Scenario one: If North Korea fired a nuclear-armed missile that devastated an American city, how would the U.S. government respond? The state-sponsored attack would fit within the Cold War paradigm; therefore, the certain American response would be an overwhelming retaliation aimed at destroying Pyongyang, Kim Jong Il’s nuclear and missile programs, and North Korea’s million-man army. Such a response would result in enormous collateral damage, killing millions of North Koreans. Despite reservations about the morality of such a response, those who established the Cold War nuclear doctrine recognized – and accepted – the unintended deaths of millions of innocents. Whoever occupied the White House during such a nuclear attack would understand this also.

Scenario Two: If North Korea were discovered to have sold a nuclear bomb to al-Qaeda, which smuggled the weapon into the United States and used it to destroy a city, how would the U.S. government respond? As things stand today, it would, again, launch an overwhelming retaliation against North Korea.

Scenario Three: Now imagine a nightmare in which an unknown terrorist group smuggles a nuclear bomb into the United States and detonates it in Manhattan or Los Angeles, killing hundreds of thousands of Americans – but the U.S. government is uncertain where the bomb came from and who delivered it.

The logic of deterrence requires a deterrer and an identified deterree. In Henry Kissinger’s formula, “Deterrence requires a combination of power, the will to use it, and the assessment of these by the potential aggressor. Moreover, deterrence is the product of those factors and not the sum. If any one of them is zero, deterrence fails.” An adversary with no known return address might calculate that it could escape retaliation. In scenario three, as an American city lies smoldering, how would the U.S. respond if Osama bin Laden announced that he had ordered the action? If the U.S. government knew where bin Laden was, it would already be there.

The terrorist nuclear weapon that destroyed an American city, or the material from which it was made, would almost certainly have originated in a state. Fortunately, producing highly enriched uranium or plutonium – the key ingredients of a nuclear bomb – is a multibillion-dollar, multiyear undertaking, something beyond the capability of nonstate terrorists. After a nuclear September 11, the U.S. government would be eager to exact swift vengeance. But against whom? It must first determine who was responsible for the attack.

Did a state make and deliver the bomb or willingly sell it to terrorists? Or was the weapon stolen from a state that had no intention of losing it? (Indeed, the material could even have come from American stockpiles.) In the first case, identification would provide a bull’s-eye for retaliation. In the second, the certainty that the weapon was stolen from, say, a Russian or Pakistani arsenal would generate a demand for the immediate global lockdown of all nuclear weapons and materials.

Today, the technological prerequisite for rethinking the unthinkable is nuclear forensics: the ability to identify a bomb’s source from radioactive debris left after it explodes. Building on Cold War techniques, the Pentagon has developed new methods for collecting samples from ground zero, measuring data such as isotopic ratios and the efficiency of the fuel burn in the detonation, and comparing that information to known nuclear data to determine the origin of the materials.

Much about foreign nuclear materials is already known, collected from Cold War military tests, commercial transactions, scientific exchanges, and covert means. But the recent debate about the uranium hexafluoride that Libya turned over after renouncing its nuclear-weapons program illustrates gaps in our knowledge. The U.S. intelligence community believes that North Korea was the source; however, according to informed press reports, it reached that determination by the process of elimination – not by identifying the uranium directly, but rather by ruling out other known sources.

The goal of a robust nuclear forensics program must be to identify nuclear material definitively and quickly. The 2002 study by the National Research Council Making the Nation Safer: The Role of Science and Technology in Countering Terrorism concluded that “The technology for developing [postexplosion nuclear attribution] exists but needs to be assembled, an effort that is expected to take several years.” Establishing this capability should be a top priority, since effective deterrence requires convincing potential perps that the United States will be able to identify the culprit.

Graham Allison, the director of Harvard University’s Belfer Center for Science and International Affairs, is the author of Nuclear Terrorism: The Ultimate Preventable Catastrophe.