I shook my head. ‘Many men have been hanged on far slighter evidence,’ I remarked.
‘So they have. And many men have been wrongfully hanged.’
– Arthur Conan Doyle, ‘The Boscombe Valley Mystery’ (1891), The Adventures of Sherlock Holmes
To repeat the question: What is science? The answer: knowledge of natural processes based on evidence. In which case there can be no science without a concept of evidence. Yet if we start looking for the word ‘evidence’ being used by seventeenth-century scientists we discover something peculiar: they have the word, but they hardly ever use it. Bacon, for example, who is of course familiar with the use of the word ‘evidence’ in a legal context, never employs it when discussing natural philosophy.1 Either they have a different concept of evidence from ours, or there is some obstacle to their use of the word.2
We need to start by recognizing that we use the word ‘evidence’ in four different senses. First, ‘evidence’ can refer to something that is evident. It is evident that 2 + 2 = 4. This is the original meaning of ‘evidence’, in that it comes directly from the Latin evidentia. Because this is etymologically the base meaning of the word, the Oxford English Dictionary lists it first, with two examples of its original use in 1665 – despite the fact that it has examples of the use of the word in other senses going back to 1300 (in its earliest sense in English, ‘an evidence’ is an example to imitate). One of the first examples given is from Robert Boyle: ‘[T]here are certain Truths, that have in them so much of native Light or Evidence that … it cannot be hidden.’3 The comparison here between what is obvious to the mind and what is obvious to the eye runs through the use of ‘evidence’ in this sense. A passage from John Locke’s Essay Concerning Humane Understanding (1690) illustrates this eye-language well:
Perception of the Mind, being most aptly explained by Words relating to the Sight, we shall best understand what is meant by Clear, and Obscure in our Ideas, by reflecting on what we call Clear and Obscure in the Objects of Sight. Light being that which discovers to us visible Objects, we give the name of Obscure, to that, which is not placed in a Light, sufficient to discover minutely to us the Figure and Colours, which are observable in it, and which, in a better Light, would be discernable. Thus our simple Ideas are clear, when they are such as the Objects themselves, from whence they were taken, did, in a well-ordered Sensation or Perception, present them …4
Although Locke does use the word ‘evidence’ (‘the degrees of its evidence’, ‘Certainty and Evidence’) elsewhere, he much prefers to use the word ‘clear’, and to write of intuitive knowledge as being ‘like the bright Sun-shine, [which] forces it self immediately to be perceived, as soon as ever the Mind turns its view that way’.5 His discussion of clear and distinct ideas thus follows the example of Descartes, who maintains that only ideas which are clear can be used in argument.ii
One reason Locke avoids using the word ‘evidence’ as far as he can is that the word in English has multiple meanings. Thus in 1654 Walter Charleton had offered a translation into English of two Latin phrases with which Gassendi had summarized the epistemology of Epicurus: ‘That Opinion is true, to which the Evidence of Sense doth either assent, or not dissent: and that false, to which the evidence of Sense doth either not assent, or dissent.’6 Since he is translating the Latin word evidentia, he ought to mean ‘obviousness’ or ‘evidentness’ by ‘evidence’ here, and his gloss implies that he does: by the ‘Assent of the Evidence of Sense, is meant an Assurance that our Apprehension or Judgment of any Object occuring to our sense, is exactly concordant to the reality thereof; or, that the Object is truly such, as we, upon the perception of it by our sense, did judge or opinion it to be’.7 So the evidence of sense is not, as one might think, the testimony of the senses but our confidence that our senses have properly grasped the object. Charleton’s example is a figure walking towards us from a distance: at a certain point it becomes obvious that it is Plato. The Oxford English Dictionary is certainly wrong to suggest that the word was not used in this sense before 1665. Here is Thomas Jackson in 1615 carefully using the word in its Latin sense:
Evidence, besides cleerenes or perspicuity (directly and formally included in its prime and native signification) collaterally drawes with it a conceit of such plenary comprehension of the object knowne, as fully satiates our desire of its knowledge: (for evident wee hardly accompt that knowledge which leaves the apprehensive faculty capable of further or better information then it already hath from the particulars which we desire to know) …’8
Following Jackson, let us call this type of evidence Evidence-Perspicuity.
Second, there is ‘evidence’ used as a term in English (and only in English) law. Initially (from 1439), English courts considered testimony and evidence – evidence being documents relevant to the case; then (from 1503), ‘evidence’ became a portmanteau term, covering both testimony and documentary evidence. ‘Evidence, (Evidentia),’ writes John Cowell in The Interpreter (1607), a book explaining legal terminology, ‘is used in our lawe generally for any proofe, be it testimonie of men or instrument.’9 There is no single Latin term for ‘evidence’ in this legal sense; documents are instrumenta and testimony is testimonium. Let us name this portmanteau legal sense Evidence-Legal. We find Charleton using ‘evidence’ in this sense, too, in his Ternary of Paradoxes: ‘For we will make it our business now, for your information, to call the action of Magnetism to the bar, and by the evidence of Meridian truths, convince the ignorance and stupidity of its adversaries.’10
From an even earlier date ‘evidence’ meant anything that gives one grounds for belief or assent (Evidence-Assent). So Cowell went on to extend his definition of evidence: in a trial, he says, the accused is called on to testify. He ‘telleth what he can say: after him likewise all those, who were at the apprehension of the prisoner, or who can give any Indices or tokens, which we call in our language (Evidence) against the malefactour’. He is quoting Sir Thomas Smith (d.1577). Smith and Cowell realize that this further sense of ‘evidence’ is peculiar to English. The Latin for indices and tokens is signa or indicia; the French is preuves. Thus we have a fourth sense of ‘evidence’, Evidence-Indices. In English courts Evidence-Indices is considered only in so far as presented via testimony or documents, as part of Evidence-Legal.
Evidence-Indices is what we are referring to when we say that science depends upon evidence. Thus a fingerprint left at the scene of a crime is an Evidence-Index, an indication, or token, that someone in particular has been there. Cowell gives an example in explaining the word ‘bankrupt’:
Bankrupt, (alias brankrowte) commeth of the French (banque route) and (faire banqueroute) with the French is as much as (foro cedere, solum vetere) with the Romanes: The composition of the French word I take to be this, (banque, i.[e.] mensa) & (route, i.[e.] vestigium) metaphorically taken from the sign left in the earth, of a table once fastned unto it, & now taken away. So that the originall seemeth to have sprung from those Roman (mensarii) which as appeareth by many wrighters, had their (tabernas et mensas) in certaine publique places, wherof when they were disposed to flie, & decieve men that had put them in trust with their monies, they left but the signes or carkasses behind them.11
You entrust some money to someone who holds a stall in the marketplace. One day you go to the market and where the stall used to be there is only a mark in the earth, a vestigium. This is a token, indication, or sign that the stall has gone; the fact that the stall has gone is a sign that your banker has gone out of business; and the fact that your banker has gone out of business implies that you have lost your money.
Charleton uses ‘evidence’ in this sense, too. He describes how if you draw a circle around a tumour with a sapphire the tumour will die, and argues that the sapphire continues to act on the tumour from a distance (magnetically): ‘[T]he place it self will afford a more certain and satisfactory evidence on the behalf of Magnetism; for it grows not black and torrid in the minute of, or by the affriction of the Saphire; but many minutes after … the virulency does successively breathe forth, in obedience to the Magnetical alliciency [attractiveness] of the absent Gem.’12
Arguments of this sort were well known to the ancient Romans. These are arguments which start from things (or from what we call facts). They are discussed in Book 5 of Quintilian’s Institutes of the Orator, a work which dates to the first century CE. Thus B is found dead, with A’s knife in him. This is a sign that A murdered B; unless of course A’s knife has been stolen from him, or B attacked A, so that A was acting in self-defence. Such signs therefore cannot be full proofs but only indications; they have to be interpreted in the context. Here is Quintilian. A sign
is that from which something else is inferred, for example murder from blood. But it may be the blood of a sacrificed animal that has got on to the [suspect’s] clothes, or just a nosebleed: a man whose clothes are bloody has not necessarily committed murder. But though this sign is not enough in itself, in combination with others it is taken as equivalent to a witness’s statement [ceteris adiunctum testimonii loco ducitur] – if the man is an enemy, or has previously made threats, or was in the same place. When the sign is added to these it makes what was only suspected seem certain.13
We say that such Evidence-Indices is circumstantial evidence, where the legal understanding of ‘circumstantial’ essentially means ‘contextual’. The word ‘circumstantial’ goes back to Quintilian, who is the only Latin author to use circumstantia to refer not to a spatial distribution (the sheep standing around the shepherd) but a problematic deduction which is context-dependent. Quintilian’s example of an appeal to circumstance is a made-up one. Imagine there is a law which says that a high priest can pardon one person from the death penalty, and another law which says that if one person committing adultery is put to death then their partner must be put to death as well. The high priest is taken in adultery and condemned to death. ‘No problem,’ he says; ‘I will pardon myself.’ ‘Not at all,’ comes the reply, ‘because if you pardon yourself your partner cannot be executed, and so you will have pardoned two people, and that you cannot do. So you must die.’ In the specific context of adultery the high priest’s right of pardon cannot be used.14
From Quintilian’s Latin, ‘circumstance’ and ‘circumstantial’ enter English to refer to an incomplete (but not insignificant) argument for belief, one that needs to be interpreted in context. Here is the Jesuit Robert Parsons in 1590:
For albeit the Apostle S. Paul declareth, the things which we beleeve, be not such in themselues, as may be made apparent by reason of humaine arguments: yet such is the goodnes and most sweete proceeding of our mercifull God toward us, as he wil not leave himselfe without sufficient testimonie, both inward and outward, as the same Apostle in another place doth testifie. For that inwardlie, hee testifieth the trueth of such things as we beleeve, by giving us light & understanding, with internall joy and consolation in beleeving them. And outwardly, hee giveth testimonie to the same, with so many conveniences, probabilities, and arguments of credibilitie, (as Divines doe call them,) that albeit the very poynt of that which is beleeved, remaine styll with some obscuritie: yet are there so many circumstaunces of likelyhoods, to induce a man to the beleefe thereof, as in all reason it may seeme against reason to denie or mistrust them.15
Probabilities, arguments of credibility, circumstances of likelihoods – these all derive from Evidence-Indices. They are not in themselves ‘testimony’, equivalent to words spoken or texts written; but when we find them in the Bible, as it were orchestrated by God, they become the equivalent of testimony. It is the Bible and the tradition of the Church which provides the textual testimony; the circumstances contextualize the testimony and accompany it. Inward light, understanding and joy, similarly, function as if they were testimony: to Parsons, it is as if they speak to the truth of our beliefs.
Between these four types of evidence (Perspicuity, Legal, Assent, Indices) there was and is plenty of scope for confusion, and for collapsing one into the others. John Wilkins in his posthumous Principles and Duties of Natural Religion (1675) has an extended discussion of evidence in the broadest possible sense of the grounds for belief (Evidence-Assent); consequently, he includes under the heading of ‘evidence’ sensation, demonstration [i.e. deduction], testimony and experience – the first two being evidence of and the second two evidence for.16 Evidence for a belief (Evidence-Legal, Evidence-Indices) could be better or worse, stronger or weaker, producing different ‘degrees of assent’ or ‘degrees of Veracity, Certainty, or Credibilitie’.17 Similarly, the evidence of a belief (Evidence-Perspicuity) could be greater or less, clearer or more obscure, producing different degrees of knowledge. For Locke there are ‘three degrees of Knowledge, viz. Intuitive, Demonstrative, and Sensitive: in each of which, there are different degrees and ways of Evidence and Certainty’.18 In Locke’s view Evidence-Indices – and he does use the word ‘evidence’ in this sense when discussing probabilities – is not a form of knowledge (which is confined to intuition, demonstration and sensation: all types of evidence of) but of opinion, which is measured in terms of ‘degrees of assent’. Nevertheless, some Evidence-Indices can be regarded ‘as if it were certain knowledge’.19
Obviously, different types of expert work with different types of knowledge. Mathematicians deal with demonstrative knowledge, or Evidence-Perspicuity. Aristotelian philosophers thought that all true knowledge could be expressed in syllogistic form, arguing from indisputable premises to undeniable conclusions, all based on Evidence-Perspicuity. On the other hand, lawyers were concerned with Evidence-Legal and Evidence-Indices, and so were theologians. Since 1400 theologians had been discussing what they called ‘moral certainty’ – evidence good enough to rely on, even if there is a great deal at stake. Thus I can be morally certain that there is a city called Rome, even if I have never been there. There is testimony, there are documents, there are maps, there are photographs, and there is plenty of other evidence to confirm the existence of a place called ‘Rome’. It is utterly implausible that all of this evidence could have been faked, so I am quite certain that Rome exists. But my certainty is not of the same sort as my knowledge that the square on the hypotenuse is equal to the sum of the squares of the other two sides, which can be rigorously proved; the evidence for Rome is an argument from experience, and so an argument from probability.iiii A standard argument was that a Christian needed moral certainty of the truths of their faith, since the fate of their soul was at stake.
Some theologians were not content with moral certainty: in 1689 the Presbyterian propagandist Richard Baxter discussed the concept of evidence at great length, and decided that the only sort of evidence that counted was Evidence-Perspicuity. One couldn’t rely, he maintained, on experience in any of its many forms. ‘Even our Experimental Philosophers and Physicians find, that an experiment that hits ofttimes, quite misseth afterwards on other Subjects, and they know not why. A course of effects may oft come from unknown causes.’20 True faith required certainty, and certainty required self-evidence, or Evidence-Perspicuity. (Baxter’s efforts to prove that the Bible was self-evidently true need not detain us.)
It should be apparent at this point that one way of characterizing the Scientific Revolution is as the displacement of Evidence-Perspicuity by Evidence-Indices, as people learning to trust circumstantial or probable evidence in place of intuitive or demonstrative evidence. Thus in the Torricellian experiment you cannot see the pressure of the air, but the height of the column of mercury is an indicator of that invisible pressure. When you look at the moon through a telescope you cannot see mountains, but the ragged terminator indicates that mountains are there. When Galileo saw the moons of Jupiter he could not see that they were moons, but their patterns of movement indicated that they were in orbit around Jupiter. In each case, what you can see points beyond itself to something you can reliably infer. The mathematicians began to handle evidence in the way that lawyers and theologians had been handling it for many centuries.
So far we have been looking at evidence mainly through English sources. In English law the jury was the judge of matters of fact: its task was to weigh the evidence and decide if the defendant was guilty or not.21 There were no set rules on how to reach that decision. The ‘beyond reasonable doubt’ rule was formulated only in the eighteenth century, and the whole point of this rule is that juries have to decide for themselves what it means. Juries were thus free to reach decisions on the basis of circumstantial evidence if they chose to. As a character in a play of 1616, entitled, with obvious sarcasm, The Honest Lawyer, puts it:
In case of Murder should we never judge
By circumstanciall likelihoods and presumptions,
No life could be secure.22
In Roman law jurisdictions the situation was, as we have seen, very different. There were clear rules on how to handle evidence.23 It was the judge who collected the evidence, applied the rules and reached a verdict. A verdict of guilty in a capital case required a complete proof of guilt, for example the testimony of two witnesses who had seen the crime occur, or a confession.iiiiii When Quintilian says that circumstantial evidence can take the place of a witness, later lawyers took him as authorizing it to be considered as half of a complete proof. In a capital case, where there was less than a complete proof, the standard procedure was to use torture in order to obtain a confession, but torture could be employed only if there were substantial grounds for suspicion amounting to half a proof. Where in English we generally use the word ‘proof’ to mean a demonstration (a mathematical proof), so that the idea of half a proof makes no sense, judges across the Continent and in Scotland accumulated proofs until they either had a complete proof or sufficient evidence to justify torture. Gossip, for example, represented one seventy-second of a proof.iviv
So the French, following their Latin sources, used the world preuve, or ‘proof’, not the word ‘evidence’; but they saw proofs as things that could be accumulated in exactly the same way as evidence can be accumulated until it amounts to a proof beyond doubt. But where in English we can speak of ‘the evidence’ as a totality – ‘the evidence of his guilt was overwhelming’ – in French it would be necessary to use a plural, les preuves. The standard French translation of ‘evidence-based medicine’ is médecine fondée sur les faits. The French were not alone in having the same word for Evidence-Indices as for proof; this was the case in all the other modern European languages apart from English and Portuguese (where evidencia is frequently found in the plural, as it was in eighteenth-century English).24
Thus there is (or there appears to be) a fundamental continuity between the way in which Quintilian talks about signs and tokens and the way in which seventeenth-century Englishmen and women talk about Evidence-Indices and seventeenth-century continental Europeans talk about proofs in law. The claim has been made, however, that there was no concept of evidence until about 1660 (which is why I have concentrated on examples of Evidence-Indices taken from before this date).25 The claim is based on distinguishing three types of evidence: the evidence of witnesses; the evidence of the senses; and the evidence of (for want of a better term) clues.26 Ian Hacking distinguishes the last two from each other by quoting J. L. Austin:
The situation in which I would properly be said to have evidence for the statement that some animal is a pig is that, for example, in which the beast itself is not actually on view, but I can see plenty of pig-like marks on the ground outside its retreat. If I find a few buckets of pig food, that’s a bit more evidence, and the noises and smell may provide better evidence still. But if the animal then emerges and stands there plainly in view, there is no longer any question of collecting evidence; its coming into view doesn’t provide me with more evidence that it’s a pig. I can now just see that it is.27
This concept of evidence (Evidence-Indices, it would seem) was missing in the Renaissance, so the claim goes. Instead, they had the concept of signs.28
This claim involves a series of errors. First, it confuses ‘signs’ (that is, indices, vestiges or tokens) with ‘signatures’.29 According to the Renaissance theory of signatures, some natural objects had their significance apparent in their form. Thus a kidney-shaped bean might be good for treating a disease of the kidneys. This doctrine, held by Platonists and Paracelsians, is quite different from the theory of signs (also known as indices, vestiges or tokens). Second, it is claimed that the doctrine of signs/signatures belongs to ‘low’ disciplines such as medicine and alchemy; no mention is made of the law or theology. Third, it is asserted that signs were ‘read’ as if they were texts and, consequently, no distinction was drawn between the evidence of clues and the evidence of witnesses. This is the point at which this argument becomes interesting, for, as noted earlier, Quintilian does take signs to be equivalent to testimonies, and so does Parsons. For them, testimony is the paradigm form of evidence (Evidence-Legal) to which Evidence-Indices is expected to conform.
Nevertheless, Quintilian carefully distinguishes between what he terms ‘technical’ and ‘non-technical’ proofs.30 Non-technical proofs are things like documents, witnesses, confessions extracted under torture: they speak for themselves. Technical proofs have to be constructed by the advocate. He notes, as we have seen, that some signs virtually speak for themselves – bloodstained clothing, a cry – but others depend heavily on interpretation. Indicia, vestigia and signa (tokens, vestiges and signs) thus do not provide the same sort of evidence as documents and witnesses, even though they can be used to do the same work as a document or a witness. The claim that signs are read as texts is held to follow from the Renaissance belief that the universe is a book, the book of nature, but this is to place the theory of signs in the wrong context. The theory of signs originates in the law, and signs are treated as if they speak because court cases are discursive performances. It is the prosecutor’s job to turn blood on the suspect’s clothes into the equivalent of a witness against them – he must make the blood speak.
Pierre Gassendi, in his Syntagma philosophicum (1656), elaborated the classical doctrine of signs into a sophisticated theory of knowledge.31 He identifies two types of sign. There are those that allow us to know something that we could have known by direct sensory experience if we had been present at the right time. These are vestiges: the banker’s table, or the pig’s trotter, or the criminal’s finger leave a vestige, a track or a trace. So, too, we would say, the craters we see on the moon are vestiges of past collisions with asteroids. A sign ‘leads us to the knowledge of something hidden in the way that tracks [vestigium] are a sort of sign indicating to a dog which way he should pursue the chase in order to catch the quarry’.32 By ‘vestige’ Gassendi means Evidence-Index. On the other hand, he says, there are signs that point to something that we can never see. Thus we cannot see the pores in the skin, but the passage of sweat through the skin proves they must be there. We can never see the legs of the scabies mite, but we can tell from the fact that it moves that it must have legs or something similar. In fact, Gassendi points out, when the microscope was invented both the pores in the skin and the legs of the mite became visible, thus confirming the validity of earlier arguments to prove their existence. Such arguments depended on analogies: comparing the skin, for example, to porous earthenware. This concept of argument from analogy had been taken up by the Epicureans from medicine, but it was also familiar to lawyers. Thus Quintilian assumes the lawyer will appeal to what we might call stereotypes: ‘It is easier to believe brigandage of a man, poisoning of a woman.’ These are arguments by analogy, based on the circumstances of the case.33
Gassendi is certainly right to emphasize the importance of arguments by analogy. Robert Boyle, when, in 1660, he wanted to explain the new doctrine of the elasticity of the air, compared the air to sheep’s wool, which could be compressed but which would bounce back if the pressure was removed: this analogy, he felt, made plausible the notion of elasticity. Torricelli compared the weight and pressure of the air to the weight and pressure of water. The distinction between two very different types of inference, between vestiges and analogies, is still important for Locke. Nobody disputed the reliability of vestiges (no scar without a wound, for example), but analogies were obviously much more problematic. Clearly, if much of our knowledge is analogical in origin it cannot be certain, and the real causes of events may always escape us.
Gassendi saw no need to make an explicit distinction between the evidence of vestiges and the evidence of witnesses, but he certainly did not think that the scar ‘testifies’ to the wound.34 He had a perfectly good concept of evidence as distinct from testimony. However, if you mistakenly assume that the key problem lay in distinguishing clues from testimony, then you might conclude that it was only in The Logic of Port-Royal that Arnauld finally differentiated the two, for we are told that he drew a distinction between ‘internal evidence’ and ‘external evidence’. The internal evidence is the evidence of clues (his knife was in the victim); the external evidence is the evidence of witnesses (his wife says he never left her side). Except, of course, Arnauld didn’t use the word ‘evidence’, as he was writing French, not English. The word he uses is circonstances. For example: ‘In order to judge of the truth of some event and to decide whether or not to believe in its occurrence, the event need not be considered in isolation – such as a proposition of geometry would be; rather all the circumstances of the event, both internal [the clues] and external [the testimony], should be considered.’35
‘Circumstances’, we have seen, is Quintilian’s coinage. Quintilian also separates the evidence of signs or clues from the evidence of witnesses; indeed, as in The Logic of Port-Royal, the evidence of signs is ‘internal’ and the evidence of witnesses is ‘external’. In the Port-Royal logic it is a bit hard to grasp what the evidence of signs is internal to. The knife may be in the body, but exactly what is the fingerprint inside? In Parsons, the distinction, as we have seen, is quite simple: internal evidence consists of my feelings, which are inside me. In Quintilian it is a little more complicated: witnesses and documents come to the lawyer from outside; technical proofs are constructed by the lawyer himself, so that they are formed within the discipline of oratory. Technical proofs are the lawyer’s own contribution to the evidence. Arnauld is not copying Quintilian, but he is reworking him in order to go beyond him.vv
Is The Logic of Port-Royal (1662) the first text to draw a clear distinction between testimony and Evidence-Indices? As we have seen, in earlier texts it is certainly assumed that the one can stand in for the other, and sometimes (as in Parsons) the two seem to be conflated. Nevertheless, here is Richard Hooker (d.1600): ‘[T]hings are made credible, eyther by the knowne condition and qualitie of the utterer, or by the manifest likelihood of truth which they have in themselves.’36 This is exactly the distinction between internal (‘in themselves’) evidence and external (testimony) evidence made in The Logic of Port-Royal. Hooker implicitly recognizes that both types of evidence have to be considered in the light of the circumstances. The accused has blood on his clothes, but he is a butcher; the testimony is explicit, but the witness is unreliable. The fact that the one can substitute for the other does not mean that Hooker thinks they are one and the same thing; they are clearly different in character, one depending on the regularities of nature, the other on the veracity of people. Another example: in 1648, well before the publication of The Logic of Port-Royal, Wilkins was already arguing that Archimedes’ discoveries (such as his famous burning mirror, with which he destroyed a fleet of ships) might seem too astonishing to be true (‘such strange exploits, as … would scarce seeme credible even to these more learned ages’; i.e. if we can’t do them, how could he?), ‘were they not related by so many and such judicious Authours’, above all Polybius, who was either an eyewitness or at least had the opportunity to speak to eyewitnesses.37 Here, internal evidence is set against external evidence, as it must often have been in a court of law. (The suspect had blood on his clothes, but his wife says he never left her side.)
Thus the claim that there is a new concept of evidence in the 1660s is mistaken; wherever we look what we find is nothing but the reworking of Quintilian’s distinctions. What is new is the transfer of concepts from one discipline to another. Evidence-Indices had been the business of lawyers and theologians; in 1660 it became the business of the Royal Society. ‘Moral certainty’ had been the language of theologians; in 1662 we find it being used by the first statisticians, Graunt and Petty.38 Just as facts moved out of the court room into the laboratory, so evidence made the same move at around the same time; and, as part of the same process of constructing a new type of knowledge, moral certainty moved from theology into the sciences. When it comes to evidence, the new science was not inventing new concepts but recycling existing ones.
The search for a new concept of evidence in the mid-seventeenth century is bound to be fruitless because there was a classic context in which the question of the reliability of inferences from facts had been debated for centuries. That context was the discussion of Ptolemaic epicycles. According to the Aristotelian philosophers, epicycles ought not really to exist: all movement in the heavens ought to be circular movement around the centre of the universe. For them epicycles were useful fictions which made computation of the position of the planets possible. The mathematicians, however, interpreted the apparently irregular movement of the planets in the heavens as evidence that some invisible reality was causing the movement. They took their observations of planetary movement to be good evidence for the reality of epicycles. Here is Clavius outlining the mathematicians’ view:
[J]ust as in natural philosophy we arrive at knowledge of causes through their effects, so too in astronomy, which has to do with heavenly bodies very far away from us, we must attain to knowledge of them, how they are arranged and constituted through our senses … It is therefore suitable and very highly rational that, from the particular motions of the planets and the various appearances, astronomers should search out the number of particular circles that carry the planets round with such varying motions, and their arrangement and shapes … But our opponents try to weaken this argument, saying that they concede that all the phenomena can be saved by postulating eccentric circles and epicycles, but that it does not follow from this that the said circles are found in nature; on the contrary they are wholly fictitious; for perhaps all the appearances can be saved in a more suitable way, though it is not yet known to us … But by the assumption of eccentric and epicyclic circles not only are all the appearances already known preserved, but also future phenomena are predicted, the time of which is altogether unknown. Thus if I am in doubt whether, for example, the full moon will be eclipsed in January 1582, I shall be assured, by calculation from the motions of the eccentric and epicyclic circles, that the eclipse will occur, so that I shall doubt no further … But it is not credible that we should force the heavens (but we seem to force them if eccentrics and epicycles are fictions, as our adversaries would have it) to obey our fictions and move as we wish or as agrees with our principles.39
Clavius’s argument here is identical with that of modern realists who claim that science must approximate to the truth, otherwise it would not be able to make successful predictions. Robert Boyle, on the other hand, sided with the philosophers and presented a line of argument that goes back to Averroes and forward to the modern pragmatists and instrumentalists:
[A]s confidently as many Atomists, and other Naturalists, presume to know the true and genuine Causes of the Things they attempt to explicate, yet very often the utmost they can attain to in their Explications, is, That the explicated Phaenomena May be produc’d after such a Manner as they deliver, but not that they really Are so: For as an Artificer can set all the Wheels of a Clock a going, as well with Springs as with Weights, and may with violence discharge a Bullet out of the Barrel of a Gun, not onely by means of Gunpowder, but of compress’d Air, and even of a Spring. So the same Effects may be produc’d by divers Causes different from one another; and it will oftentimes be very difficult, if not impossible for our dim Reasons to discern surely which of those several ways, whereby it is possible for Nature to produce the same Phaenomena she has really made use of to exhibit them.40
From the outside, a clock powered by a battery and a clock powered by clockwork look the same: the fact that the hands go round does not tell you what sort of mechanism is driving them. In scholastic terms this was a debate over the reliability of a posteriori reasoning; in our terms it is a debate over the evidence of things, or over Evidence-Indices. The debate was not new in the second half of the seventeenth century, and neither were the arguments. Here is the humanist and philosopher Alessandro Piccolomini in 1558:
[S]uppose we should see a stone strike a wall and with great force, and not knowing the origin of such fury we should imagine that the stone had come from a bow or a crossbow. And suppose that our theory was false and that as chance would have it the stone had come from a sling shot. Nevertheless it would have struck the wall with the same fury if it had come from the imagined bow. For the aforesaid fury of that stone could have derived from more than one cause. Similarly when we see many appearances in the planets in the sky, though the causes from which such appearances truly arise are hidden from us, nevertheless it is enough for us that, supposing these theories to be true, those appearances would derive from them just as we see them. This for us is more than enough for the calculations, predictions, and information which we need to have of the positions, places, magnitudes and motions of the planets.41
If this debate between realists and instrumentalists seems to echo our own debates about the nature of scientific knowledge of invisible entities such as the electron, this is because what is being put to work here is the very same concept of evidence that we use. All that is missing from these discussions is the word on which we place so much emphasis, and for which they felt no need: ‘evidence’. Their vocabulary of appearances, predictions and causes is perfectly adequate for the task.
From the 1640s, with the triumph of experiment, the sort of evidence that had been good enough for the lawyers, the doctors and the astronomers – the evidence of clues, or facts – began to be good enough for mathematicians such as Pascal when doing physics. A posteriori reasoning, reasoning backwards from appearances to causes, began to push to one side the a priori reasoning of the geometers and the scholastic philosophers, who maintained that the only reliable form of reasoning was reasoning forwards from definitions to consequences. Just as the astronomers acknowledged that, in principle, different hypotheses might do the job equally well (Clavius had no doubt that the Copernican system produced reliable predictions, but he felt sure the Earth was stationary, not moving), so many people continued to argue that there were a number of perfectly good competing explanations of the Torricellian experiment. Pascal disagreed.
It was this new sort of knowledge that Sprat had in mind when he defended the Royal Society from its critics, insisting that the starting point of the new knowledge, experimental evidence, was extremely reliable. Indeed, Sprat uses (exceptionally for a seventeenth-century author) the very word ‘evidence’ as we would use it:
[T]here is not any one thing, which is now approv’d and practis’d in the World, that is confirm’d by stronger evidence, than this, which the Society requires; except onely the Holy Mysteries of our Religion. In almost all other matters of Belief, of Opinion, or of Science; the assurance, whereby men are guided, is nothing near so firm, as this. And I dare appeal to all sober men; whether, seeing in all Countreys, that are govern’d by Laws, they expect no more, than the consent of two, or three witnesses, in matters of life, and estate; they will not think, they are fairly dealt withall, in what concerns their Knowledg, if they have the concurring Testimonies of threescore or an hundred?42
But for the most part the new scientists avoided the word ‘evidence’ because it inevitably carried with it an implicit reference to the law courts – a reference that Sprat was willing to make explicit. Thus in 1660 Boyle describes one experiment ‘as a plausible, though not demonstrative proof, that Water may be transmuted into Air’.43 Here, he uses ‘proof’ as if he were writing preuve in French, where we would use ‘evidence’ in the sense of Evidence-Indices. Nowadays, in English, we can no more have a merely plausible proof than we can have a false fact; but Boyle is using ‘proof’ differently from us.
Above all, Boyle, like any seventeenth-century scientist, realizes that most of his readers will be mathematicians. And he feels he must:
excuse my selfe to Mathematicall Readers. For some of them, I fear, will not like that I should offer for Proofs such Physical Experiments, as do not alwayes demonstrate the things, they would evince, with a Mathematical certainty and accuratenesse; and much less will they approve, that I should annex such Experiments to confirm the Explications, as if Suppositions and Schemes, well reason’d on, were not sufficient to convince any rational man about matters Hydrostaticall.44
In other words, he feels he must apologize for appealing to Evidence-Indices in a field where mathematical demonstration (Evidence-Perspicuity) seemed possible. This aspiration towards demonstration was not confined to what we would think of as the empirical sciences but was also commonplace in theology. Thus in 1593 the mathematician John Napier presented his interpretation of the Book of Revelation as being ‘in form of proposition, as neer the analytick or demonstrative manner as the phrase and nature of holy Scripture will permit’.45
Mathematics in the Renaissance looked in two directions. Aristotle had distinguished between geometry and arithmetic (which dealt with purely theoretical entities) and optics, harmonics and astronomy (which dealt with physical realities). Bacon provided labels for this distinction: ‘pure mathematics’ and ‘mixed mathematics’. (Bacon extends the list of types of mixed mathematics to include perspective, engineering, architecture, cosmography ‘and divers others’.)46 Pure mathematics deals in proofs and demonstrations; mixed in phenomena. But pure mathematics had a higher status, with the result that there was a constant hankering after the language and argumentative style of pure mathematics.
Galileo, for example, liked to keep as close to geometry as he could. He measured the length of the shadows on the moon, and used geometry to demonstrate how high the mountains that caused them were; he used geometry to prove (rather elegantly) that the spots on the sun must be on or near the surface of the sun.47 But these proofs involved arguing from things (shadows and shapes) to the applicability of geometrical theorems. Indeed, they started as analogies. Galileo thought the patches of light and dark along the moon’s terminator looked like a mountain range seen from above when the sun was rising – so like that he argued that it could only be exactly that. The spots on the sun reminded him of clouds; he knew they were not clouds, but they bore the same relationship to the sun’s surface as clouds do to the Earth’s. The new science often presented itself as a system of axioms and demonstrations, in Galileo’s Two New Sciences for example, or Newton’s Principia, but it was always grounded in facts and, less surely, in analogies.
This helps to explain the rarity of the word ‘evidence’ (which carried with it implications of disagreement and contention, of court-room conflict) in these seventeenth-century texts. It occurs three times (once as a verb) in Sprat’s History of the Royal-Society. In the Opticks (1704), the greatest triumph of the new experimental science, Newton uses it only once. In the early volumes of the Philosophical Transactions it appears only once or twice a year. Even in a work as late as Desaguliers’ Course of Experimental Philosophy (1734–44) the word appears only twice in two volumes. As we have seen, after 1660 the new scientists talked endlessly about ‘facts’ (though Newton avoided the term as not suitable for a mathematician), ‘experience’, ‘experiments’, ‘hypotheses’, ‘theories’ and ‘laws of nature’. When they used the word ‘evidence’ it was usually casually and inadvertently, and often because (as in the example quoted earlier from Sprat) they wanted to imply a comparison with law and/or theology.
If there was one term that summed up the new science for those who were involved in constructing it, it was not ‘evidence’ but ‘experience’. Pascal went further than simply insisting on the authority of experience; he argued that our knowledge of nature is capable of endless progress because it is founded in experience, and experience accumulates over time.48 The emphasis on experience was thus tied to the ideas of progress and discovery. Of course, there was something deeply problematic about relying on the measurement of the height of mercury in a tube to substantiate a theoretical understanding of nature: the measurement is a particular event, conducted with particular equipment, on a particular day, under particular circumstances, while the theory has to be universally applicable. Early experimentalists such as Galileo tried to play down this problem by reporting an experiment in general terms as having been conducted over and over again, but from Pascal on the experiment becomes a local event, described as such. Such narratives do not play down the epistemological problem of moving from the specific to the universal; they accentuate it.49 One way round this problem is to devise a series of different experiments which capture the phenomenon from different perspectives: Pascal was keen to go beyond the Torricellian experiment and devise new ones precisely in order to bridge this gap.
The correspondence between theories and facts was so crucial that Galileo and Newton were prepared to bend the facts to fit the theories, even while insisting on the priority of experience over theory. Mersenne, at any rate, was convinced that Galileo’s falling-body experiments could not be precisely replicated, and we know that Newton fiddled the figures to create a perfect fit between his theoretical physical principles and the measured speed of sound.50 We would say that Galileo’s and Newton’s sciences were always empirical (at least in aspiration), but that would be to use the word in a nineteenth-century sense – in the seventeenth century empirics were thought of as people who were untrained in reasoning, not people who grounded theories in evidence.vivi Gassendi and Locke never thought of themselves as founding an empirical philosophy, although we would say that is what they were doing.
Just as the new scientists aspired to mathematical demonstration wherever possible, and claimed an overly precise fit between theory and fact when they needed to, so, in English, they avoided the word ‘evidence’, which was inevitably associated with the law, as far as they could. The result is that where we can identify the Typhoid Mary, or the index case, for the languages of facts (Chapter 7), of laws of nature (Chapter 9) and of hypotheses and theories (Chapter 10), there is no index case for the language of evidence – or, if there is, it lies outside science.
For the language of Evidence-Indices (as distinct from the concept) was taking off towards the end of the seventeenth century not in science but in works of natural theology, such as John Wilkins’ Of the Principles and Duties of Natural Religion (1672; 75 occurrences) and Matthew Hale’s The Primitive Origination of Mankind (1677; 280 occurrences, but then Hale was Chief Justice). It became commonplace in philosophy: it occurs forty-eight times in Hume’s Treatise (1739–40). The long-term fortune of the word has a great deal to do with William Paley’s Natural Theology; or, Evidences of the Existence and Attributes of the Deity (1802). Slowly but surely we can see the language of Evidence-Indices moving from law, theology and philosophy into the sciences, but the language was far behind the concept, for experimentation was nothing other than an appeal to Evidence-Indices.
It would be wrong, however, to concentrate solely on the word ‘evidence’ rather than the concept which it expresses, for if we do so we will miss a crucial development. For Locke and everyone before him, knowledge equals truth. Defeasible knowledge, knowledge that might be corrected in the light of new observations, was merely opinion or probability. Moral certainty had been introduced as ‘hard’, reliable opinion, but the whole point about morally certain knowledge was that you could commit yourself to it without reserve; it would, in practice, never be falsified, even if it could not be demonstrated as mathematical theorems could be demonstrated. The concept of evidence (meaning relevant experience) had been introduced into science from the law. In English law, once the jury had given a verdict there could be an appeal on a question of law but there could be no appeal on a question of fact. There was, until 1907, no procedure for introducing new evidence.51 The jury thus had to be certain of their verdict. Moral certainty, grounded in probability, had, in the writings of Wilkins and Locke, joined deductive or self-evident knowledge as a form of truth. This simply conformed to the practice of the courts.
Locke had discussed a case where what looked like reliable knowledge turned out to be mistaken: that of the King of Siam, who was not willing to believe the testimony of the Dutch ambassador, who assured him that in the Netherlands water could become so hard that an elephant could walk on it. But Locke never proposed such a case as a paradigm example of how empirical knowledge (as we call it) works when it is working well. He formulated the general principle that ‘as the conformity of our Knowledge, as the certainty of Observations, as the frequency and constancy of Experience, and the number and credibility of Testimonies, do more or less agree, or disagree with it, so is any Proposition in it self, more or less probable’.52 This means that what is probable changes over time; but Locke never nerved himself up to take the further step of saying that knowledge itself changes over time. On the one hand, there is what he calls here ‘our Knowledge’, which is changeable. On the other, there is Knowledge, which is Truth.
The Oxford English Dictionary usefully distinguishes between ‘knowledge’ meaning ‘the fact of knowing or being acquainted with a thing, person, etc.’ and ‘knowledge’ meaning ‘justified true belief’ (Latin: scientia). ‘Our Knowledge’, in Locke’s phrase, which must conform to observation, experience and testimony, is knowledge as acquaintance, but Locke still hankers over justified true belief. In that respect he still thinks like Hobbes, who had written:
This taking of Signes by experience, is that wherein men do ordinarily think, the difference stands between man and man in Wisdom, by which they commonly understand a mans whole ability or power cognitive; but this is an errour: for the Signes are but conjectural; and according as they have often or seldom failed, so their assurance is more or less; but never full and evident: for though a man have always seen the day and night to follow one another hitherto; yet can he not thence conclude they shall do so, or that they have done so eternally: Experience concludeth nothing universally. If the Signes hit twenty times for one missing, a Man may lay a Wager of Twenty to One of the event; but may not conclude it for a Truth.53
A radical shift from the position of Hobbes (who clearly states Hume’s problem of induction and rejects Evidence-Indices as no basis for certainty), and even from that of Locke (who may be described as clearly defining the choice between Evidence-Perspicuity and Evidence-Indices but then fudging it) is to be found in William Wotton’s Reflections upon Ancient and Modern Learning (1694):
The new Philosophers, as they are commonly called, avoid making general Conclusions, till they have collected a great Number of Experiments or Observations upon the Thing in hand; and, as new Light comes in the old Hypotheses, fall without any Noise or Stir. So that the Inferences that are made from any Enquiries into Natural Things, though perhaps set down in general Terms, yet are (as it were by Consent) received with this Tacit Reserve, As far as the Experiments or Observations already made, will warrant.54
Wotton’s ‘Tacit Reserve’, which is the principle that all scientific reasoning is defeasible, is of fundamental importance.viivii It transforms science from knowledge of the truth, regarded as indisputable, into a form of progressive knowledge in which established truths may always be disputed and in which an ultimate truth is never attained. Where did Wotton get the concept of tacit reserve from? The phrase itself comes from moral philosophy, where, traditionally, every promise was accompanied by an unspoken reservation, ‘If I Can; If I Ought; or if things Continue in the same State: So that by the Change of Circumstances, I am discharg’d of my Obligation.’55 But the principle that intellectual systems are merely temporary constructions which may need later to be revised and improved comes directly from the mathematical language of hypotheses and theories which Wotton invokes when he writes of ‘the old Hypotheses’ which ‘fall without any Noise or Stir’.
Now, with Wotton’s formulation of a tacit reserve, moral certainty was required to give way to a new sort of temporary knowledge, a purely provisional understanding. This tacit reserve had never been so clearly stated before. Locke’s King of Siam case is one where a tacit reserve would have been appropriate, but Locke had not formulated one. What Wotton understands is that scientists can agree (‘as it were by Consent’), for the moment, to treat acquaintance knowledge, or experience, as true belief, and that this is not a mistake but rather the way in which you change an ideal, unchanging knowledge-as-truth, a final unappealable verdict, into a peculiar, progressive form of knowledge, knowledge-as-incomplete-acquaintance. Induction is always imperfect, the evidence is always incomplete, but it can be good enough to be going on with. Wotton’s formulation of the notion that in science all knowledge is accompanied by a tacit reserve gives him a claim to be the first to have an adequate understanding of the conceptual underpinnings of modern science; or, if you prefer, to be the first both to understand modern science and to acknowledge its limitations. It is only at this point that the modern theory of evidence becomes complete (although, of course, Wotton writes of experiences and observations and does not use the word ‘evidence’). So, when we finally come to Wotton’s formulation of a tacit reserve, we encounter, for the first time, a sound comprehension of the nature of scientific knowledge.56
Nevertheless, searching for the concept of evidence pays off in unexpected ways. Indeed, it brings us to a discovery which is strange and unexpected. For when scientists began to make assessments regarding the reliability of evidence they were required to exercise what they all called ‘judgement’ (for example, Locke: ‘Knowledge being to be had only of visible certain Truth, Errour is not a Fault of our Knowledge, but a Mistake of our Judgment giving Assent to that, which is not true’).57 And exercising judgement requires a specific set of virtues, the virtues you would hope to find in a jury of your peers: impartiality, assiduity, sincerity. We find these virtues everywhere in discussions of Evidence-Indices, while they are largely irrelevant in discussions of Evidence-Perspicuity. Here, for example, is a theologian in 1677 explaining the difference between certainty and faith, between Evidence-Perspicuity and Evidence-Indices:
A Mathematical Demonstration brings so strong a Light that the Mind cannot suspend its assent, but is presently overcome by the naked propounding of the Object: And hence it is that in Mathematical matters, there are neither Infidels nor Hereticks. But the motives of Faith are such, that although the Object be most certain, yet the Evidence is not so clear and irresistible, as that which flows from Sense, or a Demonstration. And ’tis the excellent observation of Grotius, God has wisely appointed this way of perswading Men the truth of the Gospel, that Faith might be accepted as an act of Obedience from the reasonable Creature. For the Arguments to induce belief, though of sufficient certainty, yet do not so constrain the mind to give its assent, but there is prudence and choice in it.58
Mathematicians don’t need prudence; Christians, lawyers and scientists do. So Sprat describes his ideal philosopher: ‘The True Philosophy must be first of all begun, on a scrupulous, and severe examination of particulars: from them, there may be some general Rules, with great caution drawn’; ‘Let us then imagin our Philosopher, to have all slowness of belief, and rigor of Trial, which by some is miscall’d a blindness of mind, and hardness of heart.’59 The scientist must be slow, scrupulous, severe, rigorous. Here is Wilkins: ‘The word Moderation is a Quality, a Habit, an Affection of intellectual virtue, whereby we are concerned for any truth according to a due measure, not more or less then the evidence and importance of it doth require, to which the notion of fierceness or fanaticalness is opposed as the deficient extreme.’60 The scientist must be moderate: this is a new type of intellectual virtue. Of its novelty Wilkins is quite clear: we have no choice, he says, but to adopt the view which we think is best supported by the evidence:
yet must it withal be granted to be a particular virtue and felicity to keep the mind in such an equal frame of judging. There are some men, who have sufficient abilities to discern betwixt the true difference of things; but what through their vicious affections and voluntary prejudices, making them unwilling that some things should be true; what through their inadvertency or neglect to consider and compare things together, they are not to be convinced by plain Arguments; not through any insufficiency in the evidence, but by reason of some defect or corruption in the faculty that should judg of it. Now the neglect of keeping our minds in such an equal frame, the not applying of our thoughts to consider of such matters of moment, as do highly concern a man to be rightly informed in, must needs be a vice. And though none of the Philosophers (that I know of) do reckon this kind of Faith (as it may be styled), this teachableness and equality of mind in considering and judging of matters of importance, among other intellectual virtues; yet to me it seems, that it may justly challenge a place among them.61
Impartiality is also now an intellectual virtue. Here is Locke:
Nevertheless, I do not yet Question, but that Humane Knowledge, under the present Circumstances of our Beings and Constitutions, may be carried much farther than it hitherto has been, if Men would sincerely, and with freedom of Mind, employ all that Industry and labour of Thought, in improving the means of discovering Truth, which they do for the colouring or support of Falshood.62
Sincerity and industry are intellectual virtues, too.
So as knowledge ceases to be a question of Evidence-Perspicuity and becomes a question of Evidence-Indices a whole new set of intellectual virtues are demanded of the knower. Eventually, with the concept of a tacit reserve, a principle from moral philosophy is introduced directly into epistemology, establishing a limit to knowledge claims. We might want to say that these virtues and this limit can be summed up in the word ‘objectivity’; but objectivity is a nineteenth-century concept and implies new ways of observing nature and recording information.63 It would be wrong to read it back into the Scientific Revolution; before the precision instrumentation of the Industrial Revolution impartiality and judgement were virtues, not ways of re-describing professional competence.
Discovery implies individualism and competition. Scientists have to be buccaneering and entrepreneurial. But, as Robert K. Merton kept pointing out, science isn’t only about individual success. Scientists are required by the culture of their profession to declare their allegiance to a quite different set of values, which he summarized as communism (knowledge is shared: often renamed ‘communalism’; we have seen the first community of experimental scientists emerging in 1640s France), universalism (knowledge must be impersonal and unbiased), disinterestedness (scientists must help each other) and organized scepticism (ideas must be tested and tested again).64 This set of values is sometimes referred to by the acronym CUDOS. Thus every scientist is subject to two competing and conflicting imperatives: they are obliged to be both competitive and cooperative. Scientists are required to be Janus-faced, both self-effacing and self-assertive, and Merton saw his task as a sociologist of science as being to work out how scientists negotiated this conflict, which he regarded as being constitutive of science as a social enterprise.viiiviii
How did this conflict come about? The answer is very simple. It is the result of combining discovery with the moral virtues associated with Evidence-Indices. The result is a structural conflict in the nature of science, a conflict whose origins are historical. You won’t find Copernicus, or Kepler, or Galileo praising moderation, impartiality, industry; but then Copernicus, Kepler and Galileo are primarily mathematicians. The generation after Galileo had to acknowledge its dependence on Evidence-Indices and so, willy-nilly, it had to adopt the virtues of the judiciary.
A great deal of play has been made with the idea that the Royal Society was committed to the pursuit of facts, to impartiality, to moderation, because it was founded in the immediate context of the Restoration.65 For twenty years people had been killing each other in the name of the truth; now they must learn to manage their disagreements in a different way. The new science has to be placed in this local context. I don’t deny that there is truth in this, but it hardly explains why Merton found the same virtues being admired by scientists in the 1940s as were admired by scientists in the 1660s. The new virtues go deeper than the immediate context of the Restoration.
Where else do we find a similar conflict between competition and cooperation? In the legal profession. The adversarial system means that lawyers want to win, and the better they are at winning, the more they get paid. At the same time, every lawyer is an officer of the court. They are bound by a code of professional standards. They must never lie on behalf of a client. They must never withhold evidence from the other side. They must be simultaneously competitive and cooperative. What happened when Evidence-Indices moved from the court room to the laboratory was that the contradictory characteristics of any evidence-based legal system (there are other legal systems – trial by ordeal, for example – which don’t have these characteristics) were imported into science, and scientists became divided against themselves just as lawyers had always been within adversarial legal systems, right back to Quintilian, who is looking all the time both for good arguments and for winning arguments, knowing full well that the two aren’t always the same thing.
With the Renaissance revival of Stoicism the word ‘philosophical’ acquired a new meaning: philosophers, it was held, were able to moderate their passions and to be unmoved by the blows of fortune.ixix They could stand back from their immediate experience and contemplate the bigger picture. Wilkins, Sprat and Locke are looking for a quite different sort of person, one who exemplifies CUDOS. This chapter began with the search for a new type of evidence; it ends with the Janus face of a new type of intellectual, one produced by philosophers being required to engage with an old type of evidence: circumstantial evidence.
There is a further step in this argument. In 1976 Thomas Kuhn published an article entitled ‘Mathematical versus Experimental Traditions in the Development of Physical Science’.66 In England, Kuhn claimed, experimental science flourished from the late seventeenth century onwards, science conducted in the tradition of Bacon. On the continent, a much more deductive style of science was preferred, science conducted in the style of Descartes. The English were preoccupied with facts, the French (for it is primarily the French he has in mind) with theories. As Kuhn states it, this argument, which sets the experimentalists against the mathematicians, seems wrong. The English had Halley and Newton, who were simultaneously experimentalists and mathematicians. The French had Pascal, the Cassinis and Huygens (the Cassinis and Huygens not being French by origin, but by choice), ditto.xx It was a Frenchman, Claude Bernard, not an Englishman, who wrote An Introduction to the Study of Experimental Medicine (1865).
But we might restate Kuhn’s argument in a different way. The English had the common law, which relied on the jury system. The jury system allowed a significant role to circumstantial evidence, as long as it was introduced in testimony. It gave prosecutors plenty of scope to argue from analogy. When the scientists reorganized science around Evidence-Indices they imported into it the virtues of the jury system, at least in an idealized form: a willingness to listen to both sides, a desire to mix proof with persuasion, an appeal to common sense. (Newton, of course, is a great exception here.) The French, by contrast, had a Roman law system. Their new science was organized around the virtues of a juge d’instruction: intellectual rigour, a set of formalized procedures, a quest for a complete proof, a confidence that one need only answer to other professionals. If there are two different scientific traditions, as Kuhn claimed, perhaps they should be seen as reflecting the importation into science of two different legal traditions, of two different ways of handling Evidence-Indices, rather than as embodying a conflict between mathematicians and experimentalists. Perhaps it is therefore wrong to say that one can safely translate the English ‘evidence’ with the French preuve, for the two terms reflect two different and incommensurable forensic cultures, trial by jury and trial by inquisition. French Evidence-Legal was different from English Evidence-Legal, and so French Evidence-Indices has always been different from English Evidence-Indices.xixi
A friend of mine was once in hospital in Paris. The doctors told him that they had an hypothesis regarding the nature of his illness which they intended to prove, where in England they would have told him that he had certain symptoms which suggested a diagnosis which they would run tests to confirm. Two cultures: one emphasizes the difference between Evidence-Indices and Evidence-Perspicuity; the other minimizes it. But, nevertheless, they have one common enterprise, that of transmuting signs and symptoms into knowledge.
What I have offered here is what Boyle would call a plausible proof – a good argument (I hope) that falls short of being conclusive. What I hope to have shown is that science modelled its new concern with Evidence-Indices on legal processes, and for as long as it could it sought to downplay the extent to which Evidence-Indices differed in character from Evidence-Perspicuity. Even David Hume, in his essay ‘Of Miracles’ (1748), still slipped and slid between the two senses of the word.xiixii
I also want to stress two things that it is easy for us to overlook, so used are we to handling Evidence-Indices and finding them convincing. First, there have been plenty of systems of knowledge that have rejected Evidence-Indices and relied on something else – on geometrical proof, for example, or on signatures, or on the identification of concealed meaning (astrology, for instance). Evidence-Indices may always have been used in an unthinking way by people going about their daily business; but to elevate them into being a reliable basis for theoretical knowledge, as happened in England in the mid-seventeenth century, was to make a claim that was culturally peculiar and far from obviously true.
Second, it is not obvious even now that relying on Evidence-Indices was bound to be a winning strategy. The whole point of Hume’s formulation of the problem of induction is that we cannot explain why induction tends to work quite well, why nature seems extremely regular in its proceedings (or at least seems regular to us, who have been trained to look for regularities). Even if one intended to rely on Evidence-Indices, how can one tell what constitutes a good argument? Doctors treat a medicine as being of demonstrated effectiveness if the results of a trial are better than anything one would obtain by chance nineteen times out of twenty; nuclear physicists say there is evidence for something if there is no more than a one in 741 chance of a false positive result – they regard something as proved if the odds are 3.5 million to one. The early scientists did not even know how to perform a test for statistical significance.
The point about evidence is not that it was a natural type of argument to rely on, nor that it was a sort of argument that was obviously bound to be successful; the point is that relying on evidence just happened to work rather well. As Evidence-Indices replaced Evidence-Perspicuity the new scientists were able to claim more and more successes (the Puy-de-Dôme experiment, Boyle’s Law, Newton’s new thory of light), and these successes further boosted the attractions of Evidence-Indices. The intellectual apparatus of the new science – facts, experiments, theories, laws of nature, evidence – did not establish its worth by philosophical arguments; its success depended upon the fact that, in practice, it produced good results. There may be inhabited worlds in which no culture ever becomes evidence-based; and there could, for all we know, be universes in which looking for evidence simply doesn’t pay off – in which the sceptics not only win the arguments but have the facts on their side. Reality and the new science happened, in certain areas of physics, to dovetail quite neatly together. This was, in the end, good luck. Locke doubted whether our sensory capacities were sufficient to allow us to develop an adequate knowledge of corporeal substances.67 It turns out he was wrong, but he could easily have been right.