An incomplete history of the double-header

My last article at THT took a look at the effect double-headers have on the future performance of teams playing them. The range of my study was just a few recent years, 2009 to 2011, which left rather a lot of ground uncovered. My curiosity piqued by the work I had already done, I dug a little deeper into the history of double-headers, looking at how they have shaped and been shaped by baseball history.

This will not be a comprehensive view of the double-header by any means. I will be skimming through, doing a bit of sabermetrics here, a bit of sightseeing there, looking at what interests me and hoping to frame it so that it interests you. And I did not have to advance very far into the history of the game—1883 to be precise—to discover one of the most wonderful baseball facts I have ever stumbled upon.

The first twin-bill, and the ultimate travel day

The first major-league double-header goes almost as far back as the major leagues themselves. On Sept. 9, 1876, the Hartford Dark Blues hosted and swept two games against the Cincinnati Red Stockings, 14-4 and 8-4. You won’t see anybody doing game-day promotions celebrating this milestone, because there is no team that can claim to have participated. Neither of the teams would exist five years later.

The Cincinnati club played its last game in 1880, after years of bad play and the occasional skipping out on league dues. The current Cincinnati Reds are a different entity, born in the American Association and later transplanted to the NL. As for the Dark Blues, their home town apparently could not sustain them. The next season, in a move that will get Angels fans nodding sympathetically, they became the Hartfords of Brooklyn. The season after that, they were defunct.

Back in the 1870s, double-headers were an emergency measure for redeeming postponed games, not deliberately planned events. Several years of seeing the popularity of these occasions got the gears turning in schedule-makers’ heads, and in 1883 the double-header finally got the endorsement of a place on the official schedule. Their proximate motivation may have been something very old and familiar even in the 19th century: competition.

1882 saw the premiere of the American Association as a rival major league to the NL. Despite only a six-team roster and a bit of a runaway pennant win by the Cincinnati Red Stockings—yes, these are today’s Reds—the AA established itself quickly as a legitimate contender to the National League. (Cheaper admission, Sunday play, and beer sales in the stands were influential factors.) The first real inter-league war had begun, and 1883 featured what appears in modern eyes to be an escalation.

The two leagues scheduled multiple double-headers to be played on the two leading holidays of the summer season: Independence Day on July 4, and Decoration Day on May 30. (Decoration Day would officially be renamed Memorial Day in 1967, and move in 1971 from a fixed May 30 date to the last Monday in May.) While the National League played two standard double-headers on each day, the American Association, now up to eight teams, got creative. And did it ever.

The AA had two standard twin-bills on Decoration Day, St. Louis at Pittsburgh and Louisville at Baltimore. Early in the day, the Cincinnati Red Stockings played at the New York Metropolitans, and the Columbus Buckeyes visited the Philadelphia Athletics. When their games were over, the Red Stockings and Buckeyes boarded trains and headed to Philly and Gotham respectively, crossing each other’s paths somewhere in New Jersey. Cincinnati then played against the Athletics, while Columbus took on the Metropolitans.

You read that right. Two teams swapped opponents and cities about a hundred miles apart to complete simultaneous double-headers, 20 years before the airplane was invented and a good 50 years before stewardesses with drink carts. It was insane. It was glorious. It was, to that date, probably the biggest publicity stunt in the history of professional sports. And as much as the average reader here knows about baseball, you almost certainly never heard about that before.

Sanity would reassert itself on the Fourth of July, when the AA played a standard full slate of twin-bills. The league did repeat the experiment in 1884 and 1886, but with the hosts being stone’s-throw neighbors New York and Brooklyn, the charming lunatic element was gone. Bi-city double-headers would never become a staple of the schedule. That was okay: the regular version had caught on nicely.

Hangover effect, across the years

In my previous article, I looked at current double-header patterns, and found that there was a tendency for teams to perform less well in the month after playing two games in one day. This came during an era when double-headers are rare, fewer than one per month for every team over the course of a season. Long ago, though, teams played them far more often, at times averaging one a week or even more.

Analyzing the data for these years the same way, checking double-headers one by one, would be daunting for even the most gluttonous of number-crunchers, and I excuse myself from that duty. That doesn’t mean I won’t take a look, just that I’ll alter my method.

I did some in-depth surveying of sample years from the 20th century, to gather data pertinent to this question. I looked at 1912-1915 (including the Federal League), 1937-1941, and 1955-1959, for a nice broad perspective. I then tracked team records against the number of double-headers they played in a season above or below the average for that year and league. (This is a fairer measure than raw totals, which fluctuated year to year and decade to decade, and more pertinent since it compares teams to their direct competitors.)

The scatter chart, with trendline, is below, and it looks rather like I expected it to look.


(Here I make a bit of a dangerous foray into self-taught statistics. If my texts have led me astray, you may wish to avert your eyes for the next three paragraphs, and just consult the above chart.)

The correlation coefficient (“r”) for the data is -0.31615, a rather weak negative relationship (“negative” meaning the more double-headers, the lower the winning percentage). The coefficient of determination, r-squared, comes out almost precisely to 0.1, meaning one-tenth of the variation in values can be explained by correlation.

Using Recurrent Neural Networks to Predict Player Performance
Technology is rapidly advancing possibilities in decision-making.

There’s an eyeball check one can run on this. Looking at the chart, one sees that a change of three in the number of double-headers matches up to a change of -.200 in winning percentage. If one-tenth of this is correlation, that means playing three more double-headers correlates to a drop of .020 in a team’s record. Given the 154-game schedules of the teams surveyed, this works out to losing roughly one extra game for each extra double-header played.

Back in “The Double-Header Hangover Effect,” I found that for each double-header a team played, it lost an extra three-fourths of a game over the next month. Okay, that’s pretty close. There may be something I haven’t taken into account: hangover effects might last longer than 30 days, for instance, or there might be a multiplier effect when you play a lot of twin-bills. For a back-of-the-envelope check, though, it’s good enough to see that the new numbers are, ahem, in the ballpark.

It’s not necessarily true that correlation leads to mono-directional causation. Instead of playing more twin-bills leading to poorer records and nothing else, it could also be that second-division teams are more likely to schedule double-headers for themselves. I will take a look at this matter, and some related ones, later.

A factor those later matters hold in common is that they involve scheduled double-headers, as distinct from unscheduled ones happening due to postponements for rain, darkness, etc. It’s a distinction to keep in mind as I sketch out how double-headers developed in the 20th century.

What war was and wasn’t good for

Once the major leagues settled into the NL-AL arrangement, the number of double-headers on the schedule stayed relatively stable for quite a long time. It usually ranged between the mid-20s and the upper 30s, with little obvious pattern. A spike to 44 in 1905 came to nothing, dropping back to 31 the next season. Another uptick came during the challenge of the Federal League in 1914-15, the numbers rising both due to more teams and to more twin-bills per team. Once the Federals folded and the spur of competition was gone, numbers returned to normal.

Despite this stability, the number of overall double-headers was on the rise, from about 150 per season before World War I to 200 or more per year before World War II. The likely reason for this is the compression of the schedule, the 154 games squeezed into 10 days to two weeks fewer. Faster trains in a more robust rail system made this ambitious scheduling possible, but this forced owners’ hands when the rains came as they always had before. There were fewer open days in the schedule, so twin-bills had to carry the load.

When a big change in scheduled double-headers happened, it coincided with the greatest upheaval major league baseball ever endured: the Second World War. Just because it happened at that time, however, may not mean it happened as a result.

The immediate effect of the war on baseball was profound in many respects. Closest to our purpose, travel across America became restricted, with resources such as gasoline and rubber being strictly rationed for wartime uses. A classic motto of the era admonishing reckless travelers was “Is this trip really necessary?”

While travel by baseball teams was necessary if the game was to survive during the war, having quite so many trips was a different matter. The baseball powers, falling in line with wartime needs, produced schedules lowering the number of road trips and miles traveled. This meant longer series in each city visited, and this ended up meaning a large increase in double-headers.

From 71 twin-bills scheduled for 1941, baseball leaped to 130 in the 1942 season, then 198, 196 and 198 for the remaining three years of the war. Nearly one game in three during those later seasons was scheduled as part of a twin-bill, and that doesn’t count those coming about from postponements for rain and darkness. This reached a remarkable peak in 1945: of the 1,230 baseball games played to an official result (including ties), 608 of them, almost half, were part of a double-header. Eight of the 16 major-league teams, four per league, played over half their games in double-headers.

If you seek more evidence that double-headers drag down a team, you can look to the National League in 1945. Here are the standings, and the number of double-headers each team played:

1945 NL         Chi.   St.L.  Bkn.   Pitt.  N.Y.   Bos.   Cin.   Phila.
Record          98-56  95-59  87-67  82-72  78-74  67-85  61-93  46-108
Double-headers   34     35     32     40     36     46     41      44

The correlation isn’t perfect, but it’s awfully close. The three best teams had the fewest two-fers; the three worst teams had the most. It didn’t decide the pennant race—twin-bills were not the Phillies’ biggest problem—but it’s a little more fuel on the fire.

The American League that year was a different matter. Standings and double-header totals were a patternless jumble, with no evidence that excessive double-headers hurt a team. Except, perhaps, at the top of the standings, where they may have cost Washington the pennant.

The Washington Nationals/Senators (the team officially went by the former, but was generally called the latter) was scheduled for 27 twin-bills in 1945, ahead of the major league average of 24.75 but not massively so. Weather and other vicissitudes forced them to play far more, a total of 44. More than half of those came from Aug. 1 onward, starting that very day with an exhausting string of five in five days.

The Nats slogged through this obstacle course to finish with an 87-67 mark. The Detroit Tigers, who played eight fewer double-headers on the season, finished 88-65, a game and a half ahead. It does not strain the imagination to suggest that a less grueling schedule for Washington could have gained that team a game here, a game there, and a pennant—or at least a playoff—at the end.

(It also did not help Washington that the Nats had to complete this schedule a week earlier than the rest of the American League. Owner Clark Griffith rented out Griffith Stadium to the NFL’s Washington Redskins for September games, and had the schedule-makers crowd his team’s games out of the last days of the month. I suppose he instead could have arranged a long road trip to end the season at the regular time, but Griffith might have thought this would be too tiring for his players. This concludes our test of the Irony Alert System.)

Once the war was over, scheduled double-headers could have returned to their pre-war levels—but war has a way of changing everything, directly related or not. Double-headers on the schedule slipped from 198 to 149, then down a little more the next few years to a nadir of 111 in 1949. For most of the 1950s, that would be the floor, as the number fluctuated in the 110s and 120s. More than ever, at least in peacetime, twin-bills were a planned part of baseball.

But … even without the war, things had been heading that way. From a mere 28 scheduled two-for-ones in 1938, baseball had jumped to 48 in 1939, the highest total since the three-league 1915 season. It rested thereabouts for a year, then leaped to 71 in 1941.

There’s no obvious exogenous reason for the increase. America was finally staggering out of Depression around that time, but economic boom and bust hadn’t notably effected the numbers before. Lights began to spread across the majors in the late ’30s, but it was some time before the day-night double-header was invented, so that had no direct effect. Pennsylvania had repealed its “blue laws,” the last ban on Sunday baseball, in 1934, so if the increase was a reaction to that, it’s a reaction that took five years to register.

That might be the winning explanation, though, because almost the entire increase in scheduled double-headers for those years came in the form of Sunday twin-bills. Before the spike, double-headers were planned almost entirely around three big holidays: Decoration (now Memorial) Day, Independence Day and Labor Day. As late as 1937, there wasn’t a single Sunday double-header on the schedule. That changed in 1939, with 17 of them, a number that inched up to 20 in 1940, then zoomed to 41 in ’41.

Sunday twin-bills became a pragmatic staple of wartime baseball. Of the 722 double-headers on the schedule from 1942 to 1945, 586 of them, over 81 percent, were set for Sunday. Most of the remainder was for holidays.

Whatever impetus was nudging the baseball powers toward Sunday double-headers starting in the late 1930s combined with the wartime adjustment (or is four years long enough to have made it a tradition?) to keep the trend going strong. Throughout the 1950s, there were 1,157 double-headers scheduled in the majors; 897 of them fell on Sundays. The 77.5 percent figure is down only a smidgen from the war, and well ahead of the 58 percent mark for the last year before Pearl Harbor.

The frequency of Sunday double-headers created a shift in managerial strategies. It became a standard maneuver to have a “Sunday pitcher” on one’s team, somebody to start the extra Sunday game without disrupting the schedules of the team’s primary starters. There weren’t yet pitching rotations as we understand them today, but it still helped to have somebody to help ride a team over that predictable bump in the road.

Even with this sustained rise in scheduled double-headers, the total numbers began to recede from their circa-1940 peak. Total twin-bills in 1957 added up to 152, a number that would have fit well in the 1910s, and just 36 more than the scheduled number for that season. Groundskeeping methods were improving; night baseball allowed some flexibility in getting a game in on its appointed day (though for some time teams were barred from turning on the lights during games, meaning contests still got called for darkness in parks with lights).

Double-headers by accident had passed their peak. Double-headers by design were not too long in following.

The surprise double-header effect

I’ve left readers hanging for a while on whether bad teams might schedule more double-headers. I’ll stop that now.

In the 1955-1959 time period I examined, there is some indication of that effect. (I did not run this check for the 1937-41 time frame, since scheduled double-headers were almost uniform for much of that time.) The fewer twin-bills you scheduled, the higher you tended to finish in the standings. The chart, with trendline, is below.


That’s about a two twin-bill difference between first and last. Laggard teams may well have felt the need to spark interest in their teams with enticements like double-headers, along with more creative plays for publicity. Can you imagine the Yankees pulling the Eddie Gaedel stunt? No, and Bill Veeck didn’t do it when he had a winner in Cleveland, either.

If I wanted to repeat my previous trick here, I would observe that every added double-header corresponded to a drop of roughly 200 points of record (from first to middle-of-the-pack, or from the middle to dead last). That’s about one-third of the change when I was measuring all double-headers. If the correlation followed the same lines, and if we subtracted this difference as being due to poorer teams deliberately playing more twin-bills (big ifs, granted), that would leave us with each double-header causing a loss of an additional two-thirds of a game. Not claiming anything officially, but I still like the ballpark figures.

This leads to an interesting point I unearthed from the data: The number of scheduled twin-bills a team played had less of an effect on its overall performance than the number of double-headers in excess of the planned ones that it had to play. I examined the 1937-41 and 1955-59 periods, leaving out 1912-15 as the overall double-header effect was almost flat for that span. Dividing them up between planned and unplanned double-headers, we get these results:

image image

That’s a pretty substantial difference, and it is even sharper if one divides the data into the two separate periods. Planned double-headers from 1937-41 had no effect on team performance: The trendline was flat. (As so many of them were holiday events that all teams were doing, this is perhaps no surprise.) The unplanned ones in that period had a strong effect, stronger than for the 1955-59 unscheduled double-headers.

This makes some sense. Planning for a double-header in advance should let you smooth out the strains it causes: setting up your starters and bullpen, deciding who gets a game off, and so forth. Unplanned double-headers could well come with little warning, giving you no time to ameliorate those troubles. Worse, you could get a cluster of them in a short time span, multiplying the strains.

This happened fairly often in the old days. One notable example came in late August of 1938, when three teams got smacked with a spate of make-up games. The New York Yankees had to play six twin-bills between Aug. 21 and 27, following four between the 12th and 18th. The Chicago White Sox played two on six straight days from the 23rd to 28th, after four in the previous 11 days. The Philadelphia A’s had seven between Aug. 21 and Aug. 28, after having four in the prior two weeks.

The Pale Hose and White Elephants struggled, going 3-9 and 6-8 respectively, and both finishing the season deep in the second division. The Yankees, on the other hand, blew through their sextet of two-fers at 9-3, and cruised to their third straight pennant. If anyone was going to buck those odds, it was going to be the Yankees.

The passing of the double-header


There was no cataclysmic event that brought the era of the double-header to a close, just a gradual diminution. Scheduled twin-bills fell below 100 in 1958, the year the Dodgers and Giants decamped for California. This may have had an effect, but likelier it was merely emblematic of the era of air travel that made those little compressions of the schedule less necessary.

Even this wasn’t a steady effect, as the 1961-62 expansion brought a rebound not only in absolute numbers of planned twin-bills but their proportion to all games played. They reached a peak close to the plateau of the ’50s, and then began fading away again. The intrusion of the NFL into the sports niche that Sunday double-headers once occupied may have become a factor late, though they were still 60 percent or more of all planned twin-bills through most of the late ’60s.

The terminal dive began around 1980, with nothing to arrest it. 1991 would be the first year in over a century with no major league double-header on the official schedule.

There were still 29 double-headers played that year due to contingency rather than intent, but even that was a great decline from decades past. The arts of groundskeeping had improved apace, and dome after dome made the elements all but irrelevant to the games played on their fields. This was the upside of the endangered double-header: Fewer games failed to be played on their appointed days, meaning fewer disappointed fans. No tears need to be shed for that.

As for scheduled double-headers, perhaps their passing was inevitable. A periodic gimmick to bring out bigger crowds might make sense when you would be boosting single-admission attendance from 10,000 to 20,000 in a stadium that seated 35,000. Today, though, when fans far outnumber empty seats at so many games? Where’s the percentage? Foregone twin-bills are a sign, not of a sickness in the game, but of its financial health.

One could also point to the competitive disadvantage I’ve unearthed as a reason not to deliberately plan double-headers—except that the early days had the right solution. When everybody is playing two on some holiday, nobody gains or loses any systematic edge. If there were any grounds to call for the return of the twin-bill on the schedule, that would be it—except, as already stated, few parks need that now to fill up on the Fourth of July (well, Rogers Centre might).

The double-header remains, but only as a ghost of its former self. The influence it has on the game today is likewise attenuated. That means, for all the work I’ve been doing analyzing the adverse effect that playing them has on teams, it measures the past far more than the present or future. Double-headers are mainly an accident of the game now, save at the highest levels of the organization, where a franchise can build a dome or locate to a dry climate.

That’s a lot of work I put into these two articles, just to achieve a split. I can only hope it doesn’t affect the quality of my next work.

References & Resources
Retrosheet and Baseball-Reference provided vital statistics as usual. Special thanks to Paul Golba for some spreadsheet work on the side.

Print This Post
A writer for The Hardball Times, Shane has been writing about baseball and science fiction since 1997. His stories have been translated into French, Russian and Japanese, and he was nominated for the 2002 Hugo Award.
Sort by:   newest | oldest | most voted

In his history of the DH, Charlie Bevis also claims that the MLBPA bargained for fewer scheduled DHs during labor negotiations in the late 60s.  (Going off memory, hope i got that right!)  But as you say, it was probably bound to happen sooner or later as the result of increased attendance at games.

Andrew Reid
Andrew Reid

Very nice article, Shane- I found a few other split DH’s in the NL in 1883, ‘84, and ‘85- each year, Boston and Providence played them on Decoration Day- a morning game and an afternoon game. There were even a couple of tripleheaders in the 1890’s…

Shane Tourtellotte
Shane Tourtellotte

Scott:  Can’t speak for the late ‘60s, but recent CBAs have been quite restrictive on how often DHs can be played, so I can fully believe this was a labor sticking point for a long time.

Andrew:  Morning/afternoon DHs were around at least into the 1900s, if my memory serves.  Patriot Day in Boston may have seen some quite recently (they traditionally play a morning game on that date).  And there was a triple-header as late as 1920.  Hey, I said up front this was an incomplete history.


very good piece.
there is a strong correlation between scheduled double headers & lower than average attendance (the trailing five years for the home team).
strongly rational decision.
also, there were the factors of opening a park once for two games, rather than twice (reducing fixed costs/game); catering/provisioning for one crowd, rather than two; accommodation expenses; et al.
a question—what was the disposition of rained-out scheduled double headers?

Paul G.
Paul G.

To add to the complication, there were some three team doubleheaders in 1899, perhaps other seasons.  The home team would host visitor A in game one and the visitor B in game two.  These were scheduled.  And unpopular, apparently.

Cliff Blau
Cliff Blau

There is no such thing as a doubleheader played in two cities, or a morning-afternoon or afternoon-night doubleheader.  The definition of “doubleheader” is two games played in immediate succession.  See

Also, the history of the Cincinnati franchises is much more complex than you suggest.