Evidence level: B
我们手上有 0.3 勒克斯这个室外照度,也有 1.5 melanopic lux 那条抑制阈值。两者之间隔着四道门,而每一道的通过率都不是常数。
满月夜,开阔地面上的水平照度约 0.1 到 0.3 勒克斯。另一头,褪黑素抑制的起始点约在 1.5 melanopic lux(PMID 30697806)。粗看这是道简单除法题:差五到十几倍,不够就是不够。但这两个数根本不在同一个位置上——前者是室外水平照度,后者是到达视网膜、按 melanopsin 作用谱逐点加权之后的剂量。中间那条链上有四道门。
第一道门:窗户和窗帘。 这一级最”人为”:低辐射镀膜玻璃、双层中空玻璃、遮光率各异的窗帘、室内多次反射,每项都在改通过率;而窗帘拉一半还是拉严,是同一个人同一夜晚里的两个实验条件。它几乎没有”标准值”,只有”当时那个窗帘的值”。
第二道门:瞳孔。 它不是固定光阑,是被光本身驱动的自反馈。施霍尔恩等人(PMID 38348477)在 90 勒克斯以下的常用室内照明里,用五基色显示器做出 L、M、S 视锥辐照度匹配、只有 melanopic 辐照度差约三倍的两组光,结果高 melanopic 条件下瞳孔中位尺寸显著变小。你把照度算好了,瞳孔会自己改;改多少,又取决于这段光里 melanopsin 含量的高低。闭眼时这级也没关掉——近红外侧照、透过闭合眼睑成像已能在临床上连续测到瞳孔光反应(PMID 34745751、PMID 41694690)。
第三道门:眼睑,四道门里最被低估的一道。 比耶曼等人(PMID 21721832)逐点测过 450 到 650 纳米的眼睑光谱透过率,两个结论都不温和:一是眼睑在短波段的光学密度远高于以往文献值,该波段平均光学密度 2.1 加减 0.3,受试者间光学密度差异范围约 1.0(相当于透过率相差十倍);二是皮肤色素沉着与眼睑透过率相关性很差,真正决定透过率的是眼睑里那些与波长无关的大分子,以及眼睑的总厚度。
第二条更值得琢磨:眼睑不是一块中性灰片,而是一块光谱有偏的滤光片,而 melanopsin 的作用峰恰在 480 纳米附近——被砍得最狠的那段,正是加权算法最看重的那段。所以”眼睑挡掉百分之多少的光”这句话本身就不成立,必须拿光谱逐点相乘再重新加权。同理,想靠肤色估眼睑透过率,这条路被数据直接堵死。
反过来,闭眼也根本不等断路。菲格罗(PMID 26985450)让健康老人与早醒型失眠者在睡眠中佩戴透过闭合眼睑发光的面罩:闪光蓝光把褪黑素起始时间显著延后(34 加减 30 分钟,红光对照 6 加减 30 分钟),睡眠起始延后约 46 分钟,而闪光红光没有这个效果。闭眼是条窄路,但蓝光仍有可测通过量。
第四道门:眼内介质,主要是晶状体。 这级的特征不是个体差异,而是年龄:晶状体对短波的透过率随年龄下降(PMID 33772847 把晶状体透过光谱与瞳孔大小并列为影响光致褪黑素抑制的两个因子;PMID 33020575 提供了在体测晶状体密度与透过光谱的方法;PMID 18757473 讨论老化的眼在昼夜光感受中的角色)。同一个月亮,二十岁和七十岁的晶状体送进视网膜的是两份光谱。
阈值那一头同样不干净。 1.5 melanopic lux 起自对既有数据集的建模(PMID 30697806),同文给出饱和点在 305 melanopic lux;更早的剂量-反应工作(PMID 10922269)测到的是另一种陡度——只用略高于最大亮光百分之一的强度,就拿到最大抑制量的一半。个体层面 ED50 可跨一个数量级(PMID 31138694);青少年与成人对不同光强的抑制不同(PMID 30803301);而性别差异只在 400 勒克斯以上出现,10 到 200 勒克斯段不显著(PMID 39041348)——在极低照度区,个体差异究竟是缩窄了,还是仅掉到检测下限以下,目前分不开。还有一条边界证据很有力:全盲者眼内受到白光暴露时,褪黑素仍可被显著抑制(PMID 29625838)。通路不依赖成像视觉,所以”闭着眼能不能收到”与”能不能看见”是两件事。
把这些摆在一起,诚实的回答既不是”够”也不是”不够”:眼睑光学密度在个体间差 1.0 就足以把结论翻转一次,窗帘、瞳孔、晶状体还会各再翻一次。真正的问题是——这条链从来没有人在真实条件下端到端量过。实验室里的月光剂量实验多半是暗室、睁眼、固定光谱;而真实的月光暴露发生在闭眼侧卧、窗帘状态随人、晶状体年龄随人的卧室里。两端的自变量不重合,等于拿在甲条件标定的尺子去量乙条件。
要闭合这条链,技术上已可行:可穿戴光谱辐射计做在体剂量测量(PMID 36433426),闭眼瞳孔测量已成熟(PMID 34745751、PMID 41694690),报告规范也已列出该报哪些参数(PMID 31281903;melanopic lux 作为测量量的有效性见 PMID 28545361)。缺的不是工具,是有人愿意在满月夜戴着传感器、拉着不同窗帘、闭着眼,把这几级折扣率一起量出来。
《素问·金匮真言论》讲”天有八风,经有五风”——天的风有八种,进到人身上只剩五种。《灵枢》到《伤寒论》讲”传变”,讲的就是气进入身体这一路会被改造。古人从不认为”天给多少”等于”人得多少”;中介在古人眼里不是管子,是车间:东西进去了,出来已不是原来的东西。眼睑那条对 480 纳米格外狠的透过率曲线,就是这间车间的第一道工序——它不是在削弱月光,是在把月光改成另一种光。
证据与局限
已验证: 眼睑光谱透过率可测且可预测,短波段光学密度显著高于早期估计、个体间光学密度差异范围约 1.0 且与皮肤色素沉着相关性差(PMID 21721832);透过闭合眼睑的蓝光干预可显著延后 DLMO 与睡眠起始(PMID 26985450);闭眼状态下瞳孔尺寸与光反应可被连续测量(PMID 34745751、PMID 41694690);瞳孔收缩显著改变 melanopic EDI 与褪黑素抑制的关系(PMID 38348477);晶状体透过光谱与瞳孔大小为影响光致褪黑素抑制的因子(PMID 33772847),晶状体密度与透过光谱可在体测量(PMID 33020575);褪黑素抑制主要由 melanopsin 驱动、可低至约 1.5 melanopic lux 起始并饱和于约 305 melanopic lux(PMID 30697806);抑制在低剂量端极陡(超过最大亮光约 1% 即达半数最大抑制,PMID 10922269);ED50 个体间跨一个数量级(PMID 31138694);全盲者眼内白光暴露仍可显著抑制褪黑素(PMID 29625838);melanopic lux 对抑制的预测优于普通照度(PMID 28545361);光暴露报告参数已有规范(PMID 31281903)。
部分验证: 住宅窗户与窗帘的可见光透过率对昼夜节律剂量的实际贡献——现有玻璃透过率数据多来自其他应用场景,缺少以 melanopic 加权为准的卧室实测;性别差异是否存在低照度边界(PMID 39041348 显示 400 lux 以上显著、10–200 lux 不显著,但该照度段统计功效本就紧张);闭眼状态下瞳孔是否仍被调制到足以改变 melanopic 剂量的程度。
待验证: 0.1–0.3 lux 室外月照在”窗户 + 窗帘 + 眼睑 + 晶状体”四级串联后的实际视网膜 melanopic EDI 是否低于 1.5 melanopic lux——尚无端到端实测;各级折扣率的方差分解(哪一级贡献最大)未知;眼睑厚度等形态学个体差异能否被外部可测指标预测。
相关阅读
证据等级: B
参考文献: PMID 21721832, PMID 34745751, PMID 41694690, PMID 26985450, PMID 38348477, PMID 33772847, PMID 33020575, PMID 18757473, PMID 30697806, PMID 10922269, PMID 31138694, PMID 30803301, PMID 39041348, PMID 29625838, PMID 28545361, PMID 31281903, PMID 36433426
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。
From Moonlight to Retina: Four Gates to Cross; 0.3 Lux Is Not the Amount Received by the Eye
Evidence level: B
We possess the outdoor illuminance value of 0.3 lux and the melanopic lux inhibition threshold of 1.5. Between these two figures lie four gates, and the transmission rate through each gate is not constant.
On a full moon night, horizontal illuminance on an open ground ranges approximately from 0.1 to 0.3 lux. On the other end, the onset point for melatonin suppression is approximately 1.5 melanopic lux (PMID 30697806). At first glance, this appears to be a simple division problem: a difference of five to ten times; if it is insufficient, it remains insufficient. However, these two numbers occupy fundamentally different positions—the former represents outdoor horizontal illuminance, while the latter represents the dose after reaching the retina and applying point-by-point weighting according to the melanopsin action spectrum. The chain connecting them contains four gates.
Gate One: Windows and Curtains. This level is the most “artificial”: low-emissivity coated glass, double-pane insulated glass, curtains with varying light-blocking rates, and multiple indoor reflections all modify the transmission rate; whether a curtain is half-drawn or fully closed constitutes two experimental conditions for the same person on the same night. It lacks a “standard value,” possessing only “the value of that specific curtain at that moment.”
Gate Two: The Pupil. It is not a fixed aperture but a self-feedback mechanism driven by light itself. Schollhorn et al. (PMID 38348477), within common indoor lighting below 90 lux, used five-primary color displays to create two groups of light where L, M, and S cone irradiance were matched yet the melanopic irradiance differed by approximately three times. Under the higher melanopic condition, the median pupil size significantly decreased. You may calculate the illuminance correctly, but the pupil will adjust itself; the degree of adjustment depends on the concentration of melanopsin within that segment of light. This gate is not closed when eyes are shut—near-infrared side illumination and imaging through closed eyelids can be continuously measured clinically as pupillary light responses (PMID 34745751, PMID 41694690).
Gate Three: The Eyelids, the most underestimated of the four gates. Bieyma et al. (PMID 21721832) measured eyelid spectral transmittance point-by-point across 450 to 650 nanometers, yielding two conclusions that are not moderate: first, the optical density of the eyelids in the short-wave band is significantly higher than values found in prior literature; the average optical density in this band is 2.1 ± 0.3, with inter-subject optical density differences ranging approximately 1.0 (equivalent to a ten-fold difference in transmittance); second, there is poor correlation between skin pigmentation and eyelid transmittance, with true determinants being large molecules within the eyelids unrelated to wavelength and the total thickness of the eyelids.
The second point warrants deeper consideration: the eyelid is not a neutral gray filter but a spectrally biased filter, and the action peak of melanopsin lies near 480 nanometers—the segment most severely attenuated, precisely the segment most valued by the weighting algorithm. Therefore, the statement “the eyelids block X% of light” is fundamentally invalid; one must multiply point-by-point across the spectrum and re-weight. Similarly, attempting to estimate eyelid transmittance based on skin tone is directly blocked by the data.
Conversely, closing the eyes does not equate to an open circuit. Figueras (PMID 26985450) had healthy elderly individuals and early-rise insomnia patients wear masks emitting light through closed eyelids during sleep. Blue light flashes significantly delayed the onset of melatonin (34 ± 30 minutes; red light control: 6 ± 30 minutes), delayed sleep onset by approximately 46 minutes, while flashing red light produced no such effect. Closing the eyes creates a narrow path, yet a measurable quantity of blue light still passes through.
Gate Four: Intraocular Media, Primarily the Lens. The characteristic of this level is not individual variation but age: transmittance of the lens to short wavelengths decreases with age (PMID 33772847 lists lens transmittance spectra and pupil size as two factors influencing light-induced melatonin suppression; PMID 33020575 provides in vivo methods for measuring lens density and transmittance spectra; PMID 18757473 discusses the role of aging eyes in circadian light sensing). The same moon sends two different spectral distributions to the retina for a twenty-year-old versus a seventy-year-old.
The threshold side is equally impure. The 1.5 melanopic lux value originates from modeling existing datasets (PMID 30697806); the same paper identifies a saturation point at 305 melanopic lux. Earlier dose-response studies (PMID 10922269) measured a different steepness—achieving half the maximum inhibition with an intensity only slightly above one percent of maximum brightness. At the individual level, ED50 can span an order of magnitude (PMID 31138694); adolescents and adults respond differently to different light intensities (PMID 30803301); and gender differences only appear above 400 lux, showing no significance in the 10 to 200 lux range (PMID 39041348)—in the extremely low illuminance zone, whether individual differences narrow or merely fall below detection limits remains indistinguishable. There is also strong boundary evidence: fully blind individuals still experience significant melatonin suppression when exposed to white light within the eye (PMID 29625838). The pathway does not rely on image-forming vision; thus, “can one receive it with eyes closed” and “can one see” are two distinct matters.
Placing these facts together, the honest answer is neither “enough” nor “insufficient”: an inter-subject difference in eyelid optical density of 1.0 alone is sufficient to flip the conclusion once; curtains, pupils, and lenses will each flip it again. The real issue is—this chain has never been measured end-to-end under real-world conditions. Most laboratory moonlight dose experiments involve dark rooms, open eyes, and fixed spectra; real moonlight exposure occurs in bedrooms where individuals lie on their sides with eyes closed, curtains vary by person, and lens age varies by person. The independent variables at both ends do not coincide; it is like using a ruler calibrated under condition A to measure condition B.
To close this chain, it is technically feasible: wearable spectral radiometers can perform in vivo dose measurement (PMID 36433426), closed-eye pupil measurement is mature (PMID 34745751, PMID 41694690), and reporting guidelines have listed which parameters should be reported (PMID 31281903; see PMID 28545361 for the validity of melanopic lux as a measured quantity). The lack is not tools, but willingness to wear sensors on a full moon night, pull curtains of varying degrees, close eyes, and measure these discount rates simultaneously.
The Su Wen · Jin Gui Zhen Lun states, “There are eight winds in heaven, five in the meridians”—there are eight types of heavenly winds, but only five enter the human body. The Ling Shu and Shang Han Lun discuss “transmission,” referring to the transformation of qi as it enters the body. The ancients never believed that “what heaven gives” equals “what the person receives”; intermediaries were not seen as pipes but as workshops: what enters is no longer the original substance upon exit. The eyelid’s transmittance curve, particularly severe at 480 nanometers, represents the first process in this workshop—it is not merely weakening moonlight but transforming it into another kind of light.
Evidence and Limitations
Verified: Eyelid spectral transmittance is measurable and predictable; short-wave optical density is significantly higher than early estimates, inter-subject optical density differences range approximately 1.0 with poor correlation to skin pigmentation (PMID 21721832); blue light intervention through closed eyelids significantly delays DLMO and sleep onset (PMID 26985450); pupil size and light response can be continuously measured under closed-eye states (PMID 34745751, PMID 41694690); pupil constriction significantly alters the relationship between melanopic EDI and melatonin suppression (PMID 38348477); lens transmittance spectra and pupil size are factors influencing light-induced melatonin suppression (PMID 33772847), and lens density and transmittance spectra can be measured in vivo (PMID 33020575); melatonin suppression is primarily driven by melanopsin, can start as low as approximately 1.5 melanopic lux, and saturates at approximately 305 melanopic lux (PMID 30697806); suppression is extremely steep at low doses (reaching half-maximum inhibition with intensity exceeding ~1% of maximum brightness, PMID 10922269); ED50 varies by an order of magnitude between individuals (PMID 31138694); fully blind individuals still experience significant melatonin suppression with in vivo white light exposure (PMID 29625838); melanopic lux predicts suppression better than standard illuminance (PMID 28545361); parameters for light exposure reporting have established standards (PMID 31281903).
Partially Verified: The actual contribution of visible light transmittance from residential windows and curtains to circadian doses—existing glass transmittance data largely originates from other applications, lacking bedroom measurements weighted by melanopic; whether gender differences exist at low illuminance boundaries (PMID 39041348 shows significance above 400 lux but not in 10–200 lux, though statistical power in this range is inherently limited); whether the pupil is still modulated under closed-eye states to a degree sufficient to alter melanopic dose.
To Be Verified: Whether the actual retinal melanopic EDI after four serial stages (“windows + curtains + eyelids + lens”) from outdoor moonlight of 0.1–0.3 lux falls below 1.5 melanopic lux—no end-to-end measurement exists; variance decomposition of discount rates at each level (which contributes most) is unknown; whether morphological individual differences such as eyelid thickness can be predicted by externally measurable indicators.
Related Reading
- 0.3 Lux Yields Two Doses for Two Individuals: How Dark Adaptation Duration Re-calibrates Melatonin Suppression Threshold
- Moonlight Has Only 0.3 Lux: Can Moonlight Suppress Melatonin?
- Light Sensitivity Polygenic Score × Lunar Phase Sleep Interaction
- Research Pool: Spatiotemporal Frequency Topology
Evidence Level: B
References: PMID 21721832, PMID 34745751, PMID 41694690, PMID 26985450, PMID 38348477, PMID 33772847, PMID 33020575, PMID 18757473, PMID 30697806, PMID 10922269, PMID 31138694, PMID 30803301, PMID 39041348, PMID 29625838, PMID 28545361, PMID 31281903, PMID 36433426
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