# Cross-Dock Facility Register and Reference Ledgers: Data Dictionary

**Version:** 1.1  
**Source verification date:** September 14, 2026  
**Compiler:** Harrisburg Dock Door Repair Research

## What the register contains

The 23-entry register contains 18 documented facility entries, four excluded source leads and one historical carrier-level benchmark. Eleven facility entries have accepted published-area and exact door/dock-count inputs. It is a purposive compilation of public documents, not a representative survey, current inventory or inspection of operations. The original contribution is the source join, evidence classification, accepted-input arithmetic and recorded exclusions.

## Files and source provenance

The input CSV feeds the calculation script. The register CSV and JSON contain the same records and derived fields. The source CSV supplies titles, URLs, document vintages and source-check dates. The geometry ledger, federal reference ledger and transaction-bundles CSV preserve separate evidence categories; they are not additional facility records.

## Facility input and output fields

| Field | Definition and interpretation limit |
|---|---|
| `record_id` | Stable identifier F01–F23; it persists even when a candidate is excluded. |
| `record_kind` | One of documented_facility, excluded_source_lead or historical_network_benchmark. Only documented_facility records can enter facility arithmetic. |
| `facility` | Source-described name or property identity; a historical name does not assert current ownership or operation. |
| `city_or_source_location` | The locality or address label supplied by the source, not automatically the parcel municipality. |
| `state_or_country` | US state abbreviation or country label retained for geographic grouping. |
| `county_or_locality_association` | County expressly identified in a primary facility source, or a qualified locality association. An empty value means no accepted county assignment. |
| `evidence_class` | Type and scope of the source, not a measured diagnosis of activity inside the building. |
| `doors` | Exact published point count, retained only where accepted from a primary document. Read with door_denominator_label. |
| `door_denominator_label` | Original category, such as doors, dock-high doors, docks or usable doors; these labels are not assumed interchangeable. |
| `published_area` | Positive area value in the source’s original unit. It is not necessarily dedicated transfer-floor space. |
| `area_unit` | ft2 for square feet or m2 for square meters. Blank when no accepted area is published. |
| `area_scope` | The source’s area concept, such as Total Sq. Ft., facility area or building area. This constrains interpretation. |
| `site_acres` | Published acreage used in calculations; conflicting or unreconciled boundaries remain in notes. Not an independent parcel survey. |
| `documented_individual_price_usd` | Accepted single-property price in US dollars. Empty throughout version 1.1; verified package amounts live in the separate bundles table. |
| `trailer_spaces` | Accepted exact point count in the source’s trailer-space category. More-than counts and possible future expansions are not entered as exact completed parking counts. |
| `trailer_count_qualifier` | Source wording, scope or reason a trailer-space value is not suitable for a point calculation. |
| `clear_height_ft` | Published clear height in feet, when provided by the selected source. |
| `building_depth_ft` | Published building depth in feet, not automatically the transfer-floor width used in a historical model. |
| `truck_court_depth_ft` | Published truck-court depth in feet; a property specification, not a universal design rule. |
| `source_ids` | Identifiers joining to cross-dock-sources.csv; multiple supporting documents are retained together. |
| `source_urls` | Primary-document URLs supporting the record. Sources and dates must accompany reused facts. |
| `source_vintage` | Publication, filing, release or historical period of the underlying statement. Undated and conflicting documents are labeled. |
| `verification_outcome` | Outcome of this source review, including accepted scope or exclusion. It is not an independent operating audit. |
| `notes` | Material qualifiers, attribution limits, source conflicts, excluded inputs and historical context. |
| `verified_date` | Date this compilation checked the cited source: 2026-09-14. It does not redetermine the measurement’s vintage. |
| `building_sqft` | Derived square-foot equivalent of published_area. Legacy field name: does not narrow a source’s broader property-area scope. m2 inputs are converted using the exact international foot. |
| `sqft_per_door` | building area equivalent ÷ doors, calculated from unrounded inputs and rounded once to one decimal. A source ratio, not utilization, dwell time or throughput. |
| `doors_per_acre` | doors ÷ site_acres, rounded to two decimals. Not comparable to square feet per door. |
| `price_per_door_usd` | documented_individual_price_usd ÷ doors. Empty in this release because individual prices were not established. |
| `published_area_to_site_area_pct` | 100 × square-foot area ÷ (site_acres × 43,560), rounded to two decimals. Not a measured footprint, land-use mix or zoning coverage determination. |
| `trailer_spaces_per_door` | Accepted exact trailer-space count ÷ doors, rounded to two decimals. Its interpretation inherits the source’s trailer count qualifier. |

## Nulls, units and precision

Missing or rejected values are empty strings in both CSV and the record objects in JSON. They are not zero, absence of the feature or proof that no source exists elsewhere. Documentary notes can describe a non-point quantity while its numeric input remains blank.

The conversion uses exactly 0.3048 meters per international foot, so square feet = square meters ÷ 0.3048². Acres are converted using 43,560 square feet per acre. Exact unit definitions do not increase the measurement precision of a company’s published building area. Source units and area scope are retained.

The script uses Decimal arithmetic and ROUND_HALF_UP. Area per door is rounded once to one decimal; doors per acre, trailer spaces per door and the site-area percentage use two decimals. The Rotterdam square-foot conversion is stored to six decimals only to preserve the calculation path. Displayed extra decimals are not claimed survey precision. The Rosedale equipped-dock share in the separate findings file is rounded to one decimal.

## Reproducing the arithmetic

With Python 3.10 or later installed, extract the dataset into one folder and run:

```sh
python cross-dock-dataset-build-script.py
```

The script locates its input CSV beside itself, validates record kinds and positive numeric inputs, and rewrites cross-dock-facility-register.csv, cross-dock-facility-register.json and cross-dock-derived-findings.json. It does not scrape websites, contact operators, validate current availability or reproduce the 2004 research simulation. The included dated inputs produce the results on the page.

The headline comparison is `(1,219,021 / 224) / (30,000 / 55) = 9.98`, rounded from the unrounded ratios. Core5’s specified area and acreage produce `1,219,021 / (196.27 × 43,560) × 100 = 14.26%`; its acreage conflict remains in the record. Pyle’s Camp Hill count and stated acres give `85 / 11 = 7.73 doors per acre`.

The script does not assert that all area or door labels share a measurement definition. The six estate inventory records use that document’s Total Sq. Ft. scope; Maersk supplies docks rather than an independently counted set of openings.

## Geometry and provenance ledger

The 28 rows distinguish the 2004 paper’s assumptions, geometry, model results and historical observations from its attributions to unpublished private correspondence. ★ means the primary paper’s statement was checked. ● means the attribution in the paper was checked, not the underlying correspondence or measurement. All rows keep the historical source vintage; none is a new 2026 survey.

Fields identify the parameter, published value, basis, paper location, verification tier, interpretation note, source URL, vintage, check scope and verification date. Approximate thresholds remain approximate. Historical floor-width and door-offset values are not current engineering standards.

## Federal reference ledger

The seven rows distinguish PHMSA interpretation, FDA rule scope and inflation thresholds, and NAICS statistical classifications. NAICS is not a safety code or operating permit. The 2026 FDA cutoff applies the agency’s stated averaging basis; it is not a new universal revenue threshold for every food-handling business. Re-check applicable primary material before relying on a rule for a specific operation.

## Transaction bundles

Two records preserve the successful-bid amounts in Yellow estate Doc. 1268, Exhibit 2: $870 million for XPO’s 28-property package and $29.4 million for Pyle’s four-property package. These are filed package values, not individual sale prices, appraised values or independently verified closing payments. No amount is allocated to a single facility or divided by that facility’s doors.

## Version 1.1 changes and limits

This version corrects record counts and units; preserves the source conflict in Core5’s spacing and acreage; converts the Rotterdam area rather than excluding it; uses direct estate inventory evidence for accepted property inputs; removes unsupported individual prices from numeric fields; and replaces a claimed operating-versus-building split with a bounded source comparison. All original candidate entries remain traceable, including excluded leads and the historical benchmark.

Checking the sources in September 2026 does not make 2019 announcements, 2023 inventory, 2024 building descriptions or 2001 correspondence current operating measurements. No surveys, site visits, shipment-log analysis, interviews, national inventory or causal price study were conducted.

## Source and rights attribution

The compiler’s original work is the assembly and arithmetic. The company, agency, court and research statements retain their original attribution. This package makes no claim of exclusive ownership over the underlying public facts and does not grant rights over third-party documents. Primary sources are linked, not redistributed in this dataset.