Gödel was shaking. 'I un-understand how you fe-feel, Professor,’ he stammered, 'but I-I'm afraid that for the time being there is no way to answer yo-your question.'

From then on, the vague threat hinted at by Gödel's Incompleteness Theorem developed into a relentless anxiety that gradually came to shadow his every living moment and finally quench his fighting spirit.

This didn't happen overnight, of course. Petros persisted in his research for a few more years, but he was now a changed man. From that point on, when he worked he worked half-heartedly, but when he despaired his despair was total, so insufferable in fact that it took on the form of indifference, a much more bearable feeling.

'You see,' Petros explained to me, 'from the first moment I heard of it, the Incompleteness Theorem destroyed the certainty that had fuelled my efforts. It told me there was a definite probability I had been wandering inside a labyrinth whose exit I'd never find, even if I had a hundred lifetimes to give to the search. And this for a very simple reason: because it was possible that the exit didn't exist, that the labyrinth was an infinity of cul-de-sacs! O, most favoured of nephews, I began to believe that I had wasted my life chasing a chimera!'

He illustrated his new Situation by resorting once again to the example he'd given me earlier. The hypothetical friend who had enlisted his help in seeking a key mislaid in his house might (or again might not, but there was no way to know which) be suffering from amnesia. It was possible that the 'lost key' had never existed in the first place!

The comforting reassurance, on which his efforts of two decades had rested, had, from one moment to the next, ceased to apply, and frequent visitations of the Even Numbers increased his anxiety. Practically every night now they would return, injecting his dreams with evil portent. New images haunted his nightmares, constant variations on themes of failure and defeat. High walls were being erected between him and the Even Numbers, which were retreating in droves, farther and farther away, heads lowered, a sad, vanquished army receding into the darkness of desolate, wide, empty spaces… Yet, the worst of these visions, the one that never failed to wake him trembling and drenched in sweat, was of 2^100, the two freckled, dark-eyed, beautiful girls. They gazed at him mutely, their eyes brimming with tears, then slowly turned their heads away, again and again, their features being gradually consumed by darkness.

The dream's meaning was clear; its bleak symbolism did not need a soothsayer or a psychoanalyst to decipher it: alas, the Incompleteness Theorem applied to his problem. Goldbach's Conjecture was a priori unprovable.

Upon his return to Munich after the year in Cambridge, Petros resumed the external routine he had established before his departure: teaching, chess, and also a minimum of social life; since he now had nothing better to do, he began to accept the occasional invitation. It was the first time since his earliest childhood that preoccupation with mathematical truths didn't occupy the central role in his life. And although he did continue his research awhile, the old fervour was gone. From then on he spent no more than a few hours a day at it, working half-absently at his geometric method. He'd still wake up before dawn, go to his study and pace slowly up and down, picking his way among the parallelograms of beans laid out on the floor (he had pushed all the furniture against the walls to make room). He picked up a few here, added a few there, muttering absently to himself. This went on for a while and then, sooner or later, he drifted towards the armchair, sat, sighed and turned his attention to the chessboard.

This routine went on for another two or three years, the time spent daily at this erratic form of 'research' continuously decreasing to almost nil. Then, near the end of 1936, Petros received a telegram from Alan Turing, who was now at Princeton University:

I HAVE PROVED THE IMPOSSIBILITY OF A PRIORI DECIDABILITY STOP.

Exactly. stop. This meant, in effect, that it was impossible to know in advance whether a particular mathematical statement is provable: if it is eventually proven, then it obviously is – what Turing had managed to show was that as long as it remains unproven, there is absolutely no way of ascertaining whether its proof is impossible or simply very difficult.

The immediate corollary of this, which concerned Petros, was that if he chose to pursue the proof of Goldbach's Conjecture, he would be doing so at his own risk. If he continued with his research, it would have to be out of sheer optimism and positive fighting spirit. Of these two qualities, however – time, exhaustion, ill luck, Kurt Gödel and now Alan Turing assisting – he had run out.

STOP.

A few days after Turing's telegram (the date he gives in his diary is 7 December 1936) Petros informed his housekeeper that the beans would no longer be required. She swept them all up, gave them a good wash and turned them into a hearty cassoulet for the Herr Professor 's dinner.

Uncle Petros remained silent for a while, looking dejectedly at his hands. Beyond the small circle of pale yellow light around us, cast by the single light-bulb, there was now total darkness.

'So that's when you gave up?' I asked softly.

He nodded. 'Yes.'

'And you never again worked on Goldbach's Conjecture?'

'Never.'

'What about Isolde?'

My question seemed to startle him. 'Isolde? What about her?'

'I thought that it was to win her love you decided to prove the Conjecture – no?'

Uncle Petros smiled sadly.

'Isolde gave me "the beautiful journey", as our poet says. Without her I might "never have set out". [13] Yet, she was no more than the original stimulus. A few years after I had begun my work on the Conjecture her memory faded, she became no more than a phantasm, a bittersweet recollection… My ambitions became of a higher, more exalted variety.'

He sighed. 'Poor Isolde! She was killed during the Allied bombardment of Dresden, along with her two daughters. Her husband, the "dashing young lieutenant" for whom she'd abandoned me, had died earlier on the Eastern Front.'

The last part of my uncle's story had no particular mathematical interest:

In the years that followed history, not mathematics, became the determining force in his life. World events broke down the protective barrier which till then had kept him safe within the ivory tower of his research. In 1938, the Gestapo arrested his housekeeper and sent her to what was still in those days referred to as a 'work camp'. He didn't hire anybody to take her place, naively believing that she'd return soon, her arrest due to some 'misunderstanding'. (After the war's end he learned from a surviving relative that she'd died in 1943 in Dachau, just a short distance from Munich.) He started to eat out, returning home only to sleep. When he was not at the university he would hang out at the chess club, playing, watching or analysing games.

In 1939, the Director of the School of Mathematics, by then a prominent member of the Nazi party, indicated that Petros should immediately apply for German citizenship and formally become a subject of the Third Reich. He refused, not for any reasons of principle (Petros managed to go through life unhampered by any ideological burden) but because the last thing he wanted was to be involved once again with differential equations. Apparently, it was the Ministry of Defence that had suggested he apply for citizenship, with precisely this aim in mind. After his refusal he became in essence a persona non grata. In September 1940, a little before Italy's declaration of war on Greece would have made him an enemy alien subject to internment, he was fired from his post. After a friendly warning, he left Germany.