Lunar and Menstrual Phase Locking

In a selected population of 312 women, prospective menses records were maintained during the autumn of 1977. Women whose menstrual cycle duration approaches the cycle duration of the earth’s moon (29.5 days) tend to ovulate in the dark phase of the lunar period. The dark phase encompasses the half-cycle of the month from last quarter, through new moon, to first quarter. Women showing irregular menses also tended to ovulate during the dark phase of the lunar period.

Although common usage refers to the human menstrual cycle as typically 28 days, repeated studies consistently show the mean and median of sample menstrual cycle data to be 29.5 days. The lunar cycle is also a 29.5-day cycle. This coincidence of the 29.5-day lunar and menstrual rhythms has intrigued a number of workers and led to subsequent small-and large-sample studies in search of a relationship.

Large-sample studies have been disappointing. Both Gunn and associates,1 in 1937, and Pochobradsky,3 in 1974, as well as Arrhenius,6 in 1898, lumped together all the reported menses onsets of all their subjects regardless of the cycle length of the individual women. Then each cycle onset was recorded in relation to its phase of the lunar cycle, in an attempt to determine whether certain phases of the moon were associated with a disproportionately high incidence of menses onsets.

The only large-scale positive association reported was that of Arrhenius,6 who argued for a periodic lunar fluctuation in frequency of menses onsets. This was later negated by Gunn and associates,1 who contrasted Arrhenius’ use of retrospectively recorded last menstrual dates of parturient and third-trimester women with prospective menstrual records of 10,416 events. Gunn and associates1 expressed their disappointment in having to conclude that there was no valid justification for associating the date of menstruation or its rhythm with lunar phenomena. Pochobradsky5 also concluded that menstrual onsets were independent of lunar cycles.

Menaker and Menaker7 showed a statistically significant, although small, synodic (29.5 days) influence on human birth rates. Rates were highest during the lunar half-cycle beginning a day or two before the full moon. Osley and associates8 reported similar findings.

Two small-scale studies have shown lunar-menstrual relationships. In one report, a subject received nocturnal illumination from days 14 to 17 of her cycle; she changed from an irregular to a regular 29-day cycle.9 Subsequently, in replication, a group of subjects received a similar regimen while also supplying control (no nocturnal illumination) cycle data. Eight of 11 subjects showed a narrower range of cycle lengths under conditions of nocturnal illumination than when unmanipulated, although only one fourth of the experimental cycles achieved the 29-day cycle length.10 Thus, the results are suggestive, although less than definitive, that nocturnal light influences cycles.

In contrast to the human studies, a clear synodic (monthly) variation in activity rhythms of animals and unicellular forms has been demonstrated. Excellent reviews by Bunning11 and Brown and Park12 show repeated monthly rhythms in planaria activity, marine worm swarming patterns, fiddler crab locomotion and color change patterns, and hamster diurnal activity patterns. The demonstration in all these studies that the rhythm persists for one to three cycles in the laboratory under constant photic conditions led Bunning11 to conclude that there was an endogenous lunar period cycle-but one could equally well consider that the maintenance of the rhythm was due to exogenous influences other than photic, for example, gravitational, which is powerfully affected by the lunar positions, as may also be the case for atmospheric electrical variation. Schneider13 has reported evidence that gravitational fields can be sensed by organisms; and it is known that, at the new moon, sun and moon exert their additive maximum gravitational effects on the earth. This is greatest at the new moon and least at the full moon.

Accordingly, it seemed reasonable to investigate, once again, the possibility of a lunar effect upon the menstrual cycle by looking at the data in a new way.

In a 14-week, double-blind, prospective study during the fall of 1977, 312 University of Pennsylvania students kept a record of their menstrual cycles on specially provided preprinted calendars. All subjects were 19 to 22 years old, gynecologically mature (menstruating for at least 7 years), white, and nulliparous; and they were not using intrauterine contraceptive devices or taking oral contraceptives, either of which might have affected the rhythm of uterine bleeding

Lunar and Menstrual Phase Locking Figure 1: One menses onset of each lunar period cycler. The calendar shows autumn 1977 dates in relation to the lunar phase. The 68 lunar period cyclers )those who show a mean cycle lenght of 29.5+ 1 day) are described here. Each dot shows the first cycle onset of one subject. Only one menses date per subject is entered to avoid weighing the data in the direction of the trend.

Data collectors were 13 specially trained female undergraduate contemporaries of the subjects who recruited participants by knocking on randomly preassigned dormitory doors to solicit participation. Only one student per dormitory room was permitted to enter the study. At the end of the 14 weeks, each recruiter met with her subjects individually for a final interview, at which time all documents were sealed and mailed in anonymous form to the investigator’s laboratory.

Furthermore, each subject was contacted to assure continued record-keeping, at least 3 weeks during the study. This large-scale prospective study followed a pilot study conducted in the autumn of 1976. The pilot study monitored menstrual cycles of 127 women with similar constraints.

Since menstrual and lunar cycles are repetitive, a circular chart (the lunar clock) for recording data was devised to tabulate the repeating cyclic phenomena in the natural form of a clock. With an almanac used to secure exact dates, the new moon was placed at the bottom (0 degrees) and the full moon at the top (180 degrees), with the first and third quarters bisecting each 180-degree semicircle. The rest of the calendar dates as they related to the lunar cycle during the fall were then arrayed (see any of the figures).

The calendar was shaded to indicate the dark half-cycle of the lunar clock which encompassed the bottom half of the graph – last quarter, through new moon, to first quarter. The top un-shaded half of the graph, then, constituted the light half-cycle of the month – the first quarter, through full moon, to last quarter. Each menstrual onset was entered by placing a dot in the appropriate sector. Since the average lunar cycle is 29.5 days and the average menstrual cycle is also 29.5 days, the many conjectures of a lunar menstrual relationship, cited above, did make sense. But once the nature of a circular chart was appreciated, it seemed possible that the methods used in investigating had actually obscured whatever relationship was there, as the following considerations show.

In the use of such a circular calendar, entry of sequential menstrual cycles of other than 29.5 ± 1 days must yield data points which distribute themselves randomly about the circle. For example, the data entries of a woman who menstruates every 34 days will distribute completely about the 29.5-day circle over time, shifting around the clock from one cycle to the next. A single cycle of a number of women is subject to similar constraints. Therefore, the question of interest is: Where do the menses onsets of the “lunar period cyclers” (29.5 ± 1 mean cycle length) fall? Do these women tend to cycle during a certain phase of the lunar cycle and, therefore, assemble in one area of the lunar clock?

This question does not appear to have been asked before. Rather, in past studies, menses onsets of lunar period cyclers were charted together with menses onsets from longer and shorter cyclers. As the above explanation suggests, one would predict that data contamination with cyclers of other than approximately 29.5 days would hide any demonstration of a high incidence of menses onsets in one phase of the lunar cycle.

 

Of the 312 women, 68 were lunar period cyclers; and these subjects are charted in Fig. 1. Only the first recorded menstrual onset of the 14-week study is tabulated in order to avoid undue weighting of events in the direction of the trend. This conservative precaution is necessary since preselected 29.5 ± 1 day cyclers would, by definition, experience their next menses in the same sectors. Note that the first menses onset occurred between September 14 (the start of the study) and October 14 (30 days later).

Figure 1

Of these 68 women, 47 menstruated in the light half of the month. This yields a significantly higher proportion of events in the light half-cycle than a random distribution would predict (z=3.16, p<0.001).* see below The pilot study produced a similar trend. Of 127 women, 29 were lunar period cyclers; and 22 of these 29 menstruated in light half of the month (z=2.79, p<0.01) in 1976.

Since others have recorded all the menses onsets of large groups of women with different cycle lengths, this type of analysis was next performed. Accordingly, Fig. 2 shows all 772 menses onsets of a random sample of 248 women with symbolic notations to differentiate active from inactive sexual behavior.

Figure 2: All menses onsets of 248 women. The calendar shows autumn 1977 dates in relation to lunar phase. All 772 onsets of these 248 subjects entered.

These particular 248 women formed a subset (see Reference 4) who had also maintained sexual behavior data in prospective fashion. Consistent with the reports described, data contaminated with other than lunar period cyclers appear to array themselves more evenly about the clock. Since more than one cycle per person is entered, statistical evaluation would be biased and accordingly is not presented.

Finally, it is asked whether women with irregular menstrual cycles (an irregular cycler was defined either as one whose standard deviation about her mean cycle exceeded 8 days or as one with an obviously long cycle of >30 days) would also show a convergence of menses onsets during the light phase of the moon. They did, as shown in Fig. 3.

Figure 3: All menses onsets of each irregular cycler. The calendar shows autumn 1977 dates in relation to lunar phase. Each dot shows one menses onset date of an irregular cycler. All 110 onsets of the 53 women are entered.

There were 53 irregular cyclers among the entire 312 women. Some of these women showed only one menstruation during the fall, whereas others showed two, three, or four menstruations, to yield a total of 110 events for 53 women. Fig. 3 shows the parceling of these 110 menses; 64 events in the light half-cycle as contrasted with 46 onsets in the dark half (z=1.72, p <0.04).* see below In the pilot group, only 18 of the women displayed irregular menstrual cycles, and too few data were produced to demonstrate any trends.

---

Notes

*Under the assumption of the null hypothesis, in this binomial situation, one would expect one half of the entries (one half of 68) to fall in each half of the lunar cycle. Here 47 appeared. The standard deviation under the binomial distribution is computed as (npq)1/2 = (68 x 1/2 x 1/2)1/2 = 4.12. The statistic then is computed as follows
observed – expected 47-34
z = standard deviation = 4.12 = 3.16 (p<0.001).

* A possible problem with the test is that some of the events may be dependent since they derive from the same individual. A chi-square test showed this dependency was not significant (x2 = 0.51, p <0.23).

 

Comment

It has been shown that menstrual cycles similar in length to the lunar cycle, in these selected populations, tended to occur during the light half-cycle of the lunar period. Since 98% of the cycles of 29.5 ± 1 day in length are ovulatory, and ovulation occurs, on average, 15 days before menses,3 ovulation is occurring in the half of the lunar hemisection opposite from menstruation occurrence. Thus, ovulation is occurring in the new-moon part of the cycle and is coincident with the greatest gravitational pull on earth.

It is important to realize that the possibility of a lunar entrainment of the menstrual cycle is not proven from the data presented. Rather these findings should be evaluated from the following perspective: a basic unit of biologic time for human cyclicity may well be the lunar, 29.5 day cycle.7, 14, 13

It is well known that lunar phases affect a variety of geophysical phenomena in nature, including magnetic and electric fields,6 as well as the tides. The fact that this same cycle length is the most common one in both human and infrahuman primate has been shown repeatedly.1,2,4,7,14

Furthermore, the physiologic individuality of biologic organisms11 suggests that each woman probably has some individual cycle length which forms the endogenous basis of her natural rhythm during the reproductive years.

A number of exogenous influences has been documented which alter the menstrual rhythm, and undoubtedly the list will be extended. To date, it has been shown that sexual behavior,4 social effects,2 as well as nutrition, seasonality, incarceration, and the stress of war,16 all contribute to the rhythm of the human cycle and subsequent fertility.

The demonstration that both lunar period cyclers (29.5±1 day) and irregular cyclers show a preponderance for menses onsets in the light half-cycle of the month suggests that a lunar effect exists. Twenty-two percent of the women of this 1977 study had the lunar period cycle. In addition, 17% of the women were irregular cyclers.

Thus, close to 40% of a random sample of women in one age group were tested and demonstrated an association between an exogenous influence (the lunar phase) and menses onsets. This suggests that a real phenomenon exists.

The remaining 60% of this sample have not been eliminated from a possible lunar effect. Rather, this form of short-term circular analysis is not suited to testing for a long-term effect. For the relatively regular (SD< 8 days) nonlunar period cycler, appropriate analysis would require long-term individual charts designed to evaluate movement toward the light half-cycle phase. An appropriate method might be similar to that of Waldron17 in which locust wing light movements were entrained to flashing-light stroboscopic stimulation. Analytic techniques were developed which showed that the closer the strobe rhythm was to the natural wing-beat rhythm, the more effective a source of entrainment the light became. Bunning has summarized similar types of analytic techniques, and these would be ideal for long-term cycle records within an individual. The short-term data sample (cross-sectional) of this report is not suited to these methods of evaluation, but such a study would be desirable for future investigation with the use of long-term menstrual records of individuals.

Thus, it has been shown that menstrual cycles similar in length to the lunar cycle occurred predominantly in the light half-cycle of the lunar period during the autumns of 1977 and 1976. That these women lived in city dwellings would seem to preclude any direct photic effect of the additional light in the sky. As Brown and Park12 have suggested, geophysical effects which result from lunar changes still affect a variety of organisms that have been removed from the direct photic influence by placement in a laboratory under constant light conditions. Entrainment of the circadian rhythm to an approximate lunar day (24.87 hours) has been show in humans who were removed from the light cycle but exposed to naturally occurring electromagnetic fields.14

Therefore, it might be considered that a natural rhythm of electromagnetic radiation has its origin in the lunar cycle, and may be reflected in phase-locking of the human menstrual cycle.

References:

1. Dunn, D.L., Jenkin, P.M., and Gunn, A.L.: Menstrual periodicity: Statistical observations on a large sample of normal cases, J. Obstet. Gynecol. Br. Emp. 44:839, 1937.

2. McClintock, M.: Menstrual synchrony and suppression, Nature 229:244, 1971.

3. Vollman, R.F.: The Menstrual Cycle, Vol. 7 in Major Problems in Obstetrics and Gynecology. Philadelphia, 1977. W.B. Saunders.

4. Cutler, W.B., Garcia, C.R., and Krieger, A.M.: Sexual behavior frequency and menstrual cycle length in mature premenopausal women, Psychoneuroendocrinology 4:297, 1980.

5. Pochobradsky, J.: Independence of human menstruation on lunar phases and days of the week, AM. J. OBSTET. GYNECOL. 118:1136, 1974.

6. Arrhenius, S.: The effect of constant influences upon physiological relationships, Skandiana Arch. Physiol. 8:367, 1898.

7. Menaker, W., and Menaker, A.: Lunar periodicity in human reproduction: A likely unit of biological time. AM. J. OBSTET. GYNECOL. 117:413, 1973.

8. Osley, M., Summerville, D., and Borst. L.H.: Natality and the moon, AM. J. OBSTET. GYNECOL. 117:413, 1973.

9. Dewan, E.M., and Rock, J.: Phase locking of the human menstrual cycle by periodic stimulation. Biophys. J. 9:A207, 1969.

10. Dewan, E.M., Menkin, M.F., and rock, J.: Effect of photic stimulation on the human menstrual cycle, Photochem, Photobiol. 27:581, 1978.

11. Bunning, E.: The Physiological Clock, Berlin. 1964, Springer-Verlag.

12. Brown, F.A., Jr., and Park. Y.J.: Synodic monthly modulation of the diurnal rhythm of hamsters, Proc. Soc. Exp. Biol. Med. 125:712, 1967.

13. Schneider, F.: Experimental evidence for magnetic, electric and ultraoptic information, Biol. Abs. 60:960, 1975.

14. Wever, R.: In Persinger, M.A., editor: ELF and VLF Electromaghtic Field Effects, New York, 1974, Plenum Press.

15. Miles, L. E.M., Raynal, D. M., and Wilson, M.A.: Blind man living in normal society has circadian rhythm of 24.9 hours, Science 28:421, 1977.

16. Cutler, W.B., and Garcia, C.R.: The psychoneuroendocrinology of the ovulatory cycle of woman. A review, Psychoneuroendocrinology 5:190. In press.

17. Waldron, I.: The mechanism of coupling of the locust flight oscillator to oscillatory inputs. Z. Vergleich. Physiol. 57:331, 1968.

From the Department of Biology, University of Pennsylvania.
Received for publication May 18, 1979.
Revised December 21, 1979.
Accepted December 31, 1979.
Reprint requests: Dr. Winnifred Cutler, Department of Physiology, Stanford University, Stanford, California 94305