跳至正文

判决书没有判死刑，执行成了死刑——再审的是判据，不是那个声称

目录

机制裁决第 153 篇 · 对称双向第 148 篇 · section B1 · 全库第 211 篇 本篇审的是「1989 年科学界已经证明冷聚变是骗局」「过量热是真实的核过程」「Google 和 ARPA-E 正在为冷聚变平反」「这是被主流利益集团扼杀的新能源」这组话——以及它背后那本 1989 年 3 月出版、收稿到修回只有十天的原始论文、那份判了「不信」却留下门缝的 ERAB 判决书、那本 35 年没涨过量级的量热账、和那台从听证会走进 ARPA-E 项目库的再审机器。先读三句红线：

  1. 本篇不裁决「冷聚变/LENR 是否对应真实的核过程」——那是实验物理学至今未结案的问题；本篇审的是「死刑」「复活」这组程序事实被读成的三层结论（骗局已被证明／声称已被平反／体系在压制真相）各自的承重方式。本篇不构成任何科研方向、投资或能源政策建议。
  2. 「判决书」与「执行」是两件不同的事。本篇的裁决落在两者之间的缝里：声称的原文内伤是真的（过量热与核产物之间 7–10 个数量级的裂缝印在 FP 论文自己的数字里），判决书没有判死刑也是真的（ERAB 明说无法断然证伪、无欺诈结论、留下「modest support」门缝），而死刑的执行同样是真的（53 天走完、APS 现场 8/9 投票、「pariah field」逐字），再审程序的落地也是真的（FOA 四条判据、八项目近千万美元、Nature 双扶手论文）——跳变不发生在电解池里，发生在「执行」被读成「判决」、「再审程序」被读成「现象平反」的那个动作里。
  3. 全篇承重句均给出可点击来源；中英文逐字引用一律取自实际取回并落盘的文件（含 Wayback 原件、OSTI/ARPA-E 官方 PDF、法庭案卷与 APS harvest 全文），自算部分写明算式与口径。凡只经二手转引到手的引语一律标注转引层级，不与一手逐字混排。

零、一句话裁决

「1989 年 3 月 23 日那场发布会公布的声称，原文自带一条七个数量级以上的裂缝」是真的——Fleischmann 与 Pons 的论文一面写下「Enthalpy generation can exceed 10 W cm⁻³…It is inconceivable that this could be due to anything but nuclear processes」，一面在同一篇论文里自报氚产率仅 1–2×10⁴ /s、中子约 4×10⁴ /s，而解释那份热需要 10¹¹–10¹⁴ /s 的核事件——声称与证据差 7 到 10 个数量级，这条裂缝不是反对者算出来的，是论文自己的数字。「1989 年科学界证明冷聚变是骗局」是半句假话——ERAB 报告逐字写的是「it is not possible at this time to state categorically that all the claims…have been convincingly either proved or disproved. Nonetheless, on balance…」，五条结论条没有任何一条判欺诈，还留下建议 2 的门缝「sympathetic toward modest support for carefully focused and cooperative experiments within the present funding system」；死刑是执行出来的，不是判出来的——发布会后第 40 天 APS 巴尔的摩会场 8/9 投票判死、Nature 八周内三大阴性背靠背、「pariah field」的社会放逐，把「on balance 的不信」执行成了永久驱逐，而那条门缝 33 年无人走（DOE 2022 年 FOA 逐字自认「few such proposals were submitted, and none were funded by DOE」）。「Google 2019／ARPA-E 2023 说明冷聚变正在平反」是误读——再审真实发生，但宾语不是现象：Google 论文 420 次量热零过量热、摘要逐字「yet to yield any evidence of such an effect」，同时写下「the debunking of cold fusion in 1989 was perhaps premature」——两只扶手焊在同一段，扶手是给「判定规范」的，不是给「现象」的；ARPA-E 的 FOA 逐字把要回答的问题写成「should this field move forward…or does it conclusively not show promise?」，四条判据里量热被降级为不作主诊断。「彻头彻尾的骗局、一切信号皆伪造」同样不立——Schenkel 的电子屏蔽 1000±250 eV 超出裸核预期百倍是全程唯一的反常正信号且被 Nature 论文自报，ERAB 的「bursts」口子从未被正式收回，Iwamura 嬗变声称的独立复现停在 0.1% 的分歧位至今未裁决。35 年过去，这个声称从未达到可复现标准（主张方自报天花板五分之一、DOE 2004 逐字「not repeatable, the magnitude of the effect has not increased in over a decade」、唯一的独立双盲为负向），而被处决的领域换了四次名字活了下来——聚变、冷聚变、LENR、CMNS、LCF：词死了，活法换了名字。

本篇的独占格在第四章（现场账：发布会早于投稿一天、收稿修回十天、两校「不再公开宣布」协议六天后被背刺的程序跳全记录）、第五章（判决书原文账：ERAB 五条结论＋门缝建议逐字，与史家三版本「fiasco of the century」的落差）、第六章（技术内伤账：裂缝与分支比 >10¹³ 的两道算术、ERAB 量热裁决与「bursts」口子、EPRI 临界参数四件套与主张方自报的五分之一天花板、Miles 30/33 氦关联与唯一独立双盲的负向）、第八章（再审程序账：2016 NDAA 报告语言→DIA「disruptive technology」→2021「stalemate」→FOA 四条判据→八项目 $9,956,549→结题稀疏→Thunderbird +15(2)% 无净增益）、第九章（Google 双扶手账：$10M、stealth 四年、420 样本零信号与「perhaps premature」同段并焊）、第十章（商业极账：Rossi 一亿美元合同、ERV 报告、开庭 48 分钟和解与永久 NDA）、第十一章（叙事生产线账：53 天化石、维基机制性死刑、词频化石——词死了，活法换了名字）。

灵魂句：判决书没有判死刑，执行成了死刑；再审程序是真的，再审的却不是那个声称。词死了，活法换了名字——而那条七个数量级的裂缝，35 年没人跨过。

一、本篇测什么

1.1 被审对象

被审的是一组互相咬合的说法及其用法：

  • 死刑句：「1989 年科学界已经证明冷聚变是骗局／病态科学」（程序裁决层——但判决书原文写了什么，是第五章的事）
  • 声称句：「过量热是真实的、化学无法解释的核过程」（现象声称层——它的内伤与 35 年账，是第六章的事）
  • 复活句：「Google／Nature 2019、ARPA-E 2022–2023 说明冷聚变正在被平反」（程序叙事层——再审程序的宾语是什么，是第八、九章的事）
  • 压制句：「这是被主流利益集团扼杀的新能源」「冷聚变研究者是被迫害的伽利略」（反向消费层——第十二、十三章称重）

1.2 结构胎记

死刑跳 × 名号跳 × 复活跳 ＋ 反向红跳。

  • 死刑跳：把「1989 年的处置结果」读成「判决书的内容」——真实档案里 ERAB 没有判欺诈、明说无法断然证伪、留下门缝建议；把「on balance 的不信」执行成永久驱逐的是随后 53 天里的社会机制（媒体狂欢→三大阴性→APS 现场投票→pariah 化），不是那份报告。死刑跳是把执行的社会学读成判决的科学（第四、五章）。
  • 名号跳：每换一次名字就换一把判据——从「聚变」（要对核产物账）到「冷聚变」（小写的聚变，仍要核产物）到 LENR（低能核反应，不再承诺聚变通道）到 CMNS（凝聚态核科学，连「反应」都软化）到 LCF（晶格约束聚变，NASA 用法）——词频化石实证：题名口径 1989 年 167 篇掉到 2018 年 5 篇，而活法在别处继续（第十一章）。名号跳的检测器只有一个问题：「这次改名，判据跟着换了吗？」
  • 复活跳：把「机构重新打开问题」读成「现象被平反」——Google 的交付物是 reference experiment（判定规范），ARPA-E 的立项理由是「should this field move forward…or does it conclusively not show promise」（判定问题），Thunderbird 自称交付的是 2019 年呼吁的那个参考实验；这三件事的宾语全部是「该怎么判」，没有一件的宾语是「热是真的」。复活跳发生在「判定规范的确立」被读成「现象的确认」那个动作里（第八、九章）。
  • 反向红跳（第十二章）：四句虚无侧读法——「彻头彻尾一切造假」「骗局已被证明所以永远不该再审」「体系 1989 年的处置公正无瑕」「商业极的一地鸡毛证明科学侧同样为零」——逐句称重，四句全不立，但每一句的诱人之处都给出交代。

1.3 落位：section B1 与分界

本篇落 section B1（物理前沿），与三篇邻近篇分界写死：

  • 与核聚变篇（2026-07-24）分界：那篇审主流磁约束/惯性约束聚变的 Q 值会计学——「永远还有三十年」话术下真实的工程账；本篇审的是 1989 年从聚变家族里被驱逐出去的那一支旁支的死刑与再审。一边是主账房里的亏空话术，一边是被逐出账房的私账——两篇合起来是「聚变」这个名号在 20 世纪后半叶的两种耗散方式。
  • 与室温超导篇（2026-07-08）方向相反：那篇审的是新声称被快速证伪的机器（LK-99 数月内全球复测闭环）；本篇审的是一台反方向机器——一个 1989 年已被执行死刑的旧声称，如何在 35 年里以改名、换判据、制度门缝的方式维持未结案状态。两篇合起来是「声称的死亡与复活」的对称样本：一个是快死，一个是不死。
  • 与 g-2 篇（2026-08-04）、惰性中微子篇（2026-08-12）、G 篇（2026-08-16）同族：那三篇审「换尺」——理论尺、通量分母尺、汇编尺在动；本篇是换尺家族最极端的变体：这里换的不是尺，是声称本身的名字与判据——每换一次名号，「算不算数」的标准跟着挪一次。详见第十三章。

1.4 去重实测

python 全库扫描 212 个 markdown／8,553,332 字符（2026-08-17 开题日实测）：冷聚变／cold fusion／LENR／Fleischmann／excess heat／SPAWAR／CMNS／氘化钯／Mizuno／McKubre 全库零命中；Pons 命中均为神经解剖学 pons（脑桥）小写义、Brillouin 命中均为 Léon Brillouin 的负熵义、重水 命中均为 SNO 中微子实验语境——冷聚变/LENR 专篇零命中，处女地确认。

1.5 边界

  • 不裁决冷聚变/LENR 是否对应真实核过程；不推荐任何实验方案、投资标的或能源政策；不评价任何在世研究者的个人信誉——Rossi 一章只登记「法庭文件与公开记录里记载的行为与程序」，不评价动机。
  • 对主张方阵营（ICCF 社群、lenr-canr 档案库、Nucleation Capital 等）与怀疑方阵营（Koonin、Lewis、Petrasso、Shanahan、Garwin 等）的材料对称称重：阵营属性一律标注，利益相关方的话不单独承重。
  • lenr-canr 是领域自办档案库（Jed Rothwell 维护、立场明确），凡引用其托管文件一律标注属性；其中 DOE 2004 报告与 ERAB PDF 恰与 ARPA-E FOA 脚注 23 的官方引用件同件，provenance 可用，属性照标。
  • 中文媒体叙事只登记话术形态，不评价具体报道与记者。
  • 本篇不构成任何科研、投资、法律或政策建议。

1.6 方法备案

  • 取证通路：DOE/OSTI 官方报告 PDF（ERAB 中期报告 OSTI 5706354、最终报告 DOE/S-0073；DOE 2004 复审报告）；ARPA-E 官方 FOA PDF（DE-FOA-0002785）与项目库；Nature 摘要层与作者分享链接（rdcu.be）、Nature 2025 Thunderbird 全文页；APS harvest 全文；JEC / Fusion Technol. 等期刊经出版社与档案库双路；Congress.gov 与 GovInfo 的 NDAA 报告语言；CourtListener/PACER 系法庭案卷转引件；New Energy Times 与 Lewan 的庭审一手文件报道；Wayback CDX＋id_ 原件快照；OpenAlex/PubMed 词频 API 自算。本轮共取回并落盘 745 件原始文件（/root/chat/tmp/cold-fusion/raw/）。
  • 承重统计一律 python3 复算（词频化石、数量级裂缝、票型比例）；凡「自算」必标注，官方数字与自算数字分行陈述不混排。
  • 任务书预设勘误 8 处（全部有落盘证据）：
    1. 「Langmuir 1989 APS 演讲」不存在——Irving Langmuir 1957 年去世；真实档案为其 1953-12-18 在 GE 的「pathological science」讲稿（档案号 68-C-035），全文 1989 年 10 月刊于 Physics Today（DOI: 10.1063/1.881205）；把「pathological science」贴到冷聚变头上的最早一手记录是 1989-05-02 APS 巴尔的摩现场 CERN 的 Douglas Morrison 之口（NYT 1989-05-03 报道逐字）。
    2. arXiv:1905.06885 与冷聚变无关（实为数学论文）；Google 论文无 arXiv 版本（Unpaywall 登记 closed）；正确卷号为 Nature 570: 45–51（569 是同日配发件所在卷）。
    3. 「2016 国会听证」不实——实为众议院军事委员会 FY2017 NDAA 报告语言（H.Rept. 114-537 第 87 页）指示 DoD 于 2016-09-22 前做一次闭门 briefing；最终协商报告 H.Rept. 114-840 零命中（条款未入法）；DoD briefing 原文 PDF（FOIA 16-F-1333）403 未取回——本篇按「委员会报告语言＋briefing 未核验」表述，不写「听证」。
    4. 「GSI 复核 Miles」查无实证——真实存在的独立复核是 Morrey 等六实验室双盲研究（Fusion Technol. 18: 659，负向：⁴He 信号可经 25 μm 表面蚀刻去除，指向污染）。
    5. 「Duncan 为 Industrial Heat 评估 Rossi 设备」无一手证据——Duncan 在 60 Minutes 复核的对象是 Energetics Technologies 的 Pd-D 工作（不是 Rossi）；IH 内部评估由 Murray 团队执行（案卷 Doc 30 ¶80）；Robert Duncan 获 Gates 方面 600 万美元资助在 Texas Tech 做 LENR 是另一件事。两个 Duncan 角色本篇拆开写。
    6. Rossi 从 IH 处所得精确为 1150 万美元（2012 年 150 万＋2013 年 1000 万两期），不写「约 1100 万」。
    7. 「ARPA-E 2023 立项」口径修正——2022-09-13 宣布 exploratory topic、2023-02-17 公布八个入选项目。
    8. 「约 1100 万美元」之外其余金额一律按文件逐字（八项目合计 $9,956,549；Woodford 敞口 $50M 按 New Energy Times 据法庭文件的报道层级标注）。

二、核验标记与层速览

核验标记沿用本库四档：[一手逐字]＝原文逐字取自实际取回文件（含 Wayback id_ 原件与官方 PDF）；[文献较稳]＝多源一致或权威评估；[需亲核]＝正文未全取回或版本存疑；[多源检索]＝方法学阴性检索（含「缺席即证据」登记）。另本篇新增一条本篇专用标注：[转引层]＝引语只经二手/三手文本到手（如 Huizenga 原书页码经 JCF17 会议论文集转引、Simon《Undead Science》经书评转引），与一手逐字分行陈述、绝不混排；阵营属性标注＝lenr-canr（领域自办档案库，立场明确）、Infinite Energy（主张方刊物）、New Energy Times（挺方调查媒体）等来源出处的材料一律在引用处注明属性。

层 名称 核心问题 裁决方向
三 守真锚 哪些是真的 六条，一条不动
四 现场账 1989 年的 53 天怎么走完 发布会早于投稿；修回次日开记者会
五 判决书原文账 ERAB 到底写了什么 五条否定＋门缝建议；无欺诈结论
六 技术内伤账 裂缝与 35 年可复现性 7–10 个数量级；自报 1/5；双盲负
七 二审账 2004 年 DOE 复审怎么投 18 人票型逐字；结论与 1989 相同
八 再审程序账 2016→FOA→八项目→结题 宾语是判定规范，不是现象
九 Google 双扶手账 420 次零信号与「perhaps premature」 两只扶手焊在同一段
十 商业极账 Rossi 一亿美元官司 开庭 48 分钟和解；独立消耗可信度
十一 叙事生产线账 化石、维基、词频 词死了，活法换了名字
十二 反向红跳 四句虚无读法 四句全不立
十三 母裁决与遗留 换尺家族的极端变体 这里换的不是尺，是名字

三、守真锚（这一侧一条都不动）

在拆任何跳变之前，先把真实的那一侧钉死。以下六条本篇不做任何弱化：

守真锚一：声称的原文内伤为真——裂缝印在 FP 论文自己的数字里。 原文摘要逐字：「(b) Enthalpy generation can exceed 10 W cm-3 of the palladium electrode; this is maintained for experiment times in excess of 120 h, during which typically heat in excess of 4 MJ cm-3 of electrode volume was liberated. It is inconceivable that this could be due to anything but nuclear processes.」（Fleischmann, Pons & Hawkins, J. Electroanal. Chem. 261 (1989) 301，全文经 lenr-canr 镜像取回，领域自档案库属性标注[一手逐字]）。同文自报的核产物率：氚 1–2×10⁴ /s、中子约 4×10⁴ /s——而按已知 D+D 聚变能量解释瓦特级热需要 10¹¹–10¹⁴ /s 的核事件率[文献较稳]。裂缝 7–10 个数量级，是论文自己的数字对不上的，不是反对者外加的。勘误另把 γ 峰位由 2.5 改为 2.225 MeV、图 1A 纵轴刻度整体缩小 25 倍（同文后附 Erratum，Vol. 263[一手逐字]）。

守真锚二：判决书没有判死刑——ERAB 原文既无欺诈结论，也明说无法断然证伪，并留下门缝。 前言段逐字：「Consequently, with the many contradictory existing claims it is not possible at this time to state categorically that all the claims for cold fusion have been convincingly either proved or disproved. Nonetheless, on balance, the Panel has reached the following conclusions and recommendations.」（ERAB 报告，NCAS 全文镜像／lenr-canr PDF 合集／中期报告 OSTI 5706354[一手逐字]）。建议 2 逐字：「The Panel is sympathetic toward modest support for carefully focused and cooperative experiments within the present funding system.」[一手逐字]。五条结论全部是否定性证据评估（「do not present convincing evidence」「not persuasive」），没有一条使用「fraud」「hoax」或任何欺诈定性—— fraud 这个词在整份报告里不出现（OCR 全文检索零命中[多源检索]）。

守真锚三：死刑的执行是真的，而且是 53 天里走完的社会事件。 发布会 1989-03-23；Nature 三大阴性（Lewis/Caltech Nature 340: 525、Williams/Harwell Nature 342: 375）八周内落地；APS 巴尔的摩专题会议 1989-05-01/02 记者会上九位主讲人被问「是否判犹他声称死刑」，八人答 yes（NYT 1989-05-03，Wayback 存档全文[一手逐字]）；在场者 Goodstein 1994 年回忆逐字：「executed between them a perfect slam-dunk that cast cold fusion right out of the arena of mainstream science」（The American Scholar 63(4), Caltech 托管 PDF[一手逐字]）。从狂欢到处决 53 天（Deseret 1989-03-24 头版「CENTURY’S GREATEST ACHIEVEMENT」→ APS 5-02；Time 1989-05-08 封面「Fusion or Illusion?」，详见第十一章）。

守真锚四：35 年过去，声称从未达到可复现标准——这是主张方自报与两审官方记录共同钉住的。 主张方旗舰账 EPRI TR-107843 自报：「Though the heat-producing phenomena were obtained in only about one-fifth of the cells…」（EPRI TR-107843，lenr-canr 镜像，领域自档案库属性标注[一手逐字]）；DOE 2004 复审逐字：「Most reviewers, including those who accepted the evidence and those who did not, stated that the effects are not repeatable, the magnitude of the effect has not increased in over a decade of work…」（DOE 2004 报告[一手逐字]）；Google 2019：「none of the 420 samples we evaluated provided evidence of excess heat」（Nature 570: 45–51，作者只读链接 rdcu.be/bEAsT[转引层双源互证]）；Miles 氦关联唯一的独立六实验室双盲复核为负向（Morrey et al., Fusion Technol. 18: 659[文献较稳]）。天花板是主张方自己报的，不是反对者估的。

守真锚五：再审程序真实发生，且全部以「判定规范」为宾语——同时反常残留也真实存在。 ARPA-E FOA DE-FOA-0002785 逐字：「establish clear practices to rigorously answer the question, ‘should this field move forward… or does it conclusively not show promise?’」；四条判据含「量热不作主诊断、量热预算 ≤10%」「排除 prosaic 解释」「A/B 队人员不得重叠」（ARPA-E 官方 FOA PDF，DE-FOA-0002785[一手逐字]，详见第八章）；八项目合计 $9,956,549（ARPA-E 项目库自算加总[自算]）；Thunderbird（Nature 644: 640, 2025）Pd 靶原位电化学加载使 D-D 聚变率 +15(2)%、逐字自报「did not generate a net energy gain」[一手逐字]。反向侧同样为真：Google 项目全程唯一反常正信号——Schenkel 组电子屏蔽 1000±250 eV、中子产额超裸核预期约百倍——白纸黑字印在同一篇 Nature 论文里（Nat. Mater. 社论称「exceeds theoretical predictions by two orders of magnitude」[一手逐字]）；ERAB 1989 的「bursts」口子（「cannot be attributed to problems with accuracy or calibration alone and are presently not understood」）从未被任何后续官方文件正式收回[一手逐字]。

守真锚六：商业极的程序记录是真的，且它独立消耗着这个领域的可信度。 Rossi v. Industrial Heat（1:16-cv-21199）合同总额 $100.5M 三期、ERV 报告声称倍增「often greater than sixty」、IH 反诉逐字「not a real test at all, but a carefully scripted effort to deceive」；2017-06-28 陪审团开庭、6-29 mistrial、7-05 重新开庭后 48 分钟内和解、with prejudice 全员撤诉＋永久保密协议（案卷：CourtListener, Rossi v. Darden, 1:16-cv-21199；庭审与和解细节经 New Energy Times/Lewan 一手文件报道，挺方调查媒体属性标注[一手逐字/转引层混合]）。USPTO 双轨同样为真：In re Swartz 232 F.3d 862 (Fed. Cir. 2000) 确立冷聚变类实用性质疑框架，而「去核化」撰写的授权案持续发生（Rossi「Fluid heater」US 9,115,913，2015；Navy US 8,419,919，2013；Brillouin US 12,590,732，2026-03-31[一手逐字]）。商业极的一切都不证明科学侧为零——但它解释了为什么「复活」叙事在商业端口总是先走一步。

四、现场账：1989 年的 53 天（程序跳全记录）

4.1 发布会先于论文：一条被新闻稿自己钉死的时间线

程序事实链，每一环都有一手件：

  • 1989-03-06，犹他大学与 BYU（Jones 组）达成书面协议，条款逐字：「C4. No further public comments on results of research until papers submitted — No further public announcements of the results of either teams’ research would be made until after the papers were submitted for publication.」（Jones 组史稿，Montclair 镜像[一手逐字]）
  • 1989-03-22，BYU 组接到 DoE 人士电话得知次日发布会，逐字记录：「if the press conference were held, we at BYU would interpret this as a violation of the agreements between the two universities. His words were that we would interpret such a press conference ‘as a stab in the back.’」（同件[一手逐字]）
  • 1989-03-23，犹他大学发布会召开；当天新闻稿自认论文尚未见刊：「Their findings will appear in the scientific literature in May.」（犹他大学新闻稿全文，New Energy Times 存档，挺方调查媒体属性标注[一手逐字]）。发布会上被问「别处是否也在做」，犹他研究副校长 Brophy 答：「…We’re not aware of any such experiments going on.」（Jones 史稿[一手逐字]）——而此时 BYU 的 Jones 中子实验已运行数月，且 3-13 Fleischmann 本人已通报 Harwell 的 Williams 并启动其实验（Lewenstein 编年[一手逐字]）。
  • 1989-03-24，Fleischmann 与 Pons 向 Nature 投稿（后撤回，转投 JEC）；Deseret News 当天头版标题逐字：「FUSION DISCOVERY AT U. COULD RANK AS CENTURY’S GREATEST ACHIEVEMENT」（Deseret 1989-03-24[一手逐字]）。

即：发布会早于任何同行评议文本的公开可得性，协议条款 C4 在六天内被己方背刺，「成果将于五月见刊」是新闻稿自己写的。程序跳不在别人手里，在这条时间线里。

4.2 FP 原文：收稿修回十天，裂缝七到十个数量级

论文扉页逐字：「(Received 13 March 1989; in revised form 22 March 1989)」（J. Electroanal. Chem. 261 (1989) 301，lenr-canr 镜像全文，领域自档案库属性[一手逐字]）——修回次日即开发布会。该收稿日期当年即有争议（Science 称修回稿 3-26 才到 JEC、Technology Review 称 3-30，Lewenstein 编年脚注 OCR 逐字），本篇两说并录、不裁判[文献较稳]。

内伤数字（论文自报）：热功率密度 >10 W/cm³、120 小时以上、累计 >4 MJ/cm³；同时自报氚 1–2×10⁴ /s、中子约 4×10⁴ /s。而 1 W 的 D+D 聚变热对应约 10¹² /s 的聚变事件率——声称与产物之间 7–10 个数量级[自算，按 D+D→⁴He+23.8 MeV 或两分支加权，算式见第六章]。Huizenga 后来的判决书式表述逐字：「The chasm between Fleischmann and Pons’ claim of ‘excess heat’ and the upper limit on the intensities of the associated fusion products continues to be many orders of magnitude!」（《Cold Fusion: The Scientific Fiasco of the Century》p.214，经 JCF17 论文集逐字转引并标注原书页码[转引层]）。

勘误两处（同文后附 Erratum，Vol. 263[一手逐字]）：γ 能量「p. 302 (3) Last line: 1H + n = 2D + γ(2.225 MeV)」——由 2.5 改为 2.225 MeV；「(10) Fig. IA: The y-axis should be labelled from 0 counts (minimum) to 1000 counts (maximum), ie. replace 0, 5000, 10000, 15000, 20000, and 25000 counts by 0, 200, 400, 600, 800, and 1000 counts respectively」——原图纵轴夸大了 25 倍。

4.3 三大阴性：八周内的三记重锤

Caltech Lewis 组（Nature 340: 525, 1989[一手逐字]），摘要逐字：「A series of experiments has been performed to determine whether nuclear fusion processes occur in palladium rods that have been electrochemically charged with deuterium. With a variety of metallurgical pretreatment procedures and different electrolytes, no evidence has been obtained for any excess enthalpy, neutron, gamma ray, tritium or helium production during electrolysis of D2O with palladium cathodes.」——五种声称产物一次扫光。Lewis 在 APS 现场的版本更短：「’We have seen no evidence whatsoever for nuclear reactions or even for unusual chemical reactions’」（LA Times 1989-05-02，lenr-canr 汇编 PDF，领域自档案库属性[一手逐字]）。注意「no evidence whatsoever」这句咒语 30 年后会在 Nature 社论里第二次出现（第九章）。

MIT Plasma Fusion Center Petrasso 组（预印本 PFC/JA-89-24 Rev., MIT DSpace 全文 PDF[一手逐字]），对 FP 的 γ 峰三点定量反驳逐字：「(1) the FPH line width at the reported 2.2 MeV γ ray to be a factor of two smaller than their instrumental resolution would allow at that energy; (2) that there is no indication of the Compton edge which should be distinctly evident at 1.99 MeV; and (3) that the neutron source strength reported by FPH is in error by a factor of 50…Furthermore, we argue their purported γ line actually resides at 2.5 MeV, not at 2.2 MeV. …We can offer no plausible explanation for the feature other than it is possibly an instrumental artifact with no relation to a γ-ray interaction.」——线宽窄于仪器分辨率允许值一半、缺少应有的 Compton 边、中子率高估 50 倍、峰位根本不在 2.2。这就是第六章「γ 峰塌方」的技术原文。

围绕这份 MIT 数据有一场延续十年的次级争议：Mallove 1999 年特别报告指控 MIT 的聚合曲线在 7-10 与 7-13 版本之间下移（Infinite Energy #24，lenr-canr 全文，主张方刊物属性[一手逐字]）；MIT 方的「stands by」回应目前只有 Mallove 报告内的转述层（MIT 原始新闻稿未取回）[转引层]。两边证据层级不同，本篇并录、标注层级差，不裁判。

Harwell Williams 组（Nature 342: 375, 1989[一手逐字]）：三种量热计设计、11 个 D₂O 电池＋14 个 H₂O/Pt 对照、最长 33 天连续电解——比 FP 声称的 120 小时窗口长约 6 倍，仍全部阴性（ERAB 表 2.1 逐字条目：「12. Harwell Lab. Williams et al. ISO A, ISO C-type, Open JM, Rod, cast, Beads. Ribbons 20-530 mA/cm2, 3-15 V none (11 cells, to 33 d) H2O, Pt (14 cells) no n, t, gammas [WIL]」[一手逐字]）。Harwell 是最早拿到 FP 内部信息的复现方（3-13 已启动），阴性结果的分量因此不同于事后追赶者。

4.4 APS 巴尔的摩现场：死刑是八比一投出来的

1989-05-01/02，APS 巴尔的摩专题会议。记者会记录（lenr-canr 汇编 PDF，领域自档案库属性；关键句有 NYT/LA Times 同期报道交叉印证[一手逐字]）：

  • 「death certificate」问答逐字：「Q: …Is anyone on the panel prepared to sign the death certificate this morning? PROF. JONES: Do you have it? PROF. MEYERHOF: I would.」——「死刑」这个词是现场记者提的，Meyerhof 签的字。
  • Jones 发起稻草投票：「…how many think we can rule that out at this time? Almost, that is fairly complete.」——针对犹他热量声称的 95% 置信度排除「几乎全体通过」；第二轮针对 BYU 低水平中子的排除则未获一致（Morrison 以「美丽的树—刚破土的小苗—种子」的比喻插话反对扩大判决——记者会记录原文逐字为「But this little sprout is still living and I do think it will grow into a pleasant flower…」，比喻链的压缩写法见工作笔记，此处不作逐字引语使用）。死刑判的是犹他的热，不是 BYU 的中子——这个差别在后来的叙事里被抹平了。
  • Morrison（CERN）定性逐字：「The entire episode, he said, was an example of ‘pathological science,’ in which an erroneous experiment initally [sic] gained some support, then prompted skepticism and finally led to denunciation.」（NYT 1989-05-03[一手逐字]）——「pathological science」贴到冷聚变头上的最早一手记录就在这里，不是 Langmuir 的鬼魂（Langmuir 1957 年已去世；其 1953 年 GE 讲稿全文 1989-10 才刊于 Physics Today，DOI: 10.1063/1.881205[一手逐字]）。Morrison 对欺诈问题的同时回答：「…this is not the first example this has happened… It very often happens that someone makes a mistake, it’s very enthusiastic, and it happens. …But it is not abnormal, it’s just human nature…」——定性为「人会犯错」，不是「人犯了罪」。
  • NYT 1989-05-03 Browne 定调篇（Wayback 存档全文[一手逐字]）：「In two days of meetings lasting until midnight, members of the American Physical Society heard fresh experimental evidence from many researchers that nuclear fusion in a jar of water does not exist.」；投票逐字：「At a news conference today, nine of the leading speakers were asked if they would now rule the Utah claim as dead. Eight said yes, and one, Dr. Johann Rafelski of the University of Arizona, withheld judgment.」；Koonin 重锤：「Dr. Steven E. Koonin of Caltech called the Utah report a result of ”the incompetence and delusion of Pons and Fleischmann.” The audience of scientists sat in stunned silence for a mement [sic] before bursting into applause.」——注意「全场沉默后爆发掌声」这个细节：掌声是给定性语句的，不是给数据的。这是社会处决的现场声学记录。
  • 程序性怨点（同文逐字）：「Top physicists directed angry attacks at Dr. Pons and Dr. Fleischmann…complaining that they had refused to provide details needed for follow-up experiments.」；Lewis：「”Pons would never answer any of our questions,”…”so we asked Los Alamos National Laboratory to put our questions to him instead”」；Oak Ridge 的 Dickens：「One published photograph of the Utah cell showed Pons’s hand, and that gave us the scale.」——复现方靠照片里的一只手估算电池尺寸。声称方不开放参数，是复现失败的共因之一，这一层责任记在声称方账上。
  • 阳性杂音同文照录：Stanford 的 Huggins 称「more confident than ever」，同段记 Meyerhof 检查其装置后认为「the experiment was flawed because of the system used to measure heat」。
  • Meyerhof 打油诗逐字：「Tens of millions of dollars at stake, Dear Brother, Because some scientist put a thermometer At one place and not another.」——这首讽刺诗本身就是第六章量热战的通俗版：温度计放错一个位置，就是全部故事。

在场者总结（Goodstein 1994，Caltech 托管 PDF[一手逐字]）：「For all practical purposes, it ended a mere five weeks after it began, on May 1, 1989, at a dramatic session of the American Physical Society, in Baltimore. …Steven Koonin and Nathan Lewis…executed between them a perfect slam-dunk that cast cold fusion right out of the arena of mainstream science.」——「五周」（发布会到 APS）与「53 天」（发布会到 NYT 定调）两个口径并存，本篇以事件链列示。

4.5 Georgia Tech 撤回：狂欢期的第一块骨牌

Lewenstein 编年逐字（OCR 原文日期字符残缺「10 A~ril」，按编年上下文补读为 10 April／12 April）：「10 April — Georgia Tech physicists lead by John Mahaffey hold 1:00 p.m. press conference to announce that they have measured neutrons and tritium.」「12 April — Georgia Tech team reports a flaw in its fusion experiment, failing to take into account the effect of temperature on the neutron detector.」（Cornell 馆存 Lewenstein 档案系录／镜像 PDF[一手逐字]）。4-10 开发布会宣布测到中子与氚，4-12 自报中子探测器的温度效应缺陷——两天撤回。这是狂欢期自我撤回的第一案，也是「开发布会先于自检」程序的第二个样本（第一个是 3-23 本身）。

五、判决书原文账：ERAB 写了什么、没写什么

5.1 委托书与程序

能源部长 Watkins 1989-04-24 委托信逐字（ERAB 附录 1.A）：「I request that the Energy Research Advisory Board (ERAB) assess this new area of research. Specifically, I would like the Board to: 1. Review the experiments and theory of the recent work on cold fusion. 2. Identify research that should be undertaken to determine, if possible, what physical, chemical, or other processes may be involved.」（NCAS 全文镜像／lenr-canr PDF，领域自档案库属性[一手逐字]）。冷聚变专门小组共同主席 John R. Huizenga（罗切斯特）与 Norman F. Ramsey（哈佛），成员含 William Happer Jr.、Allen J. Bard、Steven E. Koonin、Darleane C. Hoffman、Jacob Bigeleisen 等（名单逐字见同件）。小组 1989 年 5–11 月间五次公开会议、多个现场走访（site visits），1989-08 出中期报告（OSTI 5706354[一手逐字]）、1989-11-08 Huizenga 签交最终报告（DOE/S-0073）。

5.2 五条结论逐字：全部是否定性证据评估，没有一条是欺诈定性

  • 结论 1（热量—能源）：「the experimental results of excess heat from calorimetric cells reported to date do not present convincing evidence that useful sources of energy will result from the phenomena attributed to cold fusion.」[一手逐字]
  • 结论 2（热—核关联）：「In no case is the yield of fusion products commensurate with the claimed excess heat. In cases where tritium is reported, no secondary or primary nuclear particles are observed, ruling out the known D+D reaction as the source of tritium. The Panel concludes that the experiments reported to date do not present convincing evidence to associate the reported anomalous heat with a nuclear process.」[一手逐字]——「commensurate」（相称）是全文最关键的形容词：问题不是「一个产物都没看到」，是「产物的量级配不上热量的量级」。
  • 结论 3（新核过程）：「Based on these many negative results and the marginal statistical significance of reported positive results, the Panel concludes that the present evidence for the discovery of a new nuclear process termed cold fusion is not persuasive.」[一手逐字]
  • 结论 4（固体中氢的既有文献）：「Current understanding of the very extensive literature of experimental and theoretical results for hydrogen in solids gives no support for the occurrence of cold fusion in solids…」[一手逐字]
  • 结论 5（理论）：「Nuclear fusion at room temperature, of the type discussed in this report, would be contrary to all understanding gained of nuclear reactions in the last half century; it would require the invention of an entirely new nuclear process.」[一手逐字]

五条里最强的词是「not persuasive」与「contrary to all understanding」。全文检索 fraud／hoax／misconduct 三个词，零命中（OCR 全文机检[多源检索]）。判决书没有判死刑。

5.3 前言的自辩：它明说「无法断然证伪」

前言段逐字：「The claims of cold fusion, however, are unusual in that even the strongest proponents of cold fusion assert that the experiments, for unknown reasons, are not consistent and reproducible at the present time. However, even a single short but valid cold fusion period would be revolutionary. As a result, it is difficult convincingly to resolve all cold fusion claims since, for example, any good experiment that fails to find cold fusion can be discounted as merely not working for unknown reasons. …Consequently, with the many contradictory existing claims it is not possible at this time to state categorically that all the claims for cold fusion have been convincingly either proved or disproved. Nonetheless, on balance, the Panel has reached the following conclusions and recommendations.」[一手逐字]

这段话是双向称重的枢轴：它一方面给了主张方 35 年的生存空间（「无法断然证伪」＋「一次短暂但真实的冷聚变期都将是革命性的」），另一方面把不可证伪性的责任钉在声称自己的性质上（「任何失败的实验都可以被打折为『因未知原因没工作』」——不可证伪性不是判决书的缺陷，是声称的结构特征）。

5.4 门缝逐字：建议 2 与三条「遗留问题」

执行摘要：「The Panel also concludes that some observations attributed to cold fusion are not yet invalidated.」、「The Panel recommends against the establishment of special programs or research centers to develop cold fusion. However, there remain unresolved issues which may have interesting implications. The Panel is, therefore, sympathetic toward modest support for carefully focused and cooperative experiments within the present funding system.」[一手逐字]

建议章六条逐字节选：建议 1 反对专项拨款与专门中心；建议 2「The Panel is sympathetic toward modest support for carefully focused and cooperative experiments within the present funding system.」；建议 3 主张热量研究「focus primarily on confirming or disproving reports of excess heat」且用封闭电池与全复合条件；建议 4 指出「A shortcoming of most experiments reporting excess heat is that they are not accompanied in the same cell by simultaneous monitoring for the production of fusion products…such a claim should be supported by measurements of fusion products at commensurate levels.」；建议 5 把氚核查列为「desirable」；建议 6 给 Jones 式低水平中子留了「of scientific interest but have no apparent current application to the production of useful energy」的位置[一手逐字]。

门缝不是终稿迫于压力加的：1989-08 中期报告已逐字有「However, there remain unresolved issues and scientifically interesting questions stemming from reported cold fusion efforts. Some of these are relevant to the mission of DOE and should be handled by carefully focused and cooperative efforts within current programs by normal mechanisms for project selection.」（OSTI 5706354[一手逐字]）。判决书的执行部分（反对专项）被完整执行了 33 年；判决书的门缝部分（常轨支持聚焦实验）33 年无人走——DOE 2022 年 FOA 逐字自认「few such proposals were submitted, and none were funded by DOE」（第八章）。这不是判决书的问题，是执行史的问题。

5.5 史家三版本：判决书与定性的落差

三本同代史著给出三种定性，全部经 JCF17 论文集逐字转引并标注原书页码[转引层]：

  • Huizenga 1992《Cold Fusion: The Scientific Fiasco of the Century》p.214：「Fleischmann and Pons’ gamble to go public with their announcement of watts of ‘excess power’ from nuclear fusion at room temperature, before they had any solid evidence of fusion products, is the scientific fiasco of the century. …Room temperature nuclear fusion without commensurate amounts of fusion products is a delusion and qualifies as pathological science defined as ‘the science of things that aren’t so’.」（JCF17 论文集[转引层]）。注意判决句的宾语是「gamble to go public…before they had any solid evidence」——判的是程序冒险，落点仍是那条裂缝。
  • Taubes 1993《Bad Science》p.425：「Cold fusion—as defined by Stanley Pons and Martin Fleischmann, or Steve Jones, or as modified by John Bockris or Edmund Storms and Carol Talcott, or Bob Huggins-Stanford, or whomever—did not exist. It never had. There was at least as much empirical evidence, if not more, to support the existence of any number of pseudoscientific phenomena, from flying saucers to astrology.」[转引层]——书名从「fiasco」升级到「bad science」，定性从「世纪惨败」升级到「从未存在」。
  • Mallove 1991《Fire from Ice》p.x：「It is now reasonably clear that fusion reactions that liberate energy—near but very peculiar relatives of nuclear processes that are the lifeblood of the stars—can occur at room temperature. There is no chance whatever that cold fusion is a mistake.」[转引层]——对立极的全程在场证明：死刑执行后两年内，对立定性已经成书出版。

三个版本连同 Goodstein 的「feeding frenzy…science gone berserk」（Caltech 托管 PDF[一手逐字]）一起构成本案的判例法地层：官方文书（ERAB）最克制，史家定性两极分化，社会执行（pariah 化）比任何文字都重。死刑跳的全部材料都在这个落差里：被引用最多的是「fiasco of the century」，被忽视最多的是同一份报告的「on balance」与建议 2。

六、技术内伤账：两道算术与 35 年的可复现性

6.1 第一道算术：裂缝（热对产物，差 7–10 个数量级）

ERAB 把锚钉死在一张表里（表 1.1 逐字，OCR 上标按上下文补读）：「D + D –> 3He + n：3.27 MeV，1.90×10¹² reactions/sec per 1 W output；D + D –> T + p：4.03 MeV，1.54×10¹²」（NCAS 镜像[一手逐字]）——1 瓦聚变热必然伴随约 10¹² /s 量级的核产物。Huizenga 的对应表述（p.7，经 JCF17 转引）：「one watt of power from nuclear reactions must have associated with it approximately 10^12 (one million million) neutrons per second…」[转引层]。对照 FP 自报中子率约 4×10⁴ /s：瓦特级声称对应 10¹¹–10¹⁴ /s 所需，裂缝 7–10 个数量级[自算：10¹²/4×10⁴ ≈ 2.5×10⁷；若按 >10 W 与 4 MJ 累计外推，缝更宽]。

执行摘要对 Jones 式低水平中子的定性逐字：「Neutrons near background levels have been reported in some D2O electrolysis and pressurized D2 gas experiments, but at levels 10^12 below the amounts required to explain the experiments claiming excess heat.」[一手逐字]——低水平中子即便为真，离解释热也差 12 个数量级。裂缝是官方文书与论文自报数字共同钉住的，不是任何一方的修辞。

6.2 第二道算术：分支比（要热不要中子，得改 >10⁶；要 ⁴He 通道，得增强 >10¹³）

ERAB 逐字：「A change in the known branching ratios of more than one million, in favor of reaction (b) over reaction (a), would be required to explain experiments that claim to observe high fusion rates (either through heat or tritium production) without a corresponding high neutron flux. …However, no mechanism is known for inducing such a rapid energy-dependence in the branching ratio.」[一手逐字]

对「无中子的 ⁴He 通道」假说的逐字回应：「Some researchers have hypothesized that the D + D –> 4He + gamma (23.85 MeV) reaction, which is ordinarily some 10⁷ times weaker than reactions (a) and (b)…somehow dominates in cold fusion situations. To be consistent with the lack of neutrons, a very large enhancement of the gamma branch by a factor somewhere in excess of 10¹³ would be required. We know of no way whereby the atomic or chemical environment can effect such an enhancement, as this ratio is set by nuclear phenomena and is on a length scale some 10⁴ times smaller than the atomic scale.」[一手逐字]——原子/化学环境比核尺度大 4 个数量级，要去改一个核层面定死的分支比，需要 >10¹³ 的增强。DOE 2004 报告把这段原样复述了一遍（第七章），十五年没有新答案。

6.3 量热裁决：封闭电池阴性、「bursts」口子、与章末句

ERAB 量热节三句定调（全部逐字）：

  • 「At present, most experimenters who have performed calorimetry with closed cells under strict recombination conditions have reported no excess heat.」[一手逐字]——封闭电池＋严格复合条件下，多数实验者零过量热。
  • 标定漂移警告：「The Panel’s site visits have identified experimental uncertainties, e.g., nonlinearities of the calibration in power output vs. temperature, time dependence of calibration, and doubtful accuracy of data acquisition relative to the magnitude of the effects asserted…Even in laboratories that report excess heat, this effect, under apparently identical conditions, is not always reproducible.」[一手逐字]
  • 「bursts」口子：「However, there are reports of sporadic temperature ‘excursions’ or ‘bursts’ that apparently represent power outputs significantly larger than the input power. These events cannot be attributed to problems with accuracy or calibration alone and are presently not understood.」[一手逐字]——判决书自己留的技术口子，35 年没有任何后续官方文件正式收回或关闭它。
  • 章末句：「After assessing the reports from the different laboratories, considering the experimental difficulties and calibration problems, as well as a lack of consistency and reproducibility in observation of the excess heat phenomenon, we do not feel that the steady production of excess heat has been convincingly demonstrated.」[一手逐字]——否定落在「steady production」（稳态持续产热）上，口子在「bursts」（偶发冲激）上。后来 35 年主张方的全部正向报告几乎都长在「bursts」这个形态里——这正是判决书没被关掉的那个格子。

6.4 临界参数四件套与「五分之一」天花板：主张方自报的上限

主张方旗舰账（EPRI TR-107843，SRI/McKubre 组十年计划结题，lenr-canr 镜像，领域自档案库属性[一手逐字]）把触发条件收敛成四件套，逐字：「Four conditions were characteristic of all cells yielding episodes of excess heat: a D/Pd ratio >0.9, initial heat appearance times of 8-23 days, cathodic current densities above 0.1 A/cm2, and high D fluxes across the cathode surface. Excess powers in the range of 5-30% were observed and measured to an accuracy of ~0.4%.」

同一报告的自报天花板逐字：「Though the heat-producing phenomena were obtained in only about one-fifth of the cells, this work confirms the claims of Fleischmann, Pons, and Hawkins concerning the production of excess heat in D-loaded Pd cathodes at levels too large to be chemical transformations.」[一手逐字]——满足全部四件套的电池里，约五分之一出现过热现象。这是主张方十年国家级资助（EPRI）后的自报数字，本篇把它登记为这个声称 35 年来的实证上限。

McKubre 2009 年 ICCF15 综述（lenr-canr 镜像，领域自档案库属性[一手逐字]）把四件套写成经验公式：「Pxs = M (x-x°)2 (i-i°) |iD|」，阈值 x°≈0.875、i° 75–450 mA/cm²、iD 2–20 mA/cm²，且「a time threshold t° of at least 10 times the deuterium diffusional time constant」；同文给出主张方对可复现性的最强表述：「Also, at this point, any claim that the Fleischmann-Pons Effect is ‘irreproducible’ is not only unsound, it is unscientific. Where and when we are capable of reproducing all parameters critical to the effect, we reproduce the effect.」[一手逐字]——对称地，ERAB 1989 对同一现实的读法是：「At this time we cannot find clear differences in materials, cells, or operating conditions that can account for the widely irreproducible behavior.」[一手逐字] 同一批数据，一方读作「参数集齐了就可复现」，一方读作「找不到能解释成败差异的参数」——这个分歧从 1989 年活到今天，本篇并置两读、不裁判。

6.5 Miles 的 ⁴He–热关联：30/33 的账与唯一的独立双盲

声称本身（Miles 2003，ICCF10 综述，lenr-canr 镜像，领域自档案库属性[一手逐字]）：「A correlation between excess enthalpy and excess helium-4 was measured in 18 out of 21 experiments. The observation of no excess enthalpy was correlated with no excess helium in 12 out of 12 experiments. Thus 30 out of 33 experiments agree with the hypothesis that the excess enthalpy produced in cold fusion studies is correlated with helium-4 production: D + D → 4He + 23.8 MeV.」定量侧（McKubre 2009 转引 Miles–Bush[转引层，同一合作者网络]）：理论预期 2.5×10¹¹ ⁴He s⁻¹ W⁻¹，实测均值 1.4±0.7×10¹¹（54%）；Bush 在 SRI 的「复核」测得 1.5±0.2×10¹¹（58%）；并声称相关性随机概率仅「1:750,000」。

账本的结构性弱点：全部「确认」出自同一合作者网络。最早的确证论文（Bush-Lagowski-Miles-Ostrom, J. Electroanal. Chem. 304 (1991) 271）自述逐字：「the calorimetric electrolysis experiments reported here were performed at China Lake, and the analyses designed to establish the composition of the effluent gases were performed in Austin.」——量热在 China Lake、氦分析在奥斯汀，都是合作者自己做，非盲、非独立第三方[一手逐字]；SRI 复核的执行者仍是 Bush 本人；DOE 2004 文件列举的确认者（McKubre、Gozzi、ENEA Frascati）同属该网络。

唯一的独立盲法复核为负向：Morrey 等 1990（六家实验室双盲测量 Fleischmann 与 Pons 提供的五根钯棒，Fusion Technol. 18: 659，OSTI 官方记录[一手逐字]）摘要逐字：「All the materials, including the as-received palladium stock, contained easily measured quantities of 4He, well above amounts normally found in high-purity palladium. The 4He could be totally removed from at least two of the materials, including the as-received palladium stock, by surface etching the samples to a depth of ~25 μm. …No 3He was measured above detection limits in any of the materials by any of the six laboratories.」——原材料里就有可测的 ⁴He，且 25 μm 表面蚀刻即可完全去除：指向表面污染而非体内生成。任务书预设的「GSI 复核 Miles」查无实证（GSI 语境的 cold fusion 指超重元素合成反应道，与 LENR 无关[多源检索]）；真实存在的独立复核就是这次，且是负向。

6.6 Shanahan 的 CCS：1% 标定漂移 = 780 mW 假信号

质疑方的系统误差机制（Shanahan 2010, J. Environ. Monit., DOI 10.1039/c001299h[一手逐字]）：「In 2002, this author published a reanalysis of laboratory data claimed to have shown unequivocal excess heat…wherein a previously unrecognized systematic error was demonstrated to have the capacity to explain the observations without invoking a nuclear reaction. This error was termed the ‘calibration constant shift’ (CCS). …the variation in that term of ~1% (1 s) was found. This is a common precision level for a good analytical technique, yet it was adequate to explain an excess power signal of 780 mW, which was ~10 times the commonly assumed error level determined by baseline noise fluctuation.」——1% 的标定常数漂移（优秀分析技术的常见精度）足以凭空造出比基线噪声假定误差大 10 倍的「过量热」。注意这与 EPRI 自报「5–30% 过量、精度 ~0.4%」正面对撞：如果标定常数会漂移 1%，那么 5–30% 的信号区间与漂移区间大面积重叠。这场交锋从未被正式裁决——它直接通向第八章 ARPA-E 的第三条判据（量热不作主诊断）。

6.7 SPAWAR CR-39 径迹：三连径迹与「化学成因是完全可能的」

声称（Mosier-Boss 等 2009，Naturwissenschaften 96(1):135，DOI 10.1007/s00114-008-0449-x，摘要经 Europe PMC 取回[一手逐字]）：「Among the solitary tracks due to individual energetic particles, triple tracks are observed. …The presence of three alpha-particle tracks outgoing from a single point is diagnostic of the 12C(n,n’)3alpha carbon breakup reaction and suggests that DT reactions that produce ≥ 9.6 MeV neutrons are occurring inside the Pd lattice. To our knowledge, this is the first report of the production of energetic (> or = 9.6 MeV) neutrons in the Pd-D system.」

反驳链（Shanahan 2010 同文[一手逐字＋转引层]）：机械损伤对照——转引 Little（EarthTech）未发表报告：「We quickly discovered that mechanical damage often leads to round, track like marks after etching.」、「We were able to create various marks with sandpaper, needle points and simply by carrying around a chip in a pocket for a day.」；Little 的独立复现结论（Shanahan 转引）：「do show that chemical origin is a distinct possibility and therefore that nuclear origin is not a certainty.」[转引层]。层级标注：声称发表于 Springer 旗下同行评审期刊；反驳的关键件为未发表报告，双方证据层级不对称，本篇登记不裁决。SPAWAR 项目本身 2011 年终止（第八章）。

6.8 Iwamura 嬗变：三菱声称与丰田的 0.1%

声称（Iwamura 等 2002，JJAP 41: 4642，IOPscience 元数据取回[文献较稳]；后续 JCMNS 2011/2013 系列[一手逐字，领域期刊属性]）：D₂ 气体渗透 Pd/CaO 多层膜后 Cs→Pr 等元素转换声称。

丰田中央研究所复现（Takahashi, Kosaka, Hioki, Motohiro，JCMNS 13 (2014) 579–585，领域期刊属性[一手逐字]）——引言先交代复现史：「Few replication attempts at other laboratories have succeeded. One of the difficulties in reproducing the results of the Iwamura group may be that the amounts of the transformed element are too small to be detected by surface elemental analysis such as XPS…」；自己的结论逐字：「The amounts of Pr in the D2 permeated samples were one order of magnitude larger than those in the control samples not permeated with D2. …If we assume that the Pr atoms detected were converted from Cs atoms, the conversion rates, Pr/Cs, were in the order of 0.1%. This value is much smaller than these reported by the Iwamura group.」——看见了同方向的信号，但量级只有声称的 0.1%。两种读法并置：主张方读作「独立实验室确认了方向」，质疑方读作「复现值落在污染/本底区间」。本篇不裁判；登记为「35 年账上最后一笔未决的分歧」。嬗变声称迄今无领域外核物理实验室的独立复现记录[多源检索]。

6.9 正反馈双向：同一批数据的两种读法（本篇的技术胎记）

主张方提交 DOE 2004 评审的综述文件（Hagelstein, McKubre, Nagel, Chubb & Hekman《New Physical Effects in Metal Deuterides》§2.4，NET 镜像 PDF【立场媒体镜像，仅引正文】[一手逐字]）逐字：「A strong sensitivity of excess heat production on the operating temperature was noted and exploited early on by Fleischmann and Pons…The heat was found to increase over the course of a calibration heat pulse, which is attributed to a thermal positive feedback effect. …the results expressed in a form consistent with Pexcess = P0 e−Ea/kBT where Ea is an equivalent activation energy that was found to be about 15 Kcal/mol (about 670 meV). The observed small activation energy is thought to be associated with creating the proper chemical environment and not associated directly with the nuclear process.」

「放热随池温上升」这同一现象：主张方读作可利用的物理特征（正反馈协议、Arrhenius 拟合出 ~670 meV 等效激活能）；质疑方读作量热学的系统性危险——ERAB 1989 的标定警告（「nonlinearities of the calibration in power output vs. temperature」）与 Shanahan 的 CCS 漂移正是针对同一温度依赖性。一个依赖温度正反馈才能看到的信号，在量热学上等价于一个会自我放大的标定误差候选者。两读法指向同一批数据，并置于此，不作裁判——但它是第八章「量热被降级为辅助诊断」这一制度决定的直接技术前史。

6.10 理论侧登记（一句话级，不评判）

Widom–Larsen 2006（EPJ C，DOI 10.1140/epjc/s2006-02479-8[一手逐字]）：超低动量中子催化机制，「No Coulomb barriers exist for the weak interaction neutron production…」；Hagelstein 2025（arXiv:2501.08338[一手逐字]）：固态环境聚变速率增强框架，自称可增强「more than 40 orders of magnitude」。评审方对这类理论的官方回应（DOE 2004，Charge 2 段逐字）：「The reviewers raised serious concerns regarding the assumptions postulated in the proposed theoretical model for the explanation for 4He production.」——理论侧 35 年的任务是填那条 >10¹³ 的缝，至今没有一条机制被评审接受。

七、二审账：DOE 2004 复审——十五年后的同一张票

7.1 程序性质：纸面评审＋一天听证，无新实验

发起方是主张方自己：「The Department of Energy’s (DOE) Office of Science (SC) was approached in late 2003 by a group of scientists who requested that the Department revisit the question of scientific evidence for low energy nuclear reactions.」（DOE 2004 报告，lenr-canr 存档属性；此 URL 同时是 ARPA-E FOA 2022 脚注 23 的官方引用件，provenance 双重锚定[一手逐字]）。程序两截：9 名评审员函评一个月（核物理、材料、电化学背景）＋2004-08-23 一天评审会（另 9 名），「In total, 18 individual reviewer comments were received by DOE.」[一手逐字]。评审对象是提议方自撰综述「New Physical Effects in Metal Deuterides」（Hagelstein、McKubre、Nagel、Chubb、Hekman 列名）。三条 charge 逐字：「(1) To examine and evaluate the experimental and theoretical evidence for the occurrences of nuclear reactions in condensed matter at low energies…(2) To determine whether the evidence is sufficiently conclusive to demonstrate that such nuclear reactions occur. (3) To determine whether there is a scientific case for continued efforts…」[一手逐字]。细节考古：报告原文写「In 1987 Pons and Fleischman first reported…」——年份笔误（实为 1989）、Fleischmann 拼写亦误，引用保留原文[sic]。

7.2 票型逐字

  • Charge 1：「Two-thirds of the reviewers commenting on Charge Element 1 did not feel the evidence was conclusive for low energy nuclear reactions, one found the evidence convincing, and the remainder indicated they were somewhat convinced.」[一手逐字]——18 人里 1 人认为确证。
  • 过量热单项：「Evaluations by the reviewers ranged from: 1) evidence for excess power is compelling, to 2) there is no convincing evidence that excess power is produced when integrated over the life of an experiment. The reviewers were split approximately evenly on this topic.」[一手逐字]——热的问题上评审员对半劈开，这是 2004 与 1989 的唯一可见温差。
  • 但下一行把温差压平：「Most reviewers, including those who accepted the evidence and those who did not, stated that the effects are not repeatable, the magnitude of the effect has not increased in over a decade of work, and that many of the reported experiments were not well documented.」[一手逐字]——包括接受证据的人在内，多数评审员承认：效应不可重复、量级十年未涨。十五年，声称的瓦数没有长进。
  • Charge 2：「The preponderance of the reviewers’ evaluations indicated that Charge Element 2…is not conclusively demonstrated by the evidence presented. One reviewer believed that the occurrence was demonstrated, and several reviewers did not address the question.」[一手逐字]

7.3 推荐与结论：门缝还在，但没人推荐专项

推荐段逐字：「The nearly unanimous opinion of the reviewers was that funding agencies should entertain individual, well-designed proposals for experiments that address specific scientific issues…These proposals should meet accepted scientific standards, and undergo the rigors of peer review. No reviewer recommended a focused federally funded program for low energy nuclear reactions.」[一手逐字]——门缝语言与 1989 年建议 2 同构：常轨同行评审的门开着，专项的门关着。两个被认为值得做的方向逐字：氘化金属性质研究、「the use of state-of-the-art apparatus and techniques to search for fusion events in thin deuterated foils」；并附一句判词级的劝告：「Several reviewers specifically stated that more experiments similar in nature to those that have been carried out for the past fifteen years are unlikely to advance knowledge in this area.」[一手逐字]——再做十五年同款实验不会推进知识。这句话是 2004 评审对 1989–2004 全部方法的否决，也是 2022 年 ARPA-E 改走「薄氘化箔＋核产物诊断」路线的直接伏笔（第八章）。

结论段逐字：「While significant progress has been made in the sophistication of calorimeters since the review of this subject in 1989, the conclusions reached by the reviewers today are similar to those found in the 1989 review.」[一手逐字]——量热计进步了，结论没变。二审没有推翻一审，也没有给一审翻案：它确认的是同一状态——不可断然证伪、亦不可接受。

八、再审程序账：从 2016 报告语言到 Thunderbird——审的宾语是什么

8.1 2016：国会山的一纸报告语言（不是听证）

任务书预设的「2016 国会听证」不实，真实制度事实链三件一手文件：

  1. 众议院军事委员会 mark-up 文本与正式报告 H.Rept. 114-537（第 87 页，govinfo.gov 全文[一手逐字]）逐字：「The committee is aware of recent positive developments in developing low-energy nuclear reactions (LENR), which produce ultra-clean, low-cost renewable energy that have strong national security implications. For example, according to the Defense Intelligence Agency (DIA), if LENR works it will be a ‘disruptive technology that could revolutionize energy production and storage.’ …DIA has also assessed that Japan and Italy are leaders in the field and that Russia, China, Israel, and India are now devoting significant resources to LENR development. …the committee directs the Secretary of Defense to provide a briefing on the military utility of recent U.S. industrial base LENR advancements to the House Armed Services Committee by September 22, 2016.」——指令内容是一次闭门 briefing，不是公开听证。
  2. 最终协商报告 H.Rept. 114-840 全文检索 LENR/cold fusion 零命中[多源检索]——条款未入法。
  3. DoD 的 briefing 以 FOIA 文件形式存在（案号 16-F-1333），原文 PDF 403 未取回[需亲核]——本篇登记为「委员会报告语言＋briefing 实际发生但内容未核验」，不写「国会听证」。arXiv:2308.13533 一文把 briefing 写成「hearing」并宣称「ARPA-E 因此成立」（ARPA-E 2009 年已成立），两处失实，本篇不沿用[文献较稳]。

8.2 DIA 2009：情报口的「如果是真的」

DIA《Technology Forecast: Worldwide Research on Low-Energy Nuclear Reactions Increasing and Gaining Acceptance》（DIA-08-0911-003，2009-11-13，UNCLASSIFIED，Stanford 存档 PDF[一手逐字]）逐字：「DIA assesses with high confidence that if LENR can produce nuclear-origin energy at room temperatures, this disruptive technology could revolutionize energy production and storage, since nuclear reactions release millions of times more energy per unit mass than do any known chemical fuel.」——注意句式：「if LENR can produce」。这是一份条件句评估：高置信度的部分不是「LENR 为真」，是「若为真则影响巨大」＋「多国在投」。「日意领先、俄中印以在投入」的国家评估同上。它后来成为 2016 报告语言的直接来源。

8.3 2021 workshop：「stalemate」一词的官方化

ARPA-E 2021-10-21/22 LENR workshop 文件（议程与三份幻灯落盘）把僵局的机制逐字写明：「Despite a large body of empirical evidence for LENR that has been reported internationally over the past 30+ years…there still does not exist a widely accepted, on-demand, repeatable LENR experiment nor a sound theoretical basis. This has led to a stalemate where adequate funding is not accessible to establish irrefutable evidence and understanding of LENR, and lack of the latter precludes the field from accessing adequate funding.」[一手逐字]——没有钱所以没有铁证，没有铁证所以没有钱。这是对 1989 门缝 33 年无人走的官方诊断。

两阶段路径逐字：「1. Support targeted R&D toward establishing at least one on-demand, repeatable LENR experiment with diagnostic evidence that is convincing to the wider scientific community…; 2. If phase 1 above is successful (metrics to be determined), support a broader range of R&D activities…toward disruptive energy applications…」[一手逐字]。判据先行：先有「一个随叫随到、可重复、诊断说服力强」的实验，才谈得上扩大。

8.4 FOA DE-FOA-0002785：把「怎么判」写成四条技术判据

2022-09-13 新闻稿（ARPA-E 官方[一手逐字]）：「announced up to $10 million in funding to establish clear practices to determine whether low-energy nuclear reactions (LENR) could be the basis for a potentially transformative carbon-free energy source」；代主任 Gerbi 逐字：「we’ve set out with this LENR Exploratory Topic to conclusively answer the question ‘should this field move forward, or does it not show promise?’」

FOA 正文判定规范句逐字（官方 PDF[一手逐字]）：「A goal of this Exploratory Topic is to establish clear practices to rigorously answer the question, ‘should this field move forward given that LENR could be a potentially transformative carbon-free energy source, or does it conclusively not show promise?’.」——再审的宾语在这里白纸黑字：要回答的是「这个领域该不该往前走」，一个判定问题。

四条判据逐字（bullet 分行照排）：

• Conduct experiments that demonstrably satisfy the definition of LENR given in footnote 20

• Achieve statistically significant diagnostic evidence of nuclear reactions above background and relative to control experiments, at a level greater than 99.7% (3σ) statistical confidence level

• Carefully identify and eliminate ‘prosaic’ explanations, e.g., rogue chemical reactions resulting in excess heat, material and/or environmental contaminants, natural radiation background, etc.

• Publish results in a top-tier research journal.

[一手逐字]

三个结构性脚注/条款，每一条都对应一段 35 年的病史：

  • LENR 工作定义（脚注 20）逐字：「We define LENR as a hypothetical energy-producing process (or class of processes) with system energy outputs characteristic of nuclear physics (>>1 keV/amu/reaction) and energy inputs characteristic of chemistry (~eV/atom).」——定义里没有了「fusion」。名号跳到这一步完成制度化：判据只要求「核能级输出、化学能级输入」，不再要求任何聚变通道。
  • 量热降级（Table 1）逐字：「Calorimetry cannot be the only nor primary diagnostic, but it can be part of the diagnostic suite, provided that labor and hardware expenditures associated with calorimetry are ≤10% of the total project cost of a Category-A application」——第六章那场 1% 漂移对 780 mW 的量热战争，最终以制度形式了结：量热被限死在预算 10% 以内、不得作主诊断。
  • 机构性不可知（脚注 27）逐字：「ARPA-E is agnostic at this time regarding the existence of LENR as a physical phenomenon (as defined in footnote 20), the underlying mechanism(s) of LENR, and the specific nuclear process(es) involved, if any…」；双类别结构（A 实验队／B 诊断队，人员不得重叠，B 队逐字职责「bring a neutral, independent perspective that will bolster the credibility of any reported evidence for LENR」）[一手逐字]——把「盲法」与「独立」直接写进组织架构。

以及那句 33 年的门缝账：「DOE reviews in 1989 and 2004 both concluded that the evidence did not support the claim of D-D fusion, but that research proposals on deuterated heavy metals should be evaluated under the standard peer-review process. However, few such proposals were submitted, and none were funded by DOE.」[一手逐字]——门一直开着，33 年几乎没人递申请，递了的也没中。这是「体系压制」叙事最难消化的一行官方逐字。

8.5 八项目 $9,956,549：遴选、结构与结题稀疏

2023-02-17 遴选新闻稿（ARPA-E 官方[一手逐字]）：「announced $10 million in funding for eight projects…from universities, a national laboratory, and small business—aim to break the stalemate of research in this space.」八项目合计 $9,956,549（项目库逐项加总[自算]）：Category A 六项（Amphionic、ETC CATHODE $1,574,359、LBNL $1,500,000、MIT、Stanford、UMich 过量热）、Category B 两项（Texas Tech——Robert Duncan 实验室；UMich 辐射探测）。结构事实：8 项里 7 项的主诊断是核产物探测而非量热——2004 年评审员「薄氘化箔＋state-of-the-art 诊断」的建议路线，18 年后以组织形式落地。

结题状态（截至 2026-08-17，ARPA-E 项目库逐项目 JSON 实测）：

  • 公开产出仅两例。LBNL（Schenkel/Munday 组）：2025-03-16 APS 会议海报（eScholarship PDF；OSTI DOI 10.2172/2569797[一手逐字]）：「Our initial results show a clear increase in fusion rate in Titanium loaded electrochemically compared to Titanium loaded with only a deuterium beam.」、「Differing from Titanium, we see no increase in fusion rate when we electrochemically load Palladium foil compared to beam loading only.」——Ti 靶有增强、Pd 靶无增强；是事件率增强，不是净能量声称。期刊论文未见[多源检索]。MIT（Metzler 组）：SSRN 预印本「Probing Neutrons and Purported Fission Daughter Products from Gas-Loaded, Laser-Irradiated Metal-Hydrogen Targets」（2023，DOI 10.2139/ssrn.4411160[文献较稳]）；ICCF25/26 有会议报告。决定性期刊论文未见。
  • 其余六项（Stanford、Texas Tech、UMich 两项、ETC、Amphionic）截至本篇结题日均无公开结题论文或报告[多源检索]。
  • 状态与期限交错（项目库两字段并列表述，不互推）：Stanford/LBNL/UMich 辐射项标 active，其余五项标 alumni；而 term 字段显示 LBNL 期限 2025-12-01 已过仍 active、ETC CATHODE 期限 2027-06-26 未到已标 alumni——官方无字段语义说明，本篇只并列、不推断。八项实际执行期普遍超出 FOA 的 30 个月上限（延期）[自算]。
  • 社群侧对照：lenr-canr 2023-02-18 短评逐字：「Some cold fusion researchers feel that these eight projects were poorly chosen. The goals are framed as if cold fusion is the same as plasma fusion.」（lenr-canr，领域自档案库属性[一手逐字]）——主张方阵营对「按核产物诊断选型」本身不满。再审程序的双向摩擦在此：体系要的是核产物铁证，阵营要的是量热被重新承认。

8.6 Thunderbird（2025）：+15(2)% 与「无净增益」焊在同一篇里

Berlinguette/Schenkel/Munday 等（UBC 等，资助方 Thistledown Foundation——不是 Google；团队即 2019 Google 项目原班人马），Nature 644: 640, 2025-08-20，DOI 10.1038/s41586-025-09042-7[一手逐字]。摘要逐字：「Here we report a benchtop fusion reactor that enabled us to bombard a palladium metal target with deuterium ions. These deuterium ions undergo deuterium–deuterium fusion reactions within the palladium metal. We showed that the in situ electrochemical loading of deuterium into the palladium target resulted in a 15(2)% increase in deuterium–deuterium fusion rates. This experiment shows how the electrochemical loading of a metal target at the electronvolt energy scale can affect nuclear reactions at the megaelectronvolt energy scale.」；同文逐字：「While the Thunderbird Reactor did not generate a net energy gain, the electrochemical increase in nuclear fusion rates presents an entirely new avenue of exploration toward that ultimate goal.」

机制读法（本篇裁决）：这是真实聚变率调制实验——束流 D-D 聚变是本底，电化学加载把已知的、有核产物可测的聚变率调高了 15(2)%——电子伏特尺度的操作影响兆电子伏特尺度的核反应，这是真的新物理通道（与 Schenkel 的电子屏蔽线一脉相承）。但它同时逐字自报「did not generate a net energy gain」：它测量的恰恰是 FP 声称不承诺的那半边（核产物），而不是 FP 声称承诺的那半边（净热）。UBC 通稿逐字自称交付 2019 年论文呼吁的「reference experiment」：「The present research answers that call with a reference experiment to support the specific claim that electrochemistry can be leveraged to increase nuclear reaction rates.」（UBC 通稿[一手逐字]）——参考实验服务的声称是「电化学可以提高核反应率」，不是「冷聚变产生净能量」。这是复活跳的最新样本：实验是真的，结果被接回的名号是松动的。

8.7 制度时间线补全：处决—重启—并入再审（日、NASA、海军）

  • 日本 NEDO NHE 项目（1993-11～1998-06，约 25 亿日元，三菱重工/日立/东芝/爱信/田中贵金属/NFI 六企业参加），结题结论逐字（Kasagi & Iwamura，ICCF14 2008，lenr-canr PDF，领域自档案库属性[一手逐字]）：「(1) No apparent excess heat was observed over instrumentation limits after about 500 of electrolysis experiments by some kinds of well-developed cells. (2) This means that we failed to verify the existence of excess heat at the level which was expected at the beginning of the project. …(3) Of course, we do not deny the possible existence of solid state nuclear reactions being unknown up to now and its possible trace amount of heat. However the phenomena still remains within a scope of fundamental research.」——500 次电解零过量热＋「不否认可能存在」：与 ERAB 同构的「否证据、留门缝」。后续：东北大学 CMNR 部门 2015-04 成立（东北大＋三菱重工＋Clean Planet 产学协作，JCMNS 19 (2016) 119[一手逐字]）；JCF 年会延续至 2026-02（JCF25 程序公开）。
  • NASA Glenn：官方史（Benyo, Forsley, Steinetz，JCMNS 38 (2024) 13–41，领域期刊属性[一手逐字]）逐字：「GRC has studied LENR or Lattice Confinement Fusion under various names since the 1989 University of Utah」；AEC 项目 2011–2018、HQ 指示 verify、结题论文发表于 Phys. Rev. C 101, 044610 (2020)（摘要含「d-D nuclear fusion events were observed in an electron-screened, deuterated metal lattice…」，NTRS NASA/TP-20205001616 落盘[一手逐字]）；「The AEC Project transitioned to the LCF Project with the current funded effort (July 2021 to present)」。NASA 的存活方式就是名号跳的教科书：同一个实验室，从 LENR 改名 LCF，资助就恢复了。
  • 海军：SPAWAR LENR 研究「was terminated in 2011」（TWZ 2021-04-09[一手逐字]）；2021 年 NSWC Indian Head 牵头五政府实验室（海军、陆军研究实验室、NIST 等）重启，项目经理 Oliver Barham 逐字：「I’m not as worried about looking into something that is considered controversial as long as there’s good science there. …We’re not out to prove or disprove anything, we’re out to assemble a team of scientists who want to take it seriously.」[一手逐字]；DARPA 共同资助 NSWC Indian Head 的 HIVER 项目（NASA 史落盘件记载[转引层]）；ETC 2023 年获 ARPA-E CATHODE 项目，承接的正是 SPAWAR 的共沉积-闪烁体路线。处决（2011）—重启（2021）—并入再审（2023），链条完整：被处决的不是技术路线，是名号；路线换了判据框架后被再审程序吸收。
  • 意大利/印度（登记层）：ENEA Frascati 自 1989 持续投入（ENEA 官方史书 2009，领域自档案库属性）；印度 Current Science 2015-02-25 LENR 专刊（34 篇综述、170 页，Infinite Energy 存档 PDF，主张方刊物属性），主编逐字：「I believe that any scientific field should not be discredited without adequate verification.」——门缝语言的另一国语版本。

九、Google 双扶手账：一篇论文里的两只扶手

9.1 项目本体：stealth 四年、$10M、4 PI + 12 projects + 16 collaborations

Berlinguette 等七作者（UBC／MIT／UMD／LBNL／Google LLC 三人），「Revisiting the cold case of cold fusion」，Nature 570: 45–51（2019-05-27），DOI 10.1038/s41586-019-1256-6；体裁为 Perspective（社论称其为「a Perspective, a technical opinion piece」）；付费墙关闭，作者提供只读共享链接 rdcu.be/bEAsT——本篇正文引语以摘要一手件＋多源转引互证方式给出[一手逐字＋转引层双源互证]。

项目事实（逐字来源齐全）：内部代号 Project Charleston（Berlinguette 组官网产出清单[一手逐字]）；出资额三处逐字互证——Nature 新闻「Google’s US$10-million project aimed to test the cold-fusion claims rigorously」、Trevithick「Google has spent $10 million on the effort since 2015」、项目自述「$10 million invested in external sponsored research」（Nature 新闻[一手逐字]）；结构为 4 PI（Berlinguette/Chiang/Munday/Schenkel）＋「about 12 projects that involved some 30 researchers」（Koningstein 2024 逐字）＋「16 collaborations — 8 new academic groups, 8 experienced LENR researchers/groups; 6 more collaboration attempts were unsuccessful; 12 calorimeter designs were qualified; No lab work was conducted at Google」（Trevithick ICCF-23 幻灯逐字[一手逐字]）。Google 官方从未自行宣布该项目（无官方博客/新闻稿[多源检索]）——全程 stealth，Trevithick 逐字：「This project was carried out in stealth. We didn’t want the fact that Google was funding research in this area to become a distraction. For the first couple of years, we didn’t even tell other members of our group the real reason behind the hydrogen storage experiments going on in the lab!」[一手逐字]。选人标准逐字：「Expertise in disciplines relevant to cold fusion; No prior position on cold fusion (for or against); Willingness to collaborate…; Commitment to publish what is learned.」[一手逐字]——这是 1989 年之后第一次有人用「无预设立场」作为招人条件来重开这个问题。LENR 旗手 McKubre 2014–2015 年参与早期磋商后退出（其自述[一手逐字]），并指团队刻意与 CMNS 社群保持距离。

9.2 三条路径的结果：零、零、和一个真实的反常

论文摘要逐字：「Motivated by the possibility that such judgement might have been premature, we embarked on a multi-institution programme to re-evaluate cold fusion to a high standard of scientific rigour. Here we describe our efforts, which have yet to yield any evidence of such an effect. Nonetheless, a by-product of our investigations has been to provide new insights into highly hydrided metals and low-energy nuclear reactions, and we contend that there remains much interesting science to be done in this underexplored parameter space.」（摘要一手件[一手逐字]）

三路径分项：

  1. Pd 高加载：「Claim: Cathode loading of PdH_x where x ≥ 0.875 is required to produce excess heat. Our experience: Highly hydrided materials are difficult to produce. …We achieved x = 0.96 ± 0.02 once. …Conclusion: ‘…more work is required to produce stable samples of PdH_x where x ≥ 0.875 to comprehensively evaluate these claims.’」（图注经多源转引互证[转引层]）——第六章的临界参数第一件（D/Pd > 0.9），Google 四年只稳定达到过一次。读法分叉：质疑方读作「声称的前提条件如此苛刻以至不可检验」，主张方读作「前人报告的加载区间被独立证实难但可及」。
  2. 脉冲等离子体氚：无氚产生[一手逐字，经 Nature 新闻与论文摘要交叉]。
  3. 量热：「We designed a calorimeter capable of operating at 1,200 °C and 33 atm with less than 2% measurement uncertainty. To detect ≥ 10% excess heat events with 98% confidence, each experiment was run 4 times in parallel in identical calorimeters. For 16 months, we evaluated the effects of temperature, pressure, sample composition, particle size, and surface treatment. Conclusion: ‘…none of the 420 samples we evaluated provided evidence of excess heat.’」（图注经多源转引互证[转引层]）——420 样本、4 平行、16 个月、<2% 不确定度：这是 35 年量热史上设计余量最大的一次阴性。
  4. 唯一的反常正信号：Schenkel 组（arXiv:1905.03400，后刊 J. Appl. Phys. 126, 203302[一手逐字]）：「For palladium targets, we find neutron yields as a function of cathode voltage that are over 100 times higher than yields expected for bare nuclei fusion at ion energies below 2 keV (center of mass frame). A possible explanation is a correction to the ion energy due to an electron screening potential of 1000±250 eV, which increases the probability for tunneling through the repulsive Coulomb barrier.」Nature Materials 社论（「Coming in from the cold」[一手逐字]）定性：「Schenkel’s experiments on plasmas, meanwhile, did produce evidence of nuclear fusion. …The amount of fusion seen so far is minuscule, and nowhere near the level needed to be of practical value in energy generation. All the same, it exceeds theoretical predictions by two orders of magnitude, for reasons not yet understood.」——项目里唯一超出理论预期的信号，长在核产物侧、不长在热量侧。这个分布本身就是对 1989 裂缝的再确认：有核产物的地方没有热，有热声称的地方没有核产物。

9.3 双扶手焊在同一段：结论段逐字

结论段逐字（经 Messinger/Price、Physics World 等多源转引互证[转引层]）：「So far, we have found no evidence of anomalous effects claimed by proponents of cold fusion that cannot otherwise be explained prosaically. However, our work illuminates the difficulties of producing the conditions under which cold fusion is hypothesized to exist. This result leaves open the possibility that the debunking of cold fusion in 1989 was perhaps premature because the relevant physical and material conditions had not (and indeed have not yet) been credibly realized and thoroughly investigated. Should the phenomenon happen to be real (itself an open question), there may be good technical reasons why proponents of cold fusion have struggled to detect anomalous effects reliably and reproducibly. Continued scepticism of cold fusion is justified, but we contend that additional investigation of the relevant conditions is required before the phenomenon can be ruled out entirely.」

这是全案的承重墙：否定派拿第一句（no evidence…prosaically），验证派拿第三、五句（perhaps premature／required before…ruled out entirely）——而论文原文把两句焊在同一段里。与 ERAB 1989 的「on balance＋门缝」结构完全同构，时隔 30 年。CALL to action 逐字：「Fusion stands out as a mechanism with enormous potential to affect how we generate energy. This opportunity has already mobilized a 25 billion dollar international investment to construct ITER. Simultaneous research into alternative forms of fusion, including cold fusion, might present solutions that require shorter timelines or less extensive infrastructure. …Finding breakthroughs requires risk taking, and we contend that revisiting cold fusion is a risk worth taking.」[转引层双源互证]——再审的辩护词不是「现象大概是真的」，是「这个险值得冒」。

9.4 同一事实的两列火车：被引光谱与措辞迁移

  • 社论措辞事故：Nature 社论（d41586-019-01675-9）写「The team found no evidence whatsoever of cold fusion.」——这是社论措辞，不是论文句（论文摘要是「yet to yield any evidence」）；ITER 组织官网评论（「End of story?」[一手逐字]）把社论句误植为 Google 自认：「Google confirmed in May that they had found ‘no evidence whatsoever’ in favour of cold fusion. …Like the quest for the philosophers’ stone, the dream of cold fusion will endure and keep modern-day alchemists busy.」——「no evidence whatsoever」这句咒语 1989 年 Lewis 在 APS 现场说过一遍，2019 年社论再说一遍，30 年两见，本篇登记为叙事化石（第十一章）。
  • 媒体标题光谱（各取一条逐字）：Nature News「Google revives controversial cold-fusion experiments」；MIT Technology Review 结语句「It looks as though cold fusion really is dead this time.」；Skeptical Inquirer 标题「Google Funds Cold Fusion Research; Results Still Negative」——同一组事实，从 revives 到 dead this time 的完整光谱[一手逐字]。
  • 科学侧双向被引：反方 Frank Close（牛津，曾参与早期复现）逐字：「If somebody I was investing my money in started doing this, I would withdraw my money.」（Nature 新闻[一手逐字]）；正方 HERMES（EU Horizon 2020，约 €4M，2020–2024）项目描述逐字：「the main motivation for this work is based on the recent Nature perspective ‘Revisiting the cold case of cold fusion’」[一手逐字]；美国海军 Indian Head 首席科学家 Gotzmer 逐字：「We got our impetus from the Google paper appearing in Nature.」（IEEE Spectrum 2021-03[转引层]）；Wikipedia 条目侧则登记为一次「failed replication attempt」（第十一章）。同一篇论文：重启派拿它立项，怀疑派拿它撤资，维基拿它结案。
  • 主张方阵营的逆耳评价：Trevithick（ICCF-23 幻灯逐字[一手逐字]）：「Our experience is that the LENR community either can not teach or will not teach. Failure to share the best of what is known has impeded scientific progress.」——再审方对主张方社群的这句评价，比任何怀疑派的抨击都重：它出自花了四年和一千万美元试图学会这门手艺的人。
  • McKubre 的双面定性（Infinite Energy 146，主张方刊物属性[一手逐字]）：开篇「God bless Google!…I am delighted with the effort and the results」并列举「four startling positives」，同时批评论文把 D+D→⁴He+24 MeV 当「a straw man easily defeated」、忽视高电流密度×长潜伏期等既有知识——主张方旗舰对再审的官方回应是「感谢＋你审错了声称」。这一句值得放进天平：如果论文审的是稻草人，那么 1989 年 FP 原文那句「inconceivable…anything but nuclear」所锚定的声称，到底哪一个是本声称——这本身就是名号跳在科学内容层的版本（第十三章）。

十、商业极账：一亿美元合同、48 分钟和解、与一台永不交付的机器

商业极与科学极必须分开承重：本章全部事实只说明「商业端口消耗可信度的方式」，不构成对科学侧声称的任何方向的证据。双向标注：Rossi 方材料（JONP、ecat.com、Lewan 博客）与 IH/批评方材料（New Energy Times 等）均带立场，引用处注明归属。

10.1 2011 Bologna：两名教授的背书与它的边界

Levi 组测试报告（原载 JONP——Rossi 关联站点，立场标注；抓取件落盘[一手逐字]）：「In [Test 2] the power measured was 12686 +/- 211 W for about 40 min…」；同文自承：「The main origin of possible errors in [Test 1] measurement was that the steam was not checked to be completely dry.」——输出端的干度未验，是这类量热的第一漏洞。

Essén（瑞典皇家科学院前主席）与 Kullander（瑞典皇家科学院能源委员会主席）差旅报告（2011-03-29 测试，原载 JONP/NyTeknik[一手逐字]）结论句：「Any chemical process for producing 25 kWh from any fuel in a 50 cm3 container can be ruled out. The only alternative explanation is that there is some kind of a nuclear process that gives rise to the measured energy production.」——这是「复活」叙事在商业端口的最高规格背书。同文边界条件逐字：「Since we do not have access to the internal design of the central fuel container and no information on the external lead shielding and the cooling water system we can only make very general comments.」、「no error analysis has been done」。背书与「看不到内部」写在同一份报告里：排除化学过程的论证，是在无法接触燃料容器内部设计的前提下做出的。这个结构（外部测量＋不可接触内部）将贯穿本章。

10.2 Rossi v. Industrial Heat：合同、ERV 与「一场不是测试的测试」

案卷：Rossi v. Darden, 1:16-cv-21199, S.D. Fla.（起诉状 Doc 1、反诉 Doc 30、简易判决动议 Doc 203、案卷条目 Doc 332/333 均落盘[一手逐字]）。

  • 合同逐字：「IH agreed to pay LEONARDO One Hundred Million Five Hundred Thousand Dollars ($100,500,000.00) over three (3) payments」——$1.5M 签约、$10M 24 小时验证测试后、$89M 350 天「Guaranteed Performance Test」后。
  • 实际支付：2012-10 首期 $2M（其中 $1.5M 归 Rossi/Leonardo）＋2013 年 $10M——Rossi 实收 1150 万美元（Doc 30 逐字与 New Energy Times 核算一致[一手逐字]）。
  • ERV（Penon）结论逐字：「the amount of energy produced by the E-Cat Unit was consistently substantially greater than six (6) times the amount of energy consumed… the average energy multiplier… was often greater than sixty (60).」——倍增 60 倍，写在触发 $89M 尾款的验收文件里。
  • IH 反诉逐字：「the test that Leonardo and Rossi were conducting, in conjunction with the supposed ‘customer’ in Miami, was not a real test at all, but a carefully scripted effort to deceive Counter-Plaintiffs…」；「客户」J.M. Products 情节（Doc 203 逐字）：「These representations were unquestionably false: J.M. Products was never affiliated in any way with Johnson Matthey. …Rossi, on behalf of J.M. Products, once asked for a price quote from Johnson Matthey for the purchase of platinum sponge, and then later bought some filters from a Johnson Matthey subsidiary in the United States (to mine them for platinum sponge contained therein).」——假客户的名字仿的是一家真公司，关联的近纪录是买过滤器挖铂。
  • IH 内部评估（Doc 30 ¶80 逐字）：「In mid-2015, Industrial Heat hired Joseph Murray (‘Murray’) to serve as Vice President of Engineering, and empowered him to assemble a team of engineers and scientists… That testing demonstrated quite clearly that the results previously claimed by Leonardo, Rossi, and Penon simply could not be replicated using the E-Cat IP that Leonardo and Rossi had provided to Counter-Plaintiffs.」——注意：这是 Murray 团队的评估，不是 Duncan。「Duncan 受 IH 委托评估 E-Cat」的流传说法在本案全部法庭文件与同期报道中无一手文本[多源检索]；Robert Duncan 的两个角色（60 Minutes 复核 Energetics Technologies、Gates 资助 Texas Tech 项目）与此案无涉，本篇拆开登记。

10.3 开庭 48 分钟：和解的精确形态

案卷条目 Doc 332 逐字：「Jul 5, 2017 … Jury Trial completed on 7/5/2017. Case settled. Total time in court: 48 minutes.」；Doc 333 逐字：全员（含 Cherokee、Darden、J.M. Products、Fabiani、Bass 等十三方）「stipulate to the dismissal with prejudice of all claims, counterclaims and third-party claims…」。程序链：6-28 陪审团开庭→6-29「Mistrial declared after dismissing jurors」→6-30 继续→7-05 开庭 48 分钟和解[一手逐字]。和解实质（Lewan 据保密来源文件报道，亲 Rossi 但具一手文件接触[转引层]）：许可全部返还 Rossi——「’the indispensable condition was to get the license back because I didn’t want to collaborate with IH anymore.’」；被告方「shall return or destroy all documentation, return all reactor vessels including the 1MW E-Cat plant, also promising not to disclose the E-Cat fuel formula to anyone.」；Rossi 当日博客逐字：「The terms of the settlement will remain under NDA for ever…The war is over.」（LENR.NET 2017-07-07[一手逐字]）。一亿美元声称的终审形态：没有裁决、没有公开证据固定、永久 NDA——陪审团没有就「E-Cat 是否工作」作出任何事实认定。这台机器的可信度问题在制度上被永久搁置，而不是被解决。

10.4 资金流水的第三方核算与 2026 年的现状

New Energy Times（Krivit，反 Rossi 阵营，金额均以法庭文件为据[一手逐字]）：「According to a court document, in May 2015, Industrial Heat sold 4 percent of its company for $50 million. …approximately $50 million from the Woodford Funds…predicated upon their claims that Industrial Heat and/or IPH had ‘acquired Rossi’s intellectual property.’」（2019-06-07）——2019 年 Woodford 基金崩盘时，这笔投资被列为原因之一。即：商业极的账外溢到了主流金融市场。

2026 年现状（自我声称层，登记不承重）：ecat.com 新闻列表逐字含「May 2026 Update…March 2026 E-Cat Report…Leonardo Corporation Demonstrates On-board E-Cat Technology Continually Charging an EV Battery During 6-hour Test」（ecat.com，Rossi 官方站[一手逐字]）；对照基线 Wikipedia Energy Catalyzer 逐字：「No independent tests have ever been made, and no peer-reviewed tests of the device have ever been published.」（抓取日 2026-08-17[一手逐字]）。十五年，从 Bologna 到 EV 充电演示，独立测试数仍为零。

10.5 USPTO 双轨：核声称驳回、工程外壳授权

  • 驳回轨判例：In re Swartz, 232 F.3d 862 (Fed. Cir. 2000)（CourtListener 案卷系 Public.Resource.Org 文本落盘[一手逐字]）判词：「Here the PTO provided several references showing that results in the area of cold fusion were irreproducible. Thus the PTO provided substantial evidence that those skilled in the art would ‘reasonably doubt’ the asserted utility and operability of cold fusion.」；且堵死改名绕道：「Mr. Swartz’s attempt to show that his claims are directed to a process other than cold fusion must fail. In his written description and throughout prosecution of his application, Mr. Swartz continually represented his invention as relating to cold fusion.」——框架：不可复现 ⇒ 合理怀疑实用性 ⇒ 举证责任在申请人。
  • 授权轨实例（去核化撰写）：Rossi「Fluid heater」US 9,115,913 B1（2015-08-25）——权利要求不主张核反应；美国海军「System and method for generating particles」US 8,419,919（2013，Pd/D 共沉积）；Brillouin「Heating system and methods」US 12,590,732（2026-03-31 授权，说明书明确写 Q pulses「to excite reactions in the catalytic tubes」）[一手逐字]。
  • 现状量化（Patently-O 2026-05-09，Dennis Crouch，导语公开层[一手逐字]）：「Utility rejections under 35 U.S.C. § 101 have become something of a relic at the USPTO. The Office mailed over 650,000 office action rejections in 2025 – only 294 raised a utility ground.」——实用性驳回已极罕见（2025 年 65 万件 OA 中仅 294 件），但 2026 年样本中仍有一件 cold-fusion-adjacent 申请被此条款驳回。专利制度的实际形态：「cold fusion」四个字仍触发驳回，同一技术写成「流体加热器」即可授权——名号跳在知识产权法里的制度化版本。

10.6 产业延续面：Brillouin 与 Nucleation（登记层）

  • Brillouin Energy：SRI 的「第三方验证」是付费委托测试（Business Wire 2018-03-13 通稿逐字：「Dr. Francis Tanzella was again the principal investigator assigned to SRI’s testing of Brillouin Energy’s LENR systems and conducted all of the third party validation work.」——通稿自称 third-party validation，但合同关系是委托方付费[一手逐字]）；2022-08 ICCF24 自称「proves net-positive power from its HHT boiler」；2026-05-06 新 Executive Chairman 到任（公司通稿[一手逐字]）。
  • Nucleation Capital（2020 年成立的 506(c) 核风投）把 LENR 写进公开投资主题（官网逐字：「fission, fusion, low energy nuclear (LENR), radioisotope thermal generators…」[一手逐字]）——利益相关方标注：该基金本身是 LENR 叙事的受益者，其报告属募资材料；其《Four Year Report》（2025-10）中 LENR 仅以主题词形式出现，无逐项出资明细。登记为「LENR 已进入至少一家专门核风投的 thesis」，不登记为「资本已验证 LENR」。

十一、叙事生产线账：53 天化石、维基死刑与词频化石

11.1 狂欢化石：从「世纪最伟大成就」到封面问句

  • 1989-03-24，Deseret News 头版逐字：「FUSION DISCOVERY AT U. COULD RANK AS CENTURY’S GREATEST ACHIEVEMENT」（Deseret 存档[一手逐字]）。
  • 1989-05-08，Time 与 Newsweek 同周封面。Lewenstein（康奈尔冷聚变档案馆建库者）记述逐字：「Yet that week Time and Newsweek issued their 8 May 1989 magazines. Both chose to feature cold fusion on the cover. Though the headlines included some skepticism (Time’s was ‘Fusion or Illusion…’), the effect was to present cold fusion as a potential energy savior to millions of people around the world.」（Fax to Facts, 1998[一手逐字]；封面副题与 time.com 官方版逐字互证——两个来源对封面题渲染略异，本篇以短版为准[文献较稳]）。
  • 化石读法：头版到封面问句 45 天，到 APS 8/1 投票 40 天，到 NYT「does not exist」定调 41 天——全部狂欢与处决的符号动作在 53 天内完成。之后 35 年发生的所有事（两审、改名、再审、商业极）都活在这 53 天造好的框架里。

11.2 维基机制性死刑：导论句、仲裁委员会与修辞漂移

  • 现行导论逐字：「By late 1989, most scientists considered cold fusion claims dead, and cold fusion subsequently gained a reputation as pathological science.」（Wikipedia: Cold fusion，抓取日 2026-08-17[一手逐字]）
  • 机制层：仲裁委员会「Cold fusion」案（2008-12-15，案卷[一手逐字]）原则条：「A neutral point of view cannot be synthesised merely by presenting a plurality of opposing viewpoints, each derived from a polarised source.」——这条原则从机制上封杀了「两极来源并列」的写法；救济：亲 CF 编辑 Pcarbonn 题禁一年（7-1）。「Cold fusion 2」案（2009-09-13）把整个议题送入自由裁量制裁区（discretionary sanctions）[一手逐字]。这是制度性的死刑执行：不是某一版条目判了死刑，是条目的生产方式被永久改造了。
  • 修辞漂移标本：同一句 Koonin 引语，NYT 1989-05-03 现场稿写「sat in stunned silence for a mement [sic] before bursting into applause」，维基条目版渲染为「which was met with a standing ovation」[一手逐字对照]——「沉默后爆发掌声」在二手加工里变成「起立鼓掌」。二次加工的每一站都在给死刑加一点戏。

11.3 词频化石：词死了，活法换了名字（本篇自算）

本篇自算（OpenAlex API 逐年分组，2026-08-17 抓取；PubMed eutils 逐年复核；原始 TSV 落盘，Google Scholar 通路因验证码取不回、登记替代通路[自算]）：

  • 全文口径（search=”cold fusion”）：1988=15 → 1989=280 → 1990=244 → 缓降至 2017=141 → 2019=219 → 2020=223 → 2023=223。2019/2020/2023 三年已回到 1990 年的九成因——「复活」在文献量上可测，不是纯修辞。
  • 题名口径（title.search=”cold fusion”）：1989=167 → 1990=157 → 单调崩塌至 2018=5 → 2025=29（刚回到 1996 年水平）。
  • 改名口径（title.search=”low-energy nuclear reactions”）：1997 年首次上量 6 篇，2005–2013 维持 11–20 篇/年，2019=5，2025=10。
  • PubMed 口径：1989=23、1990=17，1991 起断崖至个位数，2019=5，2025=1。

三个口径合起来的判决书：「cold fusion」作为题名用词在 1990 年后被处决且从未平反（167→5）；同一研究活动在改名后的措辞里持续存活；2019 年起的全文口径回升由 Google/ARPA-E/Thunderbird 三案驱动。词死了，活法换了名字——这是名号跳的定量化石。

会议化石佐证（多源互证[文献较稳]）：ICCF 自 1990-03 盐湖城首届起 35 年不间断（ICCF-27 已排期 2026 尼亚加拉）；改名三步全部可逐字核——学会 ISCMNS 2004 年成立、文集书名自 ICCF-12（2005）起加冠 CMNS、会议全名自 ICCF-13（2007）起改 CMNS；改名晚于题名崩塌起点十余年：先词死，后换牌。共同体的自我叙述细节：1995 年 Monaco 届主席是 Pons 本人——放逐者在自家会议上仍是主席。

11.4 60 Minutes 2009：一次「复活」的完整生命周期与一次传播事故

CBS 60 Minutes「Cold Fusion Is Hot Again」（2009-04-19 播出，CBS 存档，五页转写全部落盘[一手逐字]）：

  • Duncan（密苏里大学副校长，受栏目委托复核 Energetics Technologies 的 Pd-D 工作——不是 Rossi）结论句逐字：「I found that the work done was carefully done, and that the excess heat, as I see it now, is quite real.」
  • 传播事故逐字记录（Skeptical Inquirer 2009-09/10 栏评[一手逐字]）：原版称 Duncan 由 APS 推荐；「An APS statement issued Wednesday says this is totally false, and the APS does not endorse the cold fusion claims on 60 Minutes. …CBS took it off. …The part where CBS says the APS picked Rob Duncan to look into the ET SuperWave is gone. CBS can change history?」——现行 CBS 转写稿中确无「APS 推荐」字样（只写「60 Minutes turned to an independent scientist」）。「复活」的第一块招牌是改出来的：独立专家的制度背书被撤除后，节目悄悄重剪。
  • 反方锚点 Garwin 的判据逐字：「I require that you be able to make one of these things, replicate it, put it here. It heats up the cup of tea. I’ll drink the tea. Then you make me another cup of tea. And I’ll drink that too. That’s not it.」——烧茶判据：100% 随叫随到。与 EPRI 自报的 1/5 天花板正面对撞：判据的分歧从来不是「有没有信号」，是「什么算数」。
  • Fleischmann 本人的墓志铭级引语（在英国乡间受访，逐字）：回顾 20 年，「A wasted opportunity.」两个遗憾：用了「fusion」这个名字（「a name coined by a competitor」）、开了那场发布会（「something he says the University of Utah wanted」）——连创始人都把病因诊断为名号与程序，而不是数据。

11.5 复活谱系 2022–2025：行业报用词迁移（逐字化石）

ANS Nuclear Newswire 三标题，三年三步（均导语层可得[一手逐字]）：

  1. 2022-09-22：No cold feet: ARPA-E wants to explore low-energy nuclear reactions——避讳词 LENR 为主，导语句「historically known as ‘cold fusion’」。
  2. 2023-02-23：ARPA-E picks eight teams to prove—or debunk—low-energy nuclear reactions——把「debunk」写进任务书，并加注「historically and sometimes disparagingly known as ‘cold fusion’」。
  3. 2025-08-25：University adds electrochemical boost in pursuit of cold fusion——标题重新直用「cold fusion」；导语逐字承认「more energy input than gained」。

四年内行业报从避讳词回到直用词。外加 Popular Science 2023-07-03「Cold fusion is making a scientific comeback」（副题「A US agency is funding low-energy nuclear reactions to the tune of $10 million.」[一手逐字]）。用词迁移本身就是化石：名号完成修复的时间点，先于任何一项被判据认可的实验结果。主流大刊对照登记：Nature/Science/NYT 对 ARPA-E 案未见专门条（本篇阴性检索登记为「未见」，非「不存在」[多源检索]）。

11.6 中文消费三层（登记话术形态，不评价报道）

  1. 科技日报 2011-11-10（科学网转载，Rossi 案）：「业已问世的几项所谓’冷聚变试验’，均被证实乃骗局一场。」[一手逐字]——比 ERAB 原文烈度高两档：「on balance 的不信」在中文官方科普口径里直接写成「骗局既成事实」。
  2. 科技日报 2019-05-28（中国日报网转载，Google 案）：「虽然研究并未发现冷聚变的证据，但研究结果被认为对诸多科学和技术领域具有借鉴意义。」「对于当时大部分科学家来说，冷聚变研究俨然成了一种’禁忌’…」[一手逐字]——「禁忌＋借鉴意义」框架。
  3. 财联社 2025-08-21 快讯（Thunderbird 案）：「名为’雷鸟反应堆’的装置以一种突破性的电化学方法，成功提升了核聚变速率。」——全讯不出现「冷聚变」三字[一手逐字]；同日微信公众号「可控核聚变」（2025-08-24）反而主动接回 1989 谱系：「1989年，弗莱施曼（Fleischmann）和庞斯（Pons）声称在氘氧化电解过程中，钯阴极产生了异常热量…为本研究奠定了探索基础。」[一手逐字]——官方口径逐级软化（骗局→禁忌→不提），行业自媒体最先接回谱系。名号跳在中文语境里以「绝口不提」的形态运行。

11.7 社会学重建：Simon 的枢轴句

Bart Simon《Undead Science: Science Studies and the Afterlife of Cold Fusion》（Rutgers UP, 2002）p.124（经 Kooistra 2003 书评带页码转引，亲 LENR 立场，[转引层]；原书未取回，本篇承重使用已降级）：「The cold fusion controversy did not end because Fleischmann and Pons were wrong; it ended because CF researchers found themselves lacking social and material resources to argue that they were right.」——社会学版本把死刑完全归因于资源与地位，不承认「证明为错」这个前提。本篇的裁决与它对齐一半：执行的死刑确实超出判决书的授权（这一半 Simon 是对的）；但裂缝与不可复现性不是资源问题造成的（这一半 Simon 的框架消化不了）。书名三连化石登记：Fiasco（Huizenga 1992）→ Bad Science（Taubes 1993）→ Undead（Simon 2002）——三本书名就是叙事史的三个地层。

十二、反向红跳：四句虚无读法称重（每句收支持/反驳双方原始件）

句①「1989 就是彻头彻尾的骗局/欺诈」——不立，但诱人之处有出处

支持侧：Koonin 现场句「the incompetence and delusion of Pons and Fleischmann」＋全场掌声（NYT 1989-05-03[一手逐字]）；Huizenga 书名「fiasco of the century」；Park 书名「…from foolishness to fraud」；科技日报 2011「均被证实乃骗局一场」[一手逐字]。

反驳侧：处决方自己的官方报告没有下欺诈结论——ERAB 逐字「not possible at this time to state categorically that all the claims…convincingly either proved or disproved」[一手逐字]；Morrison 在同一场合把性质定为「it’s just human nature」[一手逐字]；唯一一次接近欺诈立案的指控（Taubes 报道 Bockris 组氚数据疑被掺假），A&M 复核组结案词留活口：「couldn’t rule out spiking, contamination and measurements problems were more likely explanations」（Wikipedia 条目转述[转引层]）。裁决：「骗局」定性是现场修辞与书名生产的，不是评审结论。诱人之处：程序跳（发布会先于论文、拒绝答疑）是真的，它给「骗局」读法提供了全部的情感燃料——但情感燃料不是法律结论。

句②「35 年零可复现证据」——判据问题，不是事实问题

支持侧：Nature 社论 2019「no evidence whatsoever」（社论措辞[一手逐字]）；Lewis 1989 APS 现场同咒语[一手逐字]；Garwin 烧茶判据[一手逐字]；Google 420 零[转引层]；Morrey 双盲负[文献较稳]。

反驳侧：DARPA 内部备忘录（60 Minutes 转述）「no doubt that anomalous excess heat is produced in these experiments」[转引层]；Duncan「the excess heat, as I see it now, is quite real」[一手逐字]；ARPA-E FOA 承认存在「preliminary evidence of LENR that are possibly consistent with past observations but that do not yet meet the program metrics」（FOA PDF[一手逐字]）。裁决：「零证据」成立与否取决于验收线画在哪——Garwin 的线（100% 复现＋烧茶）下成立；ARPA-E 的线（初步证据存在但未达标）下不成立。双方引用的是同一堆数据，分歧在判据。这句话的真实内容是「35 年没有任何信号通过任何一条被双方共同接受的验收线」——这比「零证据」窄，也比「有证据」诚实。

句③「冷聚变已被最终证伪」——公共记忆的压缩失真

支持侧：APS 现场 8/1 投票[一手逐字]；Wikipedia 导论「By late 1989, most scientists considered cold fusion claims dead」[一手逐字]；ITER 官网「End of story?」＋哲学家之石句[一手逐字]。

反驳侧：ERAB 明说做不到断然证伪[一手逐字]；Nature 2019 论文结论段「perhaps premature…required before the phenomenon can be ruled out entirely」[转引层]；ARPA-E 立项目的本身——「需要立项去 conclusively answer」就是「尚未 conclusively answered」的机构自认[一手逐字]。裁决：「最终证伪」从来不是 1989 年官方文件的原话，它是媒体投票＋书名＋维基导论层层压缩出来的公共记忆。科学上这个声称处于 ERAB 前言精确描述的状态：结构上不可证伪（任何阴性都可被「因未知原因没工作」打折），经验上不可复现。这两个性质互为镜像，35 年没变。

句④「ARPA-E 立项证明冷聚变是真的／平反了」——不立，但制度松动是真的

支持侧：Popular Science 2023-07-03「the announcement marked vindication for their specialty」＋Nagel「the evidence has piled up」[一手逐字]；ARPA-E 入选团队成员 Messinger「for the first time in 30 years, the DOE is funding cold fusion research…reject the palpable stigma」[一手逐字]；lenr-canr 新闻帖直接把 LENR 改写回「cold fusion projects」[一手逐字]。

反驳侧：ARPA-E 局长 Wang 逐字：「does this area show promise, and if so, how? Or can we conclusively show that it does not?」[一手逐字]；ANS 标题把「debunk」写进任务书[一手逐字]；FOA 的资助逻辑是 high-risk, high-reward，不是背书[一手逐字]。裁决：ARPA-E 立项证明的是「问题值得被机构重新打开」，不是「声称属实」——把程序事件偷换成结论事件就是复活跳的定义。但反向必须写足：一个被 DOE 冻结 33 年的领域重新进入联邦资助序列，是制度性松动的事实，不是纯修辞。

十三、母裁决与遗留：换尺家族的极端变体

13.1 母裁决

把十三章的账并成一段：

判决书没有判死刑——ERAB 五条结论全部是否定性证据评估，前言明说无法断然证伪，建议 2 留门缝，「fraud」零命中。但声称的内伤是原文自带的——裂缝 7–10 个数量级、分支比需改 >10⁶ 或增强 >10¹³、勘误移动 γ 峰并缩放图轴 25 倍，全部印在 FP 论文与 ERAB 报告自己的数字里；程序跳（发布会先于投稿、协议被背刺、拒绝答疑到「靠照片里的一只手估算电池尺寸」）是声称方自己造的。死刑的执行超出了判决书的授权——53 天走完、8/1 投票、pariah 化、维基机制性死刑；门缝 33 年无人走是制度事实，「few such proposals were submitted, and none were funded by DOE」是官方自认——体系压制叙事必须消化这一行。35 年声称未达可复现标准——主张方自报五分之一天花板、两审官方票型、「not repeatable, the magnitude of the effect has not increased in over a decade」、唯一独立双盲负向、420 样本零。再审真实发生但宾语是判定规范——Google 的交付物是 reference experiment 的呼吁，ARPA-E 的立项书把「证明它不成立」列为可接受产出，Thunderbird 交付的参考实验服务的声称是「电化学调制核反应率」——没有一件的宾语是「净热为真」。商业极独立消耗可信度——一亿美元合同终结于 48 分钟和解与永久 NDA，十五年独立测试数为零。反向侧同样不能碾平——Schenkel 屏蔽 1000±250 eV 是真实反常，bursts 口子从未被关闭，Iwamura 0.1% 分歧未裁决，「骗局」与「最终证伪」两句都不是官方文件的原话。

13.2 胎记复述与换尺家族落位

死刑跳（执行被读成判决）× 名号跳（每次改名换一把判据：聚变要核产物→LENR 只要核能级输出→LCF 连「聚变」都退出题名）× 复活跳（判定规范的确立被读成现象的平反）。与本库换尺家族三连对表：g-2 篇换的是理论预言尺、惰性中微子篇换的是通量分母尺、G 篇换的是汇编尺——三篇的尺都在动，但被测对象的名字和判据不动；本篇是极端变体：尺没换，是声称本身连名字带判据一起换——每次改名都把上一把尺下失败的声称，翻译成新一把尺下「尚未检验」的声称。这就是为什么这个案子 35 年不能结案：它不是在一把尺下未决，它是在尺的更换中持续逃避结算。

13.3 遗留（活口与观察点）

  • 活口一：Schenkel 电子屏蔽线（1000±250 eV、超裸核预期百倍）→ Thunderbird +15(2)% 已落地为可复现的率调制；下一步看净能量方向的任何推进——目前逐字为零。
  • 活口二：ARPA-E 八项目的结题论文——LBNL 海报的 Ti/Pd 不对称（Ti 增强、Pd 无增强）是判据时代的第一批硬数据；Stanford/Texas Tech/UMich/ETC 四队无公开结题件，项目库状态/期限交错待官方说明。
  • 活口三：Iwamura 嬗变的 0.1% 分歧——领域外独立复现缺席；日本 CMNR 产学协作（东北大＋三菱＋Clean Planet）与 JCF25（2026-02）仍在运行。
  • 活口四：DoD 2016 briefing 原文（FOIA 16-F-1333）未取回——国会山链条缺最后一环。
  • 观察点：名号是否继续迁移（LCF/固态聚变）；财联社式「不提冷聚变」与 ANS 式「直用冷聚变」两条中文/英文消费线的下一次分岔。

不确定点（诚实清单）

  1. Huizenga/Taubes/Mallove 三本原书未取回——全部经 JCF17 论文集带页码转引[转引层]；Simon《Undead Science》p.124 经 Kooistra 书评转引且书评者立场亲 LENR，承重使用已降级；建议图书馆渠道亲核后可将相关段落升级。
  2. Google 论文正文付费墙——结论段与图注经至少两个独立二手来源交叉印证（Messinger/Price 史学论文、Physics World、Nature 新闻互证）；作者只读链接 rdcu.be/bEAsT 可达但内容未逐字落盘比对。
  3. DoD 2016-09-22 briefing 原文（FOIA 16-F-1333）403 未取回——本篇按「委员会报告语言＋briefing 实际发生但内容未核验」表述。
  4. DOE 2004 十八份匿名评审意见为 24MB 纯扫描件无文字层——本篇只用官方总结报告的统计句；个体评审员原话需 OCR 后补。
  5. Miles 三次实验序列（18/21、12/12、30/33）的原始记录散见 China Lake 内部报告与 ICCF 论文——本篇采用 Miles 自述综述口径并标注，未逐篇亲核原始记录。
  6. Rossi 案和解对价无公开证据（永久 NDA）；6-29 mistrial 的原因与恢复细节案卷仅一行；「Duncan 受 IH 委托评估」无一手证据，本篇不采用该说法。
  7. ICCF 届数口径存在两种起算点（日文维基以 1989 Santa Fe 讲习会为 ICCF-1；Montclair 书目与 lenr-canr 以 1990-03 盐湖城为 ICCF1）——本篇从后者；lenr-canr 续页 ICCF-18 年份（2012）与日文维基表（2013 Columbia）冲突，本篇取 2013。
  8. JEC 收稿日期（扉页「Received 13 March 1989; in revised form 22 March 1989」）与当年报道（Science 称修回稿 3-26 才到、Technology Review 称 3-30）冲突，本篇两说并录。
  9. lenr-canr 为领域自办档案库（立场明确）——本篇引用其托管文件时，凡同件有官方 provenance（如 DOE 2004 报告被 ARPA-E FOA 脚注 23 引用）者按官方件处理，其余均标注档案库属性。
  10. Google Scholar 逐年计数取不回（验证码）；词频化石以 OpenAlex/PubMed 为通路，口径已注明。
  11. 「Nature/Science/NYT 未见 ARPA-E 专门条」为方法学阴性检索结论——登记「未见」，不主张「不存在」。

来源清单

全部文件 2026-08-17 抓取并落盘于 /root/chat/tmp/cold-fusion/raw/（745 件）；六捆工作笔记在 /root/chat/tmp/cold-fusion/notes/bundle-A.md 至 bundle-F.md。标注：【领域自档案库】＝lenr-canr.org（Jed Rothwell 维护，立场明确）；【主张方刊物】＝Infinite Energy 等；【立场媒体】＝New Energy Times（反 Rossi）／Lewan 博客（亲 Rossi）等，均双向对冲使用。

1989 现场与判决书（第四、五章）

技术账（第六章）

制度与再审（第七、八章）

Google 案（第九章）

商业极（第十章）

叙事层（第十一章）

  • CBS 60 Minutes 2009-04-19：cbsnews.com 存档；Skeptical Inquirer 2009-09/10 栏评（CBS-APS 事故记录）
  • Wikipedia：Cold fusion／仲裁委员会 Cold fusion 案
  • 词频化石原始数据（本篇自算）：OpenAlex API 分组 TSV（F_openalex_search_cf_20260817.tsv／F_openalex_title_cf_20260817.tsv／F_openalex_title_lenr_20260817.tsv）、PubMed eutils TSV（F_pubmed_cf_tab_20260817.tsv）
  • ANS Nuclear Newswire：2022-09-22／2023-02-23／2025-08-25；Popular Science 2023-07-03
  • Simon《Undead Science》（Rutgers UP, 2002）转引件：Kooistra 2003 书评（Analog）[转引层]
  • 中文样本：科技日报 2011-11-10（科学网转载）、科技日报 2019-05-28（中国日报网转载）、财联社 2025-08-21 快讯（163.com 转载落盘）、微信公众号「可控核聚变」2025-08-24（镜像落盘）

本篇为机制裁决第 153 篇 · 对称双向第 148 篇 · section B1 · 全库第 211 篇。写作纪律：不凭记忆写；全部承重引语取自 2026-08-17 实际抓取落盘文件；逐字引语与转引层分行标注；阵营属性逐处注明。原始文件 745 件落盘于 /root/chat/tmp/cold-fusion/raw/，六捆笔记与取不回清单在 /root/chat/tmp/cold-fusion/notes/。

标签: