The Broken Chain: A Zero-Dependency Single-File Field Voucher System for Botanical Specimen Documentation

1.Introduction

The problem is old and well understood. It is not well solved.

Herbarium specimens are the foundational data layer of plant systematics, floristics, conservation biology, and climate change research. Their scientific value depends entirely on the integrity of the associated locality data: who collected it, where, when, under what conditions, with what level of identification confidence. A pressed and mounted specimen without this information is, for most scientific purposes, a curiosity. It proves a plant existed somewhere at some point. It proves nothing else.

The data break is distributed across the entire chain and time. A specimen collected today will be pressed, dried, frozen for pest control, placed in a temporary cabinet, assigned to a photography queue, one-D barcoded, returned to storage, reviewed by a collections manager, routed to a data entry station, typed into a SQL interface or its institutional equivalent, and verified before it is considered complete, or some variant of this sequence depending on the institution. This process routinely takes one to two years or more (Bebber et al., 2010, pers. obs.). Crucially, at no point in the chain, until the final step or link does the specimen acquire an identity in the database.

Herbarium digitization has never lacked ambition; the central problem has been where, when, and by whom the record first becomes computational. Herbaria remain a major frontier for species discovery, and national and international digitization efforts have long recognized the need to database, image, georeference, and mobilize specimen records at scale (Bebber et al., 2010; Barkworth and Murrell, 2012; Hardisty et al., 2020). Small-herbarium workflows and regional digitization studies have shown that this work can be done under limited resources, but also that labor, training, hardware, and repeated handling remain the limiting variables (Harris and Marsico, 2017; Powell et al., 2021). More recent tools attack the inherited downstream problem with real ingenuity: VARP improves barcode recovery from herbarium specimen images after imaging has occurred (Powell and Shaw, 2021), pyzbar exposes one-dimensional and QR decoding to Python workflows (Hudson, 2016), mvh assembles virtual herbaria from openly available specimen images in R (Vasconcelos and Boyko, 2025), Hespi detects sheet components and label fields using OCR, HTR, and multimodal models (Turnbull et al., 2025), and iNat label tools pull observation data forward into printable specimen labels (Rockefeller, 2026). These projects are not foils; they are evidence that the bottleneck is real. The field voucher system proposed here shifts the point of capture upstream, before the specimen becomes a sheet-image recovery problem. A two-dimensional QR code attached at collection time can travel with the specimen, photographs, coordinates, collector-supplied fields, uncertainty notes, and eventual export record. The hardware asymmetry matters: dedicated barcode scanners remain a specialized enterprise market, while annual smartphone shipments remain at billion-unit scale, and the phone already supplies the camera, GPS, browser, storage, network access, and QR decoder (Business Research Insights, 2026; Coherent Market Insights, 2026; IDC, 2026; IMARC Group, 2026). The contrast is therefore not simply one-dimensional versus two-dimensional symbology. It is downstream recovery versus upstream authorship, implemented as a single archival HTML file whose useful life does not depend on a server, framework, vendor, subscription, build chain, or institutional platform.

2.Materials & Methods

2.1The Zero-Dependency Constraint

The system is a single HTML file that includes ECMAScript 2025 (ES2025) JavaScript and modern CSS. No libraries, no frameworks, no CDN, no build step, no packages or services, and no server. The field voucher notebook itself, including the QR code generation engine, the EXIF parsing library, the Darwin Core export logic, the image gallery, export to HTML, JSON, CSV, XLSX, by bulk or singly, GBIF-ready UUIDs and all required fields, and the label printing system, is a self-contained primitive that opens in any standards-compliant browser. In the current reference deployment it ships alongside the Carex identification key dataset described in Section 8, but the voucher notebook does not require that dataset, or any taxon-specific data to function.

2.2The Single-File Primitive

This is not a web application in the conventional sense. It is a document that computes. The file is less than 150KB, currently at 125 KB, the complete voucher tool is smaller than the median webpage's font assets alone (131 KB), despite implementing entry capture, a timer subsystem, and Darwin Core-compliant export (Barret & Indigo, 2026). Data persists locally via the browser's IndexedDB API and remains on the device until the user exports it. Export formats are permanent and self-contained: JSON, CSV, XLSX and HTML, and a printable label sheet. An email function allows records to be transmitted to any receiving system. No data leaves the device without explicit user action. The complete file weighs under one MB: less than one half the median web page weight of 2.9 MB recorded by the HTTP Archive Web Almanac in 2025 (HTTP Archive, 2025).

2.3 Acquisition

Two ways to get the tool: an actively updated live deployment, or a fixed, citable archival snapshot.

The field voucher notebook occupies its own tab within that deployment and operates independently of the Carex identification key: no key data, regional filter, or taxon-specific input is required to create, save, or export a voucher record. Save the file to any device using the browser's native save function. The saved file is a complete local primitive. It does not call home, does not require network access, and does not communicate with any external server.

The voucher tool's timer subsystem, which records per-entry completion time locally on the user's device, was not analyzed in this study but offers a direction for future work. The timer runs entirely client-side, with no remote access; the data exists only on the user's machine unless they choose to export and send it. Voluntary, anonymized aggregation of timing data from collaborating users and students could reveal workflow patterns, such as where entry slows or accelerates, that inform future revisions to the voucher format and field protocol.

2.4 Basic Field Use

• Find mature plant material in flower or fruit. Immature material should generally be avoided for identification, regardless of taxon.
• Open Voucher. A QR code is generated automatically.
• If using a preprinted QR code, hit “Scan Now” and scan the desired code. Confirm assignment by checking the last four characters in the QR name.
• Enter Darwin Core fields, collector data, and taxon information.
• Use “Current Location” to georeference the specimen at the point of growth.
• Add habitat, associate taxa, description, population notes, and any other field observations.
• Add images already taken, or take new images of the plant, habitat, and specimen sheet.
• Save the Uncertainty Table and Key Path if an identification key was used to reach the determination. This step is optional and applies only when the collector is working through a key, such as the bundled Carex key described in Section 8; it is not required to complete a voucher.
• Save Voucher. On a phone, scroll down to view the saved voucher.
• Export records by email, local storage, or the storage system of choice.

2.5 QR Code Options

• Use the automatically generated QR code for each voucher.
• Preprint QR codes in duplicate or triplicate. Keep duplicates attached in the field, then move one clean copy to the final specimen sheet.
• Preprint collection sheets with a 5 × 5 cm QR code.
• Make duplicate QR stickers, with one placed in the notebook and one placed on the collection sheet.
• For a fast field method, number the sheets by hand, open Voucher, skip identification, generate or scan the QR code, add GPS, photograph the plant in situ, photograph the sheet, and save the voucher.

2.6 Workflow Overview

The system supports more than one effective workflow; the QR code is the anchor regardless of which path a collector takes. Section 6 catalogues the workflows identified so far, including printed-QR, sticker-duplicate, press-first, and retrospective paths for individual collectors, and aggregate paths for group events and batch imports.

2.7 Non-Botanical Applications

The workflow is not limited to plant specimens. The Kingdom field accepts Plantae, Animalia, Fungi, Protozoa, Bacteria, or Archaea, and Basis of Record accepts Preserved Specimen, Living Specimen, Fossil Specimen, Material Sample, Human Observation, or Machine Observation, so the same voucher schema covers a fungal collection, a soil sample, a fossil, a sight record with no physical specimen at all, or a camera-trap capture, with no modification to the tool. The same method applies to any physical objects that must be distinguished in the field and connected to a permanent record later, including archaeological finds, rock specimens, press bundles, or bags of sorted material.

2.8 Export and Archiving

Records export as Darwin Core JSON with required GBIF fields, CSV, XLSX, or HTML tables, all carrying the QR reference, alongside printable label sheets. Exported records can be sent by email, stored locally, or imported into an institutional system from any of these formats. The system does not maintain a connection to any receiving system and does not require one. Backups and emailed copies are recommended, since the collector holds the only copy that exists until one is made.

Georeferencing precision depends on the device and field conditions at the time of collection. Coordinates recorded at the living plant, or extracted from images taken at the living plant, document the collection event itself rather than the vehicle, camp, herbarium, or later processing location.

2.9Failure Modes Addressed

The specific failure this system addresses is the separation of the physical specimen from its documentation at any point between collection and permanent databasing. This separation occurs in several ways: the collector takes field notes and forgets to transfer them; the notes are lost, damaged, or illegible; the label is prepared days or weeks after the collection when locality details have faded from memory; the herbarium sheet and the field notebook become separated; the collector dies.

The system's response to this problem is to generate a unique specimen identifier at the moment of collection, embed that identifier in a QR code, and ensure that the QR code travels with every export format. A printed label carries the QR. A standalone HTML voucher file carries the QR. The bulk export table carries a QR image in every row. If the physical sheet and the database record ever become separated, the QR can reconnect them quickly or you can copy/paste the UUID.

3.Workflow Plurality and the Third Data Category

Or "Unresolved Records as Positive Data"

The system described above does not prescribe a single collection workflow. Section 6 catalogues eight named workflows identified so far; the list is open, not closed, and it should stay that way. A collector carrying pre-printed label pairs and a roll of cellophane tape in a vest pocket works through a different physical sequence than one who photographs everything first and assigns identifiers later at the vehicle. Another collector, four hours from home and without field materials, may salvage the chain by scanning an unused commercial QR code from a newspaper, package, receipt, or other disposable object and reassigning it to the voucher record. This is not the ideal workflow, but it is a documented one, and it is better than losing the association entirely. A bio-blitz may produce forty records from a dozen phones in one afternoon, later merged by an organizer into a single master file. A backlog of mixed material may be sorted and imported by family or kingdom rather than by collection event. A taxon that should not be collected at all may be vouchered by photograph alone, once a qualified reviewer vets it. None of these workflows were anticipated when the system was first built, and more will turn up; the author does not claim to have discovered them all. All of these paths converge on a durable, linkable record. The system's integrity derives not from procedural uniformity but from documentation: whatever path was taken, it is noted. The chain need not begin cleanly to end well. This is the operational meaning of the broken chain. A provenance record that acknowledges its own imperfections is more scientifically useful than a silent gap, because the imperfection is now queryable. The collector who writes "QR reassigned from commercial packaging; original identifier discarded; locality and date verified against field notes and device EXIF" has produced an honest record. The one who writes nothing has produced a cabinet problem.

A separate and underappreciated data type emerges from the same system: the observation that was made, documented with character states, and honestly conceded. Current biodiversity databases treat absence of a determination as silence, and silence is typically read as absence. This conflation is wrong, and its consequences compound across decades of range modeling and conservation assessment. A collector who encounters twenty members of Carex section Ovales in a Rocky Mountain canyon, records perigynium dimensions, scale color, and beak length, and then writes "unresolved; complex difficult here; characters noted" has produced something of positive scientific value. The record is not absence data. It is not confused data in the pejorative sense. It is a difficulty signal: a geolocation, a date, a set of character states, and an explicit epistemic flag indicating that the taxonomic system encountered genuine morphological compression at this site. That signal, aggregated across collectors and years, shows where the hard boundaries of a complex actually run. It tells range modelers which records require caution and which are clean. It tells the next specialist with fresh material exactly where to look. A forced determination that goes silently wrong, databased under a name that later proves incorrect, corrupts downstream analysis for as long as the record persists uncorrected. The honest concession, with characters attached, does the opposite. Expert bypass is the limit case: when Tony Reznicek looks at a specimen and says Carex frankii in Ohio, that assertion is the provenance chain. The system accommodates it directly, as it accommodates the first-year student at Cedar Bog who loses the thread on a difficult perigynium and says so. Both records mean something. Neither is nothing.

4.The Inertia Bottleneck

The scale of the backlog is not rhetorical. As of 31 December 2025, Index Herbariorum recorded 4,035 active herbaria worldwide (Thiers, accessed 2026), collectively holding an estimated 406,426,591 specimens. The number matters because every specimen is, at minimum, a physical object requiring a name, a place, a date, a collector, and a durable route into a data system. A global collection of this size cannot be rescued by meetings, intentions, or future platforms alone. It requires small tools that reduce friction at the point where the record is born.

The digitization problem is usually described as a museum problem, but the bottleneck begins earlier. A herbarium sheet can be imaged later. A label can be typed later. A database record can be imported later. The field event cannot be repeated later. The living plant, the exact slope, the wet ditch, the associated species, the collector's uncertainty, the light, the weather, the deer flies, and the memory of which specimen belonged to which bag all exist at one moment. If that moment is missed, the later institutional workflow is not digitizing a specimen. It is digitizing damage.

This system therefore attacks a different part of the chain than conventional digitization pipelines. It is not a conveyor belt for finished sheets. It is a field primitive for preventing unfinished records. The goal is not to replace herbarium databases, Specify, Symbiota, SERNEC, GBIF, or institutional cataloging. The goal is to ensure that a specimen reaches those systems with a stable identifier, field photographs, coordinates, date, collector, locality text, identification confidence, and label-ready metadata already attached.

Pressed plants are therefore the obvious beachhead. They are flat, label-driven, historically tied to locality data, and already structured around a physical sheet plus a text record. They are not the only target, but they are the cleanest first target. Once the primitive is proven in plants, the same architecture can be handed outward. Fungal packets, algae envelopes, seed lots, tissue samples, rocks, fossils, bones, archaeological bags, fish jars, bird and mammal preparations, and type cabinets all share the same core problem: a physical object must remain bound to a record through time.

A single prescribed workflow would fail because field work is plural. Organized collectors may preprint empty QR sheets or stickers before leaving and assign them to specimens by photographing the living plant with the QR in frame. Others may create the voucher in the field, generate the QR at the plant, then print duplicate labels later. Others will use the Cowboy Method: attach one photograph, capture the date and coordinates, write a rough name, and save the record before the day collapses. A fourth workflow is desk salvage: old photos, old bags, partial notes, and incomplete metadata entered later with uncertainty made explicit. The system is designed around these workflows rather than against them.

The QR code is not proposed as a replacement for institutional barcodes. It occupies a different moment. A one-dimensional barcode is excellent after the specimen enters a herbarium workflow. It is fast, cheap, and suitable for a workstation scanner moving through finished sheets. A QR code is useful before that point, when the collector has a phone but not a scanner, a living plant but not yet a mounted specimen, and a field record that must be bound to photographs and coordinates immediately. The same physical label can carry both. A human-readable identifier survives everything. A QR code supports field and phone access. A one-dimensional barcode supports legacy institutional scanning.

• Human-readable ID. The survival layer. It can be read, copied, photographed, typed, or recovered after the machines are gone.
• QR code. The field and phone layer. It binds the living plant, photograph, bag, label, and local voucher record before the institutional workflow begins.
• One-dimensional barcode. The legacy accession layer. It remains appropriate for high-throughput workstation scanning after the specimen enters a collection.
• RFID. The cabinet-scale audit layer. It is especially useful for types and other high-value specimens, where thousands of objects can be detected without line-of-sight handling.

This layered view prevents identifier warfare. The identifier is the specimen. The machine-readable forms are merely different doors into the same record. Printed text is for humans. QR is for phones and field assignment. One-dimensional barcode is for traditional scanners. Each has a place. None needs to destroy the others.

The Extended Specimen Network and Digital Extended Specimen frameworks make this more urgent, not less (Lendemer et al., 2020). Modern specimens are no longer treated only as singular physical objects. They are anchors for suites of linked data: images, DNA sequences, tissue samples, field notes, phenotypes, environmental context, annotations, loans, citations, and later determinations. These links cannot be built cleanly if the first link is missing. A specimen without a reliable collection event becomes a weak anchor for everything downstream.

The zero-dependency file described here is not the Extended Specimen Network. It is a small engine at the mouth of that network. It gives the physical specimen a durable digital handle while the plant is still alive, while the collector is still present, while the coordinates are still true, and before the institutional lag begins. The record can later move into Darwin Core, institutional databases, GBIF-mediated publication, label sheets, HTML vouchers, or RFID-backed inventory systems. The point is that the first record exists before the memory decays.

This is why the system is deliberately primitive. The advanced solution is the institutional stack: server, database, grant, account, scanner, support queue, policy meeting, and migration plan. The primitive solution is one file that opens in a browser and does the small thing immediately. It creates the record, attaches the images, captures the coordinates, preserves uncertainty, prints the label, and exports the data. It does not require the collector to join an institutional workflow before collecting. It allows the institutional workflow to arrive later without forcing the field event to wait for it, and it is built to outlast both the collector's field season and the platforms around it.

5.System Description

5.1Technical Architecture

The system is implemented in approximately 3500 lines of HTML, CSS, and vanilla JavaScript in the key file. It uses no external libraries. QR code generation is implemented via a self-contained zero-dependency encoder embedded directly in the file. EXIF data extraction from photographs uses a pure-JavaScript parser that reads GPS coordinates, capture date, and device metadata directly from image binary data in the browser, without any server-side processing.

Data storage uses the browser's IndexedDB API, providing structured local persistence with no size limits beyond device storage. Records survive browser restarts and device reboots. Data is never transmitted to any external server. The file is consistent with the 20-Year File / Digital Paratype architecture (Jones, 2025a): a single artifact, human-readable, requiring no build system, deployable by copying.

5.2The Voucher Record

Each saved voucher record in the TMJv5 notebook is built from the form fields below. The CSV and XLSX exports use the same saved keys through FIELD_DEFS; JSON carries the same record and adds a photo manifest when photographs are present.

The voucher therefore separates two identifiers that were previously easy to confuse: the visible Voucher ID / QR can be reassigned to a preprinted label, while the GBIF-ready occurrence UUID remains immutable for the record. Collector first name, collector last name, collector ORCID, institution, and country persist across consecutive saves; collector number remains manual and does not auto-increment in TMJv5. The form presents genus and species separately for field entry, but the saved and exported record carries their joined value as scientificName.

5.3QR Code Implementation

Each voucher record receives a unique identifier at the moment of form initialization, before the record is saved. This identifier is encoded in a QR code displayed prominently in the voucher interface. The QR code travels with the record through every export path.

Each individual HTML voucher export includes the QR as a 2-inch embedded image with generous white margin. The bulk HTML table export includes a QR column as the first column of every row. The label print output places the QR flush right on each label with a minimum 0.18-inch white border on all sides. This border is not decorative; it is functional, providing the quiet zone that scanning algorithms require. Labels printed without adequate quiet zones fail to scan reliably, particularly in low-light field conditions and on lower-resolution phone cameras.

5.4Export Formats

• DwC JSON A complete occurrence export conforming to Simple Darwin Core. All fields mapped per the table above. Suitable for direct import to most institutional herbarium database systems.
• HTML Table All vouchers in a single landscape-format HTML file with embedded QR images per row. Opens in any browser. Importable to most spreadsheet applications by drag and drop.
• CSV / XLSX One row per voucher, all fields from the table above, in either plain CSV or a formatted XLSX workbook. Suitable for direct import to most institutional herbarium database systems and for bulk editing before re-import.
• Label Print Direct to browser print dialog. Labels render two-across at 4 × 3 inches on legal-size paper, sized for printing onto blank adhesive sheet stock cut to size or onto card stock for hand-cutting; the layout does not match a standard pre-perforated Avery template, so test-print on plain paper before committing label stock. No popup window, no external file, no second application. The label sheet is generated inline and the browser print dialog is called directly.

5.5EXIF Extraction

When a photograph is attached to a voucher record, the system parses the image’s EXIF metadata in the browser and extracts GPS coordinates and reported positioning error, when present, populating the latitude, longitude, and accuracy fields automatically. The coordinate source field is set to "EXIF" to document the provenance of the georeference. The parser is deliberately minimal: it reads only the GPS tags out of a JPEG’s EXIF segment, nothing else, and never writes or re-embeds EXIF data of any kind. It does not extract the capture date; eventDate comes from manual entry or defaults to the day of save.

This is the most common path for field photographs taken on a modern smartphone, which embeds GPS coordinates directly in the image file. The limitation is intentional and disclosed in the source rather than hidden: HEIC originals from some iOS capture flows are not parsed for GPS in the current build, and a photograph with no GPS tag at all, an older camera, a scan, a screenshot, simply yields no coordinates, which is the expected, ordinary case rather than an error. When EXIF GPS data is unavailable for any of these reasons, the system falls back gracefully to the device Geolocation API or manual entry. The coordinate source field documents which path was used.

5.6Field Benchmark

The author is not aware of any existing field documentation system, institutional or commercial, that produces a complete Darwin Core-compliant voucher record with embedded photographs and a QR-coded identifier on a device carried to the field, without network connectivity, without a prior account, and without installation, in under three minutes.

6.The Eight Field Workflows

The system explicitly supports eight named field workflows, and certainly others not yet catalogued, reflecting the actual range of collector practice rather than assuming a single correct approach.

7.The Fifty Cabinets

The backlog exists even from collectors now infirm or deceased with incomplete or absent locality data. This figure is based on informal survey of herbaria of all sizes, and personal observations across institutions visited during three decades of field and curatorial work in Cyperaceae.

No tool, however well designed, recovers a broken chain. This paper does not claim otherwise.

What the system described here prevents is the next fifty cabinets. The chain breaks at a specific, predictable, and addressable moment: the end of a hard field day, when the documentation step has been deferred one more time. A system that makes the documentation step take > five minutes on a phone the collector already has removes the primary friction that causes the deferral.

The specimens in the author's own cabinets, several hundred of them, represent the same failure mode. They are recoverable because the author is still alive and the mental index persists, barely. The system described here exists in part because those cabinets exist.

8.Relationship to Florilegium Caricis

The field voucher system described in this paper was developed as a component of the Florilegium Caricis interactive character matrix (Jones, 2025b; DOI: 10.5281/zenodo.19138828), a zero-dependency single-file identification key for 466 North American Carex taxa based on Flora of North America Volume 23. Within that application, the voucher notebook occupies a dedicated tab and includes integration with the identification key: a collector can attach the current key filter path to a voucher record, preserving the character data that led to the identification alongside the locality record.

The voucher system functions as an independent tool and is described independently here because its scope extends beyond Carex and beyond identification keys. Any botanical field collector, regardless of taxon group, can use the system with no modification. The integration with Florilegium Caricis is additive; the voucher notebook operates without it.

The two papers cite each other. Florilegium Caricis cites this methodology paper for the voucher system architecture. This paper cites Florilegium Caricis as the demonstration context and development platform.

9.Conclusion

The documentation chain between field collection and permanent herbarium record breaks predictably, repeatedly, and at a specific point. Existing tools have not eliminated this failure mode because they impose friction at exactly the moment friction is least tolerable. A system that requires under five minutes on a phone already in the collector's pocket, with no installation, no account, no network connection, and no institutional prerequisites, addresses the failure at the correct point.

The QR code traveling with every export format addresses the chain-of-custody problem that persists even when initial documentation is complete. The Darwin Core compliance ensures that records produced in the field are directly importable to institutional systems without re-entry (Wieczorek et al., 2012). The zero-dependency single-file architecture ensures that the system remains available without maintenance burden for the lifetime of the collections it documents.

The system is designed to remain functional without modification for a minimum of twenty years. This is not an aspiration; it is an architectural constraint. The 20-Year File methodology (Jones, 2025a) identifies the single-file zero-dependency HTML artifact as the only format currently available to independent researchers that satisfies three simultaneous requirements: zero installation burden on the reader's device, no server infrastructure required by the author, and demonstrated backward compatibility across browser generations. The file is archived on Zenodo with a permanent DOI; Zenodo's preservation commitment extends to CERN infrastructure timescales. The file itself requires no Zenodo-specific behavior to function, and will open from a USB drive, an email attachment, or a local filesystem with identical behavior to the archived version.

The fifty cabinets of unrecoverable specimens are a fixed cost of past practice. The next fifty cabinets are preventable.

References

• Barkworth, M. E., and Murrell, Z. E. (2012). The US Virtual Herbarium: Working with individual herbaria to build a national resource. ZooKeys, 209, 55–73. https://doi.org/10.3897/zookeys.209.3205
• Barret, R., & Indigo, J. (2026, January 15). The 2025 Web Almanac: Page weight (Chapter 14). HTTP Archive. https://almanac.httparchive.org/en/2025/page-weight
• Bebber, D. P., Carine, M. A., Wood, J. R. I., Wortley, A. H., Harris, D. J., Prance, G. T., Davidse, G., Paige, J., Pennington, T. D., Robson, N. K. B., and Scotland, R. W. (2010). Herbaria are a major frontier for species discovery. Proceedings of the National Academy of Sciences, 107(51), 22169–22171. https://doi.org/10.1073/pnas.1011841108
• Business Research Insights. (2026). Barcode Readers Market Size, Growth: Global Report 2026–2035 (Report No. 102657). https://www.businessresearchinsights.com/market-reports/barcode-readers-market-102657
• Coherent Market Insights. (2026). Barcode Scanners Market: Share, Size and Industry Analysis. https://www.coherentmarketinsights.com/market-insight/barcode-scanners-market-3903
• GBIF Germany. (2025). Digitisation of German Herbaria. https://herbarium.gbif.de/digitisation/
• Hardisty, A., Saarenmaa, H., Casino, A., Dillen, M., Gödderz, K., Groom, Q., Hardy, H., Koureas, D., Nieva de la Hidalga, A., Paul, D., Runnel, V., Vermeersch, X., van Walsum, M., and Willemse, L. (2020). Conceptual design blueprint for the DiSSCo digitization infrastructure: Deliverable D8.1. Research Ideas and Outcomes, 6, e54280. https://doi.org/10.3897/rio.6.e54280
• Harris, K. M., and Marsico, T. D. (2017). Digitizing specimens in a small herbarium: A viable workflow for collections working with limited resources. Applications in Plant Sciences, 5(4), 1600125. https://doi.org/10.3732/apps.1600125
• HTTP Archive. (2025). Page weight. The 2025 Web Almanac. https://almanac.httparchive.org/en/2025/page-weight
• Hudson, L. (2016). pyzbar: Read one-dimensional barcodes and QR codes from Python 2 and 3. IS&T International Symposium on Electronic Imaging 2016. https://pypi.org/project/pyzbar/
• IMARC Group. (2026). Barcode Scanner Market: Global Industry Trends, Share, Size, Growth, Opportunity and Forecast 2026–2034. https://www.imarcgroup.com/barcode-scanner-market
• International Data Corporation. (2026). Worldwide Quarterly Mobile Phone Tracker: Smartphone market insights. https://www.idc.com/promo/smartphone-market-share/
• Jones, T. M. (2025a). The 20-Year File: A digital paratype architecture for zero-dependency scientific artifacts. TJID3 Research. Zenodo. https://doi.org/10.5281/zenodo.xxxxxx
• Jones, T. M. (2025b). Florilegium Caricis: An interactive character matrix and identification key for Carex of North America (FNA Vol. 23). TJID3 Research. Zenodo. https://doi.org/10.5281/zenodo.19138828
• Jones, T. M. (2025c). Human-orchestrated multi-model convergence: Čысты, a reader's companion to Boltzmann (1872), produced by collaborative AI under human editorial direction. TJID3 Research. Zenodo. https://doi.org/10.5281/zenodo.18988588
• Lendemer, J., Thiers, B., Monfils, A. K., Zaspel, J., Ellwood, E. R., Bentley, A., LeVan, K., Bates, J., Jennings, D., Contreras, D., Lagomarsino, L., Mabee, P., Ford, L. S., Guralnick, R., Gropp, R. E., Revelez, M., Cobb, N., Seltmann, K., and Aime, M. C. (2020). The Extended Specimen Network: A strategy to enhance US biodiversity collections, promote research and education. BioScience, 70(1), 23–30. https://doi.org/10.1093/biosci/biz140
• Powell, C., Krakowiak, A., Fuller, R., Rylander, E., Gillespie, E., Krosnick, S., Ruhfel, B., Morris, A. B., and Shaw, J. (2021). Estimating herbarium specimen digitization rates: Accounting for human experience. Applications in Plant Sciences, 9(4), e11415. https://doi.org/10.1002/aps3.11415
• Powell, C., and Shaw, J. (2021). Performant barcode decoding for herbarium specimen images using vector-assisted region proposals (VARP). Applications in Plant Sciences, 9(5), e11436. https://doi.org/10.1002/aps3.11436
• Rockefeller, A. (2026). iNaturalist Herbarium Label Generator. GitHub. https://github.com/AlanRockefeller/inat.label.py
• Thiers, B. [continuously updated]. Index Herbariorum: A global directory of public herbaria and associated staff. New York Botanical Garden's Virtual Herbarium. https://nybg.org (Accessed June 24, 2026).
• Turnbull, R., Fitzgerald, E., Thompson, K., and Birch, J. L. (2025). Hespi: A pipeline for automatically detecting information from herbarium specimen sheets. BioScience, 75(8), 637–648. https://doi.org/10.1093/biosci/biaf042
• Vasconcelos, T., and Boyko, J. D. (2025). mvh: An R tool to assemble and organize virtual herbaria from openly available specimen images. Applications in Plant Sciences, 13, e11631. https://doi.org/10.1002/aps3.11631
• Wieczorek, J., Bloom, D., Guralnick, R., Blum, S., Döring, M., Giovanni, R., Robertson, T., and Vieglais, D. (2012). Darwin Core: An evolving community-developed biodiversity data standard. PLOS ONE, 7(1), e29715. https://doi.org/10.1371/journal.pone.0029715