"We did run some number stats and the number of blacks [on the list] was higher than expected for our population," says Chuck Smith, a statistician for the county. Iorio acknowledged that African-Americans made up 54 percent of the people on the original felons list, though they constitute only 11.6 percent of Hillsborough's voting population.

Smith added that the DBT computer program automatically transformed various forms of a single name. In one case, a voter named "Christine" was identified as a felon based on the conviction of a "Christopher" with the same last name. Smith says ChoicePoint would not respond to queries about its proprietary methods. Nor would the company provide additional verification data to back its fingering certain individuals in the registry purge. One supposed felon on the ChoicePoint list is a local judge.

While there was much about the lists that bothered Iorio, she felt she didn't have a choice but to use them. And she's right. Section 98.0975 of the Florida Constitution states:

"Upon receiving the list from the division, the supervisor must attempt to verify the information provided. If the supervisor does not determine that the information provided by the division is incorrect, the supervisor must remove from the registration books by the next subsequent election the name of any person who is deceased, convicted of a felony or adjudicated mentally incapacitated with respect to voting."

But the counties have interpreted that law in different ways. Leon County used the central voter file sent in January 2000 to clean up its voter rolls, but set aside the one it received in July. According to Thomas James, the information systems officer in the county election office, the list came too late for the information to be processed.

According to Leon election supervisor Ion Sancho, "there have been some problems" with the file. Using the information received in January, Sancho sent 200 letters to county voters, by regular mail, telling them they had been identified by the state as having committed a felony and would not be allowed to vote. They were given 30 days to respond if there was an error. "They had the burden of proof," he says. He says 20 people proved that they did not belong on the list, and a handful of angry phone calls followed on Election Day. "Some people threatened to sue us," he said, "but we haven't had any lawyers calling yet."

In Orange County, officials also sent letters to those identified as felons by the state, but they appear to have taken little care in their handling of the list. "I have no idea," said June Condrun, Orange's deputy supervisor of elections, when asked how many letters were sent out to voters. After a bit more thought, Condrun responded that "several hundred" of the letters were sent, but said she doesn't know how many people complained. Those who did call, she said, were given the phone number of the Florida Department of Law Enforcement so that they could appeal directly to it.

Many Orange County voters never got the chance to appeal in any form. Condrun noted that about one-third of the letters, which the county sent out by regular mail, were returned to the office marked undeliverable. She attributed the high rate of incorrect addresses to the age of the information sent by DBT, some of which was close to 20 years old, she said.

Miami-Dade County officials may have had similar trouble. Milton Collins, assistant supervisor of elections, said he isn't comfortable estimating how many accused felons were identified by the central voter file in his county. He said he knows that about 6,000 were notified, by regular mail, about an early list in 1999. Exactly how many were purged from the list? "I honestly couldn't tell you," he said. According to Collins, the most recent list he received from the state was one sent in January 2000, and the county applied a "two-pass system": If the information on the state list seemed accurate enough when comparing names with those on county voter lists, people were classified as felons and were then sent warning letters. Those who seemed to have only a partial match with the state data were granted "temporary inactive status." Both groups of people were given 90 days to respond or have their names struck from the rolls.

But Collins said the county has no figures for how many voters were able to successfully appeal their designation as felons.

Recent Stories