Does menstrual synchrony really exist?

Dear Cecil: A while back I read an article about how sperm may be an antidepressant for people who ingest it in one way or another. But that’s not my question. What interests me is that a few months later, I read an article about the scientist who proposed that theory. He said what started him down that line of thinking was that he read a study about how cohabiting lesbians’ menstrual cycles don’t sync up. He theorized it was because of their lack of sperm ingestion. Last I heard (but I have been out of Catholic school for over 17 years), nuns in a convent don’t ingest sperm, yet their menstrual cycles sync up. So what’s the deal with cohabiting lesbians? Gordon, Brooklyn, NY

Cecil replies:

Whoa. So many opportunities here for … well, I was going to say “tasteless remarks,” but in discussions like this even the most innocuous comment can get you into trouble. So never mind that — and while you’re at it, forget about cohabiting lesbians, sperm ingestion, and nuns. The real news is that menstrual synchrony, first proposed in the world of science more than 30 years ago and now widely accepted as fact, may be nothing but a misperception.

The whole thing began in 1971 when Martha McClintock, then a graduate student in psychology at Harvard (now a professor at the University of Chicago), published a paper in the scientific journal Nature suggesting that the menstrual cycles of women who spend a lot of time together tend to synchronize over time. She studied 135 students in an all-female dorm and found a “significant increase in synchronization” in the onset of menses among close friends and roommates. She speculated that the cause might be pheromones, hormonal cues acting through the sense of smell.

As scientific reports go, McClintock’s was a bombshell, catching the attention of both specialists and ordinary folk. Several other investigators claimed to have replicated her findings, and soon heavy thinkers were constructing impressive theoretical edifices about primate mating patterns and such, all based on menstrual synchrony. In 1998, McClintock coauthored another Nature article reporting on experiments in which cotton pads soaked in donors’ armpit sweat were wiped on the upper lips of recipients. She alleged that this process changed the length of the recipients’ cycles, calling the effect “definitive evidence of human pheromones.”

Did that settle matters? Nope. Other researchers suggested there was less going on than met the eye. Among the objections:

Apparent clustering of menstrual onsets doesn’t necessarily mean anything. Assuming an average cycle of 28 days, the maximum time between two women’s onsets is 14 days. Since the minimum is zero, the average difference — what you’d expect purely by chance — is seven days, and half the time would be less. (In 1971 McClintock said she’d observed a decline in the average difference from seven days to five.) What’s more, women recording their onsets after the fact often misremember or are influenced by the recollections of their friends, skewing the data.
Menstrual synchrony in any meaningful sense is impossible when the women have cycles of different lengths. (Cycle length varies considerably among women not using the pill.) Though a woman with a 27-day cycle might initially have her onset on the same day as a woman with a 29-day cycle, the next month she’d be two days earlier, the month after that four days, and so on. No one has shown that supposedly synchronized women have cycles of the same length — or that their cycles, if of different lengths at first, diverge less than they should over time.
Methodological errors can easily bias a data set to show menstrual synchrony where none exists. To demonstrate one common problem: Suppose a study starts on October 1. Subject A, with a 28-day cycle, has an onset on September 27, another on October 25, and a third on November 22. Subject B, with a 30-day cycle, has an onset on October 5 and another on November 4. A naive investigator could report that these subjects were 20 days apart at the outset (October 25 vs October 5) and 18 days apart at their second onset (November 4 vs November 22). Ergo, the two are synchronizing. In fact, the two subjects were eight days apart to start with (September 27 vs October 5) and are diverging. Of course you can set up the numbers to arrive at the opposite conclusion; the point is that given the small samples commonly used in studies of menstrual synchrony, it’s easy to lead oneself astray. One skeptic (H.C. Wilson, 1992) has claimed that when you correct all the errors, including McClintock’s, the evidence for menstrual synchrony evaporates.

Currently the opposing camps are duking it out in the journals. (See the September 2002 Journal of Comparative Psychology for the latest irruption.) I won’t say the prosynchrony crowd is out of the game, but right now it’s third and long.

Cecil Adams

Send questions to Cecil via cecil@straightdope.com.