Appended below (in plain text) is a paper I gave at a conference 7 June on mobile telephones and health. First, similarities and differences in the health issues for mobile phones and power lines are summarised. Comments and criticism would be welcome.
S1 The government advisory body for both is the NRPB (National Radiological Protection Board).
S2 Both exposures are to fields below ionising frequency, which means the fields cannot damage DNA molecules by the direct hit quantum effect, and therefore cannot cause cancer by the direct hit effect. It does not mean they cannot damage DNA or cause cancer by indirect effects through intermediate chemicals nor even through accumulated quantum effects (as recent research has speculated).
S3 The NRPB has historically been overly influenced by the non-ionising nature of both exposures and therefore sceptical about any potential to cause cancer.
S4 The NRPB has also been overly influenced by an old vision of physics limited to direct energy effects on tissue, called "thermal effects", and hence dismissive of what it calls "subtle effects" or non-thermal effects.
S5 Researchers have suggested non-thermal effects in the nature of information (or signal) effects which might disturb the brain, cell communication, cell growth and death, immune systems and so on. A well known example is how synchronised flashing lights might induce epilepsy - an information effect; the energy needed for such effects is related to the (low) energy required to store or transmit information in the biological system, not to the thermal energy to heat tissue to a damaging level.
S6 Official exposure limits relate to thermal effects. The limits vary between countries, but a typical set is produced by ICNIRP (International Commission on Non-Ionising Radiation Protection). These limits are much higher than exposure levels where non-thermal effects are suggested.
S7 Harmful non-thermal effects are not accepted by most national authorities as confirmed, particularly at practical public exposure levels, and the authorities and industry tend to talk them down. Generally there is no policy at sub-ICNIRP levels to respond to potential but unconfirmed non-thermal effects.
S8 At sub-ICNIRP exposures, in respect of non-thermal effects, NRPB has called for further research but has set itself firmly against precaution.
S9 Both ELF/EMF and mobile phones have been the subject of major UK reviews recently: the Stewart Report on Mobile Phones and Health 2000 and the (second) Doll Report on ELF/EMF and Cancer 2001. Both reports are published by NRPB.
D1 The two exposures are in different frequency ranges, powerlines in ELF and mobile phones in RF/MF, although mobile phones also use modulation and pulsing which may introduce ELF frequencies.
D2 Exposure to powerline ELF fields is in the "near field" zone, which is not the same as exposure to RF/MF "radiation" which transmits energy in a more directional way, though ELF/EMF can still deliver energy.
D3 The accumulated and pooled evidence from epidemiological studies on ELF/EMF across the world has confirmed a statistical association of childhood leukaemia with exposures above 0.4 microTesla, whereas with mobile phones the epidemiology is patchy and it is too early to get this sort of confirmation. A statistical association does not prove a cause.
D4 The Stewart Report came from an ad hoc International Expert Group, not from a regular NRPB group, and it was critical of NRPB's lack of openness and of its limited vision. The Doll Report was from a long- standing NRPB group.
D5 The Stewart Report recommended precaution, particularly in the use of mobile phones by children and in keeping base station exposures as low as possible consistent with the system working. The Doll Report takes a narrow view and sets itself against precautionary policy.
D6 The Stewart Report sets out to consider all potential health effects upon all sets of people of different susceptibility. The Doll Report considers only cancer and does not consider specially susceptible people.
Here is my paper to last week's conference:
Risk Communication: Mobile Telephones and Health
Professor Emeritus M J O'Carroll University of Sunderland
City & Financial Conference on Mobile Telephones and Health - The Latest Developments Kingsway Hall, London, 6-7 June 2001
1. Risk communication
Risk communication does not equal telling the public. It is not a bolt- on end process to explain decisions. It could involve including environmental or consumer representatives on government specialist advisory panels, for the value of their input. (Note 1.69 of the Stewart Report.)
Risks may be considered in three stages: identification, assessment and control. Communication affects them all, through consultation, involvement in the assessment and decision process, and then in report, decision and monitoring. Too often, official statements and reports have appeared to be what politicians call "presentation", an attempt to reassure or to put things in the best light.
Some lessons have been learned from the BSE affair , but there seems still to be a pre-occupation with a perceived public inability to understand numerical risk rather than with public concerns. Sir Kenneth Calman declared British beef "perfectly safe to eat" at Christmas 1995, shortly before Stephen Dorrell's fateful declaration to Parliament that BSE was after all the most likely cause of vCJD. Calman later suggested definitions and descriptors for a risk scale, while addressing the task of presenting risk probabilities to the public . The mooted "Richter scales" of risk only addressed the numerical probability, not scientific uncertainty.
2. The risk perception dilemma
This preoccupation is born of the dilemma: why do the public readily accept high risks of smoking or driving, while balking at much smaller and less certain risks like CJD or GM foods?
Last week an article  set out "fright factors" to explain the dilemma. They started with risks being involuntary, but omitted the more obvious "commercial and political exploitation and dishonesty". In my experience people accept inevitable and natural involuntary risks, such as being hit by a meteorite, but react against the dishonesty of ministerial statements about CJD and the character assassination of scientific objectors such as Professor Lacy. Likewise Monsanto's achievement of mixing GM and natural soya to avoid identification and customer choice creates suspicion and negative reaction. The article, perhaps unwittingly, reflects the preoccupation with presentation of numerical risk to a non-understanding public. The HSE  claims, in similar vein, "in everyday life, though we are likely to pay more attention to consequences than to likelihoods". I wonder. Perhaps the risk industry needs a culture change with more respect for the public.
There are special explanations for smoking or driving, and for mobile phones, involving benefits and psychological needs, and advertising and peer pressures. It may not be a real dilemma, but a (mis)perception after all, though by the risk profession not by the public.
3. Scientific uncertainty
"... mathematical modellers soon learn is there is no absolute truth, not in the real world."  Even deterministic models are necessarily approximate. Models of how hazards might work may be probabilistic, but their approximate nature is more fundamental than that. The very ideas of cause, proof and even truth are artificial constructs when applied to the real world, rather than to an abstract system. The idea of an "established risk" is necessarily subjective and approximate.
Too often we hear industry and ministers claim there is "no evidence" or "no established evidence" of harm, and famously "no conceivable risk", when there may be a large body of evidence and/or concerns expressed by eminent scientists, but the official advisory bodies have not accepted proof of a cause . A better grip is needed on evidence short of proof of cause.
My main plea is to assess scientific uncertainty, with a systematic grading of credibility of causation, as a stage of risk assessment and for more consistent communication.
Consider a tree structure for risk: at the first level there are three stages: 1. identification; 2. assessment; 3. control. The first requires an exhaustive search and imagination of potential hazards - the sort of thing which clearly was inadequate in this year's Selby rail disaster, where risks from cars plunging on to the track seemed to have been neglected. But I am concerned with the second.
At the next level, split assessment into three more stages: 2.1 credibility; 2.2 likelihood; and 2.3 consequences. The HSE also introduced in  what might be a fourth stage: 2.4 tolerability. I use the HSE language of likelihood (i.e. statistical probability of incidence, sometimes described simply as "risk") and consequences (sometimes called impact). Likelihood and consequences are commonly considered, e.g. in business-risk and audit-needs analysis as well as in health and safety. It is credibility (of causation) which needs more systematic assessment and communication.
It is helpful to acknowledge the intrinsic uncertainty in cause, both in its very concept and in any particular causal model, and to reject the idea that either there is a cause or there is not. It is a matter of degree and clarity, not of yes or no. In addition to intrinsic uncertainty, there are complexities of multi-factorial and probabilistic causal relationships, and of variable human susceptibility (which may affect credibility as well as consequences).
The HSE address uncertainty in . They distinguish uncertainty from ignorance (lack of awareness of factors influencing the issue). Uncertainty is then considered as knowledge uncertainty due to sparse statistics or experimental error, modelling uncertainty due to limited validity of theory, and unpredictability due to sensitivity to initial conditions. But all of their consideration of uncertainty is then applied to the domain of likelihood and consequences only, and not to credibility.
4. Assessing credibility of cause of harm to health
The focus of the public risk perception dilemma is harm to health. The Bradford Hill criteria attempt to structure the assessment of cause, as to whether there is a cause or not. It really should be an assessment of the strength of the case - remote, possible, likely, probable. The NRPB has reflected this approach, effectively dismissing precaution while admitting the "possibility" of cause of cancer (from electromagnetic fields). Worse, it has effectively blocked local authorities from exercising any local discretion to take a precautionary approach.
The Bradford Hill criteria are neither necessary nor sufficient to establish cause, but are aspects to be considered subjectively. They group broadly into two kinds, making a third level to risk consideration: 2.1.1 biological mechanism and 2.1.2 statistical association. In  I called these aetiological and epidemiological factors, and gave rules for assessing them on 5-point scales. The grade descriptors ranged simply from Very Weak to Very Strong, with accompanying specifications. The epidemiological grading was itself determined from similar 5-point gradings or numerical measures for confounding, strength of association (Relative Risk) and significance (p value).
In this way we can construct a framework for assessment of credibility, which can then be combined with the numerical assessment of likelihood and (to an extent, with financial measures) consequences. It would be helpful, both for consistency in decision making and for communication, to have a common framework for a wide range of environmental health concerns.
5. Test question # 1 - assessment and advice
A key question of environmental hazards is the broader one: Is this hazard potentially harmful, directly or indirectly, to any aspect of the health of any minority of people exposed to it in any way? It is not enough to examine effects in the population as a whole, and not enough to advise only on established effects.
The Stewart Report  does well by this test, in contrast to the parallel NRPB Doll Report  on electromagnetic fields and cancer. An investigation needs to explore the range of types of exposure (phones, masts, radiation, distraction), the range of potential health effects (not just cancer), and human sensitivity.
The NRPB preoccupation with energy effects (either thermal or quantum) and relative dismissal of information effects (or "signal" effects - e.g. flashing light and epilepsy - which might influence brain function, cell communication or immune systems) suggests an old fashioned view of physics. Information effects are referred to as "subtle effects", which has stuck even in the Stewart Report, although it does address the physical scope for non-thermal, non-quantum bio-effects. Physical arguments to dismiss particular mechanisms for such bio-effects cannot dismiss the possibility of other mechanisms, which may involve resonance or molecular triggers; that is, it is not proved that no mechanism is possible. There is suggestive evidence (e.g. calcium efflux) that "subtle effects" are possible. Even weak epidemiological evidence should therefore be taken seriously.
It is important to acknowledge, in official communication, levels of exposure which are linked to concern about uncertain hazards, as well as levels (such as the much higher levels of the ICNIRP guidelines) linked to established effects. These lower levels (within the range of mobile phone users' exposure) may be a matter for personal choice rather than exposure limits. Official advice should be sufficient to make people aware of the concerns and the choice available. Warning signs of a generic nature, along with declaration of the specific device radiation levels, would be a reasonable requirement for all mobile phones; this goes further than the Stewart recommendations for a leaflet to be available at the point of sale.
6. Test question #2 - precaution and policy responses
A test question for developers or authorities with uncertain hazards is: will you adopt a policy that, other things being equal, exposure will be avoided?
Hypothetically, this can be a choice of tower site either close to houses or further from them, although direct radiation for many towers may not qualify as an uncertain hazard. On a personal level, you might choose to moderate your time on the mobile phone. Other things are almost never equal, but the question tests the principles of developers and officials.
The Stewart Report deals with precautionary policy at some length, so I will not. Precaution is one of a range of policy responses, from informing the public and calling for more research to interventionist policies, such as exposure limits on new developments and "retrofit" (removal of existing exposures). In  I attempted to link assessment grades with policy options, as part of a general framework for risk decision and communication.
Judging by the Stewart Report, I would assign the potential health hazard from mobile phones to the category of Weak (but not Very Weak), both for aetiology and epidemiology, with a policy response of information, research and precaution. Pretty much as Stewart did. That is in contrast to the hazard of residential exposure to powerlines, which I would classify similarly, but which NRPB and the Doll Report treat very differently - reporting with much spin suppressing public concerns and blocking precaution even as a discretionary policy for local authorities.
Risk communication runs both ways at all stages. It is not just explaining decisions. There is a gap in the systematic assessment of risk, specifically in assessing the credibility of an uncertain cause. It can be made systematic, giving a framework for consistent assessment and communication. The Stewart Report on mobile phones contrasts favourably with the NRPB Doll Report on electro-magnetic fields, addresses information effects as well as thermal or quantum energy effects, and properly recommends precaution in the face of scientific uncertainty where there are rational grounds for concern.
 Lord Phillips, BSE Inquiry Report, 2000.
 K. Calman, On the State of the Public Health for the Year 1995, HMSO, Sep 1996.
 G Mythen & C Wales, NfCG, Fright Factors and Public Anxiety, NFCG, Consumer News, 204, April/May 2001.
 J Le Guen, Reducing Risks, Protecting People Discussion Document, HSE, 1999.
 M J O'Carroll, Degrees of Proof and Policy Options in Epidemiology, BICS Conference, Electromagnetic Transmissions II, London, 20/21 April 1995.
 M J O'Carroll, Parliamentary Brief, p. 34-36, Jan 1996, and p. 35-36, May 1996.
 W Stewart, IEGMP Report, Mobile Phones and Health, NRPB, 2000.
 R Doll, ELF Electromagnetic Fields and the Risk of Cancer, NRPB, 2001.
Three Tun Web