A Calendrical and Demographic Tomb Text from Northern Peru.

The Region and

Department of Amazonas, in northeastern Peru

In general terms, khipusf which some 600 samples, mostly from coastal Peru, survive todayare composed of a main, rope-like cord (the "primary cord") to which are attached a variable number of spun and plied strings, called "pendant strings" (for an excellent overview of the structures of khipus, see Conklin 2001). Most pendant strings have knots tied into them in an arrangement that in many cases can be shown to have recorded numerical values in the decimal place system of notation used by the Inka (and presumably by the Chachapoya) for administrative record-keeping. Khipus were said by the Spanish chroniclers to have been used to record both numerical/statistical information as well as units of information recorded in some manner that we do not as yet entirely understand (see Ascher and Ascher 1997;Quilter and Urton 2001;andUrton 1998, 2002) that were consulted in reciting histories, genealogies, and other forms of narrative accounts. It appears to have been common practice in Inka and early colonial times to place khipus among the burial remains of the groups and/or of the khipukamayuqs to which they pertained.

The Chullpas of Laguna de los Condores

In 1996, a group of hacienda workmen, cutting trees in the heavily forested area around Laguna de los Condores, spotted a painting on a cliff face high above the lake. Making their way to the site, they found a half-dozen stone and mortar constructions, commonly referred to as chullpas (burial chambers), built into the overhang (Figure 2). Each of the chullpas contained numerous mummy bundles, totaling some 220 in all, as well as a wealth of grave goods, are also sometimes referred to by the term pachaqa ("one-hundred"), as well as parcialidad ("part"; see Espinoza Soriano 1981; Remy 1992:72-79; and Rostworowski de Diez Canseco 1981 :40). The Chachapoya region was brought under Inka control less than a half-century before the arrival of the Spaniards in 1532 (Espinoza Soriano 1967). The demographic make-up of Chachapoya was significantly affected by Inka imperial policies of population control. This was specifically the case with regard to the institution of the mitimae, which was the term used for the Inka practice of moving people from one place to another within the empire, primarily for economic and political motives. In regard to the Chachapoya area, the Inkas stationed garrisons manned by foreigners forexample, byWankapeoples from the southcentral highlands of Peru (Schjellerup 1997:69) within the territory, and they shipped portions of the native Chachapoya population to some 18 different locations around the empire, including a large contingent that was sent to live in the Inka capital city, Cusco (Schjellerup 1997: 66-69). In addition to the institution of the mitimae, the Inkas appear to have introduced into the Chachapoya region the full complement of imperial accoutrements, including the worship of the sun, the decimal administration of the population, and the knotted-string recording device, the khipu (Espinoza Soriano 1967:233-239).

The first Spanish entry into Chachapoya territory occurred under Alonso de Alvarado, in 1535. The Chachapoyans quickly allied themselves with the Spaniards against their former Inka overlords. The institution of the encomienda (a royal grant of patronage over a particular group of native peoples) was introduced at the beginning of the colonial era as the principal institution for the administration and exploitation of native Andean peoples. In 1538, Alvarado received the encomienda of the three parcialidades of Cochab amb a, Leymeb amb a, and Chilchos; these were three ethnic groups, each composed of multiple pachaqas, or ayllus, located in the middle and upper portions of the Utcubamba River drainage, as well as in the cloud forest between the upper Utcubamba and the Huallaga rivers. The latter region, which includes Laguna de los Condores, was the home territory of the nine ayllus that made up the Chilchos ethnic group.

Numerous visitas (fact-finding visits by colonial administrators) were made to the area of including pottery, textiles, pyro-engraved gourds, wooden sculptures, and a collection of some thirtytwo khipus.4 Although the men looted the site for a time which included their hacking open numerous mummy bundles with their machetes thanks to the quick action of Peter Lerche, a local historian and guide, most of the material from the tombs was eventually recovered (see Lerche 1999). The material was then placed in the care of the Peruvian archaeologist, Dr. Sonia Guillen, and her colleague, Adriana von Hagen. All materials mummies and grave goods were soon moved into a house in the nearby town of Leymebamba (Guillen 1999). The house was quickly transformed into a research facility, named Ce7atroMallwui ("CenteroftheAncestor"; see von Hagen and Guillen 1998; andWilford, NewYork Times, Dec. 16, 1997:F3).

Due to the fact that the mummy bundles and other grave goods originally found at Laguna de los Condores were partially plundered and badly disturbed at the time of their discovery, we do not have good information on the precise relationships between mummies and khipus. The loss of this information means that we have probably lost forever the opportunity to do a close analysis of the social organization of the disposal of the khipus that is, their distribution to specific individuals and groups of mummy bundles. Nonetheless, as one of only two archaeological discoveries of khipus for which we have (relatively) good provenience, its virtually complete aboriginal material cultural context, as well as ethnohistorical documents from the region written up by Spanish administrators from khipu readings provided by native record-keepers, the site of Laguna de los Condores represents one of the most important archaeological discoveries for advancing the study of the khipus since the time of the Spanish conquest.

The chullpas at Laguna de los Condores were utilized for disposal of the mummified remains of the dead during the several centuries-long period leading up to the Inka conquest of the region, through the Chachapoya-Inka period (ca. A.D. 145s1532) and on into the early colonial era. As for the prehispanic materials, the rich iconography adorning especially the textiles and pyro-engraved gourds found at the site suggest connections to cultures of the north Peruvian highlands and coast beginning as early as the Early Intermediate period (2000 B.C.-A.D. 700; von Hagen 2000). The continued use of the site for disposal of the dead during the early colonial period is confirTned by the presence among the burial goods of such items as a small wooden Latin cross and Spanish glazed pottery, as well as a variety of European glass trade beads.

Concerning the trade beads, during the summer of 1999, several different types of European trade beads were found inside one of the mummy bundles, which had been opened for conservation purposes. The beads were suspended on a spun cotton string that was attached to a braided cord; also suspended from the cord were several other items, including seven small, worked shell ornaments, two metal shawl pins (tupus), and an unidentified seed pod.

The glass beads included spherical red-white-bluegreen beads, as well as tubular beads of the types known as Nueva Cadiz Plain and Nueva Cadiz

Twisted. The latter two types of beads were introduced into Peru during the earliest years following the Spaxiish conquest (if not, like the germs of the conquerors, even somewhat earlier) until the early seventeenth century ( Indeed, it is for most of us, at least an arduous walk of an hour or two from Llaqtacocha around the lake to the foot of the cliff in which the tombs are located. Once at the foot of the cliff, one is faced with a difficult climb up the side of the cliff and then across a narrow trail to reach the tombs themselves. The difficulties of this excursion notwithstanding, it is nonetheless clear from the large number of dead buried in these tombs, as well as from the record of burial goods that includes both Precolumbian and colonial artifacts, that the inhabitants of Llaqtacocha and perhaps other neighboring settlements did in fact make this journey innumerable times over several centuries. What can we hypothesize about the possible significance and uses of the khipus for these people who were depositing and visiting their ancestors in the cliff face above Laguna de los Condores?

Khipus and the Dead

From the numerous published reports of previous discoveries of khipus, it appears that the final phase in the life history of the majority of these recording devices was burial, or interment with the dead. Reports of this form of disposal are found scattered throughout the archaeological and ethnohistorical literatures in the Andes. Mackey (1970) in particular has described several such cases from coastal Peru. In addition, the catalogs in the various museums where we today encounter large collections of khipus such as the Museum fur Volkerkunde, in Berlin (300 samples), and the American Museum of Natural History, in New York (100 samples) are replete with references to khipus recovered from grave sites along the Peruvian coast. The problem that is raised by this practice-particularly in those cases (like Laguna de los Condores) in which the remains of the dead remain accessible, rather than being buried in the ground is that of understanding and explaining the curious, indeed counterintuitive, habit of linking an active signifying object (a khipu) with a (passive) dead body. There is always some element of mystery associated with any object interred with a dead body. Tomb goods have significance in relation to their attachment to the life and deeds of the deceased, as well as to the interests, sentiments, and needs of the living (Rowe 1995:31-32; for an excellent overview of burial beliefs and practices in the prehispanic and colonial Andes, see Dillehay 1995). The practice of placing items of value with the dead may be relatively unproblematic for utilitarian objects, such as food vessels, hunting and artisanal equipment, etc. As with all items of material culture used daily by the members of a given society, such grave goods would represent the objects to which sentimental values were attached that were used by the deceased when living and that would continue to represent important objects of the memories and discourse of the living members of that society. These objects would be particularly powerful in their capacity to evoke memories of the living on those occasions when descendants of the mummies visited their ancestors, which was common practice in prehispanic and early colonial Peru (Doyle 1988; Isbell 1997; Rowe 1995; Salomon 1995). But what, in these terms, can we conclude about the significance of a device like a khipu, which not only had meaning as an object in the local system of remembrances and discourse, but which also had potential use as an active "signifier," a record that was susceptible of being read, interpreted, and even potentially of provoking action on the part of the living? What would have been the status of such a device in the local system of values and meanings, as well as in the systems of control and authority among the living?

To state in a direct way the position that I will adopt herein on the question of the disposal of khipus in open or accessible tombs, I think that in the prehispanic and early colonial Andes, the mummified remains of the dead were given custody of records which still had the capacity to communicate with the living because these were not just any dead bodies; rather, they were ancestors. In Andean ideology and cosmology, ancestors whose mummified remains were referred to by the term mallki were objects of great veneration and worship (see Doyle 1988; Salomon 1995). Mallkis were often kept in caves or in other built structures that afforded access to the ancestral mummies by the living (Bonnier 1993; Isbell 1997). As Doyle has noted in her excellent study of burial ritual and the ancestral cult in seventeenth and eighteenth-century central Peru, From the documentary testimonies it is apparent that cave machays [burial chambers] were almost always sealed with irregularly shaped, uncut stones, while the doorways of the other types of machays [e.g., subterranean and aboveground built structures] are rarely described.

The doorways were not designed to be permanently sealed because periodic access to the interior was necessary for the performance of ceremonies honoring the dead and the placement of new burials [Doyle 1988:110].

In prehispanic and early colonial times, the mallki of any particular social group, such as an ayllu or pachaqa, was regularly visited, given food and drink, redressed, and was asked for aid and guidance. We know a considerable amount about such activities, primarily because they form some of the central preoccupations of the "idolatry" campaigns, the investigations of idolatrous, "pagan" beliefs and practices that were carried out by Catholic clergy in the Peruvian countryside throughout the seventeenth and early eighteenth centuries. The objectives of the clergy were not only to investigate, but also to stamp out such practices and to destroy the objects of veneration, such as the ancestral mummies (see Mills 1997). The mallki were considered to be the owners and providers of all food (plant and animal) and the first owners of all agricultural fields, irrigation canals, terraces, etc.; as such, rituals and ceremonies were held to honor and worship the ancestors in the caves (mach'ay) where they were interred (Doyle 1988:68). In one very interesting document, it is reported that chicha (corn beer) was produced for ritual consumption at a mach'ay by the members of a group of ayllus that were considered to have been descended from a particular mallki. It is said that every household within each of the ayllus gave one ear of corn for making the chicha; a khipu account was created to see to it that everyone contributed (Doyle 1988:151).

Each mallki communicated with his/her living descendants through an intermediary, usually a member of the deceased's ayllu who served the mallki as its oracle (Doyle 1988:61, 117, 135-137). Mallkis were the most powerful objects validating the existence, history, social identity, as well as overseeing the well-being of the group of people descended from that ancestor. Whatever a mallki "said" through its diviner/oracle was considered to be sacred and true. By the same token, I would argue that whatever the grave goods (such as a khipu) that were buried with a mallki indicated would similarly have been taken by the living as the true, final, or at least the original/ancestral word on a particular matter.

Given the powerful position occupied by the mallkis in the life and thought of prehispanic and early colonial Andean peoples, how better than by worshiping the mummified remains of one's ancestors could one realize and validate the connection between an individual or a group and its past? I suggest that at the site of Laguna de los C6ndores (and no doubt at other chullpa sites as well) the interaction between past and present, realized in the encounter between a person and his/her mummified ancestors, was mediated in some cases by, among other things, the khipus. If this was the case, then I think this gives us grounds to argue for some rather specific traits that may have characterized the kinds of information encoded in these khipu "tomb texts" that were deposited with the dead. In particular, I hypothesize that the information inscribed in such texts would probably have concerned, or been relevant for, both the living and the dead that is, for the past and present. Such records probably would not have incorporated information that would lose its relevance over a single lifetime. Rather, the information would probably have been of a more enduring nature; for instance, it might have concerned the genealogical history of the group (e.g., an ayllu, or pachaqa) and its relations with its neighbors, as well as with supernaturals. To the degree that chronological, demographic, and other types of measurements may have been recorded on such khipus, they would probably have registered aboriginal, standardized, or ideal values or units (e.g., whole numbers, full decimal values, or powers of ten, etc; see Urton 1997). In these ways, khipu records, or what I am referring to here as tomb texts, would have served the living as standards against which to evaluate present conditions and from which to measure the changes that had taken place in the population, its organization, and perhaps its understanding of its own history and identity over time.

In regard to the latter point, I think we are warranted in making a comparison between the khipus placed in tombs like those at Laguna de los Condores with tomb texts found in other ancient societies, most notably among the ancient Mayans and Egyptians. In the former case, for instance, we have such remarkable examples as that of the Temple of the Inscriptions at Palenque in which the corpse in this case, Lord Pacal was placed in a tomb on whose walls were inscribed his dates of ascension and death, as well as other important dates of his lineage history (Lounsbury 1974; Robertson 1983:55ff.). As for the Egyptian tomb texts, we know that a whole class of such documents the so-called Pyramid Textswere inscribed on the walls of pyramid tombs for the purpose of helping the deceased royalty ascend to heaven and continue to rule there in the afterlife (David 2000:18-19; Hare 1999:65). The question raised by these comparative examples of tomb texts elsewhere in the ancient world is whether or not the disposal of khipus with the dead in the prehispanic Andes should be regarded as an indication that these documents had similar profound historical, religious, and perhaps other forms of significance for the people who placed them in the tombs. My hypothesis is that, in the case we are examining, the khipus did, in fact, have such significance for the Chilchos and other peoples who lived around Laguna de los Condores.

While my study of the collection of khipus from Laguna de los Condores is still in a preliminary stage, I will examine below the information recorded on just one khipu from this collection that, I believe, contains in its structure, organization, and apparent contents the kind of ancestral, regulatory text discussed above.

A Possible Biennial Tributary Khipu from Laguna de los Condores

The khipus from Laguna de los Condores, which are overwhelmingly made of cotton, range in size from samples having only a few pendant strings (one set of which was found inside the mummy bundle of a young woman) to several samples containing 60>800 pendant strings. In terms of its size, state of preservation, and the organization of information that was spun, strung, dyed, and knotted into it, the khipu alluded to at the end of the previous section is one of the most spectacular samples recovered from the chullpas at Laguna de los Condores (see Figure 3). I will refer to this khipu in the following discussion as khipu UR6.7

The these pendant strings are attached directly to the primary cord. The other member of each paired set is composed of 8, 9, or 10 pendant strings. These latter pendants, however, are not attached directly to the primary cord; rather, they are attached to a string whose two ends are themselves attached to the primary cord. These "intermediary" strings (i.e., as they stand between, and connect, the pendant strings and the primary cord) hang down below the primary cord in a slight arc, or loop. I will refer to these intermediary strings and the pendants attached to them as "loop pendants." There are 24 groups of these paired sets of pendant strings and loop pendants on khipu UR6 (see Figures 6 and 7). Before proceeding, I should note that in the calendrical analysis of khipu UR6 that I will present in this section, I will disregard the 32 pendant strings on this sample that do not conform to the regular arrangement of paired-that is: [20, 21, 22] + [8, 9, 10] --sets of pendant strings as outlined above. The excluded pendant strings are those that are circled in Figure 7. The knot counts on these excluded strings will come into play later in my analysis. I believe that these strings, which do not conform to the pattern of paired sets, may have been used, in some manner that I do not entirely understand at present, to provide flexibility in relation to one of the principal uses of this khipu-that is, as a biennial calendar.8 There are two clear indications of the calendrical significance of the organization of pendants strings on khipu UR6. The first indication is the number of strings in each of the paired sets; that is, if we take the mean value of the number of pendant strings in the 24 paired sets, we arrive at a (theoretical) repetitive combination of 21 + 9 (= 30) pendant strings (see Figure 5). This count is, of course, the whole number value closest to the number of days in a synodic lunar month (= 29.53 days). If each one of the 24 paired sets in khipu UR6 contained the meun number of strings (i.e., 21 + 9 = 30), this would produce a total of 720 pendant strings (30 x 24 = 720). Onehalf of this total would give us 360 pendant strings, a number which is close to an annual calendar count Figure 8 and summarized in schematic form in Table 1. In Figure 8 and

of 365 c Ways. However, when we count the actual numbers of pendant and loop pendant strings in the 24 paired sets (see Figure 7), we arrive at the number 730; one-half of this total count is 365, a number that coincides with the number of days in what is known as the "vague year" (i.e., the whole number of days in a solar year of 365.242 days). Thus, khipu UR6 appears on a superficial look to represent a calendar integrating 24 essentially synodic lunar periods into two (vague) year counts. The respective pendant string counts for these four half-year periods are illustrated in

labeled Year One (a) contains 179 pendant strings and the half-year labeled Year Two (c) contains 185 pendants. Now 179 is one-half day more than the count of the number of days covered by the perihelion; this is the period of time of the sun's (apparent) movement from the September equinox through the December solstice and back to the March equinox. The number 185 is one and one-half days less than the count of the number of days covered by the aphelion, which is the period of the sun's (apparent) movement from the March equinox through the June solstice and back to the September equinox (Aveni 1980:66). Thus, although such an interpretation would result in a biennial calendar of discontinuous, interlocking quarterly (i.e., four half-year)

periods, nonetheless, given the sums of pendant strings contained in the half-years la and 2c, we should allow for the possibility that the biennial calendar in khipu UR6 was reckoned in the pairings: 1 a + 2c (= 364) and lb + 2d (= 366), rather than of the groupings: la + lb (=362) and 2c + 2d (=368).

The main point to appreciate with respect to this discussion is that khipu UR6 appears to represent a biennial calendar count. We should pause for a moment to consider the rather unusual circumstance of finding a khipu recording a double, rather than a single, year count. It is relevant to note in this regard that the Inkas exhibited a strong emphasis on pairing and dualism in many aspects of their material culture and sociopolitical organization (for general discussions of this aspect of Inka culture, see Duviols 1973; Platt 1986; Urton 1997; and van de Guchte 1996). For instance, in terms of material culture, Cummins has shown convincingly in his studies of Inka keros, the Inka ceremonial drinking cups, that these vessels were always produced in pairs (Cummins 1988:124).

Although we do not have extensive documentation for an equal emphasis on dualism among the pre-Inkaic Chachapoya, there are certain data that do support the importance of this organizational principle in this region and in its material culture as well. For instance, we find a notable emphasis on paired artifacts in the material recovered from the chullpas at Laguna de los Condores. There are several pairs of pyro-engraved drinking gourds, each pair of which isrdecorated with virtually identical designs, found among the grave goods. I would note the very interesting ethnographic datum reported by Schjellerup in which she reports that, prior to 1960, villages in the region were paired for the purposes of maintaining the road system, the bridges, and the canals (1997:46). This information is particularly interesting in relation to the interpretation I will develop below regarding the two-year calendar of khipu UR6 as a record of labor service in the region in the late prehispanic period. Finally, Lerche detects not only a strong dualistic organizational principle in Chachapoya sociopolitical organization, but he also argues that when polities in the past were divided into dual groupings-one example of which is the dualism "Chilcho/Llaja" these entities were each further subdivided into nine subdivisions (Lerche 1995:5841). Thus, whether the calendrical khipu under consideration (i.e., UR6) was manufactured and used by the pre-Inkaic Chachapoyans or by the Inka administrators after their entry into Chachapoya territory, we ought not, in fact, be surprised to find that the calendar specialist(s) constructed a two-year calendar count. Therefore, at this first level of analysis, we can conclude that one of the messages the people from Laqtacocha (or elsewhere) would have been reminded of when they visited the tombs at Laguna de los Condores and took up khipu UR6 was a powerful representation of what I would term the ideal, or proper, structure and organization of time i.e., as a unit composed of two complementary, probably interconnected vague year counts.

As we will see below, the above interpretation has only scratched the surface of the calendrical information encoded in khipu UR6. However, in order to understand the additional temporal information that was encoded into the strings of khipu UR6, we must shift our attention to the knot counts that are contained on the pendant and loop pendant strings of this khipu. In the course of this analysis, we will identify another possible use of this knotted-string recordthat is, as an account of the organization of tributepayers in the region of Laguna de los Condores in late prehispanic times. another division is made between the two annual periods of the biennial calendar count; and c) sub-.divisions of the two yearly counts are made (as in Figure 8 and Table 1) to produce four half-year periods. I will first consider the calendrical information encoded in the knot counts illustrated in Figure 9.

The Significance of Knot Values in kkipu UR6

As is well-known, many (but by no means all)9 khipus incorporated a hierarchical decimal organization in the arrangement of knots tied into pendant strings such that there were different types of knots tied in clusters on different levels along the length of the pendant strings as place-value indicators of ls, lOs, lOOs, l,OOOs, and lO,OOOs (see Ascher and

The Calendrical Information in the Knot Values of Khipu UR6

When we study carefully the knot values tied into the pendant and loop pendant strings of khipu UR6, we encounter a rather stunning amount of calendrical information (Table 2). We are aware from earlier studies of Inka calendrics (see especially Zuidema 1977 and1989) that Inka calendar specialists had a particular interest in observing both the sidereal lunar cycle as well as the synodic lunar cycle, and in correlating these lunar periodicities with the solar year of 365.242 days. These calendrical periodicities, and the apparent interest in correlating them, are also evident in the Chachapoya-Inka calendar in khipu UR6. For example, we find in the knot values of khipu UR6 that the calendar specialists in Chachapoyas had recorded numbers that were very close numerical equivalents of sets of sidereal lunar cycles (= Ll in Table 2 This khipu also displays remarkably close approximations of multiple solar-year counts (= S in Table 2). These are seen in the periods of: a) 1,826, which is two-tenths of a day less than five solar years, and b) 730, which is one-half of a day less than two solar years. What is perhaps most striking about the calendrical periodicities encoded into khipu UR6 are those cases in which there are correlations of different lunar cycles with the solar year count(s). For example, the total knot count on the pendants and loop pendants of Year Two (= 977) is two and onehalf days more than 33 synodic lunar months and/or six-tenths of a day more than two solar years plus nine sidereal lunar cycles. One sees a similar correlation of a lunar (sidereal) periodicity and the solar region encoded into khipu UR6 is an accounting of tribute-payers that belonged to the ayllus or pachaqas in this region. This information is included primarily in the knot values of khipu UR6. That is, as we see in the knot counts (i.e., the decimal values) presented in Figure 9, the knot count forYear One totals 2,059 (i.e., 2,028 + 31 = 2,059), while the Year Two knot-count totals exactly 1,000 (i.e., 977 + 23 = 1,000). These two values give a total (biennial) knot count of 3,059. However, if we eliminate the three groups of pendant strings circled in Figure 9 (i.e., 31 + 23 = 54), which are those string groups not organized according to the pairing of pendant and loop pendant strings within the two-year counts, we end up with a total, biennial knot count of 3,005. As we will see below, this knot count is very close to the total number of tribute-payers that were counted in the first census of the population (i.e., the descendants of the mallkis at Laguna de los Condores) made in this region in early colonial times. Although there has not been extensive ethnohistorical work carried out concerning the early colonial population in the southern part of Chachapoya year count in the knot count on pendant strings in Year Two (= 446; see Table 2 for the calendrical correlation).

The organization of knot values tied into the pendants and loop pendants of this khipu sample provide the information for very complex and sophisticated calendrical reckoning and correlations. The calendrical interpretation of khipu UR6 presented here should be considered provisional. As further studies of additional khipu samples from Laguna de los Condores are completed, and as these studies are brought into relationship with studies of the historical documents from the region, it is hoped that we will arrive at fuller and more contextualized views on how and why certain astronomical observations were made, and calendrical calculations and correlations were performed, by the Chachapoya-Inka khipukamayuqs.

Khipu and Documentary Accounts of Tribute-Payers around Laguna de los Condores

The other type of information that it appears likely the khipukamayuqs of the southern Chachapoya territory around Laguna de los Condores, important colonial census data from the region are given in studies by Espinoza Soriano (1967), Lerche (1995), and Schjellerup (1997). We know that, during the brief period of Inka domination of the region, the Chachapoya population was organized into decimal units of tribute payers. Administrative oversight of the southern Chachapoya region was conducted from the Inka site of Cochabamba, located to the southwest of Laguna de los Condores (see Figure 1; see Schjellerup 1997:64).

In the Inka decimal administrative organization, the principal accounting units among the Chachapoya (as elsewhere throughout the empire; see Julien 1988 and Murra 1982) were groupings of chunka ("10"), pachaqa ("100"), waranqa ("1000"), and hunu ("10,000") tribute-payers. Subjects of the empire were required to pay tribute within these decimal units of accounting. As is well-known, in the Inka empire, the manner of "paying tribute" was through the performance of labor service for the state. This was undertaken in mit'a, "turns" of labor service performed by different tributary groups working in succession (see Murra 1982). Now, we learn from one document, which dates to 1572, that during Inka times, the cacique principal (the top political and administrative official) of the large administrative region just to the south and west of Laguna de los Condores was a man named Guaman. Guaman is described in this document as the "lord" (senor) of the three warangas (= 3000 tribute payers) of Cajamarquilla, Condormarca, and Bambamarca (Schjellerup 1997:315-316).

Soon after the Spanish entered the Chachapoya region, in 1535, Alonso de Alvarado was awarded the encomienda of Cochabamba, Leymebamba, and Chilchos (Espinoza Soriano 1967:299); this encomienda grant included the ayllus of Chilchos Indians who were removed to colonial reducciones ("towns") from the area around Laguna de los Condores. What is critical to note is that, at the time of the establishment of this encomienda grant, Alvarado is said to have taken census information from a local lord named Francisco Pizarro Guaman. As we learn from reading the colonial documents from this region carefully, Francisco Pizarro Guaman was the same man identified earlier (i.e., in the document from 1572; see above) as "Guaman," the Chachapoya-Inka cacique principal of the 3,000 tribute-payers in this region. Francisco Guaman is said to have pro-videdAlvarado with census data from khipu accounts in his (Guaman's) possession (Biblioteca Nacional Lima, A585 f93r; published in Espinoza Soriano 1967:299). Inthatcensus account, made in 1535, the total number of tribute-payers, or mit'a laborers, counted was given as three warangas i.e., 3,000 (see Schjellerup 1997:40,318).

The number 3,000, or three warangas, may have been a rounded-off, or "idealized," number, as we often find in colonial documentation pertaining to censuses and tribute records (see Remy 1992:72-79; and Urton 1997). Nonetheless, I find it remarkable and quite suggestive how closely the initial census count of 3,000 mit'a laborers approximates the numerical values encoded into the biennial calendar count in khipu UR6 from Laguna de los Condores.

To recapitulate, when we subtract the knot values tied into the circled pendant groups within each of the two "year counts" that do not conform to the calendrical organization in khipu UR6, we arrive at the total of 3,005 (= 2,028 + 977). I hypothesize that this value, and therefore khipu UR6 itself, referred to the total number of tribute-payers in the region of Laguna de los Condores, Leymebamba, and Cochabamba in late prehispanic (i.e., Chachapoya-Inka) times.

Conclusions and Questions for Future Studies

The important point to stress from the analyses given above is the coincidence between what I have interpreted as a) calendrical values, and b) census figures (for tribute-payers) on khipu UR6. The crux of my argument is that this was, in fact, a coincidence of numerical value that is, they were complementary, not contradictory. What I mean by this is that the calendrical organization of (Francisco Pizarro) Guaman's khipu would have provided the temporal pattern for the organization of mit'a labor service provided to the Inka state in the region prior to the Spanish conquest. The knot count pertaining to each paired set of pendant and loop pendant strings would have indicated the number of tribute payers that would have been responsible for performing state labor service in the region during one synodic lunar month over the two-year accounting period encoded in the strings and knots of khipu UR6.

Although it is hoped that the interpretation offered above in the form of a hypothesis has at least gone some distance toward explaining how and why khipu UR6 was constructed as it is, it is also clearly of the information encoded into khipu UR6 would have been subject to the (re-)interpretations of each successive reader, as well as each generation of readers. Each new reading of the tomb text, whether its substance was the same as or different from the last reading, would have been sanctioned by the presence of the mallki. recognized by the author that there are, in fact, many more new questions raised by this interpretation. These questions include the following: If khipu UR6 is, indeed, a biennial tributary record, then where does the first year begin? At the top or at the bottom of the khipu? At what point in the annual cycle does the calendar begin (e.g., at one or the other of the solstices? At one or the other of the equinoxes?)? Why, if this is a two-year calendar of state service that was performed by people in this region, does the record show that twice as many people worked in Year One as inYear Two?And finally, why were there such great differences in the numbers of tributaries who worked for the state from one month to the next? I cannot provide convincing answers to any of these questions at the present time. However, we are at least now asking questions of khipu records that have never been asked before, and hopefully some of these questions can help guide our future investigations of the knotted-string records in productive directions . . and 1n creat1ve ways. Finally, and as a corollary to the hypothesis articulated above with regard to the integrated nature of the calendrical and demographic information encoded into khipu UR6, I would further hypothesize that this khipu may have been the actual accounting khipu from which Lord (Francisco Pizarro*) Guaman supplied census figures to Alonso de Alvarado, in 1535. This latterhypothesis implies that the mummified remains of Francisco Pizarro Guaman may be among the 220 or so mummy bundles recovered from Laguna de los Condores.1l

When descendants of the mallki(s) at Laguna de los Condores visited the ancestor-custodian of khipu UR6, they could have taken up the khipu in order to remind themselves of the proper, or "standard," organization of time and the calendar, as well as the "appropriate," traditional division and organization of labor service to the state performed in the past by the populace in the region. During precolumbian times, these "messages from the mallkis" would have been of contemporaiy relevance, as a record of their standing labor service obligation to the Inka state. After the Spanish conquest, the accounting of time and sociopolitical organization contained in khipu UR6 would have served the people of the region as a historical document against which to measure and evaluate certain changes that had occurred in their world since the beginning of Spanish domination in the region. As in the reading of all texts, the reading Notes 1. Following Lerche (1995), I will observe the following orthographic conventions regarding the spelling of the name of the culture and geographical region under discussion in this article. Chachapoya will be used to refer to the people, or the ethnic group, as well as the prehispanic political unity (probably a chiefdom), and the culture under review here; Chachapoyas, as the name is spelled on most contemporary maps of the region, will be used for the name of the geographical region within which the prehispanic and early colonial Chachapoya ethnic group and culture were centered. 1994, 1998. 4. Concerning the condition of the khipus upon their discovery, according to Adriana von Hagen (personal communication, 1999), the khipus were found scattered among the debris left behind after the site was disturbed and looted. Several khipus that are now stored as individual samples were originally found tied together into linked bundles of khipus. These groupings were untied by Centro Mallqui staff as an essential step in cleaning and conserving the khipus. In addition, one khipu a (once) magnificient sample composed of some 266 dyed cotton pendant strings affixed to a carved wooden stick was washed in detergent by the wife of the hacendado whose workmen discovered (and plundered) the burial site at Laguna de los C6ndores. This latter khipu is now virtually completely white! I am currently working to construct an overview of the history of the khipus following their recovery. This account will be published in my complete study of the khipus from Laguna de los C6ndores (in preparation).

For archaeological studies of Chachapoya and Chachapoya-Inka sites in the region of Leymebamba and

Another

1978). Each khipu studied by the Aschers is designated in their report by an AS (=Ascher) number, thereby indicating to later researchers the source(s) of measurements and observations made on the khipu samples in question.

8. For previous studies of possible calendrical values encoded into khipus, see Nordenskiold 1925 and Zuidema 1989. 9. From my close studies of about 375 khipu samples in museum collections in Peru, Europe (Germany), and the U.S., I have found that approximately two-thirds of the samples in most collections have their knots tied in the decimal place notation system, thus strongly suggesting that the data on these samples were of a numerical-statistical nature. However, about one-third of the samples studied do not exhibit knots tied in a hierarchical, decimal-place notation manner; that is, these samples have "units" knots (i.e., figure-eight and long knots) tied above single knots, the latter of which, depending on their placement, usually designate increasingly higher powers of 10. In addition, there are examples of the so-called long-knots, which normally have a maximum of nine turns to the string (any greater value beyond which the long-knot should be replaced by a single knot, in the lOs position) that have up to 16 turns. Khipus of this latter type, which I have referred to as "anomalous" and "narrative-accounting" khipus, are described and analyzed in Urton 2001. 10. As I have reported in a study of the construction features of khipus in the American Archaeology collections of the Museum fur Volkerkunde, in Berlin (Urton 1994), khipu knots are tied in two different ways to produce either S-knots (in which the dominant axis of the knot runs from upper-left to lower-right (= \) or Z-knots (in which the dominant axis runs from upperright to lower-left (= /). 11. While admitting that it is possible, my colleague Keith Muscutt, who has spent many years exploring in the Chachapoyas region (see Muscutt 1998), doubts that Guaman would have been buried in the tombs at Laguna de los C6ndores. Instead, he favors as a final resting place for Guaman either the cliff-tomb burial site of La Petaca (which is nearer to Cochabamba) or other cliff tombs located around the lakes of La Sierpe and Mishacocha, at Atuen (Muscutt, personal communication, 2001).