With a far greater effort than it had taken to touch Gibarian’s corpse, I forced myself to touch one of the bare feet. Then I made a second bewildering discovery: this body, abandoned in a deep freeze, this apparent corpse, lived and moved. The woman had withdrawn her foot, like a sleeping dog when you try to take its paw.

“She’ll freeze,” I thought confusedly, but her flesh had been warm to the touch, and I even imagined I had felt the regular beating of her pulse. I backed out and fled.

As I emerged from the white cave, the heat seemed suffocating. I climbed the spiral stairway back to the hangar-deck.

I sat on the hoops of a rolled-up parachute and put my head in my hands. I was stunned. My thoughts ran wild. What was happening to me? If my reason was giving way, the sooner I lost consciousness the better. The idea of sudden extinction aroused an inexpressible, unrealistic hope.

Useless to go and find Snow or Sartorius: no one could fully understand what I had just experienced, what I had seen, what I had touched with my own hands. There was only one possible explanation, one possible conclusion: madness. Yes, that was it, I had gone mad as soon as I arrived here. Emanations from the ocean had attacked my brain, and hallucination had followed hallucination. Rather than exhaust myself trying to solve these illusory riddles, I would do better to ask for medical assistance, to radio the Prometheus or some other vessel, to send out an SOS.

Then a curious change came over me: at the thought that I had gone mad, I calmed down.

And yet… I had heard Snow’s words quite clearly. If, that is, Snow existed and I had ever spoken to him. The hallucinations might have begun much earlier. Perhaps I was still on board the Prometheus, perhaps I had been stricken with a sudden mental illness and was now confronting the creations of my own inflamed brain.

Assuming that I was ill, there was reason to believe that I would get better, which gave me some hope of deliverance — a hope irreconcilable with a belief in the reality of the tangled nightmares through which I had just lived.

If only I could think up some experiment in logic — a key experiment — which would reveal whether I had really gone mad and was a helpless prey to the figments of my imagination, or whether, in spite of their ludicrous improbability, I had been experiencing real events.

As I turned all this over in my mind, I was looking at the monorail which led to the launching pad. It was a steel girder, painted pale green, a yard above the ground. Here and there, the paint was chipped, worn by the friction of the rocket trolleys. I touched the steel, feeling it grow warm beneath my fingers, and rapped the metal plating with my knuckles. Could madness attain such a degree of reality? Yes, I answered myself. After all, it was my own subject, I knew what I was talking about.

But was it possible to work out a controlled experiment? At first I told myself that it was not, since my sick brain (if it really was sick) would create the illusions I demanded of it. Even while dreaming, when we are in perfectly good health, we talk to strangers, put questions to them and hear their replies. Moreover, although our interlocutors are in fact the creations of our own psychic activity, evolved by a pseudo-independent process, until they have spoken to us we do not know what words will emerge from their lips. And yet these words have been formulated by a separate part of our own minds; we should therefore be aware of them at the very moment that we think them up in order to put them into the mouths of imaginary beings. Consequently, whatever form my proposed test were to take, and whatever method I used to put it into execution, there was always the possibility that I was behaving exactly as in a dream. Neither Snow nor Sartorius having any real existence, it would be pointless to put questions to them.

I thought of taking some powerful drug, peyotl for example, or another preparation inducing vivid hallucinations. If visions ensued, this would prove that I had really experienced these recent events and that they were part and parcel of the surrounding material reality. But then, no, I thought, this would not constitute the proof I needed, since I knew the effects of the drug (which I should have chosen for myself) and my imagination could suggest to me the double illusion of having taken the drug and of experiencing its effects.

I was going around in circles; there seemed to be no escape. It was not possible to think except with one’s brain, no one could stand outside himself in order to check the functioning of his inner processes. Suddenly an idea struck me, as simple as it was effective.

I leapt to my feet and ran to the radio-cabin. The room was deserted. I glanced at the electric clock on the wall. Nearly four o’clock, the fourth hour of the Station’s artificial night-time. Outside, the red sun was shining. I quickly plugged in the long-range transmitter, and while the valves warmed up, I went over in my mind the principal stages of the experiment.

I could not remember the call-sign for the automatic station on the satellite, but I found it on a card hanging above the main instrument panel, sent it out in Morse, and received the answering signal eight seconds later. The satellite, or rather its electronic brain, identified itself by a rhythmic pulse.

I instructed the satellite to give me the figures of the galactic meridians it was traversing at 22-second intervals while orbiting Solaris, and I specified an answer to five decimal points.

Then I sat and waited for the reply. Ten minutes later, it arrived. I tore off the strip of freshly printed paper and hid it in a drawer, taking care not to look at it. I went to the bookcase and took out the big galactic charts, the logarithm tables, a calendar giving the daily path of the satellite, and various other textbooks. Then I sat down to work out for myself the answer to the question I had posed.

For an hour or more, I integrated the equations. It was a long time since I had tackled such elaborate calculations. My last major effort in this direction must have been my practical astronomy exam.

I worked at the problem with the help of the Station’s giant computer. My reasoning went as follows: by making my calculations from the galactic charts, I would obtain an approximate cross-check with the results provided by the satellite. Approximate because the path of the satellite was subject to very complex variations due to the effects of the gravitational forces of Solaris and its two suns, as well as to the local variations in gravity caused by the ocean. When I had the two series of figures, one furnished by the satellite and the other calculated theoretically on the basis of the galactic charts, I would make the necessary adjustments and the two groups would then coincide up to the fourth decimal point, discrepancies due to the unforeseeable influence of the ocean arising only at the fifth.

If the figures obtained from the satellite were simply the product of my deranged mind, they could not possibly coincide with the second series. My brain might be unhinged, but it could not conceivably compete with the Station’s giant computer and secretly perform calculations requiring several months’ work. Therefore if the figures corresponded, it would follow that the Station’s computer really existed, that I had really used it, and that I was not delirious.

My hands trembled as I took the telegraphic tape out of the drawer and laid it alongside the wide band of paper from the computer. As I had predicted, the two series of numbers corresponded up to the fourth decimal point.

I put all the papers away in the drawer. So the computer existed independently of me; that meant that the Station and its inhabitants really existed too.

As I was closing the drawer, I noticed that it was stuffed with sheets of paper covered with hastily scribbled sums. A single glance told me that someone had already attempted an experiment similar to mine and had asked the satellite, not for information about the galactic meridians, but for the measurements of Solaris’s albedo at intervals of forty seconds.