Specimen 01 / 14 · collaborator edition

UV1642

A wide Cygnus field, recovered beneath the ink.

Calibrated field

Cygnus and a second Kap 41 field are independently recovered on the same glass. The tested cubic correction belongs to the original Cygnus component; the written Pole pointing remains unconfirmed.

Original scan: 4,955 × 6,153 px
Source data: 16-bit TIFF
Angular scale: 5.532″ / native px
Sky footprint: 7.631° × 9.403°

Full plate scan — 01Complete scan, with its stored orientation retained. Display derivative: recorded 1st–99.7th percentile linear stretch, 8-bit JPEG, reduced for sharing. Click any inspection image to enlarge; click again inside the viewer for its native display pixels.

01 /

What is photographic, and what is written?

Two independently recovered sky fields; a third written pointing remains open.

The margin records Kap 40, Kap 41 and Pole in three five-minute intervals on 15 November 1916. The original Cygnus solution registers one component. Removing its catalog matches reveals another coherent sky pattern, identified near RA 329.53938°, Dec +44.64607°. A bounded search has not yet recovered the written Pole pointing. Multiple pointings explain why unmatched stars should not automatically be called distortion outliers.

The photographic field

The full plate contains a dense star field and diffuse Milky Way structure. The bright star is Deneb; NGC 7000, IC 5070 and IC 5068 provide recognizable context. These identifications follow the recovered sky coordinates rather than the handwriting.

Writing and a revised date clue

The readable margin appears to say “UV1642”, “1916, Nov. 15”, “Kap 40”, “Kap 41” and “Seed 30”, with observing times. The apparent 1916 date differs from the initial circa-1930 provenance supplied for the collection. Treat this as an image-based reading pending a plate-log check. Internal strokes still overlap genuine sky structure.

Photographic detail — The central photographic field: stars and diffuse structure remain visible around internal marks. The display stretch is for inspection, not photometry.

Written context detail — A margin crop, rotated and mirrored for reading. The full-frame image and downloadable layers keep the stored scan orientation.

02 /

Separate the observed layers.

The revised layer uses local contrast in both polarities, expanded inspected writing regions and contextual recovery of short fragments and round digits. Known photographic targets and sampled spectral apertures are protected. It is not OCR and does not reconstruct hidden sky or invent pen strokes.

The selection covers 4.42% of this scan (1,348,526 pixels). Faint marks, scratches and contacts between ink and photographic images remain ambiguous. The cyan overlay shows the wider exclusion footprint used by the photographic remainder; the stroke mask is supplied separately.

Original scan with candidate annotation selection in cyan

Cyan shows the expanded exclusion area, including edge padding. Move the slider to inspect boundaries and incidental selections.

Native transparent writing PNG ↓Native stroke mask ↓Masked display image ↓Extraction record ↓

The writing PNG is a white, contrast-rendered stencil of bright and dark observed marks. The observed-pixel PNG preserves their stretched scan appearance. Both use original scan coordinates. The photographic remainder displays excluded pixels as black holes. It is neither a clean-sky reconstruction nor a new photometric product. Keep the mask alongside any measurements. Native original-intensity TIFF/FITS products, where generated, remain in the analysis workspace and are not duplicated in this handoff.

Annotation revision 2

This revision recovers short letter fragments, round digits and inspected dark marks, with broader writing regions. The stroke selection grows from 568,828 to 1,348,526 pixels (2.37×). The photographic remainder uses a separate 3-pixel exclusion margin, covering 11.31% of the scan.

The wider selection deliberately favors removing writing. Nearby stars, scratches and ambiguous fragments can also be excluded. Explicit photographic protection regions retain known target images and sampled spectra; this does not establish perfect separation elsewhere.

Annotation mask revision comparison — Previous photographic remainder at left; revised remainder at right. Black pixels are excluded, not restored sky. Both panels show the same scan. The native stroke and exclusion masks are supplied separately.

Previous mask ←→ Revised exclusion

Photographic remainder using expanded exclusion mask

Previous photographic remainder revealed at left

Inspect what changed 50%

Native exclusion mask ↓Observed annotation pixels ↓Newly selected strokes ↓Previous mask ↓Bright/dark polarity map ↓

03 /

The evidence for a sky identification.

A blind solve using compact, round sources ranked by integrated flux identified index 4111 with odds 4.82 × 10¹⁷. The batch copy has the same SHA-256 as the original UV1642 scan, so the earlier calibration is reused rather than treated as an independent observation. Reassessment identifies a second, superposed Kap 41 sky field; its separate WCS and fresh reference checks are supplied below.

DSS2 survey context — DSS2 color survey; STScI / Digitized Sky Survey, distributed through CDS HiPS2FITS.

Historical plate ←→ DSS2 survey

DSS2 survey context in recovered plate orientation

Compare across the field 50%

The DSS2 image is resampled into the recovered plate orientation and footprint. Its photographic epochs, bands, resolution and contrast differ. Brightness and morphology need not match. It is context from the recovered WCS, not an additional independent fit check.

Catalog reference positions — Uniform point glyphs from the saved reference catalog projected through the plate WCS. This is a position-only rendering, not a survey photograph or brightness simulation.

Identified sky context — Sky labels from the saved field identification. These are coordinate-derived annotations; the original writing remains part of the scan.

DSS2 context JPEG ↓Survey request and attribution ↓

03c /

Exposure-aware reassessment.

Two independently recovered sky fields; a third written pointing remains open.

Second pointing: Kap 41

Center RA 329.53938023°, Dec +44.64607035°. After excluding reference stars near solver inputs, 578 additional associations have median/p90 residuals 0.830/1.750 native pixels. Shifted controls yield 34, 26, 21, 14, 24, 19 associations.

This is a separate coordinate frame for one sky/exposure component. Residuals use a finite association gate and no historical proper-motion propagation; they validate location rather than isolate glass distortion.

Additional reference support — Fresh photographic detections and additional catalog associations in this component. References near solver input positions were excluded before this check. Marked points are measured positions, not synthesized stars.

Component survey context — DSS2 color context for this component, distributed by STScI/CDS. Resampled through this component WCS and flipped into scan-row order. Different epochs, bands and photographic response preclude direct brightness comparison.

Component WCS ↓Additional-star checks ↓Additional associations FITS ↓

Across these scans, repeated photographic images need not imply the same mechanism. Shutter timing, guiding offsets, objective optics and different pointings can leave related patterns. Establish consistent component membership before fitting a smooth warp. No result here isolates glass deformation from telescope optics, emulsion, scanner geometry and centroid bias.

04 /

Distortion: what is measured, and what is not.

Of 1,255 initial reference associations, 959 training inliers and 232 held-out stars supported the comparison. Those held-out stars were also absent from the blind-solve input. Cubic TAN-SIP reduces the median prediction error from 0.737 to 0.524 original pixels (2.90 arcsec); the 90th percentile is 1.410 pixels. Fourth order adds little and fifth order degrades the result. These statistics describe the original Cygnus component only.

The portable correction is cubic TAN-SIP. A separate local residual interpolator reaches a 0.340-pixel median and 1.182-pixel 90th percentile, but is not baked into the rectified image. The smooth warp relative to TAN reaches 1.89 pixels inside the supported region; the whole-grid maximum of 4.71 pixels includes extrapolation. The rectified preview uses SIP-to-TAN cubic resampling, which is not flux conserving.

Model comparison · residuals in original scan pixels
Model	Held-out median	Held-out p90	Retained
TAN	0.737	1.581	—
TAN-SIP 2	0.694	1.421	—
TAN-SIP 3	0.524	1.410	Yes
TAN-SIP 4	0.515	1.399	—
TAN-SIP 5	0.720	1.668	—
SIP 3 + local residual model	0.340	1.182	Separate experiment

Held-out model-selection errors. The cyan marker is the retained portable model. These stars check the fit but are also used to choose model complexity: this is internal validation, not an external accuracy certification.

Spatial reference support — Gray marks are saved fit associations; cyan circles are held-out stars. Large white rings flag held-out residuals above two original pixels. Dimmed areas lie outside the association convex hull and should be treated as extrapolation.

Smooth cubic warp relative to TAN. Hatched regions lack reference support. A large edge value is not a measured glass displacement. Units are native scan pixels.

The held-out sample was selected within 3.5 binned pixels of the initial WCS. Model selection and that association gate limit how broadly these error summaries generalize. Catalog proper motions have not been propagated to the plate epoch. The readable UV1642 inscription appears to date the plate to 15 November 1916; the initial circa-1930 attribution is superseded for this specimen.

Rectified photographic preview — Cubic SIP-to-TAN rectification, now with revision-2 annotation exclusion applied. Black holes are excluded pixels, not reconstructed sky. The local residual interpolator is not applied. Rectification changes the pixel geometry.

Possible annotation–star intersections — Revised-mask catalog-proximity flags in original scan coordinates. These are possible star/ink intersections, not recovered hidden stars. Dark pixels are unflagged, not proof of clean signal.

Rectified writing PNG ↓Rectified exclusion mask ↓Star/ink uncertainty mask ↓Full distortion experiment ↓

Reference catalog subset (FITS) ↓Fit association table (FITS) ↓Warp grid (FITS) ↓Separate local residual model ↓Local model recipe ↓

Original-coordinate WCS ↓Solution record ↓Reference support and residuals ↓

05 /

Where the remaining questions live.

Select a numbered pin or a note. Locations are editorial review pointers, not measured confidence contours; plate-wide limitations also apply.

01Outer field

Sparse or absent reference support makes the edge correction an extrapolation. Use the dimmed coverage plot to locate the actual support boundary.

02Ink across the sky

Strokes can hide stars and diffuse emission. The revised catalog-proximity flags track the new selection; their count is recorded in the extraction JSON. They do not reconstruct what the writing obscures.

03Physical cause

The measured warp combines optical projection, emulsion, glass and scanning effects. One scan cannot isolate a glass-only distortion field.

The next useful experiment

For a physical glass measurement, scan the same plate again at a different rotation with a calibrated scanner, then compare the distortion in plate coordinates and scanner coordinates. For more precise astrometry, establish the plate date and propagate catalog proper motions to that epoch.

Present conclusion

This is the strongest distortion experiment in the set. The correction is empirically supported over the star-covered interior; annotation separation remains a candidate mask wherever ink and astronomical signal coincide.

06 /

An inspectable record for the next collaborator.

Source filename: UV1642.tif
Native dimensions: 4955 × 6153 pixels; I;16
Original SHA-256: 21c318a8897cc405bad623742c1c3749d336ba4eb95cb444c62e25565498b445
Display derivative: 2094 × 2600 px; JPEG; no photometric calibration
Display stretch: Linear; percentile 1 = 1, percentile 99.7 = 55890; clipped to 8 bits
Detail crops: Fractional original-coordinate boxes recorded in source.json. Long margins are rotated and mirrored for reading.
Layer coordinates: Native original scan size, except explicitly labeled rectified UV1642 writing
Original TIFF: Not included in this portable report; retained separately, unchanged
Analysis method: Local Astrometry.net 0.98; inspected source selection, exposure grouping where necessary; saved reference associations and internal held-out checks
Snapshot date: 27 October 2026
WCS center (RA, Dec): 312.34239040°, 46.22413013°
Survey context: DSS2 color / STScI, via CDS HiPS2FITS. Context image uses the recovered WCS and is not an extra validation set.

Source identity and display recipe ↓Full-frame display JPEG ↓

All scientific numbers are snapshots of the saved analysis products. The original 16-bit scans remain the measurement source. JPEG derivatives, contrast stretching and masked previews are inspection tools; do not derive photographic density or absolute response from their appearance.

Start of notebookR4011 →