The bay was equipped as it had been during the cruise to Saturn, with the big hab module closest to the orbiter flight deck, and then a short crawl-through tunnel to the reworked Spacelab module that had housed the CELSS farm. Behind the Spacelab lay the Topaz reactors, beneath their heavy shielding. The centrifuge cabin on its big swing arm lay across the top of all this, abandoned now. Its dismantling was on Rosenberg’s long list of things to do; that big motor ought to be useful for something.

Rosenberg jumped off the wing. Briefly, he enjoyed the childlike sensation of drifting down as slow as a snowflake in Titan’s feeble gravity, and settling gently into the slush.

Then he ploughed his way across the gumbo to the nose of the orbiter.

The upper surface of the orbiter’s flight deck, the white felt thermal tiles there, was streaked and stained by tholin gunge. Discovery looked battered, aged, as if the mushy landscape of Titan were dragging it down into terminal entropy. But if he bent down and looked underneath the chin of the orbiter, at the black thermal protection tiles sheltered there, he could still make out the scorching of entry. This had once been a spacecraft, after all, and here it still showed. Even if the payload bay had been turned into a shanty town.

But at that, he reflected, this was surely a better fate for Discovery than to have finished up as a museum piece on the lawn of some fading NASA center.

Later he was going to have to run an internal check on Discovery’s systems. They were having some trouble with balky heaters. Fixed heaters had been installed throughout the orbiter’s hull, to help insulate the life-bearing hab module at its heart. The heaters responded to commands from the command software and temperature sensors. There were also small radioisotope heating units, mounted on moveable mechanisms, that could vary the heat applied to particularly cold areas.

But the heaters ate up a lot of power. Rosenberg had ideas on how they could coat the Base’s main components — the orbiter, Bifrost — with blocks of water ice, like igloos. That would retain a lot of heat, and enable them to reduce the power output from the Topaz reactors, so extending their life. Maybe they could build airtight tunnels between the components to give themselves more space, maybe even put up independent igloos, sealed somehow…

Of course such grandiose schemes depended on getting access to water ice, which was proving a difficulty, on this ice moon. The problem was the tholin slush. As far as Rosenberg could tell they were sitting on top of a gumbo layer at least fifteen yards thick; as far as they’d dug, they hadn’t reached bedrock water ice.

When he looked up at the lighted windows of the orbiter’s flight deck, he thought he could see the blind face of Bill Angel. But the glass was obscured by purple stripes of tholin, and it was hard to be sure.

He walked over the gumbo the fifty feet to where they had dragged Bifrost, their intact Apollo Command Module. Using components from Jitterbug, they had fixed up a crude airlock over Bifrost’s side hatch. Power cables stretched from Discovery, through the slush, to the Apollo. Rosenberg had rigged up the CELSS farm in Bifrost, after stripping out the couches and other moveable gear. That had given the crew in Discovery the extra living space of the old Spacelab module.

Fifty feet further on from Bifrost was a small pile of gear, covered by another hunk of parachute fabric. This was salvage from Jitterbug, disassembled and hauled laboriously across the tholin by Rosenberg and Benacerraf. The shell of Jitterbug still lay in the gumbo where it had come down, its base and chunks of its hull chopped away, the improvised grave of Nicola Mott close by.

Rosenberg had developed a habit of peering under the stiffening fabric over the gear pile every day. He did it now. It was a waste of time, of course. He was on a lifeless world, here: there were no thieves to disturb the pile, no rats or dogs who would chew the tarpaulin. And there never would be, as long as he lived.

He walked on.

The closeness of the horizon, his immersion in this perpetual murk, made it seem as if he was stuck inside the close walls of some opaque orange bubble. Some deep part of his brain, he suspected, still believed that things must be different a little further away: a few miles from the camp, maybe over Titan’s close horizon. Some place where there were all the elements he had grown up with: people and animals and buildings and cars, and a blue sky with white, fluffy clouds…

But it wasn’t true.

It was an odd thing, a small detail, but to Rosenberg the lack of disturbance to the equipment piles emphasized their isolation on this lifeless moon more than any amount of theorizing.

Benacerraf and Rosenberg, cleaned up, sat on the flight deck of Discovery — away from Bill Angel — and chewed the fat about equipment problems.

What they were really doing, of course, was not talking about Bill.

Rosenberg was suspicious of Angel, on some deep level. Bill seemed to have stabilized since the landing, as the NASA psychologists had suggested he would. But Rosenberg didn’t think the creature in the hab module was Bill Angel any more. The way he talked, the body language… He seemed to be coming in at an angle to the rest of the world. As if his head was stuck in some fourth dimension.

Benacerraf shut Rosenberg up when he talked like this. How come scientists are so precise and picky about their specialisms, but always prepared to bullshit about stuff they know nothing about, like psychology? As far as she was concerned this situation wasn’t about trust; it was about management. Benacerraf approved of Rosenberg’s plans for their survival. But to achieve those plans they were going to need resources: time, muscle power, intellectual energy. She wasn’t convinced that the two of them, alone, were sufficient. She thought they needed Bill Angel to close the design. If they were going to survive here, they had to manage all their resources effectively. And that included Angel. She saw it as her job to manage Bill, the way they were going to have to manage other pieces of equipment, balky or otherwise.

So now Rosenberg sat here and spoke, not about Bill, but about the Sabatier unit. The Sabatier was a simple piece of kit, basically a pipe surrounded by nichrome heaters. The Sabatier cracked carbon dioxide, collected from the life support system, by reacting it at high temperatures with hydrogen, in the presence of a ruthenium catalyst, to produce methane and water. The water was collected in a condenser coil, and the methane was vented to Titan’s air.

“I think the catalyst is being poisoned by a build-up of amine vapors… The trouble must be further upstream, in the process, at the SAWD.” The solid amine water desorption unit removed carbon dioxide from the air by passing it over beads of resin inside steel canisters. Rosenberg started to list the steps he was taking to test this out. He said, “At least the systems are easier to work with, now we have a little more space to move around in. And every damn component doesn’t float off into the air every time you turn around…”

And so on, a parade of detail.

Benacerraf was chewing on a spindly carrot from the CELSS farm. “Rosenberg, how are we doing overall? What else do we need to do, that we’re not doing?”

He clasped his hands behind his head, and rocked back in his seat. “Let’s go back to basics. During the cruise, where we had to rely on nothing but the resources we carried, we tried to close all the life support loops. We recycled all our waste, solid and liquid, and fed nutrients to the plants we grew, and cleaned the air… We did well; we survived six years in interplanetary space. But even so the closure was never perfect; we lost about five percent of most materials as they passed around the life support cycle, to leakage, unrecoverable waste, whatever.