The legal and public-relations battles over the Florida presidential vote are likely to come down to a single question: Should hand-counted ballots be added to the regular (machine-read) election returns for a few counties where the Democrats have requested hand counts, or should they be added for all counties, or should they be added for none?

In the key battleground of Palm Beach County, the Democrats' argument for adding hand-counted ballots rests in turn on a single assertion. The unadjusted Palm Beach vote, it is said, was anomalous as compared to other Florida counties, with Pat Buchanan receiving an inexplicably high number of votes. The "high" Buchanan vote--combined with complaints about the "butterfly" design of the Palm Beach ballot and the statements of some voters that they mistakenly voted for Buchanan rather than Gore--suggests that the county's voting procedures artificially suppressed the Vice President's vote.

The "high" Buchanan vote is essential because, without it, all that remains are the arguments over ballot design and the morning-after complaints of voters selected by the Gore organization. These factors alone cannot be grounds for a ballot hand-count, court intervention, or any other form of post-election adjustment--unless we are prepared to do the same in hundreds of other counties in Florida and other states where the vote was close. The butterfly ballot was used in other jurisdictions, and there are both pros and cons to this (and other) ballot designs. The Bush and Gore organizations could undoubtedly produce an endless stream of voters who failed for one reason or another to register their "true" choices in the voting booth. If these issues justify post-hoc adjustment, the election result will be postponed indefinitely and, perhaps, effectively nullified.

But there was not, in fact, anything anomalous or irregular in the Palm Beach County election returns, and nothing in the returns to suggest that Buchanan "took" votes from Gore. The arguments to the contrary consist of two assertions. The first is simply that Buchanan received more votes (3,412) in Palm Beach County than in any other Florida county. But Palm Beach is one of Florida's most populous counties, and 16,695 of its residents are registered with the Reform, American Reform, or Independent parties--the second largest county registration for these parties in all of Florida. Taking account of the county's size and party registration, Buchanan's Palm Beach County vote was strong but not a statistical "outlier." In several other counties, Buchanan received a higher percentage of votes cast than his 0.79% in Palm Beach County. And in several others, he garnered more than 20 percent (3,412/16,695) of the Reform, American Reform, and Independent Party registration.

The second, more sophisticated assertion is that Buchanan's Palm Beach County vote was much higher than would have been projected by statistical methods. Last week, two economists associated with the Vice President released the results of a regression study that compared the Buchanan, Gore, and Bush votes across all Florida counties. If Buchanan had received the same votes relative to Bush in Palm Beach as in other Florida counties, his Palm Beach vote would have been only 1,200. This, they said, "appears to suggest some irregularity" in the Palm Beach County vote.

Their study, however, entirely omits party registration data. Party registration varies considerably from county to county, and it is obviously a powerful measure of voters' inclinations--unaffected by ballot design and other voting procedures. It makes much more sense that a candidate's vote would correlate, county by county, with his party's voter registration than with the votes cast for the other candidates.

Kevin Hassett of the American Enterprise Institute and John Lott of the Yale Law School have since analyzed the Florida voting data with party registration data factored in. They independently reached the following conclusions:

· First, most though not all of Buchanan's "excess" Palm Beach County votes (compared to his votes in other counties) can be explained by the high Reform, American Reform, and Independent Party registration in Palm Beach County.

· Second, Gore also received "excess" votes in Palm Beach County. Whether compared to Democratic Party registration or to the votes received by Bush and Buchanan, Gore did better, not worse, in Palm Beach County than in other Florida counties. Gore's Palm Beach "excess" was even greater than Buchanan's.

· Third, Bush received fewer votes in Palm Beach County than in other Florida counties. As in the case of the Gore "excess," the Bush "shortfall" in Palm Beach exists both when Bush's votes are correlated with the votes received by the other candidates (the Gore analysts' procedure) and when they are correlated with Republican Party registration (the procedure used by Hassett and Lott).

Statistical analysis cannot account for all of the variation in voting patterns from jurisdiction to jurisdiction. But it can and does say that there was nothing "irregular" in the Palm Beach County voting returns, and nothing to suggest that Buchanan's votes somehow suppressed Gore's. If anyone can complain about losing "mistaken" votes to Buchanan on the butterfly ballot, it is not Gore but Bush (Buchanan appeared between Gore and Bush on that ballot). There are many possible explanations for the relative performance of the three candidates that have nothing to do with ballot design. Buchanan's part-time residence in Palm Beach County may have boosted his votes. Palm Beach County is heavily Democratic, and perhaps some of the dominant Democratic voters talked their Republican friends into voting for Gore. We cannot say for sure, but we can say that nothing in the voting returns invites further scrutiny or special adjustments to add votes for Gore.

Indeed, precisely because Palm Beach County is predominantly Democratic, even "neutral" adjustments, such as hand counts that add all "partially perforated" ballots, will artificially skew the Florida vote in Gore's direction if applied only in that county (or only in other Democratic counties). The initial Palm Beach hand counts demonstrate this--and also provide further evidence that the election-night count in that county was not anomalous. The hand count has not unearthed a mountain of uncounted Gore votes, as would have happened if the "irregularity" thesis had been correct. To the contrary, and consistent with the Hassett and Lott analyses, the hand count has found relatively fewer new Gore votes and relatively more new Bush votes than would have been expected from the county's party registration and election night votes. Rather than correct an anomaly, all the procedure can do is to add more votes for Gore than for Bush because confined to a Democratic county. In sum, considerations of fairness as well as finality argue for taking the Florida recount--including absentee and overseas ballots but not hand-counted ballots from counties selected for partisan purposes--as final.