What are they?
Undervotes are ballots that contain votes in some races but not all. Sometimes undervotes are intentional: The voter simply couldn't decide on a candidate, so that part of the ballot was left blank. But sometimes the machine reading a ballot misses a vote that was cast. That can happen when, for example, a chad isn't fully dislodged from a punch card. Florida has more than 61,000 undervoted ballots from this year's presidential race, a large but not shocking number for a state that uses old Votomatics.
How can you tell if an undervoted ballot was intended by the voter to register a vote?
With most undervoted ballots three-quarters, perhaps more there's simply no indication of a vote. These ballots are true undervotes. But on punch-card ballots, even if a chad is hanging by only a corner or two, the counting machine might push the chad into its original place and wrongly call that ballot an undervote. You can tell only by looking.
So why don't we just look at them?
Usually no one looks at undervoted ballots because there aren't enough of them to change the outcome. Because election officials so rarely examine them, they haven't set uniform standards for how to count them. Clashes occur when some counters say a mere dimple in a chad even though the chad remains attached on all four sides should be counted as a vote.
What does Florida law say about which undervotes count?
Nothing, which is the problem. The legislature has said only that a legal vote exists if there is "clear indication of the intent of the voter," a frustratingly rubbery phrase. In its Friday ruling ordering counts of the undervotes, the Florida Supreme Court simply quoted that vague guideline. Nor did Judge Terry Lewis, who was implementing the hand count, offer any specifics. So on Saturday, before the U.S. Supreme Court halted the counting, each of 64 canvassing boards was making its own decisions on dimples, on who would observe the counts, on how to separate the undervoted ballots from the others, and so on. Hillsborough County, for instance, decided not to count dimples. (Says canvassing board member Marcia Reynolds, "These chads are a living nightmare.") But Miami-Dade had already started a hand count last month that included dimples.
Don't most of the undervotes come from Gore-majority areas?
Basically, yes. Of the 42,000 undervotes yet to be examined, more than 35,000 come from the 25 punch-card counties. Punch-card machines are less reliable than the pricier voting technology used in more Republican areas. Brevard County, for instance, which went 53 percent to 45 percent for Bush, uses optical scanners. "There are no dimples, crimples, pimples or anything else to interpret," says election supervisor Fred Galey. That's good for him but if the hand count resumes bad for Bush, since Brevard's 277 undervoted ballots probably contain few votes. By contrast, Pinellas County's 4,226 undervoted punch-card ballots could contain hundreds of votes, and Gore won Pinellas. Across the state, undervoted ballots tend to be concentrated in Democratic areas. One explanation: Elderly and new voters make more mistakes on ballots; many of Gore's black supporters were voting for the first time.
Is that why Bush's lawyers have fought hand counts of undervotes so doggedly?
Yes, but they have valid reasons for opposing the counts chiefly that there is no single standard for them. They also say handling the ballots "spoliates" them, a lawyerly way of saying counters can push chads through accidentally. Bush lawyer Phil Beck wanted Lewis to rule that only a chad with a stylus imprint could count, not one pushed through, say, by a finger. Lewis dismissed that request for now, but in the end his decision may not matter.