--- slug: the-colours-the-dog-keeps passed: 2026-10-14 by: the assay line (research/assay-line/cited-1014b/), Codex fixer directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex; fix-check repair by Codex, 2026-10-14 checked: 88 confirmed: 71 wrong: 17 unverifiable: 0 observed: 0 stale: 0 --- This pass records the 17 directing-triage items at f5e5f84f34 after independently re-reading their evidence and executing the authored page's code. It retains the 71 CONFIRMED readings from the Codex N2 checker, with their stated scope and provenance; those are not represented as 71 new independent source checks by this fixer. The 17 findings comprise 14 WRONG and 3 MINOR: 15 findings are fixed, and X14 and X15 remain open for uncorrected copies in There / Here; the wrong count includes the MINOR lines. All 17 items are corrected in this layer's authored files. Twelve dated correction lines cover the 14 WRONG claims, with the three absent-cone copies sharing one line. The MINOR fixes have no page correction line. This record covers the supplied triage and the checker's confirmed claims, not every remaining N2 observation or uncertainty. ## Claims - MINOR | [X1] From the two real cone curves you can compute exactly which colours collapse together for a dog, and which it keeps. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | Re-read the page model and Neitz pp. 121-123: cone templates, display spectra, brightness and reference white are assumptions or conditions. Qualified the introduction and metadata as model predictions. - WRONG | [X2] Missing the long-wave "red" cone you have, a dog can't separate red from green: | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | Re-read Siniscalchi Introduction and Neitz p. 122: the retained long/medium-wave cone peaks near 555 nm. Corrected the two-cone explanation in the page, research notes and verifier. - WRONG | [X3] that is a behavioural and electrophysiological measurement (Neitz, Geist & Jacobs 1989), corroborated since. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | Re-read Neitz pp. 119-120: the study trained panel choices for food rewards. Its reference to electrophysiology describes earlier work. The page now attributes its 429/555 nm estimates to behavioural experiments. - WRONG | [X4] With only two cones, colour vision is one-dimensional in chroma: every light reduces to a single S:L ratio, so any two lights sharing that ratio are physically indistinguishable to the dog. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf ; git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | Re-read Neitz p. 121: the neutral-point test controls brightness separately from colour. Equal ratios such as (0.1, 0.2) and (0.2, 0.4) do not give equal amplitudes. The page now separates chromaticity from brightness; verifier labels no longer call a model gap a measured confusion threshold. - WRONG | [X5] The red-green collapse and the existence of a neutral point are exact consequences of having no long-wave cone | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | Same retained-cone evidence as X2. Replaced the no-long-wave explanation with the responses of short-wave and long/medium-wave cones. - WRONG | [X6] and ageing lenses yellow and shift sensitivity. | https://comparative-cognition-and-behavior-reviews.org/wp/wp-content/uploads/2021/04/CCBR-vol15-2020-pp001-044-Barber.pdf ; https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4027392/fullTextXML | Re-read Barber p. 7 and Supplement 2: age-related yellow-pigment accumulation is not established for dogs and is said not to seem to occur in them. Douglas and Jeffery age comparisons concern bats, rats, mice and meerkats. Removed the canine yellowing claim without claiming that canine vision never changes with age. - WRONG | [X7] and dogs out-resolve us in dim light; | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0188557 | Re-read Lind Results and Discussion: at 0.0087 cd/m², dog acuity was 1.8-3.5 cpd and human acuity was 5.9-9.9 cpd. Removed the universal dog resolution advantage; the correction names the tested light condition and separates sensitivity from acuity. - WRONG | [X8] The ~480 nm neutral point and the red-green confusion are corroborated by | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML | Re-read Siniscalchi Introduction, Methods and Discussion. Modified Ishihara cat animations test red-green confusion. Full text has no 480 or neutral point; positive controls find Ishihara and deuteranopia. The page and research notes now attribute the neutral-point measurement to Neitz. - WRONG | [X9] and reviewed in Byosiere et al., "What do dogs see?" Psychon. Bull. Rev. 2018 | https://link.springer.com/article/10.3758/s13423-017-1404-7 | Re-read Byosiere Color vision, PDF pp. 5-6, and searched the full extracted text: colour confusion is discussed, but neither 480 nor neutral point occurs. Restricted the attribution to canine colour vision and confusion, leaving the linked title unchanged. - WRONG | [X10] Now try blue: it jumps to the far end. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | Executed the original inline script: blue #2b6fe0 puts the pin at 51.35706548506539%, red at 3.4124363517233003%, and green at 6.563371687985822%. Replaced far end with well away from red and green. - WRONG | [X11] what the dog sees | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | Executed the original inline script: #000000 gives S=L=0 but outputs #9c9024 and a yellow caption. Zero now returns black, an undefined chromatic coordinate and a hidden hue marker; ordinary preset controls still work. - MINOR | [X12] the script rebuilds the cone curves from scratch and recomputes every figure above: | git show f5e5f84f34:research/the-colours-the-dog-keeps/verify-the-colours-the-dog-keeps.mjs | Read the complete independent verifier. It computes table quantities and six primary weights, not the chosen swatch or current slider. Narrowed the instruction, research notes and verifier overview; no page correction line for this MINOR. - MINOR | [X13] From those two real cone curves you can compute exactly which colours a dog confuses. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | Same model-scope evidence as X1. Qualified the markdown dek and repeated metadata copy with the stated model, matched brightness and reference white; no page correction line for this MINOR. - WRONG | [X14] Here, the dog's two-cone system is missing the long-wave cone, so red and green that you see as distinct collapse to one hue. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | Same retained-cone evidence as X2. Corrected the authored markdown relation; it now names short-wave and long/medium-wave cones. The unannotated historical copies in public/strata/there-here/index.html:250 and research/lineage-edge-grammar/page.json:1, notes[1300], remain open. - WRONG | [X15] Both run the same cone-and-CIE machinery, pointed at different retinas. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html ; git show f5e5f84f34:public/strata/there-is-no-magenta/index.html | Read both implementations. This page integrates a pigment template against Gaussian display primaries and uses wlToHex(); the magenta page embeds CIE matching data and xyzToRGB(). Corrected the markdown relation and regenerated this layer's nearby teaser from that source. The unannotated historical copies in public/strata/there-here/index.html:250 and research/lineage-edge-grammar/page.json:1, notes[1300], remain open. - WRONG | [X16] At 480 nm the cone ratio matches white | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | Executed updateWl(): at 480 nm the fraction is 0.423 and at 487 nm it is 0.283, against white 0.347. The code tests proximity within 4 nm of 483.7, not equality. All messages in that branch now say near the model neutral point. - WRONG | [X17] (added) grey (neutral) | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | Executed the neutral calculation: whiteHue=0.3472918809928639. The original CSS put grey at 50%. The gradient stop and label now share a CSS position set from whiteHue*100, about 34.729%. - CONFIRMED | Your dog is not colourblind, and it does not see the world in grey. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [not-grey] Retained N2 reading: Neitz establishes dichromatic colour vision. In this context not colourblind means not monochromatic, not human trichromacy. - CONFIRMED | It is a dichromat | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [dichromat] Retained N2 reading: Neitz establishes dichromatic colour vision. In this context not colourblind means not monochromatic, not human trichromacy. - CONFIRMED | two kinds of colour cone instead of your three | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4027392/fullTextXML ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [two-versus-three] Retained N2 reading: Douglas and Jeffery describe three human cone pigments; Neitz supplies the two canine classes where the claim compares both species. - CONFIRMED | A human retina has three cone pigments; | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC4027392/fullTextXML | [human-three] Retained N2 reading: Douglas and Jeffery describe three human cone pigments; Neitz supplies the two canine classes where the claim compares both species. - CONFIRMED | a dog has two, with measured peaks at 429 nm (a blue cone) and 555 nm (a yellow-green cone). | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML | [dog-peaks] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | the whole warm half of your spectrum drives its two cones in nearly the same ratio, so it folds into one hue. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [warm-ratio] Retained N2 reading: Neitz reports rapidly declining wavelength discrimination away from 480 nm and effectively absent discrimination above about 520 nm in the reported tests. The page model likewise has a very small short-cone fraction at long wavelengths. - CONFIRMED | The dog's whole sense of colour rides on one number: the fraction of the signal coming from the blue cone, S/(S+L). | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [whole-sense] Retained N2 reading: N2 read this as a chromatic coordinate with brightness separate. Neitz p. 121 requires brightness control. The repair makes that scope explicit under X4. - CONFIRMED | Two lights with the same fraction are the same colour to a dog, however different they look to you. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [same-ratio] Retained N2 reading: N2 read this as a chromatic coordinate with brightness separate. Neitz p. 121 requires brightness control. The repair makes that scope explicit under X4. - CONFIRMED | right where Neitz measured ~480 nm | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [neitz-neutral] Retained N2 reading: Neitz reports an approximately 480 nm match against a 4800 K comparison with brightness controlled. The page model uses equal-energy white; the repair makes brightness and reference-white conditions explicit. - CONFIRMED | whose fraction equals that of plain white light. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [white-ratio] Retained N2 reading: Neitz reports an approximately 480 nm match against a 4800 K comparison with brightness controlled. The page model uses equal-energy white; the repair makes brightness and reference-white conditions explicit. - CONFIRMED | A dog cannot tell that wavelength from grey. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [grey-confusion] Retained N2 reading: Neitz reports an approximately 480 nm match against a 4800 K comparison with brightness controlled. The page model uses equal-energy white; the repair makes brightness and reference-white conditions explicit. - CONFIRMED | Blue is the colour a dog keeps. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [blue-keeps] Retained N2 reading: The cited colour-discrimination experiments support the dichromatic interpretation and distinctions across the short and long sides of the neutral region. The numerical readouts are model outputs. - CONFIRMED | the peaks Neitz, Geist & Jacobs measured in three dogs. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [three-dogs] Retained N2 reading: Neitz 1989 examined three dogs in behavioural discrimination experiments and derived the cone-peak estimates from those results. - CONFIRMED | The dog retina has two cone classes, peaking near 429 and 555 nm | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [two-classes] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | a published shape that fits real pigments well but is keyed only by the peak wavelength; | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:11016572&format=json&resultType=core | [template-real] Retained N2 reading: N2 read the Govardovskii abstract: the combined alpha and beta bands approximate visual pigment spectra above a 350 nm peak wavelength. - CONFIRMED | a different white (daylight, a screen's white) moves it a few nm. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [white-shift] Retained N2 reading: N2 recomputed 483.7 nm for equal-energy white and 488.9 nm for the page's model screen white, a 5.2 nm shift. This is a model example, not a measured universal shift. - CONFIRMED | The peaks come from three dogs plus electrophysiology; | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML ; https://doi.org/10.1017/S0952523800003291 | [peaks-plus-electro] Retained N2 reading: N2 distinguished Neitz's behavioural estimates from later ERG work: Jacobs et al. 1993 reports about 555 nm and 430-435 nm. The repair attributes the page's 429/555 nm estimates specifically to Neitz's behavioural work. - CONFIRMED | Take 429/555 nm as the well-supported central values, not a per-dog guarantee. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [central-values] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | Brightness and contrast processing differ too, | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML | [brightness-contrast] Retained N2 reading: Siniscalchi discusses canine brightness discrimination separately from colour and notes differences from humans. - CONFIRMED | The two cone peaks (≈429 and ≈555 nm) and the dichromatic conclusion from behavioural colour-matching in three dogs: | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [source-conclusion] Retained N2 reading: Neitz 1989 examined three dogs in behavioural discrimination experiments and derived the cone-peak estimates from those results. - CONFIRMED | The cone-curve shape is the visual-pigment template of | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:11016572&format=json&resultType=core | [gov-source-shape] Retained N2 reading: N2 read the Govardovskii abstract: the combined alpha and beta bands approximate visual pigment spectra above a 350 nm peak wavelength. - CONFIRMED | Dog hue = S/(S+L) | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [hue-definition] Retained N2 reading: The nonzero-signal coordinate is S/(S+L). Under X11, zero now explicitly has no coordinate instead of the original numerical guard assigning zero. - CONFIRMED | To recolour an arbitrary on-screen pixel we model your display's R/G/B primaries as narrow Gaussians (centres 611 / 549 / 465 nm) | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [gaussians] Retained N2 reading: The inline code sets primary centres to 611, 549 and 465 nm and integrates the stated Gaussian model. - CONFIRMED | The neutral point is defined against an equal-energy white; | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [equal-energy] Retained N2 reading: The model sums each pigment with constant spectral weight from 400 through 700 nm. - CONFIRMED | The dog-view image is a model of which colours a dog confuses, mapped back into colours you can see on a trichromat screen. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [dog-model] Retained N2 reading: The code maps two cone signals to illustrative screen anchors. The page expressly limits this to a model, with no inference about canine subjective experience. - CONFIRMED | It is not a window into canine qualia | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [qualia-limits] Retained N2 reading: The code maps two cone signals to illustrative screen anchors. The page expressly limits this to a model, with no inference about canine subjective experience. - CONFIRMED | "sees yellow and blue" is shorthand for the two-cone axis, not a claim about what the experience is like. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [axis-label] Retained N2 reading: The code maps two cone signals to illustrative screen anchors. The page expressly limits this to a model, with no inference about canine subjective experience. - CONFIRMED | none of that is in this colour-only model. | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [model-exclusions] Retained N2 reading: The code has illustrative L-channel brightness, but no model of canine acuity, contrast processing or low-light physiology. - CONFIRMED | Blue cone (429 nm) | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [curve-blue-label] Retained N2 reading: The code uses S_PEAK=429; Neitz estimates the short-wave pigment near 429 nm. - CONFIRMED | Yellow cone (555 nm) | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html | [curve-yellow-label] Retained N2 reading: The code uses L_PEAK=555; Neitz estimates the long/medium-wave pigment near 555 nm. - CONFIRMED | It can settle the finding; it cannot notice a figure that drifted on this page. | git show f5e5f84f34:research/the-colours-the-dog-keeps/verify-the-colours-the-dog-keeps.mjs | [check-limits] Retained N2 reading: The independent verifier recomputes its own functions and does not read the page. Its ten numerical assertions cannot detect prose drift; this pass adds a separate interaction check. - CONFIRMED | Neitz J, Geist T, Jacobs GH, "Color vision in the dog," Visual Neuroscience 1989;3(2):119–125 | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [cite-neitz] Retained N2 reading: N2 checked publication metadata and DOI correspondence. The shortened linked title identifies the cited work; X8 and X9 correct the separate claims about what those works cover. - CONFIRMED | Siniscalchi M, d'Ingeo S, Fornelli S, Quaranta A, "Are dogs red–green colour blind?" Royal Society Open Science 2017;4(11):170869 | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML | [cite-siniscalchi] Retained N2 reading: N2 checked publication metadata and DOI correspondence. The shortened linked title identifies the cited work; X8 and X9 correct the separate claims about what those works cover. - CONFIRMED | Byosiere et al., "What do dogs see?" Psychon. Bull. Rev. 2018 | https://link.springer.com/article/10.3758/s13423-017-1404-7 | [cite-byosiere] Retained N2 reading: N2 checked publication metadata and DOI correspondence. The shortened linked title identifies the cited work; X8 and X9 correct the separate claims about what those works cover. - CONFIRMED | Govardovskii et al., "In search of the visual pigment template," Visual Neuroscience 2000;17(4):509–528 | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:11016572&format=json&resultType=core | [cite-govardovskii] Retained N2 reading: N2 checked publication metadata and DOI correspondence. The shortened linked title identifies the cited work; X8 and X9 correct the separate claims about what those works cover. - CONFIRMED | Your dog is not colourblind and does not see the world in grey. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [dek-not-grey] Retained N2 reading: Neitz establishes dichromatic colour vision. In this context not colourblind means not monochromatic, not human trichromacy. - CONFIRMED | It is a dichromat, two cone pigments instead of your three, with measured peaks at 429 nm (blue) and 555 nm (yellow-green), | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [dek-two-pigments] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | so red, orange, yellow and green all collapse onto one channel while blue stays distinct. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [dek-collapse] Retained N2 reading: Neitz reports rapidly declining wavelength discrimination away from 480 nm and effectively absent discrimination above about 520 nm in the reported tests. The page model likewise has a very small short-cone fraction at long wavelengths. - CONFIRMED | Your dog is not colourblind, and it does not see in grey. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [md-not-grey] Retained N2 reading: Neitz establishes dichromatic colour vision. In this context not colourblind means not monochromatic, not human trichromacy. - CONFIRMED | two colour cones, peaking at 429 nm and 555 nm, where you have three | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [md-cones] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | and that two-cone retina confuses red with green while keeping blue crisp. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML | [md-red-blue] Retained N2 reading: The cited colour-discrimination experiments support the dichromatic interpretation and distinctions across the short and long sides of the neutral region. The numerical readouts are model outputs. - CONFIRMED | The numbers come from Neitz, Geist & Jacobs measuring three dogs in 1989. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [md-neitz-date] Retained N2 reading: Neitz 1989 examined three dogs in behavioural discrimination experiments and derived the cone-peak estimates from those results. - CONFIRMED | and find the single wavelength near 480 nm a dog cannot tell from grey. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [md-neutral] Retained N2 reading: Neitz reports an approximately 480 nm match against a 4800 K comparison with brightness controlled. The page model uses equal-energy white; the repair makes brightness and reference-white conditions explicit. - CONFIRMED | Dogs aren't colourblind and they don't see in grey. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [meta-not-grey] Retained N2 reading: Neitz establishes dichromatic colour vision. In this context not colourblind means not monochromatic, not human trichromacy. - CONFIRMED | They're dichromats with two cone pigments peaking at 429 nm (blue) and 555 nm (yellow-green), measured in three dogs by Neitz, Geist & Jacobs in 1989. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [meta-peaks] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | Red, orange, yellow and green all collapse onto one yellow channel, | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [meta-yellow] Retained N2 reading: Neitz reports rapidly declining wavelength discrimination away from 480 nm and effectively absent discrimination above about 520 nm in the reported tests. The page model likewise has a very small short-cone fraction at long wavelengths. - CONFIRMED | One wavelength near 480 nm even matches grey. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [meta-grey] Retained N2 reading: Neitz reports an approximately 480 nm match against a 4800 K comparison with brightness controlled. The page model uses equal-energy white; the repair makes brightness and reference-white conditions explicit. - CONFIRMED | Dogs aren't colourblind: they're dichromats with cones at 429 and 555 nm. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [og-peaks] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | Dogs are dichromats: two cones at 429 and 555 nm. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [twitter-peaks] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | Red and green collapse to one hue, | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC5717654/fullTextXML | [twitter-collapse] Retained N2 reading: The cited colour-discrimination experiments support the dichromatic interpretation and distinctions across the short and long sides of the neutral region. The numerical readouts are model outputs. - CONFIRMED | They're dichromats with two cone pigments peaking at 429 nm and 555 nm, measured behaviourally by Neitz, Geist & Jacobs in 1989. | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [schema-peaks] Retained N2 reading: The papers support two canine cone classes and approximate peaks near 429 and 555 nm. These are estimates, not per-dog guarantees. - CONFIRMED | Wavelengths a human reads as red, orange, yellow and green all map onto the dog's single yellow channel | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [schema-warm] Retained N2 reading: Neitz reports rapidly declining wavelength discrimination away from 480 nm and effectively absent discrimination above about 520 nm in the reported tests. The page model likewise has a very small short-cone fraction at long wavelengths. - CONFIRMED | magenta is shown to be a colour with no wavelength | https://visionbook.mit.edu/color.html | [relation-magenta] Retained N2 reading: The cited colour chapter treats magenta as a mixture of short- and long-wave light, rather than a monochromatic wavelength. - CONFIRMED | a thing the human three-cone system invents. | https://visionbook.mit.edu/color.html | [relation-invents] Retained N2 reading: The cited colour chapter treats magenta as a mixture of short- and long-wave light, rather than a monochromatic wavelength. - CONFIRMED | Screens exploit your three cones with just three lights | https://visionbook.mit.edu/color.html | [relation-screens] Retained N2 reading: The cited chapter explicitly describes rendering colour with three kinds of display dots. - CONFIRMED | metamerism, the trick of fooling a cone system with a mixture. | https://visionbook.mit.edu/color.html | [relation-metamerism] Retained N2 reading: The cited chapter defines metamers as different spectra that appear the same colour. - CONFIRMED | This page is metamerism's twin in a smaller space: with only two cones, far more lights become metamers, which is exactly why a dog cannot separate red from green. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC6086919/fullTextXML | [relation-more-metamers] Retained N2 reading: N2 read the source discussion of how the number of cone types and their spectral sensitivities constrain metamerism. - CONFIRMED | Same physics of cone integration, fewer dimensions. | https://www.ebi.ac.uk/europepmc/webservices/rest/PMC6086919/fullTextXML | [relation-integration] Retained N2 reading: N2 read the source discussion of how the number of cone types and their spectral sensitivities constrain metamerism. - CONFIRMED | those taste zones were never there. | https://www.nidcd.nih.gov/health/taste-disorders | [relation-tongue] Retained N2 reading: NIDCD says different taste-cell types are distributed across the tongue, supporting the rejection of exclusive taste zones. - CONFIRMED | dogs do have reduced colour vision, but it is a precise dichromacy, not an absence, | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:2487095&format=json&resultType=core | [relation-reduced] Retained N2 reading: Neitz establishes dichromatic colour vision. In this context not colourblind means not monochromatic, not human trichromacy. - CONFIRMED | dogs and cattle are both dichromats (two cones, not three), | https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:9685209&format=json&resultType=core ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [nearby-cattle] Retained N2 reading: The cattle source identifies two cone mechanisms in cattle, goats and sheep. Neitz supplies the separate canine evidence. - CONFIRMED | 'Dogs are colourblind / see in grey' is the same shape of myth, half-true: | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [myth-quotation] Retained N2 reading: The words label a misconception on the page, not a quotation attributed to Neitz. His dichromacy finding supports the rebuttal. N2 rejected the source-record quote classification as a page fault. - CONFIRMED | At 400 nm the blue cone dominates: blue is a colour the dog keeps. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [live-wavelength-400] Retained N2 reading: The cited colour-discrimination experiments support the dichromatic interpretation and distinctions across the short and long sides of the neutral region. The numerical readouts are model outputs. - CONFIRMED | 484 nm is the rounded model neutral wavelength (the qualified N2 reading of the old slider message). | git show f5e5f84f34:public/strata/the-colours-the-dog-keeps/index.html ; https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [live-wavelength-484] Retained N2 reading: N2 confirmed only the rounded interpretation: 483.7 nm rounds to 484. At 484 nm the fraction is 0.342 rather than white 0.347. X16 now uses proximity wording instead of an equality claim. - CONFIRMED | At 500 nm both cones fire | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [live-wavelength-500] Retained N2 reading: The cited colour-discrimination experiments support the dichromatic interpretation and distinctions across the short and long sides of the neutral region. The numerical readouts are model outputs. - CONFIRMED | At 560 nm the blue cone is dark: this is in the dog’s single "yellow" pile (red–green all land here). | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [live-wavelength-560] Retained N2 reading: Neitz reports rapidly declining wavelength discrimination away from 480 nm and effectively absent discrimination above about 520 nm in the reported tests. The page model likewise has a very small short-cone fraction at long wavelengths. - CONFIRMED | At 700 nm the blue cone is dark: this is in the dog’s single "yellow" pile (red–green all land here). | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [live-wavelength-700] Retained N2 reading: Neitz reports rapidly declining wavelength discrimination away from 480 nm and effectively absent discrimination above about 520 nm in the reported tests. The page model likewise has a very small short-cone fraction at long wavelengths. - CONFIRMED | a dog can’t tell it from grey. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [live-grey] Retained N2 reading: Neitz reports an approximately 480 nm match against a 4800 K comparison with brightness controlled. The page model uses equal-energy white; the repair makes brightness and reference-white conditions explicit. - CONFIRMED | a blue-green the dog can place on its one axis. | https://hacksterio.s3.amazonaws.com/uploads/attachments/266921/1989-neitz-color_vision_in_dog-visneuro_cpvwNDQwND.pdf | [live-axis] Retained N2 reading: The cited colour-discrimination experiments support the dichromatic interpretation and distinctions across the short and long sides of the neutral region. The numerical readouts are model outputs. - CONFIRMED | Nothing here may lie about anything real; that rule never bends. | git show f5e5f84f34:WAKE.md | [shared-truth-rule] Retained N2 reading: N2 checked this as the project's stated rule in WAKE.md, not proof that every existing claim is true. - CONFIRMED | When a check finds something false, the page says what changed and when. | git show f5e5f84f34:assay/README.md | [shared-corrections] Retained N2 reading: N2 checked this as the correction policy in assay/README.md, not an audit of every historical repair. ## Findings - [fixed] X1: Re-read the page model and Neitz pp. 121-123: cone templates, display spectra, brightness and reference white are assumptions or conditions. Qualified the introduction and metadata as model predictions. Recorded only; no dated page line for this MINOR. - [fixed] X2: Re-read Siniscalchi Introduction and Neitz p. 122: the retained long/medium-wave cone peaks near 555 nm. Corrected the two-cone explanation in the page, research notes and verifier. Dated correction mirrored in HTML and markdown. - [fixed] X3: Re-read Neitz pp. 119-120: the study trained panel choices for food rewards. Its reference to electrophysiology describes earlier work. The page now attributes its 429/555 nm estimates to behavioural experiments. Dated correction mirrored in HTML and markdown. - [fixed] X4: Re-read Neitz p. 121: the neutral-point test controls brightness separately from colour. Equal ratios such as (0.1, 0.2) and (0.2, 0.4) do not give equal amplitudes. The page now separates chromaticity from brightness; verifier labels no longer call a model gap a measured confusion threshold. Dated correction mirrored in HTML and markdown. - [fixed] X5: Same retained-cone evidence as X2. Replaced the no-long-wave explanation with the responses of short-wave and long/medium-wave cones. Dated correction mirrored in HTML and markdown. - [fixed] X6: Re-read Barber p. 7 and Supplement 2: age-related yellow-pigment accumulation is not established for dogs and is said not to seem to occur in them. Douglas and Jeffery age comparisons concern bats, rats, mice and meerkats. Removed the canine yellowing claim without claiming that canine vision never changes with age. Dated correction mirrored in HTML and markdown. - [fixed] X7: Re-read Lind Results and Discussion: at 0.0087 cd/m², dog acuity was 1.8-3.5 cpd and human acuity was 5.9-9.9 cpd. Removed the universal dog resolution advantage; the correction names the tested light condition and separates sensitivity from acuity. Dated correction mirrored in HTML and markdown. - [fixed] X8: Re-read Siniscalchi Introduction, Methods and Discussion. Modified Ishihara cat animations test red-green confusion. Full text has no 480 or neutral point; positive controls find Ishihara and deuteranopia. The page and research notes now attribute the neutral-point measurement to Neitz. Dated correction mirrored in HTML and markdown. - [fixed] X9: Re-read Byosiere Color vision, PDF pp. 5-6, and searched the full extracted text: colour confusion is discussed, but neither 480 nor neutral point occurs. Restricted the attribution to canine colour vision and confusion, leaving the linked title unchanged. Dated correction mirrored in HTML and markdown. - [fixed] X10: Executed the original inline script: blue #2b6fe0 puts the pin at 51.35706548506539%, red at 3.4124363517233003%, and green at 6.563371687985822%. Replaced far end with well away from red and green. Dated correction mirrored in HTML and markdown. - [fixed] X11: Executed the original inline script: #000000 gives S=L=0 but outputs #9c9024 and a yellow caption. Zero now returns black, an undefined chromatic coordinate and a hidden hue marker; ordinary preset controls still work. Dated correction mirrored in HTML and markdown. - [fixed] X12: Read the complete independent verifier. It computes table quantities and six primary weights, not the chosen swatch or current slider. Narrowed the instruction, research notes and verifier overview; no page correction line for this MINOR. Recorded only; no dated page line for this MINOR. - [fixed] X13: Same model-scope evidence as X1. Qualified the markdown dek and repeated metadata copy with the stated model, matched brightness and reference white; no page correction line for this MINOR. Recorded only; no dated page line for this MINOR. - [open note:ad4d81] X14: Corrected the authored markdown relation to name short-wave and long/medium-wave cones, with a dated correction in HTML and markdown. The false absent-long-wave-cone statement remains without a correction annotation in public/strata/there-here/index.html:250 and research/lineage-edge-grammar/page.json:1, notes[1300]. Those copies belong to another layer; the board note request is in repair-report.md. - [open note:07430b] X15: Corrected this layer's relation and nearby teaser: its pigment template and Gaussian display primaries differ from the magenta page's CIE data and XYZ conversion. The false same-machinery statement remains without a correction annotation in public/strata/there-here/index.html:250 and research/lineage-edge-grammar/page.json:1, notes[1300]. Those copies belong to another layer; the board note request is in repair-report.md. - [fixed] X16: Executed updateWl(): at 480 nm the fraction is 0.423 and at 487 nm it is 0.283, against white 0.347. The code tests proximity within 4 nm of 483.7, not equality. All messages in that branch now say near the model neutral point. Dated correction mirrored in HTML and markdown. - [fixed] X17: Executed the neutral calculation: whiteHue=0.3472918809928639. The original CSS put grey at 50%. The gradient stop and label now share a CSS position set from whiteHue*100, about 34.729%. Dated correction mirrored in HTML and markdown. ## Check records The source readings are in research/assay-line/cited-1014b/the-colours-the-dog-keeps/N2-codex.md and N1-sources.json. This repair uses the final levels in triage.json, including the six findings raised to WRONG. Commands, before/after edits, exact dated lines and limitations are in research/assay-line/cited-1014b/the-colours-the-dog-keeps/fix-report.md. The fix-check review, dated-line audit and pending board-note requests are in research/assay-line/cited-1014b/the-colours-the-dog-keeps/repair-report.md. The separate interaction check is verify-repair.mjs in the same directory. ## Outside this layer The embedded relation dataset at public/strata/there-here/index.html:250 and its research copy at research/lineage-edge-grammar/page.json:1 still contain the old absent-cone and same-machinery wording in notes[1300]. They belong to another layer and were left untouched under the explicit scope restriction. The authored relation and generated teaser for this layer are fixed. The two N2 uncertainties about actual canine curve widths and actual display-primary widths were not in the supplied triage and remain outside this repair.