Union ConfederateHollow = individual positions unresolved
ELAPSED RECONSTRUCTION00:00
The lines step off20,071 selected infantry identities
PENNSYLVANIA · UNITED STATES
N
CURRENT VIEWThe battlefieldLoading terrain…
G
BUILDING THE BATTLEFIELD

Reading elevation data

SPECTATORDrag to orbit · Scroll to explore
THE EVIDENCE BEHIND THE LANDSCAPE

Sources & accuracy

A measured terrain base, with a map-informed reconstruction of the July 3 landscape. The infantry replay follows the American Battlefield Trust phase-map sequence, NPS research and official reports. Selected strengths, inferred formations and timing are explained in Counts & reconstruction.

01 · MEASURED, MODERN

The shape of the ground

USGS 3DEP bare-earth elevation, sampled on a 257 × 257 grid at approximately 12 m spacing. Heights and horizontal distances use meters, with no vertical exaggeration. Modern cuts, roads and later landscape alterations may remain in the data.

USGS 3DEP elevation service ↗Retrieved October 22, 2026 · 39.794–39.822° N, 77.261–77.223° W
02 · HISTORICAL REFERENCE

Roads, farms & defensive ground

Farm lanes, orchards, woodlots, Emmitsburg Road and selected physical fence runs are approximately traced from the Warren survey. Land-cover colors are reconstructed; they do not denote property ownership or prove enclosure. The 1868–69 postbattle survey is interpreted alongside NPS research and Bachelder’s 1863 perspective. A local four-point map fit has a 24 m control residual; this is an approximate alignment, not survey-grade 1863 geography.

Warren survey · Library of Congress ↗Historic Warren survey crop showing fields, farms, orchards, roads and fence symbols in the assault sectorView the map used for tracing ↗NPS historical base-map study ↗
03 · HISTORICAL CONTEXT

The assault sector

The playback uses a fixed selected infantry roster. Unit locations are approximately registered against farm landmarks, with reconstructed two-rank deployment and internal allocations. The distant symbols summarize those same formations; they do not independently create soldiers. The source’s later maps combine actions across a time interval and cannot serve as exact simultaneous snapshots.

American Battlefield Trust · phase-map series ↗

The study frames the terrain crossed in the July 3 assault. Seminary Ridge is a broad reference area, not a precise starting position for every formation. The attack included troops from Pickett’s, Pettigrew’s and Trimble’s commands.

NPS Pickett’s Charge guide ↗1876 battlefield maps · Bachelder / Warren ↗
04 · ILLUSTRATIVE RECONSTRUCTION

Fences are not parcel boundaries

Warren’s legend identifies stone walls with continuous dots, post-and-rail fences with dots and dashes, and worm fences with zigzags. Only identified physical-fence symbols produce 3D fencing. Smooth contours, ownership labels and land-cover boundaries do not. The two east-ridge rail-to-stone transitions follow visible changes from zigzags to continuous dots. Their junctions and road-verge endpoints are now connected; color changes alone never produce a fence.

Warren map legend identifying continuous dots as stone wall, dots and dashes as post-and-rail fence, and zigzags as worm fenceInspect the original fence legend ↗Warren map detail with the two zigzag to dotted stone-wall transitions identifiedInspect the material changes ↗

How the fences are built

Worm fencing uses overlapping split rails in a broad zigzag, with slender crossed stakes and a top rider. Post-and-rail runs have five rails; low stone walls use irregular, closely fitted stones. NPS research supports 10–12-foot worm-fence rails and commonly four or five rails in post-and-rail fencing. The chosen heights, bracing pattern, weathering and individual rails are reconstructed, not measured from each 1863 fence.

What remains provisional

Bryan’s cottage silhouette follows an 1863 photograph; Codori uses the documented low English-style barn, and Bliss remains burned ruins. Building dimensions, ancillary structures, gardens, orchard spacing and density, exact tree positions, crop assignments, rail details and gate openings are reconstructed. Postbattle map features may differ from July 1863. The south end of the Angle stone wall is estimated from the NPS account of its roughly 260-yard extent; that material interface is provisional. Access openings follow farm tracks, with reconstructed widths. Some traces continue into woods beyond the detailed reconstruction. This is not an inventory of intact panels at a precise moment in the battle. The copse is not an exact tree inventory.

Bryan house · July 1863 photograph ↗NPS cultural landscape inventory · historic orchards ↗U.S. Army history · roadside fences ↗NPS · Abraham Brian farm ↗

Only the Codori, Bryan and Bliss farm groups are developed here. Orchard existence is documented; individual trees, spacing and density are reconstructed. Peripheral farm groups await further work. The infantry movement replay is an interpretation, with no person-level casualty or artillery simulation.

Prototype 0.2 · Historical landscape study · Three.js (MIT) · USGS elevation
MOVE THROUGH THE LANDSCAPE

Spectator controls

Orbit
Drag with one finger or the left mouse button.
Pan
Right-drag, two-finger drag, or use the arrow keys while the terrain is focused.
Zoom
Scroll, pinch, or use the + and − buttons.
Inspect
Select a map label, then choose “View this location.”
Get your bearings
Choose Overhead map or Reset view.

Camera movement does not advance time. Terrain is shown at its natural scale. Touch controls are supported; this first version is best explored on a larger screen.

HISTORY & RECONSTRUCTION

A ledger, not a roll call.

20,071 persistent identities represent a selected infantry scenario: 11,482 in the main Confederate attack, 1,600 in the later Wilcox–Lang advance, and 6,989 Union infantry in this sector. These estimates are not exact participant totals or counts of survivors. Officers and representative color bearers use existing identities.

What the replay reconstructs

Direction and sequence follow the American Battlefield Trust phase maps, official reports and NPS studies. The selected 40-minute elapsed timeline is not a synchronized historical clock. Maps combine actions across intervals; exact minute marks, individual routes, wheel geometry, subdivisions, rank assignments and firing cycles are inferred.

Two ranks use 0.55 m per file with supporting lines behind. Symbol widths summarize these reconstructed frontages. Buildings and tree trunks have clearance envelopes; fence crossing motions are illustrative. The number of intact fence panels on July 3 is not known.

Where the evidence stops

When an attack is repulsed, its identities remain in the ledger but their exact positions become unresolved. Hollow markers show the collective withdrawal direction. They are not all survivors marching home. Deaths, wounds, prisoners and individual return paths are not assigned.

This is the central infantry sector, not both complete armies. The preliminary cannonade, artillery crews, mounted headquarters, ambulances, teamsters and further reinforcements are not yet represented. Musicians’ precise battle locations and Confederate regimental colors remain unresolved. Uniforms and close-up drill are representative, not individual portraits.

Internal allocations

Webb’s 72nd PA uses 380; the remaining 515 are provisionally divided 257/258 between 69th/71st elements. Stannard’s selected total is divided 572/571/572 among the 13th/14th/16th VT; four 14th VT companies use 228 of those identities. Sherrill’s 150 detached New Yorkers are subtracted from the parent budget. Hall/Harrow reinforcing elements use 40%, Gates half of Biddle, and Armistead’s leading group 300. These are explicit reconstruction choices, not recovered regimental returns.

ArmyFormationSelectedBasis & limitations

Camera changes only the view. Pause, rewind and seek preserve the same identities and world time.