Petecash

Psychoneuroendocrinology 24 (1999) 449 – 459

Menstrual synchrony in a sample of working women Leonard Weller a,*, Aron Weller b, Hagit Koresh-Kamin a, Rivi Ben-Shoshan a a b

Department of Sociology, Bar-Ilan Uni6ersity, Ramat-Gan 52900, Israel Department of Psychology, Bar-Ilan Uni6ersity, Ramat-Gan 52900, Israel Received 2 February 1998; accepted 20 November 1998

Abstract Menstrual synchrony has been typically studied among women who live together: dormitory roommates or family members sharing a bedroom or living in the same house. The current study examined menstrual synchrony in 51 pairs of women working together under conditions optimally conducive to synchrony. They had been together for at least 1 year, shared a relatively small office, worked there all day full time and contact with other people during the day was minimal. Prospective records of three menstrual dates showed a significant degree of synchrony for each of the 3 months. Menstrual onsets of close friends tended to occur on the average within 3.5– 4.3 days of each other while onsets of co-workers who were not close friends were significantly more broadly ranged (7.7 – 9.0 days of each other). This is the first unequivocal demonstration of menstrual synchrony outside of the household. © 1999 Elsevier Science Ltd. All rights reserved. Keywords: Menstrual cycles; Menstrual synchrony; Menstruation; Women; Working women

1. Introduction Twenty-five years ago Martha McClintock (1971) published her pioneering paper on menstrual synchrony. The author demonstrated, for the first time, that the menstrual onset dates of college students who are roommates or close friends converged significantly over a period of 4 months. The research continued to * Corresponding author. Fax: +972-3-635-0995. E-mail address: [email protected] (L. Weller) 0306-4530/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 3 0 6 - 4 5 3 0 ( 9 8 ) 0 0 0 9 2 - 4

450

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

examine menstrual synchrony among women living together (and to smaller extent among close friends) with the bulk of the samples consisting of dormitory women. In addition to the studies affirming synchrony among roommates (e.g. Weller and Weller, 1993b,c), there are studies which did not find synchrony (e.g. Jarett, 1984; Trevathan et al., 1993), as well as studies which found synchrony only among roommates who are best friends (Goldman and Schneider, 1987; Weller et al., 1995). McClintock (1971) raised the question whether menstrual synchrony might be the result of common environmental influences, particularly food eaten in a common dining room, a similar life pattern, and common reported stress periods. She therefore randomly paired her subjects and tested for synchrony within the dormitory as a whole. Not finding a significant trend toward synchrony among the control group of randomly paired subjects, she concluded that common environmental influences were not the cause of menstrual synchrony. Since then other studies have also compared their results to control groups of random pairs (Graham and McGrew, 1980; Weller and Weller, 1993b,c). All these studies, like McClintock (1971), found significantly more synchrony among the research groups than among the random controls. These findings suggest that the phenomenon is mediated by sensory communication between the women (either olfactory, by pheromones or by other senses) and not by exposure to common environmental cues. The pheromonal explanation is particularly supported by results of a recent study which demonstrated that the axillary secretions of donors applied to female subjects affected the timing of ovulation and menstruation of the recipients (Stern and McClintock, 1998). Two sets of studies examined whether synchrony occurs among women not living together. In a major review article Graham (1991, p. 307) wrote ‘If olfactory communication underlies synchrony, then the effect might be enhanced in groups of women who are in close physical proximity in situations involving increased perspiration: for example, sports team members or sauna-users’. Following this line of thought, Weller and Weller (1995a) investigated whether synchrony occurred among women basketball teams, but did not find evidence for the phenomenon. They offered several explanations for their failure to find synchrony. The duration of social interaction among the players was relatively short ( 10 h a week); the women practiced in large halls where the air circulation may have dispersed the weakened airborne pheromonal signals; and the sweat generated during exercise (which serves a cooling function) may have diluted the pheromones in the ‘regular’ sweat whose function may be for olfactory communication. A limitation of studying women in dormitory rooms is that they may not live there together for a sufficiently long time to generate synchrony. This may very well explain at least some of the negative findings; the women in the samples for which synchrony was not found may not have been together for a sufficiently long time and/or may not have had sufficiently intensive contact with one another to generate synchrony. With the exception of one study (Weller et al., 1995), no article has attempted to report the number of hours that the women were together during the period of the study. If one discounts sleeping, then the number of hours two women

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

451

sharing a dormitory room actually spend together may have been relatively few. Therefore the case of women working together for a full 8/9 h day may offer an even better test of menstrual synchrony than that of women living in a dormitory. Matteo (1987) studied menstrual synchronization among 41 working women in five work groups: an all female university department (ten women), emergency room nurses (six women), recovery room nurses (seven women), a typing pool (eight women), and graduate students in the same department (ten women). For each work group, synchrony scores were reported for 3 months. Menstrual synchrony was found for four of the five groups. The findings of this study — if valid —are intriguing and significantly advance the area of menstrual synchrony, for synchrony would have been shown to exist among women who do not live together. The author unfortunately did not provide any information about the physical conditions under which these women worked. However, it would certainly seem that in some, if not in most of these groups, the women may not have shared a joint office space or may have spent a lot of time away from their offices. Thus the finding of synchrony among these groups of women suggests that menstrual synchrony is indeed a strong effect. The Matteo (1987) findings have been seriously criticized by Wilson (1993). Wilson (1993) contends that Matteo (1987) did not actually conduct a test for synchrony in her work groups, that her technique for determining the groups degree of synchrony is flawed and that she ‘does not demonstrate menstrual synchrony in her four groups’ (p. 537). We would add another criticism. Applying the sign test to these five groups shows that when four groups are significant and one is not, the statistical probability is .19. That is, menstrual synchrony was not demonstrated in this study. The second study of menstrual synchrony in the workplace (Weller and Weller, 1995b) consisted of two samples. The first sample was comprised of women soldiers who worked mostly as secretaries in offices on two army bases. The subjects had worked together for a mean of 11.8 months and their mean age was 19.5 years. The second sample consisted of 23 pairs of women, each pair working in different civilian office buildings. The subjects had been working together for a mean of 36 months and their mean age was 37.1 years. The data are based on 2 months of retrospective reporting of menstrual dates for the army sample and on 1 month’s retrospective reporting and 1 months prospective reporting for the civilian sample. Neither of the two samples demonstrated a significant degree of synchrony. However, women soldiers who had been working together for a longer period of time were significantly more synchronous than those who had been working together for a shorter period. Also, women soldiers who were close friends were significantly more synchronous than women who were not close friends. Neither degree of mutual activity nor degree of friendship was related to synchrony in the civilian sample. The purpose of this paper is to examine menstrual synchrony under near-optimal conditions conducive to menstrual synchrony, outside the household. Three prospective menstrual onset dates were collected from pairs of women who shared, for a complete day, a small office and had minimal contact with other workers or with clients during the workday.

452

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

1.1. Subjects The criteria for being included in the study were: (1) two women between the ages of 17 and 40 working in an office; (2) for a complete day (8–9 h); (3) for a complete work week; (4) for at least 1 year; (5) with no man present; (6) minimal contact with other workers or with clients during the workday; and (7) in an office not larger than 12 m2. One of the researchers knew of an office building which housed offices that met these criteria and gained permission to request the women to participate in the study. Another researcher sought out such offices, some of which she was aware of, but mostly by asking friends and acquaintances. In some cases, the friend who knew at least one woman in the office asked her if she would be willing to participate. In other instances the researcher simply knocked on the door, explained the study (the same explanation written on the questionnaires cover page, see below), and asked the women to participate. The setting consisted of rooms (offices) in which two women worked. A total of 19 of these offices were located in one building. The other 32 rooms were located in various office buildings located in the Tel-Aviv area. We located 181 offices which met the above criteria. In 84 of these 181 offices, one or both women refused to participate (46% participation rate). In 42 of the remaining 97 offices, both women completed the questionnaire but one or both of them did not record all 3 months of menstrual data (43% attrition rate). From the final sample of 55 offices (total participation rate 30%), the four couples in which both women used oral contraceptives were omitted from the analyses. Thus, the final sample size was 51 couples.

1.2. Instruments The cover page of the questionnaire explained that the major purpose of the research was the determination of menstrual symptoms, an extension of research performed previously in the US. Confidentiality was assured. The questionnaire, which consisted primarily of closed-ended questions, included: demographic data, questions on friendship, activities, menstruation and related issues. Questions on menstruation included: age of menarche, dates of the last and next to last onset date, average duration of menstruation, usual length of time between periods, regularity of the period, type of sanitary method (tampon, sanitary napkins), and use of contraceptives. The question of regularity was: ‘Do you have a regular period— yes, no?’ This is the same question used in virtually all studies which have assessed womens self report on regularity (Weller and Weller, 1998b). Also included was a 21-item menstrual symptom questionnaire (Chesney and Tasto, 1975), in which women were asked to indicate which symptoms they suffer from during their menstrual cycle (e.g. backache, cramps, fatigue). The friendship questions were: What is the quality of your relationship with your coworker?; How close do you feel to her? To what extent are you friends? For each question the subject responded on a 4-point scale, ranging from poor/not close

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

453

relationship to a very good/very close relationship. The seven questions on joint activity were: To what extent do you and she eat together?; study together?; wear each others clothes?; buy together?; go out together?; meet together during weekends?; do most things together? For each question, the subject responded on a 4-point scale, ranging from ‘no shared activities’ to ‘many shared activities’. Chronbachs a-coefficient of reliability for the three friendship questions was .90 and for the seven activity questions was .86. For each subject a mean friendship score and a mean activity score was computed.

1.3. Data analysis To determine whether synchrony exists, we examined the data for each of the three successive pairwise onsets. We followed the recommendation of a recent methodological article on menstrual synchrony (Weller and Weller, 1997b) that for women who are together for a considerable amount of time, menstrual onsets should be compared for each of three onsets. Menstrual onset differences were calculated according to McClintock (1971) and Wilson (1992, pp. 571 – 572) guidelines. Accordingly, the first recorded onset date of subject A is compared to the first and second recorded onset dates of subject B, and then the first recorded onset date of subject B is compared to the second recorded onset date of subject A. The least of these three values is the correct absolute onset difference between the subjects onset dates. For example, if the onset dates of subject A are June 1 and June 29 and the onset dates of subject B are June 5 and July 6, then the correct absolute onset difference between the subjects’ initial onset dates is 4 days, the least of the three comparisons. However, if the onset dates of subject A are June 12 and July 13 and the onset dates of subject B are June 30 and July 29, then the correct absolute onset difference is 13 days (June 30–July 13). The procedure utilized for calculating onset differences (described in the next paragraph) frequently results in a reduction of cases in the last comparison of menstrual onsets. In the second example, subject A’s second (July 13) onset becomes her first onset for purposes of computing the onset difference, and her first recorded onset of June 12 is discarded. Now subject A’s third recorded onset, say August 12, becomes her second onset date for computing purposes. Since there are only three recorded onset dates, subject A does not have a fourth recorded onset (third computation onset) to compare to subject B’s third recorded onset. Accordingly in the current sample, N = 51, 51, 45 for the three cycles, respectively. The smallest absolute intracouple difference could range from 0 (both women have the same onset) to 14 days, assuming both have a 28 day-cycle. For example, if one woman reported her onset days on July 1 and July 29, and the second woman on July 15, the absolute difference would be 14 days. Had the second woman reported her onset day on either July 12 or July 18, the absolute difference would be 11 days. With a mean cycle of 32 days, the range would be 0 to 16 days. Cycle length was calculated by counting the number of days between two onsets. Because each subject recorded three menstrual onset dates, two cycle lengths could be calculated. The mean pair menstrual cycle length is the mean duration of the

454

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

menstrual cycles of a pair of subjects. The expected mean onset difference (cutpoint) is one fourth the length of the mean pair cycle length. Onset differences based on random onset occurrence would be expected to average at this cutpoint, as shown by a computer simulation (50 000 runs) based on randomly generated numbers. Thus, for couples with a mean cycle of 28, 29, 30, 31, or 32 days, the cutpoints (expected mean onset differences) would be 7, 7.25, 7.5, 7.75 and 8 days, respectively. Mean menstrual onset differences below these cutpoints indicate synchrony (as in Weller and Weller, 1995b, 1997a, 1998a). One-sample ‘t’ tests were employed to determine whether menstrual synchrony exists. Here, each couples menstrual onset difference was compared with its expected mean onset difference (cutpoint). This remainder, which could be negative, zero or positive (henceforth referred to as ‘synchrony index’) was the data for the ‘t’ tests. The ‘t’ test examined whether the group’s average index was significantly different from zero, a significantly negative index indicating synchrony. If the couple had a 4-day menstrual onset difference and one woman had a 30-day cycle and the other woman had a 32-day cycle, we compared 4 (observed) with 7.75 (the expected cutpoint). The index of −3.75 suggests synchrony. If two women had a 10-day menstrual onset difference and both had a 28-day cycle, then the difference between the 10 days ‘observed’ and the 7 days ‘expected’ would be 3, suggesting no synchrony. Since we can only calculate two cycle lengths (as we do not have data prior to the first recorded menstrual onsets), the average of the two calculated cycle lengths was used for comparison with the first onset date. Thus, the mean pair cycle length based on the first and second menses was used to compare with the second menstrual onset difference. The mean pair cycle length based on the second and third menses was used to compare with the third menstrual onset difference. The mean of the two mean pair cycle lengths was used to compare with the first onset difference. This was done for each couple. Finally, the impact of activity and degree of friendship was analyzed by means of multiple regression analyses. The dependent variable was the menstrual onset difference and the independent variables were the mean friendship score and the mean activity score for each subject. To assess the impact of menstrual factors on synchrony, a number of zero-order correlations were computed. The correlations were calculated between the menstrual onset difference of the subject and her age, age of menarche, flow duration, type of sanitary method and length of time between periods. For the categorical variable (sanitary method), point biserial correlations were calculated.

2. Results The subjects (N =102) had worked together for a mean of 31.9 months (SD= 29.7). The large standard deviation represents the large range in the numbers of months the women worked together, from 12 to 260 months. Their mean age was 27.9 years (SD= 4.00). The mean age difference of the woman in each pair (N= 51) was 5.20 years (SD= 6.03). The mean cycle length was 28.2 days (SD= 5.50).

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

455

Reported age at menarche was 12.8 years (SD= 1.20). A total of 23% reported using oral contraceptives, and only 12% reported being irregular. In about twothirds of the cases, we were able to measure the size of each room, and in the other third we estimated the size. The rooms ranged from 5 to 12 m2. The mean size (and the mode) was about 10 m. The mean activity score was 1.99 (SD =.65), which indicates little joint activity between the women. The mean friendship score was 3.2 (SD= .64). On the friendship scale, 3 indicates good/close relationship and 4 indicates very good/very close relationship. Table 1 shows the results of the analysis of synchrony for the second and third successive onsets. The second row displays the index (viz. the real number difference between the pair’s onset difference and its cutpoint), averaged over the entire sample. A minus sign indicates synchrony. The results of the one-sample ‘t’ tests show the existence of synchrony for both onsets (pB.01, pB.02). The mean index for the first onset (for which cutpoint data were estimated, as described above) was − 2.42 (SD = 3.58; t = 4.63, pB .001, N= 51), providing additional support for synchrony. The number of couples who had negative indices for the first, second and third onsets, respectively, were 36, 36 and 30. The number of couples who had positive indices for the first, second and third onsets, respectively, were, 14, 15 and 15. We also examined whether there might be differences in synchrony between the women whose offices were in one building versus the women who worked in offices in various buildings. To this end, we undertook two additional comparisons. One, we examined whether menstrual synchrony was attained for each of the groups. Menstrual synchrony was found for each of these two groups for each of the three comparisons. Two, we examined whether there were significant differences in menstrual synchrony between these two groups of women. There was no significant difference between these two groups of women for each of the three menstrual onset comparisons. To determine how much synchrony exists, we adopted a formula used in previous studies (Weller and Weller, 1993a,b,c, 1997a). The index reported in the second row of Table 1 when divided by the expected mean onset difference indicates the shift toward synchrony. For the three onsets these are: 35, 25, 26%—an average of 28%. In the multiple regressions, the dependent variable was the menstrual onset difference and the independent variables were the mean activity score and the mean Table 1 Menstrual synchrony indices at two successive onsets for pairs of working womena

N Mean synchrony index SD t p a

Onset 2

Onset 3

Combined

51 −1.72 3.79 −3.25 B.01

45 −1.80 4.53 −2.66 B.02

96 −1.76 4.15 −4.15 B.001

See text for a description of synchrony index and a report on onset 1.

456

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

friendship score. The multiple regressions were calculated three times, for each of the three successive onsets. The multiple regressions were statistically significant for each of the three months (R 2 =.05, p = .02; R 2 = .04, p =.04; R 2 = .14, p= .001). For all three multiple regressions, the beta of the friendship variable was significant, but not the beta for the activity variable. We examined whether menstrual related and demographic items were related to menstrual synchrony by means of correlations, again three times for each of the three menstrual onsets. For each of three onsets, correlations were computed between menstrual onset differences and each of the following: cycle length, flow duration, age, age at menarche, and amount of time working together. None of the correlations were significant. Further analysis, by means of Mann–Whitney U-tests, were undertaken in which two groups were created: both workers said of each other that they were ‘very good friends’ or ‘good friends’ (N=39, 39 and 35, respectively, for each of the three analyses) or both workers said of each other that they were ‘not friends at all’ or ‘friends to a small degree’ (N = 10, 10 and 8, respectively, for each of the three analyses). There were only two couples where one woman said she was a good friend with the other woman, while the other woman said that they were not. For each of the three onsets, the within-pair onset differences were significantly smaller among good friends than among co-workers who did not report close friendship. The mean of all the absolute menstrual onset differences for the co-workers who were good friends versus those who did not report close friendship were 3.5 versus 8.4 days (first onset, p B .001), 3.5 versus 7.7 days (second onset, p= .001) and 4.3 versus 9.0 days (third month, p =.02). The pattern of the results shows that while mutually close friends appear to be synchronous, those who did not report close friendship had onset differences above the expected (i.e. random) mean for this sample.

3. Discussion This is the first study to have unequivocally demonstrated the existence of menstrual synchrony outside the living unit. The 28% shift toward synchrony found in this study of coworkers is similar to the estimated mean 33% shift among roommates (Weller and Weller, 1993a). Another important conclusion from our findings is that synchrony can be found also in women older (mean age 28 years) than the typical college-age subjects. We note that two reports of synchrony among mothers and daughters sleeping in different rooms but living in the same house (Weller and Weller, 1993b, 1997a) support our conclusions regarding exposure and age. The results of the current study are at variance with findings of two other samples of working women, both of which did not find synchrony, although in one of the two samples women who were closer friends were significantly more synchronous than women who were not (Weller and Weller, 1995b). There are several possibilities to resolve this inconsistency. First, the present study was

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

457

methodologically more precise than the former one. Data were based on three prospective dates, whereas in the previous study, data in one sample were based only on retrospective reporting, while the other sample was based on one retrospective date and one prospective date. More importantly, the findings of the previous samples were based on only one onset difference, whereas the findings of the current study are based on three onset differences. In a recent paper we have argued that due to the inherent variability in women’s menstrual cycles, as a reliability check, at least two and preferably three separate analyses of menstrual onset differences should be undertaken to ascertain the existence of synchrony (Weller and Weller, 1997b). Analyzing only one set of onset differences increases the chances of spuriously finding synchrony, or finding no synchrony. In addition, the present study controlled for the relatively small size of the room in which the women worked, whereas the previous study did not. Two suggestions have been offered as to why friendship should affect synchrony. One, friendship may result in physiological changes in the heart rate, blood pressure, galvanic skin response and free acid levels. Emotional involvement between close friends may lead to physiological changes that could, in turn, increase the subjects sensitivity to factors which alter the timings of their biological clocks (Jarett, 1984). Two, friendship is an indicator of time spent together, closeness and exposure. Specifically, increased friendship is associated with closer personal space (defined as the area individuals maintain around themselves into which others cannot intrude without causing discomfort). Friends maintain closer personal distances than nonfriends, and there is a positive linear relationship between frequency of contact and strength of liking (Hayduk, 1983; Bell et al., 1988). Furthermore, a parallel field of research, emotional contagion, has also shown friendship to affect synchronization in a variety of behaviors and emotions (Hatfield, et al., 1992). Another study reported findings regarding friendship and synchrony in a pattern similar to our results: Even though the sample as a whole was synchronous, roommates who were close friends were significantly more synchronous than roommates who were not (women in private residences, Weller and Weller, 1993b). Still other studies found that while the sample as a whole was not synchronous, women who were close friends were significantly more synchronous than women who were not (Goldman and Schneider, 1987; Weller and Weller, 1995b). Friendship, then, seems to be an important element in synchrony and warrants further study. We note that this summary on the friendship–synchrony relationship is based upon the results of t-tests comparing roommates who are close friends versus roommates who do not report close friendship (Goldman and Schneider, 1987; Weller and Weller, 1993b) and not upon multiple regression analyses (where friendship and activity were the predictor variables) which often produce a different pattern of results (Jarett, 1984; Matteo, 1987; Weller and Weller, 1993b). Since working in one office building may be regarded as controlling for environmental influences, the finding of no significant differences between the couples who worked in the same office building versus the women who worked in various office buildings suggests that common environmental influence is not an important factor

458

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

affecting synchrony. As such, the present findings are consistent with those based on random controls in showing that environmental influence does not underlie menstrual synchrony. In summary, the findings which demonstrate the existence of menstrual synchrony in the workplace, show that living together is not a requisite for the occurrence of menstrual synchrony. There may be other situations outside the household, in which women are together for a sufficiently long period of time for them to synchronize. While this study has found the existence of synchrony in a closed space, it is still possible that women who are in frequent contact with one another, but not in closed spaces, may synchronize. Finally, the current findings further show that the relationship among close friends may produce not only intimacy and warm mutual feelings but also a synchronization of biological rhythms.

Acknowledgements Orli Turgeman Goldshmidt’s assistance in the statistical analysis and her helpful comments on the manuscript are gratefully acknowledged. We thank Menucha Kutner for processing the manuscript.

References Bell, P.A., Kline, L.M., Barnard, W.A., 1988. Friendship and freedom of movement as moderators of sex differences in interpersonal distancing. J. Soc. Psychol. 128, 305 – 310. Chesney, M.A., Tasto, D.L., 1975. The development of a menstrual symptom questionnaire. Behav. Res. Ther. 13, 237–244. Goldman, S.E., Schneider, H.G., 1987. Menstrual synchrony: social and personality factors. J. Soc. Behav. Pers. 2, 243–250. Graham, C.A., 1991. Menstrual synchrony: an update and review. Hum. Nat. 2, 293 – 311. Graham, C.A., McGrew, W.C., 1980. Menstrual synchrony in female undergraduates living on a coeducational campus. Psychoneuroendocrinology 5, 245 – 252. Hatfield, E., Cacioppo, J.T., Rapson, R.L., 1992. Primitive emotional contagion. In: Clark, M.S. (Ed.), Review of Personality and Social Psychology, vol. 14. Sage, London, pp. 151 – 177. Hayduk, L.A., 1983. Personal space: where we stand. Psychol. Bull. 94, 293 – 335. Jarett, L.R., 1984. Psychological and biological influences on menstruation: synchrony, cycle length, and regularity. Psychoneuroendocrinology 9, 21– 28. Matteo, S., 1987. The effect of job stress and job interdependency on menstrual cycle length, regularity and synchrony. Psychoneuroendocrinology 23, 467 – 476. McClintock, M.K., 1971. Menstrual synchrony and suppression. Nature 229, 244 – 245. Stern, K., McClintock, M.K., 1998. Regulation of ovulation by human pheromones. Nature 392, 177 – 178. Trevathan, W., Burleson, M.H., Gregory, L., 1993. No evidence for menstrual synchrony in lesbian couples. Psychoneuroendocrinology 18, 425– 431. Weller, L., Weller, A., 1993a. Human menstrual synchrony: a critical assessment. Neurosci. Biobehav. Rev. 27, 427–439. Weller, L., Weller, A., 1993b. Menstrual synchrony between mothers and daughters and between roommates. Physiol. Behav. 53, 173–179.

L. Weller et al. / Psychoneuroendocrinology 24 (1999) 449–459

459

Weller, L., Weller, A., 1993c. Multiple influences of menstrual synchrony: Kibbutz roommates, their best friends, and their mothers. Am. J. Hum. Biol. 5, 173 – 179. Weller, A., Weller, L., 1995a. Examination of menstrual synchrony among women basketball players. Psychoneuroendocrinology 20, 613–622. Weller, A., Weller, L., 1995b. The impact of social interaction factors on menstrual synchrony in the workplace. Psychoneuroendocrinology 20, 21 – 31. Weller, A., Weller, L., 1997a. Menstrual synchrony under optimal conditions: Bedouin families. J. Comp. Psychol. 111, 143–151. Weller, L., Weller, A., 1997b. Menstrual variability and the measurement of menstrual synchrony. Psychoneuroendocrinology 22, 115–128. Weller, A., Weller, L., 1998a. Prolonged and very intensive contact may not be conducive to menstrual synchrony. Psychoneuroendocrinology 23, 19 – 32. Weller, A., Weller, L, 1998b. Assesement of menstrual regularity and irregularity using self-reports and objective criteria. J. Psychosom. Obstet. Gynecol. 19, 111 – 116. Weller, L., Weller, A., Avinir, O., 1995. Menstrual synchrony: only in roommates who are close friends? Physiol. Behav. 58, 883–889. Wilson, H.C., 1992. A critical review of menstrual synchrony research. Psychoneuroendocrinology 17, 565 – 591. Wilson, H.C., 1993. Reply to letter by Graham. Psychoneuroendocrinology 18, 535 – 539.

.