Heikhalot Rabbati: the first PD English of canonical Heikhalot

The shipment

On 2026-05-23 the Hekhal Targum engine shipped its first three canonical-Heikhalot translations: Heikhalot Rabbati 1 (the programmatic prologue), 19 (the seal-passage instructions), and 24 (the throne-approach climax). They form a coherent ascent-arc -- declaration, operation, climax -- and are the first public-domain English translations of canonical Heikhalot Rabbati ever shipped at full apparatus density. The site's name was always borrowed from this corpus. As of 2026-05-23 the name has primary-text content behind it.

Read in order:

• HR 1:1-5 -- the programmatic prologue: R. Ishmael's catalogue of what the merkavah-vision practice grants. The framing self-statement of the entire corpus.
• HR 19:1-3, 6 -- the seal-passage instructions: two chotamot presented at each of the seven palace gates; the sixth-palace destruction-warning as dramatic counterweight.
• HR 24:1-5 -- the throne-approach climax: Anafiel opens the seventh palace; the 512 eyes; the alphabetical King-of-praise cascade; the explicit Ezek 1:25 citation as the angelic-silencing trigger.

The Hebrew source is Solomon Aaron Wertheimer's Batei Midrashot (Jerusalem, 1893-1897), public-domain globally by virtue of pre-1929 publication and Wertheimer's death in 1935. The Wertheimer text was verified against Sefaria's hosted Hebrew edition, which sources from the same Wertheimer transcription. The translations are machine-assisted status under Hekhal's seven-state translation taxonomy; editor sign-off against the Schäfer Synopse paragraph apparatus is the next discipline step before they may flip to verified.

Why Heikhalot, and why now

The project is named for the heikhal -- the inner sanctuary of the Temple, the chamber behind the porch and before the Holy of Holies. The same Semitic root yields the Arabic haykal; the same architectural figure organizes the Heikhalot literature itself, where the yored merkavah -- the descender to the chariot -- ascends through seven heavenly heikhalot toward the throne of glory. The site has always taken its name from this corpus and from this architecture. Until 2026-05-23, the corpus's primary-text presence on the site was a codex entry and zero Targum translations. That asymmetry is now closed.

The opportunity was substantial. Heikhalot literature in English is one of the thinnest public-domain pockets in the entire history of Jewish-mystical translation. Hugo Odeberg's 3 Enoch or the Hebrew Book of Enoch (Cambridge, 1928) is the only pre-1929 PD English Heikhalot translation that exists -- and it covers Sefer Heikhalot / 3 Enoch, NOT Heikhalot Rabbati. The principal modern English Heikhalot Rabbati translation is Morton Smith's typescript (corrected by Gershom Scholem, edited by Don Karr, circulated as a copyrighted PDF through the Karr archive). The current scholarly standard is James Davila's Hekhalot Literature in Translation (Brill, 2013), also copyrighted. For the entire history of academic Heikhalot studies, no public-domain English Heikhalot Rabbati translation has been freely available; what existed lived behind paywalls or in personal academic archives.

The Wertheimer Hebrew text has been public-domain globally for over a century. What has been missing is a PD English rendering of that Hebrew at apparatus density. The Hekhal Targum engine's discipline -- controlled vocabulary, hermeneutic frame controllers, registry-grounded translation, the editor-in-the-loop machine-assisted workflow -- is purpose-built for exactly this kind of gap. The Heikhalot Rabbati ascent-arc translations are the engine's first attempt to fill it.

The methodological contribution: the merkavah-ascent frame controller

The Targum engine ships frame controllers per spec §8: per-corpus YAML modules that inject the corpus's hermeneutic grammar into the system prompt, scope the controlled glossary, and validate the output's apparatus against the frame's distinctive interpretive moves. Six frame controllers were active prior to this shipment: pardes (medieval Kabbalistic four-sense), zahir-batin (Akbarian outer-and-inner), kataphatic-apophatic (Christian negative theology), quadriga (medieval Latin four-sense), tafsir-ta'wil (Quranic exoteric-esoteric), and peshat-derash (rabbinic plain-and-homiletical).

Each of these is a post-classical exegetical frame. PaRDeS is a thirteenth-century Kabbalistic systematization; zahir-batin and tafsir-ta'wil are classical-Islamic exegetical pairings; kataphatic-apophatic, quadriga, and peshat-derash all postdate the patristic and rabbinic periods their corpora elaborate. None of them is calibrated to pre-classical Jewish mysticism. The Heikhalot corpus is third-to-seventh-century CE; PaRDeS as a hermeneutic-systematic frame postdates it by half a millennium to a millennium. Applying PaRDeS to a Heikhalot translation would be reading a medieval-Kabbalistic four-sense grid backward onto a corpus whose internal hermeneutic grammar is entirely different.

The new merkavah-ascent frame controller (backend/frames/merkavah_ascent.yaml) is the engine's first frame for pre-classical Jewish mysticism. Its six operational modes name what the Heikhalot literature actually does, not what later Kabbalah will eventually do to it.

Six modes for what the texts do at the operational level. The frame also names six scholarly-relation lenses for what modern reception has read the texts as being -- and this is where the methodology gets interesting.

Surfacing the scholarly debate honestly

Modern Heikhalot scholarship has been organized around several rival positions that disagree fundamentally on what kind of text the corpus is, what kind of milieu produced it, and what kind of practice (if any) it documents. The positions are not merely historical hypotheses; they are interpretive frames that materially shape how individual passages are read. The honest editorial response is to surface the debate rather than choose a side.

The merkavah-ascent frame controller carries six scholarly-relation lenses, named individually so apparatus can engage them per-passage:

• Scholem-continuity -- the Heikhalot literature documents authentic Jewish ecstatic practice; the texts' instructions for fasts, postures, recitations, and seal-prayers are evidence of an actual mystical-experiential tradition with reportable phenomenology. Major Trends in Jewish Mysticism (1941); Jewish Gnosticism, Merkabah Mysticism, and Talmudic Tradition (1960).
• Halperin-literary -- the corpus is primarily a literary-exegetical product of the synagogal-homiletical milieu of Late Antique Judaism, developing the Ezekiel 1 chariot-vision around the Shavuot Torah reading cycle. The yored merkavah is a literary protagonist, not necessarily an actual practitioner. The Faces of the Chariot (1988).
• Idel-indigenist -- the unified-stream reading flatters a continuity the textual record does not support. Heikhalot is one of several discontinuous Jewish-esoteric streams in Late Antiquity, distinct from the medieval Kabbalistic stream that begins with the Bahir by half a millennium. Kabbalah: New Perspectives (1988).
• Davila-sociological -- the Heikhalot tradents are a specific scribal-mystical professional class, comparable in social location to the rab-mag specialists of the broader ancient Near East and the Qumran covenanters behind the Songs of the Sabbath Sacrifice. The corpus is the technical literature of a guild, not the devotional literature of a movement. Descenders to the Chariot (2001).
• Boustan-martyrological -- the pseudepigraphic apparatus (Rabbi Ishmael and Rabbi Akiva as visionary protagonists) is shaped by the Ten Martyrs tradition rather than by Tannaitic biography. The Heikhalot mystics read the martyrological framework back into the Tannaitic past. From Martyr to Mystic (2005).
• Wolfson-visionary -- the visionary-imaginative act is constitutive rather than representational. The tsofeh (gazer) actively constitutes the merkavah-image through disciplined contemplation. Through a Speculum That Shines (1994).

The positions are not strictly mutually exclusive. A passage may be illuminated by Halperin's exegetical-Ezekiel substrate, Davila's scribal-professional milieu, Boustan's martyrological frame, and Wolfson's phenomenological-visionary cognition simultaneously. The frame controller surfaces them as alternative apparatus-level lenses on the same primary text, named individually so the editor can engage them per-passage. This is the methodological signature: the engine does not pre-commit to one position; the apparatus surfaces the relevant lens for the passage.

Concretely: the HR 1 apparatus engages Scholem-continuity (the protection-catalogue register fits an authentic-practice reading) and Boustan-martyrological (the R. Ishmael narrator-frame). The HR 19 apparatus engages Davila-sociological (the dense theurgic-instrumental register fits a scribal-class reading) and Halperin-literary (the gate-passage architecture as exegetical-elaboration of Ezek 1). The HR 24 apparatus engages Wolfson-visionary (the constitutive-visionary act foregrounded by the alphabetical cascade) and Scholem-continuity (the experiential register of the practitioner's terror-and-stabilization). Each passage gets the lenses that materially engage it; no passage gets a one-position rubber stamp.

HR 1 -- the programmatic prologue

אמר ר' ישמעאל מה הפרש שירות שהיה אומר מי שבקש להסתכל בצפיית המרכבה לירד בשלום ולעלות בשלום

The opening sentence of Heikhalot Rabbati frames the entire corpus: What is the distinction of the songs that he used to say -- the one who sought to gaze upon the vision of the chariot, to descend in peace and to ascend in peace? The interrogative form is itself the programmatic-prologue mode of the merkavah-ascent frame: the text is talking ABOUT the practice before performing it. The remaining sections of HR 1 answer the question by cataloguing what the practice grants -- five iterations of "greater than all of them" (gedolah mikulam), each naming a register of the practitioner's resulting capability or protection.

The catalogue is unusually concrete. Entry to the chambers of the firmament-palace and standing before the throne of his glory. Foreknowledge of providential fates (whom they humble, whom they exalt, from whom they take inheritance, to whom they give Torah). Recognition of every kind of human deed at sight (adulterer, murderer, accused, sorcerer). Priestly-Levitical juridical protection: anyone who raises a hand against the practitioner is clothed in tzaraat and crowned with bahereth (the technical Leviticus 13 skin-disease vocabulary, preserved transliterated rather than collapsed to "leprosy"). The bet-din-shel-maalah enforcement clause: "every one who stretches forth a hand against him, the celestial court stretches forth a hand against him."

The translation preserves the catalogue form -- five iterations of "Greater than all of them" rendered uniformly across the chapter rather than varied for English readability. The repetition is structurally generative of the chapter and structurally identifies the programmatic-prologue mode. The yored merkavah descender-paradox is preserved with the Scholem-experiential and Halperin-literary readings both surfaced in apparatus rather than collapsed; the priestly-Levitical vocabulary is preserved transliterated; the bet-din-shel-maalah clause is rendered in its juridical-actual register, not allegorized.

Three ambiguities are explicitly preserved in the published artifact: the lexical-polysemy of mah ha-perash shirot (distinction-of-songs vs. praise-of-songs); the scholarly-debate doctrinal preservation around the descent-paradox; the manuscript variant of nahshad al ha-mah (suspected concerning a matter vs. the conjectural niddah correction).

→ full HR 1 translation with apparatus

HR 19 -- the seal-passage instructions

אמר ר' ישמעאל כשאתה בא ועומד על פתח היכל הראשון טול שני חותמות בשתי ידיך אחד של טוטרוסיא"י ה' ואחד של סורי"א שר הפנים

The operational core of Heikhalot praxis. When you come and stand at the gate of the first palace, take two seals in your two hands -- one of TWTRWSY'Y YHWH and one of Suria the Prince of the Presence of TWTRWSY'Y YHWH. The procedure is procedural-instrumental: show the throne-King theonym seal to the guardians on the right of the lintel, show the sar-ha-panim seal to those on the left, at which point two of the guardians take you and convey you and entrust you to the next gate's chief guardian. The pattern iterates through the seven palaces; the theonym-pair changes at each gate; the directional discipline (right and left) replicates the priestly Temple-service convention.

The translation preserves the recited theonyms verbatim. TWTRWSY'Y YHWH, ADRYHRWN YHWH, TZWRTQ YHWH are rendered transliterated, not substituted with "God" or "the Lord." The merkavah-ascent rendering rule is non-negotiable here: the theurgic mechanism IS the precise phonetic articulation of the name; substitution destroys the operation. Schäfer (1992) traces the recitation-discipline back to the broader ancient Near Eastern incantation-as-passage tradition; Davila (2001) compares the chotam-discipline to the Greek magical papyri and the Aramaic incantation-bowl literature; Lesses (1998) emphasizes the participatory presence of the corpus in the Late Antique magical-theurgic milieu generally. None of those readings is foregrounded in the body; all three are available in apparatus.

The chapter's dramatic counterweight comes at HR 19:6 -- the sixth-palace destruction-warning. For the guardians of the sixth palace were destroying merkavah-descenders and non-merkavah-descenders without authorization (shelo bi-rshut), and they used to strike them and burn them. This is the structural anomaly of the corpus: the sixth-palace guardians persist in juridical-anomalous excess across generations of guardian-replacement. Heikhalot Zutarti preserves the parallel "water-water" warning attributed to R. Akiva (Schäfer §§407-410) -- the most-cited single Heikhalot passage in modern scholarship, though it lives in the sibling corpus rather than in HR proper. The HR 19:6 destruction-warning is the closest HR has to the dramatic-juridical heart of the corpus.

Eight gate-guardian theonyms appear in the chunk -- Bahabiel, Tofhiel, Tigriel, Matpiel, Shabbarriel, Ratzutziel, Pahadiel, Gevuratiel -- each transliterated with the partially-transparent -el ("-god") suffix preserved. The English vocalization follows established scholarly Heikhalot transliteration conventions (Davila 2001); the apparatus mentions partially-transparent stems (Pahadiel from p-h-d, "terror"; Gevuratiel from g-b-r, "might") but treats the names as theurgic-onomastic units rather than glossing them in the body. The procedural-narrative register is preserved verbatim, including the iterated "they take hold of you, one at your right and one at your left, and they conduct you and they hand you over" handoff structure.

→ full HR 19 translation with apparatus

HR 24 -- the throne-approach climax

היה ענפיא"ל השר פותח דלתות היכל השביעי והיה אותו אדם נכנס ועומד על מפתן פתח היכל השביעי

The structural climax of the corpus. Anafiel the prince would open the doors of the seventh palace, and that man would enter and stand on the threshold of the gate of the seventh palace, and the holy living creatures would bear upon him five hundred and twelve eyes. The 512-eyes count is HR's own theurgic-cosmographic arithmetic: four chayyot (per Ezekiel 1:5) times sixteen faces per chayyah (HR's elaboration on Ezekiel's four-faces-per-chayyah) times eight eyes per face yields 512. The number's cubic form may be theurgically operative without the corpus saying so explicitly; Halperin (1988) reads it as a Heikhalot-exegetical elaboration of Ezekiel's "wheels full of eyes" (Ezek 1:18, 10:12).

The descender's response is somatic-collapse: he shakes and shudders and recoils and is terrified and panicked and faints and falls backward -- seven verbs of dread in cascade, the corpus's most concentrated psychophysiological catalogue. The angelic stabilization is named in counter-cascade: Anafiel and sixty-three guardians of the seven palace gates support him, give him strength, and recite the protective formula:

"Do not fear, beloved son of the seed; enter and see the King in his beauty, and you are not destroyed and you are not burned."

The formula's structure -- al tira ("do not fear") + ben zera ahuv ("beloved son of the seed") + the ein atah nishchat ve-ein atah nisraf ("you are not destroyed and not burned") protection-clause -- echoes the angelic-encouragement formula of Daniel 10:11-12 (Gabriel to Daniel) and various rabbinic-aggadic "do not fear" speeches. The Heikhalot deployment is operationally specific: the formula activates as the practitioner's terror reaches its physiological climax. It is the dramatic protective counterweight to the sixth-palace destruction-warning of HR 19:6 (which is included in HR 19's translation precisely to set up this contrast). Boustan (2005) reads the formula within the martyrological-pseudepigraphic frame: the R. Ishmael-as-narrator inherits the angelic protection that the Ten Martyrs tradition denied his bodily life.

HR 24:4 is the alphabetical King-of-praise cascade. Approximately 65 King-epithets in alef-tav order: King mighty (melekh abir), King majestic (adir), King lord (adon), King blessed (barukh), King chosen (bahur), King lightning-flashing (baroq), and onward through every Hebrew letter to King supporting (tomekh) -- blessed is he. The merkavah-ascent hymnic-ecstatic rendering rule forbids collapse: preserving the alphabetical density preserves the doxological-cosmographic register the cascade performs. Wolfson (1994) reads such cascades as constitutive-visionary performances: the recited epithets do not describe a prior King; they actively constitute the imaginal King in the disciplined-contemplative act.

The climax is HR 24:5. And they would give him strength, and at once they would sound the horn above him: "to the firmament that was over their heads..." (Ezekiel 1:25), and the holy living creatures would cover their faces and the cherubim and the wheels would turn back their faces, and he stands, having turned, and stands before the throne of his glory. The explicit Ezek 1:25 citation is the angelic-silencing trigger. The Ezekiel verse in its biblical context narrates the cessation of the chayyot's motion-noise when the voice came from above the firmament; the Heikhalot deployment uses the verse as the textual-citational marker of the angelic-recess that allows the visionary climax. This is the most structurally generative biblical-citation in HR's seventh-palace stratum and is one of Halperin's principal evidences for the exegetical-Ezekiel reading of the corpus.

→ full HR 24 translation with apparatus

The discipline-contract status

Per Hekhal's editorial law, machine-assisted Targum drafts do not flip to verified status without editor sign-off. The discipline contract (TARGUM-BRAIN §4) names six conditions: schema validation passes; drift audit clean or incidents accepted; registry check clean or incidents resolved; citation check clean or incidents resolved; PD-comparison fired where the index has coverage; editor sign-off recorded. The three Heikhalot Rabbati outputs satisfy the schema and registry layers immediately. They surface honest incidents at the drift and citation layers.

The drift incidents are the substrate-gap surfaced honestly: thirteen Heikhalot glossary terms reference Hekhal lexicon pages that do not yet exist (heikhal, kavod, yored-merkavah, kisei-ha-kavod, tzefiyat-ha-merkavah, yeridah, anafiel, chayyot, ofanim, kerubim, sar-ha-panim, chotam, shomrei-ha-petach). Per OPERATING-MANUAL §7 discipline, those lexicon pages are NOT authored mid-translation-session; they surface as glossary_updates_proposed[] for a dedicated lexicon-authoring session. The drift incidents are not engine failures; they are the engine doing its job, surfacing a real substrate gap.

The 34 training-memory citations across the three outputs are all real, canonical works in Heikhalot studies -- Scholem 1941/1960/1991, Halperin 1980/1988, Schäfer 1981/1984/1992, Idel 1988, Wolfson 1994, Davila 2001/2013, Boustan 2005, Elior 2004, Lesses 1998, Swartz 1996, Janowitz 1989, Odeberg 1928, Wertheimer 1893, Smith/Karr 2009. Editor-page-verification against printed sources is the next discipline step before any citation may move from training_memory[] to verified[] in the manifest. The 20 new manifest entries are queued and the references are documented in the Sources and Editions page.

What was scaffolded (the heavier lift than the translations themselves)

The translations are the visible artifact. The compounding investment in this session was the scaffolding that makes every future Heikhalot translation reuse the same controlled vocabulary, the same frame controller, and the same scholarly substrate. Future passes -- Heikhalot Zutarti, 3 Enoch, Maaseh Merkavah, Shiur Qomah -- amortize against this session's work.

• New corpus glossary (glossaries/heikhalot/general.yaml): 20 controlled-vocabulary entries hitting the Tier-1 15-25 calibration bar. Each entry carries selected_sense, active_senses, forbidden_renderings, lexicon_ref, cross-tradition analogues, and editorial notes. The forbidden_renderings discipline is dense (heikhal forbids "temple"; hashmal forbids "electricity"; yored-merkavah forbids "ascender" / "mystic"; kerubim forbids "cherubs").
• New frame controller (backend/frames/merkavah_ascent.yaml): six operational senses, six scholarly-relation lenses, prompt-module rendering rules, output validators, three examples mapped to the three passages of this shipment.
• Scholarly corpus seed (corpus/scholarly/heikhalot/): six editorial summaries (~7,500 words total) covering the corpus overview, the yored-merkavah paradox, the seven-heikhalot geography, theurgic recitation, the full Scholem-Halperin-Idel-Davila-Boustan-Wolfson-Elior debate, and the kavod-throne complex. Density exceeds the existing kabbalah / akbarian / christian-apophatic collections.
• Citation manifest extension: 20 Heikhalot scholarship entries added to corpus/verified_citations.yaml training_memory[].
• Three benchmark specs at benchmarks/specs/heikhalot-rabbati-{1-prologue, 19-seal-passage, 24-throne-approach}.yaml.
• Hebrew source-of-record: three .heb.txt files in src/content/source-texts/ with the verbatim Wertheimer text per passage.

What comes next for the corpus

The HR ascent-arc triple is the first shipment, not the last. The immediate continuity items:

• Heikhalot Zutarti water-water passage (Schäfer §§407-410). The most-cited single Heikhalot passage in all of modern scholarship -- the sixth-palace warning attributed to R. Akiva: "when you reach the place of pure marble stones, do not say 'water-water,' for it is written 'he who speaks falsely will not stand before my eyes'" (Ps 101:7). Lives in the HZ corpus, not HR. The most natural next target.
• 3 Enoch / Sefer Heikhalot opening chapters. Odeberg 1928 PD English already covers this corpus -- which means a 3 Enoch Targum pass would activate the Layer 4.5 PD-comparison machinery for the first time on a Jewish-corpus passage. The cross-translator-memory layer (which has run on Akbarian and Christian-apophatic but never on Hebrew/Aramaic source material because the PD index had no Hebrew/Aramaic content) would finally fire across two scripts.
• The thirteen queued Hekhal lexicon pages. The drift incidents from this shipment named real substrate gaps; closing them in a dedicated lexicon-authoring session compounds across every future Heikhalot translation.
• Editor sign-off pass. Cross-check the three HR translations against the Schäfer Synopse §§81-93 / 219-228 / 245-258 paragraph parallels; cross-check against Smith/Karr and Davila for translational divergences; verify the 20 new training-memory citations against printed sources. Status flips to verified only after this pass.
• Shiur Qomah as an editorial-policy question. The corpus is the controversial measurement-of-the-divine-body literature; Maimonides read it as inauthentic or as anthropomorphism requiring philosophical reinterpretation. Whether to ship Shiur Qomah translations is an editorial question separate from whether the engine can produce them.

How to cite, how to reach the editor

These translations are CC-BY-SA-4.0. Reproduction with attribution is welcome. The recommended citation:

Hekhal Editorial (2026). Heikhalot Rabbati {1, 19, 24} -- Targum machine-assisted
translation from Wertheimer's Batei Midrashot (Jerusalem 1893-1897).
Hekhal Project. https://hekhal.org/texts/heikhalot-rabbati-{1-prologue,
19-seal-passage, 24-throne-approach}-targum. Frame controller:
merkavah-ascent v1.0. Glossary: heikhalot-merkavah v0.1.
Status: machine-assisted (editor sign-off pending).

Heikhalot scholars who would like to engage the apparatus -- particularly on the descender-paradox rendering choices, the gate-guardian transliteration conventions, or the alphabetical-cascade preservation discipline -- are warmly invited to correspond through the project channels. The engine's full audit packages for these three runs are available on request: spec.yaml + prompt.txt + output.json + drift_report.json + registry_report.json + citation_report.json + comparison.md per passage. The reproducibility surface is the project's signature commitment; the audit trail is what makes the machine-assisted status genuinely accountable rather than handwaved.

Companion reading

• Codex: Heikhalot & Merkavah -- the editorial codex framing of the corpus.
• Scaffolding Matters More Than the Schema -- the Phase-1 methodology paper articulating the registry-grounded translation discipline that the HR shipment ships under.
• Frame as Citable Parameter -- the Phase-2 paper on frame controllers as citable interpretive instruments.
• The apophatic-priority triangle -- the cross-tradition lexicon experiment whose Sefer Yetzirah 1:7 panel is the closest pre-Heikhalot Jewish-mysticism Targum work and the obvious editorial-juxtaposition companion to the present HR shipment.
• The full Targum catalog -- hand-curated index of every Targum output across all four traditions and the methodology section.