby-killers-what-telescopes-not-to-buy/
- verbatim: "Anything below about 3 inches (76 mm) in aperture is likely to be too small to provide a pleasing view of anything more than the Moon and a few bright star clusters."
- verbatim: "There's one design you should avoid at all costs: the 'Bird-Jones' reflector... uses a spherical primary, and a fast one at that... The view through every Bird-Jones scope I've ever looked through has been uniformly awful."
- verbatim: "Avoid scopes that use 0.965″ eyepieces... which are almost always garbage and are hard to replace with better-quality designs."
- verbatim: "A good telescope lets you reach about 50× per inch of aperture before the fuzziness gets out of hand... Does the ad... tout the telescope's magnification? If it's a 3″ aperture and they're advertising... '675×!!!' they're blowing smoke."
- verbatim: "An equatorial mount is a poor mount for a beginner... Altitude-Azimuth (alt-az) mounts are the way to go... For reflectors, the Dobsonian mount is the way to go... If you tap the end of the telescope tube and it takes more than 3 or 4 seconds to settle down, you don't want that mount."
- The article's own reader comments (dated Nov 2019-Jan 2021, read on the same page) push back on the equatorial-mount claim at length (five commenters defending equatorial mounts for their built-in tracking). Not used as a source for any fact, but used as direct evidence that practitioners disagree, recorded in the `mount_type` judgement's `sources_disagree: true`.
- Supports: the 76mm aperture floor, the Bird-Jones and 0.965" warnings, the 50x/inch ceiling (second source, independent of Celestron), and the alt-az/Dobsonian-over-equatorial judgement. Confidence high for the specialist's claims; the mount disagreement is directly observed, not inferred.

**Sky & Telescope, "How to Choose a Telescope," Adrian R. Ashford, April 16 2014** (verbatim, via Wayback Machine)
https://web.archive.org/web/20200801085445/https://skyandtelescope.org/astronomy-equipment/how-to-choose-a-telescope/
- verbatim: "the maximum practical power... 50 times your telescope's aperture in inches, or twice its aperture in millimeters" — the same ceiling as Celestron's 50-60x/inch, phrased independently (50x/in here vs 50-60x/in at Celestron; 2x/mm = 50.8x/in, consistent).
- verbatim: "if you see a small, 60-mm department-store telescope scope labeled as delivering '300 power!!!', you'll know it's advertising hype."
- verbatim: three telescope classes — refractor, reflector (Newtonian/Dobsonian), catadioptric — with their trade-offs (refractor: rugged, expensive per inch of aperture, long tube; Newtonian: best aperture per dollar, needs periodic collimation; Dobsonian: "ultimate in observer convenience for casual viewing").
- Supports: second independent source for the magnification ceiling and the deceptive-box-number pattern; `optical_design` facet's values and trade-offs; mechanism families. Confidence high.

**Cloudy Nights, "Seven Common Beginner's Mistakes," Ed Ting, originally Apr 1 2021, revisited Apr 19 2023** (verbatim, via Wayback Machine)
https://web.archive.org/web/20250316195155/https://www.cloudynights.com/articles/cat/astro-gear-today/more/beginners-guides/seven-common-beginner%E2%80%99s-mistakes-r4676
- verbatim: "As a rule, avoid any telescope sold in a department store... Almost all telescopes under about $150 USD fall under this category."
- verbatim, on Barlow lenses: "many — if not most — beginners abuse this device. Excessive high power often distorts the image to the point where you cannot see anything... low power is your friend."
- verbatim, on equatorial mounts: "An equatorial mount tracks the motion of the sky by itself... However, some effort is required... If you place your equatorial mount in a random direction, it will actually work against you."
- Supports: second, independent enthusiast-community source for the under-~$150 quality-tier judgement and the Barlow-abuse pattern, and a second voice (alongside Oltion) on equatorial mounts being harder for a beginner, even while acknowledging their real tracking benefit. Confidence high.

### Scope and schema context (not cited for a number)

**`zamana-expertise` add-on schema**, `supplies/play/educational-science/add-ons/astronomy_toy.jsonc` — defines the `astronomy_toy_type` enum this register's scope and split 1 are built against. **`educational-science-standard.jsonc`** — empty shared block. **`supplies-standard.jsonc`** — generic `sizing` and `high_signal_markdown` blocks that already carry piece counts, dimensions and build-quality prose, so this register does not duplicate them as facets. **`AGENTS.md`** — field-kind vocabulary (vertical anchor, category quality currency, hard filter, sizing) used in the "Against the existing add-on schema" table below.

### Could not reach

WebSearch reported its 200-call session budget exhausted before any query for this run returned results, so every source above was reached by direct `curl`/WebFetch of a known URL (AAS, Celestron, or a brand sitemap) or a Wayback Machine snapshot, never a search engine. Blocked or unreachable: `reddit.com/r/telescopes` wiki (requires login even via old.reddit.com, returned only "Welcome to Reddit"); `cloudynights.com` live site (403, Cloudflare challenge — reached only via Wayback); `unclemilton.com` (403); `segatoys-homestar.jp` (connection refused); NASA's `solarsystem.nasa.gov` scale-model page (loaded only site navigation chrome, no article body, likely client-rendered); the Internet Archive's CDX search API itself went down mid-session ("Temporarily Offline") when I tried to find a second source for a planetarium-projector or orrery-scale-accuracy claim specific to toys — I stopped looking rather than guess, and the representational side of the register is shorter as a result (see "Dropped" below).

## Data points used for derived metrics

- `useful_magnification_ceiling_x`: 50-60x per inch of aperture (Celestron, three articles, worked at three apertures: 76mm→180x printed = 60.2x/in; 114mm→269x printed = 59.9x/in; 130mm→307x computed = 59.98x/in); "50 times... in inches, or twice... in millimeters" (Sky & Telescope, independent phrasing, 2x/mm = 50.8x/in). The register uses the Celestron range (50-60x/in) since it is internally consistent across three of the brand's own spec sheets.
- `light_gathering_power_x`: (aperture_mm / 7)² — 7mm dark-adapted pupil (Celestron, F-Numbers article, "6 to 7 millimeters in young individuals"). Checked against two Celestron spec-sheet values it did not state a formula for: 76mm → 117.98 ≈ printed "118x"; 114mm → 265.2 ≈ printed "265x". Both within rounding.
- `magnification_from_eyepiece_x`: telescope focal length (mm) ÷ eyepiece focal length (mm) (Celestron, "What is magnification/power," worked examples at 900mm/20mm=45x and 1000mm/20mm=50x, both verified by the article's own division).

## Against the existing add-on schema

`zamana-expertise/data/schema-parts/supplies/play/educational-science/add-ons/astronomy_toy.jsonc` currently holds exactly one field: `astronomy_toy_type`, a 12-value enum. Per `AGENTS.md`'s field-kind vocabulary (vertical anchor / category quality currency / hard filter / sizing), proposals only, nothing edited:

- `astronomy_toy_type` is doing the job of a **vertical anchor** well — it is exactly the split-1 fork this register uses (`has_optics`) collapsed into one enum, and the register's `has_optics`, `primary_interest` and `has_goto` facets do not need to duplicate it. Keep it as-is.
- The add-on carries no **category quality currency** for the optical half of the category at all. `aperture_mm` is the one number this register would commit to backfilling for every `toy_telescope`/`beginner_telescope_kit` row — it is the single spec every brand page leads with (Celestron's own tags key on it: `filter-aperture::3_inches (76-100 mm)`), and it is the number that actually predicts what the instrument can show. Propose adding it as a **numeric hard filter** (buyers filter and sort on it directly), not as free text.
- No **hard filter** exists for `mount_type` (manual alt-azimuth, Dobsonian, manual equatorial, computerized/GoTo) even though Celestron's own product tags carry this as a first-class facet (`filter-mount::Manual_Alt-Azimuth`, `filter-mount::Manual_German_Equatorial`). Propose a 4-value enum.
- `solar_filter_position` is the one **safety-governed judgement** this category needs and the add-on has nothing like it. Given the acute, irreversible-injury risk documented by AAS and Celestron, this register proposes it as a categorical hard filter (`none_included` / `full_aperture_filter_included` / `eyepiece_filter_obsolete_dangerous` / `not_applicable`), not left to prose, so a listing that ships the obsolete eyepiece-cap filter can be excluded outright rather than merely annotated.
- `focal_ratio`, `optical_design`, `finderscope_type`, `eyepiece_barrel_size` and `advertised_magnification_x` are all candidates for **category quality currency** on telescopes specifically but this register would not commit to backfilling all five for every row — see each facet's `backfill_cost`. `focal_ratio` and `optical_design` are usually on the brand spec page and cheap to get; `eyepiece_barrel_size` is the one genuinely hard to find (almost never stated outright) and this register marks it `researched`/`medium` rather than pretending it is on the listing.
- Nothing in the add-on is noise by the dead-on-sight list in `AGENTS.md` rule 5 — a 12-value enum carries no certifications, warranty blocks or curated feature vocabularies to cut. The gap is entirely additive.
- The generic `sizing` block in `supplies-standard.jsonc` already covers piece counts (for puzzle models, card sets) and dimensions; this register does not propose a duplicate facet for either.

## Branches

- `has_optics` = "look through" → ask `primary_interest`, then (per its branch) `aperture_mm` or `focal_ratio`; also ask `mount_type`, `has_goto`, `eyepiece_barrel_size`, `finderscope_type`, `solar_filter_position`. Chain: these facets are properties of an optical instrument only — a projector, model, card set or puzzle has no aperture, mount, eyepiece or finderscope, and the Sun-viewing risk is specific to looking through concentrating optics (AAS: "sunlight concentrated by your optics will burn right though it"). SUPPORTED.
- `has_optics` = "shows_displays" → ask `image_source`; makes irrelevant `aperture_mm`, `focal_ratio`, `mount_type`, `has_goto`, `eyepiece_barrel_size`, `finderscope_type`, `advertised_magnification_x`, `solar_filter_position`, `optical_design`, `primary_interest`. Chain: same as above, inverted — none of the optical facets have a value for a non-optical product. SUPPORTED. (`scale_claim`, mentioned in an earlier plan for this branch, was dropped before writing `category.json`; see the 2026-10-25 correction below.)
- `primary_interest` = "faint_deep_sky" → ask `aperture_mm`. Chain: Celestron's by-object aperture guide states a floor directly for this leg, but the floor differs by object: "at least 8 inches (200 mm)" for faint nebulae (no stated ceiling) versus "8-11 inches (200 to 280 mm)" for galaxies specifically. The facet's `why_experts_care` was corrected 2026-10-25 to carry both figures instead of only the galaxy-specific range, so the branch's guidance now matches the value's own scope (nebulae + galaxies). SUPPORTED, value-scope corrected.
- `primary_interest` = "moon_and_planets" → ask `focal_ratio` (longer focal length / higher f-number favoured for high-magnification planetary views, per Celestron's by-object guide and the F-Numbers article's "slow systems... ideal for lunar and planetary"); aperture matters less below Celestron's stated 125mm floor for planets. Chain: two Celestron articles state the same preference independently. SUPPORTED.
- `primary_interest` = "star_clusters_and_bright_nebulae" → ask `focal_ratio` (short focal length / low f-number favoured for a wide, bright field, per the by-object guide's bright-nebulae band and the F-Numbers article's "lower f-numbers... desirable for gathering as much light as possible... wide field of view"). Chain: same two-article pairing as above, opposite end of the same rule. SUPPORTED.
- `has_goto` = "yes" → **(added 2026-10-25)** `makes_irrelevant: ["mount_type"]`. Chain: `mount_type` is unconditionally queued by `has_optics`'s `look_through` branch regardless of `has_goto`'s answer, so without this leg a GoTo buyer was still asked (or filtered on) a mount question the system already knows the answer to (`computerized_goto`). `has_goto` = "no" → ask `mount_type` directly, carrying the alt-az/Dobsonian-over-equatorial judgement. Chain: Sky & Telescope's "Hobby Killers" piece discusses computerized mounts and manual alt-az/Dobsonian mounts as the two live choices once "equatorial" is set aside for beginners. Both legs SUPPORTED; the missing `yes` leg was the blind check's finding 3, fixed as above.

No branch leg needed a two-step chain through facts not already stated in the register or above; none were dropped for failing to trace.

## Corrections after the blind check (2026-10-25)

Checker report: `check.md`. Eight required fixes, applied as follows. One source re-verified live against the checker's quote (AAS ISO 12312-2 page, `curl`/WebFetch, output matched `check.md` and the original notes above exactly — no new source needed for the others, since the checker's own findings already quote the pages verbatim and this register's citations already cover them).

1. `magnification_from_eyepiece_x` (finding 1, most severe): rewrote the formula from `focal_length_mm / eyepiece_focal_length_mm` (neither term a facet in this register) to `(aperture_mm * focal_ratio) / eyepiece_focal_length_mm`, using the identity focal_length_mm = aperture_mm × focal_ratio, exactly as the checker's suggested fix proposed. `depends_on` now reads `["aperture_mm", "focal_ratio"]`, matching the formula. `eyepiece_focal_length_mm` stays as a plain-arithmetic input, not a facet — it is a property of whichever eyepiece is in use, not of the telescope itself, so it varies per calculation rather than per product; the schema's formula field explicitly allows "facet ids or listing fields" for this reason.
2. `solar_filter_position.why_experts_care` (finding 2, close second): re-fetched `eclipse.aas.org/eye-safety/iso12312-2` live and confirmed verbatim that the standard "does _not_ apply to solar filters meant to fit over the aperture (front)... But any solar filter made of material that meets the transmittance, uniformity, and quality requirements... should work as an aperture filter too." Reworded the field so ISO 12312-2 is credited with the filter-material/transmittance standard, and the front-mount position rule is attributed to AAS's own guidance alone, not to ISO 12312-2.
3. `has_goto` split (finding 3): added a `yes` branch with `makes_irrelevant: ["mount_type"]`. Without it, a GoTo buyer had nothing marking `mount_type` irrelevant even though `has_optics`'s `look_through` branch queues it unconditionally. See the "Branches" trace above.
4. All four `judgements` (finding 4): added an `evidence` array to each, naming the reference ids already cited in prose in this file for that claim (magnification ceiling → `celestron_magnification_kb`, `skytel_choose`; solar filter position → `aas_optics_filters`, `celestron_solar_filter_kb`; mount ease → `skytel_hobbykillers`, `cloudynights_mistakes`; department-store tier → `skytel_hobbykillers`, `cloudynights_mistakes`). No new sourcing needed; this was a transcription gap between `notes.md` and `category.json`.
5. `advertised_magnification_x.backfill_cost` (finding 5): `none` → `low`, matching the other listing-sourced spec facets (`aperture_mm`, `focal_ratio`, `mount_type`, `optical_design`, `finderscope_type`). Added a short `backfill_note` for consistency with those facets, since it previously had none.
6. `eyepiece_barrel_size.backfill_cost` (finding 6): `medium` → `high`. The facet's own note already describes a manual per-listing lookup limited to "top sellers only" — the register's own signal elsewhere for `high`, not `medium`. Note text left unchanged; only the cost tier moved.
7. `primary_interest.why_experts_care` (finding 7): the field stated a single "200 to 280mm for faint galaxies" range, correct on its own terms, but applied under a value (`faint_deep_sky`) whose definition also covers nebulae, for which Celestron's source gives only a floor ("at least 8 inches (200mm)"), no ceiling. Reworded to give both figures separately — "200mm or more for faint nebulae, 200 to 280mm for galaxies" — rather than softening to one blended number, since the two sub-cases have genuinely different source-stated shapes (floor vs. range) and collapsing them would lose that distinction. Still single-sourced to `celestron_by_object`; no second source for the by-object aperture floors was found this pass (same gap the checker noted, not newly introduced).
8. `image_source` cold-start question (finding 8): both the split's `question` and the facet's `cold_start_question` changed from "Projects light, a model, or flat cards?" to "Shows the sky by projecting light, a model, or flat cards?" — adds a two-word anchor ("Shows the sky") so the question passes the generic test without depending on surrounding UI context. 11 words, still under the 12-word cap.

No findings rejected. All eight required fixes were applied as the checker described or as its own suggested fix proposed. `validate.py`, `render.py` and `viewer.py` all ran clean after these changes (exit 0 each); no new validator problems were introduced.

## Trimmed detail

- `advertised_magnification_x`: the 675x figure in the Seed section is this register's own arithmetic (225x from the PowerSeeker 114EQ's 4mm eyepiece × the included 3x Barlow), not a number either source states as a product of the other — Celestron's spec sheet lists the 225x and the 3x Barlow as separate line items and does not multiply them. Flagged here so the number is never read as a quoted Celestron claim.
- `useful_magnification_ceiling_x`: the lowest-useful-magnification half of the rule (3.6x/inch, below which the exit pupil vignettes) is sourced only to one Celestron article; it did not make it into the derived metric (which focuses on the upper, "box lies" ceiling that buyers actually meet) but is kept here as background for the exit-pupil vocabulary entry.
- `mount_type`: the Sky & Telescope "3 or 4 seconds to settle" wobble test is a specialist's rule of thumb, not a measured standard, and is kept in the facet's note rather than promoted into a number in the JSON.
- `optical_design`: the Bird-Jones warning (Sky & Telescope, single source) is kept as a `backfill_note` under `optical_design` rather than its own facet or a flat "avoid" rule, since no second source names the design and this register does not have a reliable way to tell a Bird-Jones scope from its spec sheet alone (it is a construction detail, not a stated spec).

## Dropped

- A `sky_accuracy_basis` facet for planetarium/star projectors (does the projected pattern map to real constellations and star positions, versus a decorative, non-mapped light pattern) — this is a real distinction enthusiasts draw (e.g., disc-and-bulb "home planetarium" projectors versus decorative LED "galaxy light" night-lights), but every source I tried for a second citation specific to toy/consumer star projectors was unreachable this run (Uncle Milton 403, Sega Toys connection refused, and the Internet Archive's own search API went down mid-session). Rather than assert it on general astronomy knowledge alone, I dropped it. If a future run can reach a planetarium-hobbyist or toy-review source on this, it is worth adding back.
- A `power_source` facet for the representational side (battery/USB/plug-in) — this is true of nearly any electronic toy and fails the generic test (it would read unchanged for a nightlight, a walkie-talkie or a toy drill), so it is left out as not category-specific expertise, even though it is a real buyer concern covered by the generic `sizing`/listing fields elsewhere.
- An age-grading facet — ASTM F963/CPSIA age labels are a legal requirement on the package, not something this register's sources add expertise to beyond what is already printed on the box, and per the image-pipeline rule in `AGENTS.md` ("visually-derived fields stay out of schemas") a box-printed age range is exactly the kind of fact the existing pipeline already handles outside this register.
- A specific "Bird-Jones = avoid" value inside `optical_design`'s enum — see Trimmed detail above; kept as a note, not a selectable value, for lack of a second source and a reliable way to detect it from spec-sheet data alone.

## Corrections

First authoring pass; no blind content check has run against this register yet.

## Corrections (2026-10-25, verification and build pass)

This folder held only `notes.md` from an interrupted earlier run (no `category.json`, `category.md` or `viewer.html` yet). Before writing the register I re-fetched and re-checked every load-bearing claim above against raw text (curl + a Python tag-strip pass for live pages, the same for two Wayback snapshots) rather than trusting the prior pass's citations on faith: AAS eye-safety page (solar-filter danger and front-mount quotes), AAS ISO 12312-2 page (the 0.00004% to 0.0032% transmittance range and the "does not apply to... front... aperture" line), the Celestron FirstScope and PowerSeeker 114EQ spec tables (every number in both tables, including the PowerSeeker's "Barlow Lens: 3x" line, which sits inside the product's own spec table and is distinct from an "Omni 2x Barlow Lens" accessory listed elsewhere on the same page as a separate add-on product — checked specifically because a careless read could have conflated the two), Sky & Telescope's "Hobby Killers" (76mm floor, Bird-Jones, 0.965in, "675×!!!", the 3-4 second mount-settle test) and Cloudy Nights' "Seven Common Beginner's Mistakes" (the ~$150 department-store line, Barlow abuse, equatorial aiming). All matched the prior notes exactly; nothing was corrected.

Two changes from the prior draft's plan, made while writing `category.json`:

- Dropped the planned `scale_claim` facet (whether a solar-system/orrery model states it is to scale) that the prior draft's "Branches" section mentioned adding for `solar_system_model`/`orrery_model`. It was never actually sourced (the prior draft's own "Dropped" section already pulled the closely related `sky_accuracy_basis` facet for the same reason — every source attempt for a toy/consumer source on this was unreachable), and on reflection it also fails the generic test: "is this model to scale" is a stock disclaimer on model trains, dollhouses and die-cast cars too, not astronomy-specific expertise. The `image_source` split (projected_light / physical_model / flat_cards_or_puzzle) is the only facet left on the representational side, which keeps that half of the register deliberately short rather than padded.
- Simplified the register to 4 splits (`has_optics`, `primary_interest`, `has_goto`, `image_source`) rather than folding `has_goto` into a `mount_type` split. `has_goto` is boolean and genuinely novice-answerable ("does it find things for you, or do you aim it yourself"); `mount_type`'s four-way categorical (alt-az/Dobsonian/equatorial/GoTo) is not something a novice can answer unprompted, so it stays a filter/annotation facet asked only once `has_goto` is "no" — matching the lamps/power-bank convention of keeping the split facet simpler than the full filter facet it feeds into.

Validator caveat: `mechanism.plain` first came in at 154 words against the 150-word cap; trimmed two clauses to fit, no content lost beyond redundant words ("everything else is built around that one number" → "...around that number", "represent the sky rather than gather its light" phrasing tightened). `primary_interest` first had `source: researched` with `backfill_cost: none`, which the validator correctly flagged as inconsistent (researched implies something to find); changed to `source: derived` to match `has_optics` and `image_source`, since all three are buyer-preference or enum-derived facets with nothing to backfill, not facts that need research. Visual-grammar `origin` fields were rewritten from full sentences to short noun phrases after checking the lamps/power-bank registers' convention (the renderer prefixes "Used today by " onto this field, so a full sentence reads as a run-on). Validate, render and viewer all ran clean with no script misbehavior to report.

Not reached (confirmed again this pass, same as the prior draft): WebSearch's session budget was exhausted before this run's own first query, so research for this register also ran entirely through direct `curl`/WebFetch of known AAS, Celestron and Wayback Machine URLs, never a search engine. No blind content check has run against this register yet.