Without aerial refueling, the global reach of U.S. airpower would be severely limited. However, nearly 90 percent of the tankers in the current refueling fleet -- 453 planes -- are Eisenhower-era Boeing KC-135s that average 50 years of age.

Nobody knows how much longer these planes can operate before metal fatigue, corrosion and parts obsolescence cause a disaster. What we do know is that if replacement were to begin tomorrow, it would take at least 20 years to modernize the whole fleet. The U.S. Air Force has been trying to acquire new tankers since the decade began but has encountered a series of obstacles. The latest obstacle is Defense Secretary Robert Gates.

During the Bush years, Gates presided over a mismanaged effort to award the contract for the first 179 next-generation tankers to one of two teams led by Boeing and Northrop Grumman. A successful protest by the losing Boeing team forced the government to start over.

Gates wants to do another winner-take-all competition while correcting the errors of the first round. But since the U.S. government has failed to consult with industry teams in formulating its requirements and strategy for the re-competition, the outlook is for another protest by whichever team loses.

Rep. Jack Murtha, D-Pa., has a better idea: split the contract between both teams, awarding a larger number of planes to the team that shows the best performance and price discipline. Gates says that will cost too much, but he's wrong. The Murtha plan will save money over the long run while avoiding further delays.

The most compelling reason for dual sourcing is operational. Boeing and Northrop Grumman are both offering planes that would cost less to operate than existing tankers while transporting more fuel, but the Northrop Grumman KC-45A plane is considerably larger than the Boeing KC-767.

What that means in operational terms is that each plane could be better suited to specific contingencies than the other. For instance, the bigger Northrop Grumman aircraft might be a better fit for the long distances of the Western Pacific, while the smaller Boeing aircraft might be a better fit for the shorter distances and denser basing network of the Middle East. You can't really say which will prove more useful without knowing where future wars will be fought, so the smart option is to buy a mix of both.

Gates doesn't like that option because he says it would cost $7 billion to $14 billion more than the winner-take-all approach. Not only would the government have to pay to develop two different tankers, but it would have to buy more tankers each year to sustain competition between two production lines.

However, Gates ignores the huge savings that would result from retiring old tankers earlier because their replacements arrive sooner. The old tankers are very expensive to operate and maintain because of their advanced age. They will require new skins and other costly modifications in the near future that could be avoided if new tankers were acquired faster, which the Murtha plan would do.

Gates also overlooks the savings that result from having two teams competing throughout the life of the program for tanker awards. A study by former Pentagon acquisition chief Jacques Gansler found that in winner-take-all competitions, costs tend to increase markedly after the award, whereas when two teams remain competitively engaged costs tend to go down.

Under the Murtha dual-sourcing plan, each team could count on getting a minimum award of planes, but the team that offers better price and performance would get a bigger share of the planes. So when Gates says dual-sourcing would cost more, he means on his watch. Over the long term, it would save a lot of money.

(Loren B. Thompson is chief operating officer of the Lexington Institute, an Arlington, Va.-based think tank that supports democracy and the free market.)