The impact of these differences on the outcome will never be known but their potential magnitude is evident in Miami-Dade County, where predominantly black precincts saw their votes thrown out at twice the rate as Hispanic precincts and nearly four times the rate of white precincts. In all, 1 out of 11 ballots in predominantly black precincts were rejected, a total of 9,904.  Had all people cast ballots that could be counted along the same lines as their neighbors, Mr. Gore would have gained nearly 7,000 votes.  Florida voters cast their ballots in a patchwork of different voting technologies. Nearly 60 percent of voters cast ballots on the punch ballot system, while more than a third use optical scanning system. Some small counties use paper ballots, or mechanical voting machines.  Each county makes its own decisions on which type of machine to use, and there have been no indications that the choice of machine is designed to favor any party or candidate.

Chuck Smith, a systems administrator in Hillsborough County, said that the county had long been aware of limitations of the punch ballot system used there and had been working for years on a successor system.  But he said that when he headed a study of alternatives for three Florida counties, he found that the optical scanning system was costly to set up, and was prone to other problems. Instead, he said, the county is waiting for the state to approve companies that could supply an electronic voting system.  He said companies had been slow to provide the state with software and equipment for testing, but he expected that process to speed up, once the books are finally closed on the presidential election.  "I think it will move much faster now," he said.

Across the country, the punch-card systems have resulted in significantly higher numbers of uncounted ballots. In 467 counties that used punch cards with pre-punched cards in the 1996 presidential election, according to a New York Times analysis, 661,000 of the 25 million ballots were not counted, or 2.6 percent of the total.  In 729 counties that used optical scanning technology, 1.9 percent of the ballots cast had no votes for president counted, according to the Times analysis of data provided by Election Data Services, a nonpartisan consulting firm in Washington.  But in areas where the voting technology prevents voting for more than one person in a race -- on lever machines, for instance -- the percentage of voters who cast no presidential vote was low this year. In Connecticut, only one half of 1 percent of those who went to the polls failed to cast a presidential vote.