Every unhappy family may be unhappy in its own way, as Leo Tolstoy observed. But new research on divorce filings suggests at least one way that many unhappy families are alike.

As it turns out, there's a consistent seasonality to divorce filings, according to research recently presented by Julie Brines and Brian Serafini of the University of Washington. They analyzed nearly every divorce filing in Washington state between 2001 and 2015.

They found that divorce filings "consistently peaked in March and August, the periods following winter and summer holidays." After August, the likelihood of divorce plummeted toward its nadir in December.

In a statement, Brines and Serafini said that their findings may reflect the calendar of "domestic rituals" that govern much of contemporary American family life. Winter and summer holidays are "culturally sacred" times, in their view — nobody wants to file for divorce on Christmas, or on the Fourth of July.

Holidays may be a time of increased stressors — high expectations of happiness, interactions with extended family members — that contribute to the dissolution of some marriages over those periods. So holidays may be a time of pent-up demand for divorce that gets released in the subsequent months as spouses file their paperwork to split up.

"People tend to face the holidays with rising expectations, despite what disappointments they might have had in years past," Brines said. "They’re very symbolically charged moments in time for the culture."

Brines and Serafini found that the pattern held true even after correcting for other seasonal factors that could put stress on a marriage and influence divorce rates, such as seasons of high unemployment. Their preliminary analysis of divorce filings in four other states shows similar patterns.

The contours of their findings generally follow some other research into the seasonality of happiness and well-being. For instance, a 2014 analysis of Google searches by economist Seth Stephens-Davidowitz found that searches for "divorce" fell off a cliff in December, only to rise dramatically in subsequent months.

And a Wonkblog analysis of Google searches related to negative mood -- "depression," "anxiety," "stress," etc. -- similarly showed two seasonal peaks in affective misery, one in the spring and the other in late fall.

One potential confounder in Brines and Serafini's divorce filing data is that there may typically be a lag of weeks or even months between a couple's decision to get divorced and the filing of paperwork for that divorce. For instance, a couple could decide to break things off in November, but not file the paperwork until January.

But that type of thing is considerably harder to track and likely varies a great deal among couples. Every unhappy family is unhappy in its own way, after all.