Brokaw, Galen. Introduction. A History of the Khipu. Cambridge: Cambridge University Press. 1-27. Introduction Traditionally, writing has been considered a major benchmark in the development of human societies: its appearance marks the boundary between history and pre-history, on the one hand, and the corresponding disciplines of history and archaeology, on the other. In more contemporary scholarship, however, the term “pre-historic” does not have the same currency as in the past, because it often implies a qualitative deficiency that is no longer politically correct. In colloquial use and increasingly in academic discourse, “history” is normally conceived as a more general term referring to past events regardless of whether or not they are documented with written texts. Nevertheless, the existence or absence of writing in a given society has inherent implications for the methodological approaches available for investigating the past. Archaeologists can certainly apply the methodology of examining the material remnants of human cultures to a period after the advent of writing. In fact, this is a very productive endeavor that attests to the fact that written documentation can never tell the entire story. The historical method of reading and analyzing documents, however, is restricted to periods in which written documentation exists. In the investigation of societies with a form of writing that is no longer in use and the knowledge of which has been lost, efforts to decipher their writing systems and to study their texts present a unique challenge that often attracts researchers from various disciplines: archaeologists and historians, of course, but also linguists, art historians, and anthropologists as well. When the system of writing is tied to verbal language, as in the case of the Maya syllabic script, researchers have a natural tendency to develop and articulate their projects in interdisciplinary terms: thorough archaeological research of the Maya area, for example, now demands training in Maya language and epigraphy. Writing systems whose conventions are either partially unknown or not tied directly to verbal language open up an interdisciplinary space with less-defined methodological constraints. The freedom of this interdisciplinary space can be extraordinarily productive for stimulating theoretical reflection, but it also has its limitations. Even partial ignorance of the underlying principles of a writing system means that decipherment projects are 1 2 Introduction hindered not only in the analysis of written conventions but also in the very recognition of conventions as such. These cases are doubly problematic, because they raise the question of what constitutes the threshold between writing systems and other forms of media. Nowhere has this issue been more controversial than in the case of Andean societies and their use of the knotted, colored cords known as khipu (also spelled quipu). A khipu, which means “knot” in Quechua, is a device of knotted cords used by the Incas and other Andean cultures to record various types of information. Throughout this book, I will refer to the Andean string device in both singular and plural forms as khipu. Although I recognize, as Tristan Platt has pointed out, that the Aymara also employed knotted cords that they called chinu (Platt 2002), they are part of the same larger Andean tradition. I spell the term “khipu” rather than “quipu,” not for any ideological reason nor to give priority to one dialect over another, but merely because this has become the more common spelling in recent scholarship.1 Although museums and private collections around the world preserve hundreds of khipu, much about this device remains unknown. Khipu cords are normally made from cotton or camelid fiber. The basic structure of a khipu includes a main cord, often displayed horizontally in museum exhibits, to which are attached any number of vertical pendant cords. In many cases, pendant cords also have their own attachments, normally called subsidiary cords. These subsidiary cords, in turn, may have their own subsidiary cords, and so forth. In most cases, the number of subsidiary cord levels is limited to one or two, but in theory a khipu could have any number of such levels. Some khipu also exhibit top cords, which are similar to pendant cords, except that they extend in the opposite direction. These top cords serve to summarize the information of a group of pendant cords with which they are associated through proximity or attachment. The colors of khipu cords include all the natural hues available in the cotton or wool itself as well as a number of colors produced using dyes. Cords may be either solid or a combination of two or more colors using various different methods to produce distinct patterns. Any given cord on a khipu, including in rare cases the main cord, may also contain knots. Although khipu exhibit a few uncommon or idiosyncratic knots, in general, they employ three types: (1) a knot tied in such a way that it creates a figureeight pattern; (2) simple overhand knots; and (3) long knots created by 1 In the colonial period, the term was always spelled “quipu” or “quipo.” The disadvantage of this spelling in English language publications is the tendency for those unfamiliar with Spanish to pronounce it KWEE-POO. In the past, some scholars have also spelled the term “kipu.” The aspiration indicated by the “h” in “khipu” reflects the way this word is pronounced in Cuzco and areas to the east and south (Alan Durston, personal communication). Introduction 3 wrapping the cord around itself normally from two to nine times and then pulling the end through the loops. The only dimension of the khipu that has been deciphered is a decimal system, documented thoroughly with archaeological khipu for the first time in the early twentieth century by Leland Locke. Locke demonstrated that the knots function in a decimal place system to convey numbers in a relatively unambiguous way. According to this system, knots and knot groupings appear at different positions along the pendant, subsidiary, and top cords. These positions correspond to decimal values. The lowest position records the value for the single units or the “ones.” Following an empty space, the next knot or grouping of knots corresponds to the tens position, the position after that signals hundreds, and so forth for higher powers of ten. Any position left blank indicates a zero value for the decimal power to which it corresponds. The lowest position, which corresponds to the single units, contains only one knot, either a figure-eight knot or a long knot; and these two knot types normally do not appear in any other position. The figure-eight knot signals a value of 1, and the various versions of the longknot indicate values two through nine according to the number of turns in the knot. All other positions may contain anywhere from one to nine overhand knots grouped closely together. Each overhand knot indicates a single unit of the decimal value of the position in which it appears. Two overhand knots in the tens position, for example, would correspond to a value of twenty.2 Many khipu, however, appear to violate this system in one way or another: figure-eight knots and long knots, for example, may appear in positions higher than the single units. Urton argues that such khipu are extranumeric: that is to say that they convey other types of information such as narratives (Urton 2002c; 2003: 55, 97–98). Another possibility is that such conventions record multiple numbers on a single string. Since Leland Locke documented the khipu decimal system in the early twentieth century, most research on the khipu has focused on the material conventions of this medium, its semiotic capacity, and the related debate about whether or not it constitutes a system of writing. Pioneering work by Carlos Radicati, Marcia and Robert Ascher, and more recently Gary Urton and Frank Salomon has greatly enhanced our understanding of the materiality of the khipu and many features of its conventional use. Here, 2 Using only the figure-eight knot and the long knots in the single units or “ones” position helps avoid the possible ambiguity of even decimal units (e.g., 20, 30, 100, etc.): if the last knot or knot grouping on the cord is a simple overhand knot, then you automatically know that the value of the “ones” position is zero. This is often useful because the actual position of each decimal power across a khipu can vary somewhat. 4 Introduction I do not propose to analyze directly the material conventions of the khipu but rather the history of this medium. The nature of khipu conventions is a fascinating and important question, but it has had a tendency to displace equally important and interesting questions about its history. Nevertheless, any historical investigation into the development of a record-keeping system inherently must discuss at times the process through which material media convey meaning. Given that we still know relatively little about khipu conventions and even less about what gave rise to them, the discussion of this process will necessarily often remain at a fairly general level. Throughout this book I employ the term “semiosis” and its adjectival form “semiotic” in order to refer in a general way to the transmission of meaning. The only alternative would have been “representation,” which I use in reference to iconographic modes of semiosis but otherwise try to avoid. In addition to being somewhat awkward in certain contexts, the notion of “representation” brings with it a great deal of conceptual baggage that can interfere in any attempt to understand the nature of non-Western media in both their synchronic aspects and their diachronic development. The history of the khipu can be divided into at least two distinct periods: the first, from its origins through the Spanish conquest, and the second from the conquest through the present. Each of these periods poses different questions and calls for different theoretical perspectives and methodological approaches. Accordingly, this book is divided into two sections. But the analysis of any form of communicative medium also raises the larger theoretical issue of the very way in which it is conceived. Before discussing the issues that arise in the historical analysis, then, it may be useful to make explicit the theoretical perspective that informs it. The Khipu and the Dialogic Model of Media Although this project does not focus directly on the conventions employed by the khipu to record information, the historical analysis of this medium requires a dialogue between an attempt to understand the semiotic conventions of the material object, on the one hand, and its historical contexts (social, cultural, political, etc.), on the other. What is at stake here is not the issue of whether or not Andean cultures had a form of writing that would make them “historical.” I find the continued use of the term “prehistory” highly problematic, but the distinction between periods in which alphabetic documentation exists and those in which it does not is certainly significant. Although we may have moved beyond a teleological perspective of history, writing is still considered a benchmark, and in many respects justifiably so. Most scholars agree that the development of extended, complex sociopolitical organizations such as states or empires is not possible without some form of writing. Introduction 5 However, pre-Columbian American states such as the Aztec and Inca Empires have always presented certain problems for this theory: they were sociopolitically and economically complex, yet they did not appear to possess a writing system. The Aztec case is relatively easily explained by their use of a form of iconographic script. The Inca Empire, on the other hand, is more problematic, because it is much more difficult to identify an Andean medium that qualifies as writing as traditionally defined. The most common solution to this problem, normally offered by scholars who do not focus on the khipu directly, involves a rather brief assessment of the khipu as a kind of anomaly, perhaps a form of “proto-writing” that somehow facilitated the development of a complex state. The interdisciplinary field of knowledge within which this type of investigation is normally carried out supplies terms such as “writing,” “literacy,” and “orality.” In most cases, debates about the nature of the khipu revolve – whether explicitly or implicitly – around the question of whether it constitutes a system of writing. At one level, this is a semantic issue that depends on the particular definition of writing that one adopts. In some cases, to insist that the khipu be considered a form of writing may be a necessary political strategy to counter ethnocentric perspectives that relegate societies without writing to an inferior position (Boone 2000:29–30). An even more radical approach, however, would be to refuse to submit to the terms of the debate. The concepts designated by the terms “writing,” “literacy,” and even “orality” originated in the particular historical context of alphabetic literacy and from the perspective of a literate mentality that has been unable to deal with the implications of other forms of semiosis. If the only two categories of society are those with alphabetic writing and those without, the Inca Empire does not fit into either of them. Researchers who seriously study the khipu and other non-Western media tend to recognize that they demand a reevaluation both of traditional historical and anthropological theory and of writing itself. I would argue that the problem presented by Andean polities, and the Inca state in particular, reveals a blind spot in traditional anthropological and historical theories of the relationship between writing and political complexity. The main weakness of such theories stems from the fact that they do not problematize sufficiently the concept of “writing.” Scholarship on writing abounds, but it tends to allow the cultural and historical determination of the concept to dictate the terms and parameters of the investigation. The problem is not merely that a universal concept of writing is difficult to define, but also that the notion of “writing” already imposes certain premises and biases that hinder such a project. The only truly successful attempt along these lines is Derrida’s recognition that the essential nature of writing resides in its iterability (Derrida 1974). To the frustration of many, however, iterability is also the essential feature of perception 6 Introduction and cognition in general, which means that if we follow Derrida, anything at all can be considered a form of writing. One might argue, then, that Derrida’s notion of writing renders the concept useless as a critical tool for projects not engaged in some form of Derridean deconstruction.3 However, Derrida’s work does not deny the possibility of making empirical distinctions between different types of writing. Nor does it deny the historical and anthropological importance of alphabetic writing in the development of modern societies. Rather, it calls into question the universality of this development and reveals that anything is potentially codifiable into a more formalized semiotic or communicative system. Certain universal characteristics of the human mind and the material world make some developments in communicative media more likely than others. Oral language is arguably a universal in human societies, but even in this case, all languages do not codify the available features of oral acoustics in the same way. Languages like Chinese and Zapotec, for example, make use of tones to determine literal semantic meaning, whereas most other languages do not. Even more important, the universal is not located in any specific feature of oral language or even in oral language itself but rather in the conditions conducive to its development. The same can be said of what I would call secondary media. All societies engage in a variety of communicative interactions through various channels. Here I draw a distinction between primary media, which inherently involve interpersonal contact such as speech or sign language, and secondary media, which do not. In other words, primary media depend on the presence of, or some form of contact between, the participants in the communicative interaction, whereas the communication made possible by secondary media may take place without such contact. Alphabetic writing, of course, would be an example of a secondary medium that does not require the presence of, or direct contact between, the parties involved. I do not wish to emphasize this distinction in any rigorous way. I realize that it is not sustainable in all contexts,4 but it nonetheless has important implications for the possibilities of social, economic, and political developments. This is because secondary media can store and transmit information over time and in most cases across space. The association between knowledge and power means that the use of such media has the potential, and perhaps inevitable tendency, to impinge on the socioeconomic and political landscape; the more versatile 3 Actually, to say that this concept of writing is absolutely useless serves to illustrate Derrida’s point. From a rigorously philosophical perspective, absolute uselessness amounts to the same thing as absolute usefulness. 4 The reason this distinction is not sustainable in all contexts is because the classification focuses on the material object rather than the practice associated with it. Thorough knowledge about any secondary semiotic system and its effects requires an understanding not only of the material medium but also of the way in which it was used. Semiosis does not occur outside of social practices. Introduction 7 the secondary medium, the more extensive its ability to store and manage various types of knowledge and hence its importance for socioeconomic and political development. This is not to suggest a causal relationship between the development of any particular form of secondary medium and particular socioeconomic or political changes: as I will explain in more detail below, these two domains are each inextricably caught up with the other. The point here is that formulating the issue in terms of secondary media without specific reference to writing attempts to avoid the problems caused by the conceptual baggage that accompanies the latter term. Moving from an emphasis on writing to one on secondary media is complicated by the fact that both the social sciences and the humanistic disciplines have had a tendency to dichotomize human societies into those that are literate and those that are oral. This dichotomy served as the original basis for the emergence of the field of orality-literacy studies, and to some extent it is still a dominant model in that field. In the early twentieth century, Milman Parry and Albert Lord inaugurated the field of orality-literacy studies with their pioneering comparative work on Homer and the Serbo-Croatian epic. Parry and Lord compared the features of the contemporary epic tradition to those of Homeric verse and concluded that the Homeric epics were originally oral compositions that had been set down in writing (Lord 1960). Subsequently this work gave rise to three related fields of study: (1) it generated a general interest in forms of oral literature, particularly poetry; (2) it served as the basis for the field of orality-literacy studies, which informed (3) the emergence of media studies and the Toronto School of Communication. The first field essentially engages in anthropologically informed literary research with a particular emphasis on poetry.5 Orality-literacy studies, on the other hand, focuses on the differences between oral and literate discourses as well as their cognitive, sociocultural, and political implications.6 The third field, media studies, also takes as its point of departure the theoretical implications of the historical transition in Greece from orality to phonographic literacy, but it also acknowledges that different forms of media correspond to different modes of thought with their own particular social and political implications. Unfortunately, since the 1960s these three fields have developed more or less independently. In so far as studies of oral literature are interested primarily in the features of specific oral discursive traditions, they would not necessarily benefit from the insights of the other two fields. Oralityliteracy studies and media studies, however, are both fundamentally based 5 See Foley 1981, 1987, 2002. 6 See work by Havelock and Hershbell 1978; Havelock 1963, 1982, 1986; Goody 1968, 1977, 1986, 1987, 2000; and Ong 1967, 1977, 1982. 8 Introduction on the theoretical argument that orality and literacy correspond to different modes of thought. The conceptual relationship between these fields has always been clear, but they have tended to focus on different contexts and to ask different questions. Orality-literacy studies tend to be historical and anthropological, whereas media studies deal with more contemporary sociological and technological issues. A research question involving communication in apparently “nonliterate” cultures such as the Inca and other Andean groups would normally adopt the critical and theoretical framework of orality-literacy studies. However, the lines of inquiry within orality-literacy studies that I am interested in here have remained locked for the most part within the binary opposition between alphabetic literacy and orality. Over the last forty years, for example, Jack Goody, one of the most prominent scholars in this field, has produced a series of books and articles developing various dimensions of this orality-literacy opposition and defending the premises of the field (Goody 1968, 1977, 1986, 1987, 2000). Such work has made significant contributions to our understanding of literacy in modern societies and of certain oral traditions. The theoretical model of orality-literacy studies works very well for understanding the nature of modern phonographically literate societies in contrast to those that do not employ such writing systems. However, it does not account for the function of other forms of media that are not recognized as writing. The orality-literacy dichotomy essentially homogenizes all societies without a medium that qualifies as writing (however this term is defined). It effectively defines “oral” societies in terms of what they are not rather than what they are.7 For this very reason Walter Ong rejects the term “illiterate” and uses “non-literate” instead (Ong 1987: 374). However, this problem is inherent to the oral-literate opposition itself. The analytical category of “oral cultures” obscures the fact that no society limits its communicative interactions to those that take place through oral language. The point here is not to equate other forms of media with alphabetic writing, but rather to recognize the way in which they function within the societies that employ them. If we maintain the comparison between “us” and “them,” the relevant opposition is not always between alphabetic literacy and orality but rather between alphabetic literacy and Mesoamerican iconography, alphabetic literacy and the Andean khipu, alphabetic literacy and Innuit pole carving, and so forth. If writing effects a cognitive transformation in the modes of thought of those who employ it, then it stands to reason that other dominant forms of semiotic or communicative media would correspond to different cognitive transformations. 7 Margaret Jackson makes this same argument specifically in reference to Moche iconography (Jackson 2008). Introduction 9 Orality-literacy theory, restricted as it is to the binary opposition indicated in its very name, is not able to address this issue;8 but this is precisely the type of question that media studies attempts to answer. Although media studies have focused primarily on the effects of modern electronic media, the fundamental theoretical basis of this field holds that the use of any given medium has particular personal and social effects (McLuhan 1994: 7). I would argue that this media-studies model, which acknowledges the transformative effect of all media, is more successful in resolving the problem identified by Ong of defining a culture in terms of what it is rather than what it is not. The application of this theoretical model to non-phonographic historical and anthropological contexts is more difficult, because typically the nature and type of communicative interactions that take place through non-phonographic media differ from those mediated by alphabetic scripts. Furthermore, the communicative functions of societies without a form of writing as traditionally defined tend to be distributed more evenly across a number of different media. In fact, this is one of the reasons why the emergence of alphabetic writing was so significant historically: it corresponded to a dramatic increase in the communicative interactions that took place through a single secondary medium. The transformation in modes of thought associated with alphabetic literacy are not due merely to the nature of the medium but also to the fact that this medium acquired such prominence, that so much semiotic activity came to be concentrated in it. Of course, the two are linked: the undeniable versatility of phonographic scripts lend themselves to use in a variety of functions and contexts, whereas most other traditional media are more limited. The nature of the medium, however, is only one part of the equation. Some scholars have argued that orality-literacy theory often gives too much credit to the role of writing, and this same criticism could be leveled at foundational media theory as well. Orality-literacy theorists such as Jack Goody and Walter Ong appear to discuss the role literacy plays in cognitive and sociopolitical transformations in causal terms: for them, writing causes transformations in thought, leads to political domination, and so forth. Ruth Finnegan argues to the contrary that the technological nature of writing or any other medium for that matter does not determine the uses to which it is put or the consequences that will follow (Finnegan 1981: 335–336; cited in Street 1987: 97). Brian Street identifies this causal argument, which treats writing as if it were an autonomous force in the transformation of society, as the autonomous model of literacy. In opposition to this autonomous model, Street proposes an ideological model of literacy, according to which the effects of literacy derive from its ideological use. For 8 For a cogent critique of the theoretical foundations of orality-literacy theory, see Biakolo 1999. 10 Introduction Street, literacy “is a social process, in which particular socially constructed technologies are used within particular institutional frameworks for specific social purposes” (Street 1984: 97). From this perspective, cognitive and social transformations often associated with literacy are results of cultural and ideological institutions rather than the technological features of the medium. It is unfortunate that the autonomous and ideological models of literacy developed in opposition to each other. They both offer interesting and valid insights for understanding the nature and effect of literacy. Many of the differences between these two models stem from the different contexts that they examine. The effects of alphabetic literacy in modern societies are the result of a long historical process in which literate technologies and practices developed in a dialog with the institutions that employ them. Literacy functions very differently in a society where it develops more or less organically over time as opposed to a context where it is introduced, adopted, or imposed, often in conjunction with political or economic imperialism. Here again, the notion of “organic development” is not meant to be overly rigorous. If, as Benjamin asserts, “there is no document of civilization which is not at the same time a document of barbarism” (Benjamin 1968: 256), then the development of writing always involves some sort of political and economic domination. The essential questions have to do with the nature and function of the institutions that employ writing and how they develop over time. Literate practices perpetuated by institutions of political control will naturally function differently than those developed by institutions of resistance, for example. One cannot generalize about the effects of literacy without taking into account such contextual differences. The technological features of a given medium are certainly conducive to certain types of use and certain cognitive transformations, but they are not restricted to those that manifest themselves in a particular sociohistorical trajectory. No universal laws determine the nature of that development: it is a dialogic process involving numerous variables, many of which we may never be able to identify. But among those variables both the nature of the medium and the ideological institutions that employ it figure prominently. I would argue that more adequate than the autonomous or ideological models of literacy, then, is a dialogic model of literacy that acknowledges the roles of both the technology of writing and the ideological institutions that develop and use it.9 Furthermore, in thinking about societies that do not employ a form of alphabetic writing, the media-studies model that I have proposed broadens the field by substituting “media” in place of “writing.” This implies a dialogic model not just of “literacy” but of media 9 This theoretical model explicitly invokes Bakhtin’s theory of dialogism (Bakhtin 1981; 1986), but it is also influenced by Heidegger’s onto-epistemology, elaborated most thoroughly in Being and Time. Introduction 11 in general: any given form of media develops in a dialogic relationship to the ideological institutions with which it is associated. The dialogic model of media does not attribute sociopolitical and economic transformations solely to a secondary medium or to ideological institutions but rather to both of them as well as to other factors in their dialogic relationship as it evolves over time. This historical process also influences the nature of the medium and its associated institutions. In other words, media and their ideological institutions are mutually constituting and interdependent. However, I have suggested that the most diagnostic historical phenomenon marking the cognitive, sociopolitical, and economic transformations normally associated with the development of complex polities is not the emergence of a medium like writing or institutions of literacy per se but rather the concentration of a large number and a high frequency of semiotic functions in a single medium, or perhaps in some cases a limited number of media. Alphabetic, and more generally phonographic, writing is clearly versatile, more versatile perhaps than any other known secondary medium. By versatility, I refer to a medium’s ability to take on different functions, to record different types of information, and so forth. Phonographic scripts certainly have their limitations. Other media are better suited for some types of semiosis, but the versatility of phonographic scripts gives them a much more dynamic role in the dialogic relationship within which they develop. I would suggest that this versatility is precisely what allowed alphabetic writing to support such a high concentration of semiotic functions. Differences between cultures, then, are evident not only in the type and number of secondary media that they employ but also in the way in which semiotic functions are distributed among them. In European societies, alphabetic script gained such importance, because the versatility of the medium supported the development of a large number and wide variety of semiotic functions, many of which were caught up either directly or indirectly in the exercising of political and economic power. The rise to prominence of alphabetic writing does not mean that other media disappear, but that their semiotic functions remain less explicit, less rigid, and less regulated by social norms relative to alphabetic writing. All societies employ multiple forms of media to one degree or another. Even in modern societies, in addition to oral language, writing, and visual media such as painting and film, a great deal of information is conveyed through clothing, hair styles, architectural structures, and so forth. The diversity of media in Andean cultures also includes pottery, textiles, architecture, and even the landscape itself. One might argue that the communicative function of such objects as clothing and architectural structures is merely incidental to their primary role of providing individual and collective shelter, protection, storage capacity, privacy, and so forth. They certainly do not appear to record the kind of knowledge that alphabetic writing 12 Introduction does. In theory it would certainly be possible for these media to encode information to the same extent as a medium such as alphabetic script, but their pragmatics and their inherent limitations constrain them from developing in this way. Nevertheless, in the Andes these types of media can, and often do, play a much larger role in social interactions than they do in modern Western cultures. Susan Niles has argued, for example, that certain Inca architectural structures served as a kind of enduring historical record (Niles 1999). The rise to prominence of one medium in relation to others may involve the transfer of functions to it from other media, but it also implies the emergence of new, original functions. In other words, the development of new media may affect the distribution of existing semiotic functions in addition to introducing new functions that develop in conjunction with them to create a new pattern of distribution. However, the greater prominence of a single medium such as alphabetic writing implies a reduction in the relative importance of other media and their semiotic functions.10 In most cases, societies that did not develop a phonographic writing system do not seem to concentrate as many semiotic functions in a single medium, but this does not mean that they distribute their semiotic activities symmetrically. The dialogic model of media leaves open the possibility that other forms of media might also support a high enough concentration of semiotic functions to create the potential for a dynamic analogous to that produced by alphabetic script in relation to its associated institutions. According to the dialogic model, the relationship between a medium, its institutions, and other factors both inform the process that leads to the concentration of semiotic activity in a single medium and are in turn affected by it. If the development of a certain level of sociopolitical and economic complexity depends on some form of secondary medium that supports a high concentration of semiotic functions and vice versa, then the existence of one implies the existence of the other. This line of reasoning leads William Burns to conclude that the Inca must have had a form of alphabetic writing, which he then locates in the textile designs known as tocapu and in the khipu (Burns 1990; 2002). The problem with this portion of Burns’s larger argument is not its form but rather the specific content of one of its premises. He assumes that only alphabetic writing is versatile enough to make possible the kind of complexity evident in 10 The rise to prominence of one secondary medium in relation to others in a society does not mean that the resulting distribution pattern of semiotic functions will remain static. Marshall McLuhan has argued that the advent of new electronic media in modern societies (radio, television) effected a new pattern of distribution that has profound social and political effects (McLuhan 1994). These shifts are as much due to socioeconomic, technological, and even political changes as they are to the proliferation of new forms of media, but this is precisely the point. These domains are inextricably connected. Introduction 13 the Inca Empire. The nature of the Inca Empire certainly may imply the existence of a secondary medium that would facilitate the administration of goods and services, but the assumption that this medium must necessarily be alphabetic is unjustified. According to the theory that I have outlined here, a sufficiently high degree of sociopolitical and economic complexity may imply the existence of a secondary medium; but it cannot determine a particular threshold of complexity nor predict the particular form of medium that will develop. To the extent possible, a history of secondary media must consider the nature of the secondary semiotic functions of a society and the way in which these functions are distributed among the various media that they employ; or to put it another way, the way in which a society deploys media to fulfill its semiotic functions. No credible evidence has surfaced to suggest that the khipu originally employed alphabetic conventions, but archaeological research and sixteenth-century historical documents make it very clear that pre-conquest indigenous Andean societies concentrated a significant number of semiotic functions in the khipu. Historical Research and “Pre-historic” Media Several major obstacles impede any investigation into the history of the khipu prior to the Spanish invasion. The historical or diachronic analysis of a secondary system of communication must posit connections between the semiotic conventions of the material medium, or at the very least its material features, and the objects and practices from which they developed. In the case of the khipu, the material nature of the textile medium did not lend itself to preservation. In contrast, the primary sign carrier involved in early Mesopotamian economic transactions that eventually gave rise to alphabetic writing consisted of dried clay tokens, which are extremely durable (Schmandt-Besserat 1992). There is a trade-off, however, between the more durable medium of dried clay and the more flexible and operable textile medium of the khipu. Clay could be inscribed only once before drying, whereas the cords of the khipu might potentially be coded and repeatedly recoded using knots. Although on occasion nonoperable sign carriers such as the Mesopotamian clay tokens might have been reused, in most cases they were probably either stored or discarded. Thus, the material nature of molded and/or inscribed clay imposes certain limitations on its use while at the same time insuring a significant and enduring archaeological record that provides evidence from which a history of writing can draw substantive inferences. Khipu, on the other hand, are made from cotton or camelid fiber, which deteriorates naturally over time, especially when not preserved in some fashion. For this reason, archaeological khipu come predominantly from the coast where the dryer climate is more conducive 14 Introduction to the conservation of these materials and almost exclusively from grave sites where they have benefitted from the careful storage and preservation procedures involved in burial customs. The fact that almost all surviving khipu come from grave sites poses a problem not only because it limits the number of archaeological specimens, but also because it may imply a generic limitation in the khipu corpus. Erland Nordenskiöld argued that ancient Peruvians would have interred only khipu with some kind of cosmological significance (Nordenskiöld 1925a; 1925b). Nordenskiöld’s premise led him to search for and find cosmologically significant numbers recorded in the knots of archaeological khipu, but his argument remains highly conjectural. Nevertheless, whether or not it is true that the semiotic content of archaeological khipu are limited to cosmological figures, it is certainly possible that only one type of khipu or set of khipu types would have been interred in a grave. Thus, we cannot assume that archaeological khipu recovered from graves constitute a representative sample of this medium. Another problem has to do with the fact that most of the khipu known or assumed to have come from grave sites have no specific contextual or provenance information that would possibly constitute clues or suggest directions for analysis. Many of the khipu located in museums were originally obtained and sold by looters who had no interest in keeping provenance records; and even those khipu with known geographic provenance are rarely accompanied by more detailed information about the specific site or other associated objects from the same grave. The late nineteenthand early twentieth-century archaeological expeditions that acquired the majority of such specimens were often more concerned with amassing a collection of artefacts for European and American museums than they were with documenting their excavations in ways that might have yielded valuable information about the objects they were collecting. The separation of khipu from their archaeological contexts also makes it difficult to establish a stylistic chronology that might inform an understanding of the nature of this medium and the direction in which it developed. The limited number of khipu specimens that have undergone radiocarbon dating has begun to shed light on certain aspects of the chronological development of this medium between the Wari and Inca periods. However, the relatively imprecise nature of radiocarbon dating, the rather short period of Wari and Inca dominance and the evidently rather rapid nature of khipu development will probably make it difficult to identify a more detailed developmental chronology based on stylistic and/or functional differences. The paucity of archaeological specimens, especially from the pre-Inca period, and the lack of contextual information make it difficult to identify and establish links to material precursors and their associated semiotic practices. Even in the case of Inca-era khipu, specimens of which appear to be the most prevalent and about which there exists the most information, Introduction 15 we have only a basic understanding of the decimal system that makes possible a numeric reading; but this is only one dimension of khipu semiosis. Moreover, the knots of many khipu, or in some cases sections of khipu, do not conform to the standard decimal conventions. It is possible that these anomalous khipu employ knots in a non-numeric way (Ascher, M. 2002; Urton 2002a, 2003; Urton and Brezine 2005); but even if this is the case, we currently have no way of knowing for sure what the specific nature of these conventions might be. Our ignorance of how the khipu functioned constitutes a significant obstacle in determining its historical development. Arguments about how the khipu developed are limited to inferences about its material relationship to other textile products, brief speculations about diachronic changes in material features, comparative analyses that consider possible analogies to the development of secondary media in other civilizations, and conjectural hypotheses that are difficult to confirm. Without an understanding of the nature of khipu semiosis or its relationship to other media in relation to which it developed, a history of this medium must make certain assumptions about the historical processes involved. Thus, several premises underlie the history of the pre-conquest period presented here. The first is that as a general rule, semiotic systems are not invented from scratch: they build on previously established cultural products and practices (Collon 1990: 14; Houston 2004: 234; Jackson 2008: 7). The emergence of any semiotic medium as complex as the khipu takes place through developmental processes of increasing sophistication that build on pre-existing practices and technologies. Second, as explained in the previous section, at the very least a general, if not universal, correlation exists between the emergence of a certain level of socioeconomic complexity and developments in secondary media; and this relationship may provide a basis for interpreting the significance of various material remains from the archaeological record. However, a corollary to this premise is that, as explained above, the development of secondary media takes place in a dialogic relationship to the development of social, economic, and political institutions. In other words, the question of which came first does not make any sense, and not just because it is difficult to tell which is the chicken and which is the egg. The chicken and the egg are both mutually dependent products of a long interrelationship, the origins of which would not be identifiable as either a chicken or an egg. Socioeconomic and political institutions and the secondary media through which they function are mutually determined. Moreover, the nature of any given stage of their development does not provide a basis for predicting further developments. Neither the specific nature of the relationship between socioeconomic institutions and their secondary media nor the specific form of their complexity is universal. Traditionally, historians of alphabetic writing have argued that writing originated with a form of mimetic pictography that eventually evolved into a system of abstract signs. In this model, the transition from mimetic 16 Introduction representation to phonography takes place through the discovery of the rebus principle according to which a pictographic or iconographic image is employed for its phonetic rather than its mimetic value. Denise SchmandtBesserat has argued, however, that pictography was an intermediate stage of development and that the long path eventually leading to the emergence of alphabetic writing began in Mesopotamia with economic record-keeping practices that employed clay tokens to represent quantities of a particular commodity (Schmandt-Besserat 1992). These Mesopotamian tokens are merely a more complex variation, if not a direct descendant, of the earliest known secondary medium of notched bone, which begins appearing in the archaeological record at least from the middle paleolithic (SchmandtBesserat 1996: 90–91). The long historical process that eventually gave rise to alphabetic writing, then, began with a simple numerical record-keeping system that arose in the context of economic institutions. As with the origins of alphabetic writing, the set of semiotic practices associated with the khipu would have emerged originally from the communicative and record-keeping needs of the economic and political institutions that motivated their development. Tristan Platt suggests that the khipu originated in the hunting and herding practices linked to the domestication of Andean camelids (Platt 2002: 226). Certainly, the context of the Andean economy was very different from that of Mesopotamia, but any form of economic exchange, partnership, cooperation, or reciprocity creates a context in which secondary media become very useful if not absolutely necessary. Ultimately, complex sociopolitical institutions are founded on, and developed in conjunction with, such economic activities. In fact, some scholars have suggested that as a matter of course, the numeracy involved in the numerical nature of early recording media has historically preceded other forms of literacy (Harris 1986: 133; Gaur 2000: 12). Thus, as SchmandtBesserat has argued, part of the history of alphabetic writing involves the transition from a primarily numerical medium to one that in one way or another builds in other kinds of semiosis. This is particularly relevant in considering the history of the khipu, because one of the few certainties that has been established is the numerical nature of this device. The fact that khipu were numerical in nature does not mean that they correlate to the early, numeric stage evident in the history of alphabetic writing in the Middle East and Europe. Again, the histories of secondary media do not all follow a universal trajectory. The development of both socioeconomic systems and semiotic media can take a wide range of possible paths. Traditional histories center on socioeconomic and political changes, but Harold Innis, Marshall McLuhan, and others have argued that communicative media play an important role in the particular nature of such developments (Innis 1951; McLuhan 1964). The approach taken here does not displace the socioeconomic in favor of an exclusive focus on media. Introduction 17 Rather, it attempts to identify the essential relationship between them. Social, economic, and political phenomena all take place through communicative interactions. Thus, an historical interest in media necessarily involves the communicative interactions of which they are a part. In order to emphasize this connection, the historical analysis in the first section of the book follows a chronological sequence in which Andean polities are discussed together with the media they employed. Although there has been very little radiocarbon dating of archaeological khipu, it appears that most surviving specimens are from Inca times. The Wari period, five hundred to one thousand years prior to the emergence of the Inca Empire, produced khipu with distinct material features, and archaeological specimens appear to reveal a link between the Inca and Wari khipu traditions. Currently we have no confirmation of the existence of khipu or khipu-like devices prior to the emergence of the Wari Empire in the first millenium CE, but various types of semiotic media of one kind or another were used. Even if we cannot identify the specific conventions employed in the early practices that eventually led to the late Inca khipu just prior to the Spanish invasion, we can identify in the archaeological record a general outline of different forms of social, economic, and political complexity that constitute the contexts within which Andean secondary semiotic media developed. Furthermore, the general features of early Andean media suggest intriguing interrelationships among themselves and with both Wari and Inca khipu. Andean Media Prior to the Spanish Conquest The first chapter of this study reviews the archaeological research on the emergence of early Andean civilizations and the implications for the development of secondary media. It then discusses in some detail the semiotic practices of the Moche evident in the fine-line drawings that appear on ceramics from the first millennium CE. The Moche arguably developed significantly more complex socioeconomic and political institutions than many of their precursors. The Moche also seem to have employed a number of different media. I argue that they distributed a number of semiotic functions across several, possibly interrelated, secondary sign systems. The most compelling of these systems was a mimetic style of fine-line painting that appears on ceramic vessels. Of course the fine-line painting itself constitutes a secondary medium, but these paintings also depict other media such as textiles, ceramic pots, and sets of inscribed beans. Although it is difficult to ascertain all the ways in which these media functioned, in some cases it is possible to make limited inferences. The various media of any given culture function in different contexts to convey different types of information, but in some cases an overlap occurs. 18 Introduction This redundancy appears to be particularly evident in societies in which semiotic functions are more evenly distributed across various secondary sign systems. Such redundancies may also facilitate the transpositioning of the semiotic function of one medium into that of another in their dialogic interaction with socioeconomic and political developments. By transpositioning, I refer to a complex relationship between two or more signifying systems by virtue of some commonality in semiotic function, which may involve common referents as well as certain semiotic conventions or principles.11 I would argue that this commonality often derives from the reworking of conventions in one medium that were originally developed in another. The pervasive nature of material media in everyday life means that mimetic arts inevitably transposition other media incidental to the mimetic operation. The iconography of Moche fine-line painting, for example, transpositions the secondary media of textiles, ceramic vessels, and inscribed beans, thus constituting a kind of meta-semiosis in which the object of representation is representation itself. I argue that this phenomenon is highly significant in the history of media, because it indicates a more self-conscious awareness of the nature of semiosis as such. In Chapter 2, I examine possible evidence of another kind of transpositioning in the Middle Horizon Period during the Wari and Chimu Empires. The specific nature of the Wari administrative state is not entirely clear, but Jeffrey Quilter argues that this is the first Andean polity to integrate large-scale religious, political, and economic activities; and it is at this point that the khipu first appears in an unambiguous way in the archaeological record (Quilter 2002b: 213–214). Middle Horizon Wari khipu are different from later Inca khipu in that their pendant cords are wrapped with colored thread to produce chromatic patterns (Image 14). Although in some cases, these khipu also contain knots, others do not. The knotless Wari khipu, then, appear to rely almost exclusively on the colored bands produced by the thread wrappings. These colored bands resonate in interesting ways with a colored checkerboard pattern that appears on ceramic vessels associated with the economic production of the Wari state (Image 13). This checkerboard pattern, in turn, suggests a relationship to the accounting device/practice known as yupana, which involves the manipulation of small stones, kernels of corn, or some other smallish objects (Images 5 and 6). And the yupana appears to be related to certain accounting practices 11 Julia Kristeva originally coined this term as an alternative to the notion of intertextuality, which had come to be understood in an overly simplistic way. Studies of verbal discourses normally limit the use of this concept to the identifiable relationship between specific texts, but it is much broader and more complicated than that. Kristeva explains it as “the passage from one sign system to another” (Kristeva 1984: 59). Kristeva did not have different writing systems in mind, but the use of the concept in this context is not formally inconsistent with the phenomenon she is attempting to describe. Introduction 19 evident in the archaeological investigation of Chimu storage facilities. These relationships are particularly important for understanding the history of the khipu, because colonial sources establish that the khipu was used in conjunction with the yupana, and both the Wari and Chimu influenced the development of the Inca Empire. The apparently non-numeric nature of most Wari khipu does not necessarily mean that the khipu originated as a non-numeric device. If the development of complex recording systems takes place gradually in a series of stages, then simpler numeric khipu practices would have predated the fully developed decimal system evident in Inca khipu. I would argue that the khipu did in fact originate as a simple numeric medium, probably in the context of the administration of economic activities analogous to those that motivated the use of clay tokens in Mesopotamia. The color bands of the Wari khipu may indicate the transpositioning of a color symbolism originally employed in the storage and accounting practices associated with the yupana-like checkerboard image on the Wari ceramic vessels and the Chimu architectural structures mentioned above. Prior to this point, the khipu would have been a comparatively more simple medium used in less centralized and hence less regulated contexts. In his analysis of changes in Chimu architectural structures over time, John Topic argues that the administration of economic resources shifts from a model of stewardship to one of bureaucracy (Topic 2003). This shift is signalled by the distancing of architectural accounting and control mechanisms from the resources with which they were associated. In other words, the material medium that conveyed and regulated certain types of knowledge acquired a greater level of independence from its referents. Based on the evidence currently available, the Chimu do not seem to have employed khipu, but the development of the Wari state also would have involved the emergence of a form of bureaucracy. If the shift from the Chimu stewardship model resulted from pressures exerted by an emerging bureaucracy, then analogous pressures in the Wari state may have inspired the adaptation of early khipu devices for use in more centralized administrative practices. In any case, the complex chromatic conventions and the fully developed decimal place system of the Inca khipu may be the direct result of the convergence between the numeric conventions of earlier knotted string records and the paradigmatic information structures and color symbolism used in the kind of accounting and storage practices evident in Chimu architectural complexes and the images on Wari ceramics. It is clear that the Inca Empire implemented the use of the khipu on a much larger scale than their Wari precursors, if for no other reason than that they controlled a much larger territory. The larger, more complex nature of the Inca state would explain the innovations in khipu construction that apparently occurred in the Late Horizon during which the Inca Empire 20 Introduction flourished. Many alphabetic texts and documents from the colonial period contain detailed information about Inca history and culture collected from, or in some cases written by, native Andeans. The criteria according to which native Andeans formulated their histories were very different from those of the colonial Spaniards, but even the alphabetic histories of the Incas produced by Spaniards in the colonial period provide valuable information that often can be taken together with archaeological data to produce significant insights into the history of the Inca Empire and the khipu medium on which it relied. Ironically, in some ways less needs be said about the khipu in the Inca period, because the importance of this medium and the uses to which it was put are so well documented in the colonial chronicles. The Inca clearly used the khipu to record numerical data related to demographics, tribute, and some form or forms of calendrics. Many colonial texts also refer to the use of the khipu for recording laws, rituals, and even histories. No real controversy exists over the idea that there was a relationship between khipu and these various types of discourse. The controversial question has to do with the nature of that relationship, that is to say the specific nature of the relationship between the material conventions of the khipu and its discursive rendering. The specifics of khipu semiosis is not the primary focus of this study, but the expansion of khipu record-keeping practices to such varied domains has important historical implications. In Chapter 3, I briefly survey what colonial chroniclers wrote about the Inca khipu, and I identify what may have been a certain historical memory of the expansion of khipu practices in conjunction with the expansion of the Inca Empire. I also propose a theory about the nature and development of imperial khipu historiography. Colonial accounts of khipu practices make it fairly clear that these devices were associated with various types of information: goods and personal belongings, censuses, laws, and ritual sacrifices and huacas. Numerous accounts also mention the khipu in relation to narrative histories. Each of the different categories of information types corresponds to a different official record-keeping functionary referred to generically as khipukamyuq. Many of the categories of information, such as commodities, censuses, and so forth, are inherently numeric in nature. Others, such as laws and lists of religious sites known as huacas, might have been incorporated into a numerical scheme, but they also would have involved other types of conventions. If narrative khipu were not merely simple mnemonic devices, as many have argued, then they certainly would have used a much more complex system of conventional signification. Either way, however, the different types of information recorded on khipu would have involved not only a difference in what was signified but also a difference in the nature of the signifying system itself. In other words, the Introduction 21 khipu appears to have been semiotically heterogenous, by which I mean that it employed different kinds of conventions in order to convey different kinds of information. Most complex media have a certain degree of semiotic heterogeneity. Even alphabetic script, which is based primarily on the principle of phonemic representation, also incorporates non-phonemic conventions such as arabic numerals, punctuation marks, formatting patterns, and so forth. We tend to think of the development of writing as a series of successive stages in which one set of practices and objects supplanted an earlier one. But this was not necessarily the case with the khipu nor with alphabetic writing for that matter. The conventional sophistication of the khipu may have developed at different levels of society, leaving previous practices in place. The innovations in khipu conventions carried out by state-level institutions for the administration of tribute, for example, would not necessarily have affected the conventions of pastoral khipu that had probably already been in use in one form or another for hundreds, perhaps thousands, of years. Thus, the various domains in which the khipu served to record information involved distinct genres that may have corresponded to different levels of khipu literacy with conventional values specific to each level, and in some cases with unique conventions.12 Numerical data may have been central to most, if not all, of these genres, but the significance of color, cord configuration, the non-numerical use of knots, and other non-numerical conventions would have differed from one genre to the next. Furthermore, different levels of khipu literacy may have exploited different material features of the khipu for conventional use. The Khipu in the Colonial Period It is important to keep in mind the nature of khipu literacy and its various genres or levels when attempting to understand colonial descriptions of the khipu. What we know of Inca khipu practices comes primarily from the reliance on these pre-Hispanic records by Spanish officials in the process of incorporating indigenous populations into the colonial system. After the conquest, information about khipu comes from specific social, economic, and political contexts that affected the way in which they were perceived and represented. The second section of this book traces the history of these contexts in order to document the continuity of khipu genres involved in the recording of demographic data, tribute payments and obligations, and historical narratives. It also identifies certain adaptations of the khipu to the new social, political, and religious contexts of the colonial period. 12 For a discussion of diversity and standardization of khipu, see Quilter 2002b: 200–204. 22 Introduction Immediately following the arrival of the Spaniards in the Andes in 1532, one of their more pressing ideological tasks was the determination of the legal status of the Indians. In theory, if they were natural lords, then they would not be subject to conquest.13 The determination of whether or not the Incas were natural lords required a historical investigation into the history of the Inca Empire. This was ironic for two reasons: first, the conquest was mostly over by that time, and second, the outcome was all but predetermined. Although the Spaniards would never have relinquished their control over the former Inca Empire, the historical investigation was still necessary in order to appease their collective conscience and to determine the status of indigenous nobility in the colonial order as well as the privileges of various different ethnic groups. Chapter 4 discusses the use of the khipu by Spaniards who conducted and wrote up these historical investigations as well as histories motivated by other concerns throughout the sixteenth and early seventeenth centuries; and these writers consistently reveal that their native informants employed khipu. After the Viceroy Francisco de Toledo’s definitive inquests of the 1570s, the number of histories produced as a result of official investigations decreases dramatically. However, the khipu continues to inform histories written by Spaniards who wished to gain fame or fortune and by mestizo and indigenous chroniclers interested in promoting their personal and political agendas. By the mid-seventeenth century, however, even these historiographical khipu disappear from the documentary record. Several factors contribute to this disappearance, first and foremost perhaps was the demise of this particular khipu genre. At the same time, seventeenthcentury historiography became more and more concerned with the history of the sixteenth-century colonial enterprise than the Inca past; and most of the seventeenth-century chroniclers who dealt with indigenous history relied primarily on written documents and earlier chroniclers, particularly Garcilaso de la Vega, rather than indigenous informants. The semiotic heterogeneity of khipu semiosis, the various khipu genres with their unique conventions, and the socioeconomic structure of the Inca Empire meant that khipu literacy was not an independent institution to the same extent as is modern alphabetic literacy. Khipu genres were dependent on the institutions that, in turn, depended on them. The dissolution of Inca political institutions, then, made inevitable the obsolescence of 13 A “natural lord” was defined as “a lord who, by inherent nature of superior qualities, goodness, and virtue, and by birth of superior station, attains power legitimately and exercises dominion over all within his lands justly and in accord with divine, natural, and human law and reason, being universally accepted, recognized, and obeyed by his vassals and subjects and acknowledged by other lords and their peoples as one who rightfully possesses his office and rightfully wields authority within his territory” (Chamberlain 1939: 130). Introduction 23 the particular khipu genres on which they depended. The historiographic projects of the Spaniards in the sixteenth century may have prompted the perpetuation of imperial Inca khipu historiography to a certain extent, but Spanish institutions would not have regulated these accounts the way the Inca had done. This may explain the rather idiosyncratic khipu that appear to inform several early seventeenth-century chronicles. In any case, whatever support that the colonial Spanish administration might have given to the institution of imperial khipu historiography incidental to its historical investigations was very limited, because the colonial regime provided no incentive or impetus for perpetuating this practice in any formal or regulated way. Administrative khipu studied in Chapter 5, on the other hand, were vital not only to the Inca but to the colonial Spaniards as well. Immediately after the conquest, the Spaniards were faced with the task of integrating the indigenous population into a colonial government. In most cases, this meant dividing the land up into encomiendas and charging the Spaniards to whom they were granted with evangelizing the Indians who lived there in exchange for the right to exact tribute. The process through which colonial officials granted encomiendas was supposed to involve an official visita or inspection that included a census of the population and an investigation of its productive capacity. But visitas were time consuming and expensive, and the turmoil of the early years after the conquest made it difficult to dedicate the resources necessary to carry them out in a thorough way. Whenever the Spaniards actually began such visitas, however, they discovered that the Indians already had khipu accounts of precisely the information they needed. In many cases, Spanish inspectors relied exclusively on khipu censuses in the determination of population figures. Khipu also provided data on both pre-conquest Inca tribute and the fulfilment of post-conquest obligations. Colonial records that transpositioned khipu accounts into alphabetic script make it evident that local record-keeping practices did not depend on Inca institutions. Local communities certainly produced khipu for their interactions with the Incas, but in many cases, khipu record keeping was already a vital part of internal community administration. The reducciones, which forced the population of a region to settle in an urban center, contributed to breaking down traditional socioeconomic structures, but many communities preserved the use of khipu in their internal administration. Colonial documents attest to these types of enduring khipu practices well into the eighteenth century. Frank Salomon’s work on the patrimonial khipu from Tupicocha indicates that in at least this case, community khipukamayuq actively employed khipu at least through the end of the nineteenth century and possibly into the early twentieth (Salomon 2004). 24 Introduction Whereas Chapters 4 and 5 trace the history of specific khipu genres or sets of genres in their relation to the colonial administration, Chapter 6 explores a more general perspective on this same relationship. This is necessary, in part, because the few studies that have been published on the history of the khipu in the sixteenth century have introduced or reinforced certain misconceptions. Building on earlier work by Pierre Duviols (Duviols 1971) and based on a very limited number of colonial documents, Carmen Loza has argued that the khipu passes through a series of stages in its relationship to the colonial legal system. The progression through these stages involves the gradual acceptance of the khipu by Spanish officials, a process of legitimation and adherence to the khipu, and later a rejection of this medium (Loza 1998a, 2000, 2001). Although Loza’s argument has a certain logic, it is based primarily on just four sets of documents and what I would argue is a misreading of both Toledo’s ordinances and the Third Lima Council’s allegedly universal condemnation of khipu issued in 1583, which are the primary bases for establishing the temporal limits of the most important historical stages in her model. Although Loza explicitly delimits her study to the history of the khipu in the colonial legal system, the implications of her argument extend to sixteenth-century khipu in general. This is not to take away from the valuable contribution Loza has made to our understanding of the khipu in the sixteenth century. Loza’s work has had a tremendous influence on the way other scholars, including myself, think about the history of the khipu. For this very reason, however, a critical engagement with her analyses is so important. Chapter 6, then, proposes a revision of Loza’s model in which the history of the khipu in its relation to the Spanish administration is much less tidy and without such clear-cut stages. One of the most common misconceptions regarding the history of the khipu in the colonial period is that in the latter part of the sixteenth century they were universally condemned by the Spaniards, sought out, and destroyed. This view has had surprising acceptance despite abundant evidence to the contrary. Most early claims along these lines were offered with no substantiating evidence. In La poesı́a quechua, Jesús Lara alleges that the Extirpación de la idolatrı́a del Perú (1621) by Joseph de Arriaga describes the destruction of khipu (Lara 1947: 50). Although a rather late work from the second decade of the seventeenth century, this is precisely the type of text where one would expect to find references to the destruction of idolatrous khipu. But I have been unable to find any such account in this text. Arriaga does record an episode in which a number of idolatrous objects are burned, but he does not list any khipu among them. Lara may have misread the term quepa [a kind of trumpet], which does appear in Arriaga’s list (Arriaga 1621: 94). Only a few pages earlier, however, Arriaga actually advocates the use of khipu for confession (Arriaga 1621: 89). Introduction 25 More rigorous scholars like Pierre Duviols and Carmen Loza have unwittingly associated assertions such as that made by Lara with a statement issued in 1583 by the Third Lima Council establishing a relationship between certain khipu records and idolatrous practices, and ordering that the khipu be destroyed (Duviols 1971: 243; Loza 1998a, 2000, 2001). In the historical investigation of phenomena for which there exists relatively little documentation, scholars have had a tendency to place an inordinate amount of weight on isolated pieces of evidence because they are often the only bases on which to construct a historical narrative and from which to draw conclusions. The assertion that the Third Lima Council issued a universal condemnation of khipu in 1583, for example, is based on a single statement taken out of its larger context. I argue that the Third Lima Council’s order was not a universal condemnation of khipu. In fact, the Third Lima Council itself explicitly advocated the use of khipu for confession. Of course, the Third Lima Council was a religious body focusing on religious issues, and its attitude with regard to the khipu was not entirely positive. Chapter 7 places the Third Lima Council’s order in its larger context and examines in a more thorough way the reaction to the khipu by Spanish priests. The fact that the Third Lima Council did not issue a universal condemnation of all khipu does not necessarily mean that there was not a widespread campaign in which many khipu were destroyed. The Council’s order probably indicates that the Spaniards had been destroying what they saw as idolatrous khipu for many years prior to 1583. The issuance of the order in the instructions formulated by the Third Council attests to the fact that they already knew about the idolatrous nature of some khipu genres and how to best deal with them. In fact, it seems that by 1583, for the most part the campaign against such khipu had already run its course. Thus, the Third Council’s order was most likely more a vestige from the early days after the conquest than an urgent call to arms. Those who would have been engaged in identifying and destroying idolatrous khipu would have been priests and missionaries; and their actions were not documented in the same way nor to the same extent as the visitas, especially in the early chaotic years during and immediately following the conquest. Even in those early years, it is clear that no general condemnation of the khipu was in effect, but Spanish missionaries may very well have been destroying “idolatrous” khipu at the same time colonial officials were drawing on khipu records in their visita inspections. The khipukamayuq who were the target of the extirpation campaign would have either abandoned their “idolotrous” khipu practices or begun hiding them from the Spaniards. Either way, this would explain the absence of any account of the destruction of such khipu in later colonial records. What sixteenthand seventeenth-century religious texts and documents do reveal is the widespread adaptation of the khipu for Christian ecclesiastical purposes. 26 Introduction These sources reveal a consistent record of support for such practices, even by extirpators of idolatry, the documentation of which begins possibly as early as 1560 and running at least through 1650. The history of all khipu genres presented in this book leaves off around this same time in the mid-seventeenth century, but this date is not intended to signal the temporal boundary of a stage or period in the history of the khipu. After 1650, tracing the history of the khipu becomes more difficult because known documentary evidence of khipu practices drops off dramatically at this point. This does not mean that after 1650 the khipu falls into disuse. The few references that are known make it clear that many Andean communities continued to use khipu throughout the eighteenth and nineteenth centuries and, in some cases, into the twentieth. The lack of known sources after the mid-seventeenth century does not even necessarily mean that there exists no documentation of enduring khipu practices in subsequent periods. In fact, I am fairly confident that many historical documents from the late colonial and early republican periods lying in the archives of Seville, Cuzco, Lima, and other Andean cities contain references to the khipu and that they collectively reveal a great deal about the history of this medium. Traditional methodologies of historical research carried out by a single individual do not lend themselves to the historical investigation of khipu record-keeping practices. The evidence of these practices consists of brief references dropped here and there in a variety of different documents dispersed among numerous archives. It is not difficult to find such references to the khipu as one peruses archival collections. Anyone who has spent any significant amount of time reading colonial Andean documents has come across them. But collecting single references here and there makes this endeavor prohibitively inefficient. The paucity of historical documentation for khipu practices after 1650 may be due merely to the fact that there has not been nearly as much historical research into this period and that far fewer primary sources have been edited and made available in print. Although the extensive research focusing on the sixteenth and early seventeenth centuries has uncovered numerous pertinent documents, surely many more remain undiscovered in this early period as well. In researching the colonial history of the khipu, I have discovered a few relevant sources by reading at random in archival collections, but by necessity the vast majority of the archival sources on which I rely were made known through the archival work of other scholars. Although I have consulted many of the original archival documents themselves, most of them have been published. What is needed now is the on-going collaboration of numerous scholars who are willing to take note of and share any references to khipu they come across as they work on other topics. To scholars engaged in researching other Introduction 27 issues, most references to khipu in colonial documents seem rather trivial and insignificant, but taken in conjunction with other sources they often constitute valuable clues that contribute to a more thorough understanding of this medium. It is very clear that many historically important sources lie unnoticed or unheeded in colonial archives. The history of the khipu presented here, then, must necessarily be a provisional account subject to revision and expansion as additional sources come to light. A book attempting to trace the history of the khipu, whether in the preHispanic or the colonial period, is in a certain way both premature and long overdue: premature because we still know so little about this medium; and overdue because the historical information that is available on this topic has never been presented in a unified or thorough way. The absence of extensive historical investigations of the khipu is not merely an unfortunate oversight. It has led to a tendency to implicitly dehistoricize the khipu and divorce it from its social, economic, and political contexts in ways that obscure both the possibility of constructing a history of this medium and the contribution that such an historical perspective might make in understanding khipu semiosis and possibly in devising decipherment strategies. Sociocultural and historical understandings generate perspectives that may have vital implications for our understanding of how the khipu worked and hence for devising possible methods, directions, or the very questions asked in decipherment projects. I would not suggest that constructing a history of the khipu will necessarily provide any immediate or easy solutions to identifying or deciphering specific khipu conventions, but in some ways it may shed light on the general nature of the medium as a system of secondary communication or artificial memory and possibly the principles on which it is based. However, regardless of whatever contributions a history of the khipu might offer to an understanding of khipu semiosis, it has a value in its own right as an essential dimension of indigenous Andean society.