The Archeology of Knowledge Michel Foucault contents part I introduction 3

6. SCIENCE AND KNOWLEDGE

A silent delimitation has been imposed on all the preceding analyses, without the principle governing it, or even its outline, being made clear. All the examples referred to belonged without exception to a very small domain. In no way could I be said to have ^'covered^', let alone analysed, the immense domain of discourse: why did I systematically ignore ^'literary^', ^'philosophical^', or ^'political^' texts? Do not discursive formations and systems of positivities have a place in them too? And if I was restricting my attention to the sciences, why did I say nothing of mathematics, physics, or chemistry? Why did I concentrate on so many dubious, still imprecise disciplines that are perhaps doomed for ever to remain below the threshold of scientificity? In short, what is the relation between archaeology and the analysis of the sciences?

(a) POSITIVITIES, DISCIPLINES, SCIENCES

First question: does not archaeology, under the rather bizarre terms of 'discursive formation^' and ^'positivity^', describe what are quite simply pseudo-sciences (like psychopathology), sciences at the prehistoric stage (like Natural History), or sciences entirely penetrated with ideol­ogy (like political economy)? Is it not the privileged analysis of what will always remain quasi-scientific? If one calls ^'disciplines^' groups of

((197))

statements that borrow their organization from scientific models, which tend to coherence and demonstrativity, which are accepted, institutionalized, transmitted, and sometimes taught as sciences, could one not say that archaeology describes disciplines that are not really sciences, while epistemology describes sciences that have been formed on the basis of (or in spite of) existing disciplines?

To these questions I can reply in the negative. Archaeology does not describe disciplines. At most, such disciplines may, in their manifest deployment, serve as starting-points for the description of positivities; but they do not fix its limits: they do not impose definitive divisions upon it; at the end of the analysis they do not re-emerge in the same state in which they entered it; one cannot establish a bi-univocal relation between established disciplines and discursive formations.

Let us take an example of this distortion. The linch-pin of Madness and Civilization was the appearance at the beginning of the nineteenth cen­tury of a psychiatric discipline. This discipline had neither the same content, nor the same internal organization, nor the same place in medicine, nor the same practical function, nor the same methods as the traditional chapter on ^{`</sup>diseases of the head<sup>'</sup> or <sup>`}nervous diseases^' to be found in eighteenth-century medical treatises. But on examining this new discipline, we discovered two things: what made it possible at the time it appeared, what brought about this great change in the economy of concepts, analyses, and demonstrations, was a whole set of relations between hospitalization, internment, the conditions and procedures of social exclusion, the rules of jurisprudence, the norms of industrial labour and bourgeois morality, in short a whole group of relations that characterized for this discursive practice the formation of its state­ments; but this practice is not only manifested in a discipline possess-ing a scientific status and scientific pretensions; it is also found in operation in legal texts, in literature, in philosophy, in political decisions, and in the statements made and the opinions expressed in daily life. The discursive formation whose existence was mapped by the psychiatric discipline was not coextensive with it, far from it: it went well beyond the boundaries of psychiatry. Moreover, by going back in time and trying to discover what, in the seventeenth and eight­eenth centuries, could have preceded the establishment of psychiatry, we realized that there was no such prior discipline: what had been said

((198))

on the subject of mania, delirium, melancholia, and nervous diseases by the doctors of the Classical period in no way constituted. an autonomous discipline, but at most a commentary on the analysis of fevers, of alterations in the humours, or of affections of the brain. However, despite the absence of any established discipline, a discursive practice, with its own regularity and consistency, was in operation. This dis-cursive practice was certainly present in medicine, but it was also to be found in administrative regulations, in literary or philosophical texts, in casuistics, in the theories or projects of obligatory labour or assistance to the poor. In the Classical period, therefore, there were a discursive formation and a positivity perfectly accessible to description, to which corresponded no definite discipline that could be compared with psychiatry.

But although it is true that positivities are not merely the doublets of established disciplines, are they not the prototypes of future sciences? By discursive formation, does one not mean the retrospective projec­tion of sciences on their own past, the shadow that they cast on what preceded them and which thus appears to have foreshadowed them? What we have described, for example, as the Analysis of Wealth or General Grammar, thus according them what was perhaps a highly artificial autonomy, was it not, quite simply, political economy in an inchoate state, or a stage prior to the establishment of a truly rigorous science of language? Is it archaeology trying, by means of a retrograde movement whose legitimacy it would no doubt be difficult to estab­lish, to regroup in an independent discursive practice all the hetero­geneous and dispersed elements whose complicity will prove to he necessary to the establishment of a science?

Again, the answer must be in the negative. What was analysed under the name of Natural History does not embrace, in a single figure, everything that in the seventeenth and eighteenth centuries might validly constitute a prototype of the science of life, and figure in its legitimate genealogy. The positivity thus revealed accounts for a num-ber of statements concerning the resemblances and differences between beings, their visible structure, their specific and generic char­acters, their possible classification, the discontinuities that separate them, and the transitions that connect them; but it ignores a number of other analyses that date nevertheless from the same period, and which

((199))

also outline the ancestral figures of biology: the analysis of reflex movement (which was to have so much importance in the constitution of an anatomo-physiology of the nervous system), the theory of germs (which seems to anticipate the problems of evolution and genetics), the explanation of animal or vegetal growth (which was to be one of the major questions of the physiology of organisms in general). More-over: far from anticipating a future biology, Natural History — a taxo­nomic discourse, linked to the theory of signs and to the project of a science of order — excluded, by its solidity and autonomy, the constitu­tion of a unitary science of life. Similarly, the discursive formation described as General Grammar does not take into account — far from it — everything that could have been said about language in the Classical period, and of which the inheritance or repudiation, development or critique, was to be found later in philology: it ignored the methods of Biblical exegesis, and that philosophy of language as formulated by Vico or Herder. Discursive formations are not, therefore, future sci­ences at the stage at which, still unconscious of themselves, they are quietly being constituted: they are not, in fact, in a state of teleological subordination in relation to the orthogenesis of the sciences.

Should it be said, therefore, that there can he no science where there is a positivity, and that positivities are always exclusive of the sciences? Should it be supposed that instead of being in a chronological relation to the sciences, they are in fact alternatives? That they are, in a way, the positive figure of a certain epistemological defect. But here, too, one could find a counter-example. Clinical medicine is certainly not a sci­ence. Not only because it does not comply with the formal criteria, or attain the level of rigour expected of physics, chemistry, or even of physiology; but also because it involves a scarcely organized mass of empirical. observations, uncontrolled experiments and results, thera­peutic prescriptions, and institutional regulations. And yet this non-science is not exclusive of science: in the course of the nineteenth century, it established definite relations between such perfectly consti­tuted sciences as physiology, chemistry, or microbiology; moreover, it gave rise to such discourses as that of morbid anatomy, which it would be presumptuous no doubt to call a false science.

Discursive formations can be identified, therefore, neither as sci­ences, nor as scarcely scientific disciplines, nor as distant prefigurations

((200))

of the sciences to come, nor as forms that exclude any scientificity from the outset. What, therefore, is the relation between the positivities and the sciences?

(b) KNOWLEDGE (SAVOIR)

Positivities do not characterize forms of knowledge — whether they are a priori, necessary conditions or forms of rationality that have, in turn, been put into operation by history. But neither do they define the state of knowledge at a given moment in time: they do not draw up a list of what, from that moment, had been demonstrated to be true and had assumed the status of definitively acquired knowledge, and a list of what, on the other hand, had been accepted without either proof or adequate demonstration, or of what had been accepted as a common belief or a belief demanded by the power of the imagination. To analyse positivities is to show in accordance with which rules a discursive practice may form groups of objects, enunciations, concepts, or theor­etical choices. The elements thus formed do not constitute a science, with a defined structure of ideality; their system of relations is certainly less strict; but neither are they items of knowledge piled up one on top of another, derived from heterogeneous experiments, traditions, or discoveries, and linked only by the identity of the subject that possesses them. They are that on the basis of which coherent (or incoherent) propositions are built up, more or less exact descriptions developed, verifications carried out, theories deployed. They form the precondi­tion of what is later revealed and which later functions as an item of knowledge or an illusion, an accepted truth or an exposed error, a definitive acquisition or an obstacle surmounted. This precondition may not, of course, be analysed as a donee, a lived experience, still implicated in the imagination or in perception, which manking in the course of its history took up again in the form of rationality, or which each individual must undergo on his own account if he wishes to rediscover the ideal meanings that are contained or concealed within it. It is not a pre-knowledge or an archaic stage in the movement that leads from immediate knowledge to apodicticity; it is a group of elem­ents that would have to be formed by a discursive practice if a scientific discourse was to be constituted, specified not only by its form and

((201))

rigour, but also by the objects with which it deals, the types of enunci­ation that it uses, the concepts that it manipulates, and the strategies that it employs. Thus science is not linked with that which must have been lived, or must be lived, if the intention of ideality proper to it is to he established; but with that which must have been said — or must be said — if a discourse is to exist that complies, if necessary, with the experimental or formal criteria of scientificity.

This group of elements, formed in a regular manner by a discursive practice; and which are indispensable to the constitution of a science, although they are not necessarily destined to give rise to one, can he called knowledge. Knowledge is that of which one can speak in a dis-cursive practice, and which is specified by that fact: the domain consti­tuted by the different objects that will or will not acquire a scientific status (the knowledge of psychiatry in the nineteenth century is not the sum of what was thought to he true, but the whole set of practices, singularities, and deviations of which one could speak in psychiatric discourse); knowledge is also the space in which the subject may take up a position and speak of the objects with which he deals in his discourse (in this sense, the knowledge of clinical medicine is the whole group of functions of observation, interrogation, decipherment, recording, and decision that may be exercised by the subject of medical discourse); knowledge is also the field of coordination and subordin­ation of statements in which concepts appear, and are defined, applied and transformed (at this level, the knowledge of Natural History, in the eighteenth century, is not the sum of what was said, but the whole set of modes and sites in accordance with which one can integrate each new statement with the already said); lastly, knowledge is defined by the possibilities of use and appropriation offered by discourse (thus, the knowledge of political economy, in the Classical period, is not the thesis of the different theses sustained, but the totality of its points of articulation on other discourses or on other practices that are not dis-cursive). There are bodies of knowledge that are independent of the sciences (which are neither their historical prototypes, nor their prac­tical by-products), but there is no knowledge without a particular discursive practice; and any discursive practice may he defined by the knowledge that it forms.

Instead of exploring the consciousness/knowledge (connaissance) /

((202))

science axis (which cannot escape subjectivity), archaeology explores the discursive practice/knowledge (savoir)/science axis.^' And whereas the history of ideas finds the point of balance of its analysis in the element of connaissance (and is thus forced, against its will, to encounter the transcendental interrogation), archaeology finds the point of bal­ance of its analysis in savoir — that is., in a domain in which the subject is necessarily situated and dependent, and can never figure as titular (either as a transcendental activity, or as empirical consciousness).

It is understandable in these conditions that we should distinguish carefully between scientific domains and archaeological territories: their articula­tion and their principles of organization are quite different. Only pro-positions that obey certain laws of construction belong to a domain of scientificity; affirmations that have the same meaning, that say the same thing, that are as true as they are, but which do not belong to the same systematicity, are excluded from this domain: what Diderot^'s Le Reve de d^'Alembert says about the development of species may well express certain of the concepts or certain of the scientific hypotheses of the period; it may even anticipate a future truth; it does not belong to the domain of scientificity of Natural History, but it does not belong to its archaeological territory, if at least one can discover in operation in it the same rules of formation as in Linnaeus, Buffon, Daubenton, or Jussieu. Archaeological territories may extend to ^'literary^' or 'philo­sophical^' texts, as well as scientific ones. Knowledge is to be found not only in demonstrations, it can also he found in fiction, reflexion, narra­tive accounts, institutional regulations, and political decisions. The archaeological territory of Natural History includes Bonnet^'s Palingenesie philosophique or Benoit de Maillet^'s Telliamed, although they do not comply to a great extent with the accepted scientific norms of the period, and even less, of course, with those that came to be required later. The archaeological territory of General Grammar embraces the imaginings of Fabre d^'Olivet (which were never accorded scientific status, and belong rather to the sphere of mystical thought) no less than the analy­sis of attributive propositions (which was then accepted as evident truth, and in which generative grammar may now recognize its prefigured truth).
((footnote))

' For the distinction between connaissance and savoir, cf. note 3, p. 16.

((203))

Discursive practice does not coincide with the scientific develop-ment that it may give rise to; and the knowledge that it forms is neither an unfinished prototype nor the by-product to be found in daily life of a constituted science. The sciences — ignoring, for the moment, the difference between discourses that have the status of scientificity, or pretensions to it, and those that really present the formal criteria of a science — appear in the element of a discursive formation and against the background of knowledge. This opens up two series of problems: what can be the place or role of a region of scientificity in the archaeo­logical territory in which it appears? In accordance with what order and what processes is the emergence of a region of scientificity in a given discursive formation accompished? We cannot, at present, pro-vide solutions to these problems: all we can do now is to indicate in what direction they might be analysed.

(c) KNOWLEDGE (SAVOIR) AND IDEOLOGY

Once constituted, a science does not take up, with all the intercon­nexions that are proper to it, everything that formed the discursive practice in which it appeared; nor does it dissipate — in order to con­demn it to the prehistory of error, prejudice, or imagination — the knowledge that surrounds it. Morbid anatomy did not reduce to the norms of scientificity the positivity of clinical medicine. Knowledge is not an epistemological site that disappears in the science that super­sedes it. Science (or what is offered as such) is localized in a field of knowledge and plays a role in it. A role that varies according to differ­ent discursive formations, and is modified with their mutations. What, in the Classical period, was offered as the medical knowledge of dis-eases of the mind occupied a very small place in the knowledge of madness: it constituted scarcely more than one of its many surfaces of contact (the others being jurisprudence, casuistics, police regulations, etc.); on the other hand, the psychopathological analyses of the nineteenth century, which were also offered as scientific knowledge (connaissance) of mental diseases, played a very different, much more important role in the knowledge (savoir) of madness (the role of model, and decision-making authority). Similarly, scientific discourse (or scientific pretension) does not perform the same function in the

((204))

economic knowledge of the seventeenth and in that of the nineteenth century. In any discursive formation, one finds a specific relation between science and knowledge; and instead of defining between them a relation of exclusion or subtraction (by trying to discover what in knowledge still eludes and resists science, what in science is still com­promised by its proximity to and the influence of knowledge), arch­aeological analysis must show positively how a science functions in the element of knowledge.

It is probably there, in that space of interplay, that the relations of ideology to the sciences are established. The hold of ideology over scientific discourse and the ideological functioning of the sciences are not articulated at the level of their ideal structure (even if they can be expressed in it in a more or less visible way), nor at the level of their technical use in a society (although that society may obtain results from it), nor at the level of the consciousness of the subjects that built it up; they are articulated where science is articulated upon knowledge. If the question of ideology may be asked of science, it is in so far as science, without being identified with knowledge, but without either effacing or excluding it, is localized in it, structures certain of its objects, systematizes certain of its enunciations, formalizes certain of its concepts and strategies; it is in so far as this development articulates knowledge, modifies it, and redistributes it on the one hand, and con-firms it and gives it validity on the other; it is in so far as science finds its place in a discursive regularity, in which, by that very fact, it is or is not deployed, functions or does not function, in a whole field of dis-cursive practices. In short, the question of ideology that is asked of science is not the question of situations or practices that it reflects more or less consciously; nor is it the question of the possible use or misuse to which it could be put; it is the question of its existence as a discursive practice and of its functioning among other practices.

Broadly speaking, and setting aside all mediation and specificity, it can be said that political economy has a role in capitalist society, that it serves the interests of the bourgeois class, that it was made by and for that class, and that it hears the mark of its origins even in its concepts and logical architecture; but any more precise description of the rela­tions between the epistemological structure of political economy and its ideological function must take into account the analysis of the

((205))

discursive formation that gave rise to it and the group of objects, con­cepts, and theoretical choices that it had to develop and systematize; and one must then show how the discursive practice that gave rise to such a positivity functioned among other practices that might have been of a discursive, but also of a political or economic, order.

This enables us to advance a number of propositions.
I. Ideology is not exclusive of scientificity. Few discourses have given so much place to ideology as clinical discourse or that of political economy: this is not a sufficiently good reason to treat the totality of their statements as being undermined by error, contradiction, and a lack of objectivity.

2. 
   Theoretical contradictions, lacunae, defects may indicate the ideological functioning of a science (or of a discourse with scientific pretensions); they may enable us to determine at what point in the structure this functioning takes effect. But the analysis of this function-ing must be made at the level of the positivity and of the relations between the rules of formation and the structures of scientificity.
   
3. 
   By correcting itself, by rectifying its errors, by clarifying its formulations, discourse does not necessarily undo its relations with ideology. The role of ideology does not diminish as rigour increases and error is dissipated.
   
4. 
   To tackle the ideological functioning of a science in order to reveal and to modify it is not to uncover the philosophical presupposi­tions that may lie within it; nor is it to return to the foundations that made it possible, and that legitimated it: it is to question it as a dis-cursive formation; it is to tackle not the formal contradictions of its propositions, but the system of formation of its objects, its types of enunciation, its concepts, its theoretical choices. It is to treat it as one practice among others.
   

(d) DIFFERENT THRESHOLDS AND THEIR CHRONOLOGY

It is possible to describe several distinct emergences of a discursive formation. The moment at which a discursive practice achieves individuality and autonomy, the moment therefore at which a single

((206))

system for the formation of statements is put into operation, or the moment at which this system is transformed, might be called the threshold of positivity. When in the operation of a discursive formation, a group of statements is articulated, claims to validate (even unsuccess­fully) norms of verification and coherence, and when it exercises a dominant function (as a model, a critique, or a verification) over knowledge, we will say that the discursive formation crosses a threshold of epistemologization. When the epistemological figure thus outlined obeys a number of formal criteria, when its statements comply not only with archaeological rules of formation, but also with certain laws for the construction of propositions, we will say that it has crossed a threshold of scientificity. And when this scientific discourse is able, in turn, to define the axioms necessary to it, the elements that it uses, the propositional structures that are legitimate to it, and the transformations that it accepts, when it is thus able, taking itself as a starting-point, to deploy the formal edifice that it constitutes, we will say that it has crossed the threshold of formalization.

The distribution in time of these different thresholds, their succes­sion, their possible coincidence (or lack of it), the way in which they may govern one another, or become implicated with one another, the conditions in which, in turn, they are established, constitute for archaeology one of its major domains of exploration. Their chron­ology, in fact, is neither regular nor homogeneous. The discursive formations do not cross them at regular intervals, or at the same time, thus dividing up the history of human knowledge (connaissances) into different ages; at a time when many positivities have crossed the threshold of formalization, many others have not yet attained that of scientificity, or even of epistemologization. Moreover: each discursive formation does not pass through these different thresholds in turn, as through the natural stages of biological maturation, in which the only variable is the latency period or the length of the intervals. They are, in fact, events whose dispersion is not evolutive: their unique order is one of the characteristics of each discursive formation. Here are a few examples of these differences.

In some cases, the threshold of positivity is crossed well before that of epistemologization: thus psychopathology, as a discourse with sci­entific pretensions, epistemologized at the beginning of the nineteenth

((207))

century, with Pinel, Heinroth, and Esquirol, a discursive practice that largely antedated it, and that had acquired its autonomy and system of regularity long before. But there are also cases in which these two stages are confused in time, when the establishment of a positivity involves at the same time the emergence of an epistemological figure. Sometimes the thresholds of scientificity are linked with the transition from one positivity to another; sometimes they are different; thus the transition from Natural History (with the scientificity that was proper to it) to biology (as a science not of the classification of beings, but of specific correlations of different organisms) did not take place at the time of Cuvier without the transformation of one positivity into another: on the other hand, the experimental medicine of Claude Bernard, then the microbiology of Pasteur, modified the type of scien­tificity required by morbid anatomy and physiology, without the dis-cursive formation of clinical medicine, as then established, being made inoperable. Similarly, the new scientificity established in the biological disciplines by evolutionism did not modify the biological positivity that had been defined at the time of Cuvier. In the case of economics the disconnexions are particularly numerous. In the seventeenth cen­tury, one can recognize a threshold of positivity: it almost coincides with the practice and theory of mercantilism; but its epistemologiza­tion did not occur until later, at the very end of the century, or the beginning of the next century, with Locke and Cantillon. However, the nineteenth century, with Ricardo, marks both a new type of positivity, a new form of epistemologization, which were later to he modified in turn by Cournot and Jevons, at the very time that Marx was to reveal an entirely new discursive practice on the basis of political economy.

If one recognizes in science only the linear accumulation of truths or the orthogenesis of reason, and fails to recognize in it a discursive practice that has its own levels, its own thresholds, its own various ruptures, one can describe only a single historical division, which one adopts as a model to be applied at all times and for all forms of knowledge: a division between what is definitively or what is not yet scientific. All the density of the disconnexions, the dispersion of the ruptures, the shifts in their effects, the play of the interdepend­ence are reduced to the monotonous act of an endlessly repeated foundation.

((208))

There is perhaps only one science for which one can neither dis­tinguish these different thresholds, nor describe a similar set of shifts: mathematics, the only discursive practice to have crossed at one and the same time the thresholds of positivity, epistemologization, scientifi­city, and formalization. The very possibility of its existence implied that which, in all other sciences, remains dispersed throughout history, should be given at the outset: its original positivity was to constitute an already formalized discursive practice (even if other formalizations were to be used later). Hence the fact that their establishment is both so enigmatic (so little accessible to analysis, so confined within the form of the absolute beginning) and so valid (since it is valid both as an origin and as a foundation); hence the fact that in the first gesture of the first mathematician one saw the constitution of an ideality that has been deployed throughout history, and has been questioned only to be repeated and purified; hence the fact that the beginning of mathematics is questioned not so much as a historical event as for its validity as a principal of history: and hence the fact that, for all the other sciences the description of its historical genesis, its gropings and failures, its late emergence is related to the meta-historical model of a geometry emerging suddenly, once and for all, from the trivial practices of land-measuring. But if one takes the establishment of mathematical discourse as a prototype for the birth and development of all the other sciences, one runs the risk of homogenizing all the unique forms of historicity, of reducing to the authority of a single rupture all the different thresholds that a discursive practice may cross, and reproduce endlessly, at every moment in time, the problem of origin: the rights of the historicotranscendental analysis would thus be reinstated. Math­ematics has certainly served as a model for most scientific discourses in their efforts to attain formal rigour and demonstrativity; but for the historian who questions the actual development of the sciences, it is a bad example, an example at least from which one cannot generalize.

(e) THE DIFFERENT TYPES OF THE HISTORY OF THE SCIENCES

The multiple thresholds that we have succeeded in mapping make

distinct forms of historical analysis possible. First, analysis at the level

((209))

of formalization: it is this history that mathematics never ceases to recount about itself in the process of its own development. What it possesses at a given moment (its domain, its methods, the objects that it defines, the language that it employs) is never thrown back into the external field of non-scientificity, but is constantly undergoing redefinition (if only as an area that has fallen into disuse or temporary sterility) in the formal structure that mathematics constitutes; this past is revealed as a particular case, a naive model, a partial and insuffi­ciently generalized sketch, of a more abstract, or more powerful the­ory, or one existing at a higher level; mathematics retranscribes its real historical trajectory into the vocabulary of vicinities, dependences, subordinations, progressive formalizations, and self-enveloping gener­alities. For this history of mathematics (the history that is constituted by mathematics itself and which mathematics recounts about itself), the algebra of Diophantus is not an experience that remains in sus-pense; it is a particular case of Algebra as we have known it since Abel and Galois; the Greek method of exhaustions was not an impasse that had to be escaped from; it is a naive model of integral calculus. Each historical event has its own formal level and localization. This is a recurrential analysis, which can be carried out only within a constituted science, one that has crossed its threshold of formalization.^'

The second type of historical analysis is situated at the threshold of scientificity, and questions itself as to the way in which it was crossed on the basis of various epistemological figures. Its purpose is to dis-cover, for example, how a concept — still overlaid with metaphors or imaginary contents — was purified, and accorded the status and func­tion of a scientific concept. To discover how a region of experience that has already been mapped, already partially articulated, but is still over-laid with immediate practical uses or values related to those uses, was constituted as a scientific domain. To discover how, in general, a science was established over and against a pre-scientific level, which both paved the way and resisted it in advance, how it succeeded in overcom­ing the obstacles and limitations that still stood in its way. G. Bachelard and G. Canguilhem have provided models of this kind of history. Unlike recurrential analysis, it has no need to situate itself within the
((footnote))

Michel Serres, Hermes ou la communication, p. 78.

((210))

science itself, to redistribute every episode in its construction, to recount its formalization in the formal vocabulary that it still possesses today: indeed, how could it do so, since it shows what the science has freed itself from, everything that it has had to leave behind in its pro-gress towards the threshold of scientificity. Consequently, this descrip­tion takes as its norm the fully constituted science; the history that it recounts is necessarily concerned with the opposition of truth and error, the rational and the irrational, the obstacle and fecundity, purity and impurity, the scientific and the non-scientific. It is an epistemological history of the sciences.

The third type of historical analysis takes as its point of attack the threshold of epistemologization — the point of cleavage between dis-cursive formations defined by their positivity and epistemological fig­ures that are not necessarily all sciences (and which may never, in fact, succeed in becoming sciences). At this level, scientificity does not serve as a norm: in this archaeological history, what one is trying to uncover are discursive practices in so far as they give rise to a corpus of knowledge, in so far as they assume the status and role of a science. To undertake a history of the sciences at this level is not to describe discursive forma­tions without regard to epistemological structures; it is to show how the establishment of a science, and perhaps its transition to formaliza­tion, have come about in a discursive formation, and in modifications to its positivity. Such an analysis sets out, therefore, to outline the history of the sciences on the basis of a description of discursive prac­tices; to define how, in accordance with which regularity, and as a result of which modification, it was able to give rise to the processes of epistemologization, to attain the norms of scientificity, and, perhaps, to reach the threshold of formalization. In seeking the level of discursive practice in the historical density of the sciences, one is not trying to place the discursive practice at some deep, original level, one is not trying to place it at the level of lived experience (on this earth, which is given, irregular and fragmented, before all geometry; in the heaven that glitters through the grid of all astronomies) ; one is trying to reveal between positivities, knowledge, epistemological figures, and sciences, a whole set of differences, relations, gaps, shifts, independences, auton­omies, and the way in which they articulate their own historicities on one another.

((211))

The analysis of discursive formations, of positivities, and knowledge in their relations with epistemological figures and with the sciences is what has been called, to distinguish it from other possible forms of the history of the sciences, the analysis of the episteme. This episteme may be suspected of being something like a world-view, a slice of history common to all branches of knowledge, which imposes on each one the same norms and postulates, a general stage of reason, a certain struc­ture of thought that the men of a particular period cannot escape — a great body of legislation written once and for all by some anonymous hand. By episteme, we mean, in fact, the total set of relations that unite, at a given period, the discursive practices that give rise to epistemological figures, sciences, and possibly formalized systems; the way in which, in each of these discursive formations, the transitions to epistemologiza­tion, scientificity, and formalization are situated and operate; the dis­tribution of these thresholds, which may coincide, be subordinated to one another, or be separated by shifts in time; the lateral relations that may exist between epistemological figures or sciences in so far as they belong to neighbouring, but distinct, discursive practices. The epis­teme is not a form of knowledge (connaissance) or type of rationality which, crossing the boundaries of the most varied sciences, manifests the sovereign unity of a subject, a spirit, or a period; it is the totality of relations that can be discovered, for a given period, between the sciences when one analyses them at the level of discursive regularities.

The description of the episteme presents several essential. character­istics therefore: it opens up an inexhaustible field and can never be closed; its aim is not to reconstitute the system of postulates that gov-erns all the branches of knowledge (connaissances) of a given period, but to cover an indefinite field of relations. Moreover, the episteme is not a motionless figure that appeared one day with the mission of effacing all that preceded it: it is a constantly moving set of articulations, shifts, and coincidences that are established, only to give rise to others. As a set of relations between sciences, epistemological figures, positivities, and discursive practices, the episteme makes it possible to grasp the set of constraints and limitations which, at a given moment, are imposed on discourse: but this limitation is not the negative limitation that opposes knowledge (connaissance) to ignorance, reasoning to imagination, armed experience to fidelity to appearances, and fantasy to inferences and

((212))

deductions; the episteme is not what may be known at a given period, due account taken of inadequate techniques, mental attitudes, or the limitations imposed by tradition; it is what, in the positivity of discursive practices, makes possible the existence of epistemological figures and sciences. Lastly, we see that the analysis of the episteme is not a way of returning to the critical question (^'given the existence of something like a science, what is its legitimacy?^'); it is a questioning that accepts the fact of science only in order to ask the question what it is for that science to be a science. In the enigma of scientific discourse, what the analysis of the episteme questions is not its right to be a science, but the fact that it exists. And the point at which it separates itself off from all the philosophies of knowledge (connaissance) is that it relates this fact not to the authority of an original act of giving, which establishes in a transcendental subject the fact and the right, but to the processes of a historical practice.

(f) OTHER ARCHAEOLOG I ES

One question remains in suspense: could one conceive of an archaeo­logical analysis that would reveal the regularity of a body of knowledge, but which would not set out to analyse it in terms of epistemological figures and sciences? Is an orientation towards the episteme the only one open to archaeology? Must archaeology be — exclusively — a certain way of questioning the history of the sciences? In other words, by confirming itself up to now to the region of scien­tific discourses, has archaeology been governed by some insuperable necessity — or has it provided an outline, on the basis of a particular example, of forms of analysis that may have a much wider application?

At the moment I am not sufficiently advanced in my task to answer this question. But I can readily imagine — subject to a great deal of further exploration and examination — archaeologies that might develop in different directions. There is, for example, the archaeo­logical description of ^'sexuality^'. And I can see very well how it might be orientated towards the episteme: one would show how in the nine­teenth century such epistemological figures as the biology and psych­ology of sexuality were formed; and how a discourse of a scientific type was established through the rupture brought about by Freud. But I

((213))

can also see another possible direction for analysis: instead of studying the sexual behaviour of men at a given period (by seeking its law in a social structure, in a collective unconscious, or in a certain moral attitude), instead of describing what men thought of sexuality (what religious interpretation they gave it, to what extent they approved or disapproved of it, what conflicts of opinion or morality it gave rise to), one would ask oneself whether, in this behaviour, as in these represen­tations, a whole discursive practice is not at work; whether sexuality, quite apart from any orientation towards a scientific discourse, is not a group of objects that can be talked about (or that it is forbidden to talk about), a field of possible enunciations (whether in lyrical or legal language), a group of concepts (which can no doubt be presented in the elementary form of notions or themes), a set of choices (which may appear in the coherence of behavior or in systems of prescription). • Such an archaeology would show, if it succeeded in its task, how the prohibitions, exclusions, limitations, values, freedoms, and transgres­sions of sexuality, all its manifestations, verbal or otherwise, are linked to a particular discursive practice. It would reveal, not of course as the ultimate truth of sexuality, but as one of the dimensions in accordance with which one can describe it, a certain ^'way of speaking^'; and one would show how this way of speaking is invested not in scientific discourses, but in a system of prohibitions and values. An analysis that would be carried out not in the direction of the episteme, but in that of what we might call the ethical.

But here is an example of another possible orientation. In analysing a painting, one can reconstitute the latent discourse of the painter; one can try to recapture the murmur of his intentions, which are not tran-scribed into words, but into lines, surfaces, and colours; one can try to uncover the implicit philosophy that is supposed to form his view of the world. It is also possible to question science, or at least the opinions of the period, and to try to recognize to what extent they appear in the painter^'s work. Archaeological analysis would have another aim: it would try to discover whether space, distance, depth, colour, light, proportions, volumes, and contours were not, at the period in question, considered, named, enunciated, and conceptualized in a discursive practice; and whether the knowledge that this discursive practice gives rise to was not embodied perhaps in theories and

((214))

speculations, in forms of teaching and codes of practice, but also in processes, techniques, and even in the very gesture of the painter. It would not set out to show that the painting is a certain way of ^'mean-ing^' or ^'saying^' that is peculiar in that it dispenses with words. It would try to show that, at least in one of its dimensions, it is discursive practice that is embodied in techniques and effects. In this sense, the painting is not a pure vision that must then he transcribed into the materiality of space; nor is it a naked gesture whose silent and eternally empty meanings must be freed from subsequent interpretations. It is shot through — and independently of scientific knowledge (connaissance) and philosophical themes — with the positivity of a knowledge (savoir).

It seems to me that one might also carry out an analysis of the same type on political knowledge. One would try to show whether the polit­ical behavior of a society, a group, or a class is not shot through with a particular, describable discursive practice. This positivity would obvi­ously not coincide either with the political theories of the period or with economic determinations: it would define the element in politics that can become an object of enunciation, the forms that this enunci­ation may take, the concepts that are employed in it, and the strategic choices that are made in it. Instead of analysing this knowledge — which is always possible — in the direction of the episteme that it can give rise to, one would analyse it in the direction of behaviour, strug­gles, conflicts, decisions, and tactics. One would thus reveal a body of political knowledge that is not some kind of secondary theorizing about practice, nor the application of theory. Since it is regularly formed by a discursive practice that is deployed among other practices and is articulated upon them, it is not an expression that more or less adequately ^'reflects^' a number of ^'objective data^' or real practices. It is inscribed, from the outset, in the field of different practices in which it finds its specificity, its functions, and its network of dependences. If such a description were possible, there would be no need of course to pass through the authority of an individual or collective consciousness in order to grasp the place of articulation of a political practice and theory; there would he no need to try to discover to what extent this consciousness may, on the one hand, express silent conditions, and, on the other, show that it is susceptible to theoretical truths; one would not need to pose the psychological problem of an act of consciousness
((215))

(prise de conscience); instead, one would analyse the formation and trans-formations of a body of knowledge. The question, for example, would not be to determine from what moment a revolutionary consciousness appears, nor the respective roles of economic conditions and theor­etical elucidations in the genesis of this consciousness; it would not attempt to retrace the general, and exemplary, biography of revo­lutionary man, or to find the origins of his project; but it would try to explain the formation of a discursive practice and a body of revolution­ary knowledge that are expressed in behaviour and strategies, which give rise to a theory of society, and which operate the interference and mutual transformation of that behaviour and those strategies.

To the questions posed above — Is archaeology concerned only with sciences? Is it always an analysis of scientific discourse? — we can now give a reply, in each case in the negative. What archaeology tries to describe is not the specific structure of science, but the very different domain of knowledge. Moreover, although it is concerned with know-ledge in its relation to epistemological figures and the sciences, it may also question knowledge in a different direction and describe it in a different set of relations. The orientation towards the episteme has been the only one to be explored so far. The reason for this is that, because of a gradient that no doubt characterizes our cultures, dis-cursive formations are constantly becoming epistemologized. It is by questioning the sciences, their history, their strange unity, their disper­sion, and their ruptures, that the domain of positivities was able to appear; it is in the interstice of scientific discourses that we were able to grasp the play of discursive formations. It is hardly surprising, there-fore, that the most fruitful region, the one most open to archaeological description should have been that ^'Classical^' age, which from the Renaissance to the nineteenth century saw the epistemologization of so many positivities; nor is it surprising that the discursive formations and specific regularities of knowledge are outlined precisely where the levels of scientificity and formalization were most difficult to attain. But that was no more than a preferential point of attack; it is not, for archaeology, an obligatory domain.

((219))