The following explanation was written by Kern Osterstock on April 24, 2011 for the Performance Years forum thread on this subject.
"Some details on the tunnel ram construction:
Once Craig and I convinced each other that we could make a Pontiac turn 9,000 rpm, we faced the challenge of finding the parts needed to let it do so. The rotating components were actually easy to order, once we convince people we were serious and handed them money. But some things, like a tunnel ram, did not exist. So, what do you do? Like a lot of other parts, we wound up building it.
The first step was to take [an early Edelbrock 1x4] P4B intake and cut it up so I was left with only the side flanges and the water crossover. I left to crossover in place to help retain the original bolt pattern. With the flanges bolted to an engine, I then welded a piece across the back of the manifold to keep the flange spacing correct. If I recall correctly, the runners came from an Edelbrock small block tunnel ram [TR-1Y]. The port size was close to correct and the spacing of the siamesed runners was about right for the P4B. A prime criteria was canting the carbs to keep the runner lengths as equal as possible. So, we bought a Chevy tunnel ram, which was stupidly expensive for something I planned to chop into little pieces. But I cut up the thing up to leave me only four small sections, which were the paired runners and about 3/4 of an inch of the plenum. all the rest went into the scrap. By keeping some of the plenum floor, I had material to create reasonable entries into the runners. Those four pieces were welded to the P4B flanges and then the plenum floor sections were patched together with weld and fabricated plates. Then came the first major grinding session to blend the runners to the Pontiac ports, as well as making the plenum floor smooth. After weld, it had looked more like a mound of mashed potatoes, yet needed to be very smooth with proper transition radii. That took about a week of milling, grinding and many trips to the welding bench.
Once the runners all came together with a reasonable floor, I welded a 3/8 inch plate to the top of them and created openings in it to let it be blended properly to the fabricated plenum floor as well as providing a place to erect the plenum walls. The first iteration had the walls about 5 inches high, which matched the Chevrolet design, but we found that the length from the valve head to the base of the carbs was too long for our intended 9,000 rpm target. So, I simply cut the walls down and used tapped holes to attach the manifold top.
The manifold was a lot of work, but was easy to visualize. The linkage for the offset carbs was a real PIA to build, but looked very much like contemporary offset linkage. "