Aztec Arithmetic Revisited: Land-Area Algorithms

and Acolhua Congruence Arithmetic

Barbara J. Williams, et al.
Science 320, 72 (2008);
DOI: 10.1126/science.1153976

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your
colleagues, clients, or customers by clicking here.

Permission to republish or repurpose articles or portions of articles can be obtained by

following the guidelines here.

The following resources related to this article are available online at www.sciencemag.org
(this information is current as of July 17, 2010 ):

Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
http://www.sciencemag.org/cgi/content/full/320/5872/72

Downloaded from www.sciencemag.org on July 17, 2010

Supporting Online Material can be found at:
http://www.sciencemag.org/cgi/content/full/320/5872/72/DC1
A list of selected additional articles on the Science Web sites related to this article can be
found at:
http://www.sciencemag.org/cgi/content/full/320/5872/72#related-content
This article cites 4 articles, 1 of which can be accessed for free:
http://www.sciencemag.org/cgi/content/full/320/5872/72#otherarticles

This article has been cited by 1 article(s) on the ISI Web of Science.

This article appears in the following subject collections:

Anthropology
http://www.sciencemag.org/cgi/collection/anthro

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2008 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.
 RESEARCH ARTICLES
Together they record over 2000 drawings of
agricultural properties pertaining to households in
16 named communities as they existed around
Aztec Arithmetic Revisited: 1540 to 1544 CE. Below we summarize the
records, which are described and analyzed in
Land-Area Algorithms and Acolhua greater detail elsewhere (8).
Milcocolli and tlahuelmantli drawings. Agri-

Congruence Arithmetic cultural fields of each property holder are de-

picted twice, once with field sidelengths in a
section labeled milcocolli, and again in a section
Barbara J. Williams1 and María del Carmen Jorge y Jorge2* labeled tlahuelmantli, which records field areas.
Milcocolli drawings are either quadrilaterals or
Acolhua-Aztec land records depicting areas and side dimensions of agricultural fields provide insight irregular, often many-sided polygons. On each
into Aztec arithmetic. Hypothesizing that recorded areas resulted from indigenous calculation, in a field side, painters indicated sidelength distances
study of sample quadrilateral fields we found that 60% of the area values could be reproduced exactly in standard linear measures with number glyphs
by computation. In remaining cases, discrepancies between computed and recorded areas were (Fig. 1). The Acolhua did not express the stan-
consistently small, suggesting use of an unknown indigenous arithmetic. In revisiting the research, we dard unit glyphically, but Nahuatl texts attest that
discovered evidence for the use of congruence principles, based on proportions between the standard it was the tlalquahuitl, hereafter referred to as

Downloaded from www.sciencemag.org on July 17, 2010

linear Acolhua measure and their units of shorter length. This procedure substitutes for computation “land rod/s” (T) (9). According to the Acolhua
with fractions and is labeled “Acolhua congruence arithmetic.” The findings also clarify variance native chronicler Alva Ixtlilxochitl (10), this mea-
between Acolhua and Tenochca linear units, long an issue in understanding Aztec metrology. sure equaled three Spanish varas (taken here to
be 0.83 m × 3, ~2.5 m). Sometimes a glyph for
he largest extant corpus of indigenous writ- and computation. Furthermore, the new analysis arrow, heart, or hand follows the standard “land

T ten data pertaining to native arithmetic in

the Americas is of Aztec origin. Two pic-
torial manuscripts, the Códice de Santa María
provides an important independent basis to
interpret and clarify often confusing Colonial-
period Spanish and Nahuatl alphabetic texts.
rods.” These refer to a distance remaining on a
field side that is shorter than one “land rod.” We
label them “monads” because they are neither
Asunción (1) and the Codex Vergara (2), treat an Data. Of the extant indigenous land docu- subdivided nor do they appear as multiples. No
important bookkeeping matter in Aztec imperial ments, many pertain to the Basin of Mexico, more than one monad glyph occurs on the same
society, that of measuring and recording household heartland of the Aztec empire constituted by a field side (11).
landholdings for tribute and other purposes. In an confederation of three powerful kingdoms (6). Other than sidelengths, the milcocolli sections
initial study of these codices, Harvey and Williams When interpreting these manuscripts, it is im- in the codices provide little data relevant to etic
(3) demonstrated that Aztec property descriptions portant to distinguish between two distinct “Aztec” analysis of area calculation. The properties are not
included the concept of areal as well as linear cultural traditions present at European contact, drawn to scale, and we find no direct evidence of
measures, but they did not pursue the question of that of the Tenochca (Mexica) of Tenochtitlan angle measurement. If other measurements such
how the area values might have been determined. and the Acolhua of Texcoco. Each produced as height and triangular subdivisions of quadri-
We addressed this issue in a partial study of the land documents that are somewhat different in laterals were made (which we suspect), they were
Codex Vergara (4), finding that recorded areas graphic style (7), in turn reflecting underlying not recorded in these codices.
for ~60% of the sample quadrilateral fields could variations in metrology. In the tlahuelmantli registers, area values in
be obtained by Length times Width and Surveyors’ The Códice de Santa María Asunción and the square “land rods” (T2) are framed by an ab-
Rule algorithms. Although exact recorded areas Codex Vergara are Acolhua pictorials depicting stract rectangle, usually with a protruding tab in
were not obtained in the residual cases, discrep- land and population records for two wards in the upper right corner. The frame provides three
ancies were consistently small, suggesting an Tepetlaoztoc, an Aztec-period city-state 6 km places in which to register numbers in base-20
unknown indigenous calculation procedure. Here northeast of Texcoco in the northeastern Basin. positional notation. The values of the number
we report findings from an expanded and more
detailed analysis of the Codex Vergara, using etic
(cultural outsider) and emic (cultural insider) per-
spectives (5). The etic analysis (framed by modern
mathematics reflecting an outsider’s view of Aztec
culture) concerns evidence for land-area calcu-
lation with algorithms and computation with
fractions. The emic analysis (framed by concepts
with meaning internal to Aztec culture) focuses
on interpreting indigenous knowledge and proce-
dures embodied in metrological categories ex-
pressed by surveyors and painters in their writings.
The results show patterns suggesting principal Fig. 1. Paired drawings of perimeter dimensions (milcocolli) and RAs (tlahuelmantli) of fields in the Codex
logical components of Aztec surveyor metrology Vergara. Sidelengths are in standard “land rods” (T); lines equal 1T and solid circles equal 20T. Monad
1
units shorter than 1T are represented by hand, heart, and arrow glyphs. One of these (arrow) appears in (A).
Department of Geography and Geology, University of
Abstract frames record RAs in T2. Lines in the upper right tab equal 1T2, lines on the bottom margin and in
Wisconsin–Rock County, Janesville, WI 53546, USA. 2Instituto
de Investigaciones en Matemáticas Aplicadas y en Sistemas, the center equal 20T2, solid circles equal 400T2. (A) The RA = 210T2, computed exactly by multiplying
FENOMEC, Universidad Nacional Autónoma de México, length times width with ACA for 2 arrows = 1T (Loc. 1-142). (B) Applying the Surveyors’ Rule and ACA for 5
Ciudad Universitaria, 04510 México D.F., Mexico. hearts = 2T yields the RA of 128T2 (Loc. 1-73). (C) The RA of 492T2 is obtained by using ACA for 5 hands =
*To whom correspondence should be addressed. E-mail: 3T and the Surveyors’ Rule (Loc. 2-30). Besides number symbols, drawings usually display a soil glyph in
[email protected] the center, and on the top margin of abstract frames a corncob glyph flags areas less than 400T2.

72 4 APRIL 2008 VOL 320 SCIENCE www.sciencemag.org

 RESEARCH ARTICLES
glyphs depend on their placement within the with procedural choices, we presumed that sur- If these algorithms failed to produce the RA,
frame line (Fig. 1). Monads do not occur on any veyors tended to use rounding in order to work we then applied the Surveyors’ Rule (Rule 3), a
tlahuelmantli drawings, signifying that area with integers of standard linear distance, and method of area computation used by Sumeri-
values were recorded only as T2 integers. that they utilized algorithms that squared-up field ans and later by Romans (12, 13). It approximates
Method. We examined the 408 milcocolli shapes or balanced one operation with another. quadrilateral area by multiplication of the aver-
quadrilateral fields of all five Codex Vergara Anecdotal evidence suggests such practices are age lengths of opposite sides: A = (a + c)/2 ×
localities and eliminated 41 from further con- quite natural, and they are readily observable in (b + d)/2.
sideration as detailed in table S1. The resultant everyday measuring procedure. Some Acolhua RAs are prime numbers (n =
Vergara database consists of 367 milcocolli and Acolhua area algorithms. We obtained most 17). These can result only if area computation
matching tlahuelmantli fields (90% of the Acolhua RAs by applying one of five area al- couples addition or subtraction with multiplica-
quadrilateral universe). In our calculations (table gorithms. The most frequent, and least complex tion operations; otherwise, multiplication in-
S2), field sides are labeled a, b, c, and d pro- arithmetically, is multiplication of two adjacent volves a fractional number. In analyzing primes,
ceeding clockwise from the left side of each sides (Rule 1). Nearly one-third of the quadri- it became apparent that some RAs of quadri-
quadrilateral. Given field sidelengths in “land laterals in the Vergara database have adjacent laterals with unequal sides are obtained by what
rods,” we tested area algorithms from simple to sides measuring either 20T by 20T or 20T by we call the Triangle Rule (Rule 4). In this pro-
complex (ranked by the number of operations 10T (n = 108 of 367). For all of these fields, cedure, a given quadrilateral is divided by a di-
required) and compared our computed area multiplication of adjacent sides yields the exact agonal, whereby the quadrilateral area equals the
(CA) with the Acolhua recorded area (RA). Ini- RA, suggesting that the field side angles are 90° sum of half the product of adjacent sides: A =
tially we ignored monads and used only integer and that the field shapes approximate squares ab/2 + cd/2, or A = ad/2 + bc/2, using the right or

Downloaded from www.sciencemag.org on July 17, 2010

sidelengths. A test algorithm was deemed and rectangles. For irregular quadrilaterals, where left diagonal, respectively (14).
successful if the error (E) = 0 (E = CA − RA). length is defined as side a or c, and width as Prime number decomposition also reveals that
If test algorithms failed, we sought new rules by side b or d, multiplication of a set of adjacent some areas are derived by adding/subtracting the
selecting the factors that could reproduce the sides also may produce E = 0. More complex is same unit (1T or 2T) to/from adjacent sides and
RAs resulting from their prime number decom- a related algorithm in which RAs are obtained multiplying the modified sidelength numbers.
position. If these two procedures failed, in a final by multiplication of the average length of one Operations with this Plus-minus Rule (Rule 5)
step we integrated monads into calculations. set of opposite sides by an adjacent side (Rule 2), balance each other, and although not always, tend
Our method presupposes that surveyor ar- thus introducing the operations of addition and to square-up field shapes. Finally, a category of
ithmetic was practical, simple, and efficient. Faced division. miscellaneous rules applies to a small number of
fields where areas are obtained with low frequen-
Table 1. Calculation examples of algorithms yielding Acolhua recorded areas. cy or unique algorithms (see sample calculations
for Rules 1 to 5 in Table 1 and miscellaneous
Sidelengths rules in table S2).
Recorded Calculated area
Field id.* a, b, c, d in (T)
area (T2) (T2)‡ Whether our methods successfully reconstruct
“land rods” and monads† Acolhua area algorithms is debatable, because at
Multiplication of two adjacent sides present we lack indigenous worksheets or other
1-207-31 20 + ht 19 + hd 20 + ht 19 + a 380 20 × 19 = 380 evidence of Acolhua calculations “in progress.” It
3-50-7 17 23 16 24 391 17 × 23 = 391 is noteworthy, however, that usually only one al-
Average length of one pair of opposite sides times an adjacent side gorithm results in E = 0 for a given set of side-
4-27-16 42 12 40 11 451 11 × (42 + 40)/2 = 11 × 41 = 451 lengths. For example, the Triangle Rule gives the
5-12-2 52 21 56 13 884 52 × (21 + 13)/2 = 52 × 17 = 884 same results as the Surveyors’ Rule only if one
5-145-31 40 8 27 24 432 27 × (8 + 24)/2 = 27 × 16 = 432 pair of opposite sides is equal. Similarly, when
Surveyors’ Rule, A = (a + c)/2 × (b + d)/2 applying the Length times Width Rule to irreg-
5-111-21 26 32 30 10 588 (26 + 30)/2 × (32 + 10)/2 = ular quadrilaterals, multiplication of each possible
28 × 21 = 588 set of adjacent sides often results in large var-
5-46-4 23 15 + hd 25 + hd 11 312 (23 + 25)/2 × (15 + 11)/2 = iances in area values. Thus, exactly obtained RAs
24 × 13 = 312 strongly suggest that our computation method is
1-2-1 16 10 11 9 126 (16 + 11)/2 = 13.5ru = 14, the same as (or functionally equivalent to) in-
(10 + 9)/2 = 9.5rd = digenous arithmetic procedures.
9, 14 × 9 = 126 Monads, Acolhua congruence arithmetic, and
Triangle Rule, A = (a × b)/2 + (c × d)/2, or (a × d)/2 + (b × c)/2 metric equivalents. Using sidelength integers, we
2-2-1 41 11 35 8+a 366 (41 × 11)/2 = 225.5ru = 226, obtain exact RAs for 218 fields with the five de-
(35 × 8)/2 = 140, scribed algorithms and miscellaneous category
226 + 140 = 366 (Table 2). In many other instances (n = 69), our
2-30-6 24 16 25 24 492 (24 × 16)/2 + (24 × 25)/2 = results closely, but not precisely, match those of
192 + 300 = 492 the Acolhua surveyors. These small discrepancies
5-34-3 49 14 47 12 + a 623 (49 × 12)/2 + (14 × 47)/2 = between RAs and CAs appear to relate to the use
294 + 329 = 623 of distance monads.
Plus-minus Rule, one sidelength +1 or +2 times another sidelength −1 or −2 Arrow: Cemmitl. We achieved the RAs for
1-106-25 16 8 16 7 126 (16 − 2) × (7 + 2) = 14 × 9 = 126 most quadrilaterals without considering sidelength
5-139-30 18 19 13 13+a 252 (19 − 1) × (13 + 1) = monads; therefore, we expected that Acolhua
18 × 14 = 252 surveyors did likewise. One special field, how-
1-189-27 14 6 13 6 75 (14 + 1) × (6 − 1) = 15 × 5 = 75 ever, demonstrated otherwise. A rectangle with
length 20T and width 10T plus an arrow has a
*Field id.: locality number 1 to 5–field number consecutive within locality–household number consecutive within locality.
†Sidelengths clockwise from left field margin; monads: a, arrow; ht, heart; hd, hand representing distances shorter than (T). RA of 210T2, which is exactly obtained by
‡Monads are not required in these computations. Abbreviations: ru, round up to integer; rd, round down to integer. ascribing a value of 0.5T to the arrow monad:

www.sciencemag.org SCIENCE VOL 320 4 APRIL 2008 73

RESEARCH ARTICLES
RA = 20T × 10.5T = 210T2 (Fig. 1A). This case for the above-mentioned rectangle, the RA is are depicted in the perimeter (Loc. 4, 43). The
provides empirical indigenous evidence, later con- calculated as area by the Surveyors’ Rule is (13T + 12T)/2 ×
firmed by many other RA computations, that (7T + 6T)/2 = (12.5T) × (6.5T) = 81.25T2. The
the arrow in the Tepetlaoztoc data has a value RA ¼ 20T
ð10 þ aÞT RA of 81T2 results by including 0.5T in the cal-
of 1=2 “land rod.” The computationally derived ¼ 200T2 þ ð20
aÞT2 culations and rounding down. In ACA the area is
value is consistent with dictionaries and
¼ 200T2 þ 10
ð2
aÞT2
Colonial-period land documents in which an ð13T þ 12TÞ=2 ¼ 12T þ a,ð7T þ 6TÞ=2
arrow glyph, or its gloss cemmitl, is often ¼ ð200 þ 10ÞT2 ¼ 6T þ a,
translated into Spanish as media de la braza, ¼ 210T2 ð1Þ RA ¼ ð12 þ aÞT
ð6 þ aÞT
“ 1=2 a standard linear measure” (15). The ¼ ð72 þ 18
a þ a
aÞT2
e72T þ 9
ð2
aÞT2 þ a
a e ð72 þ 9ÞT
2 2 2
relation between arrows and “land rods” thus This type of computation we call “Acolhua
defines a 2:1 proportion; that is, if a is arrow, congruence arithmetic” (hereafter ACA). ¼ 81T ð2Þ
then 2 × a = 1T (Fig. 2). This proportion is The 2:1 proportion of arrows to “land rods”
used to compute areas, just as in an analogous also applies in a different and very important Because Acolhua areas were recorded only
way congruence classes are found if (2 × operational context. For example, one field has as T2 integers, the a2 term is dropped. Even
arrow) were the modulus (16). In other words, sidelengths of 13, 7, 12, and 6: No monads though perimeter sidelengths did not include
arrows, this example and others provide direct
evidence that a proportion of 2 arrows = 1T
was used arithmetically in computation (see

Downloaded from www.sciencemag.org on July 17, 2010

Monad Glosses Proportion Fractional Metric
glyphs of monads equivalent equivalent sample calculations with monads in Table 3
in Acolhua Nahuatl to standard of (T) and table S2).
land documents “land rods” (1T equals Heart: Cenyollotli. Evidence that Acolhua
(T) 2.5 m) surveyors used an arrow when averaging side-
lengths clarifies possible derivations of other in-
Cemmatl 5:3 3/5 1.5 m
digenous area values. One frequently replicated
pattern is illustrated by a quadrilateral whose
one hand sidelengths are 20, 7, 20, and 6 and RA = 128T2
(Fig. 1B). According to the Surveyors’ Rule, area
Cenyollotli 5:2 2/5 1.0 m is (20T + 20T)/2 × (7T + 6T)/2 = (20T) × (6.5T) =
130T2. However, the RA is obtained exactly by
one heart multiplying 20T by 6.4T, instead of 6.5T. As-
suming 0.4T ( 2=5 T) represents a monad other
Cemomitl 5:1 1/5 0.5 m
than an arrow, of the two other monads depicted
in field perimeters—hearts and hands—the heart
one bone is the likely pictograph, because in other sources
it is similarly defined as 2=5 of a braza. Apply-
Cemacolli 3:1 1/3 0.83 m ing a proportion of 5 hearts to 2 “land rods,” 5 ×
ht = 2T, where ht is heart, in this case ACA for
one arm the RA is

ð20T þ 20TÞ=2 ¼ 20T,ð7T þ 6TÞ=2

Cemmitl 2:1 1/2 1.25 m ¼ 6T þ a e 6T þ ht,
RA ¼ 20T
ð6 þ htÞT
one arrow ¼ 120T2 þ 20
htT2
¼ 120T2 þ 4
ð5
htÞT2
¼ ð120 þ 4
2ÞT2
¼ 128T2 ð3Þ
Fig. 2. Acolhua equivalences inferred from computation of RAs. Illustrations of hand, heart, and
arrow are from (2). Bone and arm from (17) are tentatively proposed glyphic referents for
proportions 5:1 and 3:1. Alternate reading of the arm glyph may be cenciacatl (21, 22). Hand: Cemmatl. When averaging, in some
cases substitution of 0.5T by 0.6T yields the
RA. Assuming 0.6T is also monad, it would be
Table 2. Quadrilateral frequency where CA = RA, by algorithm.
longer than an arrow and probably corresponds
Computation Computation requiring monads to the hand glyph, the other of the three monads
Algorithm requiring Totals depicted in milcocolli perimeters. The hand glyph
integers only Arrow Heart Hand Bone Arm usually is translated into Spanish as a braza de
medir, “a unit of measure.” However, the hand
Rule 1: Multiplication of two 139 2 0 0 0 0 141
glyph does not symbolize the standard land
adjacent sides
measure in the Tepetlaoztoc codices because the
Rule 2: Average length of one pair of 9 4 3 1 0 4 21
hand is depicted rarely and is placed at the end
opposite sides times an adjacent side
of the circle-and-line number sequence denoting
Rule 3: Surveyors’ Rule 33 19 8 1 3 5 69
“land rods.” Rather, hd = 0.6T = 3=5 T, where hd
Rule 4: Triangle Rule 6 6 4 1 0 0 17
is hand. A proportion of 5 hands equal 3 “land
Rule 5 Plus-minus Rule 12 0 0 0 0 0 12
rods,” (5 × hd = 3T), applies for example in a
Rule 6: Miscellaneous, unique, or 19 6 0 1 1 0 27
quadrilateral whose sidelengths are 24, 16, 25,
low-frequency algorithms
24 and RA = 492T2 (Fig. 1C). The Surveyors’
Totals (Vergara database n=367) 218 37 15 4 4 9 287
Rule gives (24T + 25T)/2 × (16T + 24T)/2 =

74 4 APRIL 2008 VOL 320 SCIENCE www.sciencemag.org

 RESEARCH ARTICLES

Table 3. Examples of algorithm calculations with monad ACA where CA = RA.

Sidelengths a, b, c, d
Recorded Calculated area
Field id.* in (T) “land rods” and
area (T2) (T2)‡
monads†
cemmatl: 5 hands = 3 “land rods”
3-26-5 19 20 18 20 372 Surveyors’ Rule
(19 + 18)/2 = 18 + a ~ 18 + hd, 20 × (18 + hd) = 360 + 20×hd
= 360 + 4×(5×hd) = 360 + 4 × 3 = 360 + 12 = 372
3-86-5 15 25 17 34 470 Triangle Rule
(34×15)/2 = 255, (25×17)/2 = 25×(8 + a) ~ 25×(8 + hd)
= 200 + 25×hd = 200 + 5×(5×hd) = 200 + 5×3
= 215, 255 + 215 = 470
5-120-23 22 17 19 18 350 Average of one pair of opposite sides times an adjacent side
(22 + 19)/2 = 20 + a ~ 20 + hd, 17×(20 + hd) = 340 +3×(5×hd) +
2×hd = 340 + 3×3 + 2×hd = 349 + 2×hd ru =350
cenyollotli: 5 hearts = 2 “land rods”

Downloaded from www.sciencemag.org on July 17, 2010

3-10-2 42 17 41 9 538 Surveyors’ Rule
(42 + 41)/2 = 41 + a ~ 41 + ht, (17 + 9)/2 = 13, (41 + ht) × 13
= 533 + 13×ht = 533 + 2×(5 × ht) + 3×ht = 533 +
2×2 + 3×ht = 533 + 4 + 3×ht = 537 + 3×ht ru = 538
4-47-11 16 10 15 15 191 Triangle Rule
(16×10)/2 = 80, (15×15)/2 = 15×(7 + a) ~ 15×(7 + ht)
= 105 + 15×ht = 105 + 3×(5×ht) = 105 + 3×2 = 111,
111 + 80 = 191
5-144-31 22 20 23 21 448 Average of one pair of opposite sides times an adjacent side
(21 + 20)/2 = 20 + a ~ 20 + ht, 22× (20 + ht) = 440 + 22×ht
= 440 + 4×(5×ht) + 2×ht = 440 + 4×2 + 2×ht rd = 448
cemomitl: 5 bones = 1 “land rod”
1-179-28 22 6 21 12 191 Surveyors’ Rule
(22 + 21)/2 = 21 + a ~ 21 + bn, (12 + 6)/2 = 9, (21 + bn) × 9
= 189 + 9×bn = 189 + 5×bn + 4×bn = 189 + 1 + 4×bn
= 190 + 4×bn ru = 191
4-21-4 20 20 21 16 364 Average of two adjacent sides times another side
(21 + 16)/2 = 18 + a ~ 18 + bn, 20×(18 + bn) = 360 + 20×bn
= 360 + 4×(5×bn) = 360 + 4 = 364
5-170-36 37 30 38 28 1079 Surveyor’s Rule
(38 + 37)/2 = 37 + a ~37 + bn, (30 + 28)/2 = 29, (37 + bn)×29
= 1073 + 29×bn = 1073 + 5×(5×bn) + 4×bn
= 1073 + 5 + 4×bn = 1078 + 4×bn ru = 1079
cemacolli: 3 arms = 1 “land rod”
1-60-14 34 20 25 20 586 Surveyors’ Rule
(34 + 25)/2 = 29 + a ~ 29 + am, (20+20)/2 = 20,
(29 + am) × 20 = 580 + 20×am = 580 + 6×(3×am) + 2×am
= 580 + 6 + 2×am = 586 + 2×am rd = 586
1-182-29 5+a 9 36 8 300 Average of one pair of opposite sides times an adjacent side
(9 + 8)/2 = 8 + a ~ 8 + am, 36×(8 + am) = 288 + 36×am
= 288 + 12×(3×am) = 288 + 12 = 300
2-16-5 29 13 + a 30 + a 14 400 Average of one pair of opposite sides times an adjacent side
(13 + 14)/2 = 13 + a ~ 13 + am, 30×(13 + am) = 390 + 30 × am =
390 + 10×(3×am) = 390 + 10 = 400
cemmitl: 2 arrows = 1 “land rod”
2-14-5 36 12 37 12 438 Surveyors’Rule
(36 + 37)/2 = 36 + a, (12 + 12)/2 = 12, (36 + a) × 12 = 432 +
12×a = 432 + 6×(2×a) = 432 + 6 = 438
3-61-7 17 11 19 14 227 Triangle Rule
(17×11)/2 = 17×(5+a) = 85 + 17×a = 85 + 8×(2×a) + 1×a
= 85 + 8 + 1×a = 93 + 1×a ru = 94, (19×14)/2 = 133,
94 + 133 = 227
5-140-30 28 11 27 9+ a 257 Multiplication of two adjacent sides
27 × (9+a) = 243 + (27×a) = 243 + 13×(2×a) + 1×a
= 243 + 13 + 1×a = 256 + 1×a ru = 257
*Field id: See Table 1. †Sidelengths: See Table 1. ‡Monads: See Table 1. In addition: bn, bone; am, arm. In averaged sidelengths, 0.5 = a. Abbreviations: See Table 1.

www.sciencemag.org SCIENCE VOL 320 4 APRIL 2008 75

RESEARCH ARTICLES
24.5T × 20T = 490T2. In ACA for hands, the rilateral with sidelengths 32, 11 + ht, 31, 12, and even terrain. Calculations with the simplest al-
exact RA for this field is computed as RA = 351T2 exemplifies application of the 3 arms gorithm require multiplication; more complex
to 1 “land rod” in ACA. Where am is arm and rd ones use addition and division as well. Compu-
ð24T þ 25TÞ=2 ¼ 24T þ a e is round down to integer, multiplication of the tations yielding the exact RAs imply that the
24T þ hd, ð16T þ 24TÞ=2 ¼ 20T, average length of one pair of opposite sides by an Acolhua approximated areas by rounding and
RA ¼ ð24 þ hdÞT
20T adjacent side gives averaging. The algorithms that successfully yield
Acolhua RAs are commonly applied in western
¼ 480T2 þ ð20
hdÞT2
¼ 480T2 þ 4
ð5
hdÞT2 ð11T þ 12TÞ=2 ¼ 11T þ a e 11T þ am, utilitarian metrology, which suggests that Acol-
RA ¼ 31T
ð11 þ amÞT hua categories of field shapes (squares, rectan-
¼ ð480 þ 4
3ÞT2
¼ 341T2 þ ð31
amÞT2 gles, and right-triangles) and their ascribed
¼ 492T2 ð4Þ ¼ 341T2 þ 10
ð3
amÞT2 þ 1
am geometric properties were not unlike European
¼ ð341 þ 10ÞT2 þ 1
am rd ones.
Bone: Cemomitl. Some RAs are obtained ¼ 351T2 ð6Þ Among many unresolved questions sug-
arithmetically by substituting 0.2T ( 1=5 T) for gested by this research is an emic explanation
0.5T. For example, for sidelengths 21 + a, 12 + Discussion and conclusions. The logic of for the choice of monads substituting for 1=2 in
a, 19, and 17, and an RA = 284T2, the Sur- Acolhua land measurement is based on units of averaging procedures. More broadly, we wonder
veyors’ Rule (ignoring sidelength monads) gives standard length and monad distances shorter than what factors determined the choice of algo-
(21T + 19T)/2 × (12T + 17T)/2 = 20T × 14.5T = the standard. Using an etic perspective, l6th- rithms to use in the first place. How accurate
290T2. In ACA with a proportion of 5:1, the century Spanish and subsequent scholars often Acolhua area values were, while emically sat-

Downloaded from www.sciencemag.org on July 17, 2010

RA is calculated as defined monads in terms of fractional parts of isfactory for their purposes, nevertheless stirs
Spanish varas and brazas. In contrast, from the etic curiosity.
(21T + 19T)/2 ¼ 20T, (12T + 17T)/2 Acolhua (and most likely Tenochca) emic view,
¼ 14T þ a e 14T þ 1=5 T, each monad was conceptualized as a unitary
entity unto itself, not a fraction of a larger unit. References and Notes
RA ¼ 20T
ð14 þ 1=5 ÞT 1. Located in the Fondo reservado Ms. 1497bis, Biblioteca
The metrological role of monads generally is
¼ 280T2 þ 20
1=5 T2 presumed to indicate only native concern with
Nacional de México, Universidad Nacional Autónoma de
México (UNAM), Mexico City.
¼ 280T2 þ 4
ð5
1=5 ÞT2 precise measurement. This study suggests a con- 2. Located in the Bibliothèque Nationale de France,
¼ ð280 þ 4ÞT2 siderably more complex function than surveyed- Ms. Mexicain 37-39, Paris.
3. H. R. Harvey, B. J. Williams, Science 210, 499
¼ 284T2 ð5Þ distance units. Our calculations provide strong (1980).
evidence that the Acolhua computed field areas 4. B. J. Williams, M. C. Jorge y Jorge, Symmetry Cult. Sci.
Although undepicted in the Tepetlaoztoc with ACA based on proportions between monads 12, 185 (2001).
codices, we suspect 1=5 T = 1=5 × 2.5 m = 0.5 m and the standard linear unit. Long debated, 5. T. N. Headland, K. L. Pike, M. Harris, Eds., Emics and
Etics: the Insider/Outsider Debate (Sage, Beverly Hills,
and represents a monad unit in Acolhua me- proposed metric equivalents of monads inferred
CA, 1990).
trology. As evidence, another Acolhua document from Colonial-period texts are testable for 6. P. Carrasco, The Tenochca Empire of Ancient Mexico: The
of the same vintage, the Oztoticpac Lands Map Acolhua units by computing with ACA applied Triple Alliance of Tenochtitlan, Tetzcoco, and Tlacopan
(17), pictures not only hands, hearts, and arrows to their field area records. Our research suggests (Univ. of Oklahoma Press, Norman, OK, 1999).
but also monad glyphs that unambiguously read that among the most commonly used monads, 7. D. Robertson, Mexican Manuscript Painting of the Early
Colonial Period: The Metropolitan Schools (Yale Univ.
“bone” (omitl ) in Nahuatl epigraphy. The read- the Acolhua hand measured 1.5 m, the arrow Press, New Haven, CT, 1959).
ing is confirmed by cemomitl glosses for bone 1.25 m, and the heart 1.0 m. More problematic, 8. B. J. Williams, H. R. Harvey, The Códice de Santa María
glyphs in Tenochca documents. Thus, we the arm may have equaled 0.83 m and the bone Asunción: Households and Lands in Sixteenth Century
tentatively identify the unknown Tepetlaoztoc 0.5 m. Tepetlaoztoc (Univ. of Utah Press, Salt Lake City, UT,
1997).
monad as an Acolhua “bone” equivalent to 0.5 m, Except for the 3 arms:1T equivalence, the 9. Papeles de la embajada americana, Museo Nacional de
a length slightly longer than what Cline (18) and symmetry of the most frequent proportions is Antropología, Archivo Histórico, Instituto Nacional de
Harvey (19) suggest for the Oztoticpac bone consistent with Mesoamerican metrics in that Antropología e Historia, Mexico City.
glyph (20). they are submultiples of 20— 2:1, 5:3, 5:2, and 10. F. de Alva Ixtlilxochitl, Obras históricas de Don
Fernando de Alva Ixtlilxochitl, publicadas y anotadas
Arm: Cemacolli. In addition to the bone, possibly 5:1. They resonate with emic emphasis por Alfredo Chavero (Editorial Nacional, Mexico City,
heart, hand, and arrow (fractions 1=5 T, 2=5 T, 3=5 T, on vigesimal numerology seen in cosmovision, 1952).
and 1=2 T, respectively), another monad is sug- divisions of time, and groupings of objects or 11. With some exceptions, Tenochca and Acolhua property
gested by nine quadrilaterals in which exact RAs people. In Mesoamerica, congruence principles plans share similar numerical symbols, but with
differing values. The cemmatl, “one hand,” is the
are obtained by replacing 0.5T with 1=3T, a length had long been used in calendrics to synchronize
Tenochca standard linear measure and equals two
equivalent to 1=3 × 2.5 m = 0.83 m (one vara). solar, lunar, and Venus cycles. It is therefore not Spanish varas (0.83 m × 2 ~ 1.67 m). Thus, it is
Probably this fraction represents computation unexpected that the Acolhua would use ACA in shorter than the Acolhua three-vara “land rod.” Unlike
with “one-arm,” cemacolli. A drawing of a bent their everyday arithmetic: What is unexpected is the “land rod,” the Tenochca cemmatl is typically
arm from the shoulder to the tips of outstretched to find documentary evidence of such proce- glyphed by a hand symbol.
12. H. J. Nissen, P. Damerow, R. K. Englund, Archaic
fingers typically reads “-acolli-” in Acolhua and dures. A persistent question in etic evaluation of Bookkeeping: Early Writing and Techniques of Economic
Tenochca epigraphy, and one such monad glyph Aztec arithmetical accomplishments is whether Administration in the Ancient Near East, trans. P. Larsen
appears on the Oztoticpac Lands Map mentioned they used fractions. We suggest that the Acolhua (Univ. of Chicago Press, Chicago, IL, 1993).
above. The arm glyph in Tenochca texts is often solved the arithmetic problem in an alternate 13. E. Robson, Mesopotamian Mathematics, 2100–1600 BC:
Technical Constants in Bureaucracy and Education
glossed “media braza,” i.e., one-half of the two- manner by using ACA, thus rendering the etic (Clarendon, Oxford, UK, 1999).
vara “braza”: 1=2 × (1.67 m) = 0.83 m, a metric question moot. 14. Examples of triangular-shaped field area calculations are
distance that coincides with the Tepetlaoztoc 1=3T. Because hundreds of Acolhua RAs can be found in (4).
In this instance, the Acolhua and Tenochca “arm” replicated through computation, it seems highly 15. Unless otherwise noted, Spanish translations and metric
equivalents of native measures not based on Tepetlaoztoc
distances coincide because the relations 3:1 and improbable that areas were derived by laying out data are drawn from (21–23).
2:1 establish proportions vis a vis the 3-vara quadriculars on fields or by some other physical 16. Two integer numbers are congruent “modulus n,” where
braza and the 2-vara braza. The CA of a quad- means, especially given long distances and un- n is an integer, if their remainders are the same when

76 4 APRIL 2008 VOL 320 SCIENCE www.sciencemag.org

 RESEARCH ARTICLES
divided by n. For example, 8 and 50 are congruent measures (codos in Spanish, cubits in English, in 24. We gratefully acknowledge the Departamento de
“modulus 2”, 21 and 56 are congruent “modulus 5.” Nahuatl specifically the cenmolicpitl, cenmatzotzopaztli, Matemáticas y Mecánica at Instituto de Investigaciones en
17. Located in the Library of Congress Map Division, cemomitl), all of which vary only by a few centimeters. Matemáticas Aplicadas y en Sistemas–UNAM for research
Washington, DC; http://memory.loc.gov. 21. M. Matías Alonso, Medidas indígenas de longitud and technical support. The expertise of A. C. Pérez
18. H. F. Cline, in Actas y memorias. Thirty-seventh (Centro de Investigaciones y Estudios Superiores en Arteaga was indispensable for research, design, and
International Congress of Americanists (ICA, Buenos Antropología Social, Mexico City, 1984). various iterations of the database, and graphics
Aires, 1968), vol. 3, pp.119–38. 22. V. M. Castillo, Estud. Cult. Nahuatl 10, 195 (1972). preparation. We thank A. R. Rodríguez for data entry.
19. H. R. Harvey, in Land and Politics in the Valley of Mexico: 23. L. Reyes García, E. Celestino Solís, A. Valencia Ríos,
A Two-Thousand-Year Perspective, H. R. Harvey, Ed. C. Medina Lima, G. Guerrero Díaz, Documentos Supporting Online Material
(Univ. of New Mexico Press, Albuquerque, NM, 1991), nahuas de la Ciudad de México del siglo XVI (Centro www.sciencemag.org/cgi/content/full/320/5872/72/DC1
pp. 163–185. de Investigaciones y Estudios Superiores en Antropología Tables S1 and S2
20. In ethnohistorical sources, considerable confusion Social, Archivo General de la Nación, Mexico City, 10 December 2007; accepted 22 February 2008
surrounds the definitions and lengths of lower arm 1996). 10.1126/science.1153976

conditions of low magnesium-ion concentra-

Crystal Structure of a Self-Spliced tion and high temperature. Indeed, the intron read-
ily self-spliced during in vitro transcription by
Group II Intron T7 RNA polymerase (fig. S1). The O. iheyensis
intron belongs to the newly discovered group
Navtej Toor,1* Kevin S. Keating,2 Sean D. Taylor,3 Anna Marie Pyle1,4* IIC class of introns (fig. S2), which are highly

Downloaded from www.sciencemag.org on July 17, 2010

reactive and smaller (420 to 480 nucleotides) than
Group II introns are self-splicing ribozymes that catalyze their own excision from precursor the well-studied IIA and IIB classes (17, 18).
transcripts and insertion into new genetic locations. Here we report the crystal structure of an The IIC introns are hypothesized to be the most
intact, self-spliced group II intron from Oceanobacillus iheyensis at 3.1 angstrom resolution. An primitive of the three classes (19), and they self-
extensive network of tertiary interactions facilitates the ordered packing of intron subdomains splice in vitro through a hydrolytic pathway (20),
around a ribozyme core that includes catalytic domain V. The bulge of domain V adopts an unusual forming a linear intron (18, 21) instead of a cy-
helical structure that is located adjacent to a major groove triple helix (catalytic triplex). The clized lariat product.
bulge and catalytic triplex jointly coordinate two divalent metal ions in a configuration that is The crystallization construct contains all of
consistent with a two–metal ion mechanism for catalysis. Structural and functional analogies the six domains typically present in group II
support the hypothesis that group II introns and the spliceosome share a common ancestor. introns (Fig. 1A). The intron was crystallized in a
postcatalytic state after undergoing both steps of
splicing during in vitro transcription, which results
roup II introns are self-splicing ribo- zyme is composed of six structural domains. in an intron RNA with homogeneously cleaved

G zymes that catalyze their own excision

from precursor-mRNAs (1). They also
function as retroelements by associating with
The largest, domain I (DI), contains recognition
sequences for binding the 5′ and 3′ exons and
the branch site nucleophile. The 5′ exon forms
ends. The intron was then isolated in the native
state, without the use of any denaturation steps
that are typical for RNA purification (22). Native
intron-encoded reverse transcriptases and invad- extended base-pairings with two binding sites in gel electrophoresis revealed the purified RNA to
ing DNA targets through reverse splicing reac- DI (EBS1 and EBS2) (8), and the 3′ splice site be conformationally homogeneous (fig. S3).
tions (2, 3). Group II introns are numerous in is specified by a short pairing between EBS3 The O. iheyensis intron crystallized in space
bacteria (4), and they are also found in the and the 3′ exon (9). Domains II and III (DII and group P212121 with one molecule per asymmetric
organellar genomes of plants, fungi, protists, DIII) enhance the catalytic efficiency of splicing unit. The structure was solved by multiwave-
and some animals (1, 5). They are considered to (10), and the linker that connects them (J2/3) is length anomalous dispersion (23). Phases were
be the ancestors of nuclear introns and of the a major active-site component (11, 12). Domain independently calculated from both Yb3+ and
eukaryotic spliceosomal machinery, with which IV (DIV) contains the reverse transcriptase ORF iridium hexamine (24) derivatives and combined,
they share structural and sequence similarities and structural motifs that bind the RT protein resulting in an experimental electron density map
(6, 7). Nuclear introns are regions of eukaryotic (13). The most highly conserved substructure in of high quality (Fig. 2 and figs. S4 to S6). The
transcripts that are removed after transcription group II introns is domain V (DV), which is a phylogenetically predicted secondary structure
by a large ribonucleoprotein complex called the short hairpin that contains a bulge essential for (17) served as a guide for building the model into
spliceosome. Despite the importance of group II catalysis (14). The bulge is located near a the electron density. The final model had an Rwork
introns and their spliceosomal relatives, there “catalytic triad” of conserved nucleotides (usually of 27.6% and an Rfree of 31.0% (table S1).
are no high-resolution crystal structures of these AGC, but often CGC) at the base of domain V Density was observed for most of the intron
genetic elements in an intact state. (15). Domain VI (DVI) contains an adenosine nucleobases, although density for the k region
Most group II intron RNAs encode two ba- nucleotide that attacks the 5′ splice site during the (Fig. 1A) and the internal loop of DIII was poor.
sic components: a self-splicing ribozyme and an first step of splicing, forming lariat RNA (1). Disordered regions of the intron included the first
open reading frame (ORF) for expression of a Although biochemical studies have revealed many two nucleotides of the 5′ end, DVI, and a small
reverse-transcriptase (RT) enzyme. The ribo- aspects of group II intron architecture (1, 9, 16), section of DI (Fig. 1A).
the exact spatial organization of functional Overall structure. The dominant feature of
1 domains and molecular details of the ribozyme the O. iheyensis intron structure is a network of
Department of Molecular Biophysics and Biochemistry, Yale
University, 266 Whitney Avenue, Bass Building, New Haven, active site have remained obscure. intricate tertiary interactions that organize active-
CT 06511, USA. 2Interdepartmental Program in Computa- Here we describe the crystal structure of site components around the DV catalytic center
tional Biology and Bioinformatics, Yale University, New an intact group II intron from the halotolerant (Fig. 1, C and D, and fig. S7A). Coaxial stack-
Haven, CT 06511, USA. 3Department of Molecular, Cellular alkaliphile Oceanobacillus iheyensis at 3.1 Å ing of specific helices dictates the overall archi-
and Developmental Biology, Yale University, New Haven, CT
06511, USA. 4Howard Hughes Medical Institute, Chevy Chase, resolution. This intron was identified in a screen tectural form (Fig. 1B). For example, helices I(i)
MD 20815, USA. of group II introns from various extremophilic and I(ii) from DI are coaxially stacked on DII.
*To whom correspondence should be addressed. E-mail: bacteria and was chosen as a crystallization Stem IC lies parallel to the I(i)-I(ii) helices in an
[email protected] (A.M.P.); [email protected] (N.T.) target because it exhibited robust splicing under orientation that is capped and stabilized by the

www.sciencemag.org SCIENCE VOL 320 4 APRIL 2008 77