WASHINGTON - Excerpts of the internal draft report by the staff of the International Atomic Energy Agency (IAEA) published online last week show that the report's claims about Iranian work on a nuclear weapon is based almost entirely on intelligence documents which have provoked a serious conflict within the agency.

Contrary to sensational stories by the Associated Press and The New York Times, the excerpts on the website of the Institute for Science and International Security (ISIS) reveal that the IAEA's Safeguards Department, which wrote the report, only has suspicions – not real evidence - that Iran has been working on nuclear weapons in recent years.

The newly published excerpts make it clear, moreover, that the so-called "Alleged Studies" documents brought to the attention of the agency by the United States five years ago are central to its assertion that Iran had such a programme in 2002-03.

Whether those documents are genuine or were fabricated has been the subject of a fierce struggle behind the scenes for many months between two departments of the IAEA.

Some IAEA officials began calling for a clear statement by the agency that it could not affirm the documents' authenticity after the agency obtained hard evidence in early 2008 that a key document in the collection had been fraudulently altered, as previously reported by this writer. As journalist Mark Hibbs reported last week in Nucleonics Week, opposition to relying on the intelligence documents has come not only from outgoing Director General Mohamed ElBaradei but from the Department of External Relations and Policy Coordination.

Since September 2008, however, the Safeguards Department, headed by Olli Heinonen, has been pressing for publication of its draft report as an annex to a regular agency report on Iran.

Heinonen leaked the draft to Western governments last summer, and in September it was leaked to the Associated Press and ISIS. That has generated sensational headlines suggesting that Iran can already build a nuclear bomb.

The draft report says the agency "assesses that Iran has sufficient information to be able to design and produce a workable implosion nuclear device". But other passages indicate the authors regard such knowledge only as a possibility, based on suspicions rather than concrete evidence.

It says the "necessary information was most likely obtained from external sources and probably modified by Iran". But it cites only the 15-page "uranium metal document" given by the A. Q. Khan network to Iran when it purchased centrifuge designs in 1987.

"Based on the information in the document," it says, "it is possible that Iran has knowledge regarding the contents of a nuclear package."

The IAEA "suspects" that the 15-page document was part of "larger package that Iran may have obtained but which has not yet come to the Agency's attention", according to the leaked excerpts.

But that document only outlines procedural requirements for casting uranium into hemispheres, not the technical specifications, as the IAEA report of Nov. 18, 2005 noted. No evidence has ever surfaced to challenge the Iranian explanation that Khan's agents threw in the document after a deal had been reached on centrifuges in an effort to interest Iran in buying the technology for casting uranium.

The IAEA affirmed that it has found no evidence that Iran ever acquired such technology.

The only external "nuclear package" ever reported to have been provided to Iran is a set of flawed technical designs for a "high-voltage block" for a Russian-designed nuclear weapon, which was slipped under the door of the Iranian mission in Vienna by a Russian scientist working for CIA's Operation Merlin in February 2000.

Another far-reaching claim in the draft report is that the IAEA "has information, known as the Alleged Studies, that the Ministry of Defence of Iran has conducted and may still be conducting a comprehensive programme aimed at the development of a nuclear payload to be delivered using the Shahab 3 missile system."

It does not explain how the "Alleged Studies", which are documents on work done in 2002 and 2003, could have any bearing on whether Iran is now conducting work on nuclear weapons.

Using the same language found in published IAEA reports, the draft suggests that the Alleged Studies intelligence documents represent credible evidence. "The information, which has been obtained from multiple sources, is detailed in content and appears to be generally consistent," it says.

But that characterisation of the intelligence first shown to the IAEA by the United States in 2005 has been contested by sceptics in the agency. A senior IAEA official familiar with the documents suggested in an interview with IPS last month that the claim of "multiple sources" may be misleading.

Given the existence of "intelligence sharing networks", the official said, "one can't rule it out that one organisation got the intelligence and shared it with others." That would explain the reference to "multiple sources consistent over time", he said.

The initial U.S. account, according to the official, was that the documents came from the laptop computer of one of the Iranian participants in the alleged nuclear weapons research programme. Later, however, that account was "walked back", he said.

"There are holes in the story," said the official.

The introduction by ISIS to the excerpts from the report, evidently based on conversations with the IAEA personnel, confirms that the documents did not come from Iran on a laptop computer, as U.S. officials had claimed in the past. It suggests that the documents were smuggled out of Iran as "electronic media" by the wife of an Iranian who had been recruited by German intelligence and was later arrested.

That new explanation is highly suspect, however, because an intelligence agency would not confirm the identity of one of their agents, even if he were arrested. Asked about the ISIS account, Paul Pillar, who was national intelligence office for the Middle East when the "laptop documents" surfaced, said it "sounds unusual".

The draft report also argues that the information in the documents is credible, because it "refers to known Iranian persons and institutions under both the military and civil apparatuses, as well as to some degree to their confirmed procurement activities".

But the senior IAEA official cast doubt on that claim as well. The names of people working in the relevant Iranian military and civilian organisations are readily obtainable, he observed. "It's not difficult to cook up such a document," the official told IPS.

The draft paper states that the agency "does not believe that Iran has yet achieved the means of integrating a nuclear payload into the Shahab 3 delivery system with any confidence that it would work".

That statement hints at the fact that the reentry vehicle studies were found to have serious technical problems. The senior IAEA official told IPS that the Sandia National Laboratories, which ran computer simulation analyses of the plan, not only found that none of them would have worked, but had expressed doubt that they were genuine.

The paper makes an indirect reference to a plan for a bench-scale facility for uranium conversion, but does not mention that it had several technical flaws, as acknowledged by Heinonen in a February 2008 briefing for members.

Nor do the draft report's conclusions deal with the fact, confirmed by the senior IAEA official to IPS, that none of the intelligence documents have any security markings, despite the fact they are purported to be part of what presumably would have been Iran's most highly classified programme.

*Gareth Porter is an investigative historian and journalist specialising in U.S. national security policy. The paperback edition of his latest book, "Perils of Dominance: Imbalance of Power and the Road to War in Vietnam", was published in 2006.