University of Virginia Library

THE TRIPLE-WALLED TOWER AND ENVIRONS

When I first saw Pueblo del Arroyo, in June 1920, a broad pile of
earth and rubble was banked up against the middle west side (pl. 43).
It reached to the second-story ceiling level, yet no stonework showed
through. Building stones strewed the surface, but these plainly had
fallen from the upper, westernmost wall of the pueblo. Topping all
were several heaps of more recently turned earth and rock, thrown
out by unknown persons seeking archeological souvenirs. From the
base of the mound a less conspicuous accumulation sloped away to
merge with the valley floor.

At the west margin of this lesser accumulation, wall fragments
exposed by caving of the arroyo bank appeared, as I mentally projected

them, not only to underlie the larger mound but even Pueblo
del Arroyo itself (pl. 44, A). It was the presence of these fragmentary
walls, together with the possibility to which they seemed to point
so clearly, that persuaded me to include Pueblo del Arroyo in my initial
recommendations for a program of archeological investigation in
Chaco Canyon (Judd, 1954, p. vii).[2]

Our studies at Pueblo Bonito were inaugurated in 1921; those at
Pueblo del Arroyo, 2 years later. I had placed my chief assistant, Karl
Ruppert, in complete charge. But it was midsummer of 1926, after
he had completed scheduled excavations within the ruin, that Ruppert
led his crew to the low mound on the west side. His first exploratory
trenches cut through quantities of constructural debris—sandstone
spalls and rock-impressed chunks of adobe mud—nothing more. Not
until he dug down to the wall remnants in the arroyo bank did Ruppert
find something tangible from which to proceed. There, abutting
the masonry, was a packed clay surface easily followed. Tracing that
floor north and east brought to light a third wall, the innermost wall
of two concentric tiers of rooms encircling a central area (fig. 40).
Here, incredibly, was a triple-walled McElmo Tower in Chaco
Canyon!

Jackson (1878) and Holmes (1878), exploring southwestern Colorado
and adjoining territory in 1874 and 1875, first described the
McElmo country and its spectacular towers—circular and quadrangular,
oval and D-shaped. Both men regarded the smaller and more
numerous single-walled structures as possible lookouts; both recognized
the D-shaped and circular towers with radiating rooms as more
likely of religious than domiciliary function. Holmes, especially, was
intrigued by those with multiple walls. He reports four double-walled
towers on or near the Rio Mancos and one triple-walled tower at the
headwaters of McElmo Creek.

This latter was considered unique. Its like had not been reported
elsewhere. Fewkes (1916, p. 218, footnote 6), with understandable
skepticism, even recorded his doubt as to the reality of such a building.
To find a triple-walled tower in Chaco Canyon, therefore, was a
surprise for which we were entirely unprepared.

When Ruppert realized the significance of his discovery he proceeded
with customary skill and caution. The pavement he was following

was overlain by 3 feet of constructional debris—a conglomerate
of adobe mortar, sandstone spalls, and sand wetted by storm
waters and packed by time into an almost impenetrable mass. Pickaxes
loosened every inch. As a further handicap, the tower had been
almost completely razed; what remained was broken and disordered.
Two tiers of rooms circled a central area but neither tier was complete.
The innermost wall, more nearly razed than the other two,
was traceable for the most part only on its convex side. The highest
bit of masonry still standing, on the east, rose only 35 inches above
its floor. Sporadic fragments averaged 30 inches thick and rested
upon 21-inch-high foundations (pl. 51, upper).

Such masonry as survived consisted of salvaged building stones.
Among these, blocks of dressed friable sandstone predominated. In
some sections the blocks were laid in courses and chinked with bits
of laminate sandstone; elsewhere, courses of friable sandstone alternated
with equally thick bands composed of laminate sandstone tablets
an inch or more through. We doubt that there was a deliberate attempt
to imitate the masonry of Pueblo del Arroyo, but the use of
salvaged building blocks unquestionably resulted in stonework resembling
that in which the blocks were first employed.

The central room, 33 feet in diameter, was originally paved with
sandstone slabs but most of these had been removed, leaving only an
incomplete band at the outer edge and a broader segment in the northwest
quarter. We detected no trace of a fireplace, deflector, or other
feature. In contrast, the two encircling tiers had been floored with
adobe. Floor level appeared quite uniform throughout and just 6
inches above the base of the Pueblo del Arroyo west-wall masonry,
63 feet from the center of the Tower.

Rooms in the outer tier, perhaps 10 in number, were of unequal
length but averaged 6 feet wide; those in the inner tier averaged
6½ feet in width. The only indication of a lateral passageway anywhere
in the structure is the west jamb of an apparent door, sill
26 inches above the floor, in the outer south wall. That door, if ever
used, must have been short lived, since from the opposite side it was
blocked by the roof-level fill in the walled area enclosing Kivas "a"
and "b."

We measured the over-all outside diameter of the Tower as 73 feet
3 inches. We doubt that it was ever completed; that its rooms were
ever roofed. We believe the three concentric walls once stood considerably
higher but were razed midway of their intended height. We
observed no trace of wood, charred or otherwise, but demolition is

established by the quantities of broken wall adobe and sandstone spalls
revealed by the excavation.

Holmes (1878, pp. 398-399) describes his triple-walled McElmo
Tower as at the far side of a compact village situated on the edge of
a mesa overlooking a shallow valley. Its diameters approximated those
of "the great tower of the Rio Mancos," 25 feet inside and 43 feet
over-all; maximum height of standing walls is given as 12 feet. There
were 14 apartments about 5 feet wide in the outer tier; fallen masonry
filled the inner circle and concealed its partitions, if any. The third
and innermost wall was neither as thick nor as high as the other two
and, for this reason, Holmes assumed it to be that of a kiva. Walls
still 12 feet high after several hundred years evidence at least a
second story.

The stone towers of McElmo Creek and its tributaries, of the Rio
Mancos and the upper San Juan, have awakened the interest and
curiosity of at least three generations of archeologists, and the purpose
for which they were built is still obscure. Jackson and Holmes
supposed the single-walled structures to have been connected with the
defense of nearby communities; those of more complex construction
to have been associated with ceremonial practices. Schulman (1950)
summarizes a considerable body of fieldwork since Holmes and Jackson,
but finds nothing to corroborate or deny their early surmise. He
passes over the double- and triple-walled towers which are our sole
interest at the moment.

Holmes's great tower on the upper McElmo has never been excavated
so far as I know. He gives its over-all diameter as 43 feet. The
triple-walled structure back of Pueblo del Arroyo, partially built and
then demolished, measured 73 feet 3 inches. Holmes, with time for
examination limited, gives few constructional details, and our information
is equally meager. Data are not yet available for a third triplewalled
tower, recently discovered near Aztec, N. Mex., midway between
McElmo Creek and Chaco Canyon. Architecture identifies all
three as of Pueblo III age, and a note in American Antiquity (vol. 20,
No. 1, p. 96) reports that Gordon R. Vivian, repairing the one at
Aztec for the National Park Service in the fall of 1953, found earlier
Pueblo III remains underneath.

Architecture and pottery together identify the canyon-head ruins
of the McElmo-Yellowjacket-Hovenweep area (Morley, 1908; Morley
and Kidder, 1917; Fewkes, 1916, 1918, 1925; Kidder, 1924, p. 65)
more specifically with the distinctive Mesa Verde culture, a phase of
Pueblo III civilization that culminated in the justly famous cliff

dwellings of what is now the Mesa Verde National Park, Colorado.
Recorded tree-ring dates from those cliff dwellings lie between A. D.
958 and 1273 (Smiley, 1951, p. 23). We found no scrap of wood in
our triple-walled Chaco Canyon Tower, but 35 constructional timbers
from nearby Pueblo del Arroyo were felled in the 66-year bracket
A. D. 1052-1117 (ibid., p. 19). Since floor level in our Tower is only
6 inches above that in the pueblo we may assume that the latter antedated
the Tower by a relatively brief interval.

There can be no doubt but that the double- and triple-walled towers,
with their concentric circles and radiating rooms, are related to the
Great Kiva of late Pueblo III times as Morris (1921, p. 138) pointed
out long ago. But no one has yet fathomed the nature and extent of
that relationship. We must await future studies in Holmes's triplewalled
tower on the upper McElmo and in the newly discovered example
at Aztec. Ours at Pueblo del Arroyo provided few helpful
data and no artifacts.

Between the Tower and the west side of del Arroyo are two kivas,
almost wholly razed, each within a partially demolished rectangle.
We cleared the area to what I regarded as a safe, protective level and,
in the process, exposed a number of other walls, or sections of walls,
the significance of which was not always apparent. Doubtless we
missed still others, but all those actually uncovered, like the great
Tower, had been built with reclaimed rock, chiefly dressed blocks of
friable sandstone, and all had been razed to the last few courses.
South of the Tower are two more kivas, and I shall present the four
in the order in which they were excavated.

Kiva "a" (fig. 42) averages 12 feet 5 inches in diameter at the floor;
its bench averages 6½ inches wide and 22 inches high. Both the bench
and the wall above were constructed of fairly large sandstone blocks,
shaped and unshaped, irregularly and unevenly coursed, heavily plastered
and sooted. There were no pilasters; no south recess. At the
time of excavation the main wall rose 3½ feet above the bench at the
northeast but only 2 feet at the west.

The ventilator outlet, in the south bench with sill at floor level, is
10 inches wide by 12 inches high; lintel poles support the masonry
above. The tunnel extends S. 2° E. 23 inches to connect with a 10-inchsquare
shaft that rises 22 inches above the level of the kiva floor and
opens upon a pavement outside the enclosing wall.

An irregular, clay-lined fireplace 25 inches in diameter and 7 inches
deep lies 4½ feet north of the ventilator; it contained four sandstone
firedogs and was filled with wood ash. The fireplace had been built

above the southwest corner of a masonry-lined pit, 53 inches east-west
by 31 inches by 17½ inches deep, previously filled with constructional
waste and floored over. Although the fill included neither ash nor
charcoal, the pit walls and floor were more or less reddened from fires.
There was no deflector between ventilator and fireplace.

Fig. 42.—Kivas "a" and "b."

The Kiva "a" floor is 2½ inches above that in the Tower, but the
west side of the enclosing square abuts the Tower masonry 35 inches
above its base. At the time of excavation Kiva "a" was filled with
blown sand and rubble from its razed walls. In this we found nothing
but a handful of late-type potsherds and the broken earthenware pipe
illustrated by figure 44. Its bowl is ⅝ inch deep. The three longitudinal
punctate zigzags by way of ornamentation are reminiscent of an earlier
culture, but here it is, associated with late Chaco sherds in the fill of a

non-Chaco kiva abutting the remains of a McElmo triple-walled
Tower.

Kiva "b" lies immediately east of "a" and within the same enclosure.
It averages 11 feet 7 inches in diameter at the floor; its upper wall
had been razed to bench level. The bench, averaging 6 inches wide
and 24 inches high, is of masonry that includes shaped and unshaped
blocks of both friable and laminate sandstone, heavily plastered and
smoke stained. As in "a," there are no pilasters and no south recess.
The floor is 13 inches above that in Kiva "a" and 16 inches above that
in the Tower.

A ventilator at floor level was recessed 1 inch to receive a neatly
fitted sandstone slab measuring 13½ by 15½ inches by ¾ inch thick
(found on floor fronting outlet). It opens into a 12-by-13-inch tunnel
that is roofed with small poles and extends S. 10° W. 24 inches to
connect with an 11-by-12-inch masonry shaft the outside of which
stands 5 inches beyond the enclosing wall. A Pueblo II metate 1½
inches thick, mano groove to the south, stands embedded in the kiva
floor 29½ inches from the ventilators and forms a fireplace screen
20 inches wide by 17½ inches high (pl. 51, lower). Slab fragments
wedged in at either side provided basal support. A rectangular, claylined
fireplace lies 22½ inches north of the deflector and, like that in
"a," it contained four sandstone firedogs and was filled with wood ash.

In the corner between Kiva "b" and the Tower is an odd-shaped,
doorless cell with its adobe floor 3 inches above that in the kiva and
19 inches above that in the adjacent Tower. The east wall of this
corner room abutted the Tower masonry and was left standing free
when the latter was razed. The fact that only the convex north wall
is plastered suggests that our triple-walled Tower had been coated
externally with mud at the time of construction.

Kiva "c" (fig. 43) lies between "b" and Pueblo del Arroyo and is
surrounded by straight walls that form a generous enclosure. As in
Kiva "b," the upper wall had been razed to bench level, but in this
case the bench masonry had also been stripped away, leaving only its
rough stone-and-adobe core. Nevertheless, we learned that the bench
had been 19 inches wide and 26 inches high. Mud plaster 1¾ inches
thick and representing successive resurfacings still adhered to a section
of bench at the north side of the chamber. Kiva "c" averaged
17 feet 3 inches in diameter, and its floor, in contrast to those in "a"
and "b," is 2½ inches lower than that of the Tower.

A Chaco-type south recess, 8 feet 3 inches wide at the rear, interrupted
the Kiva "c" bench for its full height and depth. Passing under
the middle of this recess is a subfloor ventilator tunnel or duct 19

inches wide, masonry lined and roofed with small poles. It continues
S. 19° W. 4 feet, there to connect with the air intake, a 16-by17-inch
masonry shaft. Since we did not clear the middle of the kiva,

Fig. 43.—Kiva "c."

the total length of this duct and its depth were not ascertained. Neither
did we uncover the customary fireplace and the deflector, if any.

On the west side of the kiva is a subfloor vault, 8 feet 1 inch long,
25 inches wide, and 14 inches deep (pl. 50, A). Its north, west, and

south sides are lined with clay but the east side is a masonry composite—laminate
sandstone of fairly uniform thickness in the lower
courses but larger blocks of friable sandstone, both shaped and unshaped,
above. Paralleling the west side at a distance of 7 inches and
extending 16 inches beyond each end is the impression of a slender
pole and, directly beneath it, the clay-filled imprint of its predecessor.
At each end of the vault is a sort of border, 15 inches north-south,
bearing the imprints of unpeeled willows, split cedar, and cedar bark.
These imprints curve upward slightly at the west as though the materials
that caused them had once rested upon the lower pole. There is
no corresponding pole impression on the east side, and although the
"border" at the south end extends 4 inches to either side of the vault,
that at the north extends to the west only.

In no other Chaco kiva did we find equally conspicuous imprints
about one of the puzzling subfloor vaults. As we interpreted the evidence
here, this particular example had been roofed at floor level with
the materials customarily employed in house ceilings—excepting only
the supporting poles which may not have been considered necessary
in bridging a 2-foot space. But imprints of the supposed covering
occur at the vault ends only, not along the sides. The original pole
on the west edge had been removed and its impression packed with
mud to cushion a substitute. The new pole rested directly above the
imprint of the old one. Then a new floor was laid, at least from
bench to vault, rising 2 inches to top the substitute pole and the northend
"border." A well-turned hematite cylinder (U.S.N.M. No.
334781), ½ inch in diameter by 1[fraction 3 by 16] inches long, lay at the north end
of the upper pole imprint.

Unlike the majority of kiva vaults cleared in the course of our
Chaco Canyon investigations, that in "c" had not been filled and
floored over while the kiva continued in use. In this particular
instance the vault apparently served its unknown purpose until demolition
was ordered, at which time its covering was removed and wreckage
from the razed walls was allowed to fill kiva and vault indiscriminately.
In the vault fill we found several large sherds of an olla
decorated in Late Hachure and the burned abrading stone shown in
plate 40, s.

Kiva "d" lies north of "c" and within walls that abut both the
McElmo Tower and the west side of Pueblo del Arroyo. It had been
more nearly demolished than the three already described. Its main
wall and the bench, if any, had been completely razed on the east
and south, to within 10 inches of its floor on the west, and to within

18 inches on the north. We detected no trace of a south recess. A
floor diameter of 15 feet 3 inches is indicated.

The priestly builders of Kiva "d" apparently had difficulty in deciding
just what they wanted. The floor is of packed sand with a
minimum of binding clay, rounding off with the wall plaster and
blackened through use. We counted only two layers of this plaster,
an initial ¼-inch spread directly upon the stonework and a second thin
finishing wash of whitened clay. On the sandy floor, which is 5 feet
7 inches above that in the central room of the Tower, we found no
fireplace although a small quantity of wood ash was scattered over
the middle of it.

Constructional indecision
seems indicated because the
kiva masonry continues 4
feet 5 inches below floor
level with apparent work
surfaces at depths of 5
inches and 4 feet 2 inches,
respectively. The lower of
these two surfaces is in-

distinct and uneven; the upper, less irregular and firmly packed. Each
is overlain by a purposeful fill of clean sand. From top to bottom the
masonry consists of dressed blocks of friable sandstone often somewhat
angular in shape. Absence of laminate sandstone is particularly
noticable. Wall plaster does not extend to either of these work
surfaces.

Kiva "d" is enclosed on the north and south by straight walls that
abut both the Tower and the pueblo. The north enclosing wall, standing
upon a 6-inch foundation of large, irregular blocks of friable
sandstone, abuts the Tower masonry 25 inches above its foundation
and the west wall of Pueblo del Arroyo 22 inches above its base. Here,
143 feet 8 inches from the northwest corner of the pueblo, the abutting
wall had been completely razed, leaving only its foundation.
Broken masonry above the union suggests that the former enclosing
wall had been tied into that of the pueblo. At the other end of this
east-west wall, 25 inches above the Tower foundation, a pavement
extends floorlike to the north and east.

The opposite enclosing wall, 17 feet distant and razed to within
14 inches of its foot-high foundation, abuts the pueblo masonry
31 inches above its foundation. Seven feet beyond is the 16-inchthick
north wall of the Kiva "c" enclosure and, 3 feet farther, a

shorter parallel wall, 22 inches thick. The latter stands 2 feet higher
than the first, but both were erected with reclaimed building stones
and both abut the plastered exterior of the pueblo.

By themselves these two examples of rather rude construction
would excite little interest, but what we cannot comprehend is a mass
of solid rubble—chunks of sandstone in adobe mud—3½ feet from the
pueblo, standing on the same level as the two rude walls and overlying
both of them. This feature, in its present condition, defies explanation.
Between and above the two walls and, again, north of the first, stones
have been pried from the pueblo wall as though to tie in other stonework
at a higher level (pl. 52, A). North and south from this puzzling
assemblage lay a thick blanket of constructional debris or, more likely,
waste from the razing of these four small kivas and their enclosing
walls (pl. 52, B).

The reader will have noticed that not only the triple-walled Tower
but the four kivas and their individual enclosures were all built of
reclaimed building stones and that all were subsequently demolished.
Although some allowance must be made for distance, the Tower appears
to have been erected first, for its central pavement is only 6
inches above the Pueblo del Arroyo west-wall foundation while the
kiva floors, with one exception, are higher and, their enclosing walls
higher still. Kiva "c," the only one of the four with Chaco features,
was floored 2½ inches lower than the Tower. The one identifiable
floor in Kiva "d" lies 5 feet 7 inches above the Tower floor but its
associated masonry extends down to within 14 inches of that floor.
There is no doubt in my mind that the four kivas and their enclosures
were built after the Tower, but I cannot say all were razed at the
same time, however plausible this may seem.

In exploring what remained of the Tower and its environs we found
comparatively little in the way of cultural material. The RobertsAmsden
tabulation shows a preponderance of Chaco-San Juan sherds
from the Tower trenches and a high proportion of Mesa Verde blackon-white.
A number of miscellaneous artifacts were unearthed as we
removed the accumulation between the Tower and the pueblo. Most
of them came from a 40-foot space between the rubble pile northeast
of Kiva "c" and a point opposite Room 49; all, apparently, had been
discarded with floor sweepings. The following list will show the
number and character of this material: 18 bone awls, including only 1
of bird bone; 10 tubular bone "beads," ⅞-2¼ inches long; the blade
fragment of a spatulate bone knife (U.S.N.M. No. 334930) and a
dozen pieces of worked mammal bones; 5 chipped arrowheads, 3 of

them with square base and side notches; a former ¾-grooved ax, regrooved,
broken, and used as a maul (pl. 41, g); a friable sandstone
building block, smoothed on the face and incised with crossed lines
(No. 334872); a sandstone disk 2 inches in diameter by [fraction 3 by 16] inch
thick and 3 disks ⅞-1⅞ inches in diameter, made from potsherds; the
neatly squared fragment of a fine-grained sandstone tablet ¼ inch
thick with a half-inch circle gouged out at the corner (No. 334809);
8 fragments of human-effigy vessels, all but one decorated with mineral
paint (pl. 36, A); 2 fragments of a double-bowl rectangular
redware vessel, polished black in one half, red in the other (No.
334677); part of the handle from a Chaco-San Juan style ladle
(fig. 23), and the miniature half-gourd ladle with broken handle
shown on plate 27, j; and portions of several human skeletons (field
No. 596). The articulated skeleton of an adult male or female (field
No. 191; U.S.N.M. No. 327137) appeared to have been interred,
without grave furniture, among the accumulated debris south of the
Kiva "c" square.

Nearby but not immediately associated with this burial was a small
fragment of an apocynum-fiber sandal (No. 334712) and part of a
comblike bone object, concavo-convex in cross section and 2⅞ inches
long, with 5 "teeth" polished by friction on the underside (pl. 37, e).
Farther south, outside Room 16 and 21 inches above its floor level,
was a pile of burned timbers and burned debris of reconstruction, the
clean-up after a fire somewhere in the village. Room 24 is not regarded
as a likely source of the pile despite the fact that its east wall, with
the ends of several charred ceiling poles still present, had been burned
above a 33-inch-deep accumulation of collapsed masonry and blown
sand. From this same general area, between Kiva "c" and the southwest
corner of the pueblo, we collected 3,988 potsherds of which
Roberts and Amsden tabulated 138 as Corrugated-coil culinary ware,
94 as Chaco-San Juan, and 12 as of Late Hachure. Only 31 were
types that, at Pueblo Bonito, I would have listed as products of the
Old Bonitians.