Evidence level: B
上个世纪有两位研究者把女性送进地下和隔离舱,一个观察到周期变短,一个观察到变长。文献里常把这两条并列成”结论不一致”。但不一致的前提是——”方向”本该是一个可以被预测的量。这个前提本身就该先查一遍。
正文
一个方向,两个样本
1966 年发表的三个月地下隔离单例,记录直肠温度节律谱时也记下了月经事件(PMID 4955822);另一份长时间隔离的单例记录给出的是相反方向的偏移。两条都是 n=1。
n=1 的两个点不构成”矛盾”,它们只是两个点。真正的问题在后面:如果”隔离让周期延长”对人群成立,第三、第四个人的方向就该落在同一侧;可如果方向本身在个体间随机,那两个人差得再远也不奇怪——方向差得远反而是默认预期。
所以要问的不是”到底哪个方向”,而是:偏移方向是个体的固有属性(可由基线量预测),还是群体层面的统计噪声? 这是”本征周期预测方向”这个课题的合法性基础。
错的不是答案,是问法
多数隔离研究用的是单次前后比较:进舱前测几轮,出舱后测几轮,取差值。这个设计有三重污染。
一是样本量假象。个体内周期长度的天然波动很大:一项一年前瞻研究(53 人、694 周期)测得个体内周期长度方差中位数 3.1 天²,卵泡期 5.2 天²、黄体期 3.0 天²(PMID 39320898);另一项 581 人、3324 周期的汇总分析发现 43% 的女性最长与最短周期相差超过 7 天(PMID 32104920)。只取”几轮对几轮”的均值差,落在这个天然摆幅内的偏移根本不能指认为环境效应。
二是回归均值。若入舱本就是因为”最近偏长”被注意到,出舱后的”变短”有很大一部分只是随机误差回弹。
三是方向的分类化。把连续变化硬切成”延长/缩短”,既丢幅度,又让标签不稳定——基线窗口多取一轮,标签就可能翻转。
方差不在你以为的那一段
一个生理事实决定这个课题该怎么问:周期长度的变异主要来自卵泡期,不是黄体期。
那项一年前瞻研究直接比较了两段:个体内卵泡期方差显著大于黄体期(P < 0.001);55% 的女性至少经历一次黄体期短于 10 天、17% 至少出现一个无排卵周期(PMID 39320898)。而且 29% 的周期存在亚临床排卵障碍——总长度往往仍在正常范围内,但内部分配已经变了。
外推是:所谓”隔离让周期变长或变短”,严格说不是”钟走快了或慢了”,而是卵泡招募到排卵这一段耗时变了。黄体期相对守规矩(中位约 12-14 天,PMID 32104920),能变的空间小。把”方向”当成整座钟的周期读数,是量纲错配。
要检验它,设计必须长什么样
多例,不是多轮。 要把”个体间方向差异”从”个体内摆动”里分离,需要足够多的个体各自贡献多条周期,用随机效应拆开个体水平与个体内波动。
长基线。 基线窗口太短,”本征周期”估计本身就是噪声。隔离前必须积累足够多轮,才能给出可信的个体均值与个体标准差(状态空间模型刻画运动员周期即属此路,PMID 34417493)。
方向作为个体斜率,不做二值分类。 给每个个体拟合变化斜率,再问”斜率能否被某个基线量解释”。
必须有扰动侧对照。 隔离只是”缺外部授时”,不做相位强制位移。真正能区分独立振荡器与计数型从属装置的,是对昼夜节律施加可控扰动后看月级成分跟不跟着变(详见 dual-clock-free-running-independence)。
候选的个体预测因子
年龄与生殖衰老:周期变异性随年龄上升、临近绝经时结构变化,已有分层变点模型可刻画(PMID 24729638);大样本 app 数据显示年龄与早年周期不规则共同塑造变异(PMID 41297783)。
遗传背景:英国双生子样本给出周期历史的遗传/环境贡献分解(PMID 3454517);大规模 meta 分析把周期长度的遗传调控指向下丘脑-垂体-性腺轴(PMID 30202859)。”个体本征阈值”至少有一部分可遗传。
代谢与体重:个体内 BMI 变化与月经不规则的关联有真实世界队列证据(PMID 41606583)。
炎症与急性应激:前瞻队列提示全身炎症与周期长度相关(PMID 36244407);884 名感染后女性的研究显示急性系统应激会改变周期表现(PMID 42277801)。
外部节律侧的候选——周节律与月节律叠加(PMID 38206269)、月相与月经起始的关联(PMID 33571128、PMID 40991714)——更适合当自变量,用来说明外部授时信息有多强,而不是用来说明个体差异从哪来。
落到易学上
运气学说里有个常被忽略的结构:同一个”岁运”,走两个方向。
《素问·气交变大论》讲五运,每一运都分”太过”与”不及”两套话。同一个五行、同一个年份位置,落在谁身上可以是太过,落在另一个人身上可以是不及。两个方向不是两个结论,是同一套机制的两个出口。
更贴切的是用神。同一味药、同一个方位,在这个命局里是喜神,换个命局就成忌神——”这个因素让人变好还是变坏”这个问法,在易学里从来不成立,它必须先放进个体的本命。
这正是本篇要说的:方向不是环境物的属性,是个体本征与环境相位的关系。 古人说”太过不及取决于命局”,现代研究说”偏移方向需由个体本征量预测”——问的是同一件事。而两套体系眼下面对同一个空白:我们还没有一个可信的”本征量”来算这个方向。
证据与局限
已验证: 个体内周期长度波动很大——一年前瞻研究中个体内周期长度方差中位数 3.1 天²(PMID 39320898)、581 人汇总分析中 43% 女性最长最短周期相差超 7 天(PMID 32104920);个体内卵泡期方差显著大于黄体期(P < 0.001,PMID 39320898),黄体期长度相对稳定、中位约 12-14 天(PMID 32104920、PMID 28065408);亚临床排卵障碍普遍(55% 女性至少一次短黄体、17% 至少一次无排卵,PMID 39320898);周期长度遗传度有双生子证据(PMID 3454517)与大规模 GWAS 定位到下丘脑-垂体-性腺轴(PMID 30202859);年龄相关周期变异可用分层变点模型刻画(PMID 24729638)并有真实世界数据支持(PMID 41297783、PMID 38702411);个体内 BMI 变化与月经不规则关联(PMID 41606583);全身炎症与周期长度相关(PMID 36244407);急性系统应激改变周期表现(PMID 42277801);40 天洞穴群体隔离确会解耦昼夜节律(PMID 42009612);个体昼夜本征周期存在差异(PMID 32619243、PMID 22846439、PMID 17684566)。
部分验证: 人类隔离状态下月经方向改变,公开证据仅为两个方向相反的单例记录(PMID 4955822 及同期隔离记录),无任何多例设计;”偏移方向可由本征周期预测”从未被直接检验过一次;本征近月周期本身在人类中尚无可靠测量手段。
待验证: 偏移方向是多例设计下的个体固有属性还是群体随机量;若为个体属性,年龄、遗传、代谢、炎症四类预测因子中哪类贡献最大;方向作为个体斜率(而非二值标签)时,所需基线长度与个体数下界;基线窗口长度对”方向标签”稳定性的影响曲线。
相关阅读
证据等级: B
参考文献: PMID 4955822, PMID 39320898, PMID 32104920, PMID 28065408, PMID 3454517, PMID 30202859, PMID 24729638, PMID 41297783, PMID 38702411, PMID 41606583, PMID 36244407, PMID 42277801, PMID 42009612, PMID 32619243, PMID 22846439, PMID 17684566, PMID 34417493, PMID 38206269, PMID 33571128, PMID 40991714, PMID 36376588, PMID 16700687, PMID 19196217, PMID 42628374, PMID 32998909
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。
Who Owns the Direction of Shift: Can Intrinsic Cycle Duration Predict Changes in Menstrual Cycle Length Within Isolation Experiments
Evidence level: B
In the last century, two researchers sent women into underground isolation chambers; one observed a shortened cycle, the other a lengthened one. Literature often lists these as “inconsistent conclusions.” However, the premise of inconsistency is that “direction” should be a predictable quantity. This premise itself warrants verification.
Main Text
One Direction, Two Samples
A three-month underground isolation case published in 1966 recorded rectal temperature rhythm spectra and menstrual events (PMID 4955822); another long-duration isolation case recorded an opposite-directional shift. Both have n=1.
Two points with n=1 do not constitute a “contradiction”; they are merely two data points. The real issue lies later: if “isolation causes cycle lengthening” holds for the population, the direction for the third and fourth individuals should fall on the same side. However, if direction itself is random across individuals, a large discrepancy between two people is not surprising—a large directional difference is actually the default expectation.
Therefore, the question should not be “which direction,” but rather: Is the direction of deviation an individual intrinsic attribute (predictable by baseline metrics), or population-level statistical noise? This is the legitimacy foundation for the topic of “predicting direction from intrinsic cycles.”
The Error Is Not the Answer, But the Questioning Method
Most isolation studies use single pre-post comparisons: measuring several cycles before entry and after exit, then taking the difference. This design suffers from three forms of contamination.
First is the sample size illusion. Intra-individual cycle length naturally fluctuates significantly: a one-year prospective study (53 subjects, 694 cycles) found a median intra-individual cycle length variance of 3.1 days², with follicular phase variance at 5.2 days² and luteal phase at 3.0 days² (PMID 39320898); another meta-analysis of 581 subjects and 3,324 cycles found that 43% of women differed by more than 7 days between their longest and shortest cycles (PMID 32104920). Taking the mean difference of “a few cycles against a few cycles” falls within this natural amplitude; such shifts cannot be attributed to environmental effects.
Second is regression to the mean. If entry was triggered because “recent cycles were unusually long,” much of the observed “shortening” post-exit is merely random error rebound.
Third is categorical classification of direction. Forcing continuous change into “lengthening/shortening” loses magnitude and creates unstable labels—a single extra baseline cycle can flip the label.
Variance Is Not Where You Think It Is
A physiological fact dictates how this topic should be framed: variation in cycle length primarily stems from the follicular phase, not the luteal phase.
That one-year prospective study directly compared two segments: intra-individual follicular phase variance was significantly greater than luteal phase variance (P < 0.001); 55% of women experienced at least one luteal phase shorter than 10 days, and 17% had at least one anovulatory cycle (PMID 39320898). Furthermore, 29% of cycles exhibited subclinical ovulatory dysfunction—total length often remained within normal ranges, but internal allocation had changed.
Extrapolating: the claim that “isolation lengthens or shortens cycles” strictly refers not to “the clock running fast or slow,” but rather that the duration from follicular recruitment to ovulation has changed. The luteal phase is relatively regulated (median ~12-14 days, PMID 32104920), leaving little room for variation. Treating “direction” as the entire clock’s cycle reading is a mismatch of dimensions.
Experimental Design Requirements for Verification
Multiple subjects, not multiple cycles. To separate “inter-individual directional differences” from “intra-individual swings,” sufficient numbers of individuals must each contribute multiple cycles, using random effects to disentangle individual-level variance from intra-individual fluctuation.
Long baseline. A short baseline window renders the estimation of “intrinsic cycles” itself noisy. Before isolation, enough cycles must be accumulated to provide credible individual means and standard deviations (state-space models characterizing athlete cycles follow this path, PMID 34417493).
Direction as individual slope, not binary classification. Fit a change slope for each individual, then ask whether “the slope can be explained by certain baseline metrics.”
Must include a perturbation control. Isolation is merely “lack of external entrainment,” not forced phase displacement. True distinction between independent oscillators and counting-type subordinate devices requires observing whether lunar components follow changes after applying controlled perturbations to circadian rhythms (see dual-clock-free-running-independence).
Candidate Individual Predictors
Age and Reproductive Aging: Cycle variability increases with age and structural changes occur near menopause; stratified change-point models can characterize this (PMID 24729638); large-sample app data shows age and early-life cycle irregularity jointly shape variability (PMID 41297783).
Genetic Background: A UK twin sample decomposed genetic/environmental contributions to cycle history (PMID 3454517); a large-scale meta-analysis points genetic regulation of cycle length toward the hypothalamic-pituitary-gonadal axis (PMID 30202859). “Individual intrinsic thresholds” are at least partially heritable.
Metabolism and Weight: Real-world cohort evidence links intra-individual BMI changes to menstrual irregularity (PMID 41606583).
Inflammation and Acute Stress: Prospective cohorts suggest systemic inflammation correlates with cycle length (PMID 36244407); a study of 884 women post-infection showed acute systemic stress alters cycle performance (PMID 42277801).
Candidates on the external rhythm side—superposition of circadian and lunar rhythms (PMID 38206269), association between lunar phase and menstruation onset (PMID 33571128, PMID 40991714)—are better suited as independent variables to explain the strength of external entrainment information, rather than explaining the source of individual differences.
Application to Yijing Studies
In fortune-telling theory, there is a structure often overlooked: the same “Yearly Cycle” (Sui Yun) moves in two directions.
The Su Wen · Qi Jiao Da Bian Lun discusses the Five Transmissions, dividing each transmission into two sets of discourse: “Excess” and “Deficiency.” The same Five Phase and yearly position can be Excess for one person and Deficiency for another. Two directions are not two conclusions; they are two exits of the same mechanism.
More fitting is the use of “Shen” (Deities). The same herb or direction is a beneficial Shen in one destiny chart but becomes an adverse Shen in another—the question “does this factor make someone better or worse” does not hold in Yijing; it must first be placed within the individual’s destiny.
This is precisely what this article addresses: direction is not an attribute of the environment, but the relationship between individual intrinsic factors and environmental phase. The ancients said “Excess or Deficiency depends on the destiny chart”; modern research says “deviation direction must be predicted by individual intrinsic metrics”—asking the same thing. Yet both systems face the same gap today: we lack a credible “intrinsic metric” to calculate this direction.
Evidence and Limitations
Verified: Intra-individual cycle length fluctuates significantly—median intra-individual cycle length variance of 3.1 days² in a one-year prospective study (PMID 39320898), and 43% of women in a meta-analysis of 581 subjects differed by over 7 days between longest and shortest cycles (PMID 32104920); intra-individual follicular phase variance is significantly greater than luteal phase (P < 0.001, PMID 39320898), while the luteal phase is relatively stable with a median of ~12-14 days (PMID 32104920, PMID 28065408); subclinical ovulatory dysfunction is common (55% of women had at least one short luteal phase, 17% had at least one anovulatory cycle, PMID 39320898); heritability of cycle length has twin evidence (PMID 3454517) and large-scale GWAS locates it to the hypothalamic-pituitary-gonadal axis (PMID 30202859); age-related cycle variability can be characterized by stratified change-point models (PMID 24729638) with real-world data support (PMID 41297783, PMID 38702411); intra-individual BMI changes correlate with menstrual irregularity (PMID 41606583); systemic inflammation correlates with cycle length (PMID 36244407); acute systemic stress alters cycle performance (PMID 42277801); 40-day cave group isolation does indeed decouple circadian rhythms (PMID 42009612); individual circadian intrinsic cycles exhibit differences (PMID 32619243, PMID 22846439, PMID 17684566).
Partially Verified: Changes in menstrual direction under human isolation are evidenced by only two single-case records with opposite directions (PMID 4955822 and concurrent isolation records); there is no multi-subject design; “deviation direction predicted by intrinsic cycles” has never been directly tested once; reliable measurement of intrinsic lunar cycles in humans currently lacks a valid method.
To Be Verified: Whether deviation direction is an individual intrinsic attribute or population random quantity under multi-subject designs; if intrinsic, which of the four predictor categories (age, genetics, metabolism, inflammation) contributes most; required baseline length and individual count lower bounds when direction is treated as an individual slope (not binary label); impact curve of baseline window length on “direction label” stability.
Related Reading
- Constant Darkness Monitoring Protocol: Does the Human Body Hide a 29.5-Day Free-Running Near-Lunar Clock
- Two Spectral Peaks Do Not Equal Two Clocks: Are Circadian and Near-Lunar Clocks Truly Independent
- Finding a 29.5-Day Clock in Human Blood
- Environment-Life Frequency Interface
Evidence Level: B
References: PMID 4955822, PMID 39320898, PMID 32104920, PMID 28065408, PMID 3454517, PMID 30202859, PMID 24729638, PMID 41297783, PMID 38702411, PMID 41606583, PMID 36244407, PMID 42277801, PMID 42009612, PMID 32619243, PMID 22846439, PMID 17684566, PMID 34417493, PMID 38206269, PMID 33571128, PMID 40991714, PMID 36376588, PMID 16700687, PMID 19196217, PMID 42628374, PMID 32998909
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.
Leave a Reply