There can be few people concerned with radiological protection who have not heard of Professor Alice Stewart. Although there is a broad awareness of her unconventional and controversial views on the risks to health of low-level exposure to ionising radiation, I suspect that most do not have an especially deep understanding of how she came to arrive at this position. It is a long and complex story. Gayle Greene's biography attempts to tell this story, primarily on the basis of interviews with Stewart and, to a lesser extent, her colleagues and friends.

Dr Gayle Greene is Professor of Women's Studies and Literature at Scripps College, the Women's College of the Claremont Colleges, California. Her research interests are Shakespeare, women writers, feminist criticism and, more recently, women's health and environmental issues. She has written a number of books on these subjects. This background is certainly apparent in her biography of Stewart. Unfortunately, Greene's lack of knowledge of the area of science in which Stewart has worked is also apparent. Greene relies heavily on interviews with Stewart and a significant proportion of the book consists of direct quotes from Stewart. This provides one perspective, and it is of interest to `hear' Stewart tell her story; but I was left with a feeling of dissatisfaction because many intriguing issues were left unexplored, and what is presented lacks balance. I had hoped to find much more in the way of critical examination of how and why Stewart and her colleagues came to be at such odds with most of the scientific community over much of their work, and, from what I know of it, it is a fascinating tale. Disappointingly, there is little depth of discussion of these issues, which, I strongly suspect, is deliberately avoided by an author who has insufficient understanding of the subject to feel comfortable in tackling them at anything other than a superficial level. As I hope to illustrate briefly in this review, an opportunity has been missed.

Alice Mary Stewart was born in October 1906 in Sheffield to medically qualified parents who were involved in children's welfare in the grim conditions of the city at that time. She was fortunate to attend secondary school in St Andrews, and went from there to study medicine at Girton College, Cambridge, and the Royal Free Hospital, London. Her long association with Oxford began in 1941, as a clinician at the Radcliffe Infirmary. She soon became involved in `social medicine' (as non-infectious disease epidemiology was then called) which was to dominate her professional life from her appointment in 1945 as assistant to Professor John Ryle at the Institute of Social Medicine. When Ryle died in 1950, Stewart became head of the renamed and downgraded Social Medicine Unit, as a reader rather than as a professor, a time Stewart still recalls with bitterness. The first part of Greene's book chronicles the early influences on Stewart's life, and paints a vivid picture of the difficulties facing professional women in this era, particularly in medicine. It is an alluring portrait, but, in this review, I shall concentrate upon `the secrets of radiation'.

It was in the early 1950s that Stewart embarked upon the nationwide epidemiological study of childhood cancer that was to become the Oxford Survey of Childhood Cancers (OSCC), by far the most impressive and substantive work in which she has been involved. The OSCC was to become the largest case-control study of childhood cancer in the world, eventually embracing over 20,000 deaths. Initially conceived to examine the notable rise in leukaemia in Britain, the study set out to identify all deaths from cancer in childhood (about half of these being leukaemias) and to compare characteristics of these children with those of healthy control children. The 80% coverage of such deaths in the early years of the study, much of which was through her own personal efforts, demands considerable admiration of Stewart's energy and organisational abilities, especially given the conditions of the time. In 1956, Stewart and her colleagues reported the initial results from the Oxford Survey in The Lancet. They had found that mothers of affected children were twice as likely as mothers of controls to have had an abdominal diagnostic x-ray examination during pregnancy, and they suggested that these low-dose examinations were a cause of childhood cancer. This finding was received with scepticism, but Greene does not adequately explore the reasons why. Clearly, this result had uncomfortable implications for medical practice and for nuclear weapons production and testing, which must have influenced the reaction to the study; but the finding was also novel and unexpected, and, consequently, it is not surprising that the association was questioned, although the impression that is gained from Greene's book is that these doubts were unwarranted and unfair. Bearing in mind that, at this time, the results were based upon maternal interview responses only, the possibility of recall bias (mothers of dead children would have better, or different, recall of events than mothers of healthy children) had to be seriously considered. It was only later, in 1966, that the reassuring results of checks of mothers' responses against medical records were reported; but it was the study in the USA by Brian MacMahon, which was based entirely upon medical records of exposure and which confirmed Stewart's finding, that finally settled the question of recall bias. None of this is explained by Greene. (Ironically, the association with antenatal x-raying for which the OSCC is best known could not account for the rise in childhood leukaemia that was the major impetus for the study: throughout Britain, only around 10% of children experienced the intrauterine x-ray exposure that the initial results of the Survey suggested doubled the risk of leukaemia, and this collective excess risk would be unlikely to be perceptible in national rates.)

An early blow to the association found by Stewart came from a study by Michael Court Brown, Richard Doll and Austin Bradford Hill published in the British Medical Journal in 1960. This was a cohort study of leukaemia among children exposed antenatally to x-rays in hospitals in London and Edinburgh, and the findings did not support the association whereas, if the association was genuine, they should have done. Although the raised relative risk identified by the OSCC had largely been reproduced by other case-control studies, the failure of a cohort study, particularly one carried out by such eminent authors, to confirm the association had a significant impact upon the attitude of the scientific community towards the Oxford Survey. Doll has since expressed concern over the accuracy of the tracing of children carried out in this cohort study, and has concluded that its results are unreliable. He now believes that the association reported by Stewart is real and represents a cause-and-effect relationship. However, the influence that this cohort study had in casting doubt upon Stewart's results has been neither forgotten nor, apparently, forgiven by Stewart. (Doll's appointment as Regius Professor of Medicine at Oxford in 1969, when Stewart was still a reader, presumably would not have improved their relationship.) Doll is undoubtedly one of the bêtes noires of Greene's book, and it is curious that he seems to have been the only person interviewed by Greene who cannot be considered to be one of Stewart's natural allies. It is evident that Stewart still harbours considerable resentment towards Doll whom she perceives as playing a prominent role in the failure of her work to gain the recognition it deserves.

A further challenge to the OSCC came in 1970, and this most certainly influenced Stewart's attitude to the studies of the Japanese survivors of the atomic bombings of Hiroshima and Nagasaki in 1945. In the late 1960s, Stewart had been joined in her work by the statistician George Kneale. In 1970, The Lancet published their paper in which, in combination with a particular set of estimates of foetal doses, they derived from the OSCC data an excess risk of childhood cancer per unit dose of radiation received by the foetus. This was quickly followed by a rejoinder in The Lancet from Seymour Jablon and Hiroo Kato of the Atomic Bomb Casualty Commission (ABCC) who concluded that the risk coefficient derived by Stewart and Kneale was way out of line with the bomb survivor data for those exposed antenatally. In retrospect, both sets of authors can be seen to have paid insufficient attention to the uncertainties in their respective risk estimates, leading to an overstatement of the apparent contradiction between the findings. More recent papers from the Radiation Effects Research Foundation (RERF, the successor of ABCC) have been much more cautious in their conclusions on this subject, but Greene does not refer to this. However, this 1970 confrontation helped shape Stewart's critical attitude towards the value of the bomb survivor data for radiation protection purposes.

The supposed conflict between the findings of the OSCC and those of the ABCC for antenatal irradiation, which I consider to have been unduly exaggerated by both camps, led Stewart to examine the risk estimates derived from the Japanese data for those exposed after birth. She proposed that the survivors alive in 1950, when the principal ABCC studies commenced, would not be representative of those present in the two cities during the bombings because only the healthiest would have survived the dreadful conditions that existed after the explosions. This seems eminently reasonable and is, effectively, generally accepted; but the divergence in opinion between Stewart and other scientists arises over the influence that this selection effect has upon risk estimates. Stewart maintains that two `silent forces' (initial selection of healthy survivors and radiation damage of the immune system), the strengths of which are positively correlated with radiation dose, are at work in the survivor population. She contends that these act on the risk of late health effects in opposite directions, leading to a suppression of the mutational effects of radiation and giving the false impression of normality for diseases other than cancer. The presence of these `silent forces' does not allow, in Stewart's view, the application of risk estimates derived from the Japanese bomb survivors to situations where low doses are received under normal conditions of exposure, and accounts for the apparent discrepancy between the OSCC and ABCC findings. It is, of course, important that the Japanese data (often referred to as the `gold standard' for risk estimation) are examined critically, and, in this regard, Stewart's input is to be welcomed. However, Greene does not consider in any detail either the crucial question of the size of the increase in cancer risk estimates that would be predicted by Stewart's hypothesis, and whether this could account for the supposed inconsistency with the OSCC, or the work that RERF and others (notably Mark Little and Monty Charles) have done to test her ideas. In essence, although other researchers do find some evidence of selection in the late health effects experienced by the survivors, particularly in the early years of follow-up, the overall impact upon risk estimates is small and does not generate the dramatic underestimates implied by Stewart. It is unfortunate, but not altogether surprising, that Greene provides only part of this story.

The other major campaign embarked upon by Stewart in support of her risk estimates obtained from the OSCC was the study of nuclear industry workforces, particularly the workforce of the Hanford facility in the USA. Thomas Mancuso had been investigating mortality among the Hanford workers for some time when, in 1974, he approached Stewart and Kneale for assistance. Mancuso, Stewart and Kneale presented the results of their collaborative study in the November 1977 issue of Health Physics, results which, they suggested, showed that the risk of cancer produced by low-level occupational exposure to radiation had been substantially underestimated. To quote Stewart from Greene's book, `there was very harsh criticism of our 1977 paper'. Indeed there was. The authors were severely censured in a number of papers written by experienced epidemiologists and statisticians for using inappropriate methodology and making unsupported inferences. What does Greene make of this fusillade of criticism? Apparently, it was all part of an Establishment-inspired conspiracy to hide the truth about the health effects of low-level exposure to radiation. But why should so many senior scientists collude in this way, and why have almost all of the many subsequent studies of Hanford and other nuclear workforces carried out by other researchers produced results that do not agree with Stewart's? These are questions that are left largely unexplored by Greene, which is a great pity. I should point out that Stewart and Kneale have attempted to address the criticisms of the 1977 paper in later publications, although most scientists remain unpersuaded by these efforts, and the Hanford study has prompted many carefully conducted studies of nuclear industry workforces around the world. However, from what is presented in Greene's book, Stewart still cannot acknowledge the shortcomings of the 1977 paper, and this does her no credit. Incidentally, many of the fiercest critics of Stewart's work in this area have been women. They (Ethel Gilbert, Sarah Darby, Valerie Beral and Shirley Fry are singled out for mention) are relegated to the convenient category of `honorary men', and so, presumably, their work can be safely ignored, which helpfully permits Greene to avoid any serious examination of their criticisms.

Not surprisingly, Stewart has been invited to give evidence to public inquiries by activists opposing the expansion of nuclear installations in Britain. Her evidence has not been accepted by the various inquiry inspectors. Greene paints a picture of an Establishment patently ignoring the voice of reason, but little detail emerges from her book. For example, the first inquiry at which Stewart appeared was the Windscale Inquiry of 1977. The inspector, Mr Justice Parker, devoted two pages of his report to Stewart's evidence in which, inevitably, the Mancuso, Stewart and Kneale paper featured large. Parker stated that Stewart `failed to deal with a number of what appeared to me to be valid criticisms [of the paper]', giving three examples, and he found her interpretation of the Hanford data `unconvincing'. In this he was in agreement with the bulk of the scientific community. Greene's cursory treatment of Parker's criticisms is another illustration of her failure to present a balanced appraisal of an issue. More remarkable is the account of how Stewart has been, in Greene's view, inexplicably shunned by those bringing legal claims of radiation injury against the nuclear industry in Britain, in a chapter that she entitles `The Invisibilizing of Alice'. For example, Greene correctly states that Stewart was not invited to give evidence on behalf of the plaintiffs in the claim that radiation exposure due to operations at Sellafield had caused childhood leukaemia, which was heard in the High Court during 1992-93. The key issue in this case was the allegation that occupational irradiation of the father before the conception of his child was the cause of leukaemia in the child. Greene finds incomprehensible the absence of Stewart from the plaintiffs' witnesses, but if she had bothered to check Stewart's publications she would have found a paper by Kneale and Stewart, based on an analysis of OSCC data and published in the British Journal of Cancer in 1980, entitled `Pre-conception x-rays and childhood cancers'. The last sentence of the abstract of this paper concludes, `there is no support for the idea that exposure of parental gonads to diagnostic x-rays is conducive to cancer in the next generation'. This paper would hardly endear Stewart to the plaintiffs' lawyers given the nature of the claim!

Since I consider the Oxford Survey to be incontestably the major contribution to science made by Stewart, I was disappointed not to find more in Greene's book on the way that the OSCC has developed over the years. Moreover, I was dismayed by the absence of any meaningful examination of the rationale behind some of the decisions that have been instrumental in determining the risk estimates which have been produced by the study, and which have led to the, frequently ferocious, battles with other scientists. After all, Stewart has been at the helm of this study since its inception and if anyone can shed light on these matters then it is she. How, for example, did Stewart and Kneale obtain the particular foetal dose estimates used by them in their above-mentioned 1970 paper, and why have other, equally plausible, estimates not been used by them in subsequent analyses? There are other unfortunate omissions from Greene's account of the Oxford Survey. As an illustration, John Bithell is only referred to in passing, and yet I, and many others, find the 1975 paper by Bithell and Stewart to be the clearest and most comprehensive description of the OSCC data and their analysis, and Bithell has made a number of other important contributions to the interpretation of the OSCC results. Robin Mole is not mentioned at all, despite the substantial papers he wrote on the OSCC, especially on the foetal doses likely to have been received by children included in the Oxford Survey. This is a shame. However significant may be Stewart's part in the evolution of the OSCC, the major input of other scientists, who might not have been quite so closely associated with her as those who are mentioned in the book, should be recognised.

I do believe that Greene is right to say that the results of the OSCC have not been accorded the importance they deserve by the scientific community, but I do not consider that Greene pays sufficient attention to Stewart's own role in this failure to achieve proper recognition. It is on page 228, in one of the few points in the book where Greene actually gets below the surface of her subject, that the crux of the problem of the attitude of scientists towards the OSCC is to be found. An anonymous friend of Stewart is quoted as saying that the papers of Stewart and Kneale are `convoluted and difficult to read to the point where other specialists in the field have simply not been willing to invest the time it would take to follow them'. Coupled to this is the uncomfortable impression that this complication is unnecessary, or even intentional. Their tendentious style certainly does not commend itself to the reader. Further, on occasions one cannot help but feel that risk estimates are being inflated through, for example, the choice of foetal dose estimates or the nature of the statistical analysis employed. An instance of the latter is when Colin Muirhead pointed out that the risk coefficient presented in one of their papers, published in this Journal in 1987, was the product of a statistical analysis that was likely to be wrong, and when he and Kneale later jointly corrected this error (also in this Journal) the risk coefficient was found to have been exaggerated by over 50%. This does not inspire confidence and, as a consequence, considerable effort is required to convince oneself that reliable and realistic risk estimates can be extracted from the OSCC data.

Let me be clear about my view of Stewart and her work. The Oxford Survey of Childhood Cancers represents a magnificent accomplishment by Stewart. It is a significant and substantive study for which Stewart should receive due credit and I believe that the OSCC, and the similar smaller studies that it inspired, provide the only direct epidemiological evidence that acutely delivered doses of around 10 mSv produce a detectable excess risk of cancer. Clearly, this is of immense importance to radiological protection and the OSCC is likely to be pre-eminent in providing such evidence for some considerable time. This is Stewart's lasting achievement. It is unfortunate that attention has come to be concentrated on the risk coefficients that Stewart has derived from the OSCC which I am sure are unduly optimistic in their accuracy and tend to overestimation, and which, as a result, appear to place the OSCC data in conflict with the estimates derived from the Japanese atomic bomb survivors. This seeming discrepancy has led her to embark on a vigorous (and sometimes petulant) defence of her risk estimates by attacking the Japanese studies and conducting dubious analyses of nuclear workforces. I consider that the proposed inconsistency of the OSCC data with established risk estimates is spurious and that, consequently, Stewart's war has been waged on a false premise. In my view, Stewart's single-minded determination in promoting her risk estimates in this manner has led not only to her alienation from mainstream science, but also, ironically, to the unfair and undesirable marginalisation of the results of the OSCC that she was striving so hard to promote. I may be wrong in this conclusion, but if only Greene could have made at least an attempt to explore this intriguing and potentially fruitful avenue.

So, what is to be made of Greene's book? It will be read, no doubt, by activists to reassure themselves that they are right about the risks of radiation exposure. It provides, however, little for the scientist looking to know more about `the secrets of radiation', apart from giving Stewart's account of her battles with the established scientific order. Technically, it is lightweight with little evaluation of the science involved, and quite startling, and patently erroneous, claims (such as Stewart's statement that atmospheric nuclear weapons testing has `doubled background radiation worldwide') are blithely accepted without comment. This superficiality is coupled with an unhealthy and unquestioning attitude towards Stewart and her work which, at times, verges on the sycophantic. Greene is not a scientist and clearly does not have any experience in radiation risk assessment. Under these circumstances, I would have thought that an academic would have carefully researched the subject and certainly have cast her net wider than just Stewart and her friends and allies when seeking out people to interview. The approach adopted by Greene has inevitably produced an unbalanced appraisal of Stewart's findings. I believe that the work of Alice Stewart requires careful scrutiny and evaluation to sort the wheat from the chaff and to provide a proper perspective of her contribution to science. Greene's book falls a long way short of achieving this.