Dumbarton Oaks Museum

CREDITS CALENDAR

APRIL 2 – AUGUST 18
2019 Dumbarton Oaks Director LECTURE
Dr. Jan Ziolkowski
May 18, 2019 at 2pm
Co-curators Shades of Meaning: The nexus of color, knots,
Lic. Juan Antonio Murro and fiber in the Dumbarton Oaks khipus.
Dr. Jeffrey C. Splitstoser
Jeffrey C. Splitstoser
Researcher/Consultant An overview of the khipus in the Dumbarton Oaks collections
Dr. Gary Urton and what we know today about the systems of information
they contain.
Consultant Dr. Jeffrey C. Splitstoser is the leading authority on Wari khipus,
Dr. Sabine Hyland having studied all known examples in collections throughout
the world. Splitstoser is part of the Castillo de Huarmey
Exhibition Design archaeological project, directed by Dr. Milosz Giersz, which
produced the largest number of scientifically excavated Wari
Ellen Richardson
khipus to date. Splitstoser is also the textile specialist for the
Huaca Prieta Archaeological Project, directed by Dr. Tom Dillehay,
Curatorial Fellow where he discovered information encoded in warp patterning
Samuel Shapiro in 6,200-year-old cotton textiles that were dyed with the
world’s earliest known use of indigo blue. He is the cocurator
Registrar with Juan Antonio Murro of the exhibition, Written in Knots:
Joni Joseph Undeciphered Records of Andean Life.

Exhibition Diagrams CURATOR’S TOURS

Dr. Jeffrey C. Splitstoser
Every other Friday at 3pm, beginning on April 12.
Juan Antonio Murro and Jeffrey C. Splitstoser

SCHOOL VISITS
For information about school visits to Written in Knots:
Undeciphered Accounts of Andean Life, please contact
India Patel at pateli01@doaks.org

Dumbarton Oaks
Research Library and Collection
Catalog design by Christine Lefebvre Design.
 THE
BARBARA AND WILLIAM
CONKLIN COLLECTION

Written in Knots: Undeciphered Accounts of

Andean Life is a tribute to the generosity of Barbara
and William J Conklin, who gave Dumbarton Oaks
their collection of 136 Pre-Columbian and ethnographic
objects. Their gift, which includes textile fragments of
exquisite manufacture and seven remarkable khipus,
allows the Museum’s Andean collection to reach
new heights.

This exhibition explores the long-standing Andean

tradition of communicating through cords, knots, and
colors. An alternative to writing, these fiber records
persisted through time, documenting the particularities
of different cultures, while always adhering to the
William J Conklin at home, same basic tenets.
c. 2016.
Crucial to the development of the show were two
khipu workshops organized under the auspices of
Pre-Columbian Studies at Dumbarton Oaks, in which
new developments in the field and the latest research
on khipus were generously shared by its participants,
some of whom also collaborated on this exhibition.

A product of both the Conklins’ donation and innovative

scholarship, this is the first exhibition in the United
States to comprehensively bring together Wari, Inka and
Colonial khipus. We are honored to share this unique
gift with our visitors at Dumbarton Oaks.

JUAN ANTONIO MURRO

ANDEAN CORDKEEPING
AN INTRODUCTION

The Andean region of west-central South America was recordkeeping. These included staffs wrapped with
home to a succession of cultures and civilizations brightly dyed threads linked to weaving and recording
whose roots go back millennia before the Christian technologies of various pre-Inka peoples, to the initial
era. Within this region lay the rugged territory that knotted cord administrative records of the Wari, who
would eventually become home to the largest empire lived some five-hundred years to a millennium before
of the Pre-Columbian Americas, Tawantinsuyu—“the the time of the Inkas, to the elaborately knotted and
four parts intimately bound together,” the name by colorful cord records used by Inka administrators in
which the Inkas knew their empire. From the earliest Tawantinsuyu (ca. 1450–1534 CE), and down to the local
appearance of humans in this territory, beginning some cord-keeping technologies of the Colonial era following
13,000 years ago, people began experimenting with a the Spanish conquest of the Andes, beginning in 1532.
variety of plant and animal fibers to produce twisted
On display in the Textile Gallery, next to the exhibit,
and knotted cord constructions that were critical to
is an ensemble of textiles produced by Wari and Inka
making a living in this challenging environment—from
weavers. It is appropriate to view khipus in relation to
lines for fishing in the cold waters off the Pacific coast to
textiles, as the two share much in common, especially
slings and traps for binding the legs of wild camelids—
in terms of their materials (cotton and camelid fibers),
vicuña and guanaco—in the highlands. Indeed, one
as well as many of their techniques of production
could say that from the time of the earliest cultures
(spinning, plying, knotting and dyeing of threads).
occupying the central Andean region until the rise of
These similarities notwithstanding, it is clear from
the Inkas, in the mid-15th century CE, cord-making was
the archaeological record, as well as from testimony
at the heart of subsistence technologies of peoples in
recorded in the Spanish chronicles and documents, and
this region.
from close study of Wari, Inka and Colonial khipus by
The display of khipus (or quipus, Quechua: “knot”) scholars today that khipus served different ends from
recently donated to Dumbarton Oaks by Barbara and the textiles. For, in whatever period they were produced,
William J Conklin featured in the exhibit Written in khipus provided the means for recording local and
Knots: Undeciphered Accounts of Andean Life, contains state administrative information (e.g., censuses, tribute
examples of what represented the height of Andean records, etc.), as well as notations for the production of
cord-making technologies. For included within what narratives, or stories of the past, recounted by Wari, Inka,
are referred to generically as khipus is a range of and Colonial cordkeepers of the ancient Andes.
devices that were developed over time primarily for
DR. GARY URTON
Andean landscape.
 Types of cord-color patterns.

WARI KHIPUS

The Wari ruled most of present-day Peru from 600 to

1000 CE, a period of time also referred to as the Middle
Horizon, forging South America’s first empire. The Wari
developed the khipu to meet the demands of running
an empire without writing. Most likely created by
combining the metaphorical connotation of knots
(representing something that ties, binds, “commits”)
with existing emblematic systems based on color (such
SEGMENTED MONOCHROME MOTTLED BARBER POLE
as those found in Paracas and Nasca wrapped staffs
and wrapped cords), they developed a device that for
all intents and purposes served as script.

Wari khipus are the most colorful of all the khipu

traditions, perhaps reflecting a connection with earlier
color-wrapping practices. To understand the importance
of khipus to the Wari, we must imagine living in a society
where color was not merely decoration like it is for the
most part today. In a society without writing, color
most likely held far more importance as a messaging
medium than it does today.

All Wari khipus have wrapped pendant cords, a feature

not found in Inka khipus. Wari subsidiary cords are also
colorful coming in monochrome, barber-pole, mottled,
or segmented patterns. While both Wari and Inka khipus
incorporate natural colors of cotton and camelid fiber,
Wari khipus are more likely to also include dyed fibers.
Paracas wrapped staff
Wari khipus rarely, if ever, seem to have color banding, with wrapped cords.
but they frequently have color seriation, like khipus Photo by Jeffrey C. Splitstoser
courtesy of the Cerrillos
PC.WBC. 2016.067 and PC.WBC.2016.068, or random Archaeological Project.
patterning, like PC.WBC.2016.136, where pendant-cord
colors do not have a pattern. In Wari khipus with color
seriation, the first or last pendant cords are frequently
not wrapped but knotted instead.

We know almost nothing about Wari khipu numerical

systems. Unlike Inka khipus, where a knot’s type
(overhand, figure-8, and long) and position on a cord
determine its value, Wari knots are always overhand,
double overhand, or in rare instances triple overhand
(or more), and they are tied immediately after a cord’s
wrapping or attachment to a parent cord. Hints about
their numerical systems come from the number of
knots tied on cords, which never exceed ten, suggesting
a base-10 numerical system; however, almost 70% of
all Wari cords have either zero or one knot, suggesting
perhaps the simultaneous presence of a binary system
based on presence or absence. The mathematics found
in khipu PC.WBC.2016.136 seems to support this idea.

DR. JEFFREY C. SPLITSTOSER

Wrapping detail of
PC.WBC.2016.136.
Photo by Joseph Mills.
 Inka base-10 positional
knot system.

1000s

INKA KHIPUS
100s
OVERHAND
KNOTS

The Inkas offer one of the great enigmas of world

civilizations. Although they were the conquerors and 10s
overlords of what would become the largest civilization
of the ancient Americas, the Inkas did not invent the
wheel, nor did they have markets, nor, most perplexing 1
1s
(Figure 8)
of all, did they invent a system of writing, at least not
one in the form of a two-dimensional, graphic script.
2-9
Unlike every other major civilization of the ancient (Long Knots)
world—e.g., the Mesopotamians, Egyptians, Shang 1 2 3 4 5 6 7 8 9 20 1000
Chinese, and the Mayas—each of which invented a KNOT VALUES
version of a grapheme-based “logosyllabic” (i.e., ON CORDS 10 100 1532
word/syllable) writing system, the Inkas relied on a
three-dimensional, cord-based system of recording
of knots signified the place values of clusters of knots
information—the khipu.
in the base-10 system of Inka/Quechua decimal
Inka khipus were cotton or camelid (llama or alpaca) numeration. In the unit’s positions (i.e., the lowest tier
fiber constructions of spun and plied (or twisted) cords. of knots), figure-eight knots signified “1s,” while long
The core elements of a khipu were: a) the primary cord, knots signified values between 2–9. Each subsequent
which was usually the thickest cord on a khipu and was tier higher on cords held between one to nine overhand
the horizontal cord from which, b) a variable number knots, signifying increasing values of 10 (i.e., 10s, 100s,
of secondary cords, commonly referred to as pendant 1000s, etc.). The numerical values recorded on cords
cords, were suspended; and c) subsidiary cords, which generally pertained to administrative matters, such as
were second-, third- etc.-order cords attached to the censuses, tribute records, and accounts of resources
pendant cords. These three types of cords were often of stored in state storehouses. Differences in cord colors
different colors, either from the natural color differences divided the data into different categories (e.g., individual
of cotton or camelid fibers, or from the application of tribute-payer records vs. group [ayllu] records, as well
natural dyes to the construction threads. as differences in items recorded, such as corn, beans,
chili peppers, etc.).
Pendant and subsidiary cords are commonly knotted
in tiered arrays of three different types of knots: figure- Khipus were maintained by cadres of khipukamayuqs
eight knots, long knots, and overhand knots. The tiers (Quipucamayocs, “knot makers/keepers/animators”),
administrators who served the Inka at local, regional,
and imperial levels of accounting and recordkeeping.
Spanish administrators and chroniclers who questioned
khipu-keepers after the time of the Spanish conquest
note that some of the khipus recorded information for
the production of narratives of the Inka past, such as life
histories of the kings, accounts of wars of conquest, and
POST INKA KHIPUS
other such mytho-historical matters. About 1,050 Inka
khipus exist today in museums in Europe and North and
South America. GU
Khipus were involved with almost all aspects of
Andean life. When the Third Council of Lima (1583)
mandated the universal destruction of idolatrous
khipus, it simultaneously encouraged their adaptation
for ecclesiastical purposes and helped incorporate
indigenous populations into the colonial administrative
system. It is not known if the new khipu types that
followed represent new forms or the reemergence of
suppressed regional traditions, or both.

Post-Inka khipus continue many Inka traditions,

including base-10 numerical notation. However, they
also display significant variation in their methods. For
example, some post-Inka khipus use loops to represent
ten, and some use groups of ten overhand knots to
represent 100. Post-Inka khipus incorporate a wider
variety of fibers, from traditional cotton and camelid
fiber to maguey and animal fibers, such as deer, sheep
wool, human hair, horse hair, and vizcacha, that must
be identified through feel.1

Khipu boards are post-Inka devices consisting of plied,

multi-colored strings inserted through holes in a
wooden board and knotted. The knot’s twist represents
a parishioner’s moiety, or group affiliation. Khipu boards
recorded the activities of parishioners, from attendance
and the use of ritual implements and clothing to the
quality of work and worker’s enthusiasm.

Khipukamayuq. Canuto-type khipus likely represent a Colonial innovation

Adapted from Murra et. al., 1980. but are poorly understood. One was carbon dated to
 1629 CE.2 Canuto khipus have wrapped pendant cords,
yet they only use long knots, which are not in registers,
making their numerical values, if they had them, unclear.

Herding khipus are a post-Inka tradition used by

pastoralists throughout the central Andean region
to the present day. Made of a single natural or dyed
camelid hair cord, or a cord that was doubled and
knotted leaving two loose ends, they typically recorded
a single category of information, such as herd size. Two
or more were frequently joined, so categories could be
broken down into subclasses, like male and female.

“Inka-style” khipus, another post-Inka tradition, closely

resemble Inka pendant-type khipus. They frequently
have objects attached to their pendant cords such
as dolls, fragments of hide or fabric, and tufts of raw
fiber. Inka-style khipus were possibly used into the 19th
century in communities like Tupicocha.3 Much of what
we know about the significance of color banding and
seriation and pendant cord attachment comes from
post-Inka khipus.4

JCS

1
Hyland and Hyland forthcoming.
2
Personal correspondence with Gary Urton
dated 6 April 2017, based on MALI khipu 28709.
Community leaders
wearing Inka-style khipus
3
Salomon 2002.
at Tupicocha, Peru. 4
Hyland 2016; Medrano and Urton 2018;
Photo by Frank Salomon. Hyland 2015.
Hyland, Sabine
2015 How Quipus Indicate Moiety ("Hanan/Urin"):
A Study in the Three-Dimensionality of Corded Texts.
Willka Nina 3.
2016 How Khipus Indicated Labour Contributions in an
Andean Village: An Explanation of Colour Banding,
Seriation, and Ethnocategories. Journal of Material
Culture, online August.

Hyland, Sabine P. and William P. Hyland

forthcoming Quipus in an Age of Revolution:
the Epistolary Quipus of San Juan de Collata, Peru.
In Cabinet for a Global Age: New World Objects in
the History of Knowledge, edited by Mark Thurner.
LAGLOBAL.

Medrano, Manuel and Gary Urton

2018 Toward the Decipherment of a Set of Mid-Colonial
Khipus from the Santa Valley, Coastal Peru.
Ethnohistory 65(1):1–23.

Salomon, Frank
2002 Patrimonial Khipu in a Modern Peruvian Village: An
Introduction to the "Quipocamayos" of Tupicocha,
Huarochirí. In Narrative Threads: Accounting and
Recounting in Andean Khipu, edited by Jeffrey Quilter
and Gary Urton, pp. 293-319. Joe R. and Teresa Lozano
Long Series in Latin American and Latino Art and
Culture. University of Texas Press, Austin.

CATALOGUE
 Wrapped-main-cord khipu
with color seriation
Wari, 600–1000 CE
Cotton and camelid fiber

PC.WBC.2016.067

PC.WBC.2016.067 is a pendant-type khipu with a wrapped

main cord. Its pendant cords are seriated with five
different wrapping patterns that repeat, although its
colors are darkened making their patterns difficult to
discern. Its main cord is 224 cm long and its longest
pendent cord with subsidiaries is 16 cm. It has 14 groups.
Each group would have had six pendant cords, making
a total of 84 pendant cords, five of which would be
PC.WBC.2016.067 Cord attachment, cord twist, wrapped, and one would instead be knotted. Today,
and knot twist. however, several cords are missing or broken, and, like
the final groups of PC.WBC.2016.068, the last two groups
RECTO VERSO in this khipu are also unfinished: its pendants are not
MAIN CORD ATTACHMENT ATTACHMENT wrapped, and in both khipus their final subsidiary cords
are missing. The khipu also has a “blank” pendant cord
attached to it dangle end, or tail. Each pendant cord in
the first 12 groups has two subsidiary cords, and most of
them have two knots, although some have zero or one
knot, suggesting a numerical system that could signal
presence vs. absence or other binary-type information.
The last pendant cord in each group is unwrapped and
has a single knot.
All cords have a Z twist, and all subsidiary and most
CORD TWIST pendant cords have recto attachments. All knots have an
S twist. The main cord and pendant cords are mottled
brown-and-tan cotton, as are one of the two subsidiary
cords; the other is red camelid hair. The wrapping on
pendant cords includes red and golden-brown camelid
hair and black, brown, and golden-brown cotton. JCS

KNOT TWIST
(LOOSELY TIED)
 Pendant khipu with color seriation
Wari, 779–981 CE
Cotton

PC.WBC.2016.068

ABOVE:
PC.WBC.2016.068

RIGHT:
Detail of braided main This is the largest Wari khipu known in museum structure, in which the structure associated with one
cord and color seriation. collections around the world and the only with a braided shade of blue is different from the structure of another
Photo by Joseph Mills. main cord. In 2003, William Conklin had this khipu strongly suggesting that the cords were made by
carbon-dated, producing an approximate age range of different people and dyed in different vats over time.
BELOW:
779–981 CE, placing it at the end of the Middle Horizon.
Khipu in its rolled-up state. All the knots on cords in Groups 1–10 have an S twist,
It is made entirely of cotton. The khipu is color seriated
while all the other knots have a Z twist, although their
with monochrome wrapping, meaning that each
significance is elusive. When acquired by the Conklins,
pendant is wrapped with a single color that repeats. It
the khipu was rolled-up and an unidentified, red, earthy
has 101 wrapped pendant cords placed in 20 groups of
substance adhered to the braided main cord and
five each with an extra pendant in Group 4 (Pendant 21),
lower ends of its pendant and subsidiary cords, making
whose colors and wrapping are different from the rest.
them brittle. JCS
The khipu has four “blank” pendant-like cords without
wrapping and shorter and thinner than other pendant
cords. Perhaps they supported the khipu when it was
being made or displayed.
Most pendant cords have nine or ten subsidiary cords,
each with a different color, whose sequences repeat
on every pendant cord. Most subsidiary cords have no
additional subsidiaries; however, the pendant cords
in Groups 3 and 4 often have an extra subsidiary with
subsidiaries up to five levels deep. Many subsidiary
cords in the last four groups are missing, and the last
group has none at all, indicating an unfinished state. Still
the khipu originally had 1000 cords, all with a Z twist,
although several black-and-white cords, are missing,
because the black dye weakened the cotton fibers.
These black cords are better preserved in some groups.
Color shades are inconsistent across the khipu. For
example, the blue in Group 1 and 2’s subsidiary cords is
different from the blue in Groups 3 and 4. These color
differences are accompanied by differences in yarn
 Wrapped-main-cord khipu
with color seriation
Wari, 600–1000 CE
Cotton and camelid fiber

PC.WBC.2016.136

Despite its small size, this khipu has several remarkable

features, including the presence of wrapped cords down
to fifth-level subsidiary cords. Its main cord is looped and
begins with an intricate wrapping pattern with crosses
somewhat reminiscent of an Inka kayte (beginning knot).
The remainder of the main cord is wrapped between
each of its five groups of pendant cords. Additionally,
similar to many Wari khipus, its pendant cord colors do
not fall neatly into banded or seriated patterns but display
a mix of both color banding in its wrapped pendants and
ABOVE: color seriation in its unwrapped pendants and most of
PC.WBC.2016.136
its subsidiaries.
BELOW:
This khipu is special not only because it is possibly the
Detail of wrapping and
beginning loop.
only known complete Wari khipu, but also because it
Photo by Joseph Mills.
might provide clues to the enigmatic Wari numerical
system. It has five groups of pendant cords separated with
wrapping of the main cord. The first pendant cord of each
group has no wrapping, and it has subsidiary cords whose
color sequences are the same (with alterations) as the
remaining cords of the group. Interestingly, the number
of knots tied onto the subsidiaries of the first (unwrapped)
pendant cord equals the sum of the number of knots in
the other subsidiary cords of the same color of the group.
Its main cord is 50 cm long, it is 39 cm wide, and it has 41
pendant cords. This is the only known khipu with
wrapping on fifth-level subsidiaries. All cords have a final
Z twist and, except one 5th level (unwrapped) cord, verso
attachments. Subsidiary cords are solid or mottled with
and two or three colors, but it has no barber-pole cords.
This khipu’s bright colors and complex wrapping patterns
incorporating crosses and floats are similar to the wrapping
patterns found in other Wari khipus, particular those in
the American Museum of Natural History in New York. JCS
Rolled-up seriated khipu
Inka, 1450–1534 CE
Cotton and camelid fiber

PC.WBC.2016.070
When khipus were moved from a place of recording, for
instance, to an accounting center, the khipu would be
rolled up for transport in a spiral, as seen in this large and
beautifully colored example. Rolling was accomplished
with a section of the primary cord—seen here in the
spiral at the top of the bundle—called the dangle end,
or tail. Rolling in this fashion allowed for easy carrying
of the khipu, as well as for storing multiple khipus in
archives. Since it is thought that the color patterning of
primary cords may have indicated the subject matter of
what was recorded on khipus, one could see at a glance
the topics of khipus rolled and stored in an archive.
The khipu seen here has never been opened; therefore,
researchers cannot know the cord color patterning,
nor the values knotted into the strings. Nevertheless,
from what may be observed, it appears that this khipu
is organized by color seriation. This is a patterning of
cords into a repeating color series, such as: dark brown,
medium brown, light brown, white cords. Seriated khipus
generally carry higher quantitative values, as when
records from multiple lower-level accounting units (e.g.,
the Andean clans, or ayllus) were aggregated into
collective accounts. Therefore, this khipu may have been
an accounting record of a regional-level organization
within the Inka empire.
An interesting feature of this khipu is the presence of
RIGHT: multiple tufts of unspun fibers attached or pressed into
PC.WBC.2016.070 the khipu cords at various points. These may have come
into contact with, and been pressed into, the bundle of
LEFT, TOP:
Spiraled main cord at the
khipu cords in storage.
top of the bundle. Un-rolled khipus offer a conundrum to museums and
Photo by Joseph Mills. researchers alike. For while it is ideal to maintain samples
LEFT, BOTTTOM: in the pristine condition in which they are discovered,
Provincial administrator only by un-rolling a sample and spreading it out can
carrying rolled-up khipu. researchers study the cord, color, and knotting patterns
Adapted from Murra et. al., 1980. and, thereby, attempt to interpret the meaning and
significance of the khipu. Since the discovery of un-rolled
samples is unusual, it has been decided not to un-roll
this khipu. GU
 Color banded khipu
Inka, 1450–1534 CE
Cotton

PC.WBC.2016.071

There are a number of features of this khipu suggesting

that it may have been an accounting of activities of a social
group, or clan—termed an ayllu in the Andes—composed
of some 33 members. Each member of the group may
have been associated with one or the other of 33 six-cord
groupings that compose the khipu. The six-cord groups
are most evident in color differentiated groups of cords
on the left two-thirds of the khipu. There is a total of 202
cords on the khipu, which is close to what would be an
arrangement of 33 six-cord groups (33 x 6 = 198).
Color banded khipus, like the example displayed here,
are composed of cords organized into differently colored
cord groupings—e.g., six green cords, then six white cords,
ABOVE: then six brown cords, etc. Color seriated khipus have
Detail of beginning knot their cords organized in a repeating color series—dark
and banding pattern. brown, medium brown, light brown, white; dark brown,
Photo by Joseph Mills. medium brown, light brown, white; repeat…. From studies
of hundreds of banded and seriated khipus, researchers
RIGHT:
have found that banded khipus record lower numbers,
PC.WBC.2016.071
likely records of several individuals that made up a clan.
Seriated khipus record higher numbers, as would be the
case when the records of numerous banded (individual-
level) khipus were aggregated.
This khipu has a complexly constructed primary cord,
which is the thick horizontal cord from which the 202
pendant cords are suspended. It is composed of threads
of at least three colors spun and plied together into a
thick, final S-twist cord. To the left side, the primary cord
begins with a needle-work bundle wrapped with medium
brown threads. It is thought that recording began from
this beginning knot and moved toward the opposite end
to terminate (as seen here) in what is termed a “dangle
end.” The dangle end is used to roll the khipu into a spiral,
for carrying (as seen in another khipu in this exhibit). It
is thought that the primary cord color patterning and
bundle identified the subject matter of the recording. GU
Khipu with summary cords
Inka, 1450–1534 CE
Cotton

PC.WBC.2016.072
This exceptionally large and complex khipu contains an
arrangement of cords and knot values that suggests this
was an accounting record of some good or resource
whose quantities and organization were in line with
the decimal accounting principle that governed Inka
administration. The organization of cords is brought
about by spacing between groups of cords, as well as
(in some cases) by differences in cord colors.
The khipu is composed of a total of 438 cords. The most
pronounced decimal feature of the khipu is the repetitive
organization of groups of 10 + 1 = 11 cords. Two of the cord
groupings are made up of only five cords. Each of what
is therefore a total of 39 cord groupings contains 11 cords,
in which one cord of the group is a “summary cord.” The
summary cords are those in the arrangement as displayed
here that leave the horizontal primary cord straight up.
The 10-cord groups hang down from the primary cord.
The summary cord sometimes leaves the primary cord
from the middle of the 10-cord group, sometime from
one end or the other. There are also numerous subsidiary
cords attached to pendant or summary cords; these are
displayed here at oblique angles.
The general principle of the 11-cord arrangement is that
the values recorded on each group of 10 cords (and any
subsidiaries) is repeated on the associated summary cord
(and any subsidiaries). Some cords in several of the cord
groupings are broken; therefore, there are several cases
in which the sums of values on the 10-cord groups do not
equal that on the summary cord.
Although the sample appears to be a single khipu, it is, in
fact, composed to two khipus spliced together. The splice
is located between the 9th and 10th 11-cord groupings, as
counted from the left. The 10 + 1 = 11 arrangement of cords
is identical between the two khipus, although there is a
greater concentration of subsidiaries on the smaller khipu,
to the left.
Unfortunately, we do not have archaeological information
from which to determine what items were being
accounted for by means of this extraordinary khipu. GU
 Canuto-type khipu with color banding
Colonial, 1534–1821 CE
Maguey (?), camelid fiber, and cotton

PC.WBC.2016.069

This pendant-type khipu is canuto-style with color

banding. Its main cord is 58 cm long, and its longest
pendant is 67 cm. This originally had 48 pendant cords
(44 are present) arranged in twelve units, each consisting
of four pendants joined together and attached to the
main cord by doubling them over the main cord and
spiral-wrapping them. These 12 units are clustered into
three groups of four, each cluster separated by spaces
and distinguished by shared wrapping patterns that are
symmetrical about the khipu’s two central pendants.
Unlike other khipus in the collection, this khipu has no
subsidiary cords, and its pendant cords are color-wrapped
at their distal ends. The colors found in wrapping yarn at
the tops of pendant cords are white, natural-dark-brown,
and dyed red and yellow (all camelid fiber); those at
the ends of individual pendant cords are dark-brown,
red, and yellow camelid fiber and a single instance of
blue-green cotton.
The khipu’s knots are all long knots (S) representing the
numbers 3 through 9, with the majority being 6 to 9.
Characteristic of canuto-style khipus but unlike Inka
statistical khipus, the khipu has no overhand or figure-8
knots. Its knots are not arranged in decimal tiers, or
registers, potentially implying that they contain narrative
information. Unlike Inka khipus, but similar to Wari
khipu practices, the majority of this khipu’s pendent
cords have a Z twist, and its main cord has the opposite
twist of S. In addition to the cotton and camelid fiber
used to make nearly all Pre-Columbian and post-Inka
ABOVE: khipus, canuto-style khipus often include other fibers,
PC.WBC.2016.069 such sheep’s wool, horse and/or human hair, and—like
LEFT, TOP: the fiber found in this khipu—maguey (Furcraea Spp.,
Detail of distal ends a leaf fiber in the agave family) JCS
wrapping.
LEFT, BOTTOM:
Detail of long knots.
Photos by Joseph Mills.
Andean dyes.