A friend who reads a good deal of Texas history once sent me a short email. He wanted one number: how many votes did Richard Coke receive when he beat Edmund Davis for governor in December 1873? I wrote back with two numbers, and he replied, fairly, that I was hedging.
I was not hedging. The two most convenient reference points on the question disagree with each other. The Handbook of Texas entry on Coke gives the result as 85,549 to 42,663.1 The Handbook entry on the dispute that followed the election gives it as 100,415 to 52,141, and adds that the contest was marked by fraud and intimidation on both sides.2 Both entries come from the same publisher, and both are written with the calm assurance of a settled fact. They cannot both be the canvass.
The difference is not trivial. It comes to nearly fifteen thousand votes for Coke and almost ten thousand for Davis. What does not change is the proportion. In each version Coke carries roughly two votes for every one cast for Davis, which is how the Handbook entry on Edmund J. Davis describes the defeat without giving a count at all.3 So the question my friend asked has a firm answer at the level of the ratio and a soft one at the level of the tally, and the reason for the softness is sitting in a Dallas newspaper.
What the paper says and why it said it
The page is the front of the Dallas Weekly Herald for January 31, 1874, a little over two weeks after Coke was inaugurated on January 15.2 Most of it is political comment from a paper plainly on the new governor’s side: complaints about appointments, remarks on his nominations, a column of short items. Near the bottom of one column is a short paragraph. It reports that the late secretary of state “lacked about forty counties” in turning over the election returns, that many of the missing ones had since been found, and that the full count was expected to raise Coke’s majority to about fifty five thousand.4
Read that sentence as a record and it says three things. The man responsible for the paper side of the election had left office with a large part of the state unaccounted for. The count was still moving two weeks after the winner took the oath. And the Herald expected the final margin to exceed either figure the Handbook now prints, since the two versions give margins of roughly 43,000 and 48,000.
The late secretary of state was James P. Newcomb, a San Antonio newspaperman who had served as a registrar of voters under the Reconstruction Acts before joining the Davis administration.5 The same man had worked at both ends of the franchise, the registration of voters and, later, the election returns sent to his office, which is one reason I read the voter registration lists of 1867 and the tallies of 1873 as parts of one problem. The Herald had every reason to make his office look careless. It also had a reason to make the margin look large, and the next sentences of the same paragraph show it: the editor turns the expected numbers into an argument about how Black Texans voted and why. The arithmetic is a claim about a count. The paragraph around it is an argument about politics. I treat the first as evidence and the second as the paper’s stake.
None of this was new to Texas. When Davis won the governorship in 1869, the official count gave him a margin of slightly more than eight hundred votes, and he was appointed provisional governor in January 1870, about five weeks after the election and before that outcome had been confirmed.6 Twice in four years a governor of Texas took office ahead of a finished count. The difference in 1873 was that the margin was so wide that no plausible correction could reverse it.
What a canvass can hold
Here is the position I will defend, and I know a careful historian could argue with it. A canvass that was incomplete, compiled by an outgoing official from the losing party, and still being corrected after the inauguration was nevertheless sufficient to seat Coke, and the people who acted on it were right to do so. Every plausible version of the missing counties still leaves Coke ahead two to one. When the error bars on a number are narrower than the gap it measures, waiting for precision is a way of avoiding the decision rather than improving it.
I hold that view partly from the other half of my working life. I have sat in rooms where a number was known to be incomplete and a decision could not wait for it: a quarterly figure missing one field report, a reserve estimate with a well still being tested. The discipline is to ask whether any value the missing piece could plausibly take would change the answer. If none would, you decide and you write down what you did not know. The men who convened the Fourteenth Legislature on January 13, 1874, two days before the inauguration, were working through exactly that kind of week.7
The objection runs the other way, and it is serious. Davis believed he was entitled to hold office until April 1874.6 The state Supreme Court had just held the election illegal, and Davis made a last appeal for federal intervention before resigning on January 19.2 To someone standing in the capitol that week, an incomplete count in the hands of the winning side was precisely the kind of paper that ought to be questioned. I understand the argument. I do not think it survives the ratio.
Where my rule breaks is at the edge of what a canvass is. A canvass counts ballots that were cast and received. It cannot count the men who stayed home because they were threatened, or the precincts where a ballot box never reached the county seat, or the voter who was turned away at a poll. The Handbook’s own summary names fraud and intimidation on both sides.2 No tally of the ballots that arrived can tell me how many never arrived, and no amount of care with the forty missing counties recovers them. That is the one question the Herald’s paragraph cannot touch, and it is the question most worth asking about 1873.
I am the author of a biography of Coke, and this race is the kind of fact a biographer is expected to state in a single line. My decision is to print neither total as the result. I describe the outcome as a victory of roughly two to one, cite both figures where a reader might want them, and say plainly that the count had been completed after the fact by people with an interest in its size. That is less satisfying than a single number, and my friend would have preferred the number. The people in the story appear at more length in the reference entry on Richard Coke, and the documents of that winter continue with the first message he sent the legislature; the wider set of records is gathered on the Reconstruction essays page.
I would make the same decision again. A precise figure that the record cannot support is a small fiction, and small fictions in a footnote have a way of becoming facts in the next book.
References
-
Handbook of Texas Online, Texas State Historical Association, Coke, Richard. ↩
-
Handbook of Texas Online, Texas State Historical Association, The Election of 1873, Richard Coke and Edmund J. Davis. ↩ ↩2 ↩3 ↩4
-
Handbook of Texas Online, Texas State Historical Association, Davis, Edmund Jackson. ↩
-
The Dallas Weekly Herald, January 31, 1874, page one, Chronicling America, Library of Congress, The Dallas weekly herald, January 31, 1874. ↩
-
Handbook of Texas Online, Texas State Historical Association, Newcomb, James Pearson. ↩
-
Texas State Library and Archives Commission, Portraits of Texas Governors, War, Ruin and Reconstruction, part two. ↩ ↩2
-
Legislative Reference Library of Texas, Session snapshot, 14th Legislature, Regular Session. ↩