Blinding with Science
Of nearly every theory it may be said that it agrees with many facts: this is one of the reasons why a theory can be said to be corroborated only if we are unable to find refuting facts, rather than if we are able to find supporting facts.
Popper (P. H. 111n)
In most people's eyes, whatever may be said against science, it works. The facts of space travel, television, computers - to name just three of the spectacular products of inventions based on scientific discoveries - proves to most people's satisfaction that science is not nonsense.
But although, in performing his task of formulating theories and testing them, a scientist may need to amass data and possess the skills necessary to operate an electron microscope or to programme a computer, these skills of themselves do not make a scientist. It is possible for people who possess these skills to appear to be speaking with the voice of science when actually they are speaking with the voice of dogma. Genuine scientists, as well as those who are only skilled in the use of the tools of science, can lapse from the high discipline of scientific methods and begin to pontificate. To the general public it sounds like science and they are misled.
The difference between adopting a scientific outlook in the Popperian sense and an inductive one may sometimes seem very slight, but it is usually far-reaching in its consequences. A good example is the relationship between housing conditions, particularly overcrowding, and the intellectual performance of children at school, allegedly discovered by the authors of From Birth to Seven (Ronald Davie et al.), the second of the `longitudinal' studies of all the children born in the United Kingdom in one week of 1958. The fallacy inherent in the conclusion was cleverly exposed by Tyrrell Burgess, a fellow-disciple of Popper, in New Society (1975). As the argument is rather subtle and as there are probably many unexposed instances of the same kind of fallacy, I propose to go into the matter in some detail.
In their book the authors had written: `Poor housing is often mentioned as one of the contributory causes of school failure.' They then quoted the following from Professor R. N. Titmuss's introduction to the 1964 edition of R. H. Tawney's Equality: `We delude ourselves if we think we can equalise the social distribution of life's chances by expanding educational opportunities, while millions of children live in slums, without baths, decent lavatories, room to explore and space to dream.' Having dropped these prestigious names the authors duly found that over the country as a whole `the effect of overcrowding (defined as more than 1.5 persons per room) was equivalent to two or three months' retardation in reading age at age 7'. Surprisingly to them the effects of shared or absent basic amenities (hot water, indoor W.C.s, etc.) was much more - nine months retardation. By juggling the figures and taking into account family size, the retarding effect of overcrowding was brought up also to nine months. From this the authors drew their very definite conclusion: `The results have demonstrated clearly the relationship between poor housing conditions and overcrowding on the one hand and on the other educational performance . . . at the age of seven'.
However, what they did not mention, although they were contained within their own results, were other figures which Burgess extracted to fuel his fire. These show that, while Scotland as a whole contains the greatest percentage of children from overcrowded conditions (39.2 per cent compared with London and the South East's 13.4 per cent,, and Eastern England's 6.6 per cent), it also produced considerably the best reading results: 40 per cent good readers compared with 31 per cent for each of the two areas I have singled out, and 28 per cent for Wales, which was the worst reading area although one of the least overcrowded (8.7 per cent). And Scotland was no flash in the pan. The second and equal third most crowded (Northern and North West) were second and third for the percentage of good readers. The statistics of this report admittedly do not exclude the possibility that bad home conditions are incompatible with good reading ability (nor, to take a frivolous example, do they exclude the possibility that boys called Robert are never good readers). It is admittedly possible that the 40 per cent of Scottish good readers came only from the 60.8 per cent of homes that were not overcrowded. In Wales, on the other hand, that assumption - that all the overcrowded children were bad readers - would still mean that 7 out of every 8 of the bad readers (63.3 out of 72 per cent of bad readers) would have come from the 91.3 per cent of the homes that were not overcrowded. In the case of an overwhelming majority of bad readers therefore, overcrowding could not have been a factor or link. Nor could there have been in their case any `relationship between poor housing conditions and overcrowding on the one hand and on the other educational performance' as the authors had claimed. And even in Scotland at least a third (20.8 out of every 60) of `bad' readers must have come from `good' homes.
Clearly what the authors had done was to extract their own preconceived conclusion from the mass of data they had accumulated. They claimed that their data confirmed their theories when what they should have done was, as Burgess pointed out, to have formulated their hunches in a form of a precise theory such as `there is a rate of overcrowding x at which y per cent of children failed to reach a "good" standard r'. This could have been tested, and on the evidence they had already accumulated would have been found to be false.
The authors, steeped in the idea that all one has to do is to look for evidence to confirm one's hunches, genuinely could not see the force of Burgess's criticism. David Donnison, then Director of the Centre for Environmental Studies, had written in the foreword to their book: 'The patterns glimpsed in the National Child Development Study are so deeply embedded in this country's economic and social structure that they cannot be greatly changed by anything short of equally far reaching changes in that structure.' Evidently they all had their eyes fixed in this one direction. In replying to Burgess, they still maintained that `to expect schools to cope with this situation [by implication `the situation' was: being expected to teach reading successfully in the face of bad home conditions] unaided by other agencies and by improvements in social conditions is to impose an unreasonable and impossible burden and to fly in the face of virtually every piece of research on this topic'.
I thought an analogous case might help them and I cited the fight against tuberculosis, a disease universally regarded as fostered by overcrowding and other unhygienic conditions, although when most rampant it had not only spared many of the poor but also claimed its victims among the well-to-do. Yet, by tackling the disease itself and its means of spread and by early detection and treatment, it has been possible almost to eradicate it from this country (except in areas where it is constantly being reintroduced from abroad) in spite of the fact that poverty and overcrowding have not been abolished.
The holistic outlook, not very different from the revolutionary outlook, tends to encourage the adoption of vague ideas such as that bad social conditions are the cause of (or a contributary cause of - it makes little difference) failure to read. And you can easily find some facts to confirm it - just as you could confirm the theory that wealth is the result of winning the pools. It is again a case of the swans. The authors were so busy counting their white ones that they ignored the black ones that were staring them in the face.
If they had been right in their conclusion, then the consequence must be that it will take a very long time to achieve any substantial improvement in reading standards - as long as it will take to eliminate overcrowding and, provide universal basic amenities, or, if Donnison is right, even longer - until the country's whole economic aid social structure has been changed. But what their results actually show is that overcrowded home conditions are not a factor in the poor achievement of many children, and that in some areas most of the children who in fact perform badly are not overcrowded. Their results therefore suggest that it may be possible to improve the achievement of all children in spite of continuing bad conditions. Even the dissemination among teachers and parents of this piece of knowledge alone would probably immediately help children who are at school now (whereas the Donnisons of this world cannot hope to benefit the present generation of school children). For there is abundant evidence from other sources that children's attainments closely match teachers' expectations. So long as educational sociologists tell teachers that they cannot expect good results from children from poor homes, so long will many teachers not attempt what they are told is impossible. Burgess's argument, of course, does not detract in any way from the need to improve bad housing conditions. But the case for doing this is easily made on its own account and does not require the help of a bogus educational theory.
Burgess commented on the quotation from Donnison as follows:
"This kind of analysis is typical of educational sociology - particularly the conclusion that there is a 'combination' or `cycle of deprivation' which means that you cannot change anything unless you change everything. The practical consequences of these `findings' have undermined the search for effective solutions to grave social problems and encouraged the view that schools are helpless victims of an independent and destructive social process. Not only are the poor trapped in a cycle of deprivation, but the agencies that might help them are trapped in a cycle of impotence."
The 'findings' of some educational psychologists have had a similar undermining effect. Professor Denis Stott points out the fallacy in the assumption of `disability-producing deficits' as explanations of poor school performance - dyslexia and hyperactivity are the best-known examples. These deficits, he says, `have become an academic myth - a myth convenient to the college academic and the school psychologist because it ... provided them with a professional mystique ... convenient to parents because it absolved them from the shame of having a dull or retarded child, and convenient to the teacher because it excused what looked like teaching failure'. These mythical deficits (presumably in the central nervous system) have been assumed as the only explanation of poor academic performance in children whose nervous systems are, by any other test, intact. (Winston Churchill would, by modern standards, have been `deemed' hyperactive and a suitable case for treatment; and Einstein was so late in learning to talk, according to Jeremy Bernstein, that his parents were worried that he might be mentally deficient!)
It would not be true of course to say that all sociologists and educational psychologists think like the Birth to Seven authors. It just seems like it. One can hardly open a journal without catching a sociologist in the act of white-swanning. I open the Guardian and find Nicky Hart from the Department of Sociology of the University of Essex, listing, quite rightly, some of the many ways in which the vital statistics for social class V (unskilled workers) are throughout life markedly worse than those for class I (professional and managerial). 'These differences in life chances', Hart says, `reflect the distribution of material advantage in Britain' and he makes it clear that he means by this that material inequality is the cause of these differences. One can be quite sure that it is not the only cause by virtue of the fact that the life expectancy of women in each social class is so consistently greater than that for men in the same class. Not even a male chauvinist pig could allege that women enjoy a marked 'material advantage' over men in their own class. Other factors must be involved; and there is much evidence that cultural habit is more important in this regard than material wealth. Just to take one fact to support this contention: the risk of serious respiratory disease in the first two years of life is increased by half for a child whose mother smokes and doubled if the father does also. Smoking is much more prevalent among young peopleit in class V than in classes I and II and cannot be called material deprivation.
My second example of unreason masquerading as science concerns the distinguished biologist, the late C. H. Waddington, formerly Professor of Animal Genetics at Edinburgh, who at the end of his life wrote a book, published posthumously, entitled Tools for Thought.
In it, Professor Waddington makes a specific reference to Popper, but at the same time makes it clear that he does not begin to understand the force of Popper's ideas: `Karl Popper argued that the real method of science is not to try to verify statements, but to disprove them. A surprising number of scientists, including very successful ones, have expressed agreement with him.' This is a mis-statement of Popper's position. Waddington omits the very first step - the hypothesis, the bold conjecture. He goes on: ` ... suppose we have a hypothesis like "if a match is put to twigs a fire starts" ... sometimes the fire does` not start, the twigs are wet, or something; and this does not completely disprove the suggestion that matches have something to do with starting fires.' This is a case of careless and unprecise formulation. The original hypothesis is definite `a fire starts', but in order to make Popper look silly, he changes it when he talks about disproof to `something to do with starting fires', a different and vaguer proposition. `The mistake made by both sets of philosophers', he goes on, ` - those who asked for verification and those who would settle for falsification - is that they demand 100 per cent certainty: and that is something we can never have in the real world.'
I can only describe this argument as silly. Waddington does not seem to have made any effort to discover what either set of philosophers is talking about. Popper, of course, has stipulated that hypotheses must be formulated clearly, and if the hypothesis is so wide though clear as `a lighted match will always ignite a bunch of twigs', then one failure does, quite properly, falsify it. One then tries again, perhaps specifying the degree of dryness and thickness of the twigs, type of wood used, the absence of draught, temperature of the air, etc. One pursues this until one is unable to refute it. That is all that Popper is saying. Now Waddington goes on to give, as examples of how silly the philosophers are, that when `Mendel discovered the laws of heredity he was not trying to disprove them'. This is of course true; but what are now known as the Mendelian laws were the bold conjectures (the step that Waddinton omitted from the process) which invited refutation. As it happens the theory has withstood all attempts to refute it and so is generally accepted.
In his introduction to the book, Waddington goes more seriously astray through not having made an effort to understand Popper's theory of the growth of knowledge and the function of social science. He starts off well enough: `We have been trained to think, or have accepted as commonsense, that what goes on around us can usually be understood as some set of simple causal sequences in which, for instance, a causes b and b then causes c, then c causes d and so on. This is only good enough when a causes b but has very little other effect on anything else, and similarly the overwhelmingly most important effect of b is to cause c. Many of our own individual actions still have this character.' So far so good. But, `The change which has occurred, or is occurring now, is that the effects of human society on their surroundings are now so powerful that it is no longer adequate to concentrate on the primary effects and neglect all secondary influences.' And later: 'No powerful action can be expected to have only one consequence, confined to the thing it was primarily directed at.' But it never could. Clearly he has never considered the house-buying example (page 10). a practically never does nor did cause only b. The upshot of Waddington's argument is that things are now so complicated that we must use the new 'powerful' tools of systems-analysis allied with the inevitable computer.
The conclusion is wrong and if put into effect would be disastrous because it ignores the difference between human society and complex combinations of such things as industrial processes. For the latter systems-analysis is highly successful. The technique was evolved during the second world war and was successful then, even though it involved manipulating human beings, because in the circumstances of war their individual aspirations could be ignored. The purpose of the whole enterprise was to win battles. But cities and nations have no purpose in this sense and the individual aspirations of their citizens cannot be ignored. It would be a centralising of power not just in human hands but in the computer's hands, and, as has been said before, although it is easy to centralise power it cannot be wisely wielded, the reason being that the computer cannot be informed of all that knowedge in all those individual minds which is essential for the wise wielding of power.
Any attempt to do this would be up against the same objections as are being rightly made now against C.O.B.A. for the motorways. The authorities would be calling on a private line to God. It would be an attempt to answer criticism with complexity.
Rupert Crawshay-Williams has told of the setting up of the Metalogical Society in 1949 by Professor A. J. Ayer, with the object of getting philosophers and scientists together. He said that the eventual fading out of the society was in part due to the fact that the scientists tended not to recognise as their own the aims and methods which the philosophers attributed to them and this was despite the fact, he says, that 'they certainly accepted Karl Popper's famous denial of the traditional theory that scientific method uses induction'. He went on that 'The philosophers (including Popper) assumed that the task of science was to discover . "absolute-all" statements.' And he gives an example of this as 'metals expand when heated' and later mentions, to refute this statement, the new welded railway lines which do not expand when heated. This incredible muddle of imprecision is strongly reminiscent of what I have just quoted from Waddington; and one cannot help wondering whether he was the scientists' spokesman. The statement 'metals expand when heated', with no mention of conditions, is like Waddington's 'something to do with starting fires'.
It is in a way typical of what has gone wrong that it should be thought that systems-analysis is the answer. Here is something successful in war but not directly applicable to peacetime human problems. On the other hand war-time lessons of universal applicability have been ignored. Professor R. V. Jones describes how it was proposed that our heavy bombers should be fitted with equipment to prevent losses from icing-up; but the proposal was dropped when it was realised that more bombers would be needed to make up the loads lost by the weight of the de-icing equipment and that more bombers sent out would mean more shot down. It was calculated that the extra number lost to enemy action would be significantly greater than the number that could possibly be saved by prevention of icing. The fitting of the equipment would have increased casualities rather than saved them. (Just as motorways cause more casualities than they save.) Professor Jones commented: `This is an example of a phenomenon where an action can have the opposite effect from that intended, and a lesson always to be borne in mind by politicians and administrators.' Jones was one who did not suffer from tunnel-vision but, as we have seen, his lesson has not always been borne in mind.
The attraction of magic machinery and the over-emphasis on the large scale has obscured another war-time lesson pointed out by R. V. Jones. The station commander at Tangmere during the flying-bomb assault had asked one squadron commander how he succeeded in getting twice as much work out of his squadron as did the other two. The answer was that this particular squadron was organised on the old system that had operated throughout the Battle of Britain. Each pilot had his own aircraft which was serviced by a devoted ground crew who regarded themselves as part of a team with the pilot. His victories were their victories. The system was extravagant in ground crews; and one of the earliest results of 'Operational Research' was to show that substantial savings could be made by changing to a kind of central garage system into which each aircraft was sent after each operation and from which each pilot could draw a serviced aircraft. The other two squadrons at Tangmere had changed to the new system. There were a number of snags to it, Jones recalls, but the main one was the loss of the team spirit which in the old individual crew system 'somehow drew substantially more work' out of the ground crew when emergencies cropped up. 'Since this is rarely quantifiable', Jones comments, 'it is usually not taken into account by any plan to improve administrative efficiency.' Computers can only compute measurable quantities. Inevitably they leave out of account what is not measurable. Our science may thus blind us to what is not measurable but not necessarily unimportant.
On the whole, in my own profession of medicine, we have I think managed to avoid the main conceptual mistakes I have outlined because our chickens usually come home to roost rather fast, a situation that is unlike that in, for example, the civil service where the authors of an idea have usually retired or moved elsewhere by the time the effect of their actions has become apparent. In particular we are from the start taught to regard as a vice the special form of solutioneering open to us (and which we are constantly tempted by patients to indulge in), that is embarking on treatment before making a diagnosis. Once led into that trap, one is liable to find oneself unable to distinguish between the unintended effects of treatment and the unidentified disease. This again tempts one to indulge in adjustments to the treatment in a manner analogous to Popper's unplanned planning.
To doctors the making of a diagnosis is always tentative, it is a hypothesis which is open to refutation and if refuted must be changed. From the very start of our careers, we cannot help realising that nothing is certain. We can never be absolutely sure of the outcome of any treatment or operation. This fact gives rise to great difficulty with the public, who expect certainty and are always pressing doctors to commit themselves and wanting a second or third opinion if certainty is not forthcoming. Unwillingly we are often forced to reveal our conjectures, such as that the possibility of malignancy must be excluded. (i.e. One hypothesis is that this is a case of malignant disease and we will do tests with the object of refuting that theory.) But inevitably the patient's relatives will get the message that the doctors think it is cancer. The tests whose `results' are awaited with so much eagerness by patients and their relatives and received with so much disappointment when `negative', are again attempted refutations.
A British doctor sees his immediate diagnostic task as one of excluding serious, that is potentially lethal or chronically disabling, illness rather than of answering the question 'what is it?', a question which need not be pursued at all if the symptoms rapidly subside. This Popperian, but traditional, approach contrasts with American practice which tends to aim exhaustively for certainty. I think this philosophical difference accounts to a large extent for the fact that we are able to afford a medical service which, for all its faults and gaps, is broadly comprehensive, while the Americans in spite of their wealth cannot.
The British doctor on the basis of a few simple questions, answers, and observations, can in most cases make a provisional diagnosis of this excluding sort and decide either on treatment or on a course of wait and see. Because there is no financial transaction, he can see the patient as often as necessary to check on his original diagnosis and if need be change it, in what may be a series of very brief consultations. If the patient had to pay for each attendance, he would expect more time and more tangible action from the doctor.
The American doctor, aiming at certainty, is always confronted by a patient who wants his money's worth. He has to start with an impressive battery of expensive tests employing the latest technology. The difference of approach is well illustrated by the experience of a friend of mine who was taken ill, with what turned out to be infective hepatitis (jaundice), while on a professional visit to the United States. Before being allowed to see a doctor, he was subjected to the routine battery of tests which, incidentally, did not include either a test of liver function or any examination of the urine. Simply looking at the colour of the urine, without any chemical test, would at least have suggested ithe diagnosis even if it did not clinch it. Only after all this was he seen by the doctor, who was about to say that, as all the tests were normal, the diagnosis by process of elimination was influenza, when for the first time he looked at the patient and saw that his eyes were yellow. In this country the doctor would have looked first and no tests would have been necessary except as base lines for measuring subsequent progress.
Something of the American attitude does sometimes obtain in British hospitals, especially in the out-patient departments, where there are no arrangements for patients to be seen frequently and briefly. Doctors there often feel themselves under an obligation to exclude every possibility (to aim for certainty). There is a definite tendency for over-investigation in the form of the ordering of expensive and sometimes (for the patient) unpleasant tests which time may well show to be unnecessary. 'The most valuable diagnostic instrument', a wise doctor remarked, 'is the passage of time.'
But so far as treatment is concerned, medical practice is only now becoming rational. Dr John Todd has rightly written that `the supreme medical error throughout the ages has been to devise treatment from theory and deduce that it must be effective. Until very recently virtually no one compared patients who were given some remedy with those who were not'. And even recent long-term theorising has been suspect. I am thinking particularly about diet, where I believe medical advice has been wrong. A particular example was the recommendation in 1950 of the British Medical Association Committee on Nutrition. I believe the doctors concerned were blinded by their own science. Dietitians have been mesmerised by the analysis of food. So long as sufficient proteins, vitamins, etc. are contained in the food eaten, it has been assumed not to matter how these elements are combined. This was the theory by which they were guided, and they seem not to have looked for contrary evidence, nor even realised the magnitude of the assumption they were making.
They realised that the war-time `national' flour contained nutrients which would be lost if white flour were once more to be the standard; but their opinion was (probably rightly) that these lost elements could be made up because they occurred in sufficient abundance in other foods. They did not look for evidence that the change might nevertheless be for the worse although this was readily available to them in the form of the vital statistics. These showed that there had been a halt during the war years to the previously rapid increase of such diseases as coronary heart disease and diabetes. They did not ask themselves how this could have happened during a period when living conditions as a whole had deteriorated. Their mistake, if as I believe it was a mistake, has been a very costly one in terms of the nation's health.
It has also caused an enormous waste of patients' and doctors' time and unnecessary anxiety. Children and young people frequently suffer from recurrent attacks of abdominal pain. It is usually due to constipation - not to complete blockage but to hard stools which are difficult to propel through the gut. Parents tend to jump to the conclusion that this is appendicitis; and it is right to bear this possibility in mind. Sometimes the differential diagnosis is difficult and the child has to be admitted to hospital for observation. But these attacks are usually relieved over a few weeks by a change to a diet that contains more indigestible fibrous residue.
No importance has been allowed in conventional dietary theory to whether the body has to extract the essential elements from the diet or whether the extraction has taken place before in mill or refinery. Richard Wilkinson has demonstrated that the steadily worsening mortality of men in social classes IV and V relative to classes I and II is associated with differences in diet and not much else. There is no evidence at all that shortage of vitamins or trace elements are causing the deaths of the poor, and they are certainly getting enough calories (often too many) and proteins in their sugar and chips diet (a diet for a nation with ill-fitting teeth, Aneurin Bevan called it). But they are consuming their essentials of nutrition in such a way as to rot their teeth, ulcerate their guts, and clog up their arteries and veins. It is not surprising that excessive concentratipn matters. The only element that we consume from the air we breath is oxygen, the other components are simply breathed in and out again. Yet everybody knows that breathing pure oxygen is quite rapidly fatal. Unfortunately, eating nearly pure carbohydrates is only slowly fatal - after from thirty to sixty years usually, and it has taken us proportionately long to learn the lesson.
Surgeon Captain Cleave has been a pioneer in refuting the conventional medical theories in these matters. He first showed that constipation was due to an over-refined diet and then refuted the orthodox theory that varicose veins, piles, and peptic ulcers were of hereditary origin (or emotional in the latter case) and put forward his own bold conjecture that the first two were the consequences of chronic constipation, but all three due to the same unnatural diet. Gradually these views are becoming acceptable to a conservative profession. Meanwhile Cleave (together with Burkitt) has moved on to an even bolder conjecture that almost the whole range of 'diseases of civilisation' including coronary artery disease, the cause of so many deaths in middle-aged men, is a result of the excessive consumption of refined carbohydrates, i.e. sugar, white flour and polished rice.
This theory is of course impossible to prove. It is also difficult to refute. The same can be said for some of the orthodox theories. In such circumstances, in the absence of hard evidence, it is rational to act on whatever seems the best theory. Doctors in general have been inhibited from accepting the Cleave-Burkitt theory because of this persisting belief that theories can be confirmed and that one should not act on them until they have been.
For my part I regard the present generally accepted theory that a high consumption of saturated fats is the principle dietary cause of coronary artery disease as rendered unlikely, if not disproved, by the fact that a hundred years ago, when as far as we know coronary artery disease was uncommon, the average consumption of saturated fat was as high as it is now, while poly-unsaturated fats played a negligible part in the diet of this country until after coronary heart disease began to be common. (This is not to deny that a poly-unsaturated fat diet may be a good treatment for those who already have the disease.) In the absence of definite refutations therefore I opt for and act on Cleave's theory and believe in wholemeal bread, plenty of root vegetables, and a minimum of sugar, indeed a minimum of extracted foods, e.g. whole orange rather than orange juice.
Finally I turn to a medical scientist whose work, while it has had little effect on the medical profession, has to a considerable extent misled the public. It is well known that the statistical researches of Sir Arthur Bradford-Hill and Sir Richard Doll have made a strong case for the theory that something to do with cigarette smoking is the cause of most cases of lung cancer or carcinoma of the bronchus. It is important from the point of view of Popperian theory to note that when they embarked on their research they expected to find confirmation of their hunch that diesel fumes were the primary cause of the recent alarming increase in this disease. But the figures they amassed were not compatible with that possibility. That theory was conclusively refuted. They found that although there were some anomalies, on the whole the incidence of the disease was such that the possibility of getting it was roughly proportional to the number of cigarettes smoked. This does not, of course, prove a causal relationship. It remains a hypothesis. But P. R. Burch, Professor of Medical Physics at the University of Leeds, has obtained a considerable amount of publicity for his alternative theory that lung cancer is hereditary, that the same people have both a hereditary tendency to smoke cigarettes and to suffer from the disease. His theory is supported by rather complicated mathematical analysis in which the incidence of onset of hereditary disease of the kind he postulates is matched with the actual incidence of lung cancer. I do not pretend fully to understand his method. But just as I would criticise an architect whose buildings did not keep out the rain, although I myself could. not design a satisfactory house, so I criticise Professor Burch because his theory is refuted by the facts, even though I do not fully understand how he supports his theory. It is the swans again.
The statistics show that the incidence of carcinoma of the lung in 34,000 doctors who were followed from 1953 to 1965 was considerably lower than the incidence in the general population at the beginning of the study in 1953 and that in the twelve years of the survey it had fallen further, while the incidence in the same age range of the general population had risen. In the same period the consumption of cigarettes by the doctors had roughly halved while the change in consumption over the country as a whole was insignificantly small (Doll, 1972). These facts are simply not compatible with the cause being entirely hereditary unless doctors are considered to belong to an alien race.
Professor Burch has sought to get over part of this difficulty by the hypothesis that 'those who gave up were a self-selected group who were less addicted to smoking: it might be that few of them were genetically pre-disposed to lung cancer'. But Professor Doll had anticipated this suggestion some years previously. He had said (1967): 'If those who stop do so because they lack a genetic factor which causes both a strong desire to smoke and a predisposition to the disease, the fact that they stop will serve only to concentrate the incidence in those who continue, and will do nothing to alter the incidence in the whole group of people who were smoking originally.' This completely refutes Burch's theory and in so doing must be considered to amount to a severe test survived by the smoking causation theory.
I do not at all mean to discredit the genuine scientific work carried out by any one of the people mentioned in this chapter. We all succumb to the temptation to preach; and my only point in selecting these examples is to demonstrate that the most illustrious can err and so to emphasise that the most humble have the right, indeed the duty, to point out mistakes and muddled thinking, whoever may be the perpetrators. Most of the lapses from science made by scientists are cases of white-swanning, they result from the failure to realise the truth of Popper's remark quoted at the beginning of this chapter: 'Of nearly every theory it may be said that it agrees with many facts; this is one of the reasons why a theory can be said to be corroborated only if we are unable to find refuting facts, rather than if we are able to find supporting facts.' In other words it is not enough that there is a lot to say in favour of a particular scheme or theory; what matters is that there shall not be a lot to be said against it.