CEPA eprint 1340 (EVG-050)

Radical constructivism and Jean Piaget’s concept of knowledge

Glasersfeld E. von (1979) Radical constructivism and Jean Piaget’s concept of knowledge. In: Murray F. B. (ed.) The impact of Piagetian theory. University Park Press, Baltimore MD: 109–122. Available at http://cepa.info/1340
Table of Contents
The traditional approach to cognition
Experience and subjectivity
Organism and environment
The interactionist view
The contribution of feedback and control theory
The concept of knowledge in genetic epistemology
Unlike the majority of contemporary psychologists, Piaget has never dodged the epistemological question that the study of human behavior, and especially of cognitive behavior, inevitably raises. In fact, in almost everything he has written he touches upon the problem of knowledge and how we acquire it. Nevertheless it is not easy to come to a clear understanding of Piaget’s theory of knowledge. This chapter attempts to show that his genetic epistemology entails a drastic break with the traditional conception of “knowledge” that can be interpreted from a “radical constructivist” point of view. This constructivist approach is “radical” because it embodies not only the view that cognition must be considered a process of subjective construction on the part of the experiencing organism rather than a discovering of ontological reality, but also the belief that there can be no rational access to any world as it might be, prior to, and independent of, our experience.
After a short exposition of the traditional view of cognition and the concept of interaction between organism and environ‑ment, the epistemological ideas that have recently been developed in the field of cybernetics are outlined. The emerging radical constructivist model is then applied to the Piagetian concept of knowledge and it is suggested that this interpretation, though rarely explicitly stated, is implicit in Piaget’s genetic epistemology and is, indeed, necessary in order to make it an internally consistent theory of knowledge.
The traditional approach to cognition
In the main stream of philosophical tradition, ever since the pre-Socratics, the questions “What is knowledge?” and “How do we come to have it?” are inextricably tied to the concepts of truth and of reality. Whether or not it is explicitly stated, “knowledge” is conceived as the knower’s representation of things and events “in themselves” as they are supposed to exist in a “real” world, i.e., a world that is thought to “be” prior to and independent of the knower’s cognizing activity. This was the general view before Kant, and, in spite of his well-founded reservations, it still is the general view among psychologists and the bulk of other scientists. The one significant exception are the great physicists who, during the first third of this century, demolished the naive realism of classical science and revolutionized the physicist’s picture of the universe.
The traditional view of cognition, as Maturana (1970) says, begs the basic epistemological question, because it tacitly presupposes both a reality to be known and the knowing organism’s possibility of discovering it. Inevitably, any view that conceives of “knowledge” as a representation, image, or replica of structures that are there prior to the knower’s act of cognizing runs into an unsolvable problem: how could the knower’s representation ever be said to reflect, or correspond to, or approximate, reality if the only access he can possibly have to it is his very own activity of “knowing”? To establish the “objective truth” or “validity” of such a representation, some form of noncognitive comparison or match with that supposed independent “reality” would be indispensable. Such a comparison, however, is doubly impossible, first because there can be no way of comparing what we have experienced to what we have not experienced, and second because – whatever else it might mean – “to know” means to cut up experience into repeatable, recognizable segments (cf. Brown, 1969), and again, there can be no way of comparing these segments, and the structures we build out of them, to the uncut and, therefore, unstructured whole of our experience.
Experience and subjectivity
Many scientists – just as craftsmen, engineers, and technicians in general – can carry on with their jobs perfectly well without ever asking a question concerning the relation between the perceiver and the perceived, or between the knower and what he comes to know. They can do so as long as the items and events they are concerned with are all contained in their fields of experience. In that case, items, and the relations between them, can be considered the “causes” of other events or states, which then become their “effects.” Both the causes and the effects will be accessible to experience, at least in principle. That is precisely what is required by theories of “verification” or “falsification” as they have been advanced by logical empiricists{[Note 1] } and by Karl Popper (1934/1965: 33). Inferences drawn from experience are confirmed or disconfirmed by further experience. From an epistemological point of view, it is important to stress that confirmation may make an inference or hypothesis more plausible, or valid, or even more useful, in that it shows it to have held over a larger population of experiences. This may lead us to make some other inferences about how we structure our experience, but it cannot possibly justify inferences about the ontological status of the experienced items and the relations we have posited between them.
If existential inferences are proposed, they are always, explicitly or implicitly, causal inferences. What we experience, they suggest, is to be considered the effect of something that “exists” prior to our experiencing it. The causal connection, in this case, is posited on the basis of “analogy” to what we do within our experience. In order to systematize experience and to make predictions about it, we are constantly inferring causal links between segments of it, which, after Hume, we may call events A and events B. An existential inference, however, requires that we consider the elementary segments of our experience as Ds, i.e., as effects caused by Cs that are prior to our experience and, therefore, lie outside it. Clearly, this is not the kind of “generalization by analogy” that scientific methodology sanctions. Analogy can be invoked, if we believe that C is like A and D is like B (Kaplan, 1964: 107). In this particular case, we never have been, nor will we ever be, in a position to say what an item of that supposed pre-existing reality is or is not like, before we have actually experienced it, and because, afterwards, all we can possibly know is what we have experienced, the generalization of the causal connection is wholly illegitimate. Thus, though we may intuitively believe in a structured ontological reality, to invest it with the power to “cause” our experience is no less an act of mythical faith than to invest it with the power to punish or reward us or to protect us from the things we happen to dislike.
Organism and environment
When, however, a scientific investigator does ask the question concerning the relation between the perceiver and the perceived (or, more generally, the relation between the experiencer and what he experiences), he introduces an element that wholly changes the situation. If he observes an “organism,” i.e., an item that he invests with the capability of perceiving and “behaving,” he has already discriminated that item from the rest of his, the observer’s, experiential field. Thus, he can meaningfully speak of the organism’s “environment” when he wants to refer to all that part of his own experiential field that is not the organism. The relations he then sets up between the organism and its environment will still be relations between parts of his, the observer’s, experiential field. Anything he then says, or might want to say, about the organism’s experiential field must necessarily be in terms of (or derived from) his own experiences as an observer. This does not mean that the organism’s experiences must always be in every way the same as the observer’s. They may be composed of different coordinations of elementary segments of experience and/or they may be made up of a smaller set of segments; but, as far as the observer is concerned, he cannot meaningfully include in his analysis or description of the organism elements that are not bits of his own experience.{[Note 2] }
This limits the observer’s explanation and understanding of an observed organism to what Maturana (1970) has called their “domain of interaction,” a limitation partially foreshadowed in Klüver’s (1965) doubts concerning “stimulus equivalence.” Within the domain of interaction, however, the observer may speak of causes and effects, placing the causes in the observed organism’s environment as “stimuli” and attaching the effects to the organism as “observable responses.” By and large, that is what behaviorism has led psychologists to do, and insofar as they have tried to stick to the rigid program, they have striven to remain “technicians,” i.e., to limit their explanations and interpretations of “behavior” to the linear cause-effect pattern of mechanics.
But because, in mechanics, there is no room for the transformation of energy into information, nor for inferential processes, their analyses of behavior could not come to terms with the very factors that make behavior essentially different from mechanical action and interaction. In virtually everything a psychologist may want to investigate, perception and/or cognition are involved, and the reason why we want to set these processes apart from the processes covered by mechanics is precisely their “informational” aspect. In cybernetic terms, that means that we want to specify relations between events not as energy transformations and by means of energy equations, but semiotically as signs and their interpretation (Rosenblueth, Wiener, and Bigelow, 1943; Haldane, 1955). In other words, the process of coding, which is a prerequisite wherever we speak of information, inference, or cognition, involves the institution of a connection (semantic nexus) that is quite independent of the physical or mechanical characteristics of the items that are being connected.
The interactionist view
The various “interactionist” theories start with the explicit assumption of two types of entities: organisms and environments. An environment impinges on everything in it and thus provides for “selection.” Certain items will last, others will not. Living organisms are seen as a product of this principle. They have been “selected for” because, whatever it is we call “to live,” it constitutes a specific way of dealing with certain aspects of the environment. What is alive is not only “adapted” as a category or species, but also has some capability of individual adaptation. In this view, then, perception and cognition are, on the one hand, capabilities that have resulted from adaptation by phylogenic selection and, on the other, instruments of ontogenic adaptation. If this is granted, we would seem to be justified in inferring that an adapted organism “reflects” its environment – at least in the sense that the organism’s biological structure, being one that has manifestly survived, gives us some clues as to the structure of the organism’s environment. Similarly, the results of perception and cognition, insofar as they can be shown to enhance an individual’s adaptation, could be said to “correspond” in some sense to the structure of the environment. In short, with regard to the basic problem of epistemology, we would seem entitled to say that, though our knowledge of the existing world may be limited and somewhat distorted by our specific characteristics as organisms, we can acquire some knowledge of the real world.
There are two snags to this generalizing conclusion. The first we have dealt with above: what we conclude about organisms and their environments, when both are within our experiential field, can in no way be generalized to our own position, since our environment, i.e., an independently “existing” world, is not accessible to us except through our experiencing it. Hence, whatever we say about organisms we observe cannot serve as a logical analogy to ourselves qua organisms in an intuitively posited “real world” environment. It can be no more than a mythical metaphor.
It does not help either, to present the problem in crisp terms and then to say, as Attneave (1974) recently did, that its solution is the business of epistemologists and that he, the psychologist, prefers to go on believing that knowledge does reflect an “existing” reality. Such an act of faith is a rationally unfounded assumption and will sooner or later interfere with the investigator’s logic of theory construction and, hence, also with his interpretation of “data.”
The second snag is that the terms “selection” and “adaptation” are somewhat misleading. The theory of evolution from which they originate is essentially a cybernetic theory, in that it is based on “restraints” rather than on causation (Bateson, 1972: 399 ff.). “Selection” means no more than the elimination of the nonviable, and an “adapted” organism is merely one that comprises no feature, biological or behavioral, that would significantly increase the probability of the organism’s elimination (prior to its having procreated others). Adaptation, therefore, can never be said to reflect the structure of the “real” world, because, even if it did, we could not possible know it. Similarly, from maladaptation and extinction we can at best infer some organismic features that seem incompatible with the extinguished organism’s environment as we see it. That is, we may gather indications as to what that environment does not allow. But a description in negative terms cannot be turned into a positive description, because the exclusion of some possibilities, in a field of infinite possibilities, does not make that infinity finite.
The trouble with the “interactionist” view is that, in fact, it presupposes much more than the two explicitly made assumptions. The generalized separation of organism and environment takes for granted an objectively existing space in which the two can be separated. Any assertion of action, by the environment on the organism or vice versa, takes for granted the objective existence of time in which these actions and their “selective” effects take place. Even if we are not prepared to follow Kant to all his conclusions, we cannot avoid admitting that he has successfully deprived time and space of ontological reality and turned them into characteristics of our particular way of experiencing (Anschauungsweisen). This is corroborated, on the basis of very different studies, by the physicist Schrödinger:{[Note 3] } “The space-time continuum must not be assumed to be conceptually prior to what has been considered its contents; just as the corners of a triangle are not there, before the triangle.” Hence, if space and time are prerequisites of “interaction,” an interactionist view or theory can never legitimately prove the existence of an environment or of a “real” world, because it has a priori posited the spatiotemporal framework without which the separation of organism and environment, and their interaction, are meaningless.
The contribution of feedback and control theory
The birth of cybernetics as a discipline in its own right can be collocated with the publication of the paper by Rosenblueth, Wiener, and Bigelow (1943) that supplied a scientifically usable and, indeed, useful definition of “purpose.” Future historians may explain why, for 30 years and more, the behavioral sciences disregarded that formidable conceptual conquest, while the builders of mechanical and electronic control devices, who adopted it at once, proceeded from one empirical success to another.
In the context of this chapter, the relevant aspect of feedback theory is the epistemological one. The heart of the theory is the concept of negative feedback. A system is considered to incorporate a control device if it serves a homeostatic function in that it is capable (within certain practical limits) of keeping a sensory “input” signal quantitatively close to a pre-established reference value. Thus it must comprise a comparator (the material implementation of this is irrelevant) that “compares” the sensory signals to the reference value. If this comparison yields a difference (negative feedback), this difference generates an error signal that, in turn, activates an “output” function, i.e., triggers an activity apt to result in the modification of the sensory signal (adequation to the reference value).
Such a feedback control system is a highly successful and versatile model for the analysis of the behavior of living organisms (Powers, 1973). It differs from the stimulus-response model in that it posits a circular rather than a linear cause-effect chain. While S-R theory attempts to link an organism’s responses to specific stimuli, in the feedback model, on the one hand, it is the difference between a sensed quantity and the reference value that “causes” an active response and, on the other, it is this response that “causes” a change in the sensed quantity and, thus, the elimination of the error signal.
For an analysis of cognitive processes, this development of feedback theory has several important implications. If behavior can be seen as directed towards the adequation of sensory signals or percepts rather than aimed at the modification of an “outside” environment, the process of learning and the structures (motoric or cognitive) that result from it, are no longer dependent on any match with an “outside” world, but only on the experientially “correct” association of effector functions with subsequent changes in sensory signals. In other words, the logically impossible match with an objectively existing ontological reality is no longer necessary and the organism can now be seen and described as operating wholly inside its own field of experience, i.e., operating towards internal equilibration.
In this model, the concept of “knowledge” is in one important respect radically different from the concept of knowledge in traditional philosophy, science, and common sense thinking. It explicitly excludes any idea of “reflection of or “correspondence to” an independent objective reality supposed to “be” or “exist” in itself and awaiting a perceiving and/or cognizing organism to discover it. Instead, “knowledge” now designates any structure, i.e., any coordination or association of individually recurring elements of experience, that an organism uses in the ordering and systematization of its experience. Genetic epistemologists should have no difficulty in agreeing with this. Piaget has said it often enough: knowledge is whatever an organism holds invariant in the changing flow of experience.
The concept of knowledge in genetic epistemology
In Piaget’s view, a living organism begins its intellectual or cognitive career with some genetically determined “reflexes.” A reflex, by the observer’s definition, is made up of a stimulus and a response, both of which recur contiguously and in that sequence. From the regularity of their recurrence, the observer infers a “fixed” connection between the two. But what would it look like from the organism’s point of view? The stimulus, simply enough, would have to be a sensory signal or, more accurately, a constellation or pattern of sensory signals. The response, analogously, would be a signal or constellation of signals going to an effector function, i.e., a kind of signal that even the most introspectively inclined organism does not claim to be aware of. Again Powers (1973) has supplied an extremely insightful formulation: to the cognizing system everything is input. This means that an acting system can register, record, or define its activities exclusively in terms of signals from proprioceptive or kinesthetic sources and their experiential coordination (Hume’s contiguity) to subsequent changes of sensory signals. Important though this point is, it is not what interests us here. The point is that, although the terms “reflex” or “fixed action pattern” focus attention on the coordinated or patterned character of the activity, they do not make equally clear that the stimulus that triggers the activity usually also has the character of a coordination or pattern. Thus it would perhaps be better to define a reflex as consisting of a fixed connection between a specific perceptual pattern and a specific motor pattern (i.e., a scheme).
This detail is essential if we want to understand the Piagetian terms of “assimilation” and “accommodation” in an epistemologically consistent way.
Any new acquisition consists of assimilating an object or a situation to a previous scheme by thus enlarging it…. the thumb stimulus releases the sucking response only if it assumes a significance as a function of the scheme of this response, that is, if it is assimilated as a sucking object. (Piaget, 1974: 69-70)
Assimilation of the thumb as a “sucking object” means no more and no less than this: there is a pattern of sensory signals that, owing to a genetically fixed connection, triggers the sucking activity in the infant; from the observer’s point of view, this pattern, although first elicited by the nipple of the bottle or the mother, is now elicited by the infant’s thumb as well. At that level, and with reference to the scheme we call “sucking reflex,” there is, for the infant, no difference between nipple and thumb. A difference can be made only when some other sensory signals, for instance the taste of milk, are coordinated with the original sensory pattern – and such a novel coordination, resulting in a more comprehensive recurrent pattern of sensory signals, would constitute an accommodation. If we ask how this accommodation comes about, we can at once say it is induced by the very fact that, when the sucking response is triggered by the thumb, it does not lead to perception of the taste of milk, while if it is triggered by the nipple, it does; i.e., it creates a mismatch or “disturbance.” The disturbance can be eliminated either by cutting the link between the “thumb” percept and the sucking scheme, or by differentiating two sucking schemes: one that leads to perceiving the taste of milk, and one that does not. In either case, what has to occur is another accommodation. Accommodation can thus be seen as the result of discrepancies or disturbances that arise from acts of assimilation when the assimilated item in some way modifies the outcome of the scheme into which it was incorporated. In the above example, the perceptual “thumb” pattern will eventually be coordinated with other perceptual signal patterns from tactual, visual, and proprioceptive sources, and will thus come to constitute a “permanent object,” i.e., an item that is accessible and constructible in more than one sensory mode.
Although the observer of all this can meaningfully speak of “inside” and “outside” with regard to the observed organism, because he can separate the two in his own field of experience, the organism itself “assimilates” and “accommodates” and, quite generally, operates and acts exclusively on and with the signals that constitute its experience. The organism has no way of matching these signals or coordination patterns of signals with any outside reality, nor does it need anything beyond the sufficient availability of signals in order to continue with its construction.
“At eighteen months to two years this ‘sensorimotor assimilation’ of the immediate external world effects a miniature Copernican revolution” in the infant, and, as Piaget (1967: 9) goes on to say:
at the termination of this period, i.e., when language and thought begin, he (the infant) is for all practical purposes but one element or entity among others in a universe that he has gradually constructed himself, and which hereafter he will experience as external to himself.
Piaget’s meaning seems perfectly clear in this passage. The “Copernican revolution” that leads to the child’s externalization of his sensorimotor constructs, which include not only “permanent objects” but also the sensorimotor basis for the construction of the concepts of space, time, and causation, is Piaget’s key to the adult’s belief in an external “prefigured” reality. This act of externalization has to be constantly kept in mind if a logically consistent epistemology is to be gleaned from the Piagetian theory of cognitive development. Whenever we read, for instance, that the organism’s cognitive activity is “to assimilate the external world into the structures that have already been constructed, and secondly to readjust these structures as a function of subtle transformations, i.e., to accommodate them to external objects” (1967: 8), we must remain aware of the fact that this “external world” and the “external objects” it comprises are the sensorimotor constructs that we ourselves have externalized when we underwent the Copernican revolution at the age of 18 months.
If this interpretation is not altogether mistaken, Piaget’s concept of knowledge is one that in no sense involves the idea of getting to know an ontological reality that would have to be imagined as a prefabricated, fully structured world, existing by itself and waiting to be “discovered” by a cognizing organism. The concept of knowledge he has developed in his genetic epistemology seems wholly compatible and congruous with what a cyberneticist would call “knowledge” in a control system. In the feedback model it is the structures or coordinations of signals – neuronal events, if you will – that the system has come to coordinate recurrently in its effort to eliminate “disturbances,” i.e., negative feedback that creates error signals because it is different from the reference value. For Piaget, as well as for the cyberneticist, the structures or coordinations may consist of sensory signals in perceptual patterns, of effector signals in motor patterns, or of both in complex action patterns or schemata. Assimilation and accommodation, thus, take place when inconsistencies or contradictions arise through the interaction of different levels of constructs; for instance, externalized objects and specific action patterns. This interaction of cognitive constructs is, of course, not at all the “interactionist’s” interaction because it does not involve a “real” outside world. Although the observer of an organism can place that organism into an “environment,” he cannot derive ontological conclusions from this way of organizing his experience, because both the organism and its environment are parts of his own experiential field and this can contain no more than he himself has constructed and externalized.
In another paper (von Glasersfeld, 1974), using a cybernetic metaphor, it is argued that from the experiencer’s point of view, ontological reality is like a “black box,” in that he has no way of discovering what “is” and how it might be structured. Here it should be stressed once more that radical constructivism does not deny the existence of a world, but it does deny the possibility of rationally describing such a “real” world. The cognizing organism would have no way of determining or deciding whether or not its constructs in any sense reflect the structure of a “real” world, even if it could come up with structures that are not dependent on its concepts of space and time. The fact that our perceptual and conceptual structures are repeatable and seem at least to some extent dependent upon the availability of elementary segments of experience (out of which they are composed) in no way justifies the assumption that our structures reflect, or correspond to, “existing” structures; it merely shows that these structures have a relative degree of viability given the way in which we happen to have segmented our experience. If three dots are made on a blackboard in front of a person, an opportunity is created within a domain of interaction for the person to experience three “bits” that, in this context, may be called “elementary segments of experience.” The person is free to use these three bits in order to construct what we call a “triangle,” i.e., to connect them by performing a set of specific operations. Buthe is equally free, if he happens to be in a “poetic” mood, to construct out of them a horizon and the morning star. Neither of these constructions can be said to “exist” on the blackboard. Indeed, the three dots could be coordinated into innumerable other structures, and each one of them would be determined by the operations the person, the experiencer, performs, and only in a trivial sense by the three dots that, in this metaphorical example, constitute the raw material.
More complex and better examples are found in the history of physics. It is there that radical constructivism fully agrees with Kuhn’s idea of paradigm shifts (Kuhn, 1970), interpreting paradigm shifts as “change in the habitual way of constructing.” The more often a construct is repeated and the greater the number of larger structures in which it is involved, the more indispensable it becomes and the more “given” or “objective” it seems. If this is the case on the level of the natural sciences, it is a thousand times more so on the level on which we construct and externalize our everyday environments.
Finally, then, it has to be stressed that radical constructivism, like any other theory in science or common sense living, is a product of inductive inference. It does not purport to supply a model of the “world of being;” it merely hopes to develop a model for the rational organization of our experience that may make it a little easier (than do traditional models) to sort out what we can know and what must remain the unfathomable mystery of the world in which we find ourselves living.
The work that led to this paper was supported in part by the Mathemagenic Activities Program, Follow Through (C. D. Smock, Director) under Grant No. OEG-8-522478-4617 (287), HEW, Office of Education, and the Department of Psychology, University of Georgia. The opinions expressed in the paper, however, do not necessarily reflect the position or policy of the U.S. Office of Education, and no official endorsement by the U.S. Office of Education should be inferred. Comments and criticisms by C. D. Smock, John Richards, and Stuart Katz are gratefully acknowledged.
Attneave F. (1974) How do we know? American Psychologist 29 (7): 493–499.
Bateson C. (1972) Steps to an 14: Eulogy of Mind. New York: Ballantine.
Brown C. S. (1969) Laws of Form. London: Allen and Unwin,.
Glasersfeld E. von (1974) Jean Piaget and the radical constructivist epistemology. In: C. D. Smock and E. von Glasersfeld (eds.) Epistemology and Education (Report 14) Athens, Georgia: Mathemagenic Activities Program, Follow Through. http://cepa.info/1324
Haldane J. B. S. (1955) Animal communication and the origin of human language. Science Progress 43: 383–401.
Kaplan A. (1964) The Conduct of Inquiry. Scranton, Pennsylvania: Chandler.
Klüver H. (1965) Neurobiology of normal and abnormal perception. In: Hoch and Zubin (eds.) Psychopathology of Perception: 1–40. New York: Grune and Stratton.
Kuhn T. S. (1970) The Structure of Scientific Revolutions. 2nd Ed. Chicago: University of Chicago Press.
Maturana H. R. (1970) Biology of Cognition (Report 9.0) Urbana, Illinois: Biological Computer Laboratory, University of Illinois.http://cepa.info/535
Piaget J. (1967) Six Psychological Studies. New York: Random House.
Piaget J. (1974) The Child and Reality. New York: Viking.
Popper K. (1965) The Logic of Scientific Discovery. New York: Harper Torchbooks. (Originally published in German, 1934.)
Powers W. T. (1973) Behavior: The Control of Perception. Chicago: Aldine.
Quine W. van-O. (1963) From a Logical Point of View. New York: Harper Torchbooks. (Originally published 1953.)
Rosenblueth A., Wiener N. & Bigelow J. (1943) Behavior, purpose, and teleology. Journal of Philosophy of Science 10: 18–24. http://cepa.info/2691
Schrödinger E. (1956) Die Natur and die Griechen. Hamburg, Germany: Rowohlt.
Cf., for instance, Quint. (1953/1963: 41): “The factual component of statements insist, If we are empiricists, boil down to a range of confirmatory experiences.”
We may observe an organism and conclude that it is “color blind,” i.e., that the elements in our own experience that we call “color” are not part of the organism’s experience. Or we may conclude that an observed organism directly experiences elements that to us are accessible only indirectly via transformation from one sensory mode to another by some instrument, e.g., some of the sounds auditorily perceived by bats or dogs. But we cannot conceive of elements of experience to which we cannot accede either directly or indirectly.
Schrödinger (1956: 24): “Das Raum-Zeit-Kontinuum darf nicht als begrifflich früher angesehen werden als das, was bisher sein Inhalt genannt wurde; so wenig wie etwa die Ecken eines Dreiecks vor dem Dreieck da sind.”
Found a mistake? Contact corrections/at/cepa.infoDownloaded from http://cepa.info/1340 on 2016-11-21 · Publication curated by Hugh Gash