Is There a 'Divorce Season'? Not the Same One in Any Two Languages

A Wikipedia language edition is not a country. es.wikipedia is read by people in Spain, Mexico, Argentina, Colombia, and every other place Spanish is a first or second language — a Northern Hemisphere country and several Southern Hemisphere ones, mixed into one traffic stream with no way to separate them after the fact. pt.wikipedia skews heavily Brazilian by volume, but it also serves Portugal, Angola, and Mozambique. id.wikipedia comes from a country that straddles the equator. None of the eight editions below stands in cleanly for "this one nation." Read every line as "readers of this language," not "citizens of this country" — that substitution is the central weakness of this piece, and it doesn't go away no matter how the numbers are sliced.

With that said, here is what eleven years of Wikipedia pageview data actually show: the "Divorce" article peaks in six different months across the eight language editions tracked here, from May in German to July in Japanese — but Indonesian, Turkish, and German all hit their single lowest month of the year in exactly the same place: July.

The Data, Edition by Edition !July deviation across eight Wikipedia language editions, sorted from Turkish at minus 30.1 percent to Japanese at plus 14.4 percent.

The source is the Wikimedia REST Pageviews API, monthly views per article, July 2015 through July 2026 (11 full years), user traffic only (the API excludes bots). Every figure below is a percentage deviation from that language edition's own monthly pageview baseline — meaning it's already corrected for the fact that Wikipedia as a whole gets more traffic in January than in July, regardless of topic. Full method further down.

| Wikipedia edition | Hemisphere tilt | Peak month | Trough month | Total pageviews, 2015–2026 | |---|---|---|---|---| | en (English) | northern-dominant | November (+11.8%) | June (−11.3%) | 3,745,559 | | es (Spanish) | mixed | November (+29.9%) | January (−20.7%) | 1,560,781 | | pt (Portuguese) | southern-dominant (Brazil) | October (+25.4%) | December (−15.6%) | 415,688 | | id (Indonesian) | equatorial | November (+27.7%) | July (−24.2%) | 258,390 | | tr (Turkish) | northern | December (+20.6%) | July (−30.1%) | 43,925 | | ja (Japanese) | northern | July (+14.4%) | March (−11.2%) | 551,002 | | ru (Russian) | northern | June (+126.1%) | February (−38.8%) | 15,923 | | de (German) | northern | May (+5.0%) | July (−5.0%) | 441,619 |

Three editions peak in November — English, Spanish, and Indonesian — despite being about as different a set of reading populations as this dataset contains: one Northern-dominant, one genuinely split across hemispheres, one straddling the equator. The other five peaks are scattered across five separate months: October (Portuguese), December (Turkish), July (Japanese), June (Russian), May (German). Six distinct peak months, eight editions. There's no single global "divorce season" in this data, and no simple hemisphere rule either — Japanese and German are both Northern-dominant editions, and they peak five months apart.

The Trough Is Shared More Than the Peak Is

The low point of the year clusters more than the high point does, though "clusters" is doing real work in that sentence — it isn't universal. Indonesian, Turkish, and German all bottom out in July. English's "Divorce" article troughs one month earlier, in June, and English is the only edition in this dataset with a second tracked article: "Breakup," which troughs in July and peaks in March — a different peak month than its own "Divorce" article, inside the same language and (mostly) the same population. Even within one language edition, two closely related topics don't move together.

Outside that July cluster, the other four editions trough somewhere else entirely: Spanish in January, Portuguese in December, Japanese in March, Russian in February. So the honest description is: half of the eight editions bottom out within one month of each other, in the Northern Hemisphere summer, and the other half don't. That's a real pattern, not a universal one, and this piece isn't going to reach for an explanation beyond what the numbers show — the data describes a shape, it doesn't tell you why the shape is there.

Small Editions, Noisy Months

Two of these editions carry a caveat the table doesn't show on its own. Russian's "Divorce" article totals just 15,923 views across 11 years — roughly 120 views a month on average — which is small enough that a single month's random variation can swing the percentage wildly. Its reported June peak of +126.1% is the largest deviation in the whole dataset by a wide margin, and it's exactly the kind of number a small sample produces. Turkish, at 43,925 total views (about 332/month), is the next-smallest and should be read with the same discount. English, Spanish, Japanese, Portuguese, and German all sit in the hundreds of thousands to millions of total views, where month-to-month swings are on firmer ground.

The Popular Belief: "Divorce Season Is January"

Every January, a version of the same story runs in the press: family-law firms report a spike in inquiries right after the holidays, sometimes branded "Divorce Monday" for the first Monday back at work in the new year. The implied claim — that January is when interest in divorce peaks — is worth checking against both the best available research on actual filings and against this dataset. For more details, see our guide on view stories anonymously. For more details, see our guide on how long Instagram stories last.

On actual filings, the relevant research is Julie Brines and Brian Serafini (University of Washington), who presented findings at the American Sociological Association's 2016 annual meeting based on all divorce filings in Washington State from 2001 to 2015 — a full 14 years of court records, not a survey sample. They found a biannual pattern, with filings peaking in March and August, not January, and argued winter and summer holidays function as a "domestic ritual" period during which people delay filing rather than rush into it.

This dataset doesn't measure filings, only Wikipedia reading interest, so it can't confirm or refute Brines and Serafini directly — that's a limitation worth stating rather than glossing over. But none of the eight language editions here peaks in January either. Spanish's January is in fact its single lowest month of the year, the opposite of a spike. The closest thing to a match with the Brines-and-Serafini March finding is English's "Breakup" article, which does peak in March — but English's "Divorce" article, the more directly comparable topic, peaks in November instead. One of two English-language articles landing near the Brines-and-Serafini month is a coincidence worth mentioning, not a confirmation worth leaning on.

The fair summary: the "January spike" belief isn't supported by search interest in any of the eight editions checked here, and the best actual-filings research points to March and August in one U.S. state, not January anywhere.

Method

Source: Wikimedia REST Pageviews API (`https://wikimedia.org/api/rest_v1/`), CC0-1.0 licensed. Monthly views per article, "all-access" traffic, user agents only — the API itself excludes identified bots. Period covered: July 2015 through July 2026, 11 full years for every edition listed.

Every percentage in the table is a controlled figure: each article's monthly view count is divided by that Wikipedia edition's total monthly pageview count for the same month, then compared to the article's own average share across all months. This removes the fact that Wikipedia overall gets more traffic in January than July — without that step, every topic on every language edition would show a January peak, because the whole site does. The dataset also includes the uncontrolled version of each figure for comparison.

To reproduce: query the Pageviews API for `per-article` monthly views on the relevant article title in each language, and separately for `aggregate` (project-totals) monthly views for the same wiki, then divide and index. The script used here is `scripts/research/collect-pageviews.py` in this site's repository.

This piece is part of a research series on storiesfly.com auditing commonly cited relationship statistics for sourcing. The raw datasets are public: relationship-seasonality.json, relationship-seasonality.csv, and relationship-vocabulary.json (a companion dataset tracking pageview trends for terms like "ghosting" and "limerence," not analyzed in this piece).

FAQ

Is there a universal "divorce season" that applies worldwide?

Not in this data. The eight Wikipedia language editions checked here peak in six different months — November (English, Spanish, Indonesian), October (Portuguese), December (Turkish), July (Japanese), June (Russian), and May (German). There's no single month, and no clean hemisphere rule either, since two Northern-dominant editions (Japanese and German) peak five months apart.

Do Wikipedia pageviews prove people are actually divorcing more in a given month?

No. Pageviews measure reading interest, not events — someone reading the "Divorce" article is not necessarily filing for one. That's a stated limitation of the underlying dataset, not something this piece is glossing over.

Is the "January divorce spike" a myth?

It isn't supported in this dataset — none of the eight editions peaks in January, and Spanish's January is its lowest month of the year. The best research on actual U.S. filings, Brines and Serafini's 2016 analysis of Washington State court records (2001–2015), found peaks in March and August instead. Neither source points to January.

Why do Russian and Turkish show such large swings?

Both have small total pageview counts over the 11-year window — 15,923 for Russian and 43,925 for Turkish, versus hundreds of thousands to millions for the other editions. Smaller totals mean individual months are more exposed to random variation, so their reported peaks and troughs deserve more skepticism than the larger editions.