A message that had been sent with an attack.

Having decoded the first transmission, it didn't take the computer long to get to grips with the second. As with the previous message, the yrr had responded in binary code. It remained to be seen whether the data would form a picture. Until now only one sequence made sense. It was a piece of information that seemed laughably simple, given that it was supposed to have come from an alien system of thought.

It was the description of a molecule. A chemical formula. H2O.

'Very original,' Shankar said sourly. 'I think we know they live in water.'

But the formula was overlaid with other information. While the computer crunched the data, Crowe realised what the message might mean. 'Perhaps it's a map,' she said.

'How do you mean? A map of the seabed?'

'No. That would imply that they lived on the seabed. Assuming the belligerent little creatures in the lab are part of the alien intelligence, the yrr live in water. The depths are a liquid universe – homogeneous and the same from every angle.'

Shankar thought for a moment. 'Unless, of course, you examine the seawater and look at its make-up – exact levels of minerals, acids, alkalis and so on.'

'And then you see it all looks different.' Crowe nodded. 'The first time they sent us a picture composed of two mathematical solutions. This time it looks more complicated. But if we're right, there'll he limits to the variation. I can't swear to it, but I think they've sent us another picture.'

JOINT INTELLIGENCE CENTER

Weaver found Anawak sitting at the computer. Virtual amoebas were spinning over the screen, but it seemed to her that he wasn't really looking at them. 'I'm sorry about what happened to your friend,' she said softly.

'Do you know what's funny?' His voice sounded choked. "That her death's really affecting me. The last time I cried was when my mother died. My father died, and I just felt terrified because I wasn't even sorry. But Licia? God, it's not like I chased after her or anything. Until I learned to like her, she was just some student who got on my nerves.'

Tentatively Weaver laid her hands on his shoulders. Anawak's fingers reached up to touch them. 'Your program works by the way,' he said. 'So now it's up to the others to get the biology working in the lab.'

'Yes, that's the problem. Meanwhile, it's just a hypothesis.' They'd equipped the virtual amoebas with DNA that was capable of learning and could constantly mutate. Every single cell was essentially an autonomous computer that continually reprogrammed itself. Each new piece of information changed the structure of the genome. If a certain number of cells underwent a particular experience, the experience changed their genetic structure. If the mutated cells aggregated with other cells, they passed on the information, and the DNA of the other cells changed. It meant that the cells weren't merely learning constantly: whenever they aggregated they updated each other. Any new knowledge acquired by a single amoeba enriched the collective knowledge of the whole.

It was a revolutionary idea. It meant that knowledge could be inherited. They'd discussed it with Johanson, Oliviera and Rubin, but the outcome had left them more bemused than ever. 'The good news was that the theory had been accepted with enthusiasm. The bad news was that there was an almighty catch.

CONTROL ROOM

'What you have to realise,' explained Rubin, 'is that when DNA mutates, its genetic information changes – and that spells trouble for any living creature.'

While the others were still analysing the samples, Rubin had snuck out of the lab, supposedly because his migraine was returning. In reality, he'd disappeared into the hidden control room for a meeting with Li, Peak and Vanderbilt. They were working through the transcripts from the audio surveillance. By now they all knew about the computer program and about Weaver and Anawak's theory – but only Rubin understood the implications.

'Organisms rely on their DNA staying intact,' said Rubin. 'Otherwise they fall sick or produce defective offspring. Exposure to radiation, for instance, causes irreparable damage to DNA, resulting in cancer or birth defects.'

'But how does that fit with evolution?' asked Vanderbilt. 'If humans are descended from apes, our DNA must have changed.'

'Sure, but evolution takes place over a long time. And it selects those organisms whose natural mutation rate makes them best suited to the prevailing conditions. People don't often talk about evolutionary failures, yet nature gets rid of unsuccessful adaptations all the time. That said, there is another option, and that's repair. Take tanning, for instance. Sunlight leads to changes in the cells in the upper layers of our skin, resulting in mutations in the DNA. Our skin starts to tan, and if we're not careful we go red and burn. When that happens, our body sheds the cells that have been destroyed, but those remaining can be repaired. It's repairs like these that allow us to survive. Without them, we'd not only suffer continual mutations in our DNA, but our injuries wouldn't heal and we wouldn't recover from disease.'

'Fine,' said Li. 'But what about single-cell organisms?'

'The same thing applies,' said Rubin. 'If their DNA mutates, it has to be repaired. And remember, organisms like that reproduce by cell division. For a species to remain stable, its DNA has to undergo repair. It doesn't matter what kind of cells we're talking about, nature always endeavours to keep the rate of mutation within manageable limits. And that's the catch for Anawak's theory. The genome is repaired globally, along its entire length. You can picture the repair enzymes as policemen, patrolling the entire DNA strand on the look-out for errors. As soon as they find a defective area, they begin the repair. To ensure that the information corresponding to the DNA's original sequence doesn't get lost, the repair enzymes act as the guardians of the genome's data. They police the sequence and can tell immediately which genetic configurations match the original and which are defective. It's like trying and failing to teach a child to talk. As soon as it learns a new word, the repair enzymes come along and reprogram it to its original state – ignorance. It's not possible for it to learn.'

'Then Anawak's theory is nonsense,' said Li. 'It would only make sense if the amoebas could retain the changes to their DNA.'

'Well, on the one hand, that's right. Any new information would be treated as defective by the repair enzymes and, hey presto, the genome's restored to its original configuration. Back to square one, so to speak.'

'I'm guessing,' grinned Vanderbilt, 'that we're about to hear the butt.'

Rubin nodded hesitantly. 'There is one,' he said.

'Which is?'

'I don't know.'

'Hang on,' said Peak. He sat up in his chair and winced. His foot was bandaged. 'I thought you just said-'

'I know! But the theory's brilliant,' cried Rubin. 'It would explain everything. Then we'd be certain that the substance in the tank is our enemy. We'd be face to face with the yrr – the creatures that have landed us in all this shit. And I'm certain that it's them! We saw some pretty weird stuff in the lab this morning. The blob of jelly examined our robot, and you should have seen the way it did it – it had nothing to do with animal instinct or curiosity. It was pure cognitive intelligence. Anawak's theory must be right. Weaver's already got it working electronically.'

'But how are we supposed to make sense of it?' Vanderbilt sighed and mopped his forehead.

'It could be to do with anomalies.' Rubin gestured vaguely. 'Even repair enzymes sometimes make mistakes. Not often, but every ten thousand repairs or so they slip up. They miss a base pair that should have been restored to its original state. It's not much, but it's enough to cause a baby to be born hemophiliac, with a cleft palate or even cancer. We see these anomalies as defects, but they're proof that the repair mechanism doesn't always work.'