Map, Metamethodology, Aims, Inductive Presuppositions, Quarantining Sceptical Doubt, Methods


(29.5.1997) Version 3

 If you would like to download a pdf version of this essay, click here.


1. Our problem: "Is there a justifiable procedure for assessing the extent of rationality of investigations into the truth?
Our demarcation assesses (i) methods, for reasonableness, (ii) propositions, for truth-likeness and relative progress.
We quarantine sceptical doubts, and therefore accept inductive methods. Justification becomes justification/IP, conditional on an inductive presupposition.
2. Methodology cannot be tested by evidence from history, since no historical events are labelled 'rational'. But for demarcation, we cannot even rely on the current consensus of, say, Physicists. We must therefore engage in prescriptive methodology.
3. Assessment of an activity is (i) a judgement, not mechanical (ii) a matter of degree. Preliminary assessments of physics and psychoanalysis are provided as illustrations.
4. Conclusion: Investigations can be assessed for degree of rationality in a way which is intellectually defensible.

I am grateful to Stephen Clark and Harvey Siegel for vital criticisms of the even more incoherent v.1 of this paper.


   The view that there are justifiably better methods to use in the search for truth is hardly startling. If there is any originality in this paper, it is in the overall structure, not in the details.
"In the existing state of cultivation of the sciences, there would be a very strong presumption against anyone who should imagine that he had effected a revolution in the theory of the investigation of truth, or added any fundamentally new process to the practice of it" (J.S. Mill Preface p. iii). The reader should not expect to find any surprising new methods below.

Our problem

   Are fundamentalist muslims, psychoanalysis, economics, astrology, homeopathy, physics, and UFOlogy, all equally reasonable? Can we demarcate between them? Are there standards for such assessments? Is there a justifiable way of assessing the methods used, and results obtained, in such truth-seeking activities?

Our answer

Positively: We can assess the extent to which investigators:
(i) display a Critical Attitude
(ii) show general respect for evidence .
(iii) use the best available specific methods for obtaining results
(iv) honestly say how reliable their results are.
   Given evidence in these four categories, the final demarcation is the judgement of a consensus on the extent of rationality being displayed.
Negatively: There is no simpler, decisive, method of assessment. The failure of all exciting proposals for such a methods has brought defenders of demarcation, always in danger of appearing establishment and authoritarian, into disrepute.

   An activity is an abstraction, referring to a loosely linked set of people, methods, institutions, and propositions; it is therefore this set which is actually being assessed.
   Our assessment involves both the processes employed - attitudes and methods - and the results obtained - in particular, whether the claims of verisimilitude are justified.
   There are charismatic speakers who claim that miracle cures happen at their meetings; politicians who claim to know why crime is rising; psychoanalysts who claim to be able to improve the mental state of their clients. They all imply that they are intellectually respectable, that they are doing their best to find the truth, that they have obtained it, and that we should trust them.
   If the public are being conned, then philosophers have a duty to blow the whistle - to be intellectual police.  Should the public either believe Maurice Cerullo's miracle cures, or believe the doctors who deny their existence?  Do doctors investigate phenomena in a special way which entitles them to be taken more seriously than faith healers?
   This is an important practical problem. It appears in Siegel's 1985 paper (p.524) as his third question: "Q3: Is actual science (contemporary or historical) rational, in fact?". (He seems to presuppose, however, that rational is a matter of digital yes/no ascription. Instead we wish to leave open the possibility that we should assess only the extent of rationality)
   Before we answer this question, we will, as Siegel says, need to have answered two previous questions: "Q1: In what does the rationality of science consist? Q2: What is to count as evidence, or good reasons, for some hypothesis or procedure?".  Siegel, if I understand him, hopes to propose and justify some general method like commitment to evidence, to avoid bogging the discussion down in detailed questions concerning its application to particular situations - such as the infamous theory choice. He suspects that putting one's objectivist eggs in one particular basket, for example by, traditionally, trying to answer Q1 directly by providing rules for theory choice, is doomed.
   We agree with this strategy; all we intend to do is to make a few small tactical alterations and additions. First we clarify Q1. To "What shall we take rationality to be?" we give a conventional ends/means description. Then theoretically, as prescriptive methodologists, establishing what the rationality of science should consist of, we (i) choose an aim (ii) assess the value of the aim (iii) provide either one global method, or various specific methods, for assessing results and procedures in various situations (iv) assess the justification of, the reasons for, these methods. We should make clear, by considering the justifications, which of the methods are closest to timeless standards, and which are more contingent.
   Following Siegel, we emphasise that the resulting methodological system is not constrained by having to fit with contemporary or historical science, viewed as evidence. We are not conceptually analysing past science; we are not unravelling scientific method. Although we do not pretend that we devised our system in a vacuum, inventing methods ab initio ; it is not descriptive, but prescriptive . They may be widely used in physics, in which case good for physics, but if they are not, so much the worse for physics. We are providing the model by which to assess any putatively rational search for the truth. As a result, for example, we can assess the extent of rationality of recent physics just as we can assess astrology.
   This attitude to physics may seem to lack humility, but it is the only coherent position. Prescriptive methodology is independent of all activities, and physics - science - must take its chance, argue its case, beside all other activities.

   As a result of our tactical attention to ends/means, we propose a primary method, which an investigator must employ if she hopes to be assessed highly for rationality, because it is implicated in the meaning of 'rationality' in the statement of her own hope - in other words, it is justified by linguistic consistency. It is even more general than commitment to evidence:
M1 : The investigator should always have a critical attitude , doubting, questioning, looking for better methods, aware of the limitations of her best methods.
M2 : Commitment to evidence then appears in our system, as a general method.
   As long as we avoid the pretence that they provide definitive, algorithmic, guidance in situations, we can then provide a number of particular methods M3-25, based on applying the most general methods M1&2 to particular situations - particularly if sceptical doubts are quarantined (another departure from Siegel). Rationally justified guidance in some situations is much more conclusive than in others; the situation of choice between theories is notoriously one of the latter.  Sometimes we invoke evidence on human nature, or high-level guiding theories in a particular investigation.

   Since we are following much of Siegel's strategy, we must record that he objects to the end/means approach, and insists that "the rationality of science goes beyond instrumental efficacy" (p.522). He is right that this approach alarmingly requires (a) providing an aim, which seems hard (b) justifying it as rational. Nonetheless, basing our thinking here on Maxwell (1984) - though in a way he would not entirely approve either - we stick to the end/means approach, proposing the very broad Standard Empiricist aim of {Truth}. Clearly {Truth} is not sufficient to characterise the enterprise of science, but that does not matter, since we are not trying to do this . We merely state that {Truth} is the priority of an hypothetical activity; generalisations, explanations, simplicity, predictions, may also be wanted, but only if they are true. It strictly does not even matter if no such activity has ever existed - we are prescriptive methodologists, not historians. To put the same point in another way, recognisable to historians, we are starting from our own interest in the particular aim of {True X} - an aim that we value ; we should not pretend that we are neutral observers of the historical scene, who just happen to have noticed that there is a large group of people whose primary aim this is, and have therefore decided to investigate it. When we look at history, we will be selecting, from the infinity of events, those that we are interested in.
   The power of this approach is that we generate a system which should be adopted by anyone who avows this aim; our system is even-handed between science, religion, psychoanalysis, and the paranormal.
   So the first of Siegel's worries is solved by fiat , as long we are bold enough. The second, the rationality of the aim, is a problem not for the methodologist, but for the rationality of the investigation (following Maxwell (1984), one of the first people to integrate this uncomfortable issue into mainstream philosophy of science). An activity is only fully rational if it periodically considers its aims. If science is an industry for the production of Truth, then scientists should consider whether this is, in itself, a goal sufficiently humanly valuable to justify the effort and resources; maybe, given the massive problems humans have living on the Earth, {Truth}, as an unfocussed aim, is a luxury we cannot justify; maybe it is overall harmful. Has Francis Bacon's vision of the power of physics to relieve Man's estate been forgotten? Are physicists on the one hand opening Pandora's Box, and then denying responsibility for the contents, and on the other hand irresponsibly entertaining themselves at other peoples' expense?
   Siegel tries to short-circuit the need for an aim by arguing on consistency grounds that if investigators want to be assessed as rational they have to give reasons for their actions, because that is what 'rational' means; evidence is then a classic example of a reason; so rational investigators must be committed to evidence. This argument fails, like all "intrinsic conceptual connection" ((1985) p.532)) arguments, because it is just an unpacking of either everyday or personal usage. Suppose that a religious believer says that his reason for making the claim that God is in the trees is an intuitive feeling of rightness; this is a reason - does it make him rational? Only if it is a good reason. How are reasons established as good? By assessing whether they are likely to achieve their intended aim . The aim is unavoidable.

Assumptions : Given the broad sweep of this essay ("a bit ambitious", as a kind critic said), we now state some of our assumptions. If we then outline our reasons for making them, we are not pretending to prove them - some are large issues. But if philosophers are to make any vertical progress, as opposed to spreading horizontally, they must obtain some consensus results. Even though there will always be specialists working on the foundations, work should also proceed, slightly precariously, on construction much higher up. Physics is our model. If the reader judges our foundations to be too unsteady, he or she will - at best - read on with mild amusement.
A1 : Relativism, in the form that denies that any statement can be called true, or any method reasonable, independent of the perspective of a particular society, is self-referentially incoherent; it is hard to find any actual people who believe it. Nonetheless, we should not provide a hostage to fortune by suggesting that our methods are totally neutral; at best they are human-culture neutral. Perhaps Bloor (see his recent Companion To Epistemology entry) might accept that they encapsulate aspects of natural rationality ... ordinary human reasoning propensities.
A2 : Focussed Methodology (Objectivism), the attempt to establish a justifiable system based on the aim of truth, by picking on a particular situation - usually theory-choice - has failed.
A3 : Complete-Justification Methodology is a degenerating research programme. All attempts to get around the problem of sceptical doubt have failed, leaving systems notoriously unable to justify such basic uses of evidence as generalising to laws, Mill's methods, and the successful prediction of novel facts.
A4 : Bayesianism is a successful way of summarising the widespread basic uses of evidence, which are felt to be reasonable; but it does not justify them. It appeals to the consistency of the user in her application of the concept of probability (see Howson and Urbach (1990). But this concept is defined in such a way that, say, Bayes' theorem is implicit in it. Therefore consistency merely establishes that the user understands the common concept and is prepared to use it carefully. The question of whether the concept can be justifiably applied to the real world remains unanswered.


   We start with a proposal, continuing from our third assumption:

A worthwhile justification need not answer the sceptic.

   We thus quarantine the contagious problem of induction, along with all the associated problems it has infected; after two thousand years, we begin to suspect that it can only be contained, not cured. We identify the basic inductive presuppositions which seem so hard to justify, and then, without waiting for their justification, proceed with the solution of our problem.
   In the temporary absence of decisive justification of generalisations - justification which is immune to sceptical doubts - we accept justification/IP - justification based on evidence but conditional on ignoring sceptical doubts, and accepting an Inductive Presupposition.
   We propose a familiar inductive theory (Te): "The evidence we obtain through our senses is not systematically misleading; it is typical of evidence we could have obtained at other times and in other places and circumstances". "The Universe is governed by uniform laws; although these are often concealed, notably by varying initial conditions, they can be obtained by judicious generalising of limited human experience".
   As a particular case, we propose Tp: "Experienced events had as high a chance of occurring as possible, given the true laws of Nature".
   From Tp we can, for example, deduce that if hypothesis H1 has an observation statement S as a certain or very high chance consequence, while H2 makes S extremely low chance, and then S enters the consensus, then H2 is probably false; theories which can only explain observations by assuming that very low chance coincidences have occurred are not true.
Assumptions like Te and Tp enable us, not surprisingly, to justify/IP a wide range of methods which cannot be justified:
(i) Bacon/Mill's methods
(ii) preferring theories which successfully predict novel facts
(iii) using Bayes' theorem, qualitatively, to assess hypotheses.
   These assumptions do not, however, make us that elusive creature - an inductivist; they do not imply that high-level theories can be induced from low-level facts. This is fine, since hardly anyone has ever believed this - not even Bacon, and certainly not Mill.
   The lack of justification of Te and Tp is disappointing, but we are not wasting our time. Firstly, when - if - a justification of these theories is found, the following results will still hold, exactly as before, but justified back to First Philosophy, to foundations immune from doubt. If the reader is aware of such a justificationx, then he or she is already in this desirable situation.
Secondly, for practical purposes (applied philosophy) our apparent lack of absolute foundation does not matter. Human beings demonstrate their implicit assumption that Te and Tp are true with every action they make. Therefore, anyone inclined to criticise our unfavourable assessment of the reasonableness of their methods of investigation, by emphasising the unacceptable presence of these unjustified theories in our methods, is being inconsistent, since their every action displays their implicit decision to regard these same theories as acceptable. In our experience, critics accept this argument, since they recognise that the alternative, that all investigations and their results, and indeed all planned human actions, are equally unjustified, is unpalatable. The argument is ad hominem , but in the area of applied philosophy this does not matter.

{The set of methods, and their extent of justification, could have been entered at this point, but they are of more general relevance. Therefore, they are instead in a separate chapter, to which you can refer if you wish}



(1) We have chosen an aim, one which is highly valued by many people.
(2) We have established criteria for judging the best method of achieving this aim.
(3) We have used these criteria to establish the best general method, not restricted to particular situations, of checking whether particular propositions have achieved the chosen aim. We see if there are any other general methods implicit in the aim and the intention to be rational.
(4) We have shown how this best method, in particular situations, leads justifiably to the use of certain specific methods - techniques. The justification may require the presumption of non-scepticism (induction), applied both locally, and to the whole history of an investigation (Newton-Smith's meta-induction).
(5) We have assessed, using the best methods of which we are aware, the extent of verisimilitude, the level of absence/IP of doubt, that can be reasonably claimed for three broad types of propositions: particular facts, generalisations (laws), and theories. The consequent requirement that investigators should label propositions with roughly the epistemologically correct extent of truth-credit is a further method.
(If in these stages we have given examples of investigators in certain situations, these examples were pedagogical illustrations to clarify the situation and the method; they were neither claims as to what actually happened historically, nor attempts to provide evidence for the methods)
(6) We assess the extent to which specific groups of human investigators, such as scientists, who have avowed the chosen aim as a priority, have adopted the methods we have obtained. The epistemological rationality of their investigation is thus assessed. (The assessment is not primarily a personal criticism of the investigators. For example, we might assess the system of investigation of the doctors in Ur lower than that of Aristotle and his followers, and them lower than that of modern physicists. But the individuals may well have been doing the best they reasonably could have been, at that time)

Meta-Epistemology ; Testing the system: How is our epistemological system of methods checked? How is it tested? We can compare it with the methods used in investigations typically labelled 'rational'. If we find that some people have not used our methods, either we criticise them, or they criticise us. If we find that we are assessing respected investigators as irrational, we will need to discuss with them, in case we are missing something.
   Neither of us has ultimate authority - because there is no equivalent to the facts of immediate experience of nature. All there is, is a dialogue - a joint review. This is not surprising, since reflective investigators are methodologists, and therefore in direct competition with the active investigators.

How are assessments of activities to be attempted?

   We have to accept the uncomfortable fact that the epistemological methods are not algorithmic. We never had any reason to think that the methods of human enquiry would be simple, and indeed they are not. But how, then, can we hope to assess investigators for reasonableness, if the methods they are supposed to be adopting are non-mechanical, vague, and judgemental?
   Lakatos used the analogy with a trial, to indicate one possibility. Points of law do not uniquely determine the judgement, but the judge must - to be professional - remind the jury of them. Similarly investigators must - to be rational - take the guideline methods into account in making these judgements, even if, for example in the situation of theory choice, the methods do not determine the decision.
   For example, if the consensus of physicists had denied the truth significance of Fresnel's successful novel prediction of the bright dot in the middle of the shadow of a small solid object, then they would have been defying a firm guideline method, and to that extent irrational. If, however, having accepted this and other arguments, they still judged that, overall, the undulatory model of light had not yet convincingly demonstrated its superiority, in terms of probable truth, over the corpuscular theory, then we would be very hard-pressed to show that this judgement was irrational. We would be hard-pressed precisely because a judgement was needed in the first place - the guidance provided by the guideline methods was inconclusive.
   If - like us - you do not like judgement playing such a vital role in rational investigations, and hence in demarcation, if you want more authoritative methods, then the onus is on you to think of them. The best human efforts for millennia have failed to unearth them.

Investigations cannot be assessed, in a binary way, as scientific or not scientific

   No investigations are scientific; no person is an ideally rational creature. Hence investigations can only be assessed, in an analog way, on the extent to which they are rational, the extent to which they use justifiable/IP methods, presuming they are seeking explanatory truth.
Investigations can be demarcated; we suggest that it should be done, modestly but realistically, on a rough scale from, say, 1 to 5 [8]. Investigators can thus only be assessed in analog terms, not digitally.

Two outline assessments of groups of investigators

(i) We have chosen groups for average assessment. Wide individual variation is therefore ignored. We could instead have assessed individuals.
(ii) Full assessment of the extent of rationality of a group would require a more extensive inspection.
(iii) Ideally a group should make assessments of the extent of its own rationality - self-regulation - as an integral part of its investigation.

(i) Physicists

   Physicists have one advantage: the area of phenomena which they have chosen to investigate is one about which the majority of people do not intensely care. This has made it much easier for them to apply methods such as M1.
For : Physicists' record of use of the methods is very good.
Their attitude seems to me to have been broadly critical and open, with inevitable lapses (M 1).
Against : They have several times become over-confident concerning the truth of higher-level theories - notably Newton's - and been shown to be embarrassingly mistaken.
Various of their institutions work imperfectly: Funding councils, referees, and professors, tend to encourage conformity (M 19,25).
Discussions of whether subsidiary aims like realism are achievable are too rare (M 17,18) [22].
   Many investigators have displayed normal but unhelpful psychological and sociological behaviour: they have unjustifiably persisted with research programmes eg. Barkla; they have ignored inconvenient experimental evidence; they have descended to personal vituperation (Brougham on Young); they have turned against innovators (Boltzmann); they have followed dominant leaders (Newtonians); they have displayed group prejudice (Young and Fresnel, Thomson and Hertz).
Neutral : Their theories have displayed notable progress in predictive power, and - arguably - in truth-likeness, but this is irrelevant to our assessment.
Rating : Physicists, on average: 4/5.

(ii) Psychoanalysts

For : (i) There are serious professional investigators in this area.
(ii) Evidence, from everyday experience, suggests that their chosen subject-matter, which is of immense importance to humanity, may be unamenable to generalising and to theorising.
(iii) Ethical problems make testing difficult.
(iv) Financial constraints restrict opportunities for investigation.
(v) Psychoanalysts seem to operate in a reasonably open society.
Against : (i) They should not be relying on single analysts to make delicate observations; one person may suddenly perceive their client's behaviour as falling into a simple pattern, while another might perceive nothing. Individuals may deceive themselves. (M3)
(ii) Some investigators wrongly claim that the lack of a consensus does not matter; they claim that their theories are true nonetheless. (M23)
(iii) Some investigators have claimed that their unusual subject matter justifies them in abandoning some of the guideline methods. For example, they have claimed that statistical, public, outcome studies are too blunt an instrument to detect the effect they have on their patients; but these studies are essential for identifying the presence of some truth in their hypotheses (M2,8,10,11). They have claimed that the private experience of feeling that suddenly the patient's problem is understood - everything falls into place is sufficient evidence of the truth of the understanding. This claim needs to be taken seriously, but with extreme caution, given the possibility of self-delusion and wishful-thinking (M3). The fact that some psychoanalysts are aware of this possibility is in their favour.
Neutral : No reliable generalisations and theories have entered the consensus. Progress/IP is scanty.

Rating : Psychoanalysts, on average: maybe 3/5.

Is assessment decisive?

   No. Firstly, we cannot justify some of our methods decisively. A critic could judge that MX, say, is not justified by the evidence available. We cannot decisively prove him or her wrong.
   Secondly, we cannot provide any set of weightings for the relative importance of the methods. Our assessment can, at best, be a list of strengths and weakness, followed by a judged extent of reasonableness rating out of 5.
   Thirdly, some of the methods themselves involve non-mechanical judgement - for example, assessment of the merits of competing hypotheses. We cannot decisively prove that an investigator's judgement was unsound.
   Assessing the reasonableness of investigations is therefore not as simple as we might have liked. We must do it by presenting arguments, listening to responses, making our judgement, and seeking a consensus - as physicists do when they are assessing the truthfulness of a theory. We could not expect it to be simpler.
   The reader may feel that these judgemental assessments - analog demarcations - are dangerously open to debate.  But why is this a danger?  Feyerabend [21] warned of an unhealthy tendency for objectivists to want a set of rules, standards, which can be mechanically used to dismiss investigators in areas they do not like - preferably with the minimum of necessary empathy and detailed investigation. Instead we should invite physicists, Marxists, religious believers, psychoanalysts, and astrologers, to debate with us, if they feel our rating is wrong.

Other assessments are left to the reader

   The reader may like to try out this form of assessment on other investigators, individually or in groups: astrologers, for example, or religious believers.
   Low-rated groups are irrational only if they insist that their primary aim is Truth. They could practise astrology primarily to make a living; they could hold a religious belief primarily to achieve peace of mind. Truth is just one human aim amongst many. Therefore these rating are only emotively labelled with approval or disapproval - as 'high' and 'low' - if Truth is specially valued.
   A correspondent suggested that some investigators - philosophers and literary critics were the examples cited - are seeking Truth, but are doing it in a quasi-empirical way, which has different methods. We should only accept this if these investigators can explain what their methods are, and how they are justified.

Have we solved our problem?

   We have ended with a view close to common-sense - mainly by quarantining sceptical doubt, by conditionalising on the Inductive Presupposition. It was the tantalising scent of scepticism which drew the philosophical flies into their bottle; by sealing it off at the bottom, we let the flies out.

   We have shown that groups of investigators who claim to seek Truth can be demarcated from each other; they can be assessed on the extent to which they use the set of methods which we justifiably/IP claim are the best available for discovering Truth. The more they use the methods, the more rational they are. Thus we have snatched demarcation from the jaws of irrationalism.
   To save rationalism, we stepped back from philosophical extremism - in particular, by quarantining sceptical doubts. Some may agree with this attitude. Others may be aware of a justification for the inductive theories; they can just slot it in. Only those who, unaware of a justification, feel that no method that relies on a whiff of inductivism is of any interest epistemologically, will demur. We suggest that these last readers, by their insistence on impossibly decisive justifications for methodological standards, ironically damage the respectability of the claim that such standards exist.
   When Koestler endowed a chair at the University of Edinburgh for the scientific investigation of paranormal phenomena, he was intending that the investigation should be carried out in as rational a way as possible - in the way justifiably believed at the time to be the best for seeking the truth about these strange phenomena. If this way involves a complex mixture of methods, generously interwoven with unprovable judgements, then so be it. What it should not involve is methods - such as devising theories based on uncorroborated reports of exciting, unrepeatable, experiences - which a critical attitude and available evidence indicate are very poor.
   Boring temperate rationalists can justifiably/IP demarcate investigators and their products in all areas. No longer does a writer criticising, for example, fundamentalist religious believers, have to admit that the methods he is using, despite extensive analysis by philosophers, do not seem to be justified [23]; nothing, beyond unverbalisable immediate experience, is justified - but this kind of justification is of little interest. The methods listed in this essay are justified/IP, which seems to be the best that humanity can achieve. Unless fundamentalist believers can explain how authority, dogma, personal revelation, conviction, or faith, are justifiable/IP methods, they are being irrational in their search for truth. Our methods are the best presently available. Not to use them is to ignore the results of millennia of careful thought and painful experience, on the best ways of seeking the truth.


[1] NEWTON-SMITH, W. (1981) The Rationality Of Science (Boston, Routledge & Kegan Paul), pp. 8-9.
[2] Ibid., Chapter VII.
[3] NEWTON-SMITH, op.cit., p. 9.
[4] MAXWELL, N. (1984) From Knowledge To Wisdom (Oxford, Blackwell).
[5] LAKATOS, I. (1981) History Of Science And Its Rational Reconstructions , in: I. HACKING (Ed.) Scientific Revolutions (Oxford, OUP), p. 122, note 38.
[6] LAKATOS, I. (1970) The Methodology Of Scientific Research Programmes , in: I. LAKATOS and A. MUSGRAVE (Eds.) Criticism And The Growth Of Knowledge (Cambridge, CUP), p. 174.
[7] NEWTON-SMITH, op.cit., p. 270.
[8] Ibid., p.17.
[9] LAKATOS (1981), op.cit., pp. 122-3.
[10] MAXWELL, op.cit..
[11] ZIMAN, J. (1968) Public Knowledge (Cambridge, CUP).
[12] LAKATOS (1981), pp. 122-3.
[13] POPPER, K.R. (1959) The Logic Of Scientific Discovery (London, Routledge & Kegan Paul)
[14] DUHEM, P. (1974) The Aim And Structure Of Physical Theory (New York, Athenaeum), p. 218.
[15] POINCARE (1905) The Foundations Of Science (Lancaster, Science Press), pp. 131-6 and pp. 170-3.
[16] NEWTON-SMITH, op.cit., pp. 224-5.
[17] GELLNER, E. (1992) Postmodernism, Reason, and Religion (London, Routledge & Kegan Paul), pp. 75-8.
[18] BURTT, E.A. (1967) In Search Of Philosophic Understanding (New York, New American Library), pp. 186-193.
[19] NEWTON-SMITH, op.cit., pp. 245 and 272.
[20] POPPER, K.R. (1970) Normal Science And Its Dangers , in: I. LAKATOS and A. MUSGRAVE (Eds.) Criticism And The Growth Of Knowledge (Cambridge, CUP), p. 56.
[21] FEYERABEND, P. (1991) Three Dialogues On Knowledge (Oxford, Blackwell)
[22] MAXWELL, op.cit..
[23] GELLNER, op.cit., pp. 59 and 78.
Mill, J.S. (1898) System Of Logic (London, Longmans)
ZIMAN, J. (197?) Public Knowledge
TENNANT N. Changing The Theory Of Theory Change British Journal For The Philosophy Of Science 45 (1994) pp. 865-897
HOWSON AND URBACH Scientific Reasoning
SIEGEL (1985) The Rationality Of Science in Philosophy of Science

Version History :

Version 2.0 :

1. Changed title from Snatching Demarcation From The Jaws Of Irrationalism , and Six Dogmas Of Rationalism (I quite liked that one...).
2. Completely rewrote text. Removed the long introductory potted history of the problem of demarcation (inductivism, hypothetico-deductivism, falsifiability, research programmes, relativism and so on). Introduced the section in which the methods are presented quite fully, so that the complexity of a demarcation judgement is made clear.
3. Thanks to Harvey Siegel, realised that demarcation was partly about the methods , the behaviour, of investigators, but also partly about the results of their investigations. Made appropriate changes and additions.
4. Also thanks to Harvey, realised that the really basic methods, M1 for example, can be justified, and are essential to the full picture.
5. Realised, negatively thanks to Harvey, that analytic philosophy, of the kind practised by Harvey, cannot establish any useful results in this area by analysing the concepts involved.
6. Removed unnecessarily controversial assessment: The rationality of Religious Investigators. Despite the fact that it was more favourable to serious thinkers about Religion than might have been expected (though not to fundamentalists), it diverted attention from the main theme (thanks to my sister, Helena).
7. Detached a long series of Criticisms and alternative solutions into appendices:
(i) Exciting rationalists; criteria which will demarcate decisively between methodologies. Need to respond to attack by retreating to a more defensible position.
(ii) Exciting Relativists; no criteria for demarcation. Self-
referential inconsistency; affront to common-sense.
(x) Is there a neutral language or basis on which the assessment can be made?
(iii) Does Scepticism really need to be quarantined?
(ix) Should the inhumanly reasonable Epistemologist do without the inductive theories, and devise the most justified scheme of investigation on that basis?
(iv) Is Science just seeking {Truth}?
(v) Doesn't rational behaviour involve questioning the aim of the activity?
(vi) Do we assume that Science is the epitome of rationality?
(vii) Are the methods M1-25 provably the best core methods for achieving {True X}?
(viii) Are our methods timeless?

To do:

Maybe extend the references... but there again, maybe not.
Collect more comments. Yes, that means YOU!