diff --git a/docs/provenance/investigation-index.md b/docs/provenance/investigation-index.md
index 64b49ac80b4..7f03e7a74fd 100644
--- a/docs/provenance/investigation-index.md
+++ b/docs/provenance/investigation-index.md
@@ -4,7 +4,7 @@ Generated by `node packages/bake/cli/report-investigations.mts --index --write`.
 decisions, not scientific readiness.
 
 **Ledgers:** 4394 of 5329 catalogued objects.
-**Open decisions:** 1427 (1187 unresolved, 240 deferred).
+**Open decisions:** 1429 (1187 unresolved, 242 deferred).
 
 ## Where the reopening evidence would appear
 
@@ -32,6 +32,7 @@ The source each decision examined. Watching these is how a decision reopens.
 | www.hou.usra.edu | 5 | ariel, hydra, nix, varuna, annefrank |
 | astro.troja.mff.cuni.cz | 4 | eurybates, orus, patroclus, polymele |
 | echo.jpl.nasa.gov | 4 | comet-209p, comet-2p, comet-8p, nereus |
+| naif.jpl.nasa.gov | 4 | annefrank, dinkinesh, ceres, iapetus |
 | almascience.eso.org | 3 | r-doradus, eps-eridani-disc, haumea |
 | archive.stsci.edu | 3 | neptune, m95-layers |
 | github.com | 3 | dinkinesh, wasp-43b, wasp-103b |
@@ -43,7 +44,6 @@ The source each decision examined. Watching these is how a decision reopens.
 | esahubble.org | 2 | abell-2744-layers, homunculus-nebula-layers |
 | esawebb.org | 2 | bullet-cluster-layers, cassiopeia-a-layers |
 | mast.stsci.edu | 2 | hd-189733b, wd-1856-534b |
-| naif.jpl.nasa.gov | 2 | annefrank, dinkinesh |
 | opus.pds-rings.seti.org | 2 | phoebe, metis |
 | pubs.usgs.gov | 2 | ganymede, mars |
 | science.nasa.gov | 2 | mercury, thebe |
@@ -159,6 +159,7 @@ The source each decision examined. Watching these is how a decision reopens.
 | `lorri-dataset` | 2 | hydra, nix | A meaningfully different view confirms the registration with interior holdouts and the mesh-to-kernel frame binding is verified. |
 | `mvic-leisa-composition` | 2 | hydra, nix | A target-specific composition product has interpretable calibration, coordinates and qualified transfer onto this source mesh. |
 | `north-south-degeneracy` | 2 | hd-189733b, wasp-43b | The orbit's sky orientation is measured. |
+| `oblate-figure` | 2 | ceres, iapetus | The shared sphere geometry places planetocentric latitudes on a spheroid and casts feature anchors onto it. |
 | `optical-scientific-surfaces` | 2 | sn263-beta, sn263-gamma | A component-specific observation, registered scientific map or qualified spin solution is located. |
 | `optical-surface-map` | 2 | comet-209p, comet-2p | Resolved surface observations and camera or terrain controls support a reproducible mapping. |
 | `other-hrii-scans` | 2 | comet-103p, comet-9p | A specific scan adds useful source facets with independently checked registration and valid nucleus spectra. |
diff --git a/src/objects/ariel/investigations.json b/src/objects/ariel/investigations.json
index cf2ef53a5de..7eb90da079d 100644
--- a/src/objects/ariel/investigations.json
+++ b/src/objects/ariel/investigations.json
@@ -85,7 +85,7 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "unresolved",
-      "finding": "Helfenstein, Veverka and Thomas (1988) fitted Hapke models to Voyager whole-disc photometry of the five large Uranian moons. Only the abstract could be read (single-scattering albedo from 0.3 for Umbriel to 0.7 for Ariel, g near -0.3), and a whole-disc fit states no emission range. NASA's Reports of Planetary Geology and Geophysics Program for 1986 to 1989, which are open, print disc values for Miranda, Titania, Umbriel and Oberon, and none for Ariel. Schenk's 2020 mosaic history names ISIS photomet without its parameter file. Ariel keeps the shared sphere bank.",
+      "finding": "Helfenstein, Veverka and Thomas (1988) fitted Hapke models to Voyager whole-disc photometry of the five large Uranian moons. Only the abstract could be read (single-scattering albedo from 0.3 for Umbriel to 0.7 for Ariel, g near -0.3), and a whole-disc fit states no emission range. NASA's Reports of Planetary Geology and Geophysics Program for 1986 to 1989, which are open, print disc values for Miranda, Titania, Umbriel and Oberon, and none for Ariel. Schenk's 2020 mosaic history names ISIS photomet without its parameter file. Measured here on 2026-10-05, not a published law: on one calibrated clear frame at 20 degrees phase (FDS 26794.34, a 31 pixel disc) the lunar-Lambert weight that best fits the lit disc is 0.77, close to what Titania shows at that phase. Ariel keeps the shared sphere bank.",
       "evidence": [
         "https://doi.org/10.1016/0019-1035(88)90039-5",
         "https://ntrs.nasa.gov/citations/19880051927",
diff --git a/src/objects/callisto/investigations.json b/src/objects/callisto/investigations.json
index 174941780ca..643488ec378 100644
--- a/src/objects/callisto/investigations.json
+++ b/src/objects/callisto/investigations.json
@@ -46,7 +46,7 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "included",
-      "finding": "Lunar-like (Lommel-Seeliger) law of Squyres and Veverka (1981), which has no parameter. The Icarus paper is closed, but the authors' own summary in NASA's Reports of Planetary Geology Program 1980 (TM-82385, page 67) is open. It says Voyager clear-filter images at 10 to 124 degrees phase show that Ganymede's Cratered and Grooved Terrain and Callisto's Cratered Terrain are described closely by I = A F(alpha) mu0/(mu + mu0), and that bright craters follow it approximately. Flood-lit, this law is flat to the limb. F(alpha) is drawn in the report's Figure 1 and not printed, so frames with Shadows on carry no phase term. No emission range is printed; the limit 86.3 degrees is derived here from the outermost pixel of the finest clear-filter Voyager frame that holds the whole disc (Voyager 2 C2058321, 10.116 km per pixel) and is second class. Not used: USGS and the conference abstract on these mosaics (Becker et al. 2001) say they were normalized with a lunar-Lambert function with empirically derived values, which neither prints; Buratti (1991) fitted Hapke models to whole-hemisphere phase curves, which give no disc-resolved law. The Callisto photometry of Belgacem and colleagues covers regions only.",
+      "finding": "Lunar-like (Lommel-Seeliger) law of Squyres and Veverka (1981), which has no parameter. The Icarus paper is closed, but the authors' own summary in NASA's Reports of Planetary Geology Program 1980 (TM-82385, page 67) is open. It says Voyager clear-filter images at 10 to 124 degrees phase show that Ganymede's Cratered and Grooved Terrain and Callisto's Cratered Terrain are described closely by I = A F(alpha) mu0/(mu + mu0), and that bright craters follow it approximately. Flood-lit, this law is flat to the limb. F(alpha) is drawn in the report's Figure 1 and not printed, so frames with Shadows on carry no phase term. No emission range is printed; the limit 86.3 degrees is derived here from the outermost pixel of the finest clear-filter Voyager frame that holds the whole disc (Voyager 2 C2058321, 10.116 km per pixel) and is second class. Not used: USGS and the conference abstract on these mosaics (Becker et al. 2001) say they were normalized with a lunar-Lambert function with empirically derived values, which neither prints; Buratti (1991) fitted Hapke models to whole-hemisphere phase curves, which give no disc-resolved law. The Callisto photometry of Belgacem and colleagues covers regions only. McEwen and Soderblom (1983), an open summary in NASA TM-86246 (page 261), solved the limb darkening from pairs of Voyager images of the same ground: 16 pairs on Callisto and one on Ganymede, both consistent with lunar-like scattering. That is an independent disc-resolved result for the same law.",
       "evidence": [
         "https://ntrs.nasa.gov/citations/19810007392",
         "https://doi.org/10.1016/0019-1035(81)90203-7",
@@ -54,7 +54,8 @@
         "https://opus.pds-rings.seti.org/opus/api/metadata/vg-iss-2-j-c2058321.json",
         "https://astrogeology.usgs.gov/search/map/callisto_voyager_galileo_image_mosaic_map",
         "https://www.lpi.usra.edu/meetings/lpsc2001/pdf/2009.pdf",
-        "https://ntrs.nasa.gov/citations/19910063813"
+        "https://ntrs.nasa.gov/citations/19910063813",
+        "https://ntrs.nasa.gov/citations/19840015363"
       ],
       "revisitWhen": "The paper's phase functions and emission range, or a newer global disc-resolved Callisto law, are read."
     },
diff --git a/src/objects/ceres/README.md b/src/objects/ceres/README.md
index 6712dc6d186..5da785266da 100644
--- a/src/objects/ceres/README.md
+++ b/src/objects/ceres/README.md
@@ -27,7 +27,7 @@ Source selections, recorded trials and open questions are in the [investigation
 
 ## Processing
 
-The mesh is a sphere of 469.7 km with no resolved shape model. The maps share a global equirectangular grid. Exactly black pixels connected to the southern border are marked with a neutral gray grid; no terrain is inferred. The Dawn FC2 mosaic is orthorectified in the body-fixed frame by its producers ([PDS record](https://pds.nasa.gov/ds-view/pds/viewProfile.jsp?dsid=DAWN-A-FC2-5-CERESMOSAIC-V1.0)); its registration does not depend on a crater match by cssEarth.
+The mesh is a sphere of 469.7 km with no resolved shape model. Ceres is flattened by 7.5% (482 km at the equator, 446 km at the poles, [dawn_ceres_v05.tpc](https://naif.jpl.nasa.gov/pub/naif/DAWN/kernels/pck/dawn_ceres_v05.tpc)); the [ledger](investigations.json) says what in the shared geometry must change before that can be drawn. The maps share a global equirectangular grid. Exactly black pixels connected to the southern border are marked with a neutral gray grid; no terrain is inferred. The Dawn FC2 mosaic is orthorectified in the body-fixed frame by its producers ([PDS record](https://pds.nasa.gov/ds-view/pds/viewProfile.jsp?dsid=DAWN-A-FC2-5-CERESMOSAIC-V1.0)); its registration does not depend on a crater match by cssEarth.
 
 The Ammonium band is the 3.1 µm band depth between the reflectance maxima in 2.91–3.01 µm and 3.19–3.24 µm. `packages/bake/authoring/dawn/vir-mosaic.mts` builds each line's camera from the Dawn SPICE kernels, converts radiance to I/F, removes stripes and spikes, brings each cube to 30° phase, sets aside cubes that disagree with all their neighbours, and blends seams as ISIS3's `noseam` does, at 16 pixels/degree (0.5 km). Survey and HAMO set every cell they saw; LAMO, Approach and set-aside cubes fill gaps. The map is 98.7% covered. The [phase fit and rejections](source/science/vir-reduction/photometry.json), [recipe](source/science/vir-reduction/recipe.json) and [receipt](source/science/vir-reduction/receipt.json) are recorded.
 
diff --git a/src/objects/ceres/investigations.json b/src/objects/ceres/investigations.json
index 081d3878bf7..6fc734d7046 100644
--- a/src/objects/ceres/investigations.json
+++ b/src/objects/ceres/investigations.json
@@ -345,6 +345,18 @@
         "https://doi.org/10.3847/1538-3881/ab8305"
       ],
       "revisitWhen": "A published resolved thermal-inertia or emissivity map of Ceres from these ALMA data, or a new ALMA epoch whose surface-fixed residuals repeat between independent epochs."
+    },
+    {
+      "id": "oblate-figure",
+      "subject": "Figure of the globe",
+      "status": "deferred",
+      "finding": "The globe is a 469.7 km mean-radius sphere, but Ceres is flattened: the Dawn project's NAIF kernel dawn_ceres_v05.tpc gives an average equatorial radius of 482 km and a polar radius of 446 km (equivalent sphere 470 km), so the poles sit 7.5% closer to the centre than the equator, about 17 pixels of the 460 pixel globe. Tried on 2026-10-05 and withdrawn: setting the recipe to that spheroid stops the bake, and the stop is right. Three things in the shared sphere lane hold only for a nearly round body such as Mars (0.6%). The feature step compares the Gazetteer's 470 km datum sphere with the authored radius, which becomes the 482 km equatorial radius, and refuses beyond 1%. It then places anchors on a sphere, so at 7.5% labels would float above the poles. And the mesh places a map latitude at the spheroid's parametric latitude, up to 2.2 degrees from the planetocentric latitude the maps use. Drawing the figure correctly means changing that shared geometry, which also moves Mars, Uranus and Neptune.",
+      "evidence": [
+        "https://naif.jpl.nasa.gov/pub/naif/DAWN/kernels/pck/dawn_ceres_v05.tpc",
+        "https://github.com/layoutit/css.earth/blob/main/packages/bake/src/surface-geometry/surface.ts",
+        "https://github.com/layoutit/css.earth/blob/main/packages/bake/src/objects/surface-features/surface-features.ts"
+      ],
+      "revisitWhen": "The shared sphere geometry places planetocentric latitudes on a spheroid and casts feature anchors onto it."
     }
   ]
 }
diff --git a/src/objects/europa/README.md b/src/objects/europa/README.md
index c4ca71104bd..64ce6e90b1e 100644
--- a/src/objects/europa/README.md
+++ b/src/objects/europa/README.md
@@ -82,7 +82,7 @@ The withheld composition grids match the original release exactly after float32
 - **Monochrome inserts:** Smaller brightness steps remain where one gain per image cannot follow the phase-angle change in albedo, and shadows keep their original direction. Photographs beyond 85° incidence are not shown. Relative control uncertainties are about 247 m in latitude and 307 m in longitude; the older global mosaic has different registration errors. The delivered grid is about 1.20 km at the equator, so finer native detail is not shown. Atlas seams can show at extreme close zoom.
 - **False color:** 756 nm, 559 nm and 404 nm are shown as red, green and blue. This is not natural color. About 14.3% of the sphere has usable three-band coverage.
 - **Illumination:** The disk correction is approximate: no phase-angle normalization, fitted scattering model, or removal of cast shadows.
-- **Lighting law:** One law lights every terrain; the paper's chaos and crater rows differ from the ridged-plains row used. Its emission limit, 84.5°, is derived here, not stated by the paper. The fits cover 10° to 128° phase, so the flood-lit default view (0°) and frames beyond 128° use the printed lines outside that range. The line for f crosses zero at 111°; from there the lit side follows the Lambert term alone. The paper's own fit of its 10° image (A = 0.69, f = 1.17, its Figure 6) gives a darker flood-lit limb, 0.61 of the centre, than the lines of Table 2.
+- **Lighting law:** One law lights every terrain; the paper's chaos and crater rows differ from the ridged-plains row used. Its emission limit, 84.5°, is derived here, not stated by the paper. The fits cover 10° to 128° phase, so the flood-lit default view (0°) and frames beyond 128° use the printed lines outside that range. The line for f crosses zero at 111°; from there the lit side follows the Lambert term alone. The paper's own fit of its 10° image (A = 0.69, f = 1.17, its Figure 6) gives a darker flood-lit limb, 0.61 of the centre, than the lines of Table 2. McEwen and Soderblom (1983) found the same from 44 image pairs at 3° to 30° phase (0.64 at 3°), so the default view's limb is probably drawn too bright; the [ledger](investigations.json) says why it is not changed yet.
 - **Infrared:** This is a spectral color display, not an abundance map. Its 2010 registration grid is not the 2021 control grid.
 - **Feature outlines:** Craters and faculae trace a rim circle and other types their extent box; these are not published nomenclature boundaries.
 - **Composition reuse:** The Zenodo record is `other-open`, but gives no explicit terms for reusing the numerical release. The three composition views are withheld until such terms exist. SPHERE diffraction limits resolved features to about 150 km.
diff --git a/src/objects/europa/investigations.json b/src/objects/europa/investigations.json
index e2427538911..ec295065346 100644
--- a/src/objects/europa/investigations.json
+++ b/src/objects/europa/investigations.json
@@ -112,13 +112,15 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "included",
-      "finding": "Lommel-Seeliger plus Lambert law of Dhingra, Buratti and Seignovert (2021), fitted to 21 resolved Voyager 2, Galileo and New Horizons clear-filter images: I/F = A f mu0/(mu0 + mu) + (1 - A) mu0, with the ridged-plains lines of its Table 2, A = 0.871 - 0.003 alpha and f = 1.556 - 0.014 alpha (alpha in degrees), fitted at 10 to 128 degrees phase, relative to the flood-lit disc centre. A lighting overlay has one law for the whole body: the ridged-plains row is used because the paper says that unit covers more than half of Europa and that the parameters differ little between terrains. No map pixel changes. The paper states no emission range; the limit 84.5 degrees is derived here from the outermost pixel of the finest full-disc image in its Table 1 (Voyager 2 wide-angle C2065221, 14.465 km per pixel) and is second class. The same table's Minnaert lines are not used: Minnaert grows without bound toward the limb while its exponent is below 1. Buratti and Veverka (1983) introduced this form for Europa; its values could not be read. Belgacem et al. (2020) give Hapke fits for 20 regions only.",
+      "finding": "Lommel-Seeliger plus Lambert law of Dhingra, Buratti and Seignovert (2021), fitted to 21 resolved Voyager 2, Galileo and New Horizons clear-filter images: I/F = A f mu0/(mu0 + mu) + (1 - A) mu0, with the ridged-plains lines of its Table 2, A = 0.871 - 0.003 alpha and f = 1.556 - 0.014 alpha (alpha in degrees), fitted at 10 to 128 degrees phase, relative to the flood-lit disc centre. A lighting overlay has one law for the whole body: the ridged-plains row is used because the paper says that unit covers more than half of Europa and that the parameters differ little between terrains. No map pixel changes. The paper states no emission range; the limit 84.5 degrees is derived here from the outermost pixel of the finest full-disc image in its Table 1 (Voyager 2 wide-angle C2065221, 14.465 km per pixel) and is second class. The same table's Minnaert lines are not used: Minnaert grows without bound toward the limb while its exponent is below 1. Buratti and Veverka (1983) introduced this form for Europa; its values could not be read. Belgacem et al. (2020) give Hapke fits for 20 regions only. McEwen and Soderblom (1983), an open summary in NASA TM-86246 (page 261), solved the limb darkening from pairs of Voyager images of the same ground, 44 pairs on Europa: its scattering is uniform across materials and wavelength, with a Minnaert exponent K = 0.64 - 0.005a + 0.00066a^2 - 0.000012a^3 from 3 to 30 degrees phase. At 3 degrees that is 0.63, a flood-lit limb of 0.64 of the centre at 0.98 of the radius, where the Table 2 lines used here give 0.86 when carried below their 10 degrees, and the paper's own 10 degree image gives 0.61. That image's Minnaert fit is k = 0.65 for the whole disc and 0.64 for ridged plains (its Figures 4 and 6), the value the 1983 polynomial gives at 10 degrees (0.644); the paper's Table 2 line for k, 0.532 + 0.004a, gives 0.57 there. Two independent low-phase results therefore say the default view's limb is drawn too bright, and both of the paper's straight lines, fitted across 10 to 128 degrees, miss its own lowest-phase image. Not changed: the 1983 law stops at 30 degrees, so it cannot light every frame with Shadows on, and the two results cannot be joined into one published law.",
       "evidence": [
         "https://doi.org/10.3847/PSJ/ac06d6",
         "https://hal.science/hal-04259112",
         "https://opus.pds-rings.seti.org/opus/api/metadata/vg-iss-2-j-c2065221.json",
         "https://ntrs.nasa.gov/citations/19830059564",
-        "https://doi.org/10.1016/j.icarus.2019.113525"
+        "https://doi.org/10.1016/j.icarus.2019.113525",
+        "https://ntrs.nasa.gov/citations/19840015363",
+        "https://doi.org/10.48577/jpl.6IR6VA"
       ]
     },
     {
diff --git a/src/objects/ganymede/investigations.json b/src/objects/ganymede/investigations.json
index ed462907d7e..a308de8cdd0 100644
--- a/src/objects/ganymede/investigations.json
+++ b/src/objects/ganymede/investigations.json
@@ -94,7 +94,7 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "included",
-      "finding": "Lunar-like (Lommel-Seeliger) law of Squyres and Veverka (1981), which has no parameter. The Icarus paper is closed, but the authors' own summary in NASA's Reports of Planetary Geology Program 1980 (TM-82385, page 67) is open. It says Voyager clear-filter images at 10 to 124 degrees phase show that Ganymede's Cratered and Grooved Terrain and Callisto's Cratered Terrain are described closely by I = A F(alpha) mu0/(mu + mu0), and that bright craters follow it approximately. Buratti et al. (2017) and Dhingra, Buratti and Seignovert (2021) report the same result for Ganymede (lunar-like, A = 1). Flood-lit, this law is flat to the limb. F(alpha) is drawn in the report's Figure 1 and not printed, so frames with Shadows on carry no phase term. No emission range is printed; the limit 86.2 degrees is derived here from the outermost pixel of the finest clear-filter Voyager frame that holds the whole disc (Voyager 2 C2063551, 11.286 km per pixel) and is second class. Not used: USGS and the conference abstract on these mosaics (Becker et al. 2001) say they were normalized with a lunar-Lambert function with empirically derived values, which neither prints; Buratti (1991) fitted Hapke models to whole-hemisphere phase curves, which give no disc-resolved law. The near-infrared studies of Ligier et al. (2019) and King and Fletcher (2022) correct their spectra with an Oren-Nayar model, not at this map's wavelengths. The regional Hapke fits of Belgacem et al. (2021) could not be opened.",
+      "finding": "Lunar-like (Lommel-Seeliger) law of Squyres and Veverka (1981), which has no parameter. The Icarus paper is closed, but the authors' own summary in NASA's Reports of Planetary Geology Program 1980 (TM-82385, page 67) is open. It says Voyager clear-filter images at 10 to 124 degrees phase show that Ganymede's Cratered and Grooved Terrain and Callisto's Cratered Terrain are described closely by I = A F(alpha) mu0/(mu + mu0), and that bright craters follow it approximately. Buratti et al. (2017) and Dhingra, Buratti and Seignovert (2021) report the same result for Ganymede (lunar-like, A = 1). Flood-lit, this law is flat to the limb. F(alpha) is drawn in the report's Figure 1 and not printed, so frames with Shadows on carry no phase term. No emission range is printed; the limit 86.2 degrees is derived here from the outermost pixel of the finest clear-filter Voyager frame that holds the whole disc (Voyager 2 C2063551, 11.286 km per pixel) and is second class. Not used: USGS and the conference abstract on these mosaics (Becker et al. 2001) say they were normalized with a lunar-Lambert function with empirically derived values, which neither prints; Buratti (1991) fitted Hapke models to whole-hemisphere phase curves, which give no disc-resolved law. The near-infrared studies of Ligier et al. (2019) and King and Fletcher (2022) correct their spectra with an Oren-Nayar model, not at this map's wavelengths. The regional Hapke fits of Belgacem et al. (2021) could not be opened. McEwen and Soderblom (1983), an open summary in NASA TM-86246 (page 261), solved the limb darkening from pairs of Voyager images of the same ground: one pair on Ganymede and 16 on Callisto, both consistent with lunar-like scattering. That is an independent disc-resolved result for the same law.",
       "evidence": [
         "https://ntrs.nasa.gov/citations/19810007392",
         "https://doi.org/10.1016/0019-1035(81)90203-7",
@@ -106,7 +106,8 @@
         "https://www.lpi.usra.edu/meetings/lpsc2001/pdf/2009.pdf",
         "https://ntrs.nasa.gov/citations/19910063813",
         "https://arxiv.org/abs/1910.07445",
-        "https://doi.org/10.1016/j.icarus.2021.114631"
+        "https://doi.org/10.1016/j.icarus.2021.114631",
+        "https://ntrs.nasa.gov/citations/19840015363"
       ],
       "revisitWhen": "The paper's phase functions and emission range, or a newer global disc-resolved Ganymede law at visible wavelengths, are read."
     },
diff --git a/src/objects/iapetus/README.md b/src/objects/iapetus/README.md
index 2cf9402c496..26ba9b7cfea 100644
--- a/src/objects/iapetus/README.md
+++ b/src/objects/iapetus/README.md
@@ -30,7 +30,7 @@ Source selections, recorded trials and open questions are in the [investigation
 
 **Temperature.** The map is cleaned from NASA's published figure. Its frame and ticks give 2.000 figure pixels per degree, and the fit is good to about 0.25°. West longitude becomes east longitude, so the figure covers 180–360° E. The bar's 12 steps span 70 to 130 K, and the legend shows the same 12 colors. [`figures/pia07005-map.json`](source/figures/pia07005-map.json) withholds grid lines, ticks, the subsolar asterisk and the black background. `node packages/bake/authoring/figure-map/clean-figure-map.mts src/objects/iapetus/source/figures/pia07005-map.json` writes the cleaned crop.
 
-The globe uses the shared raster lane and its Lambert lighting bank, with no atmosphere, fake elevation or gap filling. [The shared preparation guide](../../../docs/surface-preparation.md#preserve-photographic-detail-through-preparation) describes how polar sprites sample the original photographs directly.
+The globe uses the shared raster lane, lit by its published law (see [Lighting law](#lighting-law)), with no atmosphere, fake elevation or gap filling. [The shared preparation guide](../../../docs/surface-preparation.md#preserve-photographic-detail-through-preparation) describes how polar sprites sample the original photographs directly.
 
 ## Lighting law
 
@@ -59,5 +59,5 @@ The temperature crop keeps 18.0% of the sphere by area. Against the base mosaic,
 - Temperature is one moment of one flyby. Local time matters as much as the surface: the Sun was overhead near 106° W, 9° S (the figure's asterisk). Places away from that point were in morning, evening or night.
 - Temperature shows NASA's published colors, not numbers. The map's colors change smoothly, but the bar has 12 steps, so a color can be read only to the nearest bar step. About 3% of the kept pixels, near noon, are paler than the bar's top step, and the figure gives no value for them.
 - Temperature's grid lines, axis ticks and asterisk are shown as missing (no data) strips and a small square, not filled in. The figure's black and its compression halo are also missing, so the dataset cannot show whether a black area was unobserved or colder than 70 K.
-- Geometry is a 734.3 km mean-radius sphere, without the oblate figure or equatorial ridge. No qualified downloadable height raster was found in the 2026-09-06 source search: the [current PDS SPC archive](https://sbnarchive.psi.edu/pds4/cassini/) has no Iapetus bundle, the [2025 author abstract](https://meetingorganizer.copernicus.org/EPSC-DPS2025/EPSC-DPS2025-115.html) says Iapetus is forthcoming, and the [USGS inventory](https://fdp.astrogeology.usgs.gov/fdp/saturn/) lists older stereo topography as unreleased. This does not establish that terrain models do not exist.
+- Geometry is a 734.3 km mean-radius sphere, without the oblate figure or equatorial ridge. The IAU figure is 745.7 km at the equator and 712.1 km at the poles ([pck00010.tpc](https://naif.jpl.nasa.gov/pub/naif/generic_kernels/pck/pck00010.tpc)), 4.5% flattened; the [ledger](investigations.json) says what in the shared geometry must change before it can be drawn. No qualified downloadable height raster was found in the 2026-09-06 source search: the [current PDS SPC archive](https://sbnarchive.psi.edu/pds4/cassini/) has no Iapetus bundle, the [2025 author abstract](https://meetingorganizer.copernicus.org/EPSC-DPS2025/EPSC-DPS2025-115.html) says Iapetus is forthcoming, and the [USGS inventory](https://fdp.astrogeology.usgs.gov/fdp/saturn/) lists older stereo topography as unreleased. This does not establish that terrain models do not exist.
 - The default photographic surface uses quarter dimensions on iPad. This reduces display detail, not the resolution of the preserved source observations.
diff --git a/src/objects/iapetus/investigations.json b/src/objects/iapetus/investigations.json
index bccf0872407..71e2d5a93d1 100644
--- a/src/objects/iapetus/investigations.json
+++ b/src/objects/iapetus/investigations.json
@@ -117,6 +117,19 @@
         "https://github.com/layoutit/css.earth/blob/main/src/objects/iapetus/source/observations/catalog.json",
         "https://github.com/layoutit/css.earth/blob/main/docs/surface-preparation.md#where-the-prepared-map-starts"
       ]
+    },
+    {
+      "id": "oblate-figure",
+      "subject": "Figure of the globe",
+      "status": "deferred",
+      "finding": "The globe is a 734.3 km mean-radius sphere, but Iapetus is flattened: the IAU radii NAIF's generic kernel pck00010.tpc prints are 745.7, 745.7 and 712.1 km, so the poles sit 4.5% closer to the centre than the equator, about 10 pixels of the 460 pixel globe. Tried on 2026-10-05 and withdrawn: setting the recipe to that spheroid stops the bake, and the stop is right. Three things in the shared sphere lane hold only for a nearly round body such as Mars (0.6%). The feature step compares the Gazetteer's 735.6 km datum sphere with the authored radius, which becomes the 745.7 km equatorial radius, and refuses beyond 1%. It then places anchors on a sphere, so at 4.5% labels would float above the poles. And the mesh places a map latitude at the spheroid's parametric latitude, up to 1.3 degrees from the planetocentric latitude the maps use. Drawing the figure correctly means changing that shared geometry, which also moves Mars, Uranus and Neptune. The equatorial ridge needs a height raster, which the README's known problems cover.",
+      "evidence": [
+        "https://naif.jpl.nasa.gov/pub/naif/generic_kernels/pck/pck00010.tpc",
+        "https://doi.org/10.1007/s10569-010-9320-4",
+        "https://github.com/layoutit/css.earth/blob/main/packages/bake/src/surface-geometry/surface.ts",
+        "https://github.com/layoutit/css.earth/blob/main/packages/bake/src/objects/surface-features/surface-features.ts"
+      ],
+      "revisitWhen": "The shared sphere geometry places planetocentric latitudes on a spheroid and casts feature anchors onto it."
     }
   ]
 }
diff --git a/src/objects/io/investigations.json b/src/objects/io/investigations.json
index 1918a10da2c..4263cba65ab 100644
--- a/src/objects/io/investigations.json
+++ b/src/objects/io/investigations.json
@@ -110,13 +110,14 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "included",
-      "finding": "Lunar-Lambert law of the USGS global mosaics: their process description says a limb-darkening correction with coefficient 0.7 was adequate for all three colors (ISIS LunarLambert, the library's form), and the pamphlet of the geologic map (USGS SIM 3168) gives the phase of those color frames as 3.5 to 4.5 degrees. The lighting puts that correction back, relative to the flood-lit disc centre. Neither document states an emission range; the limit 87.0 degrees is derived here from the outermost pixel of the finest full-disc frame at that phase (Galileo SSI C0359986578, orbit G2, 4.943 km per pixel) and is second class. The pamphlet says pixels too near the limb were removed by hand. No phase function is printed, so frames with Shadows on carry no phase term. The disc-resolved studies (Simonelli and Veverka 1986 from Voyager; Simonelli, Dodd and Veverka 2001, Hapke fits to Galileo SSI) could not be opened, and Io's photometry varies strongly between regions.",
+      "finding": "Lunar-Lambert law of the USGS global mosaics: their process description says a limb-darkening correction with coefficient 0.7 was adequate for all three colors (ISIS LunarLambert, the library's form), and the pamphlet of the geologic map (USGS SIM 3168) gives the phase of those color frames as 3.5 to 4.5 degrees. The lighting puts that correction back, relative to the flood-lit disc centre. Neither document states an emission range; the limit 87.0 degrees is derived here from the outermost pixel of the finest full-disc frame at that phase (Galileo SSI C0359986578, orbit G2, 4.943 km per pixel) and is second class. The pamphlet says pixels too near the limb were removed by hand. No phase function is printed, so frames with Shadows on carry no phase term. The disc-resolved studies (Simonelli and Veverka 1986 from Voyager; Simonelli, Dodd and Veverka 2001, Hapke fits to Galileo SSI) could not be opened, and Io's photometry varies strongly between regions. McEwen and Soderblom (1983), an open summary in NASA TM-86246 (page 261), solved the limb darkening from pairs of Voyager images of the same ground, 19 pairs on Io at 3 to 15 degrees phase, and print a Minnaert exponent per wavelength: 0.55 + 0.005a at 0.59 micrometres, 0.53 + 0.005a at 0.54, 0.52 + 0.005a at 0.48, 0.49 + 0.006a at 0.41 and 0.44 + 0.009a at 0.35, with a the phase in degrees. It is a per-color alternative, not adopted: the lighting puts back the one correction the mosaic itself received.",
       "evidence": [
         "https://astrogeology.usgs.gov/search/map/io_voyager_galileo_ssi_global_mosaic_1km",
         "https://pubs.usgs.gov/sim/3168/sim3168_pamphlet.pdf",
         "https://opus.pds-rings.seti.org/opus/api/metadata/go-ssi-c0359986578.json",
         "https://ntrs.nasa.gov/citations/19860063116",
-        "https://doi.org/10.1029/2000JE001350"
+        "https://doi.org/10.1029/2000JE001350",
+        "https://ntrs.nasa.gov/citations/19840015363"
       ]
     },
     {
diff --git a/src/objects/miranda/investigations.json b/src/objects/miranda/investigations.json
index e35b5ed1b35..0b9da43ea40 100644
--- a/src/objects/miranda/investigations.json
+++ b/src/objects/miranda/investigations.json
@@ -85,7 +85,7 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "unresolved",
-      "finding": "Helfenstein, Veverka and Thomas (1988) fitted Hapke models to Voyager whole-disc photometry of the five large Uranian moons. Only the abstract could be read (single-scattering albedo from 0.3 for Umbriel to 0.7 for Ariel, g near -0.3), and a whole-disc fit states no emission range. The summary of Hillier, Helfenstein and Veverka in NASA's Reports of Planetary Geology and Geophysics Program 1987 (TM-4041) is open: Minnaert parameters fitted to Voyager clear-filter images (0.48 micrometres) at 13 to 43 degrees phase, with the global k ranging from about 0.6 at 13 degrees to about 0.8 at 43 degrees, nearly independent of wavelength, and no Voyager data below 13 degrees. Two rounded end values are not a law: the k at each phase is in the closed paper (Hillier et al. 1989), and nothing constrains the flood-lit view at zero phase. The open scans of the 1988 and 1989 Lunar and Planetary Science abstracts hold none with Hillier as first author. Schenk's 2020 mosaic history names ISIS photomet without its parameter file. Miranda keeps the shared sphere bank.",
+      "finding": "Helfenstein, Veverka and Thomas (1988) fitted Hapke models to Voyager whole-disc photometry of the five large Uranian moons. Only the abstract could be read (single-scattering albedo from 0.3 for Umbriel to 0.7 for Ariel, g near -0.3), and a whole-disc fit states no emission range. The summary of Hillier, Helfenstein and Veverka in NASA's Reports of Planetary Geology and Geophysics Program 1987 (TM-4041) is open: Minnaert parameters fitted to Voyager clear-filter images (0.48 micrometres) at 13 to 43 degrees phase, with the global k ranging from about 0.6 at 13 degrees to about 0.8 at 43 degrees, nearly independent of wavelength, and no Voyager data below 13 degrees. Two rounded end values are not a law: the k at each phase is in the closed paper (Hillier et al. 1989), and nothing constrains the flood-lit view at zero phase. The open scans of the 1988 and 1989 Lunar and Planetary Science abstracts hold none with Hillier as first author. Schenk's 2020 mosaic history names ISIS photomet without its parameter file. A check on 2026-10-05 on the one calibrated clear frame that holds the whole disc under 90 degrees phase (FDS 26844.09, 16.5 degrees) gives a lunar-Lambert weight of 0.80; a Minnaert k of 0.6 at 13 degrees corresponds to 0.74. One frame cannot confirm how k rises with phase. Miranda keeps the shared sphere bank.",
       "evidence": [
         "https://doi.org/10.1016/0019-1035(88)90039-5",
         "https://ntrs.nasa.gov/citations/19880051927",
diff --git a/src/objects/oberon/investigations.json b/src/objects/oberon/investigations.json
index 3b974e45552..bf0a9c517a3 100644
--- a/src/objects/oberon/investigations.json
+++ b/src/objects/oberon/investigations.json
@@ -101,7 +101,7 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "included",
-      "finding": "Hapke law of Helfenstein, Hillier, Weitz and Veverka, read in their summary in NASA's Reports of Planetary Geology and Geophysics Program 1989 (TM-4210, pages 230 and 231); the Icarus paper (1991) was not read. They fitted Hapke's 1986 model to Voyager imaging and Earth-based photometry. Table I, global average, clear filter (0.48 micrometres): single-scattering albedo 0.41, asymmetry -0.29, roughness 21 degrees, with the opposition surge fixed at h = 0.007 and B0 = 1.03. Flood-lit, the limb is 0.97 of the centre; the narrow opposition surge puts the point under the Sun at 0.74 of that centre at 1 degree phase and 0.55 at 10 degrees, so frames with Shadows on are dim against the flood-lit view. The table also prints green and violet rows and three terrains; one clear-filter global row lights every channel and terrain here. It is second class: the report prints no phase or emission range. The phase range 18 to 148 degrees is the span of the resolved Voyager 2 frames the PDS node lists, and the emission limit 84.9 degrees comes from the outermost pixel of the frame the report names as its finest (FDS 26836.25, 6.130 km per pixel); both are derived here. Helfenstein, Veverka and Thomas (1988) fitted Hapke models to whole-disc photometry of the five moons; that paper could not be read. Thomas and Veverka's 1986 summary gives a Minnaert k near 0.7 for Oberon at 33 degrees phase, one phase only.",
+      "finding": "Hapke law of Helfenstein, Hillier, Weitz and Veverka, read in their summary in NASA's Reports of Planetary Geology and Geophysics Program 1989 (TM-4210, pages 230 and 231); the Icarus paper (1991) was not read. They fitted Hapke's 1986 model to Voyager imaging and Earth-based photometry. Table I, global average, clear filter (0.48 micrometres): single-scattering albedo 0.41, asymmetry -0.29, roughness 21 degrees, with the opposition surge fixed at h = 0.007 and B0 = 1.03. Flood-lit, the limb is 0.97 of the centre; the narrow opposition surge puts the point under the Sun at 0.74 of that centre at 1 degree phase and 0.55 at 10 degrees, so frames with Shadows on are dim against the flood-lit view. The table also prints green and violet rows and three terrains; one clear-filter global row lights every channel and terrain here. It is second class: the report prints no phase or emission range. The phase range 18 to 148 degrees is the span of the resolved Voyager 2 frames the PDS node lists, and the emission limit 84.9 degrees comes from the outermost pixel of the frame the report names as its finest (FDS 26836.25, 6.130 km per pixel); both are derived here. Helfenstein, Veverka and Thomas (1988) fitted Hapke models to whole-disc photometry of the five moons; that paper could not be read. Thomas and Veverka's 1986 summary gives a Minnaert k near 0.7 for Oberon at 33 degrees phase, one phase only. The same check that Titania passes, run on 2026-10-05 on two calibrated clear frames at 19 and 39 degrees phase (FDS 26809.04 and 26836.23), gives lunar-Lambert weights of 0.60 and 0.39 where this law gives 0.74 and 0.62; a Minnaert k of 0.7 at 33 degrees corresponds to 0.54. Two frames of a disc with uneven albedo do not outweigh two published fits, so the law stands, with this noted.",
       "evidence": [
         "https://ntrs.nasa.gov/citations/19900018290",
         "https://doi.org/10.1016/0019-1035(91)90065-2",
diff --git a/src/objects/titania/README.md b/src/objects/titania/README.md
index 8e377c12bd0..cc3d42877f1 100644
--- a/src/objects/titania/README.md
+++ b/src/objects/titania/README.md
@@ -49,7 +49,7 @@ The checked geology landmark differences are 0.33–0.83° (approximately 4.5–
 
 ## Known problems
 
-- **Lighting law:** The fit is to whole-disc photometry, not to resolved images, and one row lights every terrain. The surge amplitude 0.65 is computed here from the printed S(0); the phase range and the emission limit 86.2° are derived here. The opposition surge puts the point under the Sun at 0.70 of the flood-lit centre at 10° phase, so frames with Shadows on are dimmer than the flood-lit view.
+- **Lighting law:** The fit is to whole-disc photometry, not to resolved images, and one row lights every terrain. Checked against five calibrated Voyager frames from 0.8° to 70° phase, it matches the resolved limb to 0.06 in lunar-Lambert weight (see the [ledger](investigations.json)). The surge amplitude 0.65 is computed here from the printed S(0); the phase range and the emission limit 86.2° are derived here. The opposition surge puts the point under the Sun at 0.70 of the flood-lit centre at 10° phase, so frames with Shadows on are dimmer than the flood-lit view.
 - The Voyager color dataset is false color (green, violet, ultraviolet as red, green, blue) at the observations' own phase angles. A color seam at a footprint edge is a real difference in viewing geometry. Its band ratios are read from a figure at ±0.02, and the ultraviolet calibration carries a stated ±10 % uncertainty. The brightest 0.1 % of texels may clip.
 - Approximate source coverage is 44.8% for monochrome and 27.7% for elevation before interpolation; the unobserved north stays missing.
 - Monochrome processing depends on an unavailable photometric parameter file (`eu_pho10.pvl`). DN values are not calibrated albedo.
diff --git a/src/objects/titania/investigations.json b/src/objects/titania/investigations.json
index 2b005672cbd..9e9fb25242a 100644
--- a/src/objects/titania/investigations.json
+++ b/src/objects/titania/investigations.json
@@ -98,7 +98,7 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "included",
-      "finding": "Hapke law of Veverka et al. (1987), read in the table Verbiscer and Veverka print in NASA's Reports of Planetary Geology and Geophysics Program 1987 (TM-4041, page 181); the paper itself was not read. Titania's row: w = 0.48, h = 0.018, S(0) = 0.77, roughness 23 degrees, g = -0.28, Hapke's 1986 equation. The model here takes the opposition amplitude B0, so S(0) is converted with the authors' definition, B0 = S(0)/(w P(0)), printed in the 1986 report: B0 = 0.65. Flood-lit, the limb is 0.97 of the centre; the narrow opposition surge puts the point under the Sun at 0.70 of that centre at 10 degrees phase. It is second class twice over: the fit is to whole-disc photometry, taken as a limb law because the same group reports that whole-disc and disc-resolved fits agree excellently for Titania (TM-4130, page 238, citing Thomas et al. 1987); and neither range is printed. The phase range 0.8 to 152.6 degrees is the span of the resolved Voyager 2 frames the PDS node lists, and the emission limit 86.2 degrees comes from the outermost pixel of the finest whole-disc clear frame (C2684315, 3.403 km per pixel); both are derived here. Thomas and Veverka's 1986 summary gives a Minnaert k near 0.7 at 33 degrees phase, one phase only. Helfenstein, Veverka and Thomas (1988) could not be read.",
+      "finding": "Hapke law of Veverka et al. (1987), read in the table Verbiscer and Veverka print in NASA's Reports of Planetary Geology and Geophysics Program 1987 (TM-4041, page 181); the paper itself was not read. Titania's row: w = 0.48, h = 0.018, S(0) = 0.77, roughness 23 degrees, g = -0.28, Hapke's 1986 equation. The model here takes the opposition amplitude B0, so S(0) is converted with the authors' definition, B0 = S(0)/(w P(0)), printed in the 1986 report: B0 = 0.65. Flood-lit, the limb is 0.97 of the centre; the narrow opposition surge puts the point under the Sun at 0.70 of that centre at 10 degrees phase. It is second class twice over: the fit is to whole-disc photometry, taken as a limb law because the same group reports that whole-disc and disc-resolved fits agree excellently for Titania (TM-4130, page 238, citing Thomas et al. 1987); and neither range is printed. The phase range 0.8 to 152.6 degrees is the span of the resolved Voyager 2 frames the PDS node lists, and the emission limit 86.2 degrees comes from the outermost pixel of the finest whole-disc clear frame (C2684315, 3.403 km per pixel); both are derived here. Thomas and Veverka's 1986 summary gives a Minnaert k near 0.7 at 33 degrees phase, one phase only. Helfenstein, Veverka and Thomas (1988) could not be read. Checked on 2026-10-05 against five calibrated Voyager 2 clear frames at 0.8, 8.9, 16.9, 33.9 and 69.9 degrees phase (FDS 26815.17, 26825.24, 26830.39, 26836.49 and 26843.13): the lunar-Lambert weight that best fits the lit disc inside 75 degrees incidence and emission is 0.88, 0.78, 0.70, 0.56 and 0.36, and this law gives 0.94, 0.80, 0.72, 0.60 and 0.31 for the same geometry. The whole-disc fit describes the resolved limb, the flood-lit view included.",
       "evidence": [
         "https://ntrs.nasa.gov/citations/19880017749",
         "https://doi.org/10.1029/JA092iA13p14895",
diff --git a/src/objects/umbriel/investigations.json b/src/objects/umbriel/investigations.json
index 1ff8790e065..1ddf2b0c244 100644
--- a/src/objects/umbriel/investigations.json
+++ b/src/objects/umbriel/investigations.json
@@ -97,7 +97,7 @@
       "id": "limb-law",
       "subject": "Photometric law for the lighting overlays",
       "status": "included",
-      "finding": "Hapke law of Helfenstein and Veverka (1988), an open conference abstract (Lunar and Planetary Science XIX, 477). They fitted Hapke's 1986 equation to disc-resolved brightness in five Voyager clear-filter images at 10.4 to 142.9 degrees phase. Table I, disc-resolved row: w = 0.34, h = 0.06 (fixed at the whole-disc value), S(0) = 1.28, g = -0.18, roughness 28 degrees. The model here takes the opposition amplitude B0, so S(0) is converted with the authors' own definition, B0 = S(0)/(w P(0)), printed in their 1986 NASA summary on the Moon: B0 = 2.145. Flood-lit, the limb is 0.99 of the centre; the opposition surge puts the point under the Sun at 0.61 of that centre at 10 degrees phase, so frames with Shadows on are dimmer than the flood-lit view. It is second class: no emission range is printed, and the limit 84.6 degrees is derived here from the outermost pixel of the finest of the five frames (FDS 26840.04, 5.167 km per pixel). The authors' later summary in NASA TM-4130 fits whole-disc and disc-resolved data together (0.372, -0.235, 26 degrees) but prints no opposition surge values, so the abstract's complete row is used. Helfenstein, Veverka and Thomas (1988), the whole-disc fits of the five moons, could not be read.",
+      "finding": "Hapke law of Helfenstein and Veverka (1988), an open conference abstract (Lunar and Planetary Science XIX, 477). They fitted Hapke's 1986 equation to disc-resolved brightness in five Voyager clear-filter images at 10.4 to 142.9 degrees phase. Table I, disc-resolved row: w = 0.34, h = 0.06 (fixed at the whole-disc value), S(0) = 1.28, g = -0.18, roughness 28 degrees. The model here takes the opposition amplitude B0, so S(0) is converted with the authors' own definition, B0 = S(0)/(w P(0)), printed in their 1986 NASA summary on the Moon: B0 = 2.145. Flood-lit, the limb is 0.99 of the centre; the opposition surge puts the point under the Sun at 0.61 of that centre at 10 degrees phase, so frames with Shadows on are dimmer than the flood-lit view. It is second class: no emission range is printed, and the limit 84.6 degrees is derived here from the outermost pixel of the finest of the five frames (FDS 26840.04, 5.167 km per pixel). The authors' later summary in NASA TM-4130 fits whole-disc and disc-resolved data together (0.372, -0.235, 26 degrees) but prints no opposition surge values, so the abstract's complete row is used. Helfenstein, Veverka and Thomas (1988), the whole-disc fits of the five moons, could not be read. A check on 2026-10-05 on three of the five frames gives lunar-Lambert weights of 0.66, 0.62 and 0.59 at 12, 31 and 56 degrees phase where this law gives 0.74, 0.56 and 0.35. It includes the bright annulus near the limb, which the authors left out of their fit, so it does not test the law.",
       "evidence": [
         "https://ui.adsabs.harvard.edu/abs/1988LPI....19..477H",
         "https://articles.adsabs.harvard.edu/pdf/1988LPI....19..477H",