About this wing · 关于本区

Every statement the Classics Reception wing makes about itself is gathered here: where each text and number comes from, under what licence it is republished, how the counts were recomputed, what the method does not establish, whom the studies name and what was withheld to protect them, how to ask for removal, and what the passphrase curtain does and does not do. The study pages themselves carry only the material and a link back to this page. 本区关于自身的一切说明集中于此：各文本与数字之来源、再发布所依之许可、计数如何重算、方法所不能证明者、研究所涉及之人及为保护他们而未发布之项、如何请求删除，以及口令门帘之实际作用。各研究页只留材料本身与指向本页之链接。

Counts and sources计数与来源

Both studies were re-laid as house-style pages, so that what ships is only what may ship: no screenplay text, no research scratch directories, no working documents. The two source projects remain the record of how the work was done; this wing is the part of them that can be published. 两项研究皆按本站体例重新排布，所发布者仅限可以发布的部分：不含剧本文字，不含研究过程目录，不含工作文档。源项目仍是工作过程的记录，本区所载则是其中可公开的部分。

Study A — the screenplay concordance研究甲：剧本对照

Counts, each recomputed from the file it comes from.计数，每项均由其来源文件重算。

Study B — the Chinese commentary corpus研究乙：中文评论语料

Why the frozen generation, and not the larger one何以取冻结版而不取较大者

A later generation of the corpus holds 311 entries, 296 actors and 108 outlets. The analysis layer was never recomputed against it: its own declared totals still read 273 / 264 / 102, and its confidence distribution sums to 310 against 311 entries. Mounting the larger corpus would put every statistical table and every plate at a denominator that does not match the browsable entries. The 273-entry generation is the one that recomputes end to end, and every count in the ledger below is recomputed from it. It is the generation mounted here: the corpus app was rebuilt against the frozen files so the tables, the plates and the browsable entries all stand on the same 273. 语料另有较晚一版，计 311 条、296 位评论者、108 家媒体。该版之分析层从未重算：其自述总数仍为 273 / 264 / 102，置信度分布合计为 310，而条目为 311。若挂载较大者，则所有统计表与图表之分母将与可检索条目不符。273 条一版可端到端复算，下文所列各数皆由其重算。此即本站所挂载之版本：语料应用已依冻结版重新构建，故统计表、图表与可检索条目同以 273 条为准。

Counts, each recomputed from the file it comes from.计数，每项均由其来源文件重算。

Method, and what it does not establish方法及其未能确立者

Both studies state their limits in the same place as their findings. Four of those limits belong on this page, because a reader deciding whether to trust a number should meet them before the number. 两项研究皆将其界限与结论并置。其中四项宜列于本页，以便读者在见到数字之前先见其限度。

Framing agreement falls short of the target框架编码一致率未达标

Two coders overlapped on 39 entries. They agreed on the form of engagement in 39 of 39, on citation of the classics in 39 of 39, on citation of other films in 38 of 39, and on the stage of the discussion in 38 of 39. On frame type they agreed in 28 of 39, which is 71.8% against a stated target of 85%. Frame-type conclusions are reported with that figure attached, and the modules that depend on stage and engagement form are unaffected by it. 两位编码者重叠 39 条。参与形式一致 39 条，古典引用一致 39 条，影片引用一致 38 条，讨论阶段一致 38 条。框架类型一致 28 条，即 71.8%，而自定目标为 85%。凡框架类型之结论均附此数，依赖阶段与参与形式之模块则不受其影响。

Two prose statements in the dossier were wrong and are corrected底本两处表述有误，已更正

The method note stated that a visual line below OCR confidence 50 is unresolved. 42 lines fall below 50 and 40 are unresolved; the two exceptions were collated directly against the page image, and they are the only image-collated edges in the corpus. The source audit reported 23 insert cut: directions; 23 lines begin with the string and 28 contain it, so the count holds only under a direction-initial reading. Both statements are corrected in the method page rather than repeated. 方法说明称 OCR 置信度低于 50 之视觉行一律列为未决。实则低于 50 者 42 行，列为未决者 40 行；两处例外经页面图像直接核校，亦为全语料中仅有之图像核校对应边。来源审查称 insert cut: 指示 23 处；以该字符串起首者 23 行，含该字符串者 28 行，故此数仅在以起首为准时成立。两处表述已在方法页更正，不予沿用。

Figures taken from outside, and one that is corroborated外部数字，及其中一项之佐证

Several figures that circulate in commentary about the film have no stored capture anywhere in the source project: global box office, a Metacritic score of 88, the Maoyan and Taopiaopiao ratings, the Douban rater counts, and a claim of more than 200 searches performed. They are carried as external assertions, attributed and marked as unverified here. One is different: a saved page corroborates a Rotten Tomatoes critics score of 94. That capture is held offline and does not ship. 评论中流传之若干数字，源项目内并无留存的抓取记录：全球票房、Metacritic 评分 88、猫眼与淘票票评分、豆瓣评分人数，以及执行检索二百余次之说。此类数字以外部断言之形式载录，注明出处并标为未经本站核实。其中一项有别：所存页面可佐证烂番茄影评人评分为 94。该抓取件仅离线保存，不随本区发布。

What a concordance edge is, and is not对应边之所指与所不指

An accepted edge records that a panel judged a screenplay line to stand in a stated relation to a stated Greek locus, at a stated strength. It records a reading, not an authorial intention, and the strength letters carry that distinction: 7 edges are A, 191 are B, 274 are C and 3 are D. Three books of the Odyssey carry no accepted anchor at all, and the coverage report names them rather than smoothing the gap. 一条采纳边所记录者，为评议组判定某一剧本行与某一希腊语出处之间存在某种关系，并具某一强度。此为一种读法之记录，与作者意图无涉，强度字母即承此别：A 7 条、B 191 条、C 274 条、D 3 条。《奥德赛》有三卷全无采纳锚点，覆盖报告直书其事，不作弥缝。

The people this wing names本区所涉及之人

Study B quotes living people. In the frozen generation described here it holds 264 actor records and 102 outlet records; 266 of its 273 entries carry a verbatim quotation, and most carry a link to the post quoted. 136 of those actors are platform handles with no public professional identity, accounting for 136 entries, of which 132 carry a quotation, 115 a link to the original post, and 21 a locator that is not a URL. They are distributed across Douban 49, Weibo 27, Zhihu 20, WeChat 15, Bilibili 11, traditional media 7, Xiaohongshu 4, podcast 2 and HK/TW 1. 研究乙引述在世之人。此处所述冻结版有评论者记录 264 条、媒体记录 102 条；273 条条目中 266 条载有原文引语，多数附所引帖文之链接。其中 136 位评论者为无公开职业身份之平台账号，涉 136 条条目，载引语者 132 条，附原帖链接者 115 条，所附定位非链接者 21 条。其分布为豆瓣 49、微博 27、知乎 20、微信 15、哔哩哔哩 11、传统媒体 7、小红书 4、播客 2、港台 1。

The quotation policy引用体例

Quotations are short excerpts of what was published in public, with a median length of 39 characters and a maximum of 117, and each is presented with a link to the original post wherever one was recorded. No quotation is reproduced at a length that would substitute for reading the post itself. 所引皆为公开发表内容之短小节录，中位长度 39 字，最长 117 字，凡有记录者均附原帖链接。所引长度不至于取代阅读原帖。

No background was researched for the private stratum私人一层未作背景调查

The project's rule is that a tier-C actor gets no background research, and the rule held mechanically: of the 129 tier-C actors in the published generation, zero carry any education detail, institution name or education level — every one of those fields is null or marked as not researched. The same check on the later, unpublished generation returns zero of 160. No gender was inferred and no age proxy was constructed for this stratum. 项目定例为 C 层评论者不作背景调查，此例在数据上成立：本处所刊一版 C 层评论者 129 位，无一载有学历细节、机构名称或学历层级，凡此三项皆为空值或标记为未作调查。以同法检核未刊之较晚一版，160 位中亦无一例外。此一层未推断性别，亦未构造年龄替代变量。

Five things withheld from the mount挂载时保留不发布之五项

  1. The named silent register. The source appendix names 61 accounts across six buckets (26, 20, 7, 4, 3 and 1) as having not commented — an evidenced negative inferred from searches that failed, whose own notes concede login walls, six verified accounts, and a limited search window. The intention is that the bucket counts be published and the names not. All 61 names are in fact still in the mounted data/appendix.json, and that file is fetchable by direct URL. 具名之未发言名录。源附录分六组列出 61 个账号（26、20、7、4、3、1）为未曾评论者。此为由检索未果推得之否定性结论，其自注亦承认存在登录墙、仅核实六个账号、检索窗口有限。本意为各组计数予以发布，姓名不予发布。实则六十一个姓名仍在所挂载之 data/appendix.json 内，该文件可经直接链接取得。
  2. The IP-province markers. Ten entries in this generation carry a coarse geolocation inside the actor's identity label, in the form of a province or country tag on a Douban handle; the later generation carries twenty-four. The project's own audit places IP-province at platform-aggregate level under an ethics gate, so the per-entry tags are to be stripped before shipping. They have not been: all twenty-four are still in the mounted entry file, which is one consequence of the generation mismatch noted above. IP 属地标记。本版有十条条目于评论者身份标签内载有粗略地理位置，形式为豆瓣账号后所附之省份或国别；较晚一版则有二十四条。项目自身之审核将 IP 属地置于平台汇总层级并设伦理关卡，故逐条标记应于发布前移除。实则尚未移除：所挂载之条目文件仍载全部二十四条，此即上文所述版本不符之一端。
  3. An unsourced allegation against a named person. Two entries, one key event and one essay carry a third-party accusation of plagiarism against a named, identifiable individual. The dataset holds no URL for the accusation itself. Under the site's anti-fabrication rule it cannot be published as it stands, so the clause is to be cut and the reach facts around it kept. The cut has not been made: the clause is still in the mounted entry file, the key event and the essay. 针对具名个人且无出处之指控。有两条条目、一项关键事件与一篇短论载有第三方对某具名可识别个人之抄袭指控，而数据集中并无该指控本身之链接。依本站不得虚构之定例，此项按原样不可发布，故应删去该分句，保留其周边之传播事实。此项删削尚未施行：所挂载之条目文件、关键事件与短论中，该分句仍在。
  4. Fourteen mistyped increment records. Fourteen actor records added in a later increment are typed as identified persons and stamped as background-coded while every background field reads unknown; their own notes say the increment was taken without that check. Publishing the frozen generation would remove them; the mounted actor file is the later generation, so all fourteen are still there. 十四条类型有误之增补记录。较晚增补之十四条评论者记录被标为实名个人并注明已作背景编码，而其背景字段皆为未知，其自注亦称该次增补未作此项核查。发布冻结版本可排除此十四条；今所挂载之评论者文件为较晚一版，故十四条俱在。
  5. The raw coding and background-research files. The two coder files, the adjudicated coding file, the four background-research files and the research working directory stay out of the mount; the tables they produced are what ships. 原始编码与背景调查文件。两份编码者文件、裁定后之编码文件、四份背景调查文件及研究工作目录皆不挂载，所发布者为其产出之表格。

Removal requests删除请求

Anyone quoted here may ask for their entry to be removed or their handle to be replaced by a bare platform label, with no reason required and no argument offered in return. Write to the site editor, Wu Ching-Yuan, through the address on the rights and takedown page, which also carries the full policy this section summarises. 凡在此被引述者，均可要求删除其条目，或将其账号名代以纯平台标签，无须说明理由，本站亦不作辩驳。请循权利与删除页所载地址与本站编者吴靖远联系，该页并载有本节所撮述之完整政策。

Rights, licences and the copyright boundary权利、许可与著作权界限

The Greek text希腊语文本

Greek quotations are extracted from the Perseus canonical-greekLit EpiDoc file for the Odyssey, urn:cts:greekLit:tlg0012.tlg002.perseus-grc2, a digitisation of A. T. Murray's 1919 text, repository DOI 10.5281/zenodo.596309, released under CC BY-SA 4.0. This study selects and rearranges that text into book-and-line context windows, adds stable research links, and labels known lineation anomalies; the pinned source is not altered. The extracted Greek contexts remain CC BY-SA 4.0. 希腊语引文取自 Perseus canonical-greekLit 之《奥德赛》EpiDoc 文件，urn:cts:greekLit:tlg0012.tlg002.perseus-grc2，系 A. T. Murray 1919 年文本之数字化本，库藏 DOI 为 10.5281/zenodo.596309，依 CC BY-SA 4.0 发布。本研究就该文本作选取与重排，成卷行语境窗口，增补稳定研究链接，并标注已知的分行异常，所锚定之原始文件未作改动。所提取之希腊语语境仍依 CC BY-SA 4.0。

The screenplay剧本

The screenplay capture is user-supplied and copyrighted. Its OCR text is retained only in a local private directory and is excluded from every release manifest. In the public line-level tables the text column is replaced by a SHA-256 fingerprint of the line, and those fingerprints reproduce against the private ledger for all 4,406 lines with zero mismatches. The only running text published anywhere in the study is Homeric Greek. The interactive close study carries the same fingerprints with no text field, and its 5,654 text keys are all Greek. 剧本采集本由用户提供，受著作权保护。其 OCR 文字仅存于本地内部目录，并排除于所有发布清单之外。公开之逐行表格中，文字一列代以该行之 SHA-256 指纹，此指纹对内部台账全部 4,406 行复算相符，无一不合。本研究所发布之连续文字仅有荷马希腊语一种。交互式细读页所载指纹相同而无文字字段，其 5,654 处文字键皆为希腊语。

One rights claim in the source dossier is corrected here底本一项权利表述于此更正

The source project states that its public screenplay tables contain only stable locators and SHA-256 fingerprints. That holds of the tables and fails of two control files. Those two rule files carry 681 match patterns, of which 236 occur verbatim in the private OCR once normalised, and 61 of those run to four tokens or more, amounting to roughly 295 words of screenplay wording in the clear. Both files are nevertheless still in this mount. Dropping them is the obvious remedy, at the cost of candidate-generation reproducibility, and it has not been done; until it is, the wording above describes the tables and not the whole release. 源项目称其公开之剧本表格仅含稳定定位符与 SHA-256 指纹。此说于表格成立，于两个控制文件则不成立。该两个规则文件载有匹配式 681 条，其中 236 条经规范化后与内部 OCR 逐字相符，其中 61 条长达四个词以上，合计约 295 个词之剧本文字以明文存在。然此二文件仍在挂载目录内。删去二者为显然之补救，代价为候选生成之可复现性，此举尚未施行；在施行之前，上述说法只适用于表格，不适用于整个发布集。

The residual is stated rather than glossed. Measuring, for every private line, the longest run of consecutive words it shares with any file in the mount as it stands: 2 lines share 8 words or more (one of them 10), 2 share 7, 10 share 6, 26 share 5 and 60 share 4, so 100 of 4,406 lines — 2.3% — are exposed at four consecutive words or more, while 4,018 lines, 91.2%, share no run of three. Dropping the two rule files would cut that to 26 lines, 0.6%, with 4,166 lines, 94.6%, sharing no run of three. After the two rule files, the remaining sources are the adjudication file, the interactive study and the alignment edges, in that order, where a rationale paraphrases a line closely enough to echo it. 此项残余据实写明，不作掩饰。就每一内部行计其与挂载目录现状下任一文件共有之最长连续词串：8 词以上者 2 行（其一为 10 词），7 词者 2 行，6 词者 10 行，5 词者 26 行，4 词者 60 行；故 4,406 行中有 100 行，即 2.3%，在四个连续词以上有所暴露，另有 4,018 行，即 91.2%，无三词以上之共有词串。若删去该两个规则文件，则降为 26 行，即 0.6%，无三词以上共有词串者为 4,166 行，即 94.6%。除该两个规则文件外，其余来源依次为裁定文件、交互式细读页与对应边文件，皆因裁定理由之转述贴近原行而生回声。

The briefing底稿

A third page, a briefing written for a podcast episode on the film, quotes published criticism at some length: reviewers, translators, scholars and Chinese commentators, each named, dated and linked to the original. Every quotation is marked as opened and checked, relayed through a named intermediary, or unverified. The same removal route applies: anyone quoted there may ask for the passage to be shortened, attributed differently, or removed. 第三页为一期谈该片的播客所写之底稿，较长地引述已发表之评论：影评人、译者、学者与中文评论者，均具名、注日期并链接原文。每条引文标明已开页核对、经具名中介转述或未核。删除途径相同：凡在其中被引述者，均可要求缩短、改注或删去该段。

The commentary corpus评论语料

Entries record publicly posted commentary, with the author's own handle or byline, the platform, the date, a short quotation and a link. Copyright in each quoted post remains with its author; the excerpts are published as quotation for research and criticism, and the coding, the tables and the analysis are the work of this project. The rights page carries the full statement, the quotation policy, the privacy posture and the takedown route. 条目记录公开发表之评论，含作者账号名或署名、平台、日期、短小引语与链接。所引各帖之著作权仍属其作者，节录以研究与评论之引用名义发布；编码、表格与分析则为本项目之工作。权利页载有完整声明、引用体例、隐私处置与删除途径。

The curtain over the screenplay study剧本研究区之门帘

One of the two studies asks for a passphrase, and so does the dated briefing. Every page under /reception/nolan-homer/ draws the curtain, as does /reception/odyssey-briefing/; the commentary study at /reception/cn-commentary/, this page and the rights page do not, so the wing can be described, linked and read without asking for anything. 两项研究之中，仅剧本研究设有口令：/reception/nolan-homer/ 之下各页均张此帘，/reception/odyssey-briefing/ 亦然；评论研究 /reception/cn-commentary/、本页与权利页则不设，使本区可被描述、链接与阅读，无须先行索取口令。

What the curtain does, stated plainly门帘之实际作用，直说如下

The curtain is a courtesy gate. It is drawn by JavaScript in the reader's own browser, so it slows a casual arrival and stops nothing else. The JSON, CSV and JSONL files under this wing are fetchable by direct URL whether or not the curtain has been opened, and with JavaScript turned off no curtain is drawn at all. The privacy control on this wing is the decision about what ships, described above; the curtain is not that control. Real restriction would be HTTP basic authentication at the server over the study directory and the data directory, which is what the site's server runbook prescribes for the other curtained wing. 门帘仅为礼节性设置。其由读者浏览器中之 JavaScript 绘制，故可稍阻随手到访，此外别无作用。本区之 JSON、CSV 与 JSONL 文件，无论门帘开合，皆可经直接链接取得；若关闭 JavaScript，则门帘根本不会出现。本区之隐私控制在于上文所述“何者发布”之取舍，门帘并不承担此项控制。若需真正的访问限制，则应于服务器端对该研究目录及数据目录设置 HTTP 基本认证，本站服务器手册对另一设帘区域即作此规定。

The passphrase is issued separately by the site editor, and is the same word that opens the Sculpture wing. 口令由本站编者另行告知，与雕塑区所用者相同。

Page notes各页说明

The statements of source and layer that stood at the head of each generated page of the Nolan–Homer dossier, and the briefing's legend of verification marks.诺兰—荷马卷宗各生成页页首原有之来源与层次说明，及底稿之核验标记说明。

Greek context annex 希腊语上下文附录

Layers: Greek — Perseus tlg0012.tlg002.perseus-grc2 (Murray 1919 base text), verbatim from the dossier. Translation — A. T. Murray, Homer: The Odyssey, Loeb Classical Library, 1919 (public domain), from Perseus tlg0012.tlg002.perseus-eng3; segments follow his five-line milestones, so the first and last may overrun the window by up to four lines (marked). Outline — written for this page, 2026-09-06, CC BY-SA 4.0; it summarises the poem, not the screenplay. Book and line numbers are the poem's.

Base-text anomaly reminder

The annex must be read with control/anomaly_ledger.json: 3.304–305 are reversed in XML order; 10.456 and 16.101 are absent; 14.62–65 preserves the source transposition; and Book 23 omits actual 23.48 while mistagging actual 23.49 as raw 23.48. No supplementary Greek has been inserted here.

Seascapes of the Odyssey 海景

Greek: Perseus tlg0012.tlg002.perseus-grc2 (Murray 1919 base text), the same pinned file as the dossier. Translation: A. T. Murray, Homer: The Odyssey, Loeb Classical Library, 1919 (public domain), Perseus tlg0012.tlg002.perseus-eng3; segments follow his five-line milestones, so the first and last may overrun a passage by up to four lines (marked). Outlines and selection: written for this page, 2026-09-06, CC BY-SA 4.0. Line numbers are the poem's.

Technology in the Odyssey 技术

Greek: Perseus tlg0012.tlg002.perseus-grc2, the dossier's pinned text (every line asserted at build time). Chinese: line-by-line translation and notes by the dossier's expert panel, second version 2026-09-06, CC BY-SA 4.0. English: A. T. Murray, Homer: The Odyssey, Loeb Classical Library, 1919 (public domain), Perseus tlg0012.tlg002.perseus-eng3, in his five-line segments (edge overruns marked). Outlines EN/ZH and paraphrases: written for this page, CC BY-SA 4.0. Quotations from scholarship are verbatim, with page, and remain their authors'.

编订说明（自技术页移入）。日期：2026-09-06（第二版，同日校订：统一体例，改正标点，删去本地路径，补译注）。希腊文据 Perseus 底本 tlg0012.tlg002.perseus-grc2，与本档其余部分同源，逐行核对无异。中文为本次逐行自译，T04 复核既有译稿。每个核心段前后各扩展十行，扩展部分也全部翻译。跨行句子保持承接，不将截断窗口冒充完整讲话。技术名物的不确定性见各段译注。

Social classes: Odyssey and Athens 社会身份

Greek: the dossier's pinned Odyssey text (every line asserted) and the Perseus TEI of the Aristotelian Constitution of the Athenians (tlg0086.tlg003.perseus-grc2) and Demosthenes 57 (tlg0014.tlg057.perseus-grc2), each prose section asserted verbatim. Chinese: the panel's line-by-line translation, contexts, notes and comparative discussion, second version 2026-09-06, CC BY-SA 4.0. English: Murray 1919 for the Odyssey and Murray 1939 (Loeb, via Perseus) for Demosthenes; the Constitution sections are the page's own working rendering and are marked so. The glossary counts Homeric terms on the pinned text at build time. Quotations from scholarship are verbatim with page; the Wilson review is quoted from the briefing's verified transcript.希腊文据卷宗底本与 Perseus TEI 逐字核对；中文为专家小组逐行自译及对读讨论（第二版，2026-09-06）；英译取 Murray 1919、1939，《雅典政制》三节为本页工作译文并加标注；词表中荷马用语的行数据底本于构建时统计。

编订说明（自社会身份页移入）。编订日期：2026-09-06（第二版，同日校订：统一标点，删去本地路径，各段增补对读讨论，末附社会、政治、经济、文化用语的希英汉词表）。希腊文据 Perseus 底本：荷马 tlg0012.tlg002.perseus-grc2，《雅典政制》tlg0086.tlg003.perseus-grc2，德摩斯梯尼 tlg0014.tlg057.perseus-grc2，逐字核对无异。中文为本次自译，H04 复核既有译稿。荷马每个核心选段前后各扩展十行，扩展部分也全部翻译。雅典散文依自然小节完整引用，不虚构诗行编号；每段说明前后论述。共 97 组荷马逐行对照、5 个散文完整小节。研究来源与论证语境另见同档研究记录（SOCIAL-CLASSES-RESEARCH-NOTES.md）。比较沿四个维度展开：财富与生产资料、自由或奴役身份、家户与社会声望、参与公共权力的资格。荷马的伊萨卡与前四世纪的雅典相隔数百年，诗中的英雄过去、诗歌形成期与雅典民主制不能重合；对读的目的是看同一批词在两种社会里各指什么，不是把一方的制度填进另一方。

文本依据。希腊文直接取自 Perseus TEI（tlg0086.tlg003.perseus-grc2）。只将连续空白统一为单个空格，保留字词、标点与源文件中的特殊符号；中文为本附件自译。

核验。四个希腊文小节均与 Perseus TEI 按连续空白归一化后的全文逐字比对；每节均附完整自译。

Map: Nolan's locations, Homer's places 地图

Basemaps are Esri's World Physical, Shaded Relief, Terrain, Ocean and Hillshade tile services, which draw no modern political boundaries; the tiles are requested from Esri when the page opens and are the only external resource in the wing. Filming locations follow the dossier's verification of the Wikipedia production section, movie-locations.com, the FiSahara statement, the Carnegie note and Wilson's LRB review; coordinates follow the Pleiades gazetteer's representative points where it has a precise one (79 markers carry a Pleiades id), otherwise the editor's placement, with 'approx' marking a point set by description. Homeric places are classed as text (a real place the poem names), tradition (an ancient identification, quoted from Thucydides, Pausanias, Herodotus and Strabo with their public-domain translations), disputed (rival modern claims, quoted where the source is held) or unlocated (the poem gives no place; the marker holds the note). Every passage is pulled at build time from the pinned Odyssey text with Murray 1919, the Perseus Iliad with Murray 1924, and the Perseus Strabo and Thucydides; nothing is retyped. Books not held (Bittlestone 2005, Luce 1998, Bérard) are named without quotation.底图为 Esri 的世界自然地理、地貌晕渲、地形、海洋与山影瓦片，不绘现代政治边界，是本区唯一的外部资源。取景地据卷宗对维基制作节、movie-locations、FiSahara、卡内基与威尔逊书评的核对；坐标凡 Pleiades 地名录有精确点者从之（79 点附其编号），余由编者标定，approx 表示按描述放置。荷马地点分四类：诗中实名之地、古人认定（引修昔底德与斯特拉波）、今人争议、诗未给地点。所有原文于构建时自底本取出，不重打字。未藏之书只列名不引。

Footer statement, moved from the map page. Basemaps: Esri World Physical, Shaded Relief, Terrain, Ocean, Hillshade (tiles from Esri; no political boundaries drawn). Gazetteer ids and most ancient coordinates: Pleiades (CC BY), snapped to its representative points where precise. Texts: Perseus (Odyssey grc2 pinned by the dossier; Iliad grc2; Strabo grc2 with Hamilton–Falconer; Thucydides grc2 with Crawley; Pausanias with Jones; Herodotus with Godley); Murray 1919 and 1924, public domain. Site notes and coordinates: this dossier, 2026-09-06, CC BY-SA 4.0; approx marks a point placed by description. Filming locations follow the dossier's verification of the Wikipedia production section, movie-locations.com, FiSahara, Carnegie and the LRB. Generated by scripts/build_reception_map.py from map/sites.json.

Intertexts: Homer, Virgil and the Homeric apocrypha 互文读本

Greek and Latin: the pinned Odyssey text, the Perseus Iliad, Aeneid and Hymn to Aphrodite (every verse row asserted at build time), and, for the Cycle, the Contest, the parodies and the Life of Homer, the panel's transcriptions from West's Loeb editions (2003), Bassino's edition of the Contest (2019) and Evelyn-White's Epigrams, with the carrier of each fragment named. Chinese: the panel's essay and line-by-line translations, second pass 2026-09-06. English: Murray 1919 and 1924, Theodore C. Williams 1910 (Aeneid, blank verse by card), Evelyn-White 1914 (Hymn, Epigrams); eleven passages have no public-domain English on the site and are given in Greek and Chinese only. Scholarship is quoted verbatim with page from Horsfall 2013, Quint 2018, Bassino 2019, West 2003 (both Loebs), Davies 2014 and Nagy's open texts; Steiner 2010 and Faulkner 2011 could not be opened and are paraphrased with that caveat. Proclus' summaries are marked as prose summaries, not verse; West's Apollodorus supplements are excluded.希腊文、拉丁文据 Perseus 与卷宗底本逐行核对；残篇据 West 洛布本、Bassino 校本转写并注明引述载体；中文为专家小组的读本与逐行译文；英译取 Murray、Williams、Evelyn-White 公版，十一段无公版英译者只给原文与中文；学术引文注页码，两种未能打开的著作据小组记录转述并注明。

编订说明（自互文页移入）。本轮提供31组选段、594个原文—中文对照单元。原文每行下接自译；完整史诗通常给核心前后各十行，或延展为完整场景。残篇给实际保存范围，散文按句或段配译，不能伪造失传的上下十行。完整原文和前后情节见互文长选段读本（本页下方各选段）；十项现代研究的实读页码、约百词规模的中文论证语境及阅读限制见研究实读记录（本页各选段之学术语境）。本轮使用本地古典语料、Magalia同源《奥德赛》数据及网上作者、出版方资料，未把尚未读到的著作列为直接论据。

Animals in the Iliad and the Odyssey 动物

Same pattern as the seascapes page. Greek: the dossier's pinned Odyssey text and the Perseus Iliad (tlg0012.tlg001.perseus-grc2). English: A. T. Murray 1919 (Odyssey, five-line segments) and 1924 (Iliad, by card), public domain. Selection, categories and the before/after outlines: written for this page, 2026-09-06, CC BY-SA 4.0. Both poems' animal vocabulary is counted on whole word-forms at build time. Scholarship quoted verbatim with page: Kirk 1985, Bremmer 2007, van Straten 1995, Wilson LRB 2026; Scodel 2005 and the Phillippo–Waite piece are named from the briefing's record without quotation.与海景页同一做法：希腊文据卷宗底本与 Perseus《伊利亚特》；英译 Murray 1919、1924 公版；选段、分类与前后提要为本页所撰；两诗动物词汇于构建时按整词形统计；学术引文注页码，未藏者只列名。

Homer in context (Study C) 荷马语境

On 6 September 2026 the six text-in-context pages and the map were moved out of the curtained screenplay study to the open landing /reception/homer/, because none of them concerns Nolan's script. The old URLs under /reception/nolan-homer/reports/ and /reception/nolan-homer/map/ remain as redirect stubs. The pages share one chrome: the wing tab strip, a "Homer in context" line, a related-pages line into the Ancient Ecology family, a skip link and a common footer. Their figures are drawn by the generators from the same data as their tables.2026 年 9 月 6 日，六个文本语境页与地图自设帘的剧本研究区移至开放入口 /reception/homer/，因其皆与诺兰剧本无关；旧址保留为跳转页。各页共用标签栏、荷马语境导航、通往古代生态家族的相关页导航、跳转链接与统一页脚；图表由生成脚本据表格同源数据绘制。

The briefing 底稿之核验标记

底稿每条回应附来源与核验标记；标记之义如下。来源文件 drafts/20260905 odyssey podcast source-based responses v2.md，资料更新至 2026-09-05。

  • ［核］：原文页面或原文本已打开核对，引文照录。
  • ［转］：只经第二手转述核对，括号内注明转述者。
  • ［未核］：找到线索但未能打开原页，只能当作待查。
  • ［剧本］：依据本地《诺兰—荷马〈奥德赛〉剧本对照档案》（2026-08-24 建，72 页剧本 OCR，4,406 行逐行判定，475 条对应、244 条偏离）。本版已用英文维基的成片剧情梗概逐条对过，凡对上的改标［剧本＝成片］。
  • ［案］：我的判断或整理，不是来源说的话。
  • 希腊文引文来自本地 Perseus TEI（Murray 1919 本），卷行可查；中译为我的直译，只求达意。
  • 中文评论数据来自本地《诺兰〈奥德赛〉中文评论者全景调查》（Kimi，2026-08 建，273 条目，140 条高置信）。它自报的边界：小红书登录墙、微信只见搜狗快照、18 条待查。本版另加了 8 月下旬至 9 月初的材料。

The interactive close study 交互式细读页

The close study carries its own About view (sources, review procedure, verification limits, reuse rights and file checks), reached from its toolbar. Its release documents are rendered as pages: README, method and evidence contract, rights and attribution, citation audit, source audit, panel reconciliation.细读页自带关于视图（来源、复核程序、核验限度、再用权利与文件校验），由其工具栏进入；其发布文档已渲染为页面：README、方法与证据约定、权利与署名、引证审计、来源审计、专家小组调和。

Generators 生成脚本

The three text-in-context pages are generated, not hand-written: scripts/build_greek_annex.py (annex; Greek asserted byte-identical to the dossier), scripts/build_seascapes.py (47 passages from reports/SEASCAPES.json; vocabulary counted from the text at build time, epithets only beside their nouns) and scripts/build_technology.py (five windows from the panel's TECHNOLOGY-CONTEXTS.md; every Greek line asserted against the pinned text) and scripts/build_social_classes.py (eight units from SOCIAL-CLASSES-CONTEXTS.md with the glossary from SOCIAL-CLASSES.meta.json), scripts/build_intertexts.py and scripts/build_animals.py (47 passages of both poems from ANIMALS.json). Murray's translation is loaded by scripts/lib_murray.py. The figures on the landing and in the briefing are drawn by scripts/build_reception_figures.py from the data files, with a table twin under each.四个文本语境页均由脚本生成：附录、海景、技术、社会身份；希腊文逐行与底本核对，Murray 英译由共用模块载入；首页与底稿之图表由脚本据数据文件绘制，图下各附数据表。

Changes变更记录

5 September 20262026 年 9 月 5 日

The commentary study's figures were refreshed on 5 September 2026 from the sources named beside each: the Douban rating read directly from Douban's own interface, 8.6 across 466,577 raters (8.5 across 364,946 on 24 August); the worldwide gross at 1.586 billion US dollars on 3 September per the-numbers; the mainland gross past 600 million yuan on 31 August. The curve, the key-events list and two of the seven essays carry the new points with their dates; earlier figures stay where they were, dated. 评论研究之数字已于 2026 年 9 月 5 日按各自所注来源刷新：豆瓣评分直接自豆瓣接口读取，为 8.6，466,577 人评价（8 月 24 日为 8.5，364,946 人）；全球票房按 the-numbers 9 月 3 日为 15.86 亿美元；内地票房 8 月 31 日破 6 亿元。评分曲线、大事记与七篇短论中的两篇均已加入带日期的新点；旧数字原处保留，并注明日期。

Douban rating, 24 July to 5 September 2026Line from 8.2 to 8.6 with eight dated points.8.08.28.48.6mainland release 8-142026-07-24: 8.2 · 8,949 raters8.2 · 8,9492026-08-03: 8.3 · 30,000 raters2026-08-14: 8.3 · 89,885 raters2026-08-15: 8.42026-08-18: 8.52026-08-20: 8.52026-08-24: 8.5 · 364,946 raters8.5 · 364,9462026-09-05: 8.6 · 466,577 raters8.6 · 466,57707-2408-1409-05
The Douban rating from opening to the 5 September reading: 8.2 across 8,949 raters on 24 July, 8.5 across 364,946 on 24 August, 8.6 across 466,577 on 5 September. Nolan's films usually open high on Douban and drift down; this one climbed.豆瓣评分自开分至 9 月 5 日读数：7 月 24 日 8.2（8,949 人），8 月 24 日 8.5（364,946 人），9 月 5 日 8.6（466,577 人）。诺兰作品在豆瓣惯常高开低走，此片逆势爬升。stats.json · doubanCurve · Douban API snapshots
Table view · 数据表
DateScoreRatersNote
2026-07-248.28,949开分（五星37.4%、四星+五星75.3%）；来源：搜狐娱乐/腾讯新闻 2026-07-24
2026-08-038.330,000约3万人评分（约数）；来源：新浪#奥德赛豆瓣评分涨至8.3#、财新08-10
2026-08-148.389,885内地上映日；来源：腾讯新闻 2026-08-14
2026-08-158.4—来源：新浪08-15、财新08-17、网易08-16
2026-08-188.5—来源：豆瓣小组帖08-18、新浪/搜狐/腾讯08-18~20
2026-08-208.5—上映6天票房破3亿；来源：腾讯08-19、网易08-20
2026-08-248.5364,946豆瓣 rexxar API 实测 rating=8.5 / 364,946人评价（8月下旬抓取时点，仍缓涨）；08-23/24媒体报道超24万人评价，API数为抓取时点最新值
2026-09-058.6466,577豆瓣 rexxar API 实测 rating=8.6 / 466,577人评价（2026-09-05 抓取；接口需移动端 UA 与 Referer，网页版重定向至 sec.douban.com）
A third page joined the wing the same day: a sourced briefing written for a podcast episode on the film, in Chinese, following a five-part outline (Chinese audiences; Hollywood's two ancient-epic waves; classicists and the film, Emily Wilson's reviews in particular; the homecoming film and the anti-war question; memory, the dog and the slaves in Homer). Every claim in it carries one of five tags: opened and checked, relayed, unverified, from the screenplay concordance, or the author's own judgement. It sits behind the curtain, since it quotes living people at length. 同日本区新增第三页：为一期谈该片的播客准备的中文底稿，按五节提纲（中国观众；好莱坞古典题材的两次浪潮；古典学界与该片，尤其艾米莉·威尔逊的评论；战场归乡片与反战之问；荷马中的记忆、狗与奴隶）逐条作答。其中每一断言均标以五种记号之一：已开页核对、转述、未核、出自剧本对照档案、作者己见。因其大段引述在世之人，置于门帘之后。

6 September 20262026 年 9 月 6 日