Skip to content

Teardown #1 · claim verification

CLAIM VERIFICATION — NOT A TEST RESULT.

No product was bought, nothing was tested, and no sample exists. This is a verified record of what two companies publish on their own pages, and of what the number they sell on can and cannot tell a buyer.

What we did — and what we did not do

On 28 September 2026 we read the pages these products are sold on, and recorded exactly what they say. That is all we did.

We bought nothing. We tested nothing. We took no sample. No lens has been measured, no laboratory has run anything, and no instrument has been near these products. There is no result in this piece, and nothing in it says any product works or fails.

Every claim below is a company's own words, with the page's address, the capture date and a SHA-256 fingerprint of the redacted public copy we publish of that page. Each row gives the published copy's hash first, then the hash of the sealed original behind it — the sealed originals are not published, and none of them is released to anyone. Two things we verified ourselves: both of these products are sold on a numeric claim about blue light, and neither page gives the band that number is measured against. Ra Optics does name specific wavelengths in its own marketing copy — 420, 455, 550 and 600 nanometres — but does not attach any of them to the percentage it sells on; Swanwick's page names no wavelength at all.

The claim record

The two numeric blue-light claims, as each company publishes them, with the page, the capture date, and the SHA-256 of the redacted public copy we publish of that page.
#The claim, in the company's own wordsWhere it appearsDate capturedOur published copy, and its SHA-256
1"Sleep Swannies® use amber lenses that block 99% of the blue light that keeps your brain awake."swanwicksleep.com — homepage2026-09-28, 08:55 UTCswanwick-homepage-2026-09-28.public.htmlSHA-256, this published copy: 4a64fc29bc85201868f8101c06a14517f81338f7852a1b3976b6ed37b90de703Sealed original, not published — its SHA-256: d0eb3bedb224dbc0139f06a192938617bb14d75ea0c0153275e83e574907ffef
2"Our sunset lenses are designed for evening wear, blocking 100% of blue light and most green light…"raoptics.com — homepage, FAQ answer to "Should I get Daylight or Sunset?"2026-09-28, 08:20 UTCraoptics-homepage-2026-09-28.public.htmlSHA-256, this published copy: 558f00ba12fe6fc66a0c6da79a82e41c7931a52db872ac9fd8029bb01015db6eSealed original, not published — its SHA-256: 9958aa33735281952ac655e0275d744b9df781f219f51b48f2fe8075668580af
3Prices read from each store's own product data, same day: US$129.99 (Swanwick Sleep "Classic Black") and US$174.00 (Ra Optics "Clyde Sunset", non-prescription).Each store's own product feed, read live2026-09-28Read live from each store's own product data on 2026-09-28; no copy of either feed was kept, so there is no hash for one. Both figures also appear on the two captured store pages above.

On claim 2, that sentence sits inside the page's own FAQ markup, in a collapsed panel — a <details> element with no open attribute, so it is not shown until a reader opens it. It is in the page as served, and re-reading the page returns it in one line.

Neither page defines the band the number is measured against — no range, no weighting and no measurement conditions are given for the percentage.

The advertising evidence, and its limits

On the same date, Meta's Ad Library — the public archive of Facebook and Instagram advertising — showed Swanwick running active US ads: a keyword search for “swanwick” returned about 48 results, three of them creatives marked Active that started running on 5, 7 and 10 June 2026 — two shown as placed by Swanwick, and a third by an advertiser whose name our published copy withholds.

It shows Swanwick paying to advertise in the US on Meta, and nothing more. It does not show how much was spent (no spend figure is published for these advertisers), how many people saw anything, or anything about other platforms. We hold no equivalent capture for Ra Optics, so this section says nothing about its advertising. TikTok's public ad library has no US option, so it can say nothing about US advertising there.

  • meta-adlibrary-swanwick-us-2026-09-28.public.pngMeta Ad Library, United States, keyword “swanwick”, active ads — the capture behind the advertising section.184,026 bytes · SHA-256 8bbb6f407bf1860495c395a03c309258bc823a756f0212c88700b26e9e1d8e53
  • tiktok-library-country-list-2026-09-28.public.pngTikTok Commercial Content Library, “Ad target country” list open — the United States is not among the options.118,205 bytes · SHA-256 b2ca44262e398b5cb435eda6073fda7da75b95938058906fb81764103f9363d5

Why a percentage without a band cannot be checked

A percentage of blue light blocked means nothing on its own. Checking it needs four things, none of which either page gives for the number it sells on.

Which wavelengths. "Blue light" is not one colour: it names a stretch of spectrum, loosely 380–500 nanometres, with no agreed edge. A lens blocking 99% below 440 nm and 20% above it can be sold, truthfully, as "blocks 99% of blue light", while passing most of the light the buyer wanted blocked.

How they are counted. A flat average treats every wavelength as equal; the eye and the body's response to light do not. Two reasonable weightings of the same lens can return 99% or far less.

How it was measured. Curved lenses: the figure depends on the light's angle, which part was sampled, and whether it is one point or a surface average. Two labs can measure the same lens and disagree.

Which direction it runs. "Blocks 99%" and "lets through 1%" are one fact from opposite ends; only one is a figure a buyer can re-measure.

Without those four, no one can check the number — including us, until we choose the band and publish that choice before looking at the result.

The same specification, at a published ex-factory price

A public factory listing on a wholesale marketplace offers any buyer this specification: an acetate frame, a CR39 amber lens, the buyer's own logo, and the words "99% Blue Light Blocking" in the listing title. Published prices on 28 September 2026: US$6.39 a pair at 12–49 pieces, US$6.30 at 50–99, US$6.29 above that, minimum 12 pieces per colour. The retail pairs we recorded that day: US$129.99 and US$174.00.

Two things this is not. Not a claim about where either brand's glasses come from: we hold no invoice, no order, no shipping record and no shared model number, and we have contacted nobody. A shared specification is not a shared factory — an acetate frame and a CR39 lens are commodity parts, and "99% blue light blocking" is printed on listings and boxes throughout this category. Not a claim about anyone's costs: what a brand pays is unknown to us.

What it is: the same published specification, at 12 pieces for under US$7 a pair, against US$129.99 and US$174.00 at retail — and the "99%" on the listing and the "99%" on the box are the same undefined number.

  • wholesale-listing-2026-09-28.public.htmlWholesale-marketplace factory listing as served, 2026-09-28 12:51 UTC — the listing behind the price comparison, with the supplier masked.205,732 bytes · SHA-256 e53dfdfed4b3c2b95cf54b0a2d2ab13d0680770d6e14bceb96ec2547eb19c019

What this piece does not establish

  • No lens has been measured. No curve, no number, no measurement of any kind. This is not a test result.
  • Neither claim is verified or refuted. Real amber lenses do absorb strongly in the blue; "99%" may be defensible under a chosen band. We do not know, because we have not measured. Read as "the claim is false", it is being misread.
  • No supply chain is traced. The price gap is a price gap, not evidence that any brand buys from any factory.
  • Nothing here is about sleep. Whether these glasses change anyone's sleep or melatonin cannot be measured on a bench, and we claim nothing either way.
  • One badge is noted, not judged: the Swanwick homepage carries an "Independently validated" badge. The badge itself names no study, and it is not tied to any of the studies the page mentions elsewhere.

What one measurement would settle — and what it costs

One measurement stands between this piece and a test result: spectral transmittance.

Put one lens into a spectrophotometer and record, wavelength by wavelength from 380 to 780 nanometres, how much light gets through. That yields a curve — and a number with its band attached, because we choose the band before we look at the result.

It would settle whether the "99%" and the "100%" hold under a stated definition — and which one; whether the two lenses in a pair match; and what the lens does across the band, not just at one edge. It would not settle sleep, melatonin, or any effect on a person: it measures light.

Cost, read from five published facility cards by us on 28 September 2026 — the hourly cards are the facilities' own; the per-session figure below is our own reading of them and our own unit. Those are academic instrument-time cards, not commercial prices: the one quotation we hold, for this job, is €1,417 — the final price, with no tax added and shipping the only extra — for the full two-technique scope (Measurlabs offer M40708, 1 October 2026 — €490 per transmittance sample, €170 per ATR-FTIR sample, €97 per order).

  • Georgia Tech, MCF: US$28/hr for its UV-Vis instrument for industry, plus US$75/hr if its staff run it (≈US$103).
  • Texas A&M, MCF: US$32/hr for UV-Vis.
  • Duke, SMIF: US$26.53–US$66.33/hr for its instrument by user class.
  • CAMCOR, University of Oregon: US$169.75/hr, commercial and fully staffed — card re-read by us on 2026-10-03, and higher than the figure we recorded on 28 September 2026.
  • WVU, Shared Research Facilities: US$150/hr — the fifth card we read, and the only one here that prices FTIR rather than transmittance.
  • Add roughly US$20–40 to post the lens and have it returned — our own estimate of a flat-rate postal box each way, not a published rate.

One session: about US$75–200 — our own reading of those cards and our own estimate of how long a session runs, because the facilities publish hourly rates and not session prices. Measurlabs has now quoted: €1,417 — the final price, with no tax added and shipping the only extra — for the full two-technique scope (offer M40708, 1 October 2026). The band above is a card we read and remains a card we read; the quotation is an offer, not an agreement, and nothing has been ordered. That quotation is the whole cost of turning a record of what companies say into a record of what is true.

Right of reply

The companies named here are named only as the source of their own published words. Nothing here alleges dishonesty — an undefined number is a marketing-communication problem, not a legal finding — and any reply either company makes, including the band behind its number, will be published unedited and labelled.

The evidence pack, and how to check it

Five redacted public copies, and three documents that explain and check them. Every file below is served from this site, and every hash printed beside a copy is the hash of the bytes you get: download the file, run sha256sum on it, and it matches what this page says. None of these is a sealed original — the captures we hold unaltered are not published and are not released to anyone; the log identifies each of them by hash, labelled as not published.

  • raoptics-homepage-2026-09-28.public.htmlRa Optics homepage as served, 2026-09-28 08:20 UTC — carries the “100% of blue light” answer in its FAQ.406,257 bytes · SHA-256 558f00ba12fe6fc66a0c6da79a82e41c7931a52db872ac9fd8029bb01015db6e
  • swanwick-homepage-2026-09-28.public.htmlSwanwick Sleep homepage as served, 2026-09-28 08:55 UTC — carries the “99%” headline claim and the “Independently validated” badge.417,127 bytes · SHA-256 4a64fc29bc85201868f8101c06a14517f81338f7852a1b3976b6ed37b90de703
  • wholesale-listing-2026-09-28.public.htmlWholesale-marketplace factory listing as served, 2026-09-28 12:51 UTC — the listing behind the price comparison, with the supplier masked.205,732 bytes · SHA-256 e53dfdfed4b3c2b95cf54b0a2d2ab13d0680770d6e14bceb96ec2547eb19c019
  • meta-adlibrary-swanwick-us-2026-09-28.public.pngMeta Ad Library, United States, keyword “swanwick”, active ads — the capture behind the advertising section.184,026 bytes · SHA-256 8bbb6f407bf1860495c395a03c309258bc823a756f0212c88700b26e9e1d8e53
  • tiktok-library-country-list-2026-09-28.public.pngTikTok Commercial Content Library, “Ad target country” list open — the United States is not among the options.118,205 bytes · SHA-256 b2ca44262e398b5cb435eda6073fda7da75b95938058906fb81764103f9363d5
  • redaction-log.mdRedaction log — every removal and transformation applied to each copy, with the sealed and published SHA-256 of every file.
  • README.mdReadme — what this pack is, and how to check a copy you have downloaded.
  • public-copies.sha256Checksum file — the five published copies, in the format sha256sum -c reads.

Every copy here is a redacted derivative, not the page as it was captured: executable content, personal data and (in the listing) the supplier's identity were removed, the removal is recorded per file in the log, and every claim sentence, price and specification the record quotes reads identically in the sealed original. The rule this page is built to keep: we do not print a hash for a file we do not serve, and we do not link an evidence file that is not there — the site's build fails rather than publish either.

A reader who wants to check us has everything needed to do it and nothing we cannot back: the copy, its hash, and the log of every change made to it.

Contact

This record is published work, not something for sale, and it is not a funding ask. Two things a reader may want to do with it: commission an audit of a claim you are about to put live, or add a reply if you are one of the companies named here.

Both start at the same place. The audit is bought by the party that has to defend the claim — the founder, the agency or the retailer's own QA function — and the terms, the fee and the one thing no fee can buy are on the main page. Any reply from a company named here is published unedited and labelled, whether or not anything is bought.

For brands and retailers — enquiries, terms and fees

← Back to teardown #1 on the main page