An Accidental Statistician

by George E P Box

  Sweden's ties to the sea are everywhere evident. One of her less successful experiences concerns the famous warship, the Vasa. In 1628, a multitude of people assembled at Stockholm's harbor to see the maiden voyage of this brand new ship, which was one of the largest naval vessels of the age. It was 207 feet long and 36 feet wide. It displaced 1,210 tons and had a draught of nearly 15 feet. These dimensions might have been technically appropriate, but unfortunately there had been a last-minute change in plans: The Vasa was originally designed to have four decks, one of which was a gun deck, but the King had heard of a ship built in Denmark that had two gun decks, so he insisted that another deck be added and outfitted with 64 bronze guns weighing 71 tons. This interference with the original plan proved disastrous, for as the newly launched ship sailed from the shelter of the tall cliffs, she heeled over and sank.

  Fortunately the Baltic Sea has a remarkably low level of salt, and salt is a requirement of the termites that consume wooden wrecks. So when in 1956, the sunken ship was located, having spent 300 years under the sea, it was in remarkably good shape. In 1961, with great caution, it was brought to the surface and transported to a special dockyard.

  When we first saw this huge ship, it was still being sprayed intensively with polyethylene glycol to preserve it. On a later visit, the spraying had been completed and the restored ship was on display. More than 700 sculptural and ornamental pieces, some of them quite beautiful, had been recovered.

  In Sweden so many people spoke very good English that it was hard to detect those for whom it was not their native language. But Claire and I came across a shop assistant who had a very broad and unmistakable Lancashire accent. So I asked her when she had emigrated from England. “No, I'm Swedish,” she said, “but my English teacher was from Lancashire.”

  1 G.E.P. Box and J. Tyssedal, “The Sixteen Run Two-Level Orthogonal Arrays,” Biometrika, Vol. 83, No. 4, 1996, pp. 950–955.

  2 As one reviewer said of the organ, “In possession of some pugnacious reeds and a thrilling ‘pleno,’ it's an ideal choice for … some big-boned praeludia as well as Buxtehude's most extended organ work: the imposing Fantasia on the Te Deum.” Sleeve notes from the recording by organist Christopher Herrick, Buxtehude: The Complete Organ Works, Volume 2, played on the organ of Nidaros Cathedral, January, 2009. Recording released January, 2010. Captured from http://www.hyperion-records.co.uk/al.asp?al=CDA67809. See also the remarks on the organ's history by Howard Goodall at http://www.howardgoodall.co.uk/presenting/organsnu.htm.

  “I know something interesting is sure to happen.”

  Chapter Eighteen

  A Second Home in Spain

  When I first visited Spain in the 1970s, it was to give short courses that had been organized by Daniel Peña and Albert Prat. George Tiao and I taught time series in Madrid, where Daniel lived, and Stu Hunter and I taught experimental design in Barcelona, the home of Albert.

  I had never before realized that Catalonia is a state, for many purposes, separate from Spain, with its own language. Albert was proud of his heritage and was a wonderful guide. He was also an expert chef, and he knew a great deal about wine. Whenever we visited, he took us on a tour of all the restaurants, which while the best, were not necessarily the most expensive. (One place he took us to eat delicious seafood was down a narrow alley where there were a few people who had seemingly passed out on the ground, and I almost feared for my life.)

  I remember one day, when we had a free afternoon, he asked me what I wanted to do. I happened to remember that Freixenet, a maker of Catalan champagne, was not far away. When we got there, the gate was closed and it was clearly not a day when they allowed visitors, but Albert found the gatekeeper and chatted with him in Catalan. It was like magic: We were soon shown around with almost royal treatment. On another occasion, Albert's talent as an organizer reached its pinnacle when he put on a conference in Barcelona that featured a dinner at one of Franco's former palaces, replete with strolling musicians.

  The early days of my visits to Spain coincided with the end of the Franco regime. Most of our Spanish friends from Madrid and Barcelona had been in trouble with this regime as students. One of these was Agustín Maravall, who took a time series course that I taught two nights a week. He had come from Spain in 1971 to get his Ph.D. in economics. In the 15-minute break, we used to chat at the coffee machine. He told me that he had been arrested during the “troubles” and sent to a “punishment camp” for a year in Spanish Morocco. The general in charge interviewed Agustín and was happy to discover that he had studied sample surveys. The general told him that very little was known about Spanish Morocco, so he gave him a car and a couple of assistants and they spent the year making a careful survey and census.

  Although Franco died on November 20, 1975, his failing health had forced him to give up being prime minister in 1973. That the fascist influence was fading became clear when the regime was openly made fun of on the television.

  Franco had surprised his allies in Spain by announcing that Prince Juan Carlos should be his successor and serve as king. When Franco died, Juan Carlos not only assumed the throne but also instituted a democratic government, to the displeasure of the fascists. Tensions grew, and in 1981, fascist forces attempted a coup, bringing machine guns and other weapons into the parliament. Albert told me that he was packed and ready to leave the country at a moment's notice. But the King made a televised appeal for support, and the coup came to a quick end.

  Daniel and Albert were wonderful hosts, but Spanish and American ideas about time were different. I remember the first time we were there, in the early 1970s, they invited George Tiao and me to go to dinner. We were ready at about 6 p.m., but they came for us between 9 and 10. We should have known that this would happen because we had sent them the time table for the course as it was taught in the United States, starting at 8 a.m. They wrote back saying, “You may start at 8 am, but you'll be the only ones there.” So we had to start our courses much later, with a long break for lunch, resuming in late afternoon. People went to bed late, and despite, what were for us unusual hours, they were very serious about learning this stuff. Albert, for example, had about 900 engineering students studying the Design of Experiments at his university.

  In the spring of 1986, Claire and I had just flown to Madrid when someone told us that Albert Prat and Tina Roig were getting married that day near Barcelona. We flew at once to join the celebrations that had already begun in a charming outdoor hotel in a park by a lake. There were tables arranged in the garden for the many guests, and much champagne was being consumed. In fact we looked with some surprise at the stack of champagne bottles that must have measured four feet by six. None of it was labeled. I asked Albert how he had got it, but characteristically he only touched a finger to the left side of his nose and said, “I know a man.”

  The origin of the champagne was not the only mystery because, shortly after we arrived, the bride and groom whispered in our ears, “Don't tell anyone, but we aren't actually married.” They had both had previous marriages, one of which had occurred in Germany. They had had a hard time dealing with the various bureaucracies, and at the last minute, there had been some technicality that prevented them from getting a marriage license on that particular day. They decided therefore to go ahead with the wedding celebrations and to get married later. But they said that since their parents were there, they were not telling anybody except for a few close friends.

  For the rest of the day, there was merriment with feasting and dancing. I remember that the director of the orchestra wore a toupee that fell down when his directing became too vigorous.

  Tina had a large flat in Sitges, a seaside resort close to Barcelona, and we often stayed there on weekends. Sitges is famous for its beaches. It became an artsy and countercultural mecca during the Franco regime, which it remains to this day. We were there once during the “Festa Major,” which celebrates the town's patron saint, Bartholomew. The annual celebration features a street procession with outrag
eously costumed participants, some wearing huge paper machê heads that dwarf their bodies. A dragon breathing real burning coals is dragged through the streets, and firecrackers explode everywhere. At one point, sparks from the dragon burned holes in the back of Claire's dress. Finally, at night, there was the finest fireworks display we had ever seen with rockets fired horizontally over the sea.

  When I traveled to Spain in the 1970s and 1980s, Albert always welcomed me to Barcelona with his usual good cheer, and Daniel was an unerringly warm host in Madrid. If my experiences in Spain had ended there, I would have been content. In the 1990s, however, Claire and I were privileged to live in this beautiful country for two extended periods, and Spain became a second home.

  In 1991, Alberto Luceño contacted me saying he would like to visit the department at Madison for a year (Figure 18.1). He was a professor of statistics at the University of Cantabria at Santander, on Spain's northern coast, and he was particularly interested in the work that we were doing on process control. Process control meant different things to statisticians and to control engineers. Statistical process control had been initiated in 1924 by Walter Shewhart, a physicist who was working with engineers at the Western Electric Company in Cicero, Illinois, to improve the quality of telephones. Shewhart's control chart had a line showing the process average with parallel ± limit lines. Data from the process were plotted on a chart. A point outside the limit lines indicated a possible “assignable cause”—a deviation too large to be readily explained by random variation. For such deviations, it was deemed worthwhile to look for the cause, and if it was found, to take action to eliminate it. The idea was that over a period of time, common malfunctions would be eliminated. On the other hand, engineering process control automatically adjusted the process to stay close to target. A combination of the two ideas is necessary to achieve the best control. Ultimately, Alberto and I published a book about this in 1997, and my good friend, Carmen Paniagua-Quiñones, co-authored the second edition, published by Wiley in 2009.1

  Figure 18.1 Albert Luceño and me.

  Alberto and his family did indeed spend a year in Madison, and he and I worked very well together. During this time, Claire and I became close to Alberto and his wife, Marian Ros, who was a physician doing research on in vitro development (Figure 18.2). We also became very fond of their eight-year-old son, Mosqui (short for Mosquito, although he is really named Alberto). This was the beginning of a long friendship, with many visits back and forth between Madison and Santander.

  Figure 18.2 Marian, me, and Claire.

  We were delighted to see each other again in the summer of 1993, when the International Statistical Institute was holding its meetings in Florence. Claire and I were walking across the Piazza della Signoria when Mosqui spotted us and started running, calling out to us in perfect English, “My mother is pregnant!” If this had been a secret, hundreds of people now knew it. During our time in Italy, Mosqui was a policeman, watching his mother carefully to make sure she didn't drink coffee or do anything else that a pregnant woman “should not” do.

  In Italy, we once again saw Agustín Maravall, who was professor of economics at the European University Institute in Florence. He and his family lived in an apartment in the beautiful village of Fiesole, which sits in the green hills above Florence. Agustín recommended that we stay nearby, in the Villa San Giralomo, a pension run by the sisters of the Little Company of Mary. Often called the “Blue Nuns” because of the color of their habit, the sisters of the Little Company of Mary are from an Irish order that had run a hospital and a nursing home in Fiesole. The nursing home had been in the villa, which still housed a small, elderly resident population. The views of Florence from the villa and its gardens were beautiful. Less captivating but still entertaining were the shouts of the plumbers who called to each other through the floors when the plumbing failed during our visit.2

  We flew from Italy to Spain, where we went to Santander for the first time. Marian and Alberto welcomed us into their home and were wonderful guides to this special region of Spain. Santander is in the Spanish state of Cantabria, near the mountains that stretch across the north of the country and reach down to beautiful beaches. The city lies on an expansive bay, and the location has a rich history. In 1589, a year after the Spanish Armada, Queen Elizabeth received intelligence (false as it turned out) that a number of ships from the remainder of the Armada were being repaired at Santander in readiness for a second try at invading England. So she sent Drake off with a good-sized fleet to take a look. Because the bay is so large, a steady wind blows in from the sea. Drake could not see anything that looked like a threat, and his captains told him that once he entered the bay, he would have great difficulty getting out. So they stayed out. Both the history and culture of Santander pulled us in, and we knew that we would be back.

  I believe that it was on this same trip to Spain that we had a memorable boating adventure with Xavier Tort. He was my colleague in Barcelona who had helped with the Spanish translation of Statistics for Experimenters. He had recently purchased a sailboat and offered to take Claire and I out in it. He raised the sails a bit prematurely, and we started to drift perilously close to an escarpment. Xavier was unable to start the engine, and soon the wind was pushing us against this wall, although we were able to push ourselves away from it. Finally, as things were getting worrisome, he got the engine going and the rest of our trip was uneventful. He recently wrote to me that if the boat had crashed, “I would have become a very famous statistician (my only chance to become so), the statistician that killed George Box!!”

  We saw Alberto and Marian again when they spent the summer of 1994 in Madison, the first of five summer visits that allowed Alberto and me to work together. Each time Claire and I would find a house or apartment for them to rent, and in 1994, we also found a crib and other items for their baby, “Peque,” then just a few months old (Figure 18.3). These visits were filled with happy times, with many shared meals, trips to the zoo, and spontaneous games to amuse the children.

  Figure 18.3 Peque, Mosqui, Claire, George, and Alberto in Madison.

  Early in 1995 I was to receive an honorary doctorate from the University of King Carlos III in Madrid. Before the ceremony, Claire and I stole a few days in Santander to visit Alberto and Marian and their children. They greeted us with dinner and champagne and took us to the small village of Silos to visit the monastery, where we heard the monks singing Gregorian chants. We also went to Haro, in La Rioja, where we visited a very old and large winery. It was our good fortune to have as our guide the owner's granddaughter who was also the great, great, great granddaughter of the founder. She was a graduate of the Department of Viticulture and Enology at the University of California, Davis, and she spoke perfect English. We walked through seemingly endless underground tunnels stacked with wine in barrels all the way. Eventually we entered a room where there was wine more than 100 years old. The room was full of cobwebs. The guide explained that this wine was undrinkable but that the spiders were encouraged because they kept down a species of fly that ate the corks.

  Haro is a little village where everyone is involved with making wine. While there we went to a small restaurant for a lunch of tapas. The man behind the bar came over, took our order, and then asked what we wanted to drink. He was clearly surprised and upset when, after some discussion, we said that we didn't want anything to drink. He returned to his place behind the bar, but he was clearly offended, and after looking at us a number of times, he dived under the bar, came up with a bottle of wine, marched over, dumped it on our table, and spoke loudly and indignantly. Our friends translated this as, “Here, drink this. You can have it for nothing!”

  After touring the winery, Claire wrote this poem:

  The Winery at Rioja

  (January 22, 1995)

  the winery López de Heredia, windows

  connect the office to the cellar, etched in scenes

  from the south of Spain – how she would have loved

&nbs
p; these windows I think, she and this place the same age

  descend the newly swept stairs with other tourists, down

  down into the hand carved cave and thousands of miles

  away she died and I once drank this wine and still feel the pain -

  the death of her sons, I back away from the guide's racing Spanish

  resist translations, chiseled caverns alive

  need no explanation – feel the walls, smell the moss, dust thick

  the oak and wine invade my senses, swirling with smells

  dizzy with the years of richness, I mourn the loss at home

  the oak barrels few artisans left to make, the good money

  not enough to offset the hard work, what will happen when coopers

  are gone and the machine-made barrels feel no care or love

  freed to feel the decay of this century, I enter the world of bottles

  1/2 million Rioja stacked and never touched, do not disturb

  their peacefulness, they have been laid to rest and will enjoy

  a second life in 3 or 5 or 7 years – quiet , coated with dust, they wait

  last the cementerio, the oldest bottles and Rose the mother

  at their final resting, the bottles safely stacked in nichos,

  glasses are carefully placed on the lid of a huge discarded cask

  and at its center the first vine of this winery, dripping with age

  and decay, like the bridal banquet in Dicken's Great Expectations,

  yet nearby a 1987 Viño Tondonia awaits its beginning

  soon it's tenderly uncorked with instructions about tasting a Rioja

  some tourists serious, others snicker and ignore the sober ceremony

  – in the wine I taste the oak, the grapes, the decades – death and life fill my mouth

  After a few days in Santander, we all drove together to Madrid where I was to receive the honorary doctorate. Daniel Peña, who had organized this occasion, is a remarkable man (Figure 18.4). In the 1980s, he had become concerned that, although there were two universities in Madrid, there was very little available for poor people. Along with some political colleagues, he persuaded the government to build a new university. Founded in 1989, and housed initially in a former army barracks, the new university was named for King Carlos III who had been an enlightened 18th-century ruler who had promoted education and the arts.