Still sitting in the loader, he tried to hide low behind its steel door.
"I remember fighting for breath and was worried that it was gas.''
He tried to reach his self-rescuer, but then his memory goes blank. He does not believe he had time to get it on.
His next memory is vomiting in the ambulance, and in hospital, the smell of smoke lingered in his hair.
Mr Smith does not pass judgment in his statement, but he does note the tag system had become "fairly inefficient''.
Confusion in the mine
Former general manager Doug White, giving evidence to the Royal Commission of Inquiry in Greymouth earlier today, painted a picture of confusion over who was actually in charge immediately after the November 19 blast and who was underground at the time, as they repeatedly rang phones which no one answered.
There was disagreement over whether to run the conveyor and when to bring in a machine to suck the air out, with the Department of Labour and police in Wellington categorically ruling out those options.
Running the GAG machine, to prevent further explosions, was an acknowledgement the 29 men were dead because once it was running no one could breathe underground, he said.
On the night of the blast, Mr White debated with individual Mines Rescue staff who were frustrated they could not enter the mine. However, Mines Rescue did not make a formal request to go underground until the following Wednesday, he said.
The self-rescuers, fastened to the belts of everyone underground, had 30 minutes of oxygen, and the fresh air base had 50-minute ones.
Every half hour, mine staff opened the intercom system so anyone alive inside the mine would have been able to hear them speak. All phones were called. Only one did not ring; it returned static.
They urgently needed to know what gas levels were underground, and lowering a hand-held monitor on the end of a fishing line down the ventilation shaft was considered, but ruled out as people would still have to enter the fan house.
The fishing rod idea has been heavily criticised by other submitters.
"The fishing rod was my fly rod - (but) it would have put a person at particular risk (without breathing apparatus at the top of the shaft)."
Instead, St John suggested a stomach pump from an ambulance be lowered down the shaft, which sucked a sample from about 40m down.
Mr White said they had to think outside the square. Mines Rescue also lowered a radio down in a bucket, with a lamp.
Gas samples flown to the Mines Rescue station at Rapahoe for analysis indicated there had been a fire underground.
Meanwhile, staff were struggling to determine who was actually trapped underground. Some name tags had not been removed from the board, while others had not placed their tags.
A temporary seal at the portal was discussed by Mr White and Mines Rescue to reduce air flow, but the Department of Labour refused: "It was not up for discussion."
Police Superintendent Gary Knowles of Nelson became incident controller but there were no direct conversations about who was in charge, Mr White said.
There were initially frustrating delays as decisions had to be approved by people in Wellington.
Mr White wanted to run the conveyor to determine how damaged it was further in the mine. If it worked, the blast may not have been too large, but he was told he could disturb evidence.
He disagreed with the decision not to bring in the GAG machine on November 22 and was proved right "as the mine exploded a further three times before the GAG was deployed".
On November 24, he was called and told Mines Rescue was preparing to deploy. He then got a second call, and was told the mine had again exploded. They determined no one could still be alive.
The next day he was shown an image from inside the mine. It showed one open self-rescuer box, which suggested someone may have survived the first blast. He asked people not disclose the image until further analysis had been done.
"I had a sinking feeling ..."
The hearing continues.
