The military report said that surveillance footage of the beach on Friday last week showed that the explosion could not have taken place before 4.57pm, nine minutes after the military says the last shell was fired by land artillery in the area.

It said the footage showed normal behaviour on the beach between 4.54pm and 4.57pm, and that other film showed the ambulances arriving at 5.15pm.

It also said for the first time this weekend that a call was made by Palestinian security at 5.12pm to Israel asking them not to shell in the area, because ambulances were attending an emergency on the beach.

But the written hospital admission registration book at Kamal Odwan hospital in Beit Lahiya, where the dead and some of the most seriously wounded victims of the explosion were taken, times the arrival of the first eight victims as being at 5.05pm, 10 minutes earlier than when the army said last week the first ambulances arrived at the scene.

An ambulance driver based at the hospital, Khaled Abu Sada, who said he was the second to arrive at the scene, said he received mobile telephone confirmation of an emergency from a colleague working for the Palestinian Red Crescent at 4.45pm or 4.46pm, set off at 4.50pm, arrived at the scene at around 5pm and returned to the hospital after picking up one dead child and three women, two of whom were close to death, at 5.10pm.

Human Rights Watch (HRW) . . . whose senior military analyst, Marc Galasco, a former Pentagon battle damage expert, claimed last week that "all the evidence" pointed to the deaths being caused by a 155mm landbased artillery shell . . . said it had examined "the computergenerated record" from the same hospital "which documents the blood test of a victim from the beach incident being taken at 5.12pm on 9 June".

HRW also referred to admission records, and added: "If the records are accurate, based on the time needed to dispatch an ambulance and drive from the hospital to the beach and back, this suggests that the fatal explosion took place at a time when the IDF said they were firing artillery rounds."

The agency said both sets of records "call into question" the account that ambulances responding to the incident did not reach the beach until 5.15pm that day.

The register shows the body of Ali Ghalia, the dead father of the family, as having been brought to the hospital at 5.20pm, and his daughter Huda, who was seen distraught at her father's death on television footage broadcast around the world, at 5.40pm. That is consistent with her still being at the scene when local television cameras arrived.

Estimates . . . including those by the Israeli army . . . of the time for an ambulance journey to the beach are between five and 10 minutes. Abu Sada estimated that, after he had waited in vain for a nurse to join him and he left at 4.50pm, the round trip, including picking up the wounded and dead, took around 20 minutes. He said that on the outward journey he had reached speeds of up to 130kph (80mph). "I flew, " he said.

There is no proof that the written admission records were not subject to human error or even falsification. But the request to inspect them was not prompted by any Palestinian official and the timings of the entries are in sequence with those preceding and succeeding them.

There is no sign that they were tampered with, and there was no obvious motive at the time they were written to record an inaccurate time, since the timings of the explosion were not then an issue.

A complication is that a later record of all the victims of the attack by the Kamal Odwan director's office, seemingly hastily written on a word processor and printed out, marks most of the first arrivals as coming in at 5.10pm. But this record is peppered with obvious errors . . .reporting the body of Ali Ghalia, for example, as arriving at 3.40pm.

Jaber al-Attar, a doctor who was on emergency duty when the patients were admitted, said that the written registration book was always completed first and universally taken as the correct record by clinicians as well as officials.

Attar echoed other doctors by saying the admission times written in the registration book were likely to be slightly later than the actual time of arrival. He said that the entries were invariably made a few minutes after the patient had been admitted to emergency.

"It is between two minutes and a maximum of five. The person writing the entry writes the time when he is writing it, not when he thinks the patient came in."

The Israel Defence Forces spokesman remains adamant that its investigation was conclusive. Captain Jacob Dallal said on Friday that new information had brought the 5.12pm co-ordination call to light. He added: "It is completely unreasonable to assume that the call was made half an hour after [the incident] started."

He said that the timed film from rotating surveillance made by the army was bound to be more reliable than the recollection of eye witnesses.

"I have never known a more detailed and thorough investigation than this. Our whole objective has been to find out the truth of what happened and if it had been our shell to take steps to ensure it did not happen again, " he said.

The office of Mahmoud Abbas, the Palestinian president, was unable immediately to establish the timing of any co-ordination call. But it confirmed that President Abbas's office had made it clear that it was ready to cooperate fully with any independent international investigation into the deaths.

TWO SIDES OF THE STORY

WHAT THE ISRAELIS SAY 4.48: Last shell fired at beach.

4.54-4.57: Normal activity on beach.

5.12: Palestinians ask Israelis to stop shelling the beach because of medical emergency.

5.15: First ambulance arrives on beach.

WHAT THE PALESTINIANS SAY

4.45-4.46: Palestinian paramedic Khaled Abu Sada receives emergency call.

4.50: Abu Sada leaves for the beach.

5.00: His ambulance arrives.

5.05: Victims arrive at the hospital in first ambulance.