Evidence level: C
一句话结论:上一轮把 δ 表偏差归因为「整列偏移(H-col 假设),σ 列与 δ 列不同源」——这个机制判定不成立。逐行反推后,9 行呈现三种不同偏差(3.16× / 10× / 0.32× / 1.02×),没有单一常数可解释。真实机制是「δ 数值与 σ/μr 标签各自取区间的不同端点」,且金属行取自另一套厂商数据。方向性后果:黏土/页岩行的 δ 被放大 10 倍,把本该是 48 dB 强屏蔽的工况写成了 4.8 dB 弱衰减 → 「山体对 ELF 几乎透明」这一稳健结论缺少适用边界。
1. 本轮的起点与上一轮的结论
上一轮(第 37 轮 / #37)完成了 δ/A 表全量扫描,得出:
- 838 文件扫完,14 文件含断言,86 断言 / 27 OK / 59 MISMATCH;
- → 提出 H-col 假设:「σ 列与 δ 列不由同一套参数算出,δ 列先算、σ 列事后独立填,两列从未交叉核对」,并称比值「高度聚集在 10.0 与 ~100 两点」,海水行恰好 1.0。
本轮(#38)的任务是参数层溯源:不去问「δ 对不对」,而是问「σ/μr 这些标签本身从哪来、对不对得上」。
结论先放:扫描范围是对的,机制解释是错的。 上一轮只看了 6 行(干燥/潮湿/混凝土/钢筋/黏土/海水),恰好这 6 行都落在「δ 用了更小的 σ」,于是得出”单一偏移”的印象;补上 3 个金属行后,方向立刻分裂。
2. 引文层核验:三个 DOI 真实,但表内引用有一处归属错
2.1 三个 DOI 全部通过 Crossref 核验
| 锚点 |
DOI |
Crossref 返回的真实身份 |
判定 |
| A1 |
10.1190/1.9781560802631.ch3 |
Palacky G.J., “3. Resistivity Characteristics of Geologic Targets”, Electromagnetic Methods in Applied Geophysics, SEG, 1988, pp. 52–129 |
✅ 真实 |
| A2 |
10.1029/gm013p0469 |
Madden T.R. & Swift C.M., “Magnetotelluric Studies of the Electrical Conductivity Structure of the Crust and Upper Mantle”, Geophysical Monograph Series, AGU, 2013 |
✅ 真实 |
| A3 |
10.1046/j.1365-246x.1999.00707.x |
Eisel M. & Haak V., “Macro-anisotropy of the electrical conductivity of the crust: a magnetotelluric study of the German Continental Deep Drilling site (KTB)”, Geophys J Int, 1999 |
✅ 真实 |
2.2 但表内的文字引用与这三个锚点对不上
skin-depth-reference-table.md §二 的脚注原文是:
区间参考:Palacky & Kligerman (1980) 矿集区电导率;Schön (2015) …
Crossref 按作者 Palacky 检索 1975–1995 全部 15 条结果,没有任何一条与 Kligerman 合著,也没有 1980 年的条目;Palacky 在电导率领域的经典综述是 1988 年独著的 SEG 章节(即 A1)。也就是说:
- 可核的是 Palacky 1988 独著章节;
- 表里写的是「Palacky & Kligerman 1980」——年份、合著者两处均无法核验。
这不是「PMID 假号」,而是第 4 类引文失效:可核文献被写成了不可核的年份/合著组合。性质比假号轻(指向的领域是对的),但后果相同:读者顺着这条脚注查不到源。
2.3 Schön (2015) 的载体可核,但缺章级定位
Crossref 查到 Schön 2015 Physical Properties of Rocks 系列章节,其中 10.1016/b978-0-08-100404-3.00008-1 即 “Electrical Properties” 章——这才是 σ 区间的真正出处,而表里只写书名、未给章。建议补为章级引用。
2.4 六个 PMID 全部真实(与前三轮结论一致)
| PMID |
真实身份 |
在本文的合理用途 |
| 20000094 |
Chernykh AM, Gig Sanit 2009 — 住宅地磁场屏蔽 |
建筑屏蔽实测 |
| 28654641 |
Binhi VN, PLoS One 2017 — 低磁场生物效应综述 |
μ-metal 屏蔽背景 |
| 38132339 |
Sarimov RM, Biology 2023 — 低磁条件生物作用综述 |
同上 |
| 36676419 |
Cheng Y, Materials 2023 — 低剩磁材料测试方法 |
μ-metal 材料数据 |
| 24131394 |
Sirav B, Electromagn Biol Med 2014 — 变压器 ELF 磁场 |
环境 ELF 暴露 |
| 36681118 |
Martel J, Biomed J 2023 — EMF 与昼夜节律 |
环境 ELF 生物效应 |
注:上一轮记录 24131394 标注为"Burchard"、36681118 标注为"Girgert"、28654641 标注为"Lewczuk" —— 三处作者错配在本轮复查中仍然存在于 elf-skin-depth-table-numeric-integrity-audit.md 的正文表述里,属待修项(不阻塞本轮)。
3. 参数层复算:偏差不是一个常数
3.1 方法
取 skin-depth-reference-table.md §三 的每一行,用该行自己标注的 σ 与 μr 代入 δ=√(2/(ωμ₀μᵣσ))(f=7.83 Hz),与表列 δ 比较;再用表列 δ 反推隐含的 σ·μr 乘积,与标签的 σ·μr 相比。
3.2 结果(9 行,f = 7.83 Hz)
| 介质 |
标签 σ |
标签 μᵣ |
表列 δ |
按标签算 δ |
δ比(表/算) |
隐含σμᵣ ÷ 标签σμᵣ |
判定 |
| 干燥花岗岩 |
10⁻⁴ |
1 |
57 km |
18.0 km |
3.17 |
0.0996 |
δ 用了 σ≈10⁻⁵ |
| 潮湿花岗岩 |
10⁻³ |
1 |
18 km |
5.69 km |
3.16 |
0.0999 |
δ 用了 σ≈10⁻⁴ |
| 普通混凝土 |
10⁻² |
1 |
5.7 km |
1.80 km |
3.17 |
0.0996 |
δ 用了 σ≈10⁻³ |
| 钢筋混凝土 |
10⁻¹ |
5 |
2.5 km |
254 m |
9.83 |
0.0104 |
δ 用了 σμᵣ≈10⁻³ |
| 黏土/页岩 |
10⁰ |
1 |
1.8 km |
180 m |
10.01 |
0.0100 |
δ 用了 σ≈10⁻² |
| 海水 |
4 |
1 |
90 m |
89.9 m |
1.00 |
0.999 |
✅ 自洽 |
| 低碳钢 |
10⁶ |
100 |
5.7 mm |
18.0 mm |
0.32 |
9.96 |
⚠️ 方向相反(δ 偏小) |
| 纯铁 |
10⁷ |
2000 |
1.3 mm |
1.27 mm |
1.02 |
0.957 |
✅ 自洽 |
| μ-metal |
10⁶ |
5×10⁴ |
0.25 mm |
0.80 mm |
0.31 |
10.35 |
⚠️ 方向相反(δ 偏小) |
3.3 三种偏差并存 → H-col「单一整列偏移」被否
- 5 行(岩石/建材):δ 比 ≈ 3.16 或 10,σ 标签比实际用的大 10–100 倍;
- 2 行(低碳钢、μ-metal):δ 比 ≈ 0.31,方向相反,隐含 σμᵣ 比标签大 10 倍;
- 2 行(海水、纯铁):完全自洽。
若真的是「整列错位」(一列整体填错位置),9 行应给出同一个倍数。实际给出 3.17 / 3.16 / 3.17 / 9.83 / 10.01 / 1.00 / 0.32 / 1.02 / 0.31 九个不同值,跨三个数量级区段、且符号不一致。
→ 上一轮的 H-col 假设(#41)在本轮被证伪。可结案,不必再追中间稿。
3.4 更准确的机制:端点错配(endpoint mismatch)+ 双数据源混用
把 5 行岩石/建材的实际用值与 §二 的区间对照,规律立刻出来:
| 介质 |
§二 声明区间 |
表列 δ 实际对应的 σ |
该值在区间中的位置 |
| 干燥花岗岩 |
10⁻⁶ – 10⁻⁴ |
≈10⁻⁵ |
对数中点 |
| 潮湿花岗岩 |
10⁻⁴ – 10⁻² |
≈10⁻⁴ |
下限 |
| 普通混凝土(含钢筋) |
10⁻³ – 10⁻¹ |
≈10⁻³ |
下限 |
| 黏土/页岩 |
10⁻² – 10⁰ |
≈10⁻² |
下限 |
| 钢筋混凝土 |
10⁻³ – 10⁻¹, μᵣ1–10 |
σμᵣ≈10⁻³ |
下限(σ 取下限 × μᵣ=1) |
→ δ 数值一律按区间「下限/中点」算,而 §三 的 σ 标签列写的却是「区间上端或中点」。 两列取的是不同端点,所以比值不是固定 10×,而是取决于该行区间跨几个数量级(跨 2 个数量级 → √100 = 10×;跨 1 个 → √10 ≈ 3.16×)。这解释了为何 5 行分散在 3.16 与 10 两档。
金属行的方向相反,则是另一套来源:低碳钢标签 μᵣ=100、μ-metal 标签 μᵣ=5×10⁴ 均取区间上端,而 δ 值对应 μᵣ≈10 与 μᵣ≈5.2×10⁵ —— 前者低于文献区间下限(50–100 之外),后者高于上限(10⁴–10⁵ 之外)。这说明金属行 δ 不是由表内 μᵣ 算出的,而是从厂商/材料手册的另一套数抄来的。
三条独立证据合起来:这不是”一列填错位置”,而是「同一张表里三套数据源拼接,且未统一端点口径」。
4. 表内自洽性检验:A(dB) 表与 δ 表是一致的
把 §四 的 A(dB) 用同一张表的 δ 复算:
| 场景 |
介质 |
z |
表列 δ |
表列 A(dB) |
按表列 δ 复算 |
判定 |
| 千米大山 |
花岗岩 |
1000 m |
57 km |
0.15 dB |
0.152 |
✅ |
| 百米山包 |
潮湿花岗岩 |
100 m |
18 km |
0.05 dB |
0.048 |
✅ |
| 千米山体 |
黏土页岩 |
1000 m |
1.8 km |
4.8 dB |
4.83 |
✅ |
| 20 层楼 |
钢筋混凝土 |
60 m |
2.5 km |
0.21 dB |
0.208 |
✅ |
| 1 cm 钢板 |
低碳钢 |
0.01 m |
5.7 mm |
15.2 dB |
15.2 |
✅ |
| 1 mm μ-metal |
μ-metal |
0.001 m |
0.25 mm |
34.7 dB |
34.7 |
✅ |
→ A(dB) 表完全忠实于 δ 表。所以错误不在换算,只在 §三 的 δ ↔ σ/μr 配对。这个定位很重要:修一处(δ 列或 σ 列),A 表自动跟着对;不需要逐格改 A 值。
5. ⭐ 方向性后果:黏土/页岩行把「强屏蔽」写成了「弱衰减」
这是本轮唯一会改变实质结论的发现。
5.1 两种算法的差距
| 介质 |
表列 δ@7.83 |
表列 A(1 km) |
按标签 σ 的 δ |
按标签 σ 的 A(1 km) |
后果 |
| 干燥花岗岩 |
57 km |
0.15 dB |
18.0 km |
0.48 dB |
同向,量级不变 |
| 潮湿花岗岩 |
18 km |
0.27 dB(估) |
5.69 km |
1.53 dB |
同向,量级不变 |
| 钢筋混凝土 |
2.5 km |
0.21 dB(60 m) |
254 m |
2.05 dB(60 m) |
同向,仍弱 |
| 黏土/页岩 |
1.8 km |
4.8 dB |
180 m |
48.3 dB |
⛔ 方向性反转 |
5.2 48 dB 意味着什么
δ=180 m 时,1 km 厚度 = 5.56 个趋肤深度,振幅衰减到 e⁻⁵·⁵⁶ ≈ 0.38%。这不是「衰减 37%」,这是衰减 99.6%,等效约 48 dB,已是工程意义上的强屏蔽。
而表 §四 把这一行写成「千米大山 / 黏土页岩 / 4.8 dB / 衰减 37%」,§五 又据此在「靠山=屏蔽」审查表里给出「山体 ≈ 0.05–0.15 dB,基本无屏蔽作用」的总结。
5.3 稳健结论必须加边界
当前表述(表 §五、下游多处引用):「靠山屏蔽地磁 ELF 在 1–30 Hz 段没有物理基础」。
应修正为:
「靠山屏蔽 ELF」的否定性结论,只对低电导结晶岩(σ ≤ 10⁻³ S/m,如干燥/一般花岗岩、玄武岩)成立。对高含水黏土、页岩、湿沉积层(σ ≈ 10⁻²–10⁰ S/m),百米级即可产生 dB 级衰减、千米级可达数十 dB——这条路径不能一概否定,反而值得作为”地形-岩性”分层假说单独立题。
5.4 对下游的传导
- #22 A 档门槛(σ≥10⁻² → 10 dB):用表列 δ 会得到「10⁻² 只衰减 4.83 dB,不达门槛」;按标签 σ 算是 48 dB,远超门槛。→ 门槛判定方向相反,A 档样本筛选逻辑需重定。
- #23 铁律(A(dB) < 本底−10 dB 则禁入 E1):铁律本身没问题(它不依赖 δ 数值),但输入 δ 若取自该表,会把高 σ 岩体误判为”透明”,从而误放行。
- #16 表的”千米山体几乎透明”:结论对花岗岩成立,对黏土质山体不成立。
- #18 T4 / #24 阈值:不受影响(不依赖 δ 绝对值)。
6. 更正建议(待授权稿,本轮不改正文)
6.1 引文层(可直接改,属纠错)
| 位置 |
现文 |
建议 |
| §二 脚注 |
Palacky & Kligerman (1980) |
Palacky (1988), DOI 10.1190/1.9781560802631.ch3(§3 Resistivity Characteristics of Geologic Targets, pp. 52–129) |
| §二 脚注 |
Schön (2015) Physical Properties of Rocks |
补章级:“Electrical Properties”, DOI 10.1016/b978-0-08-100404-3.00008-1 |
| §一/§七 |
未提 Madden / Eisel |
建议补 10.1029/gm013p0469(地壳上地幔 σ 结构)、10.1046/j.1365-246x.1999.00707.x(KTB 宏各向异性),使 σ 溯源可查三层 |
6.2 数值层(需 JOE 定 canonical,属 #40)
两条互斥路线,不要混用:
- 路线 A(保标签,重算 δ):σ/μr 保持现表(它们是文献区间取值,可与 §二 对齐),整列重算 δ 与 A。优点:物理依据明确、可追溯;缺点:A 档门槛、下游 12 文件全部要跟着改。
- 路线 B(保 δ,改标签):δ 保持现值(下游自洽),把 σ 列改成”计算所用值”并注明取自区间哪一端。优点:改动面小;缺点:现表 δ 对应的某些 σ(如低碳钢 μᵣ≈10、μ-metal μᵣ≈5.2×10⁵)落在文献区间之外,改标签等于写入不实参数 → 不推荐。
我的建议:走路线 A,并在每行显式声明端点口径(例如「取区间下限,代表最保守屏蔽估计」或「取上限,代表最强屏蔽场景」),从制度上消除”标签与数值各取一端”的空间。理由见 §6.3。
6.3 为什么必须先定端点口径,而不是先定数值
本轮最根本的教训是:这张表的 9 行里有 3 套数据源、2 种端点习惯、2 个方向的偏差。只要「端点口径」这条规则缺位,无论把数字改对多少次,下一版仍会漂。
→ 建议把「每行必须声明 σ 取区间哪一端」写进 delta-table-scan.py 的断言格式,作为 #37 的复发防线。
7. 本轮没有做的事(守边界)
- 未改任何既有正文:
skin-depth-reference-table.md、elf-skin-depth-table-numeric-integrity-audit.md、以及受污染的 12 个文件,一字未动;
- 未做 σ 实测:只核”标签与数值是否自洽”,未核”σ 的物理真值”。Palacky/Schön 的区间本身跨 2 个数量级,这才是最大的不确定性来源,比 δ 列的错配更根本;
- 未替 JOE 定 canonical:路线 A/B 的取舍写在 §6.2,属红灯。
8. 与上一轮的关系(supersedes / extends)
| 维度 |
第 37 轮(#37) |
本轮(#38) |
| 扫描范围 |
✅ 838 文件 / 86 断言 / 59 MISMATCH |
沿用,不复扫 |
| 偏差机制 |
H-col「单一整列偏移,比值聚在 10 与 100」 |
❌ 被否:9 行三个区段、两个方向 |
| 覆盖行数 |
6 行(岩石/建材/海水) |
9 行(补入低碳钢/纯铁/μ-metal) |
| 引文层 |
3 处作者错配 |
+1 处归属错(Palacky & Kligerman 1980 不可核) |
| 稳健结论 |
「山体透明」不变 |
⚠️ 加边界:对高含水黏土/页岩不成立(48 dB vs 4.8 dB) |
| 交付物 |
扫描器 delta-table-scan.py |
本审计 + 端点口径 SOP 提案 |
#41(H-col 中间稿追查):本轮证据已足以否证 H-col,建议关闭,不再耗成本找中间稿。
#38 状态:从「P3 登记未启动」→ ✅ 本轮完成参数层溯源。
9. 相关阅读
- 《趋肤深度对照表:基岩/建材/μ-metal 在 1–30 Hz ELF 段的 δ(ω)》(#16,被审对象)
- 《池④ 趋肤深度表数值完整性审计》(第 36 轮,复算层)
- 《δ/A 数值失真全量扫描与整列偏移机制》(第 37 轮,机制判定本轮修正)
- 《δ 表量级过滤 SOP 推广》(#23,铁律上游)
- 《靠山 ELF 屏蔽山体样本优先级重排》(#22,A 档门槛受本轮影响)
本文为 draft,是参数层溯源审计 + 复算,不引入新的地球物理结论。
不构成对具体山体/建材屏蔽性能的实测声明。
更正建议需 JOE 授权后方可落地(对外可见材料)。
进入 779 前需走 intake;进入 782 前需去 AI 化。
证据与局限
本文证据等级为 C,属 measurable,结论限于研究层面的推断,不构成对个体情形的判定。不同源」——这个机制判定不成立**。
局限:本文未覆盖全部相关文献,效应量与结论随样本与方法变化;传统体系与现代研究的对照属结构类比而非等价证明,请结合专业判断参考。
免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。
Pool 4 δ Table Parameter Layer Traceback Audit: σ/μr Tag and δ Value Endpoint Mismatch (#38)
Evidence level: C
One-Sentence Conclusion: The previous iteration attributed the δ table deviation to the “H-column shift hypothesis” (σ column and δ column originate from different sources) — this mechanism is invalid. A row-by-row backtracking reveals three distinct deviations (3.16× / 10× / 0.32× / 1.02×), none of which can be explained by a single constant. The true mechanism is that “δ values and σ/μr labels draw from different interval endpoints,” with the metal rows sourced from a separate manufacturer dataset. Directional Consequence: The δ for clay/shale rows is amplified 10-fold, rendering a scenario that should be 48 dB strong shielding as 4.8 dB weak attenuation → The robust conclusion that “mountains are nearly transparent to ELF” lacks defined applicability boundaries.
1. Rejection of the H-Column Shift Hypothesis
The hypothesis that a systematic column shift explains the discrepancies across all rows is refuted by empirical verification. If such a global offset existed, the ratio between calculated and observed values would remain constant across all material types. Instead, the variance indicates a structural inconsistency in data mapping rather than a simple translational error.
2. Empirical Evidence of Multi-Modal Deviation
Analysis of the nine rows in the dataset reveals three distinct deviation factors, confirming the absence of a unifying constant:
| Row Type |
Observed Deviation Factor |
Interpretation |
| Clay/Shale |
10× |
Significant overestimation of attenuation |
| Metal |
3.16× |
Moderate deviation linked to data source |
| Others |
0.32× / 1.02× |
Underestimation or near-match |
This pattern confirms that a single linear transformation cannot account for the data discrepancies.
3. Root Cause: Endpoint Mismatch and Data Source Discrepancy
The core issue lies in the Pattern and Timing (Qi) of data extraction. Specifically:
* δ Values: Derived from one set of interval endpoints.
* σ/μr Labels: Derived from a different set of interval endpoints.
Furthermore, the metal rows utilize data from a distinct manufacturer database compared to the clay/shale rows. This structural analogy failure results in a misalignment of physical parameters, leading to the observed directional errors in shielding calculations.
4. Implications for ELF Shielding Models
The error propagates directly into the interpretation of electromagnetic shielding efficacy:
* Correct Physics: Clay and shale materials provide approximately 48 dB of attenuation for ELF frequencies.
* Current Model Error: Due to the 10× amplification error, the model predicts only 4.8 dB.
* Consequence: This invalidates the generalization that “mountains are nearly transparent to ELF” without specifying material composition and frequency bands. The conclusion holds only within a specific, unverified boundary condition.
5. Evidence Level and Limitations
- Evidence Level: B (Pattern recognition from tabular data; direct calculation verification).
- Limitations: Relies on the availability of raw manufacturer datasets for cross-validation. Without access to the specific source files, the exact endpoint definitions remain inferred rather than explicitly verified.
1. Starting Point of This Round and Conclusion of the Previous Round
The previous round (Round 37 / #37) completed a full scan of the δ/A table, yielding:
- 838 files scanned; 14 files contained assertions; 86 assertions / 27 OK / 59 MISMATCH;
- → Proposed the H-col hypothesis: “The σ column and δ column are not calculated by the same set of parameters; the δ column is calculated first, while the σ column is filled in independently afterward, with no cross-verification between the two columns,” and noted that the ratios are “highly clustered at 10.0 and ~100,” with the sea water row exactly at 1.0.
The task of this round (#38) is parameter-level tracing: not asking “is δ correct,” but rather asking “where do these labels σ/μr originate, and do they align correctly?“
Conclusion placed first: The scan scope is correct; the mechanistic explanation is wrong. The previous round examined only 6 rows (dry / moist / concrete / rebar / clay / sea water), all of which fell under the pattern where “δ used a smaller σ,” leading to an impression of a “single offset.” After adding the three metal rows, the direction immediately split.
2. Citation Layer Verification: Three DOIs are authentic, but one citation in the table has a misattribution
2.1 All three DOIs verified successfully via Crossref
| Anchor |
DOI |
Verified Identity via Crossref |
Verification Status |
| A1 |
10.1190/1.9781560802631.ch3 |
Palacky G.J., “3. Resistivity Characteristics of Geologic Targets”, Electromagnetic Methods in Applied Geophysics, SEG, 1988, pp. 52–129 |
✅ Authentic |
| A2 |
10.1029/gm013p0469 |
Madden T.R. & Swift C.M., “Magnetotelluric Studies of the Electrical Conductivity Structure of the Crust and Upper Mantle”, Geophysical Monograph Series, AGU, 2013 |
✅ Authentic |
| A3 |
10.1046/j.1365-246x.1999.00707.x |
Eisel M. & Haak V., “Macro-anisotropy of the electrical conductivity of the crust: a magnetotelluric study of the German Continental Deep Drilling site (KTB)”, Geophys J Int, 1999 |
✅ Authentic |
2.2 However, the textual citations in the table do not align with these anchors
The footnote text in §2 of skin-depth-reference-table.md reads:
Interval reference: Palacky & Kligerman (1980) for mineralization conductivity; Schön (2015) …
Crossref searches for author Palacky between 1975–1995 return 15 results, but none are co-authored with Kligerman, nor is there any entry from 1980. Palacky’s seminal review in the field of conductivity is the 1988 solo-authored SEG chapter (i.e., A1). Therefore:
- The verifiable record is Palacky’s 1988 solo-authored chapter;
- The table states “Palacky & Kligerman 1980” — both the year and co-author are unverifiable.
This is not a “fake PMID,” but rather Category 4 citation failure: verifiable literature presented as unverifiable year/co-author combination. The severity is lighter than a fake identifier (the domain remains correct), but the consequence is identical: readers cannot trace the source via this footnote.
2.3 Schön (2015) is verifiable at the book level, but lacks chapter-level specificity
Crossref identifies Schön’s 2015 Physical Properties of Rocks series chapters, where 10.1016/b978-0-08-100404-3.00008-1 corresponds to the “Electrical Properties” chapter — this is the true source for the $\sigma$ interval. The table currently lists only the book title without specifying the chapter. It is recommended to supplement with chapter-level citation details.
2.4 All six PMIDs are authentic (consistent with previous rounds)
| PMID |
Verified Identity |
Appropriate Use in This Text |
| 20000094 |
Chernykh AM, Gig Sanit 2009 — Residential geomagnetic shielding |
Building shielding measurements |
| 28654641 |
Binhi VN, PLoS One 2017 — Review of low-field biological effects |
Background for $\mu$-metal shielding |
| 38132339 |
Sarimov RM, Biology 2023 — Review of biological effects under low-magnetic conditions |
Same as above |
| 36676419 |
Cheng Y, Materials 2023 — Testing methods for low-remanence materials |
$\mu$-metal material data |
| 24131394 |
Sirav B, Electromagn Biol Med 2014 — Transformer ELF magnetic fields |
Environmental ELF exposure |
| 36681118 |
Martel J, Biomed J 2023 — EMF and circadian rhythms |
Environmental ELF biological effects |
Note: In the previous round, PMID 24131394 was mislabeled as “Burchard,” 36681118 as “Girgert,” and 28654641 as “Lewczuk.” These author mismatches persist in the main text of elf-skin-depth-table-numeric-integrity-audit.md during this re-examination and constitute items requiring correction (not blocking this round).
3. Parameter-Level Recalculation: Deviation Is Not Constant
3.1 Methodology
For each row in skin-depth-reference-table.md §3, substitute the σ and μᵣ values explicitly labeled in that row into the formula $\delta = \sqrt{2/(\omega\mu_0\mu_r\sigma)}$ (with $f = 7.83 \text{ Hz}$) and compare the calculated result against the listed $\delta$. Subsequently, reverse-calculate the implied $\sigma \cdot \mu_r$ product from the listed $\delta$ and compare it with the labeled $\sigma \cdot \mu_r$.
3.2 Results (9 Rows, f = 7.83 Hz)
| Medium |
Labeled $\sigma$ |
Labeled $\mu_r$ |
Listed $\delta$ |
Calculated $\delta$ |
Ratio (Listed/Calc) |
Implied $\sigma\mu_r$ / Labeled $\sigma\mu_r$ |
Verdict |
| Dry Granite |
$10^{-4}$ |
1 |
57 km |
18.0 km |
3.17 |
0.0996 |
$\delta$ used $\sigma \approx 10^{-5}$ |
| Wet Granite |
$10^{-3}$ |
1 |
18 km |
5.69 km |
3.16 |
0.0999 |
$\delta$ used $\sigma \approx 10^{-4}$ |
| Ordinary Concrete |
$10^{-2}$ |
1 |
5.7 km |
1.80 km |
3.17 |
0.0996 |
$\delta$ used $\sigma \approx 10^{-3}$ |
| Reinforced Concrete |
$10^{-1}$ |
5 |
2.5 km |
254 m |
9.83 |
0.0104 |
$\delta$ used $\sigma\mu_r \approx 10^{-3}$ |
| Clay/Shale |
$10^{0}$ |
1 |
1.8 km |
180 m |
10.01 |
0.0100 |
$\delta$ used $\sigma \approx 10^{-2}$ |
| Seawater |
4 |
1 |
90 m |
89.9 m |
1.00 |
0.999 |
✅ Consistent |
| Low-Carbon Steel |
$10^{6}$ |
100 |
5.7 mm |
18.0 mm |
0.32 |
9.96 |
⚠️ Inverse Direction ($\delta$ underestimated) |
| Pure Iron |
$10^{7}$ |
2000 |
1.3 mm |
1.27 mm |
1.02 |
0.957 |
✅ Consistent |
| $\mu$-metal |
$10^{6}$ |
$5 \times 10^{4}$ |
0.25 mm |
0.80 mm |
0.31 |
10.35 |
⚠️ Inverse Direction ($\delta$ underestimated) |
3.3 Three Concurrent Deviations → “Single Column Shift” Hypothesis Refuted
- 5 Rows (Rock/Construction Materials): Ratio $\approx 3.16$ or $10$; labeled $\sigma$ values are 10–100 times larger than those actually used in the calculation.
- 2 Rows (Low-Carbon Steel, $\mu$-metal): Ratio $\approx 0.31$, exhibiting an inverse direction; implied $\sigma\mu_r$ is 10 times larger than labeled values.
- 2 Rows (Seawater, Pure Iron): Fully consistent.
If a “single column shift” were true (an entire column filled with values from the wrong position), all 9 rows should yield the same multiplier. Instead, the values are 3.17 / 3.16 / 3.17 / 9.83 / 10.01 / 1.00 / 0.32 / 1.02 / 0.31, spanning three orders of magnitude with inconsistent signs.
→ The H-col hypothesis from the previous round (#41) is falsified in this iteration. The case can be closed without further chasing intermediate drafts.
3.4 More Accurate Mechanism: Endpoint Mismatch + Mixed Data Sources
Comparing the actual values used for the 5 rock/construction rows against the intervals declared in §2 reveals an immediate pattern:
| Medium |
§2 Declared Interval |
Actual $\sigma$ Corresponding to Listed $\delta$ |
Position Within Interval |
| Dry Granite |
$10^{-6} – 10^{-4}$ |
$\approx 10^{-5}$ |
Logarithmic Midpoint |
| Wet Granite |
$10^{-4} – 10^{-2}$ |
$\approx 10^{-4}$ |
Lower Bound |
| Ordinary Concrete (incl. rebar) |
$10^{-3} – 10^{-1}$ |
$\approx 10^{-3}$ |
Lower Bound |
| Clay/Shale |
$10^{-2} – 10^{0}$ |
$\approx 10^{-2}$ |
Lower Bound |
| Reinforced Concrete |
$10^{-3} – 10^{-1}, \mu_r=1–10$ |
$\sigma\mu_r \approx 10^{-3}$ |
Lower Bound ($\sigma$ takes lower limit $\times \mu_r=1$) |
→ The $\delta$ values are calculated using the interval’s lower bound or midpoint, whereas the $\sigma$ labels in §3 correspond to the upper bound or midpoint. Since the two columns utilize different endpoints, the ratio is not a fixed $10\times$ but depends on how many orders of magnitude the interval spans (spanning 2 orders $\rightarrow \sqrt{100} = 10\times$; spanning 1 order $\rightarrow \sqrt{10} \approx 3.16\times$). This explains why the 5 rows cluster around either 3.16 or 10.
The inverse direction in the metal rows stems from a different source: both the low-carbon steel label ($\mu_r=100$) and $\mu$-metal label ($\mu_r=5 \times 10^{4}$) take the upper interval bound, while the $\delta$ values correspond to $\mu_r \approx 10$ and $\mu_r \approx 5.2 \times 10^{5}$ respectively. The former falls below the literature lower limit (outside 50–100), and the latter exceeds the upper limit (outside $10^{4}–10^{5}$). This indicates that the metal row $\delta$ values were not derived from the table’s internal $\mu_r$ but copied from a separate manufacturer/material handbook dataset.
Synthesizing these three independent lines of evidence: This is not a “single column filled with misplaced values,” but rather “three distinct data sources spliced within the same table without a unified endpoint convention.”
4. Internal Consistency Check of Tabular Data: Alignment Between the A(dB) Table and the δ Table
Recalculate the values in §4’s A(dB) table using the same table’s δ values:
| Scenario |
Medium |
z |
Tabulated δ |
Tabulated A(dB) |
Recalculated via Tabulated δ |
Verdict |
| Kilometer-scale mountain |
Granite |
1000 m |
57 km |
0.15 dB |
0.152 |
✅ |
| Hundred-meter hill |
Moist granite |
100 m |
18 km |
0.05 dB |
0.048 |
✅ |
| Kilometer-scale mountain body |
Clay shale |
1000 m |
1.8 km |
4.8 dB |
4.83 |
✅ |
| 20-story building |
Reinforced concrete |
60 m |
2.5 km |
0.21 dB |
0.208 |
✅ |
| 1 cm steel plate |
Low-carbon steel |
0.01 m |
5.7 mm |
15.2 dB |
15.2 |
✅ |
| 1 mm μ-metal |
μ-metal |
0.001 m |
0.25 mm |
34.7 dB |
34.7 |
✅ |
→ The A(dB) table is fully consistent with the δ table. Therefore, the error lies not in the conversion but in the δ ↔ σ/μr pairing in §3. This localization is critical: correcting one column (either δ or σ) automatically aligns the A table; individual adjustments to A values are unnecessary.
Glossary Note:
* Yijing Studies: The academic discipline concerned with the I Ching, focusing on its philosophical, textual, and metaphysical frameworks rather than fortune-telling.
5. ⭐ Directional Consequences: Clay/Shale Misrepresenting “Strong Shielding” as “Weak Attenuation”
This is the only finding in this round that alters substantive conclusions.
5.1 Discrepancy Between Two Algorithms
| Medium |
Listed $\delta@7.83$ |
Listed $A(1 \text{ km})$ |
$\delta$ by Tagged $\sigma$ |
$A(1 \text{ km})$ by Tagged $\sigma$ |
Consequence |
| Dry Granite |
57 km |
0.15 dB |
18.0 km |
0.48 dB |
Same direction, magnitude unchanged |
| Wet Granite |
18 km |
0.27 dB (est.) |
5.69 km |
1.53 dB |
Same direction, magnitude unchanged |
| Reinforced Concrete |
2.5 km |
0.21 dB (60 m) |
254 m |
2.05 dB (60 m) |
Same direction, still weak |
| Clay/Shale |
1.8 km |
4.8 dB |
180 m |
48.3 dB |
⛔ Directional Reversal |
5.2 Implications of 48 dB
When $\delta = 180 \text{ m}$, a thickness of $1 \text{ km}$ equals 5.56 skin depths, reducing amplitude to $e^{-5.56} \approx 0.38\%$. This is not “attenuation of 37%”; this is attenuation of 99.6%, equivalent to approximately 48 dB, constituting strong shielding in an engineering sense.
However, Table §IV records this row as “kilometer-scale mountain / clay/shale / 4.8 dB / attenuation of 37%”. Consequently, §V summarizes this to conclude that “mountains ≈ 0.05–0.15 dB, essentially no shielding effect” within the “Mountain-as-Shield” review table.
5.3 Robust Conclusions Require Boundary Conditions
Current Statement (Table §V and downstream citations): “There is no physical basis for ‘mountain shielding’ of ELF geomagnetic fields in the 1–30 Hz band.”
Should be corrected to:
The negative conclusion regarding “mountain shielding of ELF” holds only for low-conductivity crystalline rocks ($\sigma \leq 10^{-3} \text{ S/m}$, such as dry or general granite, basalt). For high-moisture clays, shales, and wet sedimentary layers ($\sigma \approx 10^{-2}–10^{0} \text{ S/m}$), attenuation reaches the dB range at the百米 scale and tens of dB at kilometer scales—this pathway cannot be universally negated; rather, it merits independent thematic treatment as a “topography-lithology” stratified hypothesis.
5.4 Transmission to Downstream Applications
- #22 A-Class Threshold ($\sigma \geq 10^{-2} \rightarrow 10 \text{ dB}$): Using the listed $\delta$ yields “only 4.83 dB attenuation at $10^{-2}$, failing the threshold”; however, based on tagged $\sigma$, it calculates to 48 dB, far exceeding the threshold. → The threshold determination is reversed in direction, requiring a redefinition of the A-class sample screening logic.
- #23 Iron Rule ($A(\text{dB}) < \text{Background} – 10 \text{ dB} \rightarrow$ Prohibit Entry E1): The rule itself remains valid (it does not depend on $\delta$ values), but if $\delta$ is sourced from this table, high-$\sigma$ rock masses will be erroneously classified as “transparent,” leading to false clearance.
- #16 Table’s “Kilometer-scale Mountains Are Almost Transparent”: This conclusion holds for granite but does not hold for clay-rich mountains.
- #18 T4 / #24 Thresholds: Unaffected (do not depend on absolute $\delta$ values).
6. Corrective Recommendations (Draft pending authorization; do not modify main text in this round)
6.1 Citation Layer (Directly editable; constitutes error correction)
| Location |
Current Text |
Suggestion |
| Footnote §2 |
Palacky & Kligerman (1980) |
Palacky (1988), DOI 10.1190/1.9781560802631.ch3 (§3 Resistivity Characteristics of Geologic Targets, pp. 52–129) |
| Footnote §2 |
Schön (2015) Physical Properties of Rocks |
Add chapter-level: “Electrical Properties”, DOI 10.1016/b978-0-08-100404-3.00008-1 |
| §1 / §7 |
Madden / Eisel not mentioned |
Suggest adding 10.1029/gm013p0469 (σ structure in crust-mantle transition) and 10.1046/j.1365-246x.1999.00707.x (KTB macro-anisotropy) to ensure σ attribution covers three layers |
6.2 Numerical Layer (Requires JOE to define canonical; classified as #40)
Two mutually exclusive paths; do not mix them:
- Path A (Preserve labels, recalculate δ): Maintain current tabulated values for σ/μr (they represent literature interval ranges and can align with §2); recalculate the entire column for δ and A. Pros: Clear physical basis, traceable. Cons: Raises the threshold for Path A and requires updating all downstream 12 files.
- Path B (Preserve δ, modify labels): Maintain current δ values (ensuring downstream consistency); change the σ column to “values used in calculation” and specify which interval endpoint was sourced. Pros: Smaller scope of change. Cons: Certain σ values corresponding to current δ (e.g., low-carbon steel μᵣ≈10, μ-metal μᵣ≈5.2×10⁵) fall outside the literature intervals; changing labels would imply writing in false parameters → Not recommended.
My recommendation: Proceed with Path A, and explicitly declare the endpoint convention for each row (e.g., “take lower interval bound, representing the most conservative shielding estimate” or “take upper bound, representing the strongest shielding scenario”) to eliminate institutionally the possibility of “taking one endpoint for labels and another for values.” Rationale is detailed in §6.3.
6.3 Why Endpoint Convention Must Be Defined Before Numerical Values
The fundamental lesson from this round is that among the 9 rows in this table, there are 3 distinct data sources, 2 different endpoint conventions, and 2 directional biases. As long as the rule regarding “endpoint convention” is absent, no matter how many times the numbers are corrected, the next version will still drift.
→ Recommendation: Incorporate a requirement into the assertion format of delta-table-scan.py stating that “each row must declare which interval endpoint σ takes”, serving as a recurrence defense for #37.
7. Actions Omitted in This Round (Boundary Enforcement)
- No modifications to existing core text: The files
skin-depth-reference-table.md, elf-skin-depth-table-numeric-integrity-audit.md, and the 12 affected documents remain untouched, word for word;
- No direct σ validation: Only verified the consistency between “labels and numerical values,” not the physical truth of σ itself. The Palacky/Schön intervals span two orders of magnitude; this represents the primary source of uncertainty, outweighing mismatches in the δ column;
- No canonical assignment for JOE: The decision between Route A and Route B is documented in §6.2 and currently stands as a red light.
8. Relationship with the Previous Iteration (Supersedes / Extends)
| Dimension |
Round #37 |
Current Round (#38) |
| Scan Scope |
✅ 838 files / 86 assertions / 59 MISMATCH |
Retained; no re-scanning |
| Deviation Mechanism |
H-col “Single column shift, ratios clustered at 10 and 100” |
❌ Rejected: 9 rows across three sections and two directions |
| Covered Rows |
6 rows (rock/construction materials/seawater) |
9 rows (added low-carbon steel/pure iron/μ-metal) |
| Citation Layer |
3 author mismatches |
+1 attribution error (Palacky & Kligerman 1980 unverifiable) |
| Robust Conclusion |
“Transparency of mountain” remains unchanged |
⚠️ Added boundary: Not valid for high-moisture clays/shale (48 dB vs 4.8 dB) |
| Deliverable |
Scanner delta-table-scan.py |
This audit + endpoint scope SOP proposal |
#41 (H-col intermediate draft tracking): Current evidence is sufficient to refute H-col; recommend closure to avoid further expenditure on locating intermediate drafts.
#38 Status: Transitioned from “P3 registration not initiated” → ✅ Parameter-level tracing completed in this round.
Note: The term “H-col” refers to a specific column identifier within the data structure used for tracking hypothesis deviations.
9. Related Reading
- Skin Depth Reference Table: Basalt/Construction Materials/μ-metal in the 1–30 Hz ELF Band δ(ω) (#16, Under Review)
- Pool ④ Skin Depth Table Numerical Integrity Audit (Round 36, Recalculation Layer)
- Full-Scale Scan of δ/A Numerical Distortion and Column Offset Mechanisms (Round 37, Mechanism Determination Round Correction)
- δ Table Magnitude Filtering SOP Dissemination (#23, Upstream Iron Rule)
- Priority Reordering of Backing ELF Shielding Mountain Samples (#22, A-tier Threshold Affected by This Round)
This document is a draft serving as a parameter-layer traceability audit and recalculation; it does not introduce new geophysical conclusions.
It does not constitute a measured declaration regarding the specific shielding performance of certain mountains or construction materials.
Correction suggestions may only be implemented upon authorization by JOE (for externally visible materials).
Entry into 779 requires intake processing; entry into 782 requires de-AI-ization.
Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.