The Final Word?

After spending nearly $1 million, a consortium of big news organizations last week rendered what it once thought would be final word on last year's bitterly contested Florida recount.

The decision: a split verdict.

To the chagrin of Democratic partisans, the consortium proclaimed Bush still would have won the apparently limited statewide recount underway last December 9 even if the U.S. Supreme Court had not swooped in and stopped it. But if all disputed ballots had been manually counted--something, ironically, neither side had even asked for--Al Gore could have eked out a narrow victory.

That, at least, is how the story got played last week in front page stories in The New York Times, Washington Post and other newspapers that participated in this curious historical exercise. (Newsweek was briefly part of the consortium but dropped out because of scheduling conflicts before the project was completed.)

But is that really the last word?

Don't count on it.

Buried deep in the files of the mammoth Florida litigation, Newsweek has uncovered hastily scribbled faxed notes written by Terry Lewis, the plain-speaking, mystery-novel writing state judge in charge of the Florida recount, that potentially changed the calculations rather substantially.

The previously undisclosed notes lend firm documentary support to recent comments by Lewis that he might well have expanded the Florida Supreme Court-ordered statewide recount of "undervotes"--the disputed ballots in which machines did not record any vote for president. The notes show that--just hours before the U.S. Supreme Court issued its order--Lewis was actively considering directing the counties to also count an even larger category of disputed ballots, the so-called "overvotes," which were rejected by the machines because they purportedly recorded more than one vote for president.

This is the group of disputed ballots that the consortium has now found that--had they been manually recounted--might have yielded a rich treasure trove of votes for Gore, enough even to put him over the top.

"Judge, if you would, segregate 'overvotes' as you describe and indicate in your final report how many where you determined the clear intent of the voter," Lewis wrote in a note to Judge W. Wayne Woodard, chairman of the Charlotte County Canvassing Board on the afternoon of December 9, 2000. "I will rule on the issue for all counties, Thanks, Terry Lewis."

The issue had arisen that crazed weekend in December because many of the statewide canvassing boards found themselves confused when they sat down to implement the State Supreme Court order. The court's opinion directed them to visually inspect and then manually recount their "undervotes" to see if any of them reflected "the clear intent of the voter"--the standard the court said was the guiding principle of state election law.

One of those so confused was the canvassing board of Charlotte County, in southwest Florida. In a memorandum faxed to Lewis in Tallahassee at 1:49 p.m. that Saturday, Woodard, chairman of the board, noted that, under its optical scanner voting system, Charlotte county had more "overvotes" than undervotes. "This can result from a number of reasons," Woodard wrote, and then added: "but a manual review of the ballots can in many overvote instances determine the 'clear intent' of the voter."

Woodard wanted to know: "Is the Board restricted to the 'non-votes' or 'undervotes' or should the Board manually review the overvotes?"

The Charlotte County board wasn't the only one posing the question. Faxes located in the files of the Leon County courthouse where Lewis sits show that other canvassing boards were confronting the same issue--and dealing with it in different ways. One canvassing board, in Lafayette County, faxed a letter to Lewis at 12:48 p.m. that afternoon informing him that they had already twice counted its "overvotes" by hand and were recertifying its findings.

The chairman of the canvassing board in tiny Santa Rosa County even faxed a disputed "overvote" ballot to Lewis at 11:55 a.m. It showed a voter who had marked his ballot with an arrow pointing toward Bush and then also (apparently mistakenly) marked it with another arrow pointing toward Harry Browne, the Libertarian Party candidate. The voter then wrote over the arrow pointing to Browne the word "no." The Santa Rosa board requested Lewis's "immediate opinion" on whether such a ballot should be counted--even though it wasn't an "undervote."

Lewis's faxed handwritten reply: "For any votes like this, make a separate stack...and advise me in final report of those that you consider to be a 'clear indication of the voters' intent' even though its not a non-vote or undervote."

When first asked by Newsweek about the "overvotes" issue last April, Lewis directed a reporter to the faxed notes-- and cited the uncounted Bush-Browne ballot in Santa Rosa County as a prime example of why "overvotes" should have been looked at. "I would have addressed that issue," Lewis told Newsweek at the time, adding that he had even scheduled a hearing on the issue for the following day Sunday. (The hearing became moot once the U.S. Supreme Court stayed the recount at 2:45 p.m. Saturday.)

"Logically, everything the Florida Supreme Court said was, 'You have to look at the clear intent of the voter,'" Lewis said. "Logically, if you can look at a ballot and see, this is a vote for Bush or this is a vote for Gore, then you would have to count it....Logically, why wouldn't you count it?"

Lewis has repeated that he planned to review the issue that weekend in interviews with others, including the Orlando Sentinel (which published them for the first time in a story on Nov. 12). And if he had so ordered the counties to count the "overvotes," would that have made Al Gore president?

The news consortium review suggested the answer might be "yes." As the consortium counted the ballots, Bush would have won the statewide recount of undervotes by 243. But the consortium found that a manual inspection of the approxmiately 113,820 overvotes yielded hundreds of extra votes for Gore--primarily because they occurred most frequently in counties with more black and lower-income voters who have less experience with voting and, in some cases, were apparently penalized by poor ballot design. By one consortium count, a full statewide count of both undervotes and overvotes would have resulted in Gore winning by 171 votes.

But as one Bush lawyer put it this week, "This is never never land." Had Lewis directed the counties to begin counting the overvotes, Bush lawyers--and possibly even Gore lawyers--would have immediately jumped in and challenged his order as going beyond the boundaries of the State Supreme Court. The challenge would have created more chaos, eaten up more time and probably have required the State Supreme Court to convene another hearing, and issue another ruling, thereby adding to the circus atmosphere surrounding the Florida battle. As even Ron Klain, the lawyer in charge of the Gore forces put it in a panel discussion Saturday, efforts to reconstruct who really would have won the Florida recount are "like stopping a football game in the third quarter" and then reconvening it with the same teams a year later to try and decide who won.

Then there is another small matter overlooked by the consortium: the numbers don't add up. The consortium handed over the job of counting the disputed ballots to the National Opinion Research Center (NORC) at the University of Chicago, which in turn employed 153 ballot examiners to visually inspect the ballots in all of Florida's 67 counties. In all, the NORC counters were able to inspect 175,010 disputed ballots that the counties identified as having gone uncounted or rejected. But those 175,010 uncounted votes are 1,436 ballots less than the 176,446 ballots that, the news organizations independently determined, had originally been reported by the counties as rejected or uncounted on election night.

What explains the discrepancy? Nobody can say for sure. Kirk Wolter, senior vice president at NORC, told Newsweek that each time the counties got asked to identify their uncounted ballots, they would come up with slightly different numbers. "Each time, they did it, it turned out a little different," Wolter said. The Dade County elections supervisor last week told the conservative journal Human Events that election workers had difficulty trying to sort out ballots as they were run through county voting machines.

So does that throw a major league curveball into all the conclusions reported last week by the consortium? "I'm not going to quarrel with that," Wolter said. "There's certainly uncertainty about the results. There's no question about that."

Just like, some might say, the election itself.