# Arkle version 0003 — research and reconstruction findings

Evidence cutoff: 30 September 2026. Public information only. This is an extension of the selected-source version 0002 audit, not a complete historical drilling or tenure audit.

## What the additional research recovered

The original presentation contains embedded maps larger than the rendered page previews: pages 20 and21 are1733×975 pixels; the page 22 sampling map is652×734 and geology map556×626; the page 23 context map is1084×975. These source pixels are preserved under assets/v3. Their sample classes, bank interpretations, target boundaries, geological colours and labelled results support additional selectable geometry. The map backgrounds have no disclosed satellite acquisition date and are kept separate from the newly acquired dated geographic imagery. [Issuer presentation](https://static-media.fluxio.cloud/arkle/M68zWCjK.pdf)

The publisher's original June map is594×449 pixels, and September programme/ULG maps are477×418 and476×416. Fetching the original attachments did not reveal hidden high-resolution collar tables. Low-resolution marks can merge, so the reconstructed symbol fragments are not a count of unique holes or laboratory samples. The source imagery remains visible in Compare and available as a direct download.

The June release supports five named historical intervals: VB05,2.7 m at99.5 ppm from0.4 m; VB06,2.7 m at100 ppm from0.4 m; RC08,4.4 m at133.9 ppm from4 m; RC28,1 m at303 ppm from6.5 m; RC27,2 m at215 ppm from15.5 m. The figure labels U3O8 without an e, while the announcement identifies the method as downhole gamma-ray spectrometry and defines eU3O8. The viewer therefore labels these as radiometric equivalents, not chemical assays. The VB05/VB06 rings are close and their callouts overlap; the paired symbols are not forcibly assigned individual identities. RC27/RC28 positions were not recovered. [June announcement](https://www.investegate.co.uk/announcement/rns/arkle-resources-cdi---ark/namibia-uranium-exploration-update/9635389)

Five shared-scale downhole interval columns can be drawn without inventing a trajectory. The unreported portions of the holes are unknown, not barren intervals. No total hole depth, dip, azimuth, collar RL or complete downhole survey was recovered for these five records. The unnamed9.4 m central-east interval remains unpositioned because its from-depth and identity are missing.

Nine modern ULG collar-symbol/label associations are retained from version 0002. The September release gives programme ranges of70–200 m depth and−60° to−90° dip, and reports four deviation surveys. Those ranges cannot be assigned to individual holes. Chemical assay results and calibrated gamma values for the modern holes were not recovered through the cutoff. The hole register leaves those fields null. [September announcement](https://www.investegate.co.uk/announcement/rns/arkle-resources-cdi---ark/extensive-uranium-system-defined-at-ulg-target/9769143)

## Public research routes and remaining access gaps

- Issuer acquisition, technical updates, interim statements, presentation and original RNS image attachments: source and named-hole searches were extended. No complete collar/orientation/EOH/survey table or underlying Namibia NI43-101 report was recovered. The acquisition budgets an updated report; this does not establish that an older report is publicly available.
- Geological Survey of Namibia / Earth Data Namibia: indexed official publications establish an archive of exploration reports and drill records. The live portal and2023 publication could not be captured because of DNS/fetch failures. It is a specific next retrieval route, not evidence that Arkle's holes are present or absent. [GSN archive paper](https://www.mme.gov.na/files/publications/1c0_8_Schreiber%202023_Earth%20Data%20Namibia%20%28EDN%29.pdf)
- Historical operators and consultants: searches followed Namibia Uranium, Oakridge Vision, Chris Healey, Terratec and the named holes. No original operator-to-hole chain was established. Neighbouring-company drill records were excluded unless ownership and spatial overlap could be demonstrated.
- Environmental appendices: additional selected passages of the7986 and8298 management/audit packages describe reconnaissance, sample collection and siting drill points. They do not supply a completed drill database. Template references to other EPL numbers remain warnings against transferring an annex indiscriminately. The earlier8995/8026 tenure findings remain unchanged.
- Duplicate distributions: OTC's June RNS body was readable through web retrieval, but the separate direct capture failed. It is a duplicate of the existing captured issuer announcement, not a new drilling dataset. October2025 licence-list retrieval also failed and cannot confirm September2026 title.

Access attempts are preserved in captures/v3/access-audit.json; original image URLs and dimensions are in captures/v3/original-image-downloads.json. These attempts do not increase full-document review counts.

## Figure-level findings and limits

- ULG map: additional result callouts146,307,278 and180 ppm are retained with source-pixel endpoints. Exact sample IDs and sample support remain unresolved. The800,358 and362 ppm callouts from version 0002 remain. A362 ppm map label is a sample-linkage question, not proof of contradiction with a trench-only358 ppm maximum.
- Paleo East: selective surface classes, three labelled surface values, and partial interpreted channel banks are reconstructed. Category boundaries are copied as printed, without inventing endpoint inclusion rules or treating selected grades as deposit averages.
- HLEM: source plan traverses and12.5 m/16 m depth annotations are interpreted geophysical evidence. The caption's12.5 m potential maximum does not reconcile the16 m figure label. Neither is uranium thickness.
- Regional targets: dated licence and target outlines are native linework. Source radiometric colours represent U counts per second; no assay grades are assigned to that raster.
- Geology: eight conservative colour fragments from the source map remain flat polygons, with label/occlusion holes preserved. Boundaries outside recoverable colour cores remain in Original/Compare. No surface is interpolated between these fragments.
- North Rössing: the licence outline and partial dome trace are geological context, not a solid. The sample figure prints2924 ppm; acquisition prose reports2923 ppm. Both values remain source-specific and unresolved.

## Coordinate and display methods

Every exploration drawing remains source-local. Horizontal coordinate labels alone do not prove datum, elevation reference or surveyed registration. Independent holdout controls have not been recovered, so these layers are not draped onto the geographic terrain.

Copernicus GLO30 terrain and September2026 Sentinel-2 imagery share an explicit projected EPSG32733 frame. The DSM is EGM2008 metres and displayed at vertical exaggeration1. Its2011–2015 acquisition period differs from the satellite dates. This is regional surface context, not a surveyed collar-elevation source. The regional coverage is14.3–15.4°E and22.65–21.65°S; individual licence camera presets are display choices, not cadastral boundaries. Provider alignment does not validate Arkle's drawings.

The source-image checklist records partial reconstruction and omissions. Pixel tolerance is a tracing uncertainty, not ground accuracy. Native geometry, picking and SVG use the same feature coordinates. Geometry and public-source metadata remain separate from presentation colours and camera state.

## Repository grounding

Active drill-results extraction rules govern interval cardinality, units, null missing metadata and the separation of programme intent from completed results. Active general/cross-context guidance prevents restated results, unrelated zones and incompatible sample populations being treated as new comparable evidence. Resource guidance prevents targets becoming resources or solids. Exploration-results and exploration-intelligence drafts guide questions about coverage, bias, methods and controls; no draft scoring bands are adopted. Matrix rowsM11–M13 link these questions to the actual sources and viewer consequences.


## Final display and label review

The modern map association retained as R006 is ambiguous between R006 and R060 at the source resolution; it is not a verified hole identity. Six weak June collar candidates were removed after enlarged source review. The geometry review register identifies each exclusion.

The Sentinel display uses B04/B03/B02, common white0.6 and power0.6 without per-channel white balance. All contributing STAC items declare the BOA offset already applied, so no second −0.1 offset is applied. The corrected display and metadata were finalized before freezing. See [provider explanation](https://github.com/Element84/earth-search/issues/24), the bundled STAC items and rgb-correction.json. This processing is for visual context, not quantitative spectroscopy. Satellite pixel clarity does not validate exploration registration.