Inside the fat base of the descent stage there was a doughnut-shaped compartment. Some of this was taken up with fuel tanks, but the rest was given over to payload. Half of the doughnut was a surface shelter, a tightly curving chamber that would serve as crew quarters and laboratory for the crew while it was on the Martian surface. Then there were surface operations bays, with room for airlocks and equipment, and space for a small Mars Rover.

Sitting on top of the descent stage was a smaller cone, the cabin of the ascent stage. This was a bubble of glass, to give all-round visibility. The crew would ride down to the surface and back to orbit again inside this cabin. It held room enough, just, for four guys, lying side by side in acceleration couches. But you could fold the couches out of the way, so the pilots could stand up to control the final descent.

The rest of the ascent stage was a cylinder, impaled down the spine of the MEM’s descent stage. The ascent stage, rising from Mars, would look like a Kojak lollipop, Lee thought whimsically, a glass lollipop on a stick of propellant tanks and rocket engine, leaving behind a truncated half cone: the descent stage, scorched and decapitated.

That was the skeleton of the thing, anyhow, and the teams soon began to put on some flesh, mapping out the subsystems.

ECLSS, the environment control and life-support system: Jack Morgan’s baby, with molecular sieves to scrub out carbon dioxide, and filtration for water recycling. Electrical supply fuel cells for the descent stage, separate, smaller cells for the ascent. Guidance and control there would be inertial guidance, and rendezvous and landing radar systems, and attitude rocket clusters, and gimbaled main engines to allow thrust vector control. Communications: the engineering models of the ascent stage started to sprout antennae, S-band for TV links to the orbiter and voice to Earth, and VHF for voice links to the orbiter and EVA to MEM links…

Lee had his people borrow where they could. Details in the subsystems, for example: they put in power cells that had already served in the Apollo Lunar Module, an L-band rendezvous radar from Gemini. As for a propellant, should they use the nitrogen tetroxide/Aerojet 50 Grumman had used on the LM? That was hypergolic — meaning it would ignite on contact with its oxidizer, with no need for an ignition system — but it was low-performance, corrosive, not to mention toxic. Not the stuff you’d want to have to store for a year or more as you hauled your MEM out to Mars. There had been some studies of fluorine-based compounds, but they hadn’t been taken too far because fluorine was such difficult stuff to work with. What, then?…

As the teams worked, the design quickly evolved away from the first baseline sketched out by Lee.