I think Ford originally purchased machines that were made to weld the tank to liquid tight status but ended up modifying them further to get better results. This is Steve's PDF direct link.
http://www.plucks329s.org/pdf/gastan...production.pdf
It's a very interesting read.
I've always felt that the firewall section was stamped before it was welded on to the tank. It's not easy to stamp a piece of sheet metal that is not bolted to something pretty solid or backed up by a heavy bucking bar. It could have been done that way but it also could have been placed into the jaws of a stamping machine just prior to the part being attached to the tank. If they were finishing 2500 tanks a day, they were making a lot of tanks.
Charlie Sorensen's book mentions how a lot of Fords "storage" for parts was while the parts were being shipped to the assembly branch plants. Ford may have shipped stuff by sea to California but they may have delivered it by rail as well. They delivered a lot of stuff by rail in that time frame and by the box car loads. Shipments get tied up at rail yards when they have to go through a rail car transfer and it can slow progress down a while. There was certainly no railroad that went from Dearborn direct to the branch in California so there may have been several transfers along the way. A lot of the sea shipments were likely to foreign branches. It took a long time to get out of the Saint Lawrence seaway let alone travel down to Panama and back up to California. It could have been shipped both ways depending on how large the parts may have been. It would have been faster by rail though.