All the Record Books Are Wrong
In the 1982 launch of The National Pastime, reissued by SABR (http://goo.gl/F8vukv), a previously unpublished writer named Frank J. Williams wrote a groundbreaking article. “A breakthrough,” I called it then: “the ‘Rosetta Stone’ for deciphering won-lost decisions of the dead-ball era.” In the years since, every record-keeping book, website, and organization has been guided by the principles Williams deduced from his awe-inspiring coverage of games from 1876 to 1919. Self-described as a bank accounting officer whose special interests were the Boston Braves, Red Sox, and Joe Wood, Williams carved out an enduring place in baseball literature with this one. For more about Frank, who remains an active researcher, see this, from 1999: (http://sabr.org/content/sabr-salute-frank-williams).
Pitchers were winning games long before 1876, but were not awarded victories because in an era of nearly universal complete games and restricted substitution, there was rarely a question about which pitcher to credit or debit. In 1885, as Frank Vaccaro wrote in “Origin of the Modern Pitching Win” (http://sabr.org/research/origin-modern-pitching-win), Henry Chadwick “published National League individual totals in the 1885 Spalding Guide. The practice did not catch on. The loss came later. On July 7, 1888, The Sporting News for the first time published win-loss records, and only then after the following disclaimer: ‘It seems to place the whole game upon the shoulders of the pitcher and I don’t believe it will ever become popular even with so learned a gentleman as Mr. Chadwick to father it. Certain it is that many an execrable pitcher game is won by heavy hitting at the right moment after the pitcher has done his best to lose it.'”
I heartily recommend Vaccaro’s article, published in the Baseball Research Journal in 2013. But Williams’ monumental work came thirty years earlier and should be read first. Here it is, online for the first time.
Ready, baseball experts? Here’s a quick quiz, consisting of only three questions, and–bending over backwards to be fair–I will permit you the use of any baseball encyclopedia or record book of your choosing. If you answer all three correctly, your prize is the next tour of duty as manager of the Yankees.
1. Who was the won-lost percentage leader in the American League in 1905 and what was his record?
2. How many games did Ralph Comstock win for the Boston Red Sox in 1915?
3. How many victories did Cy Young and Walter Johnson amass over their careers?
Question 1: The answer, according to both major encyclopedias–Macmillan and Grosset & Dunlap (commonly referred to as Neft-Cohen)–is Andy Coakley of the Philadelphia A’s, with a mark of 20-7. The Sporting News Record Book lists Boston’s Jess Tannehill as the leader at 22-9. The correct answer is Rube Waddell, also of the A’s, at 27-10; this may be found only in Seymour Siwoff’s Book of Baseball Records. All other sources credit Waddell with a record of 26-11; Coakley’s correct log of 18-8 is nowhere to be found.
Question 2: Both Macmillan and Neft-Cohen show the obscure Comstock at 2-0 for Boston in the three games in which he pitched. However, the results of those three games were one victory, one defeat, and one tie. Only Turkin-Thompson gives Comstock his due at 1-0.
Question 3: Over the years, Cy Young’s victory total has been given variously between 507 and 511; Johnson’s wins have been listed as 413, 414, and 416. Currently, Macmillan credits Young with 511 and Johnson 416, as does Neft-Cohen; Turkin-Thompson lists Johnson at 416 but Young at 507. The correct figures are 510 for Young and 417 for Johnson, as derived from my year-by-year, game-by-game study of the official scoring sheets housed in the Baseball Hall of Fame Library. This research, as yet not complete for all pitchers, has revealed errors in Young’s record for 1907 and Johnson’s for 1912 which are of the same nature as those in last year’s celebrated flap over the 1910 race for the American League batting title between Ty Cobb and Nap Lajoie. How these errors crept into the record and stayed there for 70-75 years I will detail later in this discussion.
In fact, the confusion surrounding these three “trick” questions is merely the tip of the iceberg represented by the period 1901-19, one in which scoring peculiarities (by modern standards) and transcription errors are legion, affecting Hall of Famers and nonentities alike. Moreover, the random, misguided, unreconciled tinkering of the last 15 years–well-intentioned though it may have been–has piled error upon error, creating a dizzying snarl of statistics which becomes harder to untangle with the appearance of each “revised” edition.
Although the mess spills over into batting and fielding records as well, for the time being I will confine myself largely to pitchers’ won-lost records of the 1901-19 era, how they went wrong, and how they can be righted once and for all. But first, a bit of history.
Until 1967, the official scoring sheets for both the American and National Leagues were unavailable to researchers. This meant that baseball reference books compiled prior to that time (such as Moreland, Richter, Lanigan, Turkin-Thompson, Reichler, et al.) were forced to base their versions of pitchers’ won-lost decisions in 1901-19 on the Spalding and Reach Baseball Guides of that period.
This method caused a number of problems. For example, from 1902 through 1906 the Spalding Guide showed two sets of pitchers’ won-lost records for the previous season for both the American and National Leagues. There were the records according to Henry Chadwick, who edited the Spalding Guide, and there were the official records put out by the two major leagues. In the 1906 guide, on page 77 Chadwick shows Christy Mathewson with a 32-8 record for 1905, and then on page 107 the National League official record has him at 31-9. The American League was treated the same way, with Chadwick listing Cy Young as 16-18 on page 121 and the American League official record showing him at 18-19 on page 145. This discrepancy was the product of Chadwick’s idiosyncratic practices in awarding wins and losses; it must be remembered that his records were unofficial.
The guides published by Spalding and Reach from 1907 through 1913 were based solely on the official records of both leagues. This continued with respect to the National League in the 1914 guides, but a strange development had occurred in the American League for the season of 1913. Ban Johnson, A. L. president, omitted won-lost decisions from the official records released to the public, believing that these did not reflect the true worth of a pitcher, and that earned run averages did. (Earned run average was an official statistic in the American League for the first time in 1913.)
In the 1914 Reach Guide, the editor, Francis Richter, put it this way:
It will be seen that in the above official record the pitchers are ranked according to percentage of earned runs, and the old way of ranking them according to games won and lost is omitted altogether. As that custom had been too well established to be discontinued at once, the Editor of the Reach Guide takes the liberty for the benefit of the readers of the Reach Guide to append the following unofficial, but substantially covered record of games won and lost and the pitchers’ rating thereunder.
In the 1915 Reach Guide, Richter did not give even an unofficial won-lost list, simply mentioning that the decisions were omitted from the official record. In the 1916 guide, Richter went back to showing an unofficial won-lost list; but instead of showing just the 1915 season, he also offered the 1914 season with the following explanation:
During the 1914 season the pitchers’ won and lost records were omitted, which had become so well established that they were regarded as indispensable alike by fans and critics. The omission created such a general protest that President Johnson announced that he would restore that pitching feature to future records. Never the less, we find the won and lost records again absent from the official figures. In obedience to public demands, we therefore append the unofficial records for both 1914 and 1915.
Ban Johnson’s policy continued right through the 1919 season, and each year the Reach Guide carried the unofficial won-lost records; Richter was always very careful to keep these separate from the regular official pitching records (E.R.A., strikeouts, etc.). After all, the Reach Guide was the American League publication and felt an obligation to keep its readers informed.
The Spalding Guide was a National League publication, however, and its editor felt no such obligation. The 1914 and 1915 Spalding Guides offered no explanation for the omissions from the official record and did not bother to show any won-lost decisions for the American League.
Unofficial won-lost records did appear in the 1916 Spalding Guide–but were thrown in with the official pitching records, accompanied by a footnote which read, “The won and lost columns are not included in averages compiled by the American League, but are inserted unofficially as a matter of record.”
This approach by the Spalding Guide, which continued through 1919 (no won-lost records were shown in the 1920 guide), was very confusing for two reasons. First, if the reader did not see the footnote, he thought he was looking at the official won-lost pitching records; and second, the footnote implied that the American League did not compile any official won-lost records during the seasons of 1915 through 1918. The American League did compile these records, but just didn’t release them to the public.
More confusion was added in the 1918 and 1919 Spalding Guides when the wording of the footnote was altered. It now read, “The won and lost and percent columns are
not included in the official averages compiled by the American League, but are obtained from official scores.”
During this period, both guides obtained their unofficial won-lost records from the weekly list of pitchers’ decisions published in The Sporting News, Sporting Life, and the Sunday edition of such newspapers as the New York Times and the Washington Post. These lists were based on what the official scorer recommended to the league secretary or president. (He could never do more than recommend: it was the secretary or president who officially compiled the pitchers’ won-lost records during the season.)
Often, when two or more pitchers were involved in a game, the official scorer’s recommendation was overruled by the league president. It was widely known that Ban Johnson, after reviewing the situation, often disagreed with his official scorers. Sometimes the dispute was made public; usually it was not. This compelled statisticians like George Moreland, who compiled many of the weekly lists that appeared in newspapers, to rely solely on the scorers’ unofficial recommendations rather than the final, official decision rendered by Johnson. Of course, the Reach and Spalding Guides were also forced to use these unofficial lists at the end of the season because Johnson did not release the official won-lost decisions.
Such, then, was the data base for the 1901-19 period which was to be used in record books and encyclopedias between 1920 and 1967. It was the best and only information available.
In the fall of 1967, the official sheets of both the American and National Leagues were made available to researchers from Information Concepts Incorporated, the organization responsible for the first edition of the Macmillan Baseball Encyclopedia. (The I.C.I. group, incidentally, disbanded shortly after the 1969 publication of “Big Mac,” but David Neft, Richard Cohen, and Jordan Deutsch of that crew went on to compile a rival encyclopedia for Grosset & Dunlap.) The researchers made a sincere and honest effort to clear up any discrepancies that existed in past major-league records. One of the major problems confronting them was the won-lost pitching records prior to 1920, particularly in the American League. The official sheets for the American League prior to 1905 had not survived and the same situation obtained in the National League prior to 1903. This meant the I.C.I. group had to reconstruct day-by-day pitching and hitting statistics for those periods.
In doing the pitching records for the American League, I.C.I. discovered that the won-lost columns on the 1914 A.L. official sheets were blank. In the National League, too, there were games in which no pitcher had been awarded a win or loss; or a pitcher was awarded a win when it should have been a loss, or vice versa; or two pitchers had been awarded the same win or loss. The I.C.I. researchers corrected most of these mistakes and were able to reconcile the individual pitchers’ won-lost records to those of the teams.
Although there were some errors of this nature, the majority of the won-lost decisions for the American League, 1905-1919, had been recorded correctly on the official sheets. Yet the researchers were perplexed by these records too: they found that in games in which two or more pitchers were used, the win or loss was awarded on a basis which did not conform to pitching rules in effect from 1920 to 1949, nor to those prevailing from 1950 to the present.
Evidently convinced that there was no consistency in these pitching practices, I..C.I. chose to apply modern standards, as is indicated on page 2328 of Macmillan I (1969):
Scoring rules governing won and lost decisions by a pitcher did not become official until 1950. It was decided that all pitching decisions during the period 1920-1949 shall stand as they are in the official records, but that for the period 1876-1919 the 1950 ruling shall be in effect. The reason for this was that since 1920 the official scorer did exist, and he had the explicit authority to award the victory based on common practice, which was very close to the rule adopted in 1950. In the pre-1920 period, however, there was no official scoring rule or common practice for wins by a pitcher and for many years no official scorer.
This wholesale ravaging of the official records was as if a team of archaeologists had come upon the monoliths of Stonehenge and, not fathoming the reason for the complex astronomical arrangement of the stones, had rearranged them into a pattern they could understand.
Of course, this switching around of wins and losses caused quite a few changes in pitchers’ won-lost records, including those of Young and Johnson. Young’s wins went from 511 down to 509 and his losses went up from 313 to 316. Johnson’s wins decreased from 416 down to 413, as did his defeats, from 279 to 277.
In 1978, I undertook a research project to verify the Boston Red Sox won-lost pitching records day by day from 1901-62, comparing my figures with the statistics compiled in the various editions of Macmillan (the current edition, published in 1982, is the fifth). I had no problem in agreeing with Macmillan’s records post-1920, but for the 1901-19 era, it was a different story. I realized that my totals for Red Sox pitchers, gleaned from a variety of sources, differed so much from Macmillan’s that I would have to go to the Hall of Fame Library and go through the official sheets for the American League.
Despite the lack of official sheets from 1901 through 1904, I did not find those four seasons that hard to check because most of the games featured only one pitcher per team and the official won-lost records were in the Reach and Spalding Guides. The 1905-19 period was not so easy, as I had to start matching written newspaper accounts against the official sheets in order to ascertain the official scorer’s thinking in awarding a decision. This prodedure worked out amazingly well: a consistent pattern emerged on all won-lost decisions for Red Sox pitchers. Many of these practices were completely foreign to anything in use today.
I began to wonder if these practices might apply to other American League teams, and if they were common in the National League, too. This started me on a course of doing other teams’ pitchers on a day-by-day basis for 1905-19 and, sure enough, I found the same common practices in effect. I also found that Irwin Howe, A. L. statistician, had released pitching won-lost records in 1914 to The Sporting News, Sporting Life, New York Times, Washington Post, etc. This solved the dilemma of the blank won-lost columns on the 1914 A. L. official pitching sheets.
All of this plus invaluable information, advice, and help from SABR members Cliff Kachline, Ed Walton, Bob Wood, Pete Palmer, John Thorn, Paul Doherty, Don Luce, Bill Gavin, and former Boston Red Sox pitcher Smoky Joe Wood, brought about a list of common scoring practices used in both the American and National Leagues between 1901 and 1919. Had these practices been known to the I.C .I. researchers 15 years ago, we would have a perfect set of won-lost records today.
This story continues tomorrow.