Khipu Decipherment

KHIPU DECIPHERMENT by Gary Urton Introduc)on to the Problem of Khipu Decipherment In earlier studies, I and others have described and analyzed the principal physical proper)es of the Inka kno=ed-string recording device, the khipu (or quipu, "knot;" Ascher and Ascher, 1997 ; Chirinos, 2010; Urton, 2003, 2017; Figure 1). These studies have included analyses of string materials, cord structures, and colors, in addi)on to close considera)on of the principal modes of manipula)ng those various elements, such as the direc)onality of spin/ply, a=achments, knoRng; color use and pa=erning, and other construc)on features of khipus. Figure 1 - Inka Khipu (Centro Mallqui, Leymebamba, Peru; INC-LDC-108/LC1-497) Concerning the la=er, it has been established that a cri)cal feature of the produc)on and manipula)on of the strings of khipus was the repeated recourse to binarism (right vs. leY, S vs. Z, up vs. down, etc.; Urton, 2003; Clindaniel, 2018). These proper)es and processes were all under 2 the purview of the khipukamayuqs, the Inka administra)ve officials, in their produc)on of state records. It has long been presumed that these are the core features that cons)tute the elements of coded informa)on in the khipus (Urton, nd). Therefore, if any "deciphering" of the khipus is to take place, it will presumably be in terms of the transla)on into some target language (e.g., Spanish, English, etc.) of these features. In short, while researchers over the past century have succeeded in analyzing and interpre)ng the major structural and numerical proper)es of the khipus (Locke, 1923; Ascher and Ascher, 1997; Urton, 2003, 2017), what remains to be accomplished is determining whether or not these kno=ed-string and colorful accounts can be deciphered and, if so, how that might be accomplished. In this paper, I will take up the ques)on of the major challenges I think we face in deciphering the Inka khipus and what prospects I see for the success (or not) of these efforts. It will be helpful ini)ally to consider the nature and meaning of the act and process of decipherment. The Meaning of "Decipherment" By "decipherment," I mean the act and process of decoding a set of nota)ons and/or a script to the degree that one can accurately, verifiably and repeatedly render the encoded informa)on into a spoken or wri=en "transla)on" -- that is, so that one can speak and/or read the coded message(s) in some target language. Although we commonly think of decipherment as an esoteric act performed on some (oYen) ancient, coded script or recording system, it is important to recognize that this is, in fact, a mundane ac)vity -- one that we perform daily. 3 For instance, when I type the following line of text: "I am typing this text," a reader's ability to understand, interpret and speak what I have just typed, in order to report the significance and meaning of this line of text to another person, is an act of decipherment. You, the reader, are now deciphering this text -- that is, if you indeed know the English language and are able to "decipher" from the nota)ons I am typing by means of my computer keyboard an intelligible, interpretable message. It is in this generalized sense that decipherment is an on-going, daily process that most literate people engage in for much of our lives. My ques)on here then, again, is: Are the Inka khipus suscep)ble of being deciphered in the manner just described? My provisional answer to this ques)on, up-front, is: I doubt it, at least not in the way in which I just defined and described the ordinary, everyday process of decipherment -- i.e., decoding from some previously untranslated text the equivalent of a spoken narra)ve. Explaining my answer will require that we examine closely what kind of system of communica)on I think the khipu nota)ons represent and, therefore, precisely what kind of challenges I think the decipherment of the khipus would entail. Before moving to my current thinking on the ques)on of khipu decipherment, I must make a diversion to discuss briefly claims I made about this ma=er in an ar)cle published over 25 years ago (see Urton, 1998). In that ar)cle, en)tled, "From Knots to Narra)ves: Reconstruc)ng the Art of Historical Record Keeping in the Andes from Spanish Transcrip)ons of Inka Khipus," I argued that before the )me of the Spanish conquest of the Inka Empire, beginning in 1532, at least a por)on of Inka khipu records were probably characterized by the recording of informa)on in a rather fulsome, if not complete, script-like system of recording. That is, that the khipu nota)onal system was, at least in part, composed of fully gramma)cal, narra)ve-type informa)on. This 4 would (supposedly) have been in a manner similar to other ancient and modern wri)ng systems -- i.e., with the full panoply of nouns, verbs, etc., and the full gramma)cal entailments of subjects, objects, and other components of grammar. It was, I argued in that ar)cle, due to the domina)on and systema)c suppression of the Inka recordkeepers -- the khipukamayuqs ("knotkeepers/makers/animators;" Figure 2) -- by Spanish colonial officials, as well as what was a major Figure 2 - Khipukamayuq with assistant reading a khipu, with text reading: "le=er and knot [quipo recording] of the Inka" (Murúa, 1591:124) transforma)on in the types of informa)on that was recorded from before to aYer the Spanish conquest of the Inka Empire (especially in terms of a radical change in the nature of tribute) that transformed this supposed "original" script-like recording system into a significantly less than fully gramma)cal system of nota)ons. This change, I argued, led over )me to a reduc)on of the 5 previously fully readable (hence, ul)mately, poten)ally decipherable) wri)ng system into one that lacked full legibility in the post-conquest, Colonial khipu recording tradi)on. I will not discuss that earlier ar)cle in any detail here. The arguments that I made there, and then, had their own logic and merits. I leave it to the interested reader to evaluate the arguments made therein as he/she desires. Perhaps the arguments there were correct; I don't wish to reject or argue against them explicitly now. However, aYer a quarter of a century of further working with and thinking about the Inka khipus, I no longer think I can support those arguments. It is, therefore, in this revisionist spirit -- i.e., of having had many more years to reflect on these ma=ers -- that I now offer the arguments below, in which I cast doubt on the full decipherability of even the pre-conquest khipus, at least in terms of our ability today to produce fully gramma)cal, syntax-rich, readable, text-like narra)ve renderings directly from the khipu nota)ons. I have already discussed, in a provisional way, certain aspects of my current views concerning the non-fully gramma)cal nature of khipu recording in two previous book-length studies (Urton, 2017 and nd.). In this paper, I will explain what kind of nota)onal system I think the pre-conquest khipus represented, what encoding features appear to have existed in terms of their gramma)cality before the )me of the Spanish conquest and the effects of European colonialism, and whether or not and to what degree I think we might be able -- even par)ally -to decipher those messages today. The Language(s) of Encipherment 6 Two ini)al notes of clarifica)on are required here. One concerns the possible language(s) of any coded informa)on in the khipus; and the other concerns what circumstances might have caused Andean peoples to feel the need to invent a decipherable system of coding informa)on in some durable medium. The ques)on of the "readability," and thus the poten)al decipherability today, of the khipus seems to me to rest in the first place on what we know about the communica)on system(s) of any one of the three languages that are the best candidates for any gramma)cal messages encoded in the khipus. The three relevant languages are: 1) Puquina: the language that was commonly spoken around Lake Ti)caca (which is said in the Spanish chronicles to have been the original homeland of the Inkas), located to the south of the Inka capital city of Cuzco, which linguists have argued was the original language spoken by the Inkas before their arrival in Cuzco and even (as a "secret language") during their occupa)on of the valley of Cuzco; 2) Aymara: the probable language spoken in Cuzco, especially by Wari residents in the Cuzco region (the Wari having preceded the Inkas as the dominant ethnic/poli)cal group in the region; Bauer and Covey, 2002), at the )me the Inkas arrived in Cuzco from the south and the language which the Inkas adopted for local communica)on soon aYer their arrival in Cuzco; and 3) Quechua: the language which the Inkas adopted aYer their expansion out of the Cuzco region toward the northwest (in the direc)on of Chinchaysuyu), in the highlands above presentday Lima (i.e., the region of the original home of the proto-Quechua language); this was the language which served as their administra)ve lingua franca throughout the short life of the empire (on these linguis)c ma=ers, see Cerrón-Palomino, 2008; and Torero, 2002) 7 Presumably, any fully gramma)cal messages encoded in the khipus would be in one or another of these three languages. The ques)on then becomes: What do we know about narra)ve modes, styles and renderings in any one of these three languages before the interven)on of Europeans (Spaniards) into the Andean world? The central point regarding the above is that, since no known system of "readable" wri)ng was invented in the Andes before the arrival of literate Europeans,1 and since any wri=en texts or vocabularies in any one of the above three languages that we do have were wri=en under the oversight and influence, if not in the language, of the conquerors -- i.e., Spanish -- we have no indigenous, pre-conquest "guiding texts" in any one of the three languages men)oned above as a basis for determining the proper reading of fully gramma)cal messages that might be encoded in that language in the khipus. The nearest thing we have to such a guiding text is the so-called Huarochirí Manuscript (Salomon and Urioste, 1991). This is a text wri=en in the Quechua language near the beginning of the 17th century in the central highlands of Peru, with oversight and probable contribu)ons from a local Franciscan curate and priest (Francisco de Avila); this occurred long aYer the introduc)on of the Spanish language into the Andes. Thus, the Huarochirí Manuscript is the best poten)al "reading guide" we have to any fully gramma)cal, narra)ve messages in the khipus that may have been encoded in Quechua. However, the fact that the authors of this text would have been at least moderately influenced by Spanish grammar, narra)ve styles, and other such seman)c conven)ons -- as well as being at the )me under colonial domina)on -- means that I, at least, am dubious of the use of this text as a guide to reading any poten)al pre-conquest Quechua texts encoded in the khipus. It may indeed be helpful, at some stage in our efforts at 8 decipherment, but I doubt that it represents anything approaching an authen)c, full-blown preSpanish conquest rendering of an indigenous khipu-encoded narra)ve. Why Did the Inka Need a Recording/Wri)ng System? The second ma=er for considera)on in this discussion concerns the administra)ve, poli)cal and intellectual contexts in which the need for a system of recording informa)on, in some standardized, durable form, would have been needed by the Inkas. From an abundance of evidence provided by the Spaniards beginning in the early years following their conquest and occupa)on of Tawan)nsuyu ("the four parts in)mately united," the Inka's name for their empire), the informa)on retained in the khipus represented the communica)ve life-blood of administra)ve officials and the indigenous intelligentsia in what was the largest state of the PreColumbian Americas. It should be noted that the Inkas inherited much of their state apparatus, including the use of khipu-like kno=ed-string recording devices, from their predecessors in the valley of Cuzco, the Wari (see Urton, 2014). As I have argued previously (Urton, 1984), since the khipus represented the principal form of recordkeeping in the Inka state, we must presume that there would have been a need for a high degree of standardiza)on of records (e.g., census data, informa)on on labor demands for state projects and the mili)a, resources stored in state storehouses, etc.) within the Inka state administra)on. In my view, one cannot imagine a state-level society in which each recordkeeper invented and deployed his2 own, individual, unique system of recording informa)on --i.e., one that could not be reviewed, checked and surveilled by other (especially higher-level) state recordkeepers. Thus, the khipukamayuqs were the primary administrators and recordkeepers at 9 all levels of Inka state administra)on, from the court officials in Cuzco down to provincial- and poten)ally even local-level authori)es. I would note here that the need for standardiza)on across the Inka administra)on, at least in the later stages of the life of the empire, argues for the likelihood that the primary language that was used in any gramma)cal messages encoded in the khipus would most likely have been in Quechua, the shared, lingua franca of Inka state administra)on at the )me of the Spanish conquest. The above concerns and considera)ons are offered as two of the major issues that should be taken account of in our effort here to determine whether or not the khipus may be decipherable and what the nature -- linguis)c, gramma)cal, poli)cal -- of any principles and projec)ons for decipherment of the khipus need to take account of. A Hypothesis for the Encoding and Decipherment of Khipus As for what kind of recording system may be represented in the khipus, and, therefore, what par)cular challenges we may face in our efforts at decipherment, we also need to consider how any coded informa)on in the khipus might be cons)tuted, or composed. For instance, might those message have been -- and might they s)ll be (i.e., in the surviving ca. 1,050 extant samples in museums) -- composed in a manner similar to the kind of text the reader is reading here, and now? In my view, I see no reason to suppose that the Inkas (or the Wari, before them) would have conceived of the need for recording informa)on in extensive, narra)ve form, as the reader is reading here. The system I am currently using to produce this text -- both the le=ers on this page and the narra)ve, linear mode of represen)ng my thoughts on this page of paper -- was passed 10 down from centuries, actually millennia, of the development of wri)ng and narra)on in the ancient Near East and the eastern Mediterranean, prac)ces and processes which were subsequently adopted by the socie)es of western Europe. As far as I am aware, the Wari and Inka knew nothing of this; nothing like this form of recorded, linear, narra)ve composi)on was in their intellectual DNA nor in their developed systems of symbolizing and signing from pre-Wari and pre-Inka )mes. What we know both the Wari and Inka had in their inherited technologies of produc)on and communica)on were well-developed tradi)ons of working with co=on and camelid fibers to craY exceedingly complex fabric-based cloths of various shapes, sizes and degrees of complexity. These produc)on tradi)ons carried very complex, usually dyed, mul)-colored designs that were woven, s)tched or otherwise constructed (e.g., such as the tapestry tucapu designs) in those fabrics (see Stone-Miller, 1992). These things they knew, and performed, exceedingly well. Therefore, as these socie)es became increasingly more complex, leading to the emergence of state-level socie)es -- which the Wari and Inka undoubtedly represented -- there would have been an increasing need to record informa)on. The ques)on to consider, then, is: How would they have met this need? I begin by assuming that they would not -- certainly not necessarily -- have conceived of a narra)ve-based manner of recording -- that is, that they would have conceived of producing prose-like narra)ves about what was going on in their socie)es. What they needed to know and retain, for managing their states, were numerous, fact-based pieces of informa)on, such as: How many people live here and over there? How many households are people divided into? Who was the head of each household? How old were the inhabitants of each household? What kind and 11 what quan))es of goods -- crops, clothing, animals, etc. -- was each household capable of producing? How many males were of the appropriate age for military service? How many people were available to provide labor in the Inka system of tribute labor? Who showed up for a labor event, and who was absent? Who was responsible for the different sec)ons of an irriga)on canal? Who owned and used this piece of land and what were its boundaries? And finally, how did these values and iden))es change over )me? The la=er ques)on is the substance and essence of historical chronicles. However, such chronicles need not have been "wri=en" down (or out) as historical narra)ves. It would have been sufficient to maintain local archives that provided detailed accounts of these ma=ers from one year to the next. Such archives of khipu accounts have been retrieved, archaeologically, at such Peruvian sites as Limatambo (Díaz, 2008), and at Puruchuco and Lake of the Condors (Urton, nd.). A comparison of the informa)on retained in such archives over the course of years would have provided the informa)on for what we might term historical chronicles -- not narra)ve histories, but annota)ons of historical con)nui)es and changes. All of the informa)on prompted by the ques)ons raised above could have been accounted for by a system of knoRng strings in a hierarchical, decimal-based system of numera)on coupled with iden))es signified by certain combina)ons of material, binary construc)on techniques, and color. Such a system would have been perfectly adequate for recording all of the informa)on that was needed by state administrators. A gramma)cally more complex system of construc)ng/signing narra)ves -- explaining all of these things in a fully-gramma)cal, narra)ve form -- would have been superfluous, as their iden))es, quan))es and even histories (i.e., last year's quan))es compared to those enumerated this year) could be efficiently recorded and 12 represented for all )me on strings, knots, numbers and colors -- and all of this informa)on could be retained, for compara)ve/historical purposes, in khipu archives. If the Wari or Inkas wanted to tell stories, they could narrate them, orally; however, there would have been no reason, certainly not for state recordkeeping purposes, of recording narra)ve informa)on in the kno=ed records. AYer all, even in our own, modern-day states, administra)ve records are kept primarily in account books and other such formal instruments (e.g., tax receipts, census figures, licensing forms, etc.), not in novelis)c-like narra)ve renderings! The la=er is a part of our own literary and historical tradi)ons, with roots that go all the way back to Homer and Herodotus. However, we have no reason to assume that such narra)ve-producing individuals, or the technologies and seman)c/gramma)cal/intellectual tradi)ons that produced and sustained them, ever existed in the ancient central Andes. Perhaps they did, but we have no basis on which to presume that these ancient South American socie)es developed in a parallel way to those of the ancient Near East, Western Europe, China and even central America -- places where narra)ve histories were developed and persisted for millennia. On the other hand, as Andean socie)es con)nued through )me, from their Wari and Inka beginnings, and as the tradi)on and technology of recording became increasingly more complex, and as archives of records built up over )me, some clever individual might well have hit on the idea of using this technology to record graphical (i.e., string, knot, color) configura)ons that could be assigned seman)c values that could underlie narra)ve construc)ons. The la=er, however, was not necessary nor inevitable. What was necessary were the nota)ons of iden))es, quan))es, and different structural proper)es characterizing these ma=ers, discussed above, as they existed across the landscape of Tawan)nsuyu over )me. 13 I suggest that the Wari and Inka khipus developed over )me in a straight-forward manner, such as that described above (i.e., before the preceding paragraph). In that case, then, both Wari and Inka records would have been "about" iden))es, numbers (quan))es), and structures. Our efforts at decipherment, then, should be aimed at char)ng, comparing, and analyzing representa)ons of these classes and types of iden))es, quan))es and structures in the kno=edstring records. I suggest that the best keys, or guides, for learning how to interpret and translate these values, quan))es, structures and iden))es are those that were recorded in the earliest Spanish chronicles, which described what the Spaniards encountered soon aYer their conquest of Tawan)nsuyu. These documents recount for us what the Spaniards observed and heard about concerning the structure and organiza)on of Inkaic society, in the court in Cuzco, and in the myriad villages in the countryside. These Spanish wri=en accounts detail such forms of organiza)on as: dualism (hanan/upper vs. hurin/lower), triadic hierarchies (collana, payan, callao), quadripar))on (e.g., the four suyus of Tawan)nsuyu), decimal organiza)on, and highly complex local and imperial ayllu (kin group) organiza)ons. These (and perhaps other) principles, structures and values should be the principal elements at the heart of the informa)on encoded in the Inka (and probably Wari) kno=ed-string records. I have produced transla)ons of numerous such khipu recordings in a previous publica)on (Urton, 2017). Summary and Conclusion The sugges)ons for the contents of the Inka khipus outlined above represent my current views on what is recorded in the khipus -- that is, what was recorded on them both in the pre- 14 Spanish conquest era and in early colonial )mes. I think that it is these structural, quan)ta)ve, hierarchical nota)ons that cons)tute the substance of informa)on encoded and recorded on the surviving khipus that exist in museum collec)ons around the world today. Deciphering this informa)on can provide us with the knowledge and understanding of what the Inkas knew, thought, and recorded about their world -- seen from their eyes, from their points of view -before the arrival of Europeans. In short, discerning the structure and contents of these nota)ons represent our only means of seeking an understanding of this ancient American civiliza)on that is not a product of the minds, aRtudes and morals of the sixteenth century European invaders of the central Andes. Therefore, while the challenges of deciphering the khipus are indeed daun)ng, the poten)al benefits and contribu)ons to compara)ve world history make the effort well worth undertaking. 15 Notes 1) In all cases that I am aware of, at least in the early sources, khipu keepers are identified as male (but see Salomon’s account of what was apparently one female cord reader, in 18th century Peru; 2004:122-5). Female khipu keepers likely were common in pre-conquest times, however, just as what appears to have been an entire, parallel hierarchy of female officials (e.g., female equivalents of kurakas) seems to have been systematically disregarded by Spanish administrators (see Silverblatt, 1987), the same lack of attention from the almost exclusively male colonial officials might explain why we hear nothing of female khipu keepers. 2) But see Szeminski's argument for a possible system of Inka wri)ng, principally known only from what is now Bolivia (Szeminski, 2019). 16 References Ascher, Marcia and Robert Ascher 1997 Mathema)cs of the Inca: Code of the Quipu [1981]. New York: Dover. Bauer, Brian S., and R. Alan Covey 2002 "State Development in the Inca Heartland (Cuzco, Peru)." American Anthropologist 10, no. 3:846-864. Cerrón-Palomino, Rodolfo, 2008 Voces del Ande: ensayos sobre onomástica andina. Lima: Fondo Editorial de la Pontificia Universidad Católica del Perú. Chirinos Rivera, Andrés 2010 Quipus del Tahuantinsuyo: Curacas, Incas y su saber matemático en el siglo XVI. Lima, Editorial Commentarios. Clindaniel, Jonathan N. 2018 Toward a Grammar of the Inka Khipu: Inves)ga)ng the Produc)on of Non-Numerical Signs. Ph.D. Disserta)on. Department of Anthropology. Harvard University, Cambridge, MA. Díaz, Luisa 2008 “Aproximaciones hacia la problemá)ca del territorio Ychsma.” Arqueología y Sociedad No. 19:115-127 Locke, Leland L. 17 1923 The Ancient Quipu, or Peruvian Knot Record. New York: American Museum of Natural History. Murúa, Mar~n de 2001 Historia General del Perú [1613]. Madrid: Das)n Historia. Salomon, Frank 2004 The Cord Keepers: Khipus and Cultural Life in a Peruvian Village. Duke University Press, Durham and London. Salomon, Frank and George L. Urioste 1991 The Huarochirí Manuscript: A Testament of Ancient and Colonial Andean Religion [ca. 1598-1608]. Aus)n: The University of Texas Press. Silverbla=, Irene 1987 Moon, Sun, and Witches: Gender Ideologies and Class in Inca and Colonial Peru. Princeton, N.J.: Princeton University Press. Stone-Miller, Rebecca 1992 To Weave for the Sun: Ancient Andean Tex)les in the Museum of Fine Arts, Boston. New York: Thames and Hudson. Szeminski, Jan 2019 Qué Sabemos de las Qillqas Bajo el Gobierno del Inca? Arequipa: Ediciones del Lector. Torero, Alfredo, 2002 Idiomas de los Andes: lingüística e historia. Lima: IFEA-Editorial Horizonte. Urton, Gary 18 1998 “From Knots to Narra)ves: Reconstruc)ng the Art of Historical Record-Keeping in the Andes from Spanish Transcrip)ons of Inka Khipus.” Ethnohistory 45(3):409438. 2003 Signs of the Inka Khipu: Binary Coding in the Andean Kno=ed-String Records. Aus)n: University of Texas Press. 2014 “From Middle Horizon cord-keeping to the rise of Inka khipus in the central Andes.” An)quity 88:205-221. 2017 Inka History in Knots: Reading Khipus as Primary Sources. Aus)n, University of Texas Press. nd. Khipus: A Historical Guide to Making, Encoding, and Reading the Inka Kno=ed-String Records. Arequipa: Ediciones El Lector (in press).