Jessie and I lived in Sale, about 12 miles from Blackley where I worked. A number of people from ICI lived nearby, so we jointly hired a deluxe coach to take us to and from work. At times these trips offered unexpected entertainment. After the war, new cars were almost impossible to get, so it was regrettable when one morning a brand new car was badly damaged in a collision with our bus. While we waited for the police to come, the driver of the new car sat down disconsolately on the sidewalk. A small crowd of somewhat grubby children who, I suppose, were on their way to school, quickly gathered. After a time someone in the bus must have been telling a joke because a small urchin peering through the door of the bus and greatly impressed by the damage that our vehicle had done to this beautiful car, said to the hapless car owner, “Here, mister, this one's laughing!”
On another bus journey, I was sitting next to a very dapper little man who said to me, “I understand you're going to America.” I said, “Yes.” “What are you going to do with your house?” he asked. I replied that we intended to let it. He said, “Our family did that when we went abroad and we had a most unfortunate experience.” “How so?” I asked. “Well,” he replied, “the first people who rented were not so bad, although they did steal the silverware. But the people who rented after them were worse. It seems they were running a disorderly house.” He then went on to say, “Our house rather lent itself to that because it had so many bedrooms.”
Later I was in a car pool. In those days, everyone heated their houses with coal. This and the climate inversions that occurred in Manchester periodically produced “pea soup” fogs. These fogs were like nothing I have encountered anywhere else, and when they happened, we would be told to quit work early and try to get home. The visibility on such occasions was literally no more than four feet. To navigate we had someone, who was just visible to the driver, walk by the side of the car and someone else, who was just visible to the first man, walk on the sidewalk.
I remember on one such expedition the road must have turned sharply. But we didn't, and together with several other cars, ended up in someone's garden. The owner of the house was jumping around solidly cursing us as we maneuvered across his flower beds, backing and turning in order to find our way out.
At some point, “smokeless fuel” became available, the burning of coal was banned, and the pea soup fogs disappeared.
1 G.E.P. Box and G.C. Tiao, “A Canonical Analysis of Multiple Time Series,” Biometrika, Vol. 64, 1977, pp. 355–365.
2 G.E.P. Box, “Evolutionary Operation: A Method of Increasing Industrial Productivity,” Applied Statistics, 1957.
3 G. E.P. Box, J.S. Hunter, and W.G. Hunter, Statistics for Experimenters, 2nd ed., Wiley, New York, 2005, p. 163.
“When I use a word…it means just what I choose it to mean—neither more nor less.”
Chapter Four
George Barnard
George Barnard was a brilliant mathematician (Figure 4.1). He had come from a poor family and on his own merits went to Cambridge to the best math department in Britain. For some reason, when you were with George Barnard, interesting and unexpected things tended to happen.
Figure 4.1 George Barnard making tea.
Imperial College, where George had his office, was one of the institutions built by Prince Albert with the proceeds from the Great Exhibition of 1851. At one time, its rooms must have been very large, but expansion of the faculty had, over the ages, produced many extra walls. Consequently, it now had a number of very narrow rooms with very high ceilings. One of these was George Barnard's office. I remember one particular day when I had gone to see George to discuss a problem, but we never really got to it. Soon after I arrived, George said, “You must find it stuffy in here. I'll open a window.” That window must have been closed for a very long time for it proved impossible to move it. George's secretary, Miss Mills, who was used to coping with all kinds of emergencies, brought a screw driver and a hammer and every possibility was tried. Eventually, after a lengthy struggle, the window was opened. But then, almost immediately, two pigeons flew in.
There was a passageway that passed through George's office, and the pigeons settled on its roof. For a long time George, his secretary, and I stood on chairs with pointers and other instruments trying to get the pigeons to go back out of the window. In the end we were successful, but then George said, “Let's go to lunch.” We strolled down the road, no doubt very deep in conversation, and we came to a pedestrian crossing. Now in England, the strictly enforced rule was that no vehicle was allowed on a crossing if a pedestrian was already anywhere on it. But on that particular day, there was a very expensive car, all chromium plate and headlights, impatiently edging its way onto our crossing. The driver looked very angry, for he, no doubt, regarded our progress on the crossing as too slow. When we came level with the car, George became aware of this vehicle inching its way onto the crossing and he immediately stepped in front of the car, banged with both hands on the hood and shouted, “Back! Back! Get off the crossing. I'm not going to move until you do!” Eventually of course the motorist did retreat, but he didn't much like it.
Many of my most memorable experiences with George Barnard occurred when we were both members of the Research Committee of the Royal Statistical Society. The committee was responsible for spotting talent and finding speakers with something new to say in “read” papers for the Society. George had become a consultant to ICI, and from time to time, he visited the Dyestuffs Division and was familiar with what we did there.
I had been working in particular with a chemist, Dr. K. B. Wilson, on a general technique called “response surface methodology” for improving a process. George suggested that Wilson and I write up our work so that it could be presented and discussed as a read paper at a Research Committee meeting. This idea astonished me at first, but eventually we wrote the paper and submitted it. One of the referees, a very well-known statistician who had been a student of Fisher,1 said the paper should be rejected. He felt so strongly about this that he threatened to resign from the Society if it was accepted. However, in the end, in some way that I have never discovered, his objections were overcome and the paper was presented and discussed and duly appeared in the Journal of the Royal Statistical Society.2 The paper was a great success and, as I will later explain, was the cause of my first coming to the United States.
When I was elected a member of the Research Committee, there were about eight of us and we came from all over England, Scotland, and Wales. George drove a van called a “Dormobile,” which could carry a number of people. Typically, after the meeting in London, he took all of us to our appropriate railway stations from which we could travel home. At that time, a topic about which there was a great deal of dispute (sometimes heated) concerned theories of statistical inference. Among these were the Neyman–Pearson theory, Bayes' theory, likelihood, and fiducial inference.
This last was a gallant (but I think unsuccessful) attempt by Fisher to avoid certain difficulties encountered with the others. Now George was a great admirer of Fisher and thought he could do no wrong (indeed it did turn out that Fisher was right about most things). However, Dennis Lindley, who was one of George's passengers, believed that Bayes was right, and Fisher was wrong, and expressed his opinion forcibly. George was incensed and went through two red lights. When the cries of “Oh my God!” had died down, we all took a solemn vow never again to discuss theories of inference while traveling in George's dormobile.
David Cox (later Sir David Cox) was a member of the Research Committee at the same time that I was. We often heard, “Ha Ha! Box and Cox! You must write a paper together.” So we thought we would do that, but what should the paper be about?
In the late 1880s, there had been a comic play called “Box and Cox.” Box had a night job and Cox had a day job. The woman who kept the house where they boarded got paid double by renting a room to Box during the day and the same room to Cox at night. We got rather fed up with the teasing, but knowing the story, we decided that obviously
our paper should be about transformations.3 In this spirit, we deliberately included two different derivations of our results, one using likelihood and the other Bayes. At the meeting, the discussants tried hard to find out who wrote what, but we were not about to tell them. From a practical point of view, it didn't make much difference anyway.
After our research meetings, we usually took the speaker out to dinner at an Italian restaurant called Bartorelli's. At one particular meeting, George had strongly criticized the speaker on a specific point. As we were walking over to the restaurant, the speaker said to George, “I thought you were rather hard on me tonight. I heard you say the same thing yourself not more than 12 months ago.” George looked surprised and slightly indignant and said, “You don't imagine I'd necessarily think this year what I thought last year, do you?”
In 1953, when I first came to the United States, Senator McCarthy was very much in the news, so I made a point of going to Washington to observe the activities of the House Un-American Activities Committee. I witnessed first hand how the epithet “communist” could be used to destroy a person's reputation. During this time, it happened that many statisticians on the American side of the Atlantic wanted to get a chance to talk with George Barnard. But there was a problem. Immediately before WWII, George had spent some time at the math department at Princeton. The FBI claimed that while he was there, he had formed a Communist cell.
As a young student, George, like many, had sympathies with the Communist Party. Later, when I knew him, he was liberal in his outlook and not exceptionally so. What the “cell” at Princeton had been is unclear, but the FBI refused him a visa numerous times. There was an outcry by many U.S. statisticians who worked very hard to show that George had statistical ideas they needed to know more about, and that he did not constitute a danger to the United States. Over the next year or two, there was much writing to senators and various attempts to get him a visa.
At one point, we got around the difficulties by having George present a lecture course on the Canadian side of Niagara Falls, and many American statisticians came to hear him. But George had greatly enjoyed being in the United States and had looked across the falls with nostalgia. Eventually the FBI said they would give him a visa if he would give them the names of all who had been in the group at Princeton. This, of course, George refused to do, but in 1961, when George was invited to visit the National Institutes of Health, the FBI agreed to allow him in with a “waiver.” I think someone presented a document to George, covered it with his hand, and said, “Don't try to read this, just sign it.” And eventually George did.
George's problems with the FBI did not end, however. From 1975 to 1981, when George spent a part of each year as professor of statistics at the University of Waterloo in Canada, difficulties entering the United States still plagued him.
The adventures with George went far beyond statistics. One of the most memorable houseboat trips I organized was with the Barnard family. George and Mary had five children, so we needed rather a large boat to provide sleeping accommodations for everyone. We usually started the trip near Windsor, traveled the Thames to Oxford, and then back again, going about 160 miles in all. Every few miles along the river there was a lock. The lock keepers were ex-Navy men, and we were expected to observe proper protocol. If you were going upstream, you would wait for the lock gates to open, then you moored the boat with the appropriate length of rope, and without bumping the sides of the lock, kept the boat steady as the lock filled up. Most lock keepers were friendly and helpful and would sell you fresh produce from their gardens. No one was in a hurry, and we spent about three days going up to Oxford and another three days coming back.
The children were very watchful of their prerogatives. If, for example, one child was permitted to steer for a bit, the other four children watched very carefully, and noted the time the amateur helmsman was at the wheel. Then each insisted that they should be allowed to do the same and for exactly the same amount of time. If their time was half a minute short, he or she protested that they had “not had a proper go.”
On a different occasion, we went very early, in April. One chilly morning, I started the engine just as George released the mooring rope, which wound itself very tightly around the propeller shaft. The Thames is cold in April, but there was nothing to do but to take turns diving under the boat with a penknife, and bit by bit we cut through the layers of rope until our breath gave out. Neither of us had brought swimming gear so Mary, George's wife, leant each of us a pair of her drawers. Although it took a long time, we were finally successful, but both of us were frozen to the bone. Mary, who was a Scot and a nurse, gave each of us a half-tumbler of whisky, which made us warmer but a bit unsteady.
There were many places to tie up along the river and to explore the surrounding countryside or buy provisions. On one occasion, we got off the boat, walked across some fields, and came to a village with a very small pub. The person managing the pub seemed to think that we were posh people, so he opened up another room for us that was reserved for special occasions. There was a piano there, and I sat down and played. Soon after our arrival, a large bus of people with strong cockney accents arrived and it was clear that they were out celebrating. One of them peeked into the room and asked, “Can anyone come in here?” When we told them yes, apparently so, I was soon joined around the piano by a boisterous group, and we sang songs like, “Knees Up Mother Brown,” a tune that inspired others in a vigorous dance around the floor.
On one occasion when I was making a brief visit to England from the United States, I was anxious to talk to George, but upon inquiry, it turned out that he was chairing a meeting of the Industrial Applications Section of the Royal Statistical Society at Jesus College, Oxford. I drove to Oxford, and we had a chance to talk. I arranged to stay at the College, but I kept in mind that I had to get up very early the next morning to drive to Heathrow to catch a plane. The person who had organized the meeting was Miss Joan Kean, the secretary of the Industrial Applications Section who was staying at the College. I told her of my need for an early start in the morning, and she lent me her alarm clock.
The following morning I was up and dressed at 5 a.m., but I found I could not get out of the college. There was an extremely large and formidable locked door and a high wall that had been there since the year 1500 or so. So I started to look for another exit. I went all around the college twice but could not find a way out. There were stairs leading down to some sort of a dungeon, so I wandered about down there, but again there was no exit. It finally occurred to me that over the centuries the college had wanted to keep the undergraduates inside and they had done a first-rate job.
I then remembered that I had to return the alarm clock to Miss Kean, and that this would be a good opportunity to consult with her. I knocked on her door, and she eventually appeared in pajamas and dressing gown. I told her my problem. She said, “There surely must be a way out. I'll come with you.” So we repeated the search I had already made without success.
After some time, we eventually found a ladder and discovered that there was an outer building attached to the wall, so we climbed up onto the roof of this building and we were just lowering the ladder down to the road on the other side when we saw a policeman looking up at us. We must have looked rather strange: I was carrying a suitcase, and she was wearing pajamas and a dressing gown. I said to the policeman, “I am Professor Box of the University of Wisconsin, and this is Miss Joan Kean, the secretary of the Industrial Applications Sector of the Royal Statistical Society who is helping me. I need to leave and catch a plane, but there seems to be no way out of this college.” I think this kind of thing must have happened before, for the policeman didn't look very surprised. After a little thought, he said, “Right 'o. I'll 'old the ladder and ‘elp you get down.” So I said goodbye to my companion on top of the wall and was soon on my way to Heathrow.
Later when I was in England, I told this story at a dinner party. As we were putting on our coats getting ready to
leave, a very, very proper English lady took me to one side and said, “George,” and I said, “Yes, Elizabeth” and she whispered to me, “I just wanted to let you know that I think you were quite right to introduce Miss Kean to the policeman.”
I started smoking when I was first in the Army. When it came time for a break, the sergeant would say, “Smoko.” There were few, if any, abstainers. Later, when I met George Barnard, he encouraged me to give up the habit because he had just become aware of the research done by Bradford Hill and Richard Doll that revealed the possible link between smoking and lung cancer. Their 1954 paper, which has been unjustifiably overshadowed by later studies in the United States, gave conclusive proof of the link.4
Hill was a brilliant medical statistician. In the 1940s, he conducted the first randomized clinical trial in a study of the effect of streptomycin in the treatment of tuberculosis. Fisher, who had pioneered randomization in field trials at Rothamsted 20 years before, was on friendly terms with Hill. In 1954, he proposed that Hill be made a member of the Royal Society, and Hill became a fellow that year. But Fisher was also an avoid smoker but by the late 1950s, Fisher openly criticized the smoking study. He argued that correlation, which had been proved, and causation, which had not, were not the same thing—for example, a genetic factor in an individual that predisposed him to smoke might also predispose him to cancer. More generally, x may be correlated with y because both are correlated with z. The relationship between Fisher and Hill cooled considerably, but the causative evidence for a link between smoking and cancer grew stronger as Hill and Doll's study continued. I found the evidence compelling, and with George Barnard's encouragement, I stopped smoking.
An Accidental Statistician Page 7