Salviati: If this question we are arguing about was some point of law or of one of the other disciplines in the humanities, where there is neither truth nor falsehood, then one could justifiably rely on intellectual subtlety, on verbal fluency, and on breadth and depth of reading, and hope that whoever had the advantage in these respects would succeed in making his argument seem, and be accepted as, the stronger; but in the natural sciences [scienze naturali], the conclusions of which are true and necessary, and in which the opinions of human beings are irrelevant, you have to be careful not to give your support to error, because a thousand Demosthenes and a thousand Aristotles will find themselves defeated by a mediocre intellect who has the good fortune to attach himself to the truth. Therefore, Signore Simplicio, give up on that idea and that hope that you have, that there can be men so much better educated, so much more sophisticated, and with so much more book-learning than the rest of us that they can, in defiance of nature, turn falsehood into truth.
Galileo, Dialogue Concerning the Two Chief World Systems (1632)1
Shakespeare had – to return to Borges’s comment with which this book began – no sense of history. He read classical authors as if they were his contemporaries. He had plenty of experience of change, sometimes for better and sometimes for worse, but he had no notion of irreversible change, and he had no notion of progress. And this is hardly surprising, for there was, in his world, little evidence of progress; when Shakespeare retired from the stage in 1613 Bacon had published only one book on the new science, The Advancement of Learning (1605), and it was only three years since Galileo had published his telescopic discoveries. But since then progress has been uninterrupted. I see no reason to revise John Stuart Mill’s opinion that one of the main drivers of economic development has been ‘the perpetual, and so far as human foresight can extend, the unlimited, growth of man’s power over nature’, and that (as we saw in Chapter 14) this power results from increasing scientific knowledge.2
All sorts of taboos have developed around the use of the word ‘progress’; indeed, it has become a word that can no longer be used in the humanities without its use being penalized by what Pierre Bayle called ‘the law of opinion’, a heavy sanction indeed in the academic world since it means the denial of tenure and of promotion.ii So let me stress that my views on this question coincide with those of many of the sharpest critics of the idea of progress. Here is the philosopher John Gray in a book subtitled Against Progress and Other Illusions: ‘In science progress is a fact, in ethics and politics it is a superstition. The accelerating advance of scientific knowledge fuels technical innovation, producing an incessant stream of new inventions; it lies behind the enormous increase in human numbers over the past few hundred years. Post-modern thinkers may question scientific progress, but it is undoubtedly real.’3
This view used to be entirely conventional. As George Sarton, the founder of the [American] History of Science Society and of its journal, Isis, put it in 1936: ‘[T]he history of science is the only history which can illustrate the progress of mankind. In fact, progress has no definite and unquestionable meaning in other fields than the field of science.’4 Such statements have led to Sarton becoming someone who is quoted only to show how naive we once were. Alexandre Koyré’s reputation has survived better than Sarton’s, but he said exactly the same thing the year before: the history of science, he claimed, is the ‘only history (along with that, linked to it, of technology) which gives any sense to the idea, so often glorified and so often decried, of progress.’5
Sarton and Koyré were right. A history of modern science without progress fails to capture science’s unique feature. Moreover, the best of the supposed ‘relativists’ know this. Kuhn denied that science made progress towards the truth,iiii or could ever claim to have grasped the truth, but he always insisted that a place had to be found for the idea of progress in science, even if he had great difficulty explaining how this might be the case.6 The last chapter of Structure was entitled ‘Progress through Revolutions’. In it he writes: ‘[A] sort of progress will inevitably characterize the scientific enterprise so long as such an enterprise survives’; and he goes on to argue that progress must be understood in evolutionary terms.7 Richard Rorty, the boldest defender of pragmatism, insisted that there are no epistemological foundations on which we can build an incontrovertible knowledge, but he was also an admirer of Kuhn, and like Kuhn acknowledged that science makes progress within its own terms: ‘To say that we think we are heading in the right direction is just to say, with Kuhn, that we can, by hindsight, tell the story of the past as a story of progress.’8 A form of knowledge which aims at prediction and control gets better at prediction and control. Progress is part of the story. This book is not aimed at the mitigated relativism of a Kuhn or a Rorty but at the strong relativism which presents progress in science as an illusion, the consequence of a misunderstanding of what is in fact taking place when scientists disagree with each other. The public – and the scientists themselves – imagine that the quality of the evidence determines the outcome; actually, we are told, it is the status, power and rhetorical skill of the combatants.
This emphasis on the contingent and local character of scientific knowledge is supported by what many take to be a philosophical argument of profound importance, the so-called Duhem-Quine thesis, named after Pierre Duhem (1861–1916), a physicist and historian of science, and W. V. O. Quine (1908–2000), an American philosopher.9 The thesis is misnamed, because, as it is usually formulated, Duhem did not hold it and Quine abandoned it, but it is the fundamental conceptual underpinning of much modern history and philosophy of science.iiiiii
The thesis takes two forms. First, it is argued that a scientific theory cannot be refuted by experiment: not just that it cannot be refuted by a single experiment, no matter how often repeated, but that it cannot be refuted by a whole series of different experiments. Scientific theories are complex things, for they are constituted by interlocking bundles of theories, facts, and equipment. If an experiment produces a result which is at odds with the theory then something is wrong; but you cannot simply say that the theory is wrong. Some other theory that this theory depends on may be at fault, some fact that has been taken for granted may be mistaken, or some piece of equipment may not operate as envisaged. Consequently the results of experiments cannot refute a theory. This is called ‘holism’.
But let us take the example of sailing to America. This was in effect an experiment, and it was a crucial experiment: it straightforwardly refuted the two spheres theory. The only way of rescuing the theory in the light of the new evidence would have been to say that all the navigators were wrong: America was not where they thought it was. Nobody thought it worth pursuing this line of argument. This would not have puzzled Duhem, who specifically formulated his thesis to address modern physics; he acknowledged that it did not apply, for example, to nineteenth-century biology.
Second, it is argued that theories have a very loose relationship to the facts. Given any set of facts, there are countless theories that can account for them, just as there are countless lines that can be drawn through any set of dots to join them together. This means that scientists, although they may not be aware of it, are never obliged to adopt any particular theory: there are always alternatives that would work as well, indeed, for all we know, better.iviv And of course facts and theories have an intimate relationship: what counts as a fact depends on the theories you hold, and whether a theory appears to be valid depends on the facts you acknowledge. This loose, slippery, and incestuous relationship between facts and theories is called the principle of underdetermination.
Again, the example of sailing to America presents problems for the principle of underdetermination: we have seen that although Bodin proposed an alternative to the terraqueous globe theory, it was never viable; not a single person came to its support. The terraqueous globe theory was not underdetermined; in this case the relationship between the theory and the facts was a tight one, not a loose one. The same is true of the phases of Venus: once their existence was acknowledged, the conclusion that Venus orbited the Sun was inescapable.
These two principles – holism and underdetermination – are what is being invoked when appeal is made to the Duhem-Quine thesis. The standard argument is that this thesis proves that evidence does not determine what scientists hold to be true; consequently, it is claimed, scientific beliefs are primarily shaped by cultural and social factors. If science is largely culturally determined, then the now familiar conclusion follows: its procedures and conclusions will reflect a purely local consensus.vv This conviction provides yet another reason for insisting that there was no such thing as ‘the Scientific Revolution of the seventeenth century’. Science, we are to understand, was one thing in Florence, something quite different in Paris or London. Historians of the book seek to block the obvious objection – that scientists in London were reading books written by scientists in Florence and Paris, and so belonged to a single intellectual community – by maintaining that books mean different things to different readers, so that reading Galileo in Florence in the 1640s was a quite different enterprise from reading him in London in the 1660s.10
This contrast between local meanings and cosmopolitan messages is a perfectly sound one. One just has to get the balance right. Galileo may never have left Italy, but he had English and Scottish students, and his Two New Sciences was first published in Leiden; William Harvey, the discoverer of the circulation of the blood, studied medicine in Padua; René Descartes moved from France to Holland, Christiaan Huygens from Holland to France, Thomas Hobbes from England to France; Robert Boyle may have spent his working life in Oxford and London, but he went to Italy and learnt Italian; his associate, Denis Papin, worked in France, England, Italy and Germany. And of course almost all the early scientists shared a common language. Galileo published his Dialogue Concerning the Two Chief World Systems in Italian in 1632, but it appeared in Latin in 1635; Boyle published his New Experiments Physico-Mechanical Touching the Spring of the Air in English in 1660, but it appeared in Latin in 1661; Newton published his Opticks in English in 1704, but it appeared in Latin in 1706. Of the first 550 fellows of the Royal Society, elected between 1660 and 1700, seventy-two were foreigners (and the proportion of foreigners rose in the eighteenth century, until it reached one third).11 The new science knew no boundaries of language or nationality, at least not within western Europe, within the world of the printing press, gunpowder weaponry, the telescope and the pendulum clock.
A moderate interpretation of the Duhem-Quine thesis leads to a mixed constructivism, in which evidence and culture each have a part to play in the construction of scientific beliefs.12 Kuhn’s Copernican Revolution (1959) provides an example of this. According to Kuhn, Copernicanism won out over the alternative systems (the Ptolemaic and Tychonic) before the invention of the telescope, but this cannot be explained simply in terms of the mathematical elegance of the Copernican system; other cultural factors, such as neo-Platonism, which might have encouraged people to revere the Sun, were also important.vivi Another example would be Newton’s willingness to accept the idea of action over a distance. To Cartesians, Newton’s theory of gravity made no sense; but in England, where Cartesianism had never been adopted without reservations, and where arguments from design were widely accepted, resistance to the theory was much weaker. But once a science establishes itself it becomes, to a remarkable degree, autonomous, immune to influence from other fields.13 This is not to say that it is not shaped by culture as well as evidence; but the culture which shapes it is, first and foremost, the culture of science itself. Thus Kepler, because he was familiar with Gilbert’s work on magnets, could use magnetic force as a model, and on the basis of it identify his laws of planetary motion: Gilbert made it possible for Kepler to conceive of an astronomy grounded in physics rather than mere geometry. Thus Newton, in England, could propose his theory of gravity, but only because he already had (unlike the Cartesians) the notion of a theory: something more than an hypothesis but distinct from a proof.
An uncompromising interpretation of the Duhem-Quine thesis leads to the conclusion that science is entirely a social construction, or at least should be studied as if it is, and that reality (the source of the Danube, the existence of America, the phases of Venus) is of no concern to historians and sociologists. If this is correct, then then there is no way of distinguishing good science from bad science because all theories are, or are to be regarded as being, equally adequate (this has been called ‘cognitive egalitarianism’), and therefore it is meaningless to discuss progress in science.viivii I call this relativism.viiiviii For some considerable period this uncompromising interpretation has been the dominant position in the history of science. It is precisely because the Duhem-Quine thesis, so interpreted, cannot handle either the demise of the two spheres theory or of Ptolemaic geocentrism that these crucial historical events have been invisible to those who are convinced of the truth of the thesis. Its proponents behave just like the philosopher Cesare Cremonini when he refused to look through Galileo’s telescope: they hold to their convictions even when the evidence shows they are wrong by the simple device of ignoring anything that does not fit their theory.
Exactly the same relativist approach is employed with regard to facts as to theories; indeed, the two are often indistinguishable. According to Ian Hacking there is no necessary agreement on basic measurements, such as the speed of light.14 He claims to show this by pointing out that the first person to measure the speed of light came up with a very different figure from the one that we now rely on, and so feels entitled to dismiss the argument that agreement on the speed of light was inevitable as ‘dreadful’. In fact the ‘dreadful argument’ is (quite uncharacteristically) Hacking’s, and in order to see this we need only look at the evidence.
Hacking, following established convention, presents the astronomer Ole Rømer (1644–1710) as the first to measure the speed of light. In fact, Rømer never calculated a figure for the speed of light.15 His goal was to establish accurate figures for the periods of the satellites of Jupiter (the eclipses of the satellites were to be used to set a standard time against which to measure the longitude of different places on the earth’s surface). Rømer concluded, on the basis of a very small set of observations, that when the earth is at its maximum distance from Jupiter the moment of an eclipse appears to be delayed by twenty-two minutes compared to when it is at its minimum distance. Thus it takes light twenty-two minutes to cross the diameter of the earth’s orbit, or eleven minutes to cross the radius (i.e. to travel from the sun to the Earth). Claims that Rømer measured the speed of light depend on then introducing a figure for this distance, which Rømer never did (and there is no reason to think he would have regarded any figure he could have used as reliable).ixix Direct measurements of the speed of light come much later, in the nineteenth century. The two tables opposite summarize the history of these two sorts of measurement.16
Table of time taken by light from the sun to Earth
| Date | Author | Method | Time |
| 1676 | Rømer | Moons of Jupiter | 11 mins |
| 1687 | Newton1 | Moons of Jupiter | 10 mins |
| 1693 | Cassini2 | Moons of Jupiter | 7 mins 5 secs |
| 1704 | Newton | Moons of Jupiter | 7 or 8 mins |
| 1726 | Bradley | Stellar aberration | 8 mins 12.5 secs |
| 1809 | Delambre | Moons of Jupiter | 8 mins 13.2 secs |
| Modern figure | 8 mins 19 secs3 |
1 Newton’s figures are not based on independent measurements; they are the best available, in his judgement, at the time.
2 Cassini never accepted that light was not projected instantaneously but, like Rømer, he came up with a figure for the correction to be employed in calculating the timing of eclipses of the moons of Jupiter, and this figure was adopted by others as a figure for the speed of light.
3 This figure is an average, as the Earth’s orbit is an ellipse.
Table of speed of light
| Date | Investigator | Method | Result (km/s) |
| 1849 | Fizeau | Rotating toothed wheel | 313,000 |
| 1850 | Foucault | Rotating mirror | 298,000 |
| 1875 | Cornu | Rotating mirror | 299,990 |
| 1880 | Michelson | Rotating mirror | 299,910 |
| 1883 | Newcomb | Rotating mirror | 299,860 |
| 1907 | Rosa and Doresey | Electromagnetic constants | 299,790 |
| 1926 | Michelson | Rotating mirror | 299,796 |
| 1928 | Mittelstaedt | Kerr cell shutter | 299,778 |
| 1932 | Pease and Pearson | Rotating mirror | 299,774 |
| 1940 | Huttel | Kerr cell shutter | 299,768 |
| 1941 | Anderson | Kerr cell shutter | 299,776 |
| 1950 | Essen and Gordon-Smith | Cavity resonator | 299,792 |
| 1951 | Bergstrand | Kerr cell shutter | 299,793 |
| 1958 | Froome | Radio inferometry | 299,792 |
| Modern figure | 299,7921 |
1 This is now true by definition as, since 1983, the length of a meter has been set in relation to the speed of light rather than vice versa.
There are two conclusions to be reached from these tables. The first is that a pretty reliable figure for the time light takes to travel from the sun to the Earth was available seventeen years after Rømer’s first estimate of eleven minutes. And the second is that measurements of the speed of light improved steadily until 1928; then for about twenty years they settled around a figure which now seems slightly inaccurate; and then progress resumed in 1950, since when measurements have remained more or less identical.xx
What makes Hacking’s argument ‘dreadful’ is that he takes one figure on its own, and that figure is the first in a long succession of attempts to measure the speed of light. Of course Rømer’s figure for the time light takes to travel from the sun to the earth was a very crude approximation! But science is not an enterprise conducted by isolated individuals; it is, as argued in Chapter 8, a collective undertaking in which progress is driven by competition (and cooperation).17 Indeed John Flamsteed, the Astronomer Royal, noted that there was ‘emulation’, and even bad blood, between Cassini, the established authority, and Rømer, the upstart.18 Over time, competition ensures progress. And of course competition is imperfect, and for a time scientists may set out on the wrong path; but over time good results will drive out bad ones.xixi The claim that an alien society, if it had a sufficiently advanced technology, would come up with almost exactly the same figure as us for the speed of light is perfectly sound. The question of the speed of light is not just an arbitrary one invented to entertain a theoretical physicist: it would impose itself on anyone doing astronomy designed to predict the future apparent location of planetary bodies to a high degree of accuracy, no matter whether their purpose was astrology, chronometry (as in Rømer’s case), or space navigation.xiixii
A relativist would respond to this argument by saying that there is no reason to think that scientists were getting better at measuring the speed of light; they were simply getting better at agreeing on how to measure the speed of light. This is obviously false, because one test of measurements of the speed of light is whether, when used in conjunction with Kepler’s laws of planetary motion, they enable you to predict the apparent position of planets in the sky. Rømer’s figure fails that test, and modern figures pass it. Still, a classic example of this line of argument is Simon Schaffer’s essay ‘Glass Works: Newton’s Prisms and the Uses of Experiment’ (1989). Schaffer’s claim is that Newton had not, as is traditionally claimed, demonstrated by experiment that white light is made of rays of differently coloured light which are differently refrangible, for his experiments could not be successfully replicated except under arbitrary and unreasonable conditions (the use of prisms made in England, for example). Newton’s supposed discovery only imposed itself on the scientific community because Newton acquired ‘control over the social institutions of experimental authority’. His authority became ‘overwhelming’. We believe Newton’s theory of colour not because of the experimental evidence, but despite it; we believe it because Newton successfully imposed himself on the scientific community, and experiments were then ‘staged’ to produce the required results.19 Nowadays, Newton’s experiments come pre-packaged, so that they can be reliably reproduced to educate children in school; but that is because the equipment has been devised to produce the required result.
One might think that Schaffer or his readers would have rejected these arguments as inherently implausible. One might think they would have wondered about the various and varied technologies that depend on Newton’s theories of refraction and of colours: the reflector telescopes that Newton himself designed, which avoid the problem that different colours of light are differently refracted, causing a penumbra of colours around objects seen through a lens; or the colour televisions which had been in widespread use for twenty years when Schaffer published his article, which make a full range of colours out of red, green and blue. On the contrary, Schaffer’s claims were accepted as confirming a by-then-well-established theory of how science works – that it works not through evidence, but through power and persuasion. His essay was admired because it seemed to demonstrate that the strong theory could be put into practice: one could write the history of what we now think of as good science (Newton’s new theory of light) using exactly the same intellectual moves as the ones one would use for writing about what we now think of as bad science (say, alchemy). Unfortunately, it was Schaffer’s evidence, not Newton’s, that was ‘staged’ to produce the required results. Step by step, detail by detail Schaffer’s argument was refuted by Alan Shapiro in 1996: it turned out that plenty of people had successfully replicated Newton’s experiment without any particular difficulty, and without any need to fiddle the results. But since the year 2000 Schaffer’s paper has been cited seven times for every two times that Shapiro’s has, and the gap is growing, not closing: over the last four years Schaffer has had ten citations for every two to Shapiro. Bad knowledge, at least temporarily, has driven out good.20
Schaffer’s essay is not an isolated case. There is a significant group of intellectuals, working within the same tradition as Schaffer, who claim that experiments can never be replicated in a straightforward way. Whenever experiments are repeated genuinely independently, they would argue, divergent results are obtained; to learn to get the ‘right’ results you have to be trained to do the experiment under very peculiar and particular conditions, and at first this involves learning it directly from people who have conducted it successfully in the past. Eventually an experiment can be reliably mass-produced by manufacturing special equipment designed to obtain precisely that result – the equipment and the results are interdependent. This is what is called ‘black boxing’. Once an experiment has been black boxed the experiment is no longer a test of the result; rather, getting the right result has become a test of the reliability of the equipment. So the whole notion of replication, in the view of those who argue in this way, is misleading, and there is nothing transparent about experimental knowledge.21 Building consensus around what an experiment demonstrates is thus primarily a social process of persuading people to act and think in the way you want them to, not an impartial process of discovering an objective aspect of the real world. These claims, of course, run into trouble if one tries to apply them to the experiment that had more influence on the behaviour of scientists than any other, the Torricellian experiment – or, indeed, to Newton’s prism experiments, or to measurements of the speed of light.
Robert Boyle summed up the alternative view of science from that propounded by the devotees of holism and underdetermination and by deniers of the independent replication of experimental results:
Experience has shown us, that divers very plausible and radicated Opinions, such as that of the Unhabitableness of the Torrid Zone, of the Solidity of the Celestial part of the World, of the Blood’s being convey’d from the Heart by the Veins (not the Arteries) to the outward parts of the body, are generally grown out of request, upon the appearing of those new Discoveries with which they are inconsistent, and would have been abandon’d by the Generality of Judicious Persons, though no man had made it his business purposely to write Confutations of them: so true is that Vulgar saying, that Rectum est Index sui & Obliqui. [‘The line which shows itself to be straight shows also what line is crooked.’]22
In other words, as with the terraqueous globe and the phases of Venus, very often a new theory triumphs rapidly and without resistance, because new evidence simply eliminates the viability of all known alternatives.
If the relativist account of science were correct, every major paradigm shift ought to be accompanied by bitter disputes between competing intellectual communities: indeed it was Kuhn’s view that this is exactly what happens. Some are, but others take place silently, as Boyle says, without anyone bothering to write confutations of the old theory. One army abandons the field almost immediately after the first blow has been struck; its opponents declare victory and are rapidly joined by deserters from the other side. What causes this sudden transformation? When Vadianus maintained in 1507 that Aristotle did not know everything, that he was a fallible human being (the question immediately at issue was the source of the Danube, but of course the two-spheres theory was also at stake), the claim seems so obvious to us as to be trivial; it would not have been at all obvious to Vadianus’s contemporaries. Why did Aristotle make mistakes? Because of experientiae penuria, insufficient experience.23 The victory of the terraqueous globe theory following the discovery of America is the first great triumph of experience over philosophical deduction, and thus the beginning of a revolution.xiiixiii
But it would be dangerous to rely on examples such as this to support an oversimplified view of the role of experience. Experience, we might say, comes in three kinds. Sometimes, as we have just seen, it falsifies beliefs and in doing so immediately imposes an alternative; sometimes it confirms beliefs that are already held (the measurement of the shape of the Earth by French expeditions to Peru and Lapland (1735–44) confirmed Newtonianism); and sometimes it represents only one step along a path that leads to an outcome which cannot be foreseen. Of this third kind, there are both answers to scientific questions that might have been right but turn out to be wrong, but nevertheless are a crucial step towards a successful answer; and right answers whose full significance only slowly becomes apparent in the light of further experience. Kuhn held that the fact that the outcome of a revolutionary crisis is unpredictable while it is going on means that it cannot be explained with the benefit of hindsight. On the contrary, there is often only one path through the debate which is capable of producing a stable outcome. Getting it right can be like finding the way out of a maze.
In the late Middle Ages, for example, the Venetians became wealthy importing spices from Asia; the spices were carried overland from the Red Sea to Alexandria, which meant that the Venetian merchants, who bought them in order to ship them across the Mediterranean, had to pay a high price for them. The Portuguese, hoping to undercut the Venetians, sought a route to the spice islands by sailing around Africa, and in the end they succeeded. They were followed by the Dutch, who made the spice trade the foundation of a great commercial empire. Columbus sought a route to the West, but his successors discovered that circumnavigating South America was arduous and time-consuming; as a commercial route to Asia his discovery proved to be a failure, but for this there was more than compensation in the discovery of gold and silver in South America. In the early 1600s the French explorer Samuel Champlain thought he might find a route by water across Canada – up the St Lawrence, through the Great Lakes, and on.24 He carried Chinese court dress with him in his canoe in case he should meet Chinese agents coming from the East. His transcontinental route too was a failure. Until 1794 ships sought a Northwest Passage, but none was to be found.25
Here we have a series of attempts to answer the same question, against a background of changing geographical knowledge: indeed it was the attempt to find a better route to Asia which was the main driver in improving that knowledge. As it turned out, between the end of the fifteenth century and the end of the nineteenth century efforts to find a new route failed over and over again. It was impossible to know in advance that this would be the outcome; but we can be sure, as they could not, that Columbus would not reach China, that Champlain was never going to meet an emissary from the Chinese court, and (until the advent of global warming) that the search for a commercially viable Northwest Passage was doomed to failure. By 1800 all the possible alternatives had been eliminated, and the question of the best route to Asia had finally been settled (at least until the opening of the Suez Canal in 1869).
Such examples of path dependency are the rule and not the exception. Once Copernicus had suggested that the earth was not the centre of the universe but a planet orbiting the sun, people were bound to puzzle over what sort of planet it could be. In the Aristotelian universe the earth had been the recipient of light but had given no light. It was easy to imagine looking down on the earth, but what one would see would be a miniature earth. There was, surprisingly, an extensive discussion among Aristotelian philosophers as to what the earth would look like seen from the heavens, but no one envisaged it as one of the brightest stars in the night sky. Nicholas of Cusa had turned the earth into a true star, but only at the price of turning the sun into an earth – hardly anyone was prepared to follow him.26 For Digges and Benedetti, even though they were Copernicans, from a vast distance the Earth, which received light but did not transmit it, would become a dark star. Leonardo, Bruno and Galileo realized that the Earth would look like a vast moon if seen from the moon, and recognized that when the moon was new you could see that it was being illuminated by the Earth – Harriot, after reading Galileo, named this ‘earthshine’, which is the term we use today.27 Galileo devised some elementary experiments to show that the Earth would reflect light, and that the land would reflect more light than the sea (which is why the moon shines so brightly by reflected light). Turning his new telescope towards Venus in 1610 he found that it had phases – proof that it too shone by reflected light. Moreover, it had a full set of phases, proof that it orbited the sun, as in the systems of Copernicus and Tycho Brahe.28 At this point it became apparent that seen from Venus the Earth would be one of the brightest stars in the sky.
Thus Copernicanism posed a straightforward question: What sort of planet is the earth? A full range of answers to that question was canvassed. Only one answer, that all the planets shine by reflected light, proved robust and stable. It took seventy years for that answer to establish itself, but once the telescope had been invented and turned into a scientific instrument it was the only answer with any chance of survival. The answer was entirely unpredictable in 1543, but entirely inevitable after 1611.
Once a scientific question is on the agenda of a community of scientists – once it becomes ‘alive’ – then over a period of time we may expect a range of possible answers to be explored; indeed, sometimes all possible answers may be proposed for consideration.29 In this early period it may be impossible to get agreement about which is the right answer. But over time a stable consensus will emerge that one answer is right and all the others are wrong. This consensus does not just depend on a rhetorical or political process of winning agreement; it also depends on the capacity over time for the supporters of one particular theory to rebut criticism and provoke fruitful lines of enquiry.30 A ‘robust’ or ‘stable’ answer comes to be regarded as quite simply the right answer. This does not mean that its rightness was always apparent, although careless historians and scientists often imply as much; it does mean that its rightness becomes incontestable, at least for a period of time.
The discovery of antipodes led straight to the concept of the terraqueous globe; but Copernicanism did not lead directly to the view that all planets shine by reflected light – the telescope had to intervene. Between the recognition that the Torricellian tube is a pressure gauge and the invention of the atmospheric steam engine there was no intervention of an extraneous factor. Boyle’s law was a natural development of Pascal’s Puy de Dôme experiment, and the atmospheric steam engine was a natural development of Boyle’s law (even if the technical problem of building an engine that worked was considerable). Torricelli did not for a moment imagine the steam engine, any more than Columbus imagined America; the path that led from the barometer to the steam engine was not as straight or as short as the path that led from Palos de la Frontera to the Bahamas; but the path was there, waiting to be found.
The search for the Northwest Passage, or the attempt to explain planetary motion on the model of magnetism, were mistaken but useful. There are plenty of other examples of scientific enterprises which have been doomed from their inception, but whose practitioners have simply refused to learn from experience: the attempts to turn base metal into gold or to cure infectious diseases by letting blood both lasted for more than two millennia, but neither could be made to work, and indeed neither produced valuable new knowledge, as both the quest for the Northwest Passage and Kepler’s new astronomy did. And it is this which is the fundamental problem with a relativist approach – either anything can be made to work, in which case the philosopher’s stone may yet be found, or some things can never work, in which case there is an external reality that constrains which beliefs are viable and which are not. Of course ‘made to work’ is a slippery concept: plenty of alchemists thought they had seen base metal turn into gold, and plenty of doctors thought they had cured their patients through bloodletting. People delude themselves in all sorts of ways. The idea that America was really Asia died within a generation, while the alchemists’ project was much more resilient.
Naive realists, those who think that science always establishes incontrovertible truths about the world (a view difficult to sustain, given the evidence that scientific theories change radically as the evidence they are based upon is revised),31 assume that scientific enquiry is always going to ask similar questions and produce identical answers; relativists assume that both the questions and the answers are infinitely variable. In truth the questions may be variable, but sometimes the answers are not. You do not have to sail west, but if you do you will end up in America. And once you have found America, if you were trying to get to Asia, then the search will begin for ways round it. One question leads to another; scientific enquiry is path-dependent.32
There is a common-sense view that takes this idea to an extreme. It holds that once you set out to answer a question the answer you will arrive at is entirely predetermined – just like Columbus’s discovery of America. Just as any two carpenters will agree on the length of a table, though one may measure in inches and the other in centimetres, so a Martian and an Earthling will agree on the speed of light, though they will surely have different units of measurement.
Thus according to the common-sense view the science of extraterrestrial beings, if they exist and if they are intelligent, must, where it overlaps with our science, agree with it. Steven Weinberg, a Nobel prize-winning physicist, expressed this view when he wrote ‘when we make contact with beings from another planet we will find that they have discovered the same laws of physical science as we have.’33 Science is thus a cross-cultural language which any culture can in principle learn to speak, and which any technologically sophisticated culture will already have learnt to speak. This was the assumption underlying a message that was broadcast into space by the Arecibo radio telescope in 1974. The message consisted of the numbers one through to ten, the atomic numbers of hydrogen, carbon, nitrogen, oxygen and phosphorus, the formulas for the sugars and bases in the nucleotides of DNA, the number of nucleotides in DNA, the double helix structure of DNA, a figure of a human being and its height, the population of Earth, a diagram of our solar system, and an image of the Arecibo telescope with its diameter. The assumption was that any extraterrestrial intelligence capable of receiving the message would recognize the maths and the science and quickly make sense of the Earth-specific information. The great mathematician Christiaan Huygens discovered the law of the pendulum in 1673; he believed that there were inhabited planets scattered across the universe; and by the time he died in 1695 he had persuaded himself that this law was known throughout the universe.34
The opposing view is that science is shaped by a whole series of cultural and social factors which ensure that no two societies would produce the same knowledge, just as no two societies produce the same religious beliefs. Scientific knowledge is not, in reality, unalterable truth, even if it appears to be. Consequently two different scientific communities will always produce two radically different bodies of scientific fact and theory, and science is not a cross-cultural form of knowledge but a local consensus, specific to a particular community. You might think that Boyle’s law is a bit like the New World – it was there waiting to be discovered. But the cultural determinists do not accept this. They think it is more like a signature dish – Escoffier’s peach melba, for example – the product of a very specific, local technology and culture (the Savoy Hotel in 1892).35 Just as certain dishes – peach melba, prawn cocktail – have managed to spread across continents and survive over time, so some scientific doctrines are successfully disseminated, while others remain firmly tied to their time and place of origin.
This book seeks to recognize that proponents of both positions have some good arguments and some bad arguments. Its criticisms are not only directed at relativists (who are given more space simply because their views are more prevalent in history of science); they are also directed at realists who fail to take to heart the evidence on which relativism is founded, the evidence of cultural difference across time and space – the evidence of history and anthropology. It is a standard ploy of opponents of relativism to argue that we all share certain commonsensical capacities for reasoning, and thus we can all recognize better knowledge when we see it.36 They take the view that science is, in essence, common sense applied systematically, or, as Karl Popper put it, ‘common-sense knowledge writ large, as it were’.xivxiv In my view, explaining science in terms of common sense is simply to go round in a circle. Evidently there are certain fundamental experiences and modes of reasoning that are universal and can be regarded as valid in any and every human culture. If there weren’t, cross-cultural communication would often be impossible.37 For example, there is some experience of hunting wild animals in every human culture. We have seen that Roman lawyers regarded vestigia, which originally meant the tracks left by an animal, as a form of evidence. The word ‘investigate’ has as its root meaning to follow tracks when hunting. Our word ‘clue’, in its meaning, new in the nineteenth century, of something that detectives are on the lookout for, is a metaphor derived from Ariadne’s thread which allowed Theseus to escape from the Minotaur’s labyrinth, an indication that there is nothing new about following the evidence and seeing where it will lead – the detective too is following a track, just as Theseus retraced his own steps. All human beings are capable of the sort of intellectual activity that tracking requires, and when we investigate a problem we are engaged, as the realists claim, in a sophisticated version of the same activity.38
But, and it is a big ‘but’, although there are certain cross-cultural uniformities common to all human societies, this doesn’t help us very much, most of the time. Firstly, although there may be a very few truly universal experiences and modes of thought, each culture naturalizes the experiences and ways of thinking that its members have in common as their own, local version of common sense – common to us, if not to those other people over there. Thus G. E. Moore thought it was common sense to believe in an external reality, but not in a creator God or a life after death, while these beliefs have seemed commonsensical to many people.39 In practice, the range of beliefs that can come to be generally shared and effectively unquestioned within a society is enormous. Discussions of common sense generally fail to distinguish adequately universal and local definitions of the concept.
Take just one example: through the Middle Ages and the Renaissance, scholastic philosophers followed Aristotle in thinking that earth is naturally heavy, and tends downwards, while fire is naturally light (it has negative weight, we might say) and tends upwards. Air and water were heavy or light depending on the circumstances in which they found themselves – depending on whether they were above or below their proper location. Aristotelians also held that solids are denser and heavier than liquids. Thus ice is heavier (because it is ‘denser’) than water. So why does it float? It floats because ice on a pond is flat, and the water resists its sinking. Wood, they claimed, was heavier than water, since the element out of which it was primarily made was earth. You can make a boat out of wood, they would say, but only if it has a flat (or at least flattish) bottom.40
The works of Archimedes were widely known in the Middle Ages, but the philosophers simply didn’t accept that his account of buoyancy was correct: Archimedes argued as if all substances have weight and tend downwards, and as far as the philosophers were concerned this represented a fundamental misconception. When they dismissed Archimedes as irrelevant they thought common sense was on their side; and their own experience of the world corresponded exactly to their claims. They were aware of no significant anomalies where their theories failed to correspond to reality. They sailed in ships, walked on pontoon bridges, and stepped across frozen puddles without ever experiencing anything that seemed to oblige them to rethink their theories. Nevertheless, as Galileo pointed out, it would have been easy to test their claim that ice floats because it is flat by breaking it up into small pieces and seeing if the pieces still floated.xvxv And in any case, why does a flat sheet of ice bob back to the surface if you push it under?
A reading of the philosophers who replied to Galileo will quickly establish that they thought they had reason, common sense, and the authority of Aristotle all on their side, although they were far from sharing a consistent position. One argued that Galileo’s ice bobbed to the surface because, although pure ice would be heavier than water (which would not prevent it floating, but would prevent it surfacing once submerged), his ice had air trapped within it, which explained why it would not stay submerged.41 Others pointed out that Galileo’s theory presented its own problems. Galileo appealed to sensory experience, but sensory experience established, or so they claimed, that boats float higher in the water when far from shore, and lower as they approach port. Neither Archimedes nor Galileo could explain this basic fact, but an Aristotelian, they were happy to report, could. (A larger body of water pushes more firmly against the bottom of the boat, and so the puzzle is solved.)42 Meanwhile they disagreed on the most fundamental questions: one thought wood heavier than water, another lighter; one held that water had no weight in its proper place, another denied it; one acknowledged that water expands when it freezes, another disputed it. One thing, though, they agreed on: Aristotle was always right.
There is a key difference between Galileo and his opponents: both appealed to experience, but Galileo had engaged in a programme of experimentation. His was an applied knowledge and theirs was not. Ultimately, the difference between him and them lies in the fact that they were prepared to make claims that they were sure were true but which they had never tested (wood is heavier than water; boats float higher in the water as they move away from the shore), while Galileo had tested each and every one of his claims. Their failure to conduct tests was reflected in their inability to agree on the most basic points. For the most part they did not dispute the truth of what Galileo stated as a matter of fact; but they expected him to accept their own factual claims, which of course he was not prepared to do. Thus they reported, on the authority of Seneca, that there was a lake in Syria where the water was so thick that bricks floated on top of it; invited to explain this, Galileo dismissed it as nothing but a tall tale, so that no explanation was necessary; at which they grew indignant, insisting that one should believe ‘authors worthy of faith, such as Seneca, Aristotle, Pliny, Solinus, and so forth’.43 In other words, the basic difference between Galileo and his opponents was that they were philosophers, while he was a mathematician in the process of becoming a scientist (rather than being, as one opponent claimed, a mathematician falsely claiming to be a competent philosopher).44
It seems pointless to maintain that Aristotle and his followers were deficient in common sense, or deficient in experience of how the world goes. They had plenty of both, according to the standards of their own time. What they lacked was the right intellectual tool-kit: in this case, a procedure for devising tests to confirm (or rather falsify) theoretical claims. They lacked this procedure for the fundamental reason that they thought it unnecessary: what could go wrong if one argued from uncontested premises to conclusions that necessarily followed? What could go wrong if one relied on the common experiences that we all share? So we come to a dilemma. Either common sense is amorphous and malleable, to the extent that all the beliefs that are shared within a community are compatible with common sense. Or, if you claim that many societies are lacking in common sense because they hold beliefs that could easily be refuted, then it seems there are communities in which, much of the time, nobody exhibits any common sense at all. The idea of ‘common sense’ either proves too much or too little. Either all societies have enough of it to get by, in which case the concept hardly helps one understand what makes for reliable knowledge; or only the arguments that accord with our own exhibit common sense, in which case across history and across cultures common sense has been in very short supply.
When Susan Haack writes, ‘Our standards of what constitutes good, honest, thorough inquiry and what constitutes good, strong, supportive evidence are not internal to science. In judging where science has succeeded and where it has failed, in what areas and at what times it has done better and in what worse, we are appealing to the standards by which we judge the solidity of empirical beliefs, or the rigor and thoroughness of empirical inquiry, generally,’ she is, I submit, confusing two distinct issues.45 Yes, in our society certain standards of enquiry (including an emphasis on the acquisition of empirical information) are not peculiar to science, but are widely shared – that is only because the Scientific Revolution, and the wider cultural shifts that made it possible, have shaped the whole of our culture. It would however be simply wrong to think that we and Aristotle share the same views about what constitutes a justified true belief, and about how to acquire such beliefs. So the problem is: Who is ‘we’ here? Is it, to use her examples, we (contemporary) historians and we (contemporary) detectives? Or is it we as human beings who share a certain capacity for common sense with all other human beings? The first interpretation is largely true, but not very significant; the second would be significant, but is untrue.
Moreover, notions of how to judge empirical evidence vary from one enterprise to another. Boyle and Newton believed in the transmutation of base metals into gold; but they were excellent judges when dealing with other empirical questions. In the sixteenth century Jean Bodin wrote a book entitled Method for the Easy Comprehension of History (1566), which is often regarded as a founding text of modern historical reasoning; he also wrote a book, On the Demon-mania of Witches (1580), arguing that there are witches everywhere and that human beings regularly metamorphose into wolves. Both sets of views seemed to him equally commonsensical. A century later, Thomas Browne campaigned against the epidemic of false beliefs (that elephants have no knees, for example); but he continued to believe that Lot’s wife had been turned into a pillar of salt, and (in his professional capacity as a doctor) he appeared in a witchcraft trial to certify that supernatural forces were at work, thus ensuring the conviction of the accused.46 Browne was a pretty sensible chap, at least by the standards of the day, and we will not get far by trying to argue that his views on witchcraft show that he was incompetent to make judgements about empirical beliefs. If he had views about what constitutes evidence that were different from ours, then that simply shows that ‘common sense’ varies from one culture to another.
To watch our modern conception of common sense (as applied to science and other types of empircal enquiry) being born we must turn again to the example of Galileo’s revision of Archimedes’ principle. Galileo was pretty much the same person when he got the right results as he had been only a few days previously when he was still getting the wrong results. What had changed? The answer is straightforward: he had begun an experimental programme of testing and refining his theories. He had started with one anomaly, the floating chip of ebony, and discovered another, the floating needle; as a consequence he had done his best to make sense of surface tension. He had also set out to illustrate Archimedes’ principle in action, and as a consequence had discovered a further anomaly, that a weight of water could lift a weight greater than itself. Boldly, he had gone back, reanalysed Archimedes’ principle, and revised it. He had then tested his new theory with a quite different experiment. This back and forth movement between theory and evidence, hypothesis and experiment has come to seem so familiar that it is hard for us to grasp that Galileo was doing something fundamentally new. Where his predecessors had been doing mathematics or philosophy, Galileo was doing what we call science. The difference between Galileo at the start of this process and Galileo at the end is that at the end evidence was being used to impose much tighter constraints on argument than at the beginning; evidence and argument were interacting in a new way. In arguing for such an interaction Galileo could indeed appeal to what both he and we would think of as universal principles of common sense, even though both his experimental practice and his conclusions were novel; but such appeals are always problematic when they run against convictions so entrenched that it has come to be assumed that they are indisputable truths. All parties to a dispute always claim that common sense is on their side.xvixvi Galileo seems to have had absolutely no success in his efforts to convince the philosophers that he understood why bodies float better than they did.
The whole of the Scientific Revolution is encapsulated in Galileo’s little treatise on floating bodies. The subject had been discussed by brilliant philosophers and mathematicians for 2,000 years. The views of the philosophers and mathematicians, although they were sharply at odds with each other, corresponded satisfactorily to daily experience, or at least so they believed. Neither theory was in crisis. Before Galileo, no competent mathematician had doubted that Archimedes provided a complete explanation of what happens when bodies float, and no competent Aristotelian philosopher had disputed the view that shape is crucial in determining which bodies float. And yet Galileo realized, not only that ice was lighter than water, but also – it must have been an extraordinary shock – that bodies often float without displacing their own weight in water. Everybody had been wrong.
We cannot explain the timing of this intellectual revolution simply by appealing to some local, contingent event, an argument between Galileo and some Aristotelian philosophers (though such an argument did indeed take place). Nor was Galileo, when he invented the hydraulic press, working on some new, practical problem that only a competent engineer could tackle. He was simply addressing an age-old question: Why do some bodies float and others sink? If he came up with new answers it was because he employed new methods and was evolving a new intellectual tool-kit. Pascal went on to clarify and systematize Galileo’s new understanding. His situation was a little different: he was studying pressure in fluids in order to understand how the weight of air supports a column of mercury in a barometer. (Actually, he was inventing the concept of pressure – Galileo had thought in terms of weights, not pressures.)47 Pascal’s new hydraulics is an extension of his vacuum experiments, but Galileo’s work on floating bodies is not path-specific in this way. It springs, not from a new problem, but from a new type of practice and a new way of thinking.
Aristotle might have had difficulty in following Galileo’s argument that the size of the container matters when trying to understand floating bodies, but Archimedes would not. So what separates Galileo from Archimedes? In what sense is this new thinking if Archimedes would have had no difficulty in understanding it? First, Galileo lived in a culture where even the most authoritative beliefs could be questioned – this was the legacy of Columbus. He lived in an age of discovery. Second, Galileo was constructing a new science in which everything was, at least in principle, subject to measurement, even the rise in the level of a pond when a duck enters the water, or of an ocean when a ship is launched. This principle of ever-more-exact measurement derives from Tycho Brahe; in tightening up the relationship between evidence and theory in physics, Galileo was extrapolating from the practices of the astronomers.xviixvii Third, Galileo had the example of Gilbert, who had used the careful manipulation of experimental apparatus to establish new and unsuspected truths. Columbus, Brahe and Gilbert did not supply arguments that Galileo used; they supplied role-models that he was bold enough to follow. They did not contribute directly to his new hydraulics, but the new hydraulics was made possible by the intellectual culture they had helped to shape – specifically, Galileo’s intellectual culture, for most of his contemporaries were still happy to defer to Aristotle and were uninterested in challenging orthodox beliefs. It is only, as we have seen, with Pascal that this peculiar, idiosyncratic culture began to become widely adopted and generally respected.
Archimedes had been convinced that the real world is mathematically legible, even though oceans and ships are not shaped like circles, triangles and squares; he had been sure that Euclid’s geometry is more powerful than Aristotle’s syllogisms. There is no way of proving that in every possible world the mathematicians will be better equipped to understand the world than the philosophers: it is something that can only be established through the success of mathematical practice, and that requires a culture in which mathematicians are permitted to challenge the claims of philosophers and are rewarded for their successes.48 But Galileo was not simply a second Archimedes; he was also an experimental scientist. Nothing guarantees that mathematicians will be interested in taking this further step. In fourteenth-century Oxford, mathematically inclined philosophers had hypothesized that falling bodies accelerate continuously; but they had made no effort to confirm this theoretical possibility through experiment. That was left to Galileo. Indeed, the Oxford philosophers had hypothesized that colour and temperature might in principle be quantified and might change with increasing or decreasing rapidity; they had no way of measuring colour or temperature, just as for all practical purposes they had no way of measuring the speed of falling bodies. Their speculations were purely abstract and theoretical; they applied to any possible world and not to our actual world. They studied mechanics in theory but they had no practical interest in machines.xviiixviii
Archimedes had been available in Latin since the twelfth century, and in print since 1544: before Galileo every mathematician was happy to think about ships floating in a boundless ocean, but no one had floated model boats in model oceans and studied exactly what happened. The finest mathematicians were entirely satisfied with Archimedes’ principle, which seemed to them both coherent and complete. Galileo was the first to turn Archimedes’ account of how bodies float into a theory to be tested with an experimental apparatus; at which point the theory proved incomplete.
Galileo’s determination to embody mathematical theory in a corresponding apparatus is fundamental. He even hoped to build a mechanical model that would illustrate his theory of the tides.49 From it follows the great advantage of Galileo’s analysis over Aristotle’s and even over Archimedes’, which is that Galileo’s analysis gives you better prediction and more control: the proof of the pudding is in the eating. And then, of course, success once achieved and disseminated is redescribed as common sense, and we are assured that any sensible person can see that there is only one way – the right way, in this case Galileo’s way – of tackling the problem, by moving back and forth between induction and deduction, by devising thought experiments and real experiments, by never trusting a theory until it has been tested and attempts have been made to falsify it.xixxix ‘Common sense’, if we mean the practices which we think embody it, did not exist before the publication of Galileo’s Discourse on Floating Bodies in 1612.xxxx Galileo is the first person who is one of us in the sense of ‘us’ that is required when Susan Haack writes of ‘our standards of what constitutes good, honest, thorough inquiry and what constitutes good, strong, supportive evidence’.50 In seeking to understand a case like Galileo’s study of floating bodies we can, in short, have too much realism, in which case we will never understand Galileo’s originality or the opposition he faced; but we can equally have too much relativism, in which case we will never acknowledge that he was right, and his opponents were wrong.
Thus we are obliged to recognize that both realists and relativists have a good point. We must run with the hare and hunt with the hounds. Like the relativists, we must acknowledge the dangers of arguing from universal standards of human rationality (which is not to say that such arguments are always invalid – Galileo’s knowledge of buoyancy was much more reliable than those of his opponents, as they would have discovered if they had embarked on a programme of experimentation). Like the realists, we must insist that it is not so difficult to tell good science from bad, as long as you agree that knowledge should be carefully and systematically tested against experience.
I started the previous section of this chapter by presenting two alternative views of science, and I have argued that both have merit. From one point of view, the knowledge we end up with seems to be culturally relative, contingent, peculiar; from another it seems to be commonsensical, predictable, inevitable. Kuhn tried to maintain both these views, despite the tension between them, by distinguishing revolutionary science from normal science. The outcome of a revolution is, he argues, culturally relative, contingent, peculiar; but it leads to a period of stability during which progress is the norm. Kuhn’s distinction between two types of science was too neat, but his basic approach was sound: sometimes, as we have seen, one discovery leads to another in a way which is, with hindsight, entirely inevitable. Between Copernicus and Newton a number of revolutions took place, and there is no simple path that leads from one to the other; but from Torricelli to Newcomen the route is fairly straightforward, and once the barometer had been invented and experimentation had become accepted as the best route to new knowledge, the steam engine could straightforwardly follow. In the history of science there is no simple solution to the apparently binary choice between radical contingency and predictable evolution. For the choice is a false one. The answer is always somewhere between the two extremes, and the balance between the two has to be struck afresh with each new topic.
We have been studying the origins of science. We have seen that science is an enterprise we have invented and agreed to play by certain rules. There are plenty of enterprises where we invent the rules and we change them when we want to. The age of majority used to be twenty-one; now it is eighteen. Women used to be denied the vote; now they have it. There are other enterprises where our ability to change the rules is constrained by factors over which we have no control.
Gardeners, for example, create a peculiar micro-environment for their plants. If the gardener stops working, nature takes over. A garden is thus both natural and artificial, and it is both of these things entirely and simultaneously. It is easy to assume that the law of the excluded middle requires something to be either natural or artificial: thus a shirt is made of a material which is either natural (cotton, linen, wool) or synthetic (nylon, polyester). But rayon is both natural and artificial, being made from wood. So, too, a boat capable of sailing into the wind uses natural forces to achieve a result that would not occur in nature. In gardening, in cooking, in naval architecture there are lots of culturally specific choices that we can make, but there are plenty of other things that simply cannot be done. Plants die, mayonnaises curdle, ships sink. Wishing and willing will not make it otherwise.xxixxi Such activities depend on a complex collaboration between the natural and the social. Thus it must be wrong to say, as Andrew Cunningham does, that science is ‘a human activity, wholly a human activity, and nothing but a human activity’:51 it is wholly a human activity, but it is not ‘nothing but a human activity’. Poetry and Scrabble are nothing but human activities. Science belongs to the very extensive class of activities which combine the natural and the artificial, which are constrained by both reality and culture.
The peculiar feature of science is that it claims not simply to cooperate with nature (as gardeners, cooks, and naval architects do), but to discover a truth that existed before that cooperation began. It is not surprising that the history of science is a problematic activity, because science itself constantly claims to escape from its own temporal specificity, its own artificiality. In claiming to escape from its own process of production, science presents itself as being natural, not artificial; it is only to be expected that, in opposition to this obvious misrepresentation, some want to claim that science is entirely artificial and not at all natural. But the simple truth is that it is both, and that the scientists are right to claim that this artificial enterprise can discover what happens in nature.xxiixxii
Some would simply deny that it is possible to escape culture and discover nature in the way that scientists claim. Bruno Latour maintained that the fact that the bacterium that causes tuberculosis has been discovered in the lungs of an ancient Egyptian pharaoh, Ramses II, did not mean that Ramses died of tuberculosis. Tuberculosis was only discovered in the nineteenth century. Before its discovery there was no such thing as tuberculosis, and consequently nobody could die of it. This is just wrong: of course Ramses II did not know he was dying of tuberculosis, but nevertheless we now know that it was tuberculosis that killed him. Latour’s historicism misses a key point about science, which is that it is about matters which are the case whether we believe them to be so or not. The bacterium which causes tuberculosis was discovered not invented by Robert Koch in 1882. Latour says that Ramses II could no more die of tuberculosis than he could be killed by a Gatling gun (Richard Gatling invented the Gatling gun in 1861), making clear that he thinks discovery and invention are the same thing. They are not. The Gatling gun involves a new sort of cooperation between nature and society; but the bacterium that causes tuberculosis requires no deliberate cooperation on our part, even if identifying it and killing it do require techniques which originate in the laboratory and involve a complex cooperation between nature and society.xxiiixxiii52
Science, as a method and practice, is a social construct. But science as a system of knowledge is more than a social construct because it is successful, because it fits with reality.53 This fit cannot be shown to be necessary or inevitable, which is what the realists do not understand. Aristotle thought his method was necessarily trustworthy; he was wrong. If our method works better than his it is because it fits better with the world as it is, not because the world was bound to be like that.xxivxxiv Nevertheless, wherever this fit is established (and it has to be established anew in each new scientific discipline) it establishes a positive feedback loop. Outside the narrowly mathematical disciplines (including astronomy and optics), that loop was first closed in 1600. Consequently, we need to think of science as being the result of an evolutionary process where good science has had, over the last five centuries, a better prospect of survival than bad science. As Kuhn rightly put it, ‘Scientific development is like Darwinian evolution, a process driven from behind rather than pulled toward some fixed goal towards which it grows ever closer.’54
The problem with the relativists is that they explain bad science and good science, phrenology and nuclear physics, in exactly the same way – advocates of ‘the strong programme’ explicitly insist on this equivalence.xxvxxv The problem with the realists is that they assume there is nothing peculiar about the method and structure of science. According to them the scientific method is somehow natural, like walking, not artificial, like a watch. This book will look, I trust, realist to relativists and relativist to realists: that is how it is meant to look. It stands in the tradition of Kuhn’s 1991 lecture ‘The Trouble with the Historical Philosophy of Science’. There Kuhn criticized the relativists (who had taken much of their inspiration from his own work), saying that their mistake was
taking the traditional view of scientific knowledge too much for granted. They seem, that is, to feel that traditional philosophy of science was correct in its understanding of what knowledge must be. Facts must come first, and inescapable conclusions, at least about probabilities, must be based upon them. If science does not produce knowledge in that sense, they conclude, it cannot be producing knowledge at all. It is possible, however, that the tradition was wrong not simply about the methods by which knowledge was obtained but about the nature of knowledge itself. Perhaps knowledge, properly understood, is the product of the very process these new studies describe.55
The task, in other words, is to understand how reliable knowledge and scientific progress can and do result from a flawed, profoundly contingent, culturally relative, all-too-human process.
One of the obstacles to understanding knowledge (to echo Kuhn) lies in the vocabulary we use to discuss our difficulties. There is a satisfactory label for people who think there is not really such a thing as knowledge at all (but merely belief systems that pass as knowledge): they are relativists. But there is no collective term for all the different positions which have in common a recognition that some forms of knowledge of nature are more successful than others, and that consequently knowledge can progress. One could of course adopt the term ‘progressivist’, but that would elide all the difficulties that are associated with the idea of progress. For progress often comes to a halt, and in many fields of life one step forward results in two steps back. Progress is not linear or incremental, and it is often difficult to reach agreement on the standard by which it should be measured. Nevertheless, it happens.
What all these groups – whether they call themselves realists, pragmatists, instrumentalists, fallibilists, or whatever – have in common, apart from a willingness to recognize progress when they see it, is a recognition that nature (or reality, or experience) establishes practical constraints on what can pass as successful prediction or control – that nature ‘pushes back’.56 These people recognize that scientific knowledge is not fully determined, nor is it undetermined; it is semi-determined. It is not possible to be a full-blooded relativist and to acknowledge that nature pushes back; but it is possible to be a constructivist (to say that we make knowledge out of the cultural resources available to us) and to acknowledge the resistance of nature. Indeed scientific knowledge, properly understood, must be seen to be both constructed and constrained. Hasok Chang has proposed the label ‘active realism’ for this double recognition.xxvixxvi57
Anyone who tries to occupy this ‘best of both worlds’ position needs to face up to a further challenge, that of fleshing out the notion that nature pushes back. Kuhn saw this challenge, but he misdescribed it. He complained about those who ‘freely acknowledge that observations of nature do play a role in scientific development. But they remain almost totally uninformative about that role – about the way, that is, in which nature enters the negotiations that produce beliefs about it.’58 If you try to grasp Kuhn’s meaning you will find it slips through the fingers. For there is a sense in which science itself is the account of how nature enters into negotiations that produce beliefs about it; in which case the question that Kuhn is asking is not really an historical or philosophical question, but a request to have some bit of science explained to him.
So we have to turn Kuhn’s formulation around. In order to understand the way in which the physical world enters the negotiations that produce beliefs, we must look at the ways in which we communicate with and about it. On one level, this is a question about equipment: the telescope transformed the way in which astronomers negotiated with nature. Secondly, it is a question about intellectual tools: the concept of laws of nature, for example, shapes the sorts of questions scientists ask and the sorts of answers nature provides. In the dialogue between the scientist and the physical world, the physical world (by and large) stays the same, while what scientists bring to the dialogue changes, and this transforms the role that the physical world plays. The ways in which nature pushes back alter as we alter. Hence the need for an historical epistemology which allows us to make sense of the ways in which we interact with the physical world (and each other) in the pursuit of knowledge. The central task of such an epistemology is not to explain why we have been successful in our pursuit of scientific knowledge; there is no good answer to that question. Rather it is to track the evolutionary process by which success has been built upon success; that way we can come to understand that science works, and how it works.
Earlier in this chapter I quoted Popper’s claim in 1958 that science ‘is common-sense knowledge writ large, as it were’. In that, as we have seen, he was wrong. A few months later he added a new epigraph to the second printing of The Logic of Scientific Discovery, a quotation taken from the papers of the historian Lord Acton (1834–1902): ‘There is nothing more necessary to the man of science than its history …’59 And just what should the scientist and the citizen learn from the history of science? That nothing endures. That just as the theories of Ptolemy and Newton seemed perfectly satisfactory for centuries, so too our most cherished theories will one day be supplanted. As Kuhn pointed out over and over again, one of the central purposes of an education in the sciences is to hide this basic truth from the next generation of scientists.60 Science replicates itself by indoctrination, since scientific communities work most efficiently when they are agreed about what they are trying to do.
But, as Kuhn also grasped, the fact that even the best-established scientific theories may not endure does not mean that they are unreliable, and it does not mean that science does not progress. Ptolemy gave the astrologers the information they needed, and Newton explained Kepler’s laws of planetary motion. We demonstrate the reliability of modern science at every moment of every day. Recognizing both science’s limitations and its strengths requires a peculiar mixture of scepticism and confidence; relativists overdo the scepticism and realists overdo the confidence.