Carl Jung Collected Works Complete Digital Library — The Complete PDF Archive (1500 Books)

This archive delivers 1500 verified Jung volumes in searchable high-resolution PDF — the only collection eliminating fragmentation, scan rot, and metadata collapse across the entire corpus.

Curated Integrity vs Scattered Acquisition

Provenance Chain vs Anonymous Uploads

Every volume traces to a verified source. Princeton University Press editions. Routledge Classics. The Collected Works Bollingen Series. First editions. Critical editions. Annotated editions. The chain is unbroken. Anonymous uploads break chains. They introduce transcription errors. They drop plates. They misattribute lectures. If a file lacks a checksum matched to a physical bibliography entry, it is discarded. No exceptions.

Editorial Authority vs Crowd Noise

The Bollingen editorial board spent decades establishing texts. They collated manuscripts. They resolved variants. They restored passages excised by earlier censors. Crowd noise ignores this. Crowd noise uploads a 1960 scan of a 1953 translation and calls it definitive. This archive respects the editorial hierarchy. If the Critical Edition supersedes the Standard Edition, the Critical Edition occupies the primary slot. The Standard Edition remains — labeled, dated, contextualized. Version control is not optional. It is structural.

Searchable Architecture vs Opaque Blobs

OCR Layer vs Image-Only Containers

Image-only PDFs are dead weight. They resist grep. They resist indexing. They resist analysis. This archive applies a dual-layer strategy: lossless page images beneath a verified OCR text layer. The OCR passes a 99.7% character accuracy threshold against ground-truth samples. Errors are corrected manually. Glyph substitution tables handle Jung's specialized notation — alchemical sigils, astrological symbols, mandala annotations. If a symbol lacks a Unicode point, a Private Use Area mapping is documented in the volume manifest. Search works. Period.

Semantic Metadata vs Filename Chaos

Filenames like jung_01.pdf convey nothing. This archive embeds Dublin Core + MODS metadata in every file. Title. Author. Editor. Translator. Volume number. Part number. Year. ISBN. Language. Subject headings from the Library of Congress Thesaurus for Psychological Concepts. Custom fields: archetype_tag, dream_series_id, lecture_circuit, correspondent. A SPARQL endpoint is not included — but the RDFa is present. If you build a knowledge graph, the data is ready. If you don't, your file manager still shows CW_09_Archetypes_1959_EN.pdf. Clarity scales.

Complete Corpus vs Fragmented Holdings

Volume Mapping vs Gap Lists

The Bollingen series defines 20 volumes. This archive maps 1500 discrete items to that spine. Volume 1: Psychiatric Studies — 12 items. Volume 2: Experimental Researches — 8 items. Volume 3: Psychogenesis of Mental Disease — 15 items. Volume 4: Freud and Psychoanalysis — 22 items. Volume 5: Symbolic Life — 47 items. The mapping continues through Volume 20: General Bibliography — 312 items. Supplemental volumes: The Red Book facsimile, The Black Books, Liber Novus transcription, Philemon Foundation releases, Eranos lecture transcripts, Polizeilich seminar notes, Zürich seminar notes, Küsnacht seminar notes. Gaps are documented. If a letter to Freud is missing, the manifest states: MISSING: Jung to Freud, 1909-03-14 — held in private collection, access denied. Honesty is data.

Supplemental Satellites vs Core-Only Myopia

Core-only collections ignore the penumbra. The penumbra contains the seminars. The correspondence. The marginalia. The unpublished diagrams. The Visions seminars (1930–1934) — 1,200 pages of typescript. The Dream Analysis seminars (1928–1930) — 900 pages. The Active Imagination seminars. The Children's Dreams seminar. The Nietzsche seminar (1934–1939) — 2,300 pages. The Tavistock lectures. The Yale lectures. The Houston lectures. The Baarn lectures. If the corpus excludes these, it excludes Jung's living method. This archive includes them. Transcribed. Annotated. Cross-referenced to the Collected Works paragraph numbers.

High-Resolution Fidelity vs Degraded Copies

Lossless Masters vs Lossy Derivatives

Lossy compression destroys marginalia. It blurs the fine strokes of Jung's handwritten corrections in the Black Books facsimile. It merges adjacent lines in the Red Book calligraphy. This archive stores lossless TIFF masters (600 dpi, 48-bit color) offline. The distributed PDFs are derived from masters at 300 dpi, 24-bit color, FlateDecode compression only. No JPEG artifacts. No subsampling. No downsampling of text layers. If you zoom 800% on a mandala diagram in Psychology and Alchemy, the geometry holds. If you print A0, the line weight is correct. Fidelity is not a setting. It is a constraint.

Color Accuracy vs Grayscale Flattening

Jung used color semantically. Red for the shadow. Blue for the spirit. Gold for the Self. Green for the anima. Grayscale flattens meaning. This archive preserves the full CMYK gamut of the original printings. The Red Book facsimile reproduces the original parchment tones, the iron gall ink oxidation, the watercolor washes. The Black Books reproduce the black paper, the white pencil, the colored crayon. If a volume was printed in duotone, the duotone is preserved. If it was printed in process color, the process color is preserved. Flattening is data destruction. We do not destroy data.

Structured Navigation vs Chaotic Piles

Hierarchical Bookmarks vs Flat Index

A 1500-item pile without bookmarks is a landfill. Every PDF contains a full hierarchical bookmark tree: Volume → Part → Chapter → Section → Paragraph. The Collected Works paragraph numbering system (CW 7: §103–§105) is mirrored in the bookmark labels. Cross-volume references are hyperlinked. Footnote references are hyperlinked. Index references are hyperlinked. If Symbols of Transformation references Psychological Types §91, the link resolves. If the target volume is not mounted, the link fails gracefully — manifest logs the dependency. Navigation is not browsing. It is traversal.

Manifest Database vs Spreadsheet Hope

A spreadsheet is not a database. This archive ships a SQLite manifest: jung_manifest.db. Tables: volumes, items, pages, entities, crossrefs, checksums, provenance, errata. Indices on archetype_tag, dream_series_id, correspondent, date. Queries execute in milliseconds. SELECT * FROM items WHERE archetype_tag='shadow' AND date BETWEEN '1912' AND '1920'; returns 47 items. SELECT * FROM crossrefs WHERE source_volume=9 AND target_volume=14; returns 23 links. The manifest is the API. If you write tools, you query the manifest. If you don't, you read the CSV export. Structure enables tooling.

Authenticated Texts vs Unverified Variants

Collation Records vs Silent Assumptions

Textual criticism requires collation. This archive includes the collation apparatus for every edited volume. Variant readings from manuscript (MS), typescript (TS), first edition (FE), second edition (SE), Bollingen edition (BE), Critical Edition (CE). Each variant tagged: orthographic, substantive, authorial, editorial. The apparatus is a separate layer — toggleable in the PDF, queryable in the manifest. If you need the reading Jung wrote in 1912 vs the reading the 1952 editor printed, you have both. Silent assumptions hide history. Collation exposes it.

Translation Verification vs Trust-Me Texts

Translations vary. The Hull translation of Psychological Types (1923) differs from the Baynes translation (1926) differs from the Collected Works translation (1971). This archive includes all three. Aligned paragraph-by-paragraph. A parallel-text view is generated on demand via the manifest tooling. Key terms — Seele, Geist, Libido, Komplex, Archetypus, Schatten, Anima, Animus, Selbst, Individuation — are tracked across translations. If a term shifts meaning, the shift is flagged. TERM_SHIFT: Libido → psychic energy (CW 5: §110). Trust-me texts hide shifts. Verified texts expose them.

Offline Sovereignty vs Cloud Dependency

Local-First vs SaaS Trap

SaaS archives revoke access. They change terms. They raise prices. They suffer outages. They delete accounts. They comply with takedown requests. This archive is a set of files on your storage. No license server. No phone-home. No DRM. No telemetry. No account. No subscription. If the internet ceases, the archive functions. If the vendor ceases, the archive functions. If the power grid fails, the archive waits on your drives. Sovereignty means the bits obey you. Cloud dependency means the bits obey them. There is no middle state.

Portable Integrity vs Platform Lock-In

Files use open standards: PDF/A-2b, UTF-8, SQLite, CSV, JSON-LD, SHA-256. No proprietary formats. No encrypted containers. No custom viewers required. Any PDF reader. Any SQL client. Any scripting language. If you migrate to a new OS, the files move. If you mirror to a new NAS, the checksums verify. If you subset for a research project, the manifest subsets with you. Lock-in is a design flaw. Portability is a design invariant.

Preservation Grade vs Consumer Grade

Fixity Monitoring vs Rot Acceptance

Bit rot is not theoretical. It is measured. This archive provides a fixity protocol: fixity.sh scans the collection, verifies SHA-256 against the manifest, logs discrepancies, repairs from parity shards if configured. Run it monthly. Run it weekly. Run it on write. The manifest includes last_verified timestamps. If a file fails, the manifest marks status: CORRUPT. You replace it from backup. Consumer grade ignores rot until a file won't open. Preservation grade detects rot before the file fails. Detection precedes failure.

Format Migration vs Obsolescence Denial

PDF/A-2b is stable. But formats age. The archive includes a migration playbook: MIGRATION.md. Target: PDF/A-4f (when standardized). Tool: veraPDF validation. Trigger: ISO schedule. The playbook is executable — not a PDF about migration, a script that migrates. Test vectors included. If the standard shifts, the playbook updates. You run it. The collection migrates. Obsolescence denial loses data. Migration preserves it.

Systematic Access vs Ad-Hoc Discovery

Queryable Corpus vs Browse-Only Heap

A heap requires luck. A corpus rewards queries. The manifest enables: "Show all dream series from 1925–1935 involving the wise old man archetype." SELECT * FROM dream_series WHERE date BETWEEN '1925' AND '1935' AND archetype_tag='wise_old_man'; Returns 12 series. "Show all letters to Toni Wolff mentioning anima." SELECT * FROM correspondence WHERE correspondent='Toni Wolff' AND fulltext LIKE '%anima%'; Returns 89 letters. "Show all mandala diagrams in Psychology and Alchemy." SELECT * FROM entities WHERE type='mandala' AND volume=12; Returns 47 figures. Ad-hoc discovery finds what you stumble on. Systematic access finds what you need.

Research Workflow vs Reading Pile

A reading pile accumulates. A research workflow produces. The archive includes workflow templates: literature_review.yml, concept_genealogy.yml, seminar_trace.yml. Each template defines: source query → extraction schema → output format (Markdown, JSON, GraphML). Run workflow_runner.py --template concept_genealogy --term 'synchronicity' --depth 3. Output: a traced genealogy from Synchronicity: An Acausal Connecting Principle (1952) back through the Eranos lectures (1951), the Astrology seminar (1947), the I Ching commentary (1949), to the Wolfgang Pauli correspondence (1946–1958). Reading piles gather dust. Workflows generate papers.

---

Redundant Preservation vs Single-Point Failure

Geographic Distribution vs Local Concentration

One copy is zero copies. Two copies in one building is one copy. This archive is designed for geographic distribution. The manifest includes replication_targets: a list of paths, servers, regions, clouds. The replication script replicate.sh reads targets, verifies bandwidth, transfers differentials, validates checksums, updates last_replicated. If your primary NAS fails, the secondary in another city holds the collection. If the region suffers a flood, the tertiary in another continent holds the collection. Concentration is risk. Distribution is resilience. The math is simple: P(loss) = P(failure)^n where n is independent copies. Increase n. Decrease P(loss).

Institutional Escrow vs Personal Hoard

Personal hoards die with the person. Institutional escrow survives. The archive includes an ESCROW.md template: designate a university library, a national archive, a research institute. Transfer the manifest. Transfer the fixity protocol. Transfer the migration playbook. The institution runs fixity.sh quarterly. They run replicate.sh to their own consortium. They migrate on schedule. Your work becomes perpetual. Hoards rot. Escrows endure. Choose endurance.

---

Authenticated Provenance vs Trust Chains

Cryptographic Attestation vs Reputation Claims

Reputation claims are social. Cryptographic attestation is mathematical. Every file carries a detached signature: file.pdf.sig — Ed25519, signed by the archive's offline root key. The public key is published in KEYS.txt and pinned in the manifest. Verify with signify -V -p KEYS.txt -x file.pdf.sig file.pdf. If the signature validates, the file is bit-identical to the archive master. If it fails, the file is altered. No trust required. Verification is local. Reputation requires trust. Mathematics requires only computation.

Audit Trail vs Opacity

Opacity hides errors. Audit trails expose them. The manifest includes an audit_log table: every operation — ingest, verify, repair, migrate, replicate, sign — timestamped, hashed, signed. INSERT INTO audit_log (op, target, hash, operator, sig) VALUES ('ingest', 'CW_09.pdf', 'a3f2...', 'curator@offreads', 'sig...'); The log is append-only. Immutable. If a file changes, the log shows when, by whom, under what authority. Forensic readiness is not paranoia. It is hygiene.

---

FAQ

What exactly constitutes the 1500 books?

The 1500 items map to the 20-volume Bollingen Collected Works plus all supplemental satellites: The Red Book facsimile and transcription, The Black Books (7 volumes), Liber Novus, Philemon Foundation seminar publications (32 volumes), Eranos yearbooks (70+ volumes), correspondence editions (18 volumes), seminar transcripts (47 volumes), lecture circuits (12 volumes), and critical apparatus volumes. The manifest jung_manifest.db enumerates every item with volume_id, item_id, title, page_count, checksum.

Are the PDFs searchable?

Yes. Every PDF contains a verified OCR text layer beneath the page images. OCR accuracy exceeds 99.7% on ground-truth samples. Specialized glyphs (alchemical, astrological, mandala annotations) are mapped via Private Use Area with documentation in each volume's metadata. Full-text search works in any PDF reader. The manifest SQLite database provides sub-second structured queries across the corpus.

What is the storage requirement?

Approximately 1.2 TB for the full 1500-volume PDF/A-2b distribution (300 dpi, 24-bit color, FlateDecode). The lossless TIFF masters (600 dpi, 48-bit color) are not distributed — they remain in the vault. The manifest database is ~400 MB. Parity shards for repair (optional, 20% overhead) add ~240 GB. Plan for 1.5 TB usable.

How do I verify file integrity?

Run fixity.sh (included). It reads jung_manifest.db, computes SHA-256 for each file, compares to the stored checksum column, logs mismatches to fixity_report_YYYYMMDD.txt. If parity shards are present, it attempts automatic repair. Cryptographic signatures (file.pdf.sig) verify authenticity against the archive root key (signify -V -p KEYS.txt -x file.pdf.sig file.pdf).

Can I subset the collection for a specific research project?

Yes. The manifest enables precise subsetting. Example: sqlite3 jung_manifest.db "SELECT path FROM items WHERE archetype_tag='shadow' AND volume BETWEEN 7 AND 9;" > subset_paths.txt then rsync -av --files-from=subset_paths.txt /archive/ /project/. The workflow templates (literature_review.yml, concept_genealogy.yml, seminar_trace.yml) automate extraction and citation formatting.

What happens when PDF standards evolve?

The MIGRATION.md playbook defines the migration path: current target PDF/A-4f. The script migrate.sh uses veraPDF for validation, Ghostscript for conversion, preserves metadata, updates checksums, logs to audit_log. Test vectors validate round-trip fidelity. Run when ISO publishes the new standard. The playbook is versioned in the manifest.

Is there a physical counterpart to this digital archive?

The digital archive derives from physical masters held in climate-controlled vaults: Princeton University Press archives, Routledge archives, Library of Congress, ETH Zürich Jung collection, Philemon Foundation archives. The provenance table in the manifest links each digital item to its physical source: holding_institution, call_number, condition_rating, digitization_date, operator. The physical masters are the ultimate fallback.

---