Recovering historical data from physically sealed documents presents a fundamental trade-off: breaking the physical seal destroys structural evidence of early modern security practices, while leaving the artifact intact leaves the textual data inaccessible. The Brienne Collection—a postal trunk containing 3,148 cataloged items from the late 17th century, including 577 entirely unopened letterpackets—exemplifies this bottleneck. Historically, extracting information from these artifacts required mechanical forcing, destroying the geometric configuration known as letterlocking. Modern computational analysis resolves this dilemma by substituting destructive physical intervention with non-invasive volumetric imaging and algorithmic structural flattening.
The operational architecture capable of reading these intact documents rests on two interdependent technological pillars: high-sensitivity X-ray microtomography (XMT) and automated virtual unfolding algorithms. This methodology preserves the structural cryptography of the paper while extracting high-fidelity digital representations of hidden text. Recently making news lately: What Most People Get Wrong About AI in Hong Kong Classrooms.
The Volumetric Data Acquisition Framework
Extracting text from inside a multi-layered paper packet requires an imaging system capable of resolving micro-scale structural variations without altering the physical substrate. Standard medical CT scanners lack the spatial and contrast resolution required for this task. Instead, the process utilizes high-sensitivity X-ray microtomography (XMT) systems originally developed for dental research to map mineral densities in teeth.
[XMT Volumetric Scan] ---> [Voxel Intensity Grid] ---> [Segmentation] ---> [Algorithmic Flattening]
The data acquisition phase relies on a specific chemical contrast mechanism. The writing materials used in 17th-century Europe were primarily iron gall inks, which contain high concentrations of heavy metal ions, specifically iron ($Fe$). The paper substrate consists of organic cellulose fibers with low atomic mass. When high-intensity X-ray beams pass through the folded packet, the iron atoms in the ink absorb significantly more radiation than the surrounding paper, creating a stark differential in X-ray attenuation. Further insights regarding the matter are detailed by Wired.
The scanner records a series of two-dimensional angular projections, which are computationally reconstructed into a 3D volumetric dataset composed of voxels (three-dimensional pixels). The resolution of these voxels must be fine enough to distinguish individual paper layers, which typically measure between 100 to 150 micrometers in thickness. This stage generates a high-density voxel matrix where intensity values correspond directly to material density, mapping the physical topography of both the paper and the ink distribution.
The Computational Unfolding Pipeline
Once the 3D voxel grid is established, the data must be transformed from a complex, interleaved geometric state into a legible 2D plane. This transformation is executed through an automated computational pipeline that isolates the writing surface and calculates its flattened topology. The pipeline operates via four sequential engineering phases.
1. Segmentation and Layer Tracking
The algorithm analyzes the volumetric data to identify the continuous sheet of paper within the compressed packet. Because the paper is tightly folded, multiple layers lie in direct contact. The software uses intensity gradients and spatial continuity rules to segment the single primary writing substrate from the air gaps and adjacent layers, tracking the sheet across complex folds, tucks, and slits.
2. Crease Pattern Mapping
As the algorithm maps the geometry of the paper, it identifies and classifies folds based on surface curvature. Highly localized, sharp deformations indicate interior creases, which appear in the data as thin, saturated lines of high curvature. Rounded, gentle deformations indicate exterior folds where the paper bent around other layers. This generates a definitive digital map of the document's crease pattern.
3. Text-to-Surface Projection
The voxels representing the iron-rich ink must be mapped precisely to their corresponding location on the segmented paper surface. Because text is written on both sides of the sheet, the algorithm must calculate the depth of the ink within the paper volume to assign it to either the front or the back face. This prevents the legibility errors that occur when overlapping text from opposing pages bleeds together in a simple vertical projection.
4. Isometric Flattening
The final phase projects the 3D surface mesh onto a 2D coordinate system. The algorithm minimizes geometric distortion—specifically stretching and tearing—by optimizing for an isometric mapping, which preserves the true surface distances between points on the paper. The output is a flat, highly legible image of the original letter accompanied by an exact blueprint of its structural folds.
Classifying Physical Cryptography
The utility of virtual unwrapping extends beyond text extraction; it allows for the systematic analysis of "letterlocking," an analog form of physical cryptography used from the 1500s until the widespread adoption of mass-manufactured envelopes in the 1830s. Letterlocking used specific folding sequences, slits, and paper locks to ensure that any unauthorized interception and reading of a document would result in irreversible physical damage, alerting the recipient to the security breach.
Analysis of over 250,000 historical documents has revealed that letterlocking operates as a structured security system. The Unlocking History Research Group codified these techniques into a standardized matrix categorized by explicit design variables:
- Folds and Tucks: The foundational geometry. Documents are scored and folded into narrow packets, with the final edges tucked into internal flaps to prevent easy opening.
- Slits and Holes: Structural penetrations cut into the paper substrate. A secondary strip of paper, or a corner torn from the letter itself, is threaded through these slits to bind the layers together.
- Paper Locks: Mechanisms where an external paper strip or thread is passed through the document and secured, often bound with sealing wax. Tearing the lock is mandatory to access the text.
The virtual unfolding of letterpacket DB-1627 from the Brienne Collection demonstrated the execution of a low-security locking format. The document, a letter dated July 31, 1697, from Jacques Sennacques to his cousin Pierre Le Pers, was secured using basic folds and a paper lock. Because researchers could map the entire internal structure computationally, they verified that this specific format was designed to deter casual prying rather than to withstand dedicated espionage, as opening it required fewer destructive steps than more secure variants like the spiral stitch.
Operational Bottlenecks and Constraints
While virtual unfolding successfully extracts data from closed volumes, the methodology possesses clear technical and economic limitations that prevent immediate, large-scale deployment across global archives.
The primary constraint is computational density and processing velocity. While data collection via XMT can be completed relatively quickly, processing the resulting volumetric datasets through segmentation and flattening algorithms historically required up to several days per document. Optimizations in computational mathematics have reduced this timeline toward several hours per packet, yet the process remains too resource-intensive for processing millions of uncataloged items.
Furthermore, the technique relies entirely on specific material properties. If a historical document was written using carbon-based inks (such as lampblack) rather than iron gall ink, the atomic contrast between the ink and the paper drop significantly. Carbon inks do not possess the heavy metal concentrations required to create a distinct X-ray attenuation signature, rendering the text invisible under standard XMT configurations. Additionally, documents heavily degraded by mold, moisture, or chemical decay often exhibit structural fusing, where separate layers of paper merge into a single solid mass, causing the segmentation algorithms to fail.
The final strategic move for institutions managing compromised archival material is to employ a tiered assessment model. Virtual unwrapping should not be viewed as a universal replacement for physical conservation, but as a high-precision tool reserved for objects whose structural security configuration holds equal historical value to the text written upon them. Mass digitization efforts must prioritize documents with iron-based inks and stable physical geometries, while developing accelerated segmenting models to drive down per-unit computational costs.
For a visual demonstration of how these computational layers separate and unfold a historical document without destroying its physical seal, the virtual unfolding sequence of letter DB-1538 illustrates the exact transition from a 3D XMT scan to a legible 2D plane.
