As students of the process are beginning to observe, President Bush and Sen. John Kerry could end up in an electoral tie. There are, after all, an even number (538) of electoral votes available, and 2000 reminded us how close it can get: After the recounts were complete, the candidates were separated by a mere four electoral votes.

More than two months before the election, it is too early to use polling data to attempt to predict with any accuracy whether the election will be as close as most people seem to think.

The very situation that makes an electoral tie a real possibility — an almost equally divided electorate — also makes such a result difficult to predict. And the small shifts among voters that will ultimately determine the results stubbornly resist capture.

Trying to avoid simple guesswork, we wondered if it would be possible to calculate the probability that the electoral vote might end in a tie. Given the remarkable events that would ensue after such a tie, the temptation to do the math was irresistible. So here goes.

Based on the 2000 election results, Bush could be considered the favorite in about 22 states that have a combined total of 190 electoral votes. Kerry could be given the edge in 11 states, with 168 electoral votes.

This scenario leaves 17 "swing" states, accounting for 180 electoral votes, in play this time around: Arizona, Arkansas, Florida, Iowa, Maine, Michigan, Minnesota, Missouri, Nevada, New Hampshire, New Mexico, Ohio, Oregon, Pennsylvania, West Virginia, Washington and Wisconsin.

The magic electoral number is 269; therefore, any combination of votes among those 17 states that yields a total of 101 more votes for Kerry (or, conversely, 79 more votes for Bush) will do the trick. Because there are only two potential outcomes in each of these 17 states (Kerry or Bush — sorry, Ralph Nader), there are 131,072 (2 to the 17th power) possible formulations of the vote, ranging from all the states preferring Kerry to all of them supporting Bush.

Using a computer program, we calculated that there are exactly 1,969 different scenarios under which the swing-state votes actually yield a tie. This may sound large, but it amounts to only about a 1.5 percent chance. If that seems too unlikely to worry about, think about this: The probability of the electoral count turning out the way it did in 2000 (271 to 267) was only 1.4 percent.

We are dealing with probabilities, not real voters going to real polls. But if we use 2000 as a baseline, we find that a switch in as few as two swing states could result in both candidates being frozen at the magic number of 269. For example, if Nevada and New Hampshire (Bush states in 2000) switched allegiances in 2004, a tie would result.

What then? According to the Constitution, if no candidate receives a majority of electoral votes, the House of Representatives is to immediately choose a president from among the top three candidates. The vote is done by delegation; so, for example, Ohio's 20 House members would cast a single vote for president. Because Republicans have a majority in the Ohio delegation, it's probably safe to assume the state would vote to re-elect Bush.

In the current House delegations, Republicans control 30 states and Democrats 16, with 4 states evenly divided. But the current House would not conduct the vote. According to federal law, the electoral votes are actually counted (before a joint session of Congress) Jan. 6. The Constitution has the new Congress taking office three days earlier, on Jan. 3.

Nevertheless, given that there are few competitive House races nationwide, it's unlikely the Republicans will lose control of the majority of the state delegations on Nov. 2.

Therefore, in the event of an electoral college tie, it's all but certain Bush would be selected to serve for another four years.

Senate selection

The proceedings in the Senate might be even more interesting. The Constitution, in its complicated wisdom, gave the Senate the role of selecting the vice president.

Unlike the House, the Senate votes not by state delegation but by individual senator.

It has 51 Republicans, 48 Democrats, and 1 independent (who caucuses with the Democrats). Obviously, the loss of one seat by the Republicans would result in a deadlocked Senate, something that occurred for a time in 2001 until Vermont Sen. Jim Jeffords left the Republican Party and declared himself an independent.

If a 50-50 Senate were to cast a party-line vote for vice president, who would break the tie? Well, the Constitution calls for the sitting vice president, as president of the Senate, to break ties. There might be a debate over whether this tie-breaking role ought to be exercised during a nonlegislative vote, but an argument could be made that Vice President Dick Cheney would be constitutionally empowered to place himself back into the No. 2 slot.

Bush and Edwards?

This is largely uncharted territory. It's been 180 years since Congress was faced with the task of acting in the wake of an election that did not produce a majority vote.

Back in 1824, the House chose John Quincy Adams over Andrew Jackson, the candidate who had won the popular vote (although at the time, every state did not hold a popular vote for president).

Perhaps the correct precedent is even older. In 1796, John Adams, a member of the Federalist Party, was elected president. His vice president was Thomas Jefferson, leader of the Federalists' opponents, the Republicans. (This occurred, in part, because before the 12th Amendment, the candidate with the second-highest number of electoral votes became vice president.)

Jefferson would defeat Adams four years later in an election that was itself fraught with controversy and which resulted in our first electoral tie. The tie — between Jefferson and Aaron Burr — went to the House, which voted 36 times before declaring Jefferson the winner.

Far-fetched as this may sound, we could end up with Bush being re-elected president and Democrat John Edwards serving as his vice president.

If the Democrats gain two Senate seats in November, they would hold a majority. In the event of a tie, there would be nothing to prevent them from selecting Edwards over Cheney. Indeed, they would have every reason to do so, because, as president of the Senate, Edwards would then be able to protect Democratic control of that body in the event that his party somehow lost a seat (as happened to the Republicans when Jeffords bolted).

Faithless electors

Another possibility would involve a so-called faithless elector.

Realizing that a tie vote would send the election to Congress, one elector might simply change his or her vote. While unusual, it's happened in the past.

And imagine the pressure on a single elector, particularly when his or her vote might well lead to a chain of events that would once again result in a White House occupied by someone who had failed to secure a majority — and perhaps even a plurality — of the popular vote.

From the perspective of those who wrote the Constitution, however, such an elector would be considered faithful, not faithless. The framers expected those who would ultimately select the president to be free agents, not bound by a popular vote or by the choice of organized political parties.

Indeed, they spent quite a bit of time at the convention considering what to do in the expected event that, without organized parties or a candidate with the reputation of a Washington, no individual would garner a majority of the electors.

Yet so far, this has happened only once in our history. Perhaps we should be glad that George Mason, a delegate to the Constitutional Convention, was wrong when, on Sept. 4, 1787, he predicted that such a result would occur "19 times out of 20."