Send To Printer Email to Friend

The Florida Overvote: Tragic Mistake, or Katharine Harris with Tweezers?
Sharman Braff sharmanbraff@yahoo.com

"You say overvote, I say Katharine Harris with tweezers"
(from a website I could not find again or I'd give credit)

It is a crime that the Republican Party stopped the counting of Florida's 60,000 undervotes. But perhaps the real crime is in the overvote.

There were 110,000 overvotes in Florida, ballots rejected because more than one choice was marked for president. Overwhelmingly, the spoiled ballots were Gore votes. It was the overvote, much more than the dangling and dimpled chads of undervote, that cost America the president we really elected.

Did tens of thousands of voters screw up their ballots by mistake, or was the second hole punched by someone other than the voter?

1) Now it's official: Gore won by 50,000. But for the overvote, that is.

Just six counties, Duval (22,000), Palm Beach (19,100), Miami-Dade (18,000), Broward (8,000), Hillsborough (5,000) and Pinellas (4,000), account for 76,000 of the overvotes. We'll call them the suspicious counties. [1]

The Washington Post was able to secure computer records from the machine counts of all the suspicious counties except Duval, together with three other small counties. This is the first time we've been able to learn, not just how many overvotes there were and what precincts they came from, but specifically which candidates are marked on the double-punched ballots.

The Post determined that 46,000 Gore votes were spoiled by double-punches, compared to 17,000 Bush votes. The article also reported that 5,800 overvotes showed punches for both Gore and Bush. If all the 5,800 Bush/Gore ballots were altered Gore votes, that's 46,000 spoiled Gore votes to 11,000 of Bush's. If you split that bunch 50/50, that's 43,000 Gore votes to 14,000 of Bush's. (I assume the 5,800 Bush/Gore ballots were included in the 46,000 and 17,000 totals. The numbers add up correctly (79,000 total with Duval)-see Footnote 1.) [2]

The ratio of spoiled ballots ran more than 4:1 to Gore's detriment. If Duval's 22,000 overvotes went in approximately the same ratio (say, 18,000/4,000), that's 64,000 spoiled Gore votes to 15,000 for Bush. Almost 50,000 net. Remember the Miami-Herald's statistical analysis of the rejected ballots, which showed Gore winning by 23,000 votes? Al did much better than that.

If you wanted to spoil 50,000 of your opponent's votes, you might also double-punch 15,000 of your own (then another 15,000 of his), to make the overvote look less suspicious.

And tell me this, is there any other state in the country that saw 5,800 ballots marked for both Gore and Bush?

2) 76,000 is an extraordinarily high number for punchcard overvote

The public now knows that many votes go uncounted in every election. But most rejected ballots are undervotes. Overvote is relatively rare, and it is particularly rare with punchcard ballots. All six of the suspicious counties use punchcard ballots.

One expert put the normal rate for punchcard overvote at 0.1% of total ballots cast, [3] while another expressed the rate as 2% of total rejected ballots. [4] By either measure, one would expect only a few thousand such ballots out of Florida's 6,000,000 total. In the words of the two experts, Florida's punchcard overvote was "phenomenal," "outrageous."

3) However much the Republican Party tries to downplay the overvote numbers, no one disputes that the overvote in Duval (22,000) and Palm Beach (19,100) is remarkable.

As the overvote anomalies came to light during the election contest, the Bush campaign spun a number of arguments to minimize its significance. A common ploy was to recite statistics about total rejected ballots, for example, comparing Palm Beach's 14,000 rejected ballots in 1996 (combined over and undervote) with its 19,100 overvotes in 2000. My personal favorite was the GOP chairman who cited the overvote in Duval, a Republican county, as proof that there was nothing suspicious about the overvote in Democratic Palm Beach. [5] (Duval is in fact Republican by majority. But it has a large black population in urban Jacksonville, which is-surprise-where the overvote came from.)

No one really disputes that the overvote numbers in Duval and Palm Beach are extraordinary. This has been the subject of many a news report. And no one disputes that the rejected ballots came disproportionately from Democratic precincts. (And as the Washington Post now confirms, were overwhelmingly Gore votes.) In some black precincts, the spoilage rate ran as high as 30%. [6]

But the public has been given an explanation for the unfortunate disenfranchisement of these voters: Confusing ballot designs, coupled with inexperienced and uneducated first-time voters. The get-out-the-vote effort (raising Afro-American participation in Florida's election from 10% in 1996 to 15% in 2000) was impressive, but tragically undermined by voter error.

I will discuss the allegedly confusing ballots of Duval and Palm Beach in more detail below. Suffice it to say here, Al Gore must be really unlucky to be on the losing end of not one but two bizarre instances of voter confusion, two different ballot designs, 40,000 lost votes in all, in the one state that was key to the whole election.

4) But it's not just Duval and Palm Beach. There were 18,000 overvotes in Miami-Dade, and an additional 18,000 from Broward, Hillsborough and Pinellas combined. Again, all Democratic precincts.

I am aware of no ballot design problems, no voter confusion or other circumstance that would account for the extraordinary overvote in Miami-Dade, Broward, Hillsborough or Pinellas.

In fairness, I know of one prior election in which Miami-Dade experienced an extraordinary overvote rate, the election of 1988. That militates in favor of an innocent explanation for the 2000 overvote. There were, however, peculiar circumstances surrounding the 1988 election, discussed more fully below.

5) Comparison with optiscan overvote casts further suspicion on the six punchcard counties

Reports of the hand counts in optiscan counties shed some light on how innocent overvote occurs. [7] 1) Voters erased incorrect choices, not realizing the machine would read the erasure as a mark. 2) Smudges or other marks were read by the machine as votes. 3) Voters crossed through several names and circled their chosen candidate. 4) Voters marked the wrong candidate initially, then wrote "mistake," circled the correct name, or made similar notations to indicate the error. 5) Voters thought the "write-in" line required them to both mark where the candidate's name was listed and also write the candidate's name out.

Two things are notable here. First, these errors are much more likely to occur with optiscan ballots, where the voter sees the candidates' names on the ballot and votes by marking the ballot with a pencil. That bears out the statistical evidence that overvote with punchcards is relatively rare.

Second, with many of the optiscan overvotes, the cause of the overvote, what was going on in the voter's mind, could be figured out by looking at the ballot. In fact, the voter's intended candidate was often clearly discernable, as when the voter noted a mistake, or wrote the candidate's name on the write-in line. Voter intent could be discerned in as many as 1,000 of Lake County's 3114 overvotes.

By contrast, the 76,000 punchcard overvotes offer few insights into the voter's mind. We know this because both Broward and Palm Beach (and Miami-Dade for several precincts) did full manual recounts, with Republican and Democratic observers scrutinizing every ballot. If there had been any overvotes that revealed the voter's intent, or even shed light on how so many voters came to spoil their ballots, we would have heard about it. In fact, it was Palm Beach's policy to count write-in overvotes where the voter's intent could be discerned. [8] Yet few (if any) of Palm Beach's 19,100 overvotes were recovered in the hand count. [9]

Presumably, what the canvassing boards in Broward and Palm Beach found when examining the overvote was simply ballots with two chads punched out for president, and no explanatory or other marks. Given the messy, even chatty ballots we see with optiscan overvote, the cleanliness of the punchcard overvote is suspicious.

6) The "voter confusion" stories are not convincing.

One lesson learned from the optiscan overvote is that voter error does occur, but it is usually understandable. (Note bene, I have heard of no optiscan ballots marked for both Gore and Bush.) In addition to the errors seen in the optiscan overvote, there was the Buchanan error in Palm Beach, and for another example, a few thousand ballots were punched one-off the correct hole, because the voter did not insert the ballot in the machine (the machine aligns the ballot with the first hole not to be punched). I can believe that a handful of voters might make these mistakes.

I cannot believe that 40,000 voters made the mistakes supposed to have occurred in Duval and Palm Beach.


The presidential race in this county's ballot ran onto two pages, and newspaper ads showing sample ballots before the election instructed voters to "vote every page." Get-out-the-Vote volunteers with the Democratic Party, working with poorly educated, first-time voters, repeated the bad advice. Faithfully following directions, voters punched for president on both the first and second pages, voiding their votes.

Did this really happen? Of all the evidence for an innocent explanation of the overvote, this is the most convincing to me. Not because it sounds plausible, but because Democratic Party and elected officials, and groups such as the NAACP, have publicly repeated the story. [10] Undoubtedly they have interviewed voters in Duval, so there must be some anecdotal evidence to back it up. But the numbers bother me. Would 22,000 voters make such a flagrant mistake, however inexperienced? Would that many GOTV volunteers give such ridiculous advice?

Duval has a large black population (urban Jacksonville), but is Republican by majority, with Republican elections officials. Several things are noteworthy about Duval's elections officials. When directly asked by the county Democratic Party leader how many ballots had been rejected, the elections supervisor answered "Not many. 200 - 300." (There were 27,000 total, including the undervote.) Whether or not he deliberately lied (the official later claimed to have misunderstood the question), wouldn't fundamental fairness dictate that he notify the Democratic Party immediately on seeing the extraordinary reject numbers? The Democrats did not learn the real numbers until the story appeared in a local paper, just hours before the deadline to request a recount. [11]

There is now talk that Duval County may refuse to cooperate with the news media hand count. And Duval was the only large punchcard county that did not furnish computer records for the Washington Post study. That is at best an indication of where the County's sympathies lie, at worst, a determined effort to keep the evidence under wraps.

Palm Beach

I have heard two explanations for the Palm Beach overvote, both blaming the error on the butterfly ballot.

To be sure, the butterfly ballot was confusing, causing 3400 elderly Jewish voters to vote for Buchanan. That is an understandable error (I will call it the "Buchanan error"). But that is not overvote. I can understand how a voter could punch the wrong hole. I do not understand how ballot confusion could make a voter punch twice. Even if the voter punched for Buchanan and then realized that was the wrong hole, no reasonable person would think he could correct the mistake by punching the ballot a second time. Indeed, we see from the optiscan overvote that voters who make mistakes often write a note on the ballot, to let the counters know their true intent. We know from the hand count there were no notes on Palm Beach's overvotes.

The other explanation I have heard is that both the Buchanan and Gore holes lined up inside the Gore/Lieberman box, causing voters to think they had to vote for both president and vice-president. I believe I heard of one voter publicly admitting that she made this mistake. But I find it hard to believe that there were 19,000 voters thus misled. Certainly, no GOTV volunteer gave that advice.

In fact, Palm Beach furnishes a particularly damning piece of evidence against the "voter confusion" theory. Among the precincts with the highest overvote rates were several populated by Jewish retirees. [12] Say what you will about uneducated, first-time black voters, Haitian immigrants, etc. Jewish seniors are highly educated, second-generation citizens who have been voting for decades.

7) If 79,000 voters screwed up their ballots, changing the outcome of the most publicized and scrutinized election of our lifetimes, why haven't we heard from them?

The key evidence to me (aside from the sheer numbers) is the lack of voter testimonials, especially in Palm Beach.

Because of the publicity generated by the Buchanan error, which led to quick publicity for the overvote too, Palm Beach was swarming with reporters the day after the election. Both the Buchanan error and the 19,000 overvotes were front page news, and voters were asked to search their memories. We heard from many hundreds of voters who realized that they had made the Buchanan error. But even though there were far more overvotes than Buchanan errors, I have heard of only one Palm Beach voter coming to realize she had voided her vote by double-punching.

I had the opportunity to talk with an attorney volunteer for the Democratic Party who conducted voter interviews in Palm Beach after the election. He told me that there were "a few thousand" affidavits from voters who "weren't sure," but thought they might have punched their cards twice. I tried to probe him as to what mental process caused the voters to punch twice. What did they remember, that made them think they might have punched twice? The volunteer talked vaguely about the Buchanan error, but I could not elicit any reasonable story from him that would explain how the Buchanan error could cause overvote. He emphasized that none of the affiants had a definitive memory of overvoting; they were all "unsure."

I am ultimately persuaded that the voters who gave affidavits about overvoting were trying to give information helpful to the cause, or just looking for attention. The Democratic Committee website solicited testimonials from voters confused by the ballot, and specifically asked to hear from voters who might have been so confused that they punched their ballots twice. No doubt there were other outreach programs during this period of time, on radios, churches, etc. In short, the "unsure" affidavits were solicited, and are dubious evidence of actual voter error. Rather, the failure of the effort to elicit solid testimony of voter confusion is proof of its absence.

To be sure, there appears to be anecdotal evidence in Duval that some voters were confused by the vote-every-page instruction. Democratic elected officials, as well the NAACP and similar groups, have lent their credibility to the story. But the evidence of Palm Beach is more convincing to me, perhaps because I saw it with my own eyes, on TV, right after the election. There were dozens of Jewish seniors recounting how they had voted for Buchanan by mistake. But despite all the attention from the media, from the many black and Jewish leaders on the scene, I saw not a single face--black or elderly-come forward with personal testimony of overvoting. To say nothing of the silence from Miami-Dade, Broward, Hillsborough and Pinellas.

8) Miami-Dade: Either evidence that overvote can occur innocently, or a rebuttable presumption of fraud any time this county conducts an election

And then there's Miami-Dade.

There was significant overvote in at least one previous presidential election, 1988, and again the overvote occurred in black precincts. So it may be that punchcard overvote is not uncommon after all, at least in the black precincts of Miami-Dade. But also noteworthy about the election of 1988, there were strong suspicions of fraud in the US senate race, with the suspicion centering--once again--on the counties of Miami-Dade, Palm Beach, Hillsborough and Broward.

The statistics from the '88 Florida senate race are even weirder than Election 2000's. In that race, Democrat Buddy MacKay lost a close race which featured an unprecedented "dropoff rate" of 200,000 votes, predominantly centered in the four Democratic counties listed above. (The "dropoff rate" is the dropoff between total votes cast for president and total votes cast for senator.) To highlight the extraordinary absence of senate votes in those counties, in Hillsborough, even a referendum that drew the lowest number of votes statewide- 'Amendment Number 3: Assessment of High Water Recharge Lands' - recorded 10 percent more votes than the senate race. [13]

And here's déjà vu all over again: Can you guess the official explanation for the anomalous dropoff in the 1988 senate race? Ballot design! And what was the design flaw that caused so many voters to vote for president but skip the senate race? The senate candidates were listed on the bottom of the page.

The large presidential overvote in Miami-Dade could have been part of the scheme to steal the senate race. Reducing the total votes cast for president reduced the dropoff rate, thus mitigating that suspicious statistic. It is also possible that ballots might be tampered with on the precinct level, not with an eye to changing the election (1988's presidential race was a foregone conclusion), but as malicious mischief.

For a final gloss, recall that there was proven election fraud in Miami-Dade in one recent election (mayor elected with fraudulent absentee ballots), to say nothing of the skullduggery the nation observed in plain view during the election protest. (The canvassing board could not physically count 10,600 undervote ballots in four days? During the short-lived hand count ordered by the Florida Supreme Court, 8 judges completed half the task in one hour.)

9) It is recommended that voting machines tally the number of overvotes, specifically as a safeguard against ballot tampering. [14]

The idea is, you should investigate if you see suspicious numbers. I see suspicious numbers.

10) A girl can dream, can't she?

A month before the election, an amateur astrologer predicted that "secrets will be uncovered to Gore's advantage," and "Bush may be declared the president, only to have it taken away later." [15]

Allow me the fantasy.


1. It is difficult to get exact numbers for the overvote. The counts have varied as absentee ballots came in, and as machine recounts and hand counts recovered previously rejected ballots. Various tallies have been published as the election contest went on.

The best sources I have are (1) an Orlando-Sentinel survey, published 11/14/00, that shows total rejected ballots (both under and overvotes) county by county, and also indicates what voting method (punch, optiscan, a very few others) the county used and (2) a county-by-county survey of the undervotes that had not been counted as of the date the Florida Supreme Court ordered the recount. I saw the latter published in the San Francisco Chronicle. There was no data for a few of the counties in the undervote survey.

The Duval (22,000) and Palm Beach (19,100) overvote numbers have been reported in numerous articles, including many cited here. I have calculated the overvotes in the other 4 counties by subtracting the reported undervote totals from the reported rejected ballot totals.

To add all the numbers up:

The total number of rejected ballots is just under 180,000, according to the Orlando-Sentinel survey. Most published reports refer to the overvote as 110,000, and the undervote as 62,000. So there are clearly gaps.

According to my sources, there are 31,674 total rejected optiscan ballots, of which 6,671 (perhaps more) were undervotes. Thus, about 25,000 of the overvote is optiscan.

The Washington Post study, which included all 6 of the suspicious counties except Duval, found 57,000 punchcard overvotes. Duval is another 22,000, for a total of 79,000. In addition to the 5 suspicious counties, the Post study included Highlands, Pasco and Marion counties (all punchcard). These three counties had combined numbers (my sources, the Post did not break it down) of 8,271 rejected ballots, 4,710 undervotes, and therefore 3,561 overvotes.

Finally, there were 17 small punchcard counties not included in the Post study, for a total of about 20,917 total rejected ballots, 11,370 (or more) undervotes and therefore 9,547 overvotes.

So the figures about add up: 76,000 - 79,000 suspicious punchcard overvotes, 25,000 (probably, mostly) innocent optiscan overvotes, and 5,000 - 10,000 hard to peg. The 5,000 - 10,000 are at least partially undervote, and may include punchcard overvote that is accounted for by normal error (the 0.1% normal rate) or absentee ballots with their own peculiar problems, or may not be overvote at all (the discrepancy between the 180,000 figure for total rejected ballots, and the reported undervote (62,000) and overvote (110,000) totals).

2. "Multiple voting hurt Gore in Florida," Dan Keating, Washington Post, Jan. 27, 2001.

See also "State's double-punched ballots far outnumber those elsewhere," Andrea Robinson, Geoff Dougherty and Curtis Morgan, Miami-Herald, Nov. 18, 2000. In addition to Palm Beach and Duval, the piece mentions Miami-Dade, Broward, Hillsborough and Pinellas as having abnormally high overvote rates, predominantly from Democratic precincts.

There may be one more suspicious punchcard county, Pasco, one of the counties included in the Washington Post study. Pasco had total rejected ballots of 3,917, of which 1776 were undervote, for a total overvote of 2,141. Pasco is adjacent to Pinellas, and may share the demographics of its Democratic-voting neighbor, making it a target for tampering.

3. "The disappearing ballots of Duval County," Eric Boehlert, Salon.com Nov. 13, 2000.

4. "The State's double-punched ballots far outnumber those elsewhere," supra

5. "The disappearing ballots of Duval County," supra

6. "The disappearing ballots of Duval County," supra. "Democrats Rue Ballot Foul-Up in a 2nd County," Bonner and Barbanel, New York Times, Nov. 16, 2000.

7. Orlando Sentinel, Dec. 19, 2000, Untitled? piece by David Damron, Ramsey Campbell and Roger Roy. And see "Board discovers that over-votes could present a problem," Kathleen Chapman, Palm Beach Post, Dec. 6, 2001.

While for the most part the optiscan overvote looks innocent to me, there is some question about Gadsden, a rural county that is nonetheless a majority black. Gadsden saw about 10% of its votes lost to overvote, almost 2,000 ballots, and one group suspected about 500 Gore votes were tampered with. See "The Florida recount continues," Salon.com, Jan. 18, 2001.

It is hard to tell what went on in Gadsden-innocent voter error, part of a larger fraud conspiracy, even tampering at the precinct level just for malicious mischief. Gadsden gets specific mention in "Florida's Typical Election Day is Often a Nightmare," Mark Fineman and Lisa Getter, LA Times, Nov. 12, 2000: "Gadsden, the only one of Florida's 67 counties with a black majority, has a long history of vote fraud, which many residents say is grounded in racism and carried out through intimidation."

8. "Board discovers that over-votes could present a problem," supra

9. Palm Beach recovered about 1000 ballots in its hand count. I recall hearing recovered votes reported only in the hundreds as the hand count results were made public on TV. Most, if not all, appear to have been undervotes.

Broward confirms that rough estimate. Broward, with roughly the same number of total ballots cast as Palm Beach and a more liberal standard, recovered only about 1500 ballots in its hand count. Since Broward finished its count before the extended deadline, these figures are available in the state-certified totals.

And it appears that the Washington Post study of Palm Beach County's overvote confirms that there were no "write-in" overvotes in Palm Beach.

10. "Democrats Rue Ballot Foul-Up in a 2nd County," Bonner and Barbanel, New York Times, Nov. 16, 2000. "Blacks Cry Foul In Duval County," Robert Saladay, San Francisco Chronicle,

11. "The disappearing ballots of Duval County," supra

12. "Nearly half of the tossed ballots came from black, elderly precincts." Stephen Kiehl and Elliot Jaspin, The Palm Beach Post, Nov. 18, 2000. See also "The state's double-punched ballots far outnumber," supra

13. Carl Bernstein, "Problems in Florida 2000 Vote Echo 1988 Senate Race," Salon.com Dec 6, 2000; David Beiler, "A Short in the Electronic Ballot Box," Campaigns and Elections July/August 1989

14. "A Short in the Electronic Ballot Box," supra

See also "Not the First Case of Ballot Box Backlash," Rinker Buck, The Hartford Courant, Nov. 12, 2000. "Polling experts are also critical of the antiquated system of punch-card ballots because of their propensity for abuse. Precinct workers in the past have been charged with punching extra holes in the ballots, but those charges have been exceedingly difficult to prove in court."

15. "Personals" column, Leah Garchik, San Francisco Chronicle, Nov. 14, 2000

Katharine Harris with Tweezers: Postscript


The evidence continues in.

1) A recent study by the Orlando-Sentinel casts further doubt on the "voter confusion" story

The Orlando-Sentinel recently published a study of the rejected ballots of "Florida's most error-prone counties." "Small counties wasted more than 1,700 votes," Orlando-Sentinel Jan 28, 2001. This study focused on 15 small rural counties, all optiscan. Without knowing many specifics, I believe the prevailing demographic of these error-prone voters is lower middle class.

The Orlando-Sentinel study found 4,268 overvotes that exhibited what the paper called the "broken ballot" error. Florida had a very long presidential slate (10 candidates and a write-in) and many counties broke the list so it ran two pages on the ballot. The broken-ballot voters cast a vote on both pages, apparently believing the candidates listed on the second page were in a separate race.

"Broken ballot" is the ballot error that allegedly tripped up Duval's 22,000 overvoters. So the 22,000 overvotes of Duval are innocent, right? Voters do make that mistake, after all.

But look closer. 4,268 broken ballot overvotes is 2.27% of the Orlando-Sentinel's 188,000 total. (The article said the 15 counties represented 4.6% of Florida's 6,100,000 ballots (280,600), and Lake County (92,000) did not have a broken ballot.) Duval's broken ballot overvotes were 6.42% of its total ballots cast (22,000/291,000). That is three times the rate experienced by Florida's 14 most error-prone rural counties.

But even that is not a true measure. If we can posit that the 14 error-prone counties are relatively homogeneous, if only half of their 188,000 voters fall into the less well-educated category, that is still a broken ballot error rate of under 5%. By contrast, Duval has widely divergent population groups, voting in some black precincts as much as 93% for Gore, while still having enough Bush strongholds to vote majority Republican. 291,000 is not the right denominator. The error-prone voter group that is responsible for Duval's broken ballot overvotes is the black precincts of Jacksonville. There, the spoiled ballot rate run from 10% to as high as 30% in some precincts.

I do not know if Miami-Dade, Broward, Hillsborough or Pinellas used a broken ballot. The Post study did not point out any such pattern from its review. Palm Beach, of course, used the butterfly ballot. There is still no logical explanation (to my mind) for 19,000 voters to double-punch a butterfly ballot.

The Orlando-Sentinel study also found that more ballot errors were made by Gore voters than Bush's. That bears out the theory, the underprivileged that make up a good part of Gore's constituency are more prone to error.

But once again, look closely. The Orlando-Sentinel found 2416 Gore votes lost to the broken ballot error, to 1852 for Bush. Total overvotes where the voter's intent was apparent ran 1,033 Gore to 677 Bush. That's an overall error rate of 1.5:1. In stark contrast, the double-punched ballots of the Washington Post study, the suspicious punchcard counties, ran 4:1 against Gore.

2) Let's talk about those 5,800 Gore/Bush votes again.

The Washington Post reported that fully 5,800 of the 57,000 overvotes in its study were punched for both Gore and Bush. The Post study did not include Duval's 22,000 overvotes, so there may be more than 5,800.

I knew the 5,800 Gore/Bush votes were suspicious. But it was not until I read Roger Miller's piece at legitgov.org, "Coup d'etat 2000," that I realized these were ballots pre-punched for Bush.

I recall (have not yet been able to find again) reports of black voters claiming they were given pre-punched ballots at the polls. At the time, I thought little of it--What's the good of a scheme that nets only a couple dozen votes? The perpetrator wouldn't dare pull this trick with thousands of ballots, and risk having it noticed. But on reflection, most rigged ballots would probably go undetected. Punchcard voters do not scrutinize their ballots during voting, as optiscan voters do, and chad-holes are small. And if the prepunched ballots are spread over many precincts, with voters coming in over the course of several hours, if one or two do notice a pre-punched ballot, it will go unremarked in the general din. Which is what happened.

One final observation. Apparently, NOT ONE SINGLE VOTER in the Orlando-Sentinel study, the most error-prone voters in the state, made the mistake of voting for both Bush and Gore.

3) Democratic ballots from Palm Beach were among the last to be counted election night.

"A rush of Democratic votes from Palm Beach County then [after 4am election night] brought the race to an effective tie on the Florida election website." Michael Hammerschlag, Mediachannel.org 1/03/01

So our 19,100 Palm Beach overvotes were among the very last ballots to be counted (or reported) that late election night.

Here's a theory.

The Republicans had a number of tricks, but hoped to get away with as little risk as possible. They hoped the voter roll purges, pre-punched ballots and other tricks would give them a good head start. They set up Duval from the beginning, printing the fake sample ballot in the newspaper, maybe putting some plants among the Democratic Get-out-the-Vote volunteers. Duval would have an abnormally high rate of overvote, but there would be an explanation.

They hoped to get in under the radar screen with their hits to the other Democratic counties. And they have largely succeeded. Who among the general public knows about the overvote in Miami-Dade, Hillsborough, Broward and Pinellas? Have these statistics been reported even once on national TV or press? The right wing owns the media.

But Palm Beach was a gift. A large, heavily Democratic county that by chance experienced a spate of true ballot confusion. A second cover story handed to the Republicans ready-made, an opportunity to go for a spoiled ballot number that would be too large to escape notice, but still explainable.

And so they held back the Democratic Palm Beach ballots as long as possible, waiting to compute how many they needed to spoil.

4) The demeanor evidence.

As this story unfolded, perhaps we were all percipient witnesses to one last piece of evidence, as it played out on our TVs. The demeanor evidence.

There is the boil, of course. George Bush's corrupt soul festering to the surface.

But did any one else notice Bush Senior? First, the Gala Dinner at the White House, the night after the election, all the living presidents gathered to celebrate the 100th anniversary of the White House. There was George Senior in his tux, on his way to the fete as the newsguys caught up with him. Did he seem stressed to you? He couldn't wait to get away from the cameras. Then a few weeks later, he was interviewed on TV (I can't remember by whom-Barbara Walters, or the like). Asked about his son the President-Apparent, Dad stumbled through something with the word "pride" in it, then made it clear he did not want to talk about the election, or his son. I swear I remember him saying something like "I want to talk about fishing."

Finally, there was this TV moment that a friend recounted to me. Can any one confirm it? Reportedly, the family Bush was caught on camera, at dinner, the moment the networks called Florida for Gore. Jeb left the table abruptly. What was the urgent errand?