He looked as me as if I were stupid. "The need to have a full moon or be near one," he explained patiently, "is to ensure a time of low tide at sunrise on the invasion beaches, so that mines and tank traps and so on can be cleared. The RAF and US Air Force would prefer an outright full moon, so that gliders and other planes can land before sunrise-which doubles the odds again."

"Not exactly making it easy, are they?"

Stagg shook his head. "I don't think it's any less daunting for the bra.s.s hats than it is for me. I have to present to them regularly now: Eisenhower, Air Chief Marshal Tedder, General Bull, Admiral Creasy, Air Vice-Marshal Wigglesworth...and all the other chiefs and deputies of SHAEF divisions. The first time was terrifying. Eisenhower looked at me and said: 'Whenever you see a good spell that would be suitable coming along during the next month or so I want you to tell us. Give us as much notice as you can.'"

"And have you?"

"What?"

"Suggested a date."

"Not yet. None of the forecasters can agree. And there isn't enough data. We don't have enough weather ships in the Atlantic. Sir Peter has promised more."

He stood up and went to a map on the wall, showing me where ships, marked out with flags, were dispersed across the ocean. "This whole area has no weather ships. The only good news is that the Germans are in a worse position. More or less the whole remaining Atlantic U-boat fleet is now concerned with sending weather information. They know just as we do that the Atlantic weather is what will determine the weather in the Channel."

He sat down, taking off his gla.s.ses and rubbing his eyes with the backs of his fingers. "Where was I? We have three teams of forecasters. At Widewing-that's the main US air base near here-there is a man called Krick and another, Holzman. Both colonels."

I smiled ruefully, remembering the Glasgow hotel. I hadn't realised they held such high rank.

"Krick has compiled a statistical index of weather patterns in northern Europe going back forty years or so. He uses the a.n.a.logue method."

The a.n.a.logue method involved selecting weather types from past periods that most closely matched the current weather and seeing what happened before. It was a bit like case law. The future is extrapolated from the past, with the forecast extending to as long as six days ahead.

"I've met them," I explained, remembering with a nauseous twinge the poker game and the terrible hangover which followed it. "Krick and Holzman. By chance, at Prestwick airport. Krick seems a jolly fellow, but as for the a.n.a.logue method, if that is what he practises, I'm not persuaded. Nature does not repeat itself like a workshop press; identical patterns do not develop identically; and it's not really possible to forecast more than two days ahead. Three days, max."

"Exactly," said Stagg. "That is just what Charles Douglas says. Well, he's opposed to anything over two, in fact. He's pretty much made that the rule at Dunstable, as you will know from your time under him there. Fancy some tea?"

He stood up again, unbundling his long limbs like a praying mantis on a leaf.

"Yes please."

He flipped a switch on an electric kettle. "A gift from the Americans," he explained.

I remembered Douglas running round the table, his tie and the tails of his suit jacket flying behind him. I suppose it must have been what's now called stress which made him do this, as well as the aeroplane crash in which he had been involved during his combat training as a fighter pilot. He was wounded five times in incidents after the crash, and that can't have helped either.

Like Stagg, Douglas had a thin face and a moustache. Well, lots of people had moustaches in those days. I considered him a man of tremendous skill and judgement, and to some people's mind he is still the greatest British practical weather forecaster of the century, with Ryman taking the palm for theory.

Very sound, very careful. He tended to start with the present weather data then would apply weather memory and weather theory by common sense, rather than according to a particular philosophy.

"Douglas doesn't apply past situations religiously, like Krick, or rely totally on theory, like Petterssen," Stagg continued, shaking loose tea into a pot. "He allows a kind of jiggle, a wrinkle, into his system, a s.p.a.ce for his own intuition, and he admits of theory whatever he is personally convinced by."

All this tallied with what I knew of Douglas from my own experience. "That's why, even though he stammers and stutters and sometimes can hardly speak his mind," Stagg continued, "I listen to him most-he is very aware of the complexity of any given situation, having more experience of the vicissitudes of British summer weather than the others. He is less likely to stick his neck out, which is Krick's preferred method. If you can call it that. And the Norwegian, well, he just seems to believe he's infallible."

"That would be Petterssen?" I ventured.

"Sverre Petterssen, yes. You'll speak to him soon. The third member of the team. Rather academic, an expert on the upper atmosphere. A member of the Bergen school who has spent time in America. Pa.s.sage of fronts, deductions from the upper air..."

Stagg's voice trailed off wearily. With sad-looking eyes he stared at the tendrils of steam coming from the kettle.

"I know more about the upper air affecting the surface than I used to," I said, trying to be helpful. "Ryman did a lot of work on that at one stage."

"Did he now?" said Stagg, musing. "Well, I wish he was here now, because I often don't have a clue what Petterssen is talking about. It would be good to have someone to vet his a.s.sertions, which are made as if backed up by tons of data. Whereas actually his findings are based on quite new stuff. And as for his habit of revisiting his successes, well, that just gets everyone's back up. Most of all Krick. He just loves it when Petterssen's forecasts are wrong."

He took off his gla.s.ses for a third time, this time rubbing his cheeks with his palms, like someone using a flannel to clean their face.

"It sounds as if you have a lot on your plate."

"Yes. Krick and Petterssen are both tricky customers, They both have irreconcilable, fixed ideas, seemingly logically developed. I can hardly make them agree on the time of day, never mind next week's weather. Oh, there's another lot, too. Naval forecasters, Wolfe and Hogben, at the Admiralty centre in London. Very skilled on wave conditions, as you'd expect, and lower air. They don't have such dogmatic views as Krick and Petterssen, and tend to agree with one another, which is a G.o.dsend in one way, but..."

His voice trailed off again. I realised I was looking at an almost broken man. "So, you all meet once a week?"

Stagg straightened up immediately. "Good G.o.d, no. These people don't meet in person, except Douglas and Petterssen, who work together in Dunstable. No, we do it all by telephone-twice a day."

I was puzzled. "Why by telephone...and so often?"

Stagg chuckled. "You know how Sir Peter is short of forecasters? Well, these people-among the best forecasters in the world-are doing lots of other work for their respective services, in different locations, as well as preparing the forecast for the invasion. And we have to talk twice a day to keep up with Eisenhower's plans. What we say to him affects a vast network of troops and vehicles, all of them waiting to go, not to mention a host of ships hiding round Britain, from the coves of north Devon to the sea lochs of Argyll...That's where you were, isn't it?"

I thought of the ships and subs moored outside Ryman's house. Once again, I felt deep astonishment at my role in the death of probably the one man on earth who might have been able to reconcile the competing views of Stagg's warring forecasters. It suddenly occurred to me that Sir Peter had ordered me here in spite of, not because of, my frank letter to him. He still hoped I had learned something from Ryman. Even if I had more or less given up, he was still looking for a single all-explaining answer. It was the wrong approach; but how could the multidimensional picture which Ryman conveyed to me be conveyed in turn to military men who needed relatively simple instructions?

The generals were the least of it. The thought of casualties filled me with dread again. After my Scottish calamity, was I now going to be responsible for sending thousands of men to their deaths on the beaches of Normandy because of an incorrect forecast?

"I'll take you through the charts before the phones go," said Stagg. "We've got about two hours."

"Right."

He finally produced the promised cup of tea, and we sat down in front of the charts. They showed a map of Europe and the Atlantic, covered with isobars and fronts, together with specific pressure and temperature readings from weather ships and other sources.

Once we had finished going over the charts, which were more complicated than any others I had previously seen, Stagg brought up again the subject of my joining up.

"Now, I thought flight lieutenant would be the rank appropriate to your Met Office grade. I hope that's all right. You should have time to pick up a uniform from the commissary before the conference. Follow signs for Web 51. They should be able to fix somewhere for you to sleep, too. Don't be too long."

Four.

After collecting my new blue serge uniform, which was rather itchy, and sorting out the logistics of a billet, I retraced my steps to Stagg's office. I joined him at the big oak table with the three telephones. Their chrome dials looked like flowers waiting to open. I was hungry. The timing of events was such that I had missed lunch and no one had yet mentioned dinner. I looked at the table. Next to each phone was a little black box housing a scrambler. Our conversation would be encrypted.

Stagg and I were joined by his American deputy, Don Yates. He was a spare, dark-haired little man, who would often amuse us by telling fantastic tales of his hiking, hunting and fishing exploits back home in the States. He came from a wooded, mountain area of Maine, near the Pen.o.bscot river. If he was to be believed, the area was still as full of deer and fish as it had been in the days of Buffalo Bill. It sounded like paradise: sheltered coves and mossy forests where Yates had learned how to catch his supper with his bare hands. I remember him once saying how he had reached down into a stream and felt the quivering ma.s.s of a salmon there, 'like a piece of pure muscle'.

He was a patient fellow, Yates, and a good handler of men. Like Holzman, he had been a student of Krick's at Caltech, before rising quickly to become head of the US Army's weather operations in Europe. He often had to face down Krick as perhaps only a fellow American would be able to do. He had a lot of presence and, I suspect, carried great influence in the presentations to Eisenhower. He knew when to speak and when to keep quiet. When I saw Yates and Stagg arguing, as they often did, there were times when I would quite cheerfully have belted Stagg over the head with a ruler, but Yates always kept his cool.

My very first conference call followed a pattern that would become familiar. First we set up the phones, routing the calls through a knot of exchanges run by intelligence staff. Nowadays it would be a matter of pressing a couple of b.u.t.tons, but at that time to arrange a conference call on secret lines was quite a feat.

Once we had gone though this frustrating and at times amusing process, which involved a lot of "yes, yes, yes..." and took about twenty minutes, Stagg picked up his handset and dialled. Immediately the two other phones rang and we picked up.

I heard a series of disembodied voices check in: 'Dunstable' (the Met Office), 'Widewing' (the USAAF and RAF base nearby), 'Citadel' (the Royal Navy at the Admiralty Forecasting Unit in Whitehall).

This telephone circuit became a major part of my life during May and June 1944. Krick andor Holzman speaking for Widewing; Petterssen andor Douglas speaking for Dunstable; and one or other of Lieutenant Hogben or Commanders Wolfe and Thorpe speaking for the Royal Navy from the Citadel.

Other parts of the military establishment listened in-to ensure that our top-level D-Day forecasts did not conflict with those regularly given to lower-level naval, air and army formations.

On that first day, Stagg introduced me, saying, "You'll all be pleased to hear I have a new a.s.sistant, Henry Meadows, a bright Cambridge natural sciences graduate who I hope will pitch in from time to time. He has worked with me at Kew and trained as a Met observer under Mr Douglas."

I said h.e.l.lo to Douglas, who I think was pleased to hear from me, and reacquainted myself with Krick, hoping he wouldn't mention that we had played poker and got drunk together, which I doubted Stagg would approve of.

But he didn't, just drawling, "Well I'll be d.a.m.ned, Henry. Welcome aboard."

The first job of the conference was to agree on a map of current conditions, and I would soon discover that not everyone always turned up with the same map, let alone the same forecast. It often took about half an hour to sort all this out.

Once actual forecasting got underway, Petterssen at Dunstable was first to speak, his strong Norwegian accent interrupted by the occasional clicks and static of the telephone wires. It took me a while to become familiar with all the codes they were using to describe areas of high and low pressure...H1, H2, H3...L1, L2, L3...

It was standard that H stood for a high pressure area, L for a low, but the numbers to which they were attached were altered from time to time as a further security precaution, should the enemy be listening. Given the transitory nature of weather, our counterparts at the Zentral Wetterdienstgruppe were going to be hard pressed to interpret any intelligence they might receive. I wondered whether Sir Peter had been able to get anything more out of Heinz Wirbel, the scientist who had bailed out of the Junkers.

"As I forecast last time," began Petterssen, "L1 has moved east-north-eastwards, bringing further deterioration from the west in its wake as the week progresses. There will be increase in cloud and freshening west-north-westerly wind through the week, switching to west-south-west up to force four or five on Wednesday, as an interval, deterioration continuing into Sat.u.r.day, when there is risk of rain..."

"Patches of low, low, low...," interrupted an English voice.

"Lowish cloud along s-south-west coasts on Tuesday morning, some of these, er, at a base of one thousand feet, mixing with, with fog patches in the western Channel." It was Douglas. "Mainly fair to Wednesday, then G-G-G.o.d knows."

Someone else on the line grunted. I heard Stagg sigh beside me. The echo repeated in the handset against my ear.

"I can't go along with this," said an American, unmistakably Krick. "You're far too gloomy, Petterssen. I see quiet, fair weather in all areas, especially from Wednesday. Considerable fine intervals, especially in eastern areas. Good visibility except for those local morning fog patches Douglas mentioned. They'll burn off quick."

"What about that low?" countered Petterssen. "Surely you can see that low coming? High pressure in the north-east Atlantic is bound to force it through."

"That cyclonic cell, not very considerable in my view, will anyway collapse within two days, allowing the warm period I mentioned," said Krick. "There are many a.n.a.logues for a settled period like this, in May 1929, for instance, and the following year. This is how it will play."

He spoke with great confidence, with bravado, in fact. That was the thing about Krick. He did not have the intellectual power or ethical rigour of the others who sat, at least figuratively speaking, round that table, but he had something none of the rest of us had. Conversational force, and the ability to make a narrative of a scientific forecast. The latter, especially, is a really important quality in a forecaster.

But if the story's wrong then the whole team is in the soup. And there were other voices round that table that were often convinced Krick's predictions were way off beam. In their own minds, these speakers were actually just as confident as Krick, even if they didn't sound so.

"Um, not n-n-n-necessarily," said Douglas. "We had a development like this in May 1931. Pressure over Europe was a little lower than now, and not so high in the north-east Atlantic. But the upshot was a period of north to-north-east winds which continued for ten to twelve days; they r-r-r-reached gale force at times in the eastern Channel."

There was a pause in which the telephone wires clicked and whirred as if, somewhere in the depths of the exchange, a mechanism was running down.

"G.o.d almighty," said Stagg. "We'll come back to this. Navy?"

"We lean towards Petterssen-Dunstable on the general forecast," said a good-natured voice with a New Zealand accent. This was Lieutenant Hogben. "Fine weather but risk of rain on Sat.u.r.day. On the maritime side of things, and I remind you this is an amphibious operation, we expect no appreciable swell. Waves less than two feet at first, probably increasing to four feet in the eastern Channel and six feet in the western Channel."

"Right," said Stagg. "Well now, either Dunstable or Widewing will have to relax its view. It seems to me-and you must remember that my job is to present a single, confident reliable forecast to General Eisenhower-that the divergence rests on what happens at the end of the week. We all know how, by its very nature, the structure and processes of weather can produce interminable discussion and still spring-"

There was an unearthly moan throughout the whole complex as the generators went down and the lights were extinguished.

"Potash!" shouted Stagg.

"What's that?" said Yates's American voice in the darkness.

"I say potash so as not to swear," said Stagg. "It's a bombing raid," he explained to me. "We always just shut off the electricity because if they hit SHAEF-well, it's all over then."

The phones themselves were clearly on a different circuit from the mains, and this enabled us to keep talking. I heard some distant explosions, but could not estimate how near the bombs were falling or of what magnitude they were.

"Some way off," said Yates, as if reading my thoughts.

The lights came back on, but the discussion had stalled. The experts still could not agree. Stagg became quite angry, as he had to deliver a five-day forecast to Elsenhower the following morning.

Douglas disparaged the whole idea of five-day forecasts. "You can have as m-m-many conferences as you like," he said. "They will make no difference: it is just not p-possible to make regular forecasts five or six days ahead that can have any real v-v-value for military operations or any other p-p-purpose."

"Scientifically speaking, there are no reasons why long-range forecasting should not be possible," said Petterssen calmly.

"Of course it's possible!" bl.u.s.tered Krick. "Precise long-range weather forecasting requires day-by-day prediction for years ahead, and that is what my a.n.a.logue sequence method provides. Look at the chart and the comparison of previous weather sequences from 1930 I sent through."

"Pure guesswork from t-t-two days out," mumbled Douglas.

"How dare you!" exploded Krick through the earpiece. "I've been through half a century of northern hemisphere weather maps. Because of that, I am able to give a mathematically reliable five-day forecast."

"There is only one man in Britain I know able to do weather prediction by d-d-direct attack with mathematics," said Douglas, "and even he would admit it is a p-p-process very liable to error. His name is Wallace Ryman."

A chill went through me. I was wrong that everyone in the meteorological community knew what had happened. "He's dead," I said immediately, hearing my own voice in my ear a second later. "Ryman is dead. He died in an accident in Scotland. I was working with him there."

"Oh dear," said Douglas. "What a p-p-pity. I remember going to Norway with him to see your p-p-people in Bergen, Sverre."

"Yes. I heard about him from Bjerknes, my tutor," Petterssen said. "I'm afraid to say he was regarded as a strange sort of character. He brought a gun with him, to measure wind shear. Many considered the gun a toy and the man himself an overgrown Boy Scout. It is a shame, though, that he abandoned meteorology before his numerical weather process could be put into practice."

I thought of Ryman with his gun in the field. Was he not the great man I'd thought? "Well," I bristled, feeling the need to stick up for him, "until his death he devoted himself to the application of mathematics to peace studies. I think many of his meteorological ideas are still valid, nonetheless."

"Devoted himself to what? what?" said Krick, incredulous.

"Peace studies. He applied mathematics to the relationships between opposing forces to see how war might have been avoided."

"Gentlemen, can we please make a forecast?" said Stagg. "On exactly what issues are we divergent?"

There was a babble of voices from all sides.

"My past a.n.a.logues are right," said Krick. "The whole US weather service is run on this basis."

"It must be informed by theory," said Petterssen. "Otherwise it is worthless."

"You must look at the prevailing pattern before you consider other factors," said Douglas.

I felt the need to speak again, but it was as if the voice coming through me was not my own. "Future weather is a judgement of probabilities based on physical principles which are reducible to mathematical formulae. There is one I know which relates temperature and wind speed to produce an index of turbulence. The Ryman number. You may have heard of it. Well, I could try and find its values for the Channel weather in the relevant period."

Silence ensued. Of course, I now feel it was a mistake to have brought up the number at the first conference, but I suppose I was trying to prove myself.

"Very good, Meadows," said Stagg eventually, like a schoolmaster congratulating a pupil. I perceived a slightly embarra.s.sed tone in his voice. "Do so, though in my experience weather is less reducible to numerical process than Ryman and, clearly now you, believe."