Figure 1 Khipu Cord from the Inka Period (1300-1532) [Museo Larco, "Los Quipus, Época Imperial,"ML60004, Lima, Peru]
The khipu knot records of the Andes are mysterious objects whose exact use and decipherment elude scholars to this day. The word “khipu,” Quechua for “knot” (alternatively rendered “quipu;” Aymara: “chino”), refers to bunches of dyed and knotted cords that record various types of information. Khipu were most famously and extensively used by administrators of the Inka empire but were also used both in earlier and later times, as well as by people outside the central state.  However, their use as a means of recording has died out and their system of coding is lost to us, with no one alive today capable of reading them.
Knowledge of the khipu comes to us from a relatively small archeological record and various colonial accounts. Though there was never an extended investigation of khipu under the Spanish colonial government, there are many colonial authors who discuss khipu at greater or lesser length. Early chroniclers witnessed Inka officials consult khipu, but accounts from after the Toledan period (post-1581) generally draw from earlier sources. However, even first-hand accounts are only descriptions of oral readings of khipu produced by native record-keepers; these record-keepers never explained how the khipu system worked, nor showed how one might record something “in khipu.” These khipu readings recorded by Spanish conquerors and other native oral histories were initially treated with authority and were the primary source for European understandings of the Inka past, but in later periods, this respect turned to distrust as khipu records were used against Spanish authorities in legal conflicts with the colonial government. Where the khipu records differed from Spanish written records, the latter were considered the correct version. This suggests that Spanish records must be read with a grain of salt, as conquistadors and chroniclers had a vested interest in doctoring the historical record to their own advantage.
In many ways, the historical record in regards to the khipu has to be deciphered as well. Through this muddle of sources, scholars have tried to figure out what exactly the khipu recorded and what they were capable of recording. There is also a sustained debate as to whether or not the khipu represent a “writing system.” This essay attempts to explain current understandings of these questions.
I will first look at the form and physical system of the khipu. I will then look at their known use in accounting and delve into the question of their use in narrative accounts: how much was recorded by knots and how much was memorized by the khipukamayuq, the khipu-keeper? I will then use Rosaleen Howard’s study of Quechua narrative to understand what recording and telling a narrative looks like in the cultural context of the Andes and what a khipu-recorded narrative might look like. Finally, I will examine the extent to which khipu were mutually intelligible throughout time and space.
What do khipu look like and how were they made?
It is important to understand the khipu as a three-dimensional object. The material, color, and twist of the cord, the types and placement of knots on the cord, and the spacing between cords and knots are all important elements and ways of conveying information. In this way, the khipu are an extension of folk craft, deeply related to the Andean tradition of weaving which has also been an important form of information storage and transmission.
Figure 2 Basic khipu terminology and structure [Ascher and Ascher 1997, 12, fig.2.7]
The main cord is the primary suspension line off of which all other cords hang. There are two ends of the primary cord: The closed end, which is looped from the doubling of the cord, and the open end which has free threads which may be knotted and wrapped to form a ball. A single cord may dangle from the open end of the main cord and will generally consist of a summary of the khipu’s information. The main cord would have been the first element of the khipu to be constructed and generally announces the nature of the khipu through its materials, colors, and ply direction, much like a title or heading. It has been speculated that the primary cord was held taut by the closed end wrapped around the big toe of the khipu-maker, much the same way braids and narrow cords are constructed in the weaving tradition.
Figure3 Attaching accessory cords to the main cord [Kate Derosier, 2011]
Secondary cords are attached to the primary cord, moving from closed to open end. Secondary cords also have a closed and open end and attach to the main cord by their closed end. They are generally arranged in groups of five, six, or ten, with an open space of primary cord separating their grouping. These cords are further distinguished and classified by color. In about ten percent of the khipu corpus, top cords are attached to groups of secondary cords and sum the values of that group. It is commonly accepted that khipu knots represent a base-10 counting system, and top cords provide internal corroboration of that reading.
Tertiary cords can be attached to secondary cords, again by their closed end. There can be multiple tertiary cords coming off of one pendant, or other cords coming off of tertiary cords, creating further groupings and levels of information. These cords would have had to have been completed and attached after the knotting of the secondary cords.
Figure4 Types of knots used in the khipu [L. Leland Locke, 1923]
The cord layout, with its multiple layers of information, is the array into which the knot-data is placed. The knots represent numbers, sometimes as magnitudes, sometimes as labels, depending on the purpose of the khipu. Each knot has a fixed end (closer to the primary cord) and a free end (closer to the open end of the cord), analogous to the open and closed ends of the cord. Reading of the knot and cords would have progressed from closed to open end, with the highest order first, as in the Quechua language.
In the making of the khipu, then, knotting would have occurred last in the process. The primary information of the cord is its color, material, and ply direction which is “acted upon” through knotting. In addition, each cord was made for the unique purpose of each individual khipu; there was no ball or hank of pre-made cord to cut from. The first step in making khipu, therefore, was the selection of materials for the cords.
Figure 5 Plying cord and Twist Direction [British Standards Institution, 1963]
The primary material for most khipu was cotton in at least eight natural un-dyed colors: white, beige, light, dark, and medium brown, chocolate, reddish-orange, and mauve. Occasionally, blue-dyed cotton was used. Sometimes “accessory threads” of brightly-dyed alpaca fiber, much more receptive to color than cotton, were used to encode special elements. In addition, multiple colors were sometimes plied together, creating cords with different strands of color, providing yet more ways of encoding information. Garcilaso de la Vega, the famous half-Inkan chronicler, provides us with some examples of color correspondences for state-record khipu: yellow stood for gold, white for silver, and red for warriors.
After choosing color and fiber, the cords had to be twisted. Inka khipu are generally triple-plied, that is, the fiber is spun in one direction and the thread is folded in half and the ends are spun together in the opposite direction; the cord is then halved and spun together again, quadrupling the thickness. Most khipu in the corpus have a final S-ply direction. Only after spinning and plying would the cords have been attached to each other and then knotted.
The Use of Khipu in Accounting
Despite certain textual incongruities between colonial sources as to the various uses of the khipu, all agree that the system was used for accounting in the Inka state. Cieza de León, contemporary of Francisco Pizarro, noted that the Inka used their knotted-cord records to keep track of reserves and expenditures of states as well as tribute accounts and population censuses. He claimed that bookkeepers balanced the accounts every four months. Hernando Pizarro corroborates his account with a description of this book-balancing: “When [Pizarro] and his men removed these goods from the storehouse, the record keepers ‘untied some of the knots which they had in the deposits section [of the khipu] and then [re-]tied them in another section.’” This is further corroborated by the archeological record which shows evidence of some khipu having their knots untied at some point after completion and storage.
In the early years of the colonial government, khipu readings produced by native administrators formed the basis of Spanish inspections and reports. In 1549, Pedro de la Gasca, president of the Real Audiencia in Lima, ordered a round of visitas, or general inspection, throughout the empire. Inspectors record census figures and statistical data on resources, all drawn from the khipu as told by the khipukamayuq.
Colonial accountant and administrator Agustín de Zárate, writing in 1555, gives a relatively very detailed description of the khipu system, perhaps driven by his own closeness to the subject of accounting. He states:
Important matters are preserved with the help of some cotton strings that the Indians call quipos, denoting the numbers by differently tied knots, going up along the space of the string from the units to the tens and higher, and using strings of different colors according to what they want to show. And in each province there are people that are in charge of remembering by these strings general things, whom they call Quipucamayos. And so there can be found public houses full of these strings, which are easily interpreted by the person in charge of them, even though they are many ages older than he is.
This is an incredibly detailed account compared to other colonial sources and shows that at least some Spanish administrators asked questions as to exactly how the khipu recorded information. Here, too, is written evidence that accounting khipu used a base-ten system. However, Zárate is looking only at accounting-type khipu, as they are the only type of interest to his position. He does not mention other types of khipu, let alone go into such a detailed description of them.
Nonetheless, from these early accounts, we can assuredly say that some khipu were used as accounting “books.” In these khipu, the knots stood for numbers and used a decimal system. These accounting khipu were intelligible to specialized administrators, signifying that there was a consistent system throughout time and space. They were used to keep track of the accountable stores of the state, including population demographics, and therefore played an essential role in the highly organized and centrally-controlled Inkan state.
These accounting-type khipu are also the most easily decipherable by modern scholars. However, as Marcia Ascher, one of the leading voices in the study and decipherment of khipu, states:
A concept-based system as general as khipu writing does not have a single encompassing resolution. Decipherment of khipu writing can only mean a solution to a khipu or set of related khipu. This is so because khipu writing is adaptable to so many different systems of meaning, and for each system a different khipu or set of khipu comes into being.
Therefore, when looking at khipu used for purposes other than accounting, we are back to square one.
Khipu and Oral Narratives: How much was recorded? How much was memorized?
Though scholars are certain that narrative renderings were produced by khipukamayuq with the aid of khipu, we do not know the manner in which information was recorded, nor how much information was recorded and how much was retained in memory. Miquel de Esteta, a conquistador contemporaneous with Pizarro, writes that “[the Inka] recall the memory of things by means of certain cords and knots, though the most notable things are remembered in songs.” Cieza de León, writing in the same period, states that reserves and expenditures of the state were recorded on khipu as well as “other things that occurred many years in the past.” He also describes a class of men charged with composing and memorizing songs commemorating great deeds of kings; these are probably the court poet-philosophers called amauta. We do not know the relationship between khipukamayuq and amauta, nor if the amauta used khipu to aid their memory.
During the Toledan era, chronicler Cristóbal de Molina writes, “They had Indians who were very skilled and knowledgeable in the quipos and accounts; these [Indians] passed knowledge down from generation to generation; and in fixing in the memory all that they had learned, miraculously nothing was forgotten.” This account is interesting in that the khipu are specifically linked to inherited oral knowledge. Memory, however, is emphasized. It is also strange that, during a period in which native accounts are being discredited in Spanish courts, the author lauds the incredible indigenous historical memory.
In 1582, just after the Toledan period, Bernabé Cobo states that “all manner of information was retained on [khipu], but understanding the records required the memorization of information beyond that which was actually registered. On the basis of this memorized information, the khipukamayuq constructed a full narrative.” This is a very important account as it specifies that khipu do not record all relevant information; in this statement, the khipu seem much closer to a mnemonic device than a writing system as we understand it. It is interesting as well in that it emphasizes “missing/memorized” information, right on the heels of Toledo’s viceroyalty, where the khipu were regularly discredited and condemned.
Writing almost thirty years after the end of Toledo’s viceroyalty and Cobo’s account, Garcilaso de la Vega insists that the khipu “only assisted the recollection of already memorized histories; they were only ‘perishable expedients’ […, for they could not serve] as letters that perpetuate the memory of events beyond the lives of their authors.” This account states for the first time that khipu are only mnemonic devices and calls into question mutual intelligibility between khipukamayuq on the basis of the scant information the khipu actually contain.
In these accounts we see the shift in how the Spaniards viewed the khipu. First as a “miraculous” information-storage device. Then, as simply a mnemonic aid. Finally, even de la Vega, someone with direct connections to khipukamayuq as the son of an Inkan noblewoman, questions their accuracy. However, even before the Toledan-era shift in regard for the khipu, early chroniclers still imply a relationship between recorded information and memorized information. Though no sources help us understand the ratio between the two types of information, it is clear that the khipu did not act as a form of direct transcription of speech for the purpose of creating a physical manifestation of spoken language. Now it is helpful to turn to Rosaleen Howard’s conclusions on Quechua narrative to show that direct transcription would have never been a priority in the cultural context of the Andes, anyway.
Remembering, Factuality, and Storytelling in the Andes
Khipu of the accounting type record “an objectively perceived state of things in the past which was relatively closed to reinterpretation at a later date.” However, when used as mnemonic devices in oral storytelling, there was almost certainly more latitude on reinterpretation, especially considering how memory and remembering operate culturally in the Andes. To the Quechua mind, remembering is not about being able to recall set data pieces, but is rather a “culturally vital activity involving not only the telling of stories but also the performance of rituals and participation in fiestas.” Remembering (yarpiriy/yuyariy) through ritual, song, and storytelling is about keeping culture and society alive and in harmonious interaction with the worlds of the saints, the souls of the dead, and the powers of the landscape. Remembering in the Andes is regenerative: the past provides “the symbolic resources for making sense of the present and projective towards the future, in a way that allows at once for continuity and change.”
When telling history, Andean peoples do not dwell on historical fact but rather on a sense of “what should have been” and an ideal future that “could come to be” is made out of this imagined ideal past. The past is constantly being remade and reimagined in the present; “memory is a continuous process of reactivation and reformulation of the past relative to the present circumstances.” This cultural valuation of the past is evident in a specific grammatical tense in Quechua that allows one to talk about the past as it affects and is mirrored in the present. Direct transcription of a story is not an Andean ideal, so of course, the khipu were not designed to do this. It is much more important in the cultural context to be able to remember key events, settings, persons, etc. in order to use them in a narrative constructed in the present, with a specific lesson or interpretation applicable to present circumstances. Khipu used as a mnemonic device, recording key information but not every detail, is ideal for this purpose.
In all places, including the Andes, narratives have a formulaic structure. When one understands this structure, one can hypothesize how a khipu could have encoded events in a narrative: color variations and groups to distinguish episodes and locations; knots encoding events and characters; the order of the cords showing narrative chronology, and so on. A narrative recorded like this would look like one of the many highly patterned, highly rhythmic, redundant, and symmetrical khipu in the corpus.
Regarding Mutual Intelligibility
As with all other aspects of the khipu, the degree to which the khipu were mutually intelligible between khipukamayuq or between regions is a murky question. Our colonial sources are divided. We know there were many khipukamayuq and various levels of authority, and even local khipu held by community elders. However, Bernabé Cobo insists that “one was unable to understand the registers and recording devices of others” and suggests that each khipukamayuq held secret, individually memorized information.
On the other hand, Felipe Guaman Poma identifies a khipu as a letter being run from one locale to the capital, suggesting that there was in fact mutual intelligibility between regions. And, to refer back to a previous quote, de Zaraté claimed that khipu kept in storehouses could be read by a khipukamayuq generations after they were made, suggesting mutual intelligibility through time.
The reason for these discrepancies lies in a very Andean cause: regional variation. Certainly, the accounting-type khipu used by the Inkan state would have had to have had a consistent system where a khipu constructed in a province and then transported to the capital could be read by a royal khipukamayuq. However, for other types of khipu less important to the central state, such as khipu recording locally important myths and stories, it is likely that regional variation in structure occurred and that therefore, these khipu could not be read outside of that locale. Late Toledan chronicler Martín de Murúa made note of local variation and the fact that different populations used different logics in recording. This is corroborated by the archeological record. “The khipu collection of the AMNH seems to reveal the existence of regional styles of khipu. Such apparent regional groupings imply the existence of local, non-official khipu.”
The Khipu: Is it a Writing System?
Certainly, the khipu represent a complicated system of recording information. They are known to be able to record exact numbers and magnitudes of many items with great detail. They were also likely used to record important narrative information and used in oral storytelling. As a three-dimensional object unconcerned with transcribing speech sounds, it is hard to draw analogies to Eurasian writing systems. In fact, asking to compare the two goes against our purpose of understanding the khipu, as trying to fit them into European definitions of writing is bound to fail. Andean society is an oral one, and although the khipu are the recording and information-storage device of that society, European ideals of transcription, fact, and history, and marking are foreign and unnecessary to this culture. The system of khipu and khipukamayuq were perfectly suited and highly effective for Inkan society. Accounting khipu were consistent and exact enough for use in the elaborative Inkan state and narrative khipu were flexible enough to be a perfect complement to oral history and storytelling methods. To call the khipu a “writing system” would miss the point. They are the three-dimensional information storage system of an oral society.
ABOUT THE AUTHOR:
Taryn DeLeon Mendiola is an artist and researcher based in New York. Having earned her undergraduate degrees in Art and History, her passion is researching and recreating ancient textile techniques. She is also critically interested in the workings of the garment industry and the place of textiles and craft in the fine arts market and museum space. You can find her work @t.dl.mendiola on Instagram.
Ascher and Ascher, Code of the Khipu: A Study of Media, Mathematics, and Culture. New York: Dover Publications, 1997.
Ascher, Marcia. “Reading Khipu: Labels, Structure, and Format.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, 87-102. Austin, TX: University of Texas Press, 2002.
Ascher, Robert. “Inka Writing.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, 103-15. Austin, TX: University of Texas Press, 2002.
Assadourian, Carlos Sempat. “String Registries: Native Accounting and Memory According to the Colonial Sources.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, 119-50. Austin, TX: University of Texas Press, 2002.
Conklin, William J. “A Khipu Information String Theory.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, 53-86. Austin, TX: University of Texas Press, 2002.
Derosier, Kate. “Quipu Anatomy.” Mathimages.com, 2011. Accessed October 27, 2021. https://mathimages.swarthmore.edu/index.php/File:Quipu_anatomy.jpg
Howard, Rosaleen. “Spinning a Yarn: Landscape, Memory, and Discourse Structure in Quechua Narratives.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, 26-49. Austin, TX: University of Texas Press, 2002.
Hyland, Sabine P. “Woven Words.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, 151-70. Austin, TX: University of Texas Press, 2002.
L. Leland Locke. The Ancient Quipu. New York: The American Museum of Natural History, 1923.
“Methods of Test for Textiles.” British Standards Institution, Handbook No.11. London: British Standards House, 1963.
Museo Larco, “Los Quipus, Época Imperal,” ML60004. Lima, Peru. Accessed October 27, 2021. https://www.museolarco.org/catalogo/ficha.php?id=20039
Quilter Jeffry. “Preface.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, xiii-xix. Austin, TX: University of Texas Press, 2002.
Urton, Gary. “An Overview of Spanish Colonial Commentary on Andean Knotted-String Records.” In Narrative Threads: Accounting and Recounting in Andean Khipu, edited by Jeffrey Quilter and Gary Urton, 3-25. Austin, TX: University of Texas Press, 2002.
 Urton, “An Overview of Spanish Colonial Commentary on Andean Knotted-String Records,” 3.  Ibid., 4.  Ibid., 4.  Quilter, “Preface,” xv.  Ascher, “Reading Khipu: Labels, Structure, and Format,” 95.  Conklin, “A Khipu Information String Theory,” 66.  Ibid., 59.  Ibid., 66-67.  Ascher, “Reading Khipu,” 95.  Ibid., 95.  Ibid., 98.  Conklin, “A Khipu Information String Theory,” 68.  Ascher, “Reading Khipu,” 90.  Conklin, “A Khipu Information String Theory,” 72, 78.  Ibid., 59.  Ibid., 60-61, 72.  Assadourian, “String Registries: Native Accounting and Memory According to Colonial Sources,” 124.  Conklin, “A Khipu Information String Theory,” 70.  Urton, “An Overview of Spanish Colonial Commentary,” 7.  Assadourian, “Native Accounting and Memory,” 121.  Urton, quoting H. Pizarro, “An Overview of Spanish Colonial Commentary,” 6.  Conklin, “A Khipu Information String Theory,” 80.  Urton, “An Overview of Spanish Colonial Commentary,” 6.  Assadourian, quoting A. Zárate, in “Native Accounting and Memory,” 120.  Ascher, “Inka Writing,” 106.  Urton, quoting M. de Esteta, in “An overview of Spanish Colonial Commentary,” 6.  Ibid., quoting C. de León, 7.  Ibid., 7.  Ibid., quoting C. de Molina, 9.  Ibid., quoting B. Cobo, 16-17.  Hyland, quoting G. de la Vega, in “Woven Words,” 158.  Howard, “Spinning a Yarn: Landscape, Memory, and Discourse Structure in Quechua Narratives,” 29.  Ibid., 29.  Ibid., 46.  Ibid., 28.  Ibid., 45.  Ascher, “Inka Writing,” 110-112.  Assadourian, “Native Accounting and Memory,” 122.  Urton, quoting B. Cobo, “An Overview of Spanish Colonial Commentary,” 16-17.  Conklin, “A Khipu Information String Theory,” 55.  Assadourian, “Native Accounting and Memory,” 120.  Urton, “An Overview of Spanish Colonial Commentary,” 9; Hyland, “Woven Words,” 160.  Conklin, “A Khipu Information String Theory,” 63.