DATE DOWNLOADED: Mon Jul 15 12:54:31 2024

SOURCE: Content Downloaded from HeinOnline

Citations:
Please note: citations are provided as a general guideline. Users should consult their preferred
citation format's style manual for proper citation formatting.

Bluebook 21st ed.

Niklas Luhmann, Operational Closure and Structural Coupling: The Differentiation of
the Legal System, 13 CARDOZO L. REV. 1419 (1991).

ALWD 7th ed.

Niklas Luhmann, Operational Closure and Structural Coupling: The Differentiation of
the Legal System, 13 Cardozo L. Rev. 1419 (1991).

APA 7th ed.

Luhmann, Niklas. (1991). Operational closure and structural coupling: the
differentiation of the legal system. Cardozo Law Review, 13(5), 1419-1442.

Chicago 17th ed.

Niklas Luhmann, "Operational Closure and Structural Coupling: The Differentiation of
the Legal System," Cardozo Law Review 13, no. 5 (March 1992): 1419-1442

McGill Guide 9th ed.

Niklas Luhmann, "Operational Closure and Structural Coupling: The Differentiation of
the Legal System" (1991) 13:5 Cardozo L Rev 1419.

AGLC 4th ed.

Niklas Luhmann, 'Operational Closure and Structural Coupling: The Differentiation of
the Legal System' (1991) 13(5) Cardozo Law Review 1419

MLA 9th ed.

Luhmann, Niklas. "Operational Closure and Structural Coupling: The Differentiation of
the Legal System." Cardozo Law Review, vol. 13, no. 5, March 1992, pp. 1419-1442.
HeinOnline.

OSCOLA 4th ed.

Niklas Luhmann, 'Operational Closure and Structural Coupling: The Differentiation of
the Legal System' (1991) 13 Cardozo L Rev 1419 Please note:
citations are provided as a general guideline. Users should consult their preferred
citation format's style manual for proper citation formatting.

-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and
Conditions of the license agreement available at
https://heinonline.org/HOL/License
-- The search text of this PDF is generated from uncorrected OCR text.
-- To obtain permission to use this article beyond the scope of your license, please use:
Copyright Information
 OPERATIONAL CLOSURE AND STRUCTURAL
COUPLING: THE DIFFERENTIATION OF
THE LEGAL SYSTEM

Niklas Luhmann *

I.

Mainstream systems theory conceives of systems as "open sys-

tems" feeding upon exchanges with their environment. These ex-
changes can be structured and limited to the needs of the system by
restricting it to "inputs" and "outputs." This concept of open sys-
tems has been the answer, first from biologists and then from sociolo-
gists, to the laws of thermodynamics and to the probability of decay
in the sense of vanishing differences. The problem was how to avoid
this probability-that is, how to reverse this tendency towards en-
tropy-and above all, how to explain a world which is obviously able
to build up order and to maintain negentropy. Since thermodynamic
law presupposes closed systems, open systems seemed to be the solu-
tion. However, an annoying question remains: What is a system? Or
more to the point: How does a system operate so that it can be called
an "open system"? Or, if a system depends upon its environment,
what prevents a growing confusion between system and environment
and, again, thermodynamic decay?
The earliest response, fashionable during the 1950s and accepted,
for instance, by Talcott Parsons, was to formulate the problem as one
of system maintenance or boundary maintenance.' The answer was
that a system that cannot fulfill the requisite of system maintenance
will dissolve itself and will no longer be available for investigation.
This led to a revival of teleology (the new brand name was functional-
ism) and again, as we know from the famous Moliresque calming
power of opium, to a tautological reformulation of the problem.
Hence, we stay with our question: What kind of operations produce
the unity of the system with the effect that it can maintain itself, that
is, continue to operate and maintain connections to its environment?
How are we to understand that there is something that can select in-
puts and outputs and survive irritating information about its environ-
ment according to internal needs?

* Professor of Sociology, Bielefeld University.

1 See TOWARD A GENERAL THEORY OF ACTION 108-09 (Talcott Parsons & Edward A.
Shils eds. 1951).

1419
1420 CARDOZO LAW REVIEW [Vol. 13:1419

Recalling all these attempts to avoid the fate of entropy, the the-
ory of autopoietic, self-reproducing systems does not reinvent the idea
of complete self-causation in empirical isolation. 2 This should be ob-
vious. The description of a system as autopoietic, as autonomous, as
operationally closed, refers to the network of its operations and not to
the totality of all empirical conditions, that is, the world. The ques-
tion is not how a system can maintain itself without any environmen-
tal support. Rather, it is what kind of operations enable a system to
form a self-reproducing network which relies exclusively on self-gen-
erated information and is capable of distinguishing internal needs
from what it sees as environmental problems.
The answer is no longer a disguised tautology but an obvious
one. The unity of the system is produced by the system itself.3 The
methodological task becomes to "unfold" (as logicians would say) this
tautology, and this has to be done by empirically identifying the oper-
ations which produce and reproduce the unity of the system (for ex-
ample, the biochemical processes within a living cell).
It would be easy to advance from here by distinguishing between
subjects and objects, or scientific perspectives and realities. But let us
first take another look at the question of how a system produces its
own unity. Asking "how" leads to the further question of who asks
the question; or, what are the systemic conditions of asking this ques-
tion; or simply, who is the observer? Is it the system itself or some
external unit? Perhaps it is the famous subject or simply another
system?
This technique of reflexive questioning dissolves-or should I
say, deconstructs?-the distinction between subject and object and
leaves us with the assumption of "observing systems" in the double
sense of the English -ing form.4 We ourselves may be observing sys-
tems observing observing systems; and our question remains. Are
self-maintaining, self-reproducing systems necessarily observing sys-
tems, able to distinguish themselves from their environment? How
can they maintain themselves if they cannot distinguish themselves
and what is the operational basis for this self-observation? Is the

2 Objections which focus on this point reject a theory that nobody proposes. See, e.g.,
MICHEL VAN DE KERCHOVE & FRANCOIS OST, LE SYSTtME JURIDIQUE ENTRE ORDRE ET
DtSORDRE 154-59 (1988).
3 The same tautology appears in evolutionary theories. Systems survive by adapting
themselves to their environment, and adaptation is proven by survival. The problem is the
solution because the solution is the problem. And here, too, the tautology has to be broken up
by empirically distinguishing between the internal operations that change the systems and the
resulting survival. We have to admit, therefore, the possibility that maladapted systems will
survive.
4 See HEINZ VON FOERSTER, OBSERVING SYSTEMS (1981).
1992] CLOSURE & STRUCTURAL COUPLING 1421

classical problem of reference (as a condition of meaning and truth)

itself a meaningful question, or must it be transformed into a question
of how to distinguish between self-reference and external reference?
If so, what kind of systems are able to make such a distinction and by
what kind of operation? The distinction between self-reference and
external reference is by necessity an internal distinction. But logi-
cally, how can the distinction between internal and external be an
internal distinction? And empirically, what kind of operation pro-
duces and reproduces this distinction within the system and thereby
solves the logical puzzle (whereas logicians would need to look for
higher logical types or levels to unfold the problem)?
All this may convey the impression of being unnecessarily com-
plicated and of epistemological interest only. But my intention is to
prepare the ground for understanding the concept of operational (or
systemic) closure, and I have to deconstruct all kinds of epistemologi-
cal obstacles first so that you may see the point. We could take the
route of Ferdinand de Saussure and Jacques Derrida,6 or George
Spencer Brown,7 and follow the injunction to always start with differ-
ence and not identity, with distinction and not unity. But we could
also accept the results of empirical sciences, namely cognitive sciences
and systems theory, which dissolve the classical privileges of ontology
and epistemology by organizing questions and answers.
Our first question will always be: Who is the observer; or, what is
the system that distinguishes itself from its environment? The con-
cept of operational closure will provide an answer to this question. Of
course, as a single concept, it will not "explain" anything, neither the
state of the system nor its structural drift. The same holds true for
the concept of autopoiesis.8 But any explanatory theory will have to
describe how systems cope with the problems of closure and openness,
and this cannot be done in causal terms. If you continue to mix up
questions of causal isolation with questions of operational closure--or
in more abstract terms, questions of causality and questions of refer-
ence-you may observe other things with other instruments, but you
will not follow this article.

5 See generally FERDINAND DE SAUSSURE, COURSE IN GENERAL LINGUISTICS (Charles

Bally & Albert Sechehaye eds., Roy Harris trans., 1983).
6 JACQUES DERRIDA, WRITING AND DIFFERENCE 154-68 (Alan Bass trans., 1978).

7 GEORGE S. BROWN, LAWS OF FORM (1979).

8 This impressive neologism in particular seems to mobilize exaggerated expectations
which then are easily refuted. See, for example, the objections of BERND MARIN, CON-
TRACTING WITHOUT CONTRACTS-ECONOMIC POLICY CONCERTATION BY AUTOPOIETIC
REGIMES BEYOND LAW? (European University Institute, Working Paper No. 87/287, 1987).
1422 CARDOZO LAW REVIEW [Vol. 13:1419

II.
We are now ready to start on our way through the legal system,
that is, ready to meet the first obstacle. The first surprise will be that
the famous protagonists of operational closure at the level of general
systems theory strongly reject its application to social systems. Heinz
Von Foerster finds this idea shocking; 9 nor would Humberto
Maturana and Francisco Varela join in calling social systems auto-
poietic systems. 10
Why not? The answer is easy. Not being sociologists, these au-
thors think of social systems as consisting of concrete people, individ-
uals with bodies and minds. Of course, it is impossible to admit the
closure of social systems which include the reproduction of molecules
in cells, or of cells in bodies, or of thoughts in minds, as social opera-
tions within the social system." But this mistake simply points to
unexplored possibilities of clarification. We have to be very precise in
defining the type of operation which reproduces, within a closed net-
work of its own productions, the unity of a social system. This opera-
tion can never be defined as the biochemical production of life, nor
can it be defined as the reproduction of thoughts by thoughts within
the internal darkness of a conscious system.' 2 The operation we are
looking for can only be communication.
In fact, the theory of autopoietic systems could bear the title
Taking Individuals Seriously, certainly more seriously than our hu-
manistic tradition. Taken as an individual, no human being can be
part of any other systems. Critics frequently miss this point. Arthur
Jacobson, for one, states that certainly the common law (if not the
continental law) includes "[tihe role and needs of the individual."13
But actually, the position of the individual in both versions does not
differ at all. Jacobson goes on to say that:

9 Heinz von Foerster expressed this view in the context of an interview, Gdndalogies de
l'auto-organisation, 8 CAHIERS DU CENTRE DE RECHERCHES ET D'UTUDES APPLIQUES 263
(1985).
10 See Humberto R. Maturana, Biologie der Sozialitdt, 5 DELFIN 6 (1985).
I1 Even the political reasons which make the idea unacceptable are obvious. Maturana one
day objected to the idea that a social system like Chile's under the rule of Pinochet produces
and reproduces individuals by its own structure and process.
12 Almost a Hegel citation. Hegel calls it "diefinstere Innerlichkeit des Gedankens." GE-
ORG W. F. HEGEL, Vorlesungen uber die Asthetik, in 13 WERKE 18 (1970).
13 See Arthur J. Jacobson, Autopoietic Law: The New Science of Niklas Luhmann, 87

MICH. L. REV. 1647, 1683 (1989). Compare Christophe Grzegorczyk, Systime Juridique et
Rialiti: Discussion de la Thiorie Autopoidtique du Droit, in CONTROVERSES AUTOUR DE
L'ONTOLOGIE DU DROIT 179 (Paul Amselek & Christophe Grzegorczyk eds., 1989) with the
comments of Gunther Teubner, How the Law Thinks: Toward a ConstructivistEpistemology of
Law, 23 LAW & Soc'Y REV. 727, 739-41 (1989).
19921 CLOSURE & STRUCTURAL COUPLING 1423

Individuals figure in the common law only in the character they

display through interaction oriented toward the values expressed in
prior applications of norms. The individuals applying norms may
have hosts of attitudes (personality, emotion) toward the applica-
tion. The attitudes do not matter: Only the display of character in
interaction matters. 14
This can be translated as: Only communication matters because legal
systems are social systems!
But why communication? Communication is one of those rare
types of operation that has the capacity to solve the riddle of self-
transcendence.15 Seen as an operation, it cannot leave the society as
the system that integrates all communications, for no system can op-
erate outside of its own boundaries, and communication cannot be
noncommunication. But observing itself and the ensuing system of
communication, the observing operation needs, similar to Fichte's
Ego,I6 a distinction between communication and non-communication,
or system and environment. The communicative system is or is not
self-transcending depending on whether you (or the system itself) ob-
serve it as an operating or an observing system. The evolutionary suc-
cess of communicative systems depends upon introducing the
difference between system and environment into the system-a "re-
entry of the form into the form" in the words of George Spencer
Brown. 17
Of course, the concept of communication has to be adapted to
this task because, as we know, different concepts make different dis-
tinctions and different distinctions make different observers. Commu-
nication cannot be defined as communicative action because this
would require the unthinkable: an actor without body and mind.'"
And it cannot be defined as transmission of messages or information
because this would presuppose non-transmitted codes and shared
frames from which information is selected. 19 In any case, communi-

14 Jacobson, supra note 13, at 1684.

15 See Jean-Pierre Dupuy, On the Supposed Closure of Normative Systems, in AUTO-
POIETIC LAW: A NEW APPROACH TO LAW AND SOCIETY 51 (Gunther Teubner ed., 1988),
for this problem with reference to the legal system.
16 JOHANN G. FICHTE, Grunlage der Gesamten Wissenschaftslehre, in 1 AUSGEWAHLTE
WERKE §§ 1-3 (1964).
17 See BROWN, supra note 7, at 56-57, 69-76.
18 Renate Mayntz, for one, likes to remind me that there are no actors without bodies and
minds. Certainly not, but since we can not include all biochemical, neurophysiological,
psychic, etc. activities as part (link? step?) of the communication processes, we have to exclude
actors. The question is whether and how to take individuals seriously.
19 See Benny Shanon, Metaphorsfor Language and Communication, 3 REVUE INTERNATI-
ONALE DE SYSTtMIQUE 43 (1989). See also TERRY WINOGRAD & FERNANDO FLORES, UN-
DERSTANDING COMPUTERS AND COGNITION: A NEW FOUNDATION FOR DESIGN (1986).
1424 CARDOZO LAW REVIEW [Vol. 13:1419
cation should not be thought of as including something-the reper-
toire of possibilities from which information is selected-that is not
communicated. Rather, communication seems to be an emergent re-
ality of its own, a kind of autopoietic network of operations which
continually organizes what we seek, the coincidence of self-reference
(utterance) and external reference (information). Communication
comes about by splitting reality through a highly artificial distinction
between utterance and information, both taken as contingent events
within an ongoing process that recursively uses the results of previous
steps and anticipates further ones.
When we begin to use this concept of communication, we imme-
diately notice a new problem. Once started and engaged in reproduc-
ing itself, the operation of communication produces the societal
system, the encompassing system of society, and, nowadays, the
global system of the world society-nothing more and nothing less.
Our conceptual framework produces a clear concept of society as the
closed system of connectable communications, reproducing communi-
cation by communication. Given the present state of sociology, this
may be an important step toward a theory of society. But how to
advance from here? If the society is an operationally closed auto-
poietic system, can we conceive of systems within the societal system
as closed systems as well? Biologists have to face the question of
whether, if cells are autopoietic systems, they can also conceive of
organisms as autopoietic systems. Sociologists may have to start glob-
ally and doubt whether, if societies are autopoietic systems, subsys-
tems of the societal systems can be autopoietic systems as well. Can
there be autopoiesis within autopoiesis, closure within closed systems?
And what exactly does "within" mean in this case?
Consider our topic of the legal system as a social system within
the societal system, as a subsystem of society. A description of the
legal system as an autopoietic system would require us to say that the
states of the system are exclusively determined by its own operations.
The environment can eventually destroy the system, but it contributes
neither operations nor structures. The structures of the system con-
dense and are confirmed as a result of the system's own operations,
and the operations are in turn recursively reproduced by structural
mediation. In this sense, the system is a structure-determined system.
But how can the society tolerate, or even promote, the emergence of
such a degree of autonomy within itself? It is obvious that the legal
system uses communication as its own basal operation and that it uses
language, although employing special terminology. It is also obvious
that we find a flow of communication crossing the boundaries of the
19921 CLOSURE & STRUCTURAL COUPLING 1425

legal system (which, of course, does not mean that the legal system
can communicate as a collective actor on its own behalf). On the
other hand, we could hardly think of a legal system as being unable to
recognize its own boundaries, as an arbitrarily designed analytic ob-
ject of outside observers. Such a system would be unable to distin-
guish between the legal and the external consequences of a legal
decision, to mention a famous and controversial issue.2 ° It would be
unable to find reasons for staying with precedents, for distinguishing,
or for overruling. It would even be unable, as we shall see, to separate
conceptual (internal) issues and interest (external) issues. As a social
system within the societal system, it reproduces society by communi-
cation. There is no sense in separating law and society as if these were
two different objects, and not even good sense in treating the society
as the environment of the legal system. The legal system itself is an
inseparable part of the societal system-it does not simply depend on
external sources for social support and legitimation, but is an inextri-
cable part of the network that reproduces the society by recursively
connecting communication with communication. Nevertheless, the
legal system is a closed system, producing its own operations, its own
structures, and its own boundaries by its own operations; not by ac-
cepting any external determination nor, of course, any external delim-
itation whatsoever. To put it even more pointedly, just because the
legal system operates as part of the social network of societal opera-
tions, there is no sense in looking for external sources of determina-
tion and delimitation. As part of the societal system, the legal system
is a self-organizing, self-determining system. There is nothing else, no
external system that could do the job.

III.

Reviewers and critics of systems theory who start with a different

conceptual framework--or, for the most part, without any-are noto-
riously unable to identify the hard questions. They normally operate
at the level of superficial plausibilities or implausibilities, and "clo-
sure" is, of course, a highly improbable evolutionary achievement.
They point to obvious causal connections and to important historical
and semantical correspondences between the societal system and the
legal system. Hierarchical societies of the past, of course, had a differ-
ent law than modern society, and the constitutional realities in the

20 See Bernard Rudden, Consequences, 24 JURID. REV. 193 (1979); Neil MacCormick, On
Legal Decisions and Their Consequences: From Dewey to Dworkin, 58 N.Y.U. L. REV. 239
(1983).
1426 CARDOZO LAW REVIEW [Vol. 13:1419

United States differ from constitutional realities in Brazil. No doubt,

but also, no objection.
It should be possible to discuss a theory at the level of its own
problems-I almost feel the temptation to recall the romantic notion
of critique 2"-and these problems are more serious and more difficult
than a critique aiming at rejection would ever imagine. As in all cases
of operational closure, the problem is how to define the operation that
differentiates the system and organizes the difference between system
and environment while maintaining reciprocity between dependence
and independence. This indeed is a most difficult question; it is the
core problem of a theory of the legal system.
Within the traditional framework of functionalism and division
of labor one would tend to point to a specific function, and this re-
mains part of the truth. The function of the legal system may be de-
fined as producing and maintaining counter-factual expectations in
spite of disappointments.2 2 The communication of this intention to
maintain one's own expectations even if they are not fulfilled (that is,
to refuse to learn from facts) uses the symbols of normativity, for ex-
ample, the word "should." But this does not suffice for closing the
system because there are many nonlegal norms, and the concept of
function itself suggests looking for functional equivalents and asking
the question of whether learning would not be the better (or at least a
functionally equivalent) way to equip the system with stable expecta-
tions. As we know, this question had been answered by recourse to
nature and Aristotelian ethics which teach only how to avoid errors
with respect to the good as the objective of action. Nowadays, people
look for moral rules or values to found the legal system in nonlegal
norms-the famous Jellinek-Weber-Habermas line of "legitima-
tion,"' 23 or the American discussion of moral aspiration versus origi-
nal intent as a guideline for an interpretation of the constitution. My
advice would be to unask such questions---or to "deconstruct"
them-and to replace them with the question of how the system or-
ganizes its own closure, its own social autonomy, and its own immu-
nity in fulfilling its function.
Legal reasoning uses the distinction between norms and facts, be-

21 Critique requires a Reflexionsmedium or an imaginary space in which identity and dif-

ference, unity and duality join in the attempt to improve on the work of art--or the work of
theory. See WALTER BENJAMIN, Der Begriff der Kunstkritik in der deutschen Romantik, in 1
GESAMMELTE SCHRIFTEN 7 (Hermann Schweppenhiuser ed., 1974).
22 For further elaboration, see NIKLAS LUHMANN, A SOCIOLOGICAL THEORY OF LAW 22-
49 (1985).
23 See, e.g., JORGEN HABERMAS, LEGITIMATION CRISIS (Thomas McCarthy trans., 1977)
(1973).
1992] CLOSURE & STRUCTURAL COUPLING 1427

tween normative and cognitive expectations. It has to know in which

respects it is supposed to learn (did somebody kill another woman?)
and in which respects not (should she have been killed?). Legal rea-
soning would not get along very well by confusing these questions. In
this sense, the system is normatively closed and cognitively open at
the same time.2 4 But the legal system has to anticipate (that is, to
know in advance of every specific operation) which norms are legal
norms and which norms are simply opinions in its environment, for
example, beautiful images of economic and ecological rationality. In
other words, the legal system has to organize the legal validity of
norms, validity represented as a circulating symbol, moved by legisla-
tion and contract, and with continually shifting contents. The recur-
sive autopoietic network of reproducing normative expectations with
reference to normative expectations uses the distinguishing mark of
legal validity. Acceptable legal reasoning has to restrict itself to legal
norms (including norms of moral standards referred to by legal
norms, professionally sound practice, and so on. 2 ' This is how clo-
sure is recognized--or "observed"-in the system (every legal theorist
will immediately recognize H.L.A. Hart's secondary rules of recogni-
tion). 26 However, this does not quite satisfactorily explain how clo-
sure is produced in the first place.
The structure that actually organizes the autopoiesis of the sys-
tem as an unavoidable outcome of its own operations is the system's
binary code; that is, the continuous necessity of deciding between
legal right and wrong. This code is a strictly internal structure. To
declare something illegal does not mean that it belongs to the environ-
ment of the system. We have to differentiate the distinction between
self-reference and external reference on the one hand, and legal and
illegal on the other.2 7 Moreover, the code of the system is not a norm;
this assumption would only lead to the paradoxical question of
whether the distinction between legal and illegal is itself legal or ille-

24 1 share some of the objections of Arthur J. Jacobson, supra note 13, against this highly
reduced formula, and the text tries to elaborate on the problems as I see them. But I would
maintain that this is not a question of explaining or organizing dynamism. Closed systems are
inherently temporalized, dynamic systems. They look for occasions, irritations, opportunities
in their environment. But even the classification as occasion, irritation, opportunity (e.g., as a
complaint) is an internal classification and not something which exists independently of the
system in its environment.
25 See NEIL MACCORMICK, LEGAL REASONING AND LEGAL THEORY (1978) (proposing
for this refined positivistic approach the brand name of "institutional theory of law").
26 See H.L.A. HART, THE CONCEPT OF LAW (1961).
27 Even epistemology has to recognize that problems of reference should not be confused
with problems of truth and that there are no definitional links between the concepts of reality,
meaning, and truth as the Vienna circle presupposed.
1428 CARDOZO LAW REVIEW [Vol. 13:1419

gal. The code is simply a rule of attribution and connection. If the

question arises whether something is legal or illegal, the communica-
tion belongs to the legal system, and if not then not. There is no
further sanction and no natural or societal sorting of topics and com-
munications as belonging or not to the legal system; it is a purely
factual matter. The question of legality is or is not picked up in com-
munication,2 8 and by this very fact the communication takes part in
the recursive network of legal communication. 29 Nobody, not even
the legal system, can refuse to ask the question. It may, of course, be
possible to avoid asking the question by suppressing communication.
The legal system can expand or shrink according to societal violence;
but this does not change the rule of attribution and connection, it
simply changes its range of application.
Identifying itself by its binary code, and distinguishing itself from
its environment by the specificity that code, the legal system knows of
no fundamental norm (Grundnorm) representing its unity within the
societal system. There is no way to introduce the unity of the system
into the system. The system is an open-ended, ongoing concern struc-
turally requiring itself to decide how to allocate its positive or nega-
tive value. The bifurcation necessitates decisions and thereby further
operations, and decisions require the construction of normative rules
(programs) to connect them in a network for reproducing decisions.
Norms, then, are purely internal creations serving the self-generated
needs of the system for decisional criteria without any corresponding
"similar" items in its environment. Nothing else is meant by "auto-
poiesis." Historically, there is no beginning except an always renewed
reconstruction of the past.30 Logically, there are no apriorities, but
simply a circular, reciprocal conditioning of the code and programs.
Only the programs have a normative status and convey a normative
quality to concrete expectations. They serve the function of the sys-

28 This, of course, is not to say that it is simply a matter of chance. We can also recall that
some societies allow for asking the qumstio iuris even in the context of highly political issues.
The Roman republic was famous for that.
29 See Niklas Luhmann, Communication About Law in Interaction Systems, in ADVANCES
IN SOCIAL THEORY AND METHODOLOGY: TOWARD AN INTEGRATION OF MICRO- AND
MACRO-SOCIOLOGIE 234 (K. Knorr-Cetina & V. Cicourel eds., 1981).
30 At this point we meet the "historical school," and in particular, Friedrich C. von Savi-
gny, for whom a conscious formulation of legal rules is only possible when such a rule can be
found in practice. FRIEDRICH C. VON SAVIGNY, SYSTEM DES HEUTIGEN RbMISCHEN
RECHTS (SYSTEM OF MODERN ROMAN LAW) 14 (1840). We would say that the autopoietic
closure of the legal system is only possible if sufficient preadaptive advances, experiences in
handling conflicts, model cases, etc. are available. For the common law, see SIR MATTHEW
HALE, A HISTORY OF THE COMMON LAW OF ENGLAND (photo. reprint 1987)(1713)
(decoupling the question of the validity of the law and the legitimacy of the Norman conquest
as its origin).
1992] CLOSURE & STRUCTURAL COUPLING 1429

tem. The closure of the system is produced by realizing the structural

difference of the code and programs, and this difference alone gives
legal norms a distinctive flavor.
Normative closure means, above all, that morality as such has no
legal relevance-neither as code (good/bad, good/evil), nor in its spe-
cific evaluations. The law provides ample space for immoral behav-
ior. Non omne quod licet honestum est.3 This is not simply a moral
weakness of the law, but a condition of free and unrestrained develop-
ment of moral communication. Such was the common opinion in the
eighteenth and nineteenth centuries, and it was the unavoidable con-
sequence of religious and moral pluralism produced by the printing
press. Of course, this does not prevent the legal system from incorpo-
rating moral constraints as legal constraints; but this has to be done
within the system and has to be checked by the usual references to
legal texts, precedents, or rulings that limit the realm of legal argu-
ment. 2 Even theories that dare to assert that all legal decisions in so-
called "hard," and even "easy," cases need a moral justification main-
tain that looking for "grounds drawn from outside the law" is, "in-
deed, . . . required by law."33 Whatever import one claims for
external references, these references are aspects of internal operations.
The system has to take care of itself.

IV.
The daily problem of closed social systems is how to connect in-
ternal and external references by internal operations. At the opera-
tional level this problem is solved by distinguishing norms from facts;
that is, in terms of internal structures, distinguishing normative from
cognitive expectations. The distinction of normative and cognitive ex-
pectations, and this holds true for any distinction, has to be made; in
our case, it has to be made by the legal system. It cannot be found in
the natural or created world. It is not a "categorical" property of the
world. This means that even "facts" that are relevant for the legal
system are not facts for everybody.34 Facts are constructions, state-
ments about the world, and careful sociological investigations show
that scientific facts and facts which serve as components of legal or
31 DIG. 50.17.144 (Paulus, Ad Edictum 62) ("Not everything which is lawful is

honorable.")
32 See Neil MacCormick, Why Cases Have Rationes and What These Are, in PRECEDENT

IN LAW 155, 166-182 (Laurence Goldstein ed., 1987).

33 David Lyons, Justification and Judicial Responsibility, 72 CAL. L. REV. 178, 186, 188
(1984) (introducing further distinctions to "make this conclusion less paradoxical.").
34 And we may add, it may appear so at the level of first order observers, but not for us, not
for observers observing these observers.
1430 CARDOZO LAW REVIEW [Vol. 13:1419
political-administrative decision making differ in remarkable ways. 5
In other words, knowledge has different "credibility profiles" 36 inside
and outside the legal system. Legal facts are made to fit the legal
framework; they have to facilitate as far as possible the deductive use
of legal norms. They have to support the presentation of legal validity
by conveying the impression that, given the rules, the decision follows
from the facts of the case. They have to be certified facts.3 7
At the level of juridical doctrine, the same problem can be for-
mulated in terms of concepts and interests. We are used to thinking
of analytic jurisprudence and interest-orientation as two different,
competing schools of legal thought. The interest approach was born,
at least in Germany, by inventing an opponent called Begriffjuris-
prudenz (Conceptual Jurisprudence). 38 Roscoe Pound copied this po-
lemical style. 39 The tribal rules of the academic system favor such
distinctions of schools and controversies. The legal system operates
under different conditions of autopoietic reproduction. It has to com-
bine internal and external references and, to the extent that general-
izations are useful, may refer to well-tried concepts and to well-known
interests. There is no contradiction in using both references simulta-
neously. Concepts articulate the self-referential aspects of legal deci-
sions; interests, on the other hand, are environmental facts to be taken
as given. The task of the system as formulated by the system consists
of distinguishing interests protected by the law from interests to be
suppressed and combatted.'

35 See EXPERT EVIDENCE: INTERPRETING SCIENCE IN THE LAW (Roger Smith & Brian
Wynne eds., 1989) [hereinafter EXPERT EVIDENCE].
36 Brian Wynne's term. Brian Wynne, Establishingthe Rules of Laws: ConstructingExpert
Authority, in EXPERT EVIDENCE, supra note 35, at 52.
37 For a very different handling of certainty/uncertainty questions within the system of
science, see Brian L. Campbell, Uncertainty as Symbolic Action in Disputes Among Experts,
supra at 429; Susan L. Star, Scientific Work and Uncertainty, 15 Soc. STUD. OF SCi. 391
(1985).
38 Today we are well aware that this was, to a large extent, fictional history. See, e.g.,
ULRICH FALK, EIN GELEHRTER WIE WINDSCHEID: ERKUNDUNGEN AUF DEN FELDERN
DER SOGENANNTEN BEGRIFFSJURISPRUDENZ (1989).
39 See Roscoe Pound, MechanicalJurisprudence, 8 COLUM. L. REV. 605 (1908).
40 See 3 ROSCOE POUND, JURISPRUDENCE (1959). He writes: "A legal system attains the
ends of the legal order (1) by recognizing certain interests individual, public, and social; (2) by
defining the limits within which those interests shall be recognized and given effect through
legal precepts .... " Id. at 16. So some (or almost all?) interests are left without protection, or
are even repressed. This positive/negative distinction can, of course, not be deduced from the
interests as such, for even bad interests are interests. A theory which looks exclusively at
interests cannot give good reasons for this distinction except by saying that the legal system
has an interest in distinguishing protected and unprotected interests. See PHILIPP HECK,
Gesetzesauslegung und Interessenjurisprudenz (1914), reprinted in PHILIPP HECK, DAS PROB-
LEM DER RECHTSGEWINNUNG 102 (1968); BENJAMIN CARDOZO, THE NATURE OF THE JU-
19921 CLOSURE & STRUCTURAL COUPLING 1431

This distinction cannot be derived from the environment of the

system, nor can it be "seen" as an inherent quality of systems. The
distinction must be constructed by the internal operations. This is not
to say that the dividing line separating protected from suppressed in-
terests can be drawn arbitrarily. The guarantee of "justice" is not the
correspondence with external qualities or interests, but the consis-
tency of internal operations recognizing and distinguishing them. It is
this requisite of consistency which, under conditions of sufficient com-
plexity, leads to the elaboration of concepts.
Concepts are also internal operators referring to internal differ-
ences, or at least this has to be presupposed. Even if a concept-like
the bona fides of the Romans-has moral connotations in everyday
language, these do not become part of the law except by explicit refer-
ence. In other words, bonafides is not a "source of law."4 The same
holds true for terms that are used in legal texts but also have scientific
meanings (often more than one) outside the law. If a concept is used
explicitly as citation, it refers to environmental norms, rules, and cus-
toms as facts that must be proven.
At the level of single cases, there may be a choice between giving
priority to the urgency of interests or to the purity of legal concepts,
but the system as such cannot choose in this way. If its decisions are
presented as evaluations of interests, this amounts to understating
conceptual issues and neglecting for the time being the control of con-
sistency. If the reasons given are conceptual ones, this will be a dis-
guised way of favoring or disfavoring certain interests more than
others. In theoretical terms, the ultimate problem always consists of
combining external and internal references, and the real operations
which produce and reproduce such combinations are always internal
operations. Nothing else is meant by closure.

V.
To repeat again and again this trivial point, closure does not
mean empirical isolation. Closure is a highly selective, improbable,
artificial achievement-not in the sense of intentional design, but as
an outcome of evolution.
The emergence of closed systems requires a specific form of rela-
tions between systems and environments; it presupposes such forms

DICIAL PROCESS 112 (1921) ("One of the most fundamental interests is that the law shall be
uniform and impartial.").
41 See ANTONIO CARCATERRA, INTORNO Al BONAE FIDEI IUDICIA (1964); Yan Thomas,
La langue du droit romain: Problemes et mdthodes, 19 ARCHIVES DE PHILOSOPHIE DU DROIT
103 (1974) (asking for more refined semiotic analyses of this question).
1432 CARDOZO LAW REVIEW [Vol. 13:1419

and is a condition of their possibility as well. The theory of "open

systems" describes these forms with the categories of input and out-
put. This model postulates a causal chain in which the system itself
serves as the connecting part linking inputs and outputs. The theory
of autopoietic systems replaces the input/output model with the con-
cept of structuralcoupling.42 It renounces the idea of an overarching
causality (admitting it, of course, as a construct of an observer inter-
ested in causal attributions), but retains the idea of highly selective
connections between systems and environments.
The structural coupling of system and environment does not con-
tribute operations (or any other components) for the reproduction of
the system. It is simply the specific form in which the system presup-
poses specific states or changes in its environment and relies on them.
Walking presupposes the gravitational forces of the earth within very
narrow limits, but gravitation does not contribute any steps to the
movement of bodies. Communication presupposes awareness states
of conscious systems, but conscious states cannot become social and
do not enter the sequence of communicative operations as a part of
them; they remain environmental states for the social system.43 Struc-
tural couplings are forms of simultaneous (and therefore, not causal)
relations. They are analogical, not digital, coordinations.
The system in its normal dealings does not observe its structural
couplings, but it has to contend with perturbations, irritations, sur-
prises, and disappointments channeled by its structural couplings. It
must also assimilate and accommodate (Piaget's terms!) such ambigu-
ities. But perturbations are purely internal constructs because they
appear only as deviations from expectations; that is, in relation to the
structure of the system. The environment does not contain perturba-
tions or anything that in a semantical sense is similar to them. Nor is
there any transmission of perturbations from the outside into the sys-
tem. The twin concepts of closure and structural coupling exclude
the idea of information "entering" the system from the outside. Even
in the case of cognitive expectations this is impossible because selec-
tions of information are always internally constructed, and cognitive
expectations are nothing but specific forms to be prepared for irrita-

42 See HUMBERTO R. MATURANA, ERKENNEN: DIE ORGANISATION UND

VERK(RPERUNG VON WIRKLICHKEIT: AUSGEWAHLTE ARBEITEN ZUR BIOLOGISCHEN Epis-
TEMOLOGIE 143-45, 243-44, 251-52 (1982) [hereinafter ERKENNEN]; HUMBERTO R.
MATURANA & FRANCISCO VARELA, EL ARBOL DEL CONOCIMIENTO 49-53, 64, 154-55 (1984).
43 See NIKLAS LUHMANN, Wie ist Bewusstsein an Kommunikation beteilgt?, in MATERI-
ALITAT DER KOMMUNIKATION 844 (Hans Gumbrecht & K. Ludwig Pfeiffer eds., 1988);
NIKLAS LUHMANN, DIE WISSENSCHAFT DER GESELLSCHAFT (1990).
1992] CLOSURE & STRUCTURAL COUPLING 1433

tions (surprises, unpredictabilities).44 But without structural coupling

there would be no perturbation and the system would lack any chance
to learn and transform its structures. Hence, structural coupling, to-
gether with sufficient internal complexity, is a necessary precondition
for building up regularities45 to construct order from noise or redun-
dancy from variety. Communication never becomes thought, but
without being continually irritated by communication, an individual
would not become a socialized individual; it46 would remain depen-
dent upon its flow of perceptions, that is, dependent upon structural
couplings and internally constructed regularities of another type.
In this way, structural couplings provide a continuous influx of
disorder against which the system maintains or changes its structure.
Memory depends upon this tendency towards entropy. Memory is
not a storage of past facts (the past can never be present), but a form
for mediating order and disorder-very frequently by forgetting, in
other cases by constructing a spatial or temporal order to dissolve
incompatibilities. However, this would not be possible if structural
couplings did not exclude most of the environmental facts from im-
mediate relevance. Structural coupling presupposes and organizes
decoupling. Communication (which is to say, society) is coupled to
consciousness, but not to the immense mass of physical, chemical, and
biological facts. These facts can prevent communication and they can
destroy it, but they cannot irritate communication. In this sense,
coupling has to be conceived as a difference, as a form with two sides:
an internal side that admits irritation and an external side to which
the system remains indifferent. Structural couplings arise with sys-
tems; they are not physical, chemical, biological facts that exist before
systems emerge (although these facts as such preexist as preadaptive
advances for the emergence of systems).
Applying this complex conceptual apparatus at the level of socie-
tal differentiation, and in particular to the differentiation of a legal
system, we immediately see the structural coupling of social commu-

44 This responds to a remark of William Evan, that the theory of autopoietic systems does
not explain how information (expectations, demands) is transmitted to the legal system. WIL-
LIAM M. EVAN, SOCIAL STRUCTURE AND LAW: THEORETICAL AND EMPIRICAL PERSPEC-
TIVES 42 (1990). There is normal communication as the operation of the societal system, to be
sure, but no contribution of external sources to what the closed system can handle as
information.
45 Francisco J. Varela, On the Conceptual Skeleton of Current Cognitive Science, in BE-
OBACHTER: KONVERGENZ DER ERKENNTNISTHEORIEN? 13 (1990), postulates a modular
(non-hierarchical) order of information processing as an additional prerequisite. See also Jo-
seph A. Goguen & Francisco J. Varela, Systems and Distinctions:Duality and Complementar-
ity, 5 INT'L J. GEN. Sys. 31 (1979).
46 It = the system, whether he or she.
1434 CARDOZO LAW REVIEW [Vol. 13:1419

nication reproducing society on the one hand, and special legal mean-
ings as normative projections claiming legal validity-the legal code
and the special programs (laws, regulations, contracts)-on the other.
Communication is the "domain" 4 7 in which the differentiation of a
legal system becomes possible. This does not (and cannot!) require a
communication of the legal system to the society as a relation between
sender and receiver. The legal system cannot communicate as a unity
and the society has no address. However, by operating within its own
boundary, the legal system reproduces itself and the societal system
without making this simultaneity a topic of communication, without
using it as an argument in pleading before the court, and, of course,
without needing any "legitimation" by the societal system. It hap-
pens as an unavoidable fact because (not although!) the legal system
reproduces itself under the condition of operational closure.

VI.
Given the fact of the structural coupling of the societal system
and its legal system, further structural couplings can evolve between
the legal system and other functional subsystems. All subsystems use
the same domain, "communication." 4' 8 They could not be subsystems
of the societal system on the basis of other types of operations-say
biological or conscious ones-but this does not solve all the problems
of coupling and decoupling which arise in the relations between the
subsystems. In traditional societies we find devices to represent social
order as relations between susbsystems, for example, as relations of
center and periphery, city and country; or as relations of rank be-
tween castes or estates. The transition to modem society dissolved
this order without replacing it-Foucault cites a "loss of representa-
tion" occurring in the eighteenth century.4 9 Under the regime of
functional differentiation, the societal system loses its integrative ca-

47 ERKENNEN, supra note 42, at 154.

48 This statement excludes theories which oppose a semantic level of communication and
macro-economic structures. Authors who use this or similar distinctions for rejecting the id-
iom of autopoiesis should be asked to clarify their distinction at an adequate conceptual level.
See, e.g., Peter Nahamowitz, Autopoiesis oder dkonomischer Staatsinterventionismus?,in 9
ZEITSCHRIFr FOR REHTSSOZIOLOGIE 36 (Erhard Blankenburg et al. eds., 1988). It is, of
course, not acceptable to rush into "empirical" arguments without sufficient explication of the
theoretical description of the issue, because this will only lead to talking about different ob-
jects. The critics would also have to answer the questions about whether they can think of
social systems without the capacity of self-transcendence or re-entry, without self-observation
and self-description, and the kind of operation upon which such systems would be based. See
infra, Part I.
49 MICHEL FOUCAULT, LES MOTS ET LES CHOSES: UNE ARCH]tOLOGIE DES SCIENCES
HUMAINES 229-61 (1966).
19921 CLOSURE & STRUCTURAL COUPLING 1435

pacity. Reduced to its mechanism of structural coupling, it continues

to autopoietically reproduce itself by communication. But language
as such contains and spreads the possibility of refusing all kinds of
proposals. Under these conditions, the social order requires auto-
poietic closure, self-organization, and autonomy of the most impor-
tant function systems as well as the development of new forms of
structural coupling for the relations between these systems.
The economic system depends on the codes of property and
money. Without a clear divide between having and not having prop-
erty rights, no transaction would be possible. Nevertheless, the eco-
nomic and the legal consequences of a transaction are completely
different because they occur in different systems in different recursive
networks under different criteria and concrete conditions. The eco-
nomic and the legal systems are and remain separate, and both oper-
ate under the condition of operational closure; but this needs a specific
mechanism of structural coupling, above all in the form of property
and contract. 50
There is much historical evidence that, beginning in the four-
teenth century, the legal system adapted to these requirements and
enlarged-slowly, and with many scruples-the permitted degrees of
freedom in property and contract. The core meaning of property has
included since Bartolus,5" in addition to defense and use and enjoy-
ment (usus, fruitio), the right of disposal (dispositio). The same im-
probabilities deform and generalize the institution of contract;
contract was the most important legal invention of Roman civil law,
providing a legal instrument not only for solving actual conflicts, but
for regulating and avoiding them. During the transition from stratifi-
cation, based on real estate, to functional differentiation, the society
created needs, motives, and legitimation for enlarging the scope of
employing property and the scope of possible contracts with legal pro-
tection. The limitations of access to courts which the traditions of
civil law and common law arranged under the names of actio and writ
vanished, and finally-in England only in the nineteenth century-
contracts became legally valid, even without "consideration," on the
50 Other important requisites include corporate law-which, during the eighteenth & nine-
teenth centuries, became independent from political privileges and monopolies-and patent
law, and, last but not least, the banking system. The latter provided a sufficient separation and
recoupling of money and other (real) property for the credit mechanism which the economy
needed: the capacity to pay even without sufficient wealth.
51 See DIETMAR WILLOWEIT, Dominium und Proprietas: Zur Enwicklun des Eigentum.
sbegriffs in der mittelalterlichen und neuzeitlichen Rechtwissenschaft, 94 HISTORISCHES
JAHRBUCH DER GtORRESGESELLSCHAFT 131 (1974). Bartolus was a twelfth century Italian
jurist and commentator on Roman law. See generally JOSEPH H. BEALE, BARTOLUS ON THE
CONFLICT OF LAWS (1914).
 1436 CARDOZO LAW REVIEW [Vol. 13:1419

basis of private will.5 2

Modem concepts of property and contract do not integrate or
even de-differentiate the legal and economic systems. As mechanisms
of structural coupling, they organize the reciprocal irritation of these
systems and influence, in the long run, the natural drift of structural
developments in both systems. In particular, the regulatory state pre-
supposes this connection and uses it as a medium of political interven-
tion into both systems by limiting once again the degrees of freedom
for using property and contract.
If we look for a parallel mechanism coupling the legal system
and the political system, we find it in the form of constitutions-in the
modem sense-emerging from revolutionary movements during the
second half of the eighteenth century. 53 The historical innovation, the
mutation of legal forms, had been occasioned by concrete political
circumstances and by the possibilities they provide for an instrumen-
tal use of conceptual variations.54 This has been favored by the mod-
em concept of the "state," which suggests a unity of political and
legal "sovereignty." This unity makes it difficult to see the distinc-
tion, but the political system and the legal system are separate, opera-
tionally closed, autopoietic systems. A political operation as such
never has legal relevance if not endowed with it by the legal system
and vice versa. Otherwise no political discourse, no political bargain-
ing, and no policy planning would be possible without immediate
legal effects. The constitution separates the systems and provides for
their structural coupling. The ultimate paradoxes and tautologies of
the legal system (that law is whatever the law arranges to be legal or
illegal) can be unfolded by reference to the political system (for exam-
ple, the political will of the people giving itself a constitution), and the
paradoxes and tautologies of the political system (the self-inclusive,

52 For the common law, see PATRICK S. ATIYAH, THE RISE AND FALL OF FREEDOM OF
CONTRACT (1979); MORTON J. HORWITZ, THE TRANSFORMATION OF AMERICAN LAW,
1780-1860 (1977); MAX RHEINSTEIN, DIE STRUKTUR DES VERTRAGLICHEN
SCHULDVERHALTNISSES IM ANGLO-AMERIKANISCHEN RECHT (1932); ALFRED W.B. SIMP-
SON, A HISTORY OF THE COMMON LAW OF CONTRACT: THE RISE OF THE ACTION OF As-
SUMPSIT (1975); EIKE VON HIPPEL, DIE KONTROLLE DER VERTRAGSFREIHEIT NACH
ANGLO-AMERIKANISCHEM RECHT: ZUGLEICH EIN BEITRAG ZUR CONSIDERATIONSLEHRE
(1963). On the continent, the nudum pactum was recognized earlier and the problem was to
reduce the amount of state intervention into contracts in the context of "mercantilistic" poli-
cies. See DIETER GRIMM, Soziale, wirtschaftliche und politische Voraussetzungen der Vertrag-
sfreiheit, in RECHT UND STAAT DER BORGERLICHEN GESELLSCHAFT 165 (1987).
53 For a more extensive treatment, see Niklas Luhmann, Verfassung als evolution/ire Er-
rungenschaft, 9 RECHTSHISTORISCHES J. 176 (1990).
54 See, e.g., CONCEPTUAL CHANGE AND THE CONSTITUTION, (Terence Ball & John G.A.
Pocock eds., 1988); POLITICAL INNOVATION AND CONCEPTUAL CHANGE (Terence Ball et al.
eds., 1989).
1992] CLOSURE & STRUCTURAL COUPLING 1437

binding, sovereign power) can be unfolded by reference to the positive

law and by supercoding the legal system with the distinction of consti-
tutional and unconstitutional legality. 5"
The institutional justification for constitutions cannot be reduced
to either a political or a legal function, and it does not lie in the value-
laden language of the constitution itself, nor in the value of its values.
The constitution serves the dual function of including and excluding
reciprocal perturbations of political and legal operations. Its two-
sided form of including and excluding influence maintains the separa-
tion of the systems and allows for separate autopoietic reproduction
without any confusing overlap. It also characterizes the ways in
which the legal system (and on the other side, the political system)
avoids isolation (which means entropy) and constructs on its internal
screen what can serve within the system as information. As previ-
ously discussed, the system operates as a non-self-transcending system
on the level of its operations, and as a self-transcending system on the
level of its observations; this in spite of the fact that observations are
operations.
To fully understand this apparently contradictory, even paradox-
ical, statement, it is necessary to take time into account. Operations
are events without duration; they vanish as soon as they appear. Ob-
served as single events, they can participate in different systems. A
payment can be at the same time (but only at the same time) the ful-
fillment of a contractual obligation in the legal system and part of an
economic transaction which transfers the capacity to make further
payments in the economic system. The same holds true for an act of
legislation which may have both political and legal relevance. An ob-
server may identify these aspects as one event and may even find him-
self unable to see two different operations. The systems, however,
that generate their own elementary operations also need their own
recursive networks and connections with earlier and later operations
within the same system in order to construct meaningful identities.
They condense and confirm identities for repetitive use within the
same system5 6 and, for them, preconditions and consequences of
events differ completely according to system reference. Since auto-

55 For the decisive invention of the term "unconstitutional" during the eighteenth century,
see Gerald Stourzh, Constitution:Changing Meanings of the Term from the Early Seventeenth
to the Late Eighteenth Century, in CONCEPTUAL CHANGE AND THE CONSTITUTION, supra
note 54, at 35. Cf GERALD STOURZH, WEGE ZUR GRUNDRECHTSDEMOKRATIE: STUDIEN
ZUR BEGRIFFS-UND INSTITUTIONENGESCHICHTE DES LIBERALEN VERFASSUNGSTAATES
50-74 (1989).
56 For further elaboration, see NIKLAS LUHMANN, Identita't-was oder wie?, in 5 Sozio-
LOGISCHE AUFKLARUNG 14-30 (1990).
1438 CARDOZO LAW REVIEW [Vol. 13:1419
poietic systems are temporalized systems depending on self-generated
dynamic forms of stability, 57 they necessarily differentiate and recog-
nize their own operations by temporal orientations. Therefore, any
observer who cares for the perspective of the system itself--of course,
there can be other observers with other frames and interests--cannot
cross-identify events over boundaries.

VII.
One of the most frequent objections to the theory of autopoietic
systems in general, and its application to legal systems specifically,
states that this theory-if it is a theory at a115 8-- does not care for
empirical verification. 9 This criticism needs two comments. First,
the repertoire of empirical methods in present day sociology is very
limited and completely inadequate for objects like the legal system;
that is, self-observing objects with highly structured complexity. Re-
stricting the access to objects by available empirical methods-and I
take this denotation in its usual meaning-would simply mean not
seeing the society and its legal system as it contructs and presents
itself. It requires going out of the market and leaving the business to
others-to mass media and to sociological writers. This should not be
the last word. Secondly, one can look for new combinations of (1)
well-known and uncontroversial (maybe obvious or even trivial) facts;
and (2) theoretical descriptions which present the obvious in unusual
illumination. The distinction between operational closure and struc-
tural coupling is one of these theoretical instruments. In other words,
given the structure of its domain, sociology cannot reduce its concept
of social reality to self-created data. We may even doubt whether
there is any correlation between empirical research and social reality,
except by the methodologically unguided "active interpretation" of
results.
If these considerations suggest (to repeat, for the present situa-
tion) a primacy of refined theoretical research, they do not exclude
57 It is a frequent, but very crude and uninformed, error to say that the theory of auto-
poietic systems does not have the possibility of taking dynamism (on the level of operations)
and change (on the level of structures) into account. On the contrary, the theory has no place
for any kind of non-dynamic, unchangeable, "essential," "substantive" components.
58 Within the context of American sociology, this level of theorizing is sometimes called
"metatheoretical"-as in the section title under which a paper of the present author is printed,
in DIFFERENTIATION THEORY AND SOCIAL CHANGE: COMPARATIVE AND HISTORICAL PER-
SPECTIVES (Jeffrey C. Alexander & Paul Colomy eds., 1990). Seen from a European perspec-
tive, this reflects the rather modest level of theoretical aspirations in present day American
sociology. Cf DAG OSTERBERG, METASOCIOLOGY: AN INQUIRY INTO THE ORIGINS AND
VALIDITY OF SOCIAL THOUGHT (1988).
59 See, e.g., EVAN, supra note 44, at 46.
1992] CLOSURE & STRUCTURAL COUPLING 1439

empirical research. We have outlined a hypothesis indicating rela-

tions between a transition from stratification to functional differentia-
tion and the invention of new or the reformulation of old mechanisms
of structural coupling during the transition period. There could be
many similar suggestions, for example, concerning the forms of de-
paradoxification of the legal system's self-description 6 or the replace-
ment of the unified notion of iurisdictioby the separation and circular
recombination of legislation and jurisdiction as a condition for the
transition from natural law to positive law as the dominant form of
self-validation of the legal system. There could be many empirical
projects exploring the sensitivity (or limits thereon) of the autopoiesis
of the legal system to social and political changes. 6 1 There are no
fundamental incompatibilities between the theory of self-referential
systems and empirical research, but there is an uncomfortable tension
between theoretical conceptions and the present possibilities of empir-
ical research. Instead of rejecting theory as unverifiable, critics
should see the insufficiencies on both sides.

VIII.
Finally, I return to the concept of operational closure. Focussing
on the operation that autopoietically reproduces the system as the net-
work which reproduces its operations offers new insights with respect
to the relation of structure and process. An autopoietic system does
not consist of two different kinds of entities, namely structures and
processes. It is not composed of two different kinds of matter or sub-
stance. The enzymes of the living cell are at the same time outcomes
of production, productive factors, and programming factors which or-
ganize the reproduction of the cell. The human mind does not consist
of two different kinds of entities which have been called-by the
Logic of Port Royal62 and by Locke, 63 as well as their followers-
ideas and representations. There are not two different qualities of a
communicative system, langue and parole, as the linguistic theory of
Saussure would have it.6M Finally, it is questionable what it means to

60 See Niklas Luhmann, The Third Question: The Creative Use of Paradoxesin Law and
Legal History, 15 J. L. & Soc'y 153 (1988).
61 See, e.g., MICHAEL HUTER, DIE PRODUKTION VON RECHT: EINE SELBSTREFEREN-
TIELLE THEORIE DER WIRTSCHAFT, ANGEWANDT AUF DEN FALL DES ARZNEIMIT-
TELPATENTRECHTS (1989).
62 ANTOINE ARNAULD & PIERRE NICOLE, LA LOGIQUE OU L'ART DE PENSER (1972)
(1662).
63 See JOHN LOCKE, AN ESSAY CONCERNING HUMAN UNDERSTANDING (Peter H. Nid-
ditch ed., 1975).
64 See F. DE SAUSSURE, supra note 5.
1440 CARDOZO LAW REVIEW [Vol. 13:1419

say that a legal system consists of "norms and activities. ' 65 Rather,
the system uses the same type of operation-be it biochemical replica-
tion, thought, or communication-in the dual function of (1) produc-
ing subsequent operations; and (2) confirming or changing the
structure used to select the next operation. In the sense of Heinz Von
Foerster, it is a non-trivial, self-referential machine-a machine that
uses every operation to construct itself anew.66 As observers, and for
analytic purposes, we may distinguish these two functions of produc-
ing operations and using, confirming, or changing structures. In real-
ity, these are only two aspects which necessarily require each other.
In a complex system you cannot fix the next operation without select-
ing it, you cannot select without narrowing the choice in the first
place, and you cannot restrict the possibilities at hand without ac-
cepting frames, as some people say, or structures, or in the case of
social and legal systems, expectations.
Operational closure, then, means that a system has to rely on all-
purpose operations of a specific type (and, of course, not anything
goes!). It has to use the same type of operation for confirming and
changing (or simply forgetting) structures, and each operation is al-
ways determining the next operation. It is simply a prime fact that
the autopoiesis goes on and on. There is no operation available that
could stop it, because all operations gain their own unity by produc-
ing subsequent operations. The system may select, condense, confirm,
change, or forget structures-but all this is a way to continue its own
autopoiesis. Of course, this does not prevent destruction, but it goes
on as long as it goes on. If a system can organize structural changes,
it can increase its adaptive capacity, but also its maladaption. Under
a condition of sufficient complexity, a system can differentiate proce-
dures for changing structures, and structures for organizing proce-
dures for changing structures. This all requires and uses the type of
operation that produces and reproduces the unity of the system.
I conclude with a trivial remark: There is no conservative bias in
such a theory. The system has no preference for maintaining itself,
there is simply no choice. It can continue by confirming or by chang-
ing its structure if operations are available to focus on such an out-
come; otherwise it just happens. An observer (which may be the

65 See TORSTEN ECKHOFF & NILS K. SUNDBY, RECHSSYSTEME: EINE SYSTEMTHE-

ORETISCHE EINFHRUNG IN DIE RECHTSTHEORIE 41 (1988). I suppose that most legal theo-
rists would tend to accept this without giving further thought to what it could mean to say
"and"-norms and activities.
66 See Heinz Von Foerster, Principles of Self-Organization-In a Socio-Managerial Con-
text, in SELF-ORGANIZATION AND MANAGEMENT OF SOCIAL SYSTEMS: INSIGHTS,
PROMISES, DOUBTS, AND QUESTIONS 2, 8-10 (Hans Ulrich & Gilbert J.B. Probst eds., 1984).
1992] CLOSURE & STRUCTURAL COUPLING 1441

system itself) may or may not discover that structures have changed
over time. He may see that intentional changes are conservative with
respect to the frames, values, procedures, and constitutions which are
needed for, and confirmed by, intentional changes. An observer may
use the distinction of static and dynamic systems for himself. Systems
theory, however, will advise him that this scheme is a crude simplifi-
cation. Autopoietic systems are systems organizing dynamic stability.
If an observer does not heed to this advice, and his own autopoiesis
does not force him to do so, it would be advisable to change the topic
of interest and observe the observer.