Why Nine Innings?

More about 19th-century baseball from Richard Hershberger.

In my previous post, on the 1882 Troys and the 1883 New Yorks, I explored a question of baseball organizational history. Here I am taking a different tack, peering into a corner of baseball rules history: Why does a game last nine innings? This might seem a fundamental aspect of the game, an eternal verity brought down from Mount Sinai. In reality, it was a clever solution to a problem, a solution so successful that today we have forgotten the problem, and imagine the solution simply to be part of the natural order of things.

How do you know when the game is over? There are three ways this can be determined.

1. By time: this might be a game clock, but it can also be a pre-determined end time or whenever it gets too dark to continue play.
2. By score: the game ends when one side reaches a certain score.
3. By innings: the game ends after a certain number of innings have been played.

Pre-modern versions of baseball used all three approaches in endless (and sometimes deeply mysterious) variants.

Of interest here is the version played by the Knickerbocker Club of New York and codified in 1845. This version took the second approach, ending once one side had reached 21 runs (provided that both sides had batted the same number of innings). At least this was the idea. In practice, darkness could end the game prematurely. Baseball never had a tradition of carrying a game over from one day to the next, so in reality it used a mixture of score and time to determine when the game was over.

This worked well enough for ten years or so. This was because the very early clubs weren’t very competitive. By this I don’t mean that they weren’t very good (though they weren’t) but rather that competition wasn’t the point. They existed to give young urban professionals and merchants with sedentary occupations a medium to take their exercise together in a socially congenial context. The vast majority of games were within a club, with the two sides picked for that day. Competition was the snake in the garden. Boys being boys, it was inevitable that if there were two clubs in one city, they would agree on a date, each pick their best team, and go at it. It was all very sociable and fraternal, culminating in a big dinner, but they also knew who won. Now jump forward to the mid-1850s. Baseball is a bona fide fad in New York. Clubs are popping up all over. They are challenging each other to matches, and they are playing to win. This included looking for any edge. The rules which had been fine for the old style of play turned out to have some holes that needed plugging.

The one that matters here was the draw. This is not a matter of playing to win, but of playing to not lose. This requires explanation. In present day American English a tie and a draw are the same thing. This has not always been true, and it isn’t true today in British English. Here we turn to baseball’s older cousin, cricket. A cricket match (and I am writing here of cricket in its traditional form, not a modern limited overs format) can in principle end in a tie, meaning the two sides have the same number of runs, but this is unlikely. A draw is another matter. A full cricket match consists of two innings. These might stretch over a number of days, but a match is not open-ended. A modern international test match is limited to five days. So what happens if the two innings are not complete at the end of those five days? This is a draw. It is like a tie in that neither team wins or loses, but it is unlike a tie in that the number of runs scored is irrelevant.

What this means in practice is that if you are the captain of a side, and the awful realization has crept upon you that your side is not going to win this one, you can play for a draw. In American sports if we say that one team is playing to not lose, we mean this to describe timid play, with the implication that the feared loss will be the result. In cricket, playing for a draw is literally playing not to lose, and it is a perfectly respectable strategy.

The technique is, to put it bluntly, to stall. This works within the rules because the stalling side still has to play the game. A batsman can be dismissed (i.e. put out) in several ways, three of them relevant here: He can be bowled, meaning the bowled ball strikes the wicket (the three vertical “stumps” with two cross pieces, the “bails,” laid atop them) with enough force to knock one or both bails down. This very roughly corresponds to a strikeout in baseball. He can be run out, meaning the wicket is knocked down by the ball while the batsman is running, and is on the wrong side of a line called the “crease.” This corresponds to throwing a runner out. Finally, he can be caught out, which exactly corresponds to being caught out on a fly ball. A key difference between cricket and baseball is that in cricket the batsman is not required to run on a ground ball. So when he hits a grounder, he judges whether or not it is safe to run. Putting these together, a cricketer playing for a draw bats very conservatively, defending his wicket but hitting the ball into the ground. He won’t score many runs this way, but at this point he isn’t trying to score runs. He is trying to kill time. But he still has to hit the ball, lest he be bowled, and he might put the ball in the air by mistake. So he is still playing the game.

Baseball of the mid-1850s was different. In particular, it did not have the called strike. The swinging strike was an established feature, but the concept had not yet been extended to the called strike. There was no penalty for not swinging at a pitch, regardless of its location. This had not been a problem previously. It was understood that the competition and the fun of the game lay in the fielding and throwing and running. Pitching and batting were simply how the ball was put in play. But when a team went into stall mode, there was nothing to impel them to swing at the ball, and the game ground to a halt. Winning, or at least not losing, had become more important than playing a fun or interesting game. The result was that fully one fifth of the games played in 1856 ended in draws. This was a problem (one of several) that needed fixing.

The Knickerbockers issued a call for a convention to be held in early 1857 to update the rules. They proposed a draft version, which included a rather brilliant solution to the drawn game problem. It created two distinct end points, with a distinction between a complete game and an official game: a larger number of innings for a complete game, and a smaller number for an official game. Once five innings were played, the game would be official regardless of whether or not they got the rest of the game in.

The idea was that by the time the trailing team foresaw an inevitable loss in a complete game, it would be too late to stall out the official game. It might be obvious by the sixth inning how the game was going, but there would be no reason to stall out the game at that point. Five innings for an official game today seems like an afterthought, but it actually was a key to making the game work. The new rule did not completely eliminate stalling, but it did marginalize the practice.

So far I have actually answered why the game is official at five innings. But why five, and why nine for a complete game. Why, for that matter, go by innings? Why not make a game complete at 21 runs and official at 11? Here things get murkier.

Predictable game length is a highly desirable trait in any sport. This is the great advantage of ending the game based on a clock. Matters get muddied when you adopt a game clock which can be stopped and started, rather than just agreeing to stop play at five o’clock, but you still have a pretty good idea how much of your day this is going to take. Ending a game based on score works well if the rate of scoring is predictable, as in volleyball. The rate of scoring in baseball is wildly unpredictable, however, making scoring a terrible basis for ending the game. The length of innings are based on the rate of getting outs, not of scoring, and this tends to be more nearly constant. Why not just go by time? We don’t know, but it is suggestive that cricket ends by innings. Cricket was far more prestigious than baseball at the time, and several of the Knickerbockers’ proposed rule changes would make baseball more like cricket. So that may well be the explanation of why they went by innings.

But why nine innings? Here we have two mysteries. The Knickerbockers’ draft rules made seven innings a complete game, with five for an official game. An amendment making nine innings a complete game was offered from the floor during final session of the convention. To further muddy matters, the amendment was offered by one of the Knickerbocker delegates, Louis Wadsworth. Wadsworth seems to have been a contentious fellow, and this looks like an internal club debate that went public. Sadly, there is no record of the arguments made.

The thing is, even seven innings is odd. Games the previous season had ranged from two to twelve innings, with an average of four. Setting the complete game at seven, much less nine, innings would seem a recipe for games to drag on until called for darkness. But it worked out. Partly this was because fielding was improving rapidly, leading to quicker innings. We don’t have the data for length of game in 1856, so we don’t have a direct record of how long an inning took, but they presumably had a pretty idea how long an inning lasted when neither side was stalling. They did the math, and concluded that seven innings would be about right. Perhaps Wadsworth was a visionary and realized that fielding would continue to improve, and so argued for nine innings. We can only speculate.

The new rule was spectacularly successful. It remains to this day with only a few tweaks. It compares very favorably to many other rule changes that set off cascades of unintended consequences followed by further rule changes, and it seems weird to consider any other way.