Down for the recount: no one can know who won Washington State's gubernatorial race.

'Rossi Re-Vote."

That's how they answer the phones these days at the headquarters of Dino Rossi, the Republican who, after one election-night count, one machine recount, one excruciating manual recount, and more than a few electoral oddities, lost the race for governor of Washington State by 129 votes. Rossi, a former state senator, is now in court, contesting the election results and asking that a new vote be held.

There seems little doubt that Rossi deserves to be governor at least as much as Democrat Christine Gregoire, the state's three-term attorney general who was declared the winner and sworn in on January 12. But at this point, after all the counting, nobody claims to know precisely how many votes were cast for each candidate last November. Nobody claims that it is even possible to know that number. The only question now is what to do about it.

About 2.8 million people went to the polls in Washington State on November 2, and what happened after the voting became one of the most vertiginous political roller-coaster rides in recent history. On election night, Rossi was slightly ahead. By the next day, as the votes from heavily Democratic King County, home of Seattle, continued to come in, Gregoire took a lead that grew as large as 16,000 votes. By the morning of November 9, that lead had been cut in half. But at that point, more than 200,000 ballots had not yet been counted. Later that night, results came in from a number of Republican-leaning counties, and Rossi took a lead of about 2,000 votes. The day after that, it stretched to 3,500.

About 85,000 ballots were still uncounted. Most of them were provisional ballots, meaning they were cast by people who came to the polls and could not show that they were registered to vote. They were allowed to cast ballots anyway, with the understanding that the ballots would be counted later if it could be shown that those voters were indeed registered.

On November 15, the count moved in Gregoire's favor when King County announced that it had 10,000 more ballots to count than election officials originally thought (it's still not entirely clear how that happened). Those votes went heavily to Gregoire, putting her in the lead by 158 votes. It was a kick in the teeth for Republicans, but Rossi remained calm. "We still knew how many ballots were outstanding in the counties that we had," Rossi told NATIONAL REVIEW recently, "and we were confident that we were going to win."

Sure enough, when those Rossi-leaning counties sent in their final totals, the Republican pulled back into the lead. On November 17, the counting was finished, and Rossi had won by 261 votes.

COUNT AND COUNT AGAIN

A recount was guaranteed: Washington State law requires a recount if an election is decided by less than one-half of one percent of the total votes cast. The recount was done by machine. It might seem obvious to say that it was intended to recount those ballots that had already been counted in the first round, but that fact was not obvious to officials in King County. They decided to "enhance" some ballots that had not been legible or otherwise countable during the first go-round. More and more ballots went into Gregoire's column. The King County votes cut into Rossi's lead, but not enough to put Gregoire into the lead. By the time the machine recount was finished, Rossi won by 42 votes.

Gregoire refused to concede. Democrats wanted a hand recount of all the votes, but the law required that they pay for it, which meant they would have to raise nearly $1 million. For a while, Democrats wondered where the money would come from, but their worries were eased when checks came in from John Kerry, who donated $250,000 in unused campaign funds; from MoveOn.org, which pitched in another $250,000; from the Democratic National Committee, which sent still another $250,000; and from former Democratic presidential candidate Howard Dean, who raised about $200,000 for the cause.

The recount began December 8. At first, it seemed to make little net change in the vote totals. Rossi appeared to be headed for a final, unquestionable victory when King County officials announced they had found 573 "new" ballots, which--they said--had been mistakenly disqualified. Republicans went to court, arguing that the time for counting "new" ballots was past, and that to count them now would require changing the rules in the middle of an election. But the state supreme court ruled against the GOP. The "new" votes were counted, and in the final tally Gregoire won by 129 votes. On December 30, she was certified the winner.

Republicans were stunned. "We thought that if it was a straightforward operation, a hand recount would work for us," Rossi says. "But King County kept finding ballots and finding ballots and finding ballots until they had enough to win."

What should the GOP do? Rossi and his allies began collecting the voting records from several counties around the state, and found what might be charitably called serious irregularities. They found counties where the number of votes cast exceeded the total number of registered voters. They found counties that had made mistakes in the counting of provisional ballots. And they found counties in which--shades of Daley's Chicago--dead people apparently cast ballots.

ONCE MORE UNTO THE BREACH 

The GOP resolved to challenge the election results in court. Rossi decided not to insist that had he in fact won the race (although he believes that he did), but to argue instead that the whole count was such a mess that the real winner can never be established. "[We] believe that the number of illegal votes counted, and the number of valid votes improperly rejected in this election, are so great as to render the true result of the election uncertain and likely unknowable," Rossi's court petition said. "Because the true results cannot be ascertained, a new election must occur promptly to restore the integrity of Washington's election process."

The challenge is based on three categories of evidence. The first is the discrepancy between votes and voters. In King County alone, there were about 1,800 more ballots counted than there were voters who were credited with voting--that is, people who, as the law requires, signed the poll book or who signed their absentee ballots. In an election decided by 129 votes, the significance of that is obvious. And--unlike a lot of election contests in the past--no one disputes that there is a problem here. County election officials concede that there is indeed a difference in the two totals; after first claiming that it was 1,200 votes, they later raised it to 1,800. The officials defended themselves by saying that they were "99.99 percent accurate in terms of administering of this election," suggesting that the discrepancy is statistically insignificant.

Republicans disagree. "It's incredibly significant," state GOP chairman Chris Vance told NATIONAL REVIEW. "What is the number one safeguard we have to insure a fair election? How do we prevent someone from just stuffing a ballot box? It's the basic requirement that somebody has to sign something. Every single voter either has to sign a poll book or sign the absentee ballot, and in this case you have more votes than people who signed in."

The second category of evidence in the Rossi challenge is the odd set of circumstances surrounding the handling of provisional ballots. Anyone can go to a polling place and request--and receive--a provisional ballot. After that person votes, the ballot is supposed to be kept in a secure envelope while election officials check on the person's eligibility to vote. At all times, that vote should be segregated from the larger pool of votes. But in King County, officials have found nearly 350 cases in which those ballots were mistakenly counted. "They were fed into the counting machines, went into the sea of ballots, and can never be traced," says Rossi campaign spokeswoman Mary Lane. The Rossi campaign believes there are another 75 or so such ballots in Pierce County, meaning more than 400 ballots improperly entered the system--again, in a race decided by 129 votes.

The third category of evidence is the number of instances in which felons and corpses voted, while other people voted more than once. There is still no total number of ballots that might fit into that category--Rossi's researchers are painstakingly matching voter names to death records, criminal-conviction records, and the like--but so far, Republicans have come up with more than 129 examples.

With all of that, Rossi believes he has a slam-dunk case for a re-vote. "We have so much evidence that the court will have to strain its neck to look the other way," he told NATIONAL REVIEW. "The bottom line is, we don't really know who won the race." Rossi stresses that he is not simply trying to have enough votes reversed to make him governor. The drawn-out process of recounting and "discovering" previously unknown votes has shaken the public's faith in the integrity of the system, and yet another turnaround, Rossi says, would do nothing to remedy that situation. "If I just turned it over by 130 votes, I just don't think people's faith in the process would be restored," he explains. "Under those circumstances, I'd rather not be governor."

Rossi concedes that there is no precedent in Washington State for holding a statewide re-vote. In fact, the only precedent he points to is a 1974 race for county commissioner in Washington's Adams County in which, after bungled recounts, a judge ordered a new vote. "There's no statewide precedent," Rossi says, "but courts have upheld the principle." But there seems little doubt that if Rossi wins in court, and a judge orders a new vote for governor, there will be plenty of future races in which losing candidates call for re-votes, and not all of them will have as strong a case as Rossi. Where will the line be drawn? The possibilities for litigation--and the disruption of government--will be endless.

In addition, it might serve Rossi's political future better if he, like John Thune--who narrowly lost a Senate race in South Dakota in 2002 after probable fraud on the state's Indian reservations--bowed out gracefully and fought another day. Thune later won a Senate seat, knocking off Tom Daschle last year. Rossi rejects the comparison. "In Thune's case, I don't think that they had anybody on the reservation admitting they had more votes than voters, or admitting that they put through provisional ballots," he says. "King County did that. It's a different issue."

Another reason not to re-vote is that Gregoire has been badly hurt by her actions. Polls show that most people in Washington State do not believe she won office by legitimate means, and she will likely govern under a cloud of suspicion. Republicans will be free to remind the public of that at every appropriate time. And then, in four years, the injustice of 2004 will likely be on the voters' minds as they go to the polls. But Rossi rejects that argument, too. "This is a lot bigger than me and her at this point," he says. "If I just roll over, they're going to sweep this under the rug, and we'll have the same problem next time. This is the only way this process is going to be cleaned up."

Now, it's up to a judge. Just as there is no way to know who won the election held on November 2, there is no way to know whether holding a re-vote is the correct solution to the problem. Yes, justice might be done, but the precedent might well lead to more litigation, which might further undermine the public's faith in the system that Rossi wants to restore. If hard cases make bad law, then hard elections might do the same thing, too.