Within the middle Verde region there are thousands of cavate lodges, sometimes in clusters of two or three, oftener in small groups, and sometimes in large groups comprising several hundred rooms. One of these large groups, located some 8 miles south of Verde on the eastern side of the river, has been selected for illustration.

The bottom lands of the Rio Verde in the vicinity of Verde have been already described, and the cavate lodges in question occur just below the southern end of this large area of tillable land, and some of them overlook it. The river at this point flows southward, and extending toward the east are two little canyons which meet on its bank. North and south of the mouth of the canyons the bank of the river is formed by an inaccessible bluff 180 or 200 feet high. These bluffs are washed by the Verde during high water, though there is evidence that up to a recent time there was a considerable area of bottom land between the river and the foot of the bluff. Plate XXIV shows the northern end of the group from a low mesa on the opposite side of the river; the eastern bank of the river can be seen in the foreground, while the sandy area extending to the foot of the bluff is the present high-water channel of the Verde. The map (plate XXV) shows the distribution of the cavate lodges composing the group, and plate XXVI shows the character of the site. The cavate lodges occur on two distinct levels—the first, which comprises nearly all the cavate lodges, is at the top of the slopes of talus and about 75 feet above the river; the second is set back from 80 to 150 feet from the first tier horizontally and 30 or 40 feet above it. The cavate lodges occur only in the face of the bluff along the river and in the lower parts of the two little canyons before mentioned. These canyons run back into the mesa seen in the illustration, which in turn forms part of the foothills rising into the range of mountains hemming in the Rio Verde on the east.



The walls of the canyon in the cavate-lodge area are composed of three distinct strata, clearly defined and well marked. The relations of the strata, at points on the northern and western sides of the north canyon, are shown in figure 289 and plate XXVI. The lowest stratum shown in the figure is that in which almost all the cavate lodges occur. It is about 8 feet thick and composed of a soft, very friable, purple-gray sandstone. Above it lies a greenish-white bed a few inches thick, followed by a stratum of a pronounced white, about 12 feet thick. This heavy stratum is composed of calcareous clay, and the green bed of a calcareous clay with a mixture of sand. The white stratum is divided at two-thirds its height by a thin belt of greenish-white rock, and above it there is another belt of purple-gray sandstone about 12 feet thick. The top of this sandstone forms the ground surface south of the point shown in the diagram, while on the north and east it forms the floor of the upper tier of cavate lodges.



On the southern side of the canyon the lower purple stratum shows three distinct substrata; the upper is reddish purple and about 3 feet thick, the middle is purple gray, about 7 feet thick, and apparently softer than the upper and lower strata. The lodges occur in the middle purple substratum, their floors composed of the upper surface of the lower stratum and their roofs of the under surface of the upper stratum. Those on the north side are similarly placed, their roofs being about 3 feet below the white, except that in several instances the upper part of the purple up to the white has fallen, making the cavity larger. This has occurred, however, since the abandonment of the caves, and the debris, still fresh looking, is in situ.

The formation in which the lodges occur is not of volcanic origin, although the beds composing it were perhaps deposited by hot springs during the period of great volcanic activity which produced San Francisco mountain in central Arizona and the great lava flows south of it. In view of the uncertainty on this point and the further fact that almost all the cavate lodges heretofore found were excavated in tufa, ash, or other soft volcanic deposits, the report of Mr. Joseph S. Diller, petrographer of the U.S. Geological Survey, will be of interest. It is as follows:

The coarse-grained specimen is sandstone, that of medium grain is argillaceous sandstone, and the fine-grained one is calcareous clay. The coarse-grained friable sandstone, in which the lodges have been excavated, consists chiefly of subangular and rounded grains of quartz and feldspar with a small proportion of black particles. Many of the latter are magnetite, while the others are hornblende and various ferromagnesian silicates. I did not detect any fragments of volcanic origin.

The specimen of argillaceous sandstone is made up of thin layers of fine-grained sand of the same sort as the first, alternating with others containing considerable clay. In the clay layers, a trace of carbonate of lime was found here and there, forming a transition of the calcareous clay.

The calcareous clay when placed in acid effervesces vigorously, but when allowed to stand the effervescence ceases in a few minutes and the insoluble white clay remains.

All the strata composing this formation are very soft; the purple-gray material of the middle layer is so soft that its surface can be rubbed off with the hand. They are also minutely stratified or laminated, and the lamin are not well cemented together, so that a blow on the roof of a cavity with a stone or other implement will bring off slabs varying from half an inch to an inch and a half in thickness. These thin strata or lamin are of unequal hardness, weathering in places several inches into the face of the rock in thin streaks of a few inches or less. The middle purple stratum exhibits this quality somewhat more decidedly than the others, and this fact has doubtless determined the selection of this stratum for the location of the lodges, as a room can be excavated in it more easily than a room of a similar size could be built up with loose rock.

The almost absolute dependence of the native builder on nature as he found it is well illustrated by these cavate lodges. At a point in the northern wall of the northernmost canyon, shown in the diagram (figure 289) and in plate XXVI, there is a small fault with a throw of about 2 feet, and the floors of the lodges west of the fault are just that much lower than the floors east of it. Furthermore, where the purple-gray stratum in which the lodges occur is covered up by the rising ground surface, the cavate lodges abruptly cease. In the northern and southern ends of the group the talus encroaches on and partly covers the purple-gray stratum, and in these places the talus has been removed from the face of the rock to permit the excavation of lodges. In short, the occurrence of the cavate lodges in this locality is determined absolutely by the occurrence of one particular stratum, and when that stratum disappears the lodges disappear. So far as can be ascertained without actually excavating a room there is no apparent difference between the stratum in which the lodges occur and the other purple strata above and below it. That there is some difference is indicated by the confinement of the lodges to that particular level, but that the difference is very slight is shown by the occurrence in two places of lodges just above the principal tier, a kind of second-story lodge, as it were. It is such differences in environment as these, however, often so slight as to be readily overlooked, which determine some of the largest operations carried on by the native builders, even to the building of some of the great many-storied pueblos, and, stranger still, sometimes leading to their complete abandonment.

In the region under discussion cavate lodges usually occur in connection with and subordinate to village ruins, and range in number from two or three rooms to clusters of considerable size. Here, however, the cavate lodge is the feature which has been most developed, and it is noteworthy that the village ruins that occur in connection with them are small and unimportant and occupy a subordinate position.

There are remains of two villages connected with the cavate lodges just described, perched on the points of the promontories which form the mouths of the two canyons before mentioned. The location of these ruins is shown in plate XXV. The one on the southern promontory is of greater extent than that on the northern point, and both are now much broken down, no standing wall remaining. A general view of the ruin on the northern promontory is given in plate XXVII, and the same illustration shows the remains of the other village on the flat top of the promontory in the farther part of the foreground.



The cavate lodges are generally rudely circular in shape, sometimes oblong, but never rectangular. The largest are 25 and even 30 feet in diameter, and from this size range down to 5 or 6 feet and thence down to little cubby-holes or storage cists. Owing to their similarity, particularly in point of size, it is difficult to draw a line between small rooms and large storage cists, but including the latter there are two hundred rooms on the main level, divided into seventy-four distinct and separate sets. These sets comprise from one to fourteen rooms each. On the upper level there are fifty-six rooms, divided into twenty-four sets, making a total of two hundred and fifty-six rooms. As nearly as can be determined by the extent of these ruins the population of the settlement was probably between one hundred and fifty and two hundred persons.



There is great variety in the rooms, both in size and arrangement. As a rule each set or cluster of rooms consists of a large apartment, entered by a narrow passageway from the face of the bluff, and a number of smaller rooms connected with it by narrow doorways or short passages and having no outlet except through the large apartment. As a rule two or more of these smaller back rooms are attached to the main apartment, and sometimes the back rooms have still smaller rooms attached to them. In several cases there are three rooms in a series or row extending back into the rock, and in one instance (at the point marked E on the map, plate XXV) there are four such rooms, all of good size.

Attached to the main apartment, and sometimes also to the back rooms, there are usually a number of storage cists, differing from the smaller rooms of the cluster only in size. These cists or cubby-holes range in size from a foot to 5 feet in diameter, and are nearly always on a level of the floor, although in some instances they extend below it. Storage cists are also sometimes excavated in the exterior walls of the cliffs, and occasionally they are partly excavated and partly inclosed by a rough, semicircular wall. An example of the latter type is shown in figure 290.

As a rule the cavate lodges are set back slightly from the face of the bluff and connected with it by a narrow passageway. Another type, however, and one not uncommon, has no connecting passageway, but instead opens out to the air by a cove or nook in the bluff. This cove was used as the main room and the back rooms opened into it in the usual way by passageways. A number of lodges of this type can be seen in the eastern side of the northern promontory or bluff. Possibly lodges of this type were walled in front, although walled fronts are here exceptional, and some of them at least have been produced by the falling off of the rock above the doorway. The expedient of walling up the front of a shallow cavity, commonly practiced in the San Juan region, while comparatively rare in this vicinity, was known to the dwellers in these cavate lodges. At several points remains of front walls can be seen, and in two instances front walls remain in place. The masonry, however, is in all cases very rough, of the same type as that shown in plate XXVIII.

In this connection a comparison with the cavate ledges found in other regions will be of interest. In 1875 Mr. W. H. Holmes, then connected with the Hayden survey, visited a number of cavate lodges on the Rio San Juan and some of its tributaries. Several groups are illustrated in his report.[5] Two of his illustrations, showing, respectively, the open front and walled front lodges, are reproduced in plates XXIX and XXX. The open front lodges are thus described:

I observed, in approaching from above, that a ruined tower stood near the brink of the cliff, at a point where it curves outward toward the river, and in studying it with my glass detected a number of cave-like openings in the cliff face about halfway up. On examination, I found them to have been shaped by the hand of man, but so weathered out and changed by the slow process of atmospheric erosion that the evidences of art were almost obliterated.

The openings are arched irregularly above, and generally quite shallow, being governed very much in contour and depth by the quality of the rock. The work of excavation has not been an extremely great one, even with the imperfect implements that must have been used, as the shale is for the most part soft and friable.

A hard stratum served as a floor, and projecting in many places made a narrow platform by which the inhabitants were enabled to pass along from one house to another.

Small fragments of mortar still adhered to the firmer parts of the walls, from which it is inferred that they were at one time plastered. It is also extremely probable that they were walled up in front and furnished with doors and windows, yet no fragment of wall has been preserved. Indeed, so great has been the erosion that many of the caves have been almost obliterated, and are now not deep enough to give shelter to a bird or bat.

[Footnote 5: Tenth Ann Rep. U.S. Geol. Survey, 1876, pp. 288-391.]



Walled fronts, the author states, were observed frequently on the Rio Mancos, where there are many well-preserved specimens. He described a large group situated on that stream, about 10 miles above its mouth, as follows:

The walls were in many places quite well preserved and new looking, while all about, high and low, were others in all stages of decay. In one place in particular, a picturesque outstanding promontory has been full of dwellings, literally honeycombed by this earth-burrowing race, and as one from below views the ragged, window-pierced crags [see plate XXX] he is unconsciously led to wonder if they are not the ruins of some ancient castle, behind whose moldering walls are hidden the dread secrets of a long-forgotten people; but a nearer approach quickly dispels such fancies, for the windows prove to be only the doorways to shallow and irregular apartments, hardly sufficiently commodious for a race of pigmies. Neither the outer openings nor the apertures that communicate between the caves are large enough to allow a person of large stature to pass, and one is led to suspect that these nests were not the dwellings proper of these people, but occasional resorts for women and children, and that the somewhat extensive ruins in the valley below were their ordinary dwelling places.

It will be noticed that in both these cases there are associated ruins on the mesa top above, and in both instances these associated ruins are subordinate to the cavate lodges, in this respect resembling the lodges on the Verde already described. This condition, however, is not the usual one; in the great majority of cases the cavate lodges are subordinate to the associated ruins, standing to them in the relation of outlying agricultural shelters. Unless this fact is constantly borne in mind it is easy to exaggerate the importance of the cavate lodges as compared with the village ruins with which they are connected.

The cavate lodges near San Francisco mountain in Arizona were visited in 1883 by Col. James Stevenson, of the Bureau of Ethnology, and in 1885 by Maj. J. W. Powell. Major Powell[6] describes a number of groups in the vicinity of Flagstaff. Of one group, situated on a cinder cone about 12 miles east of San Francisco peak, he says:

Here the cinders are soft and friable, and the cone is a prettily shaped dome. On the southern slope there are excavations into the indurated and coherent cinder mass, constituting chambers, often 10 or 12 feet in diameter and 6 to 10 feet in height. The chambers are of irregular shape, and occasionally a larger central chamber forms a kind of vestibule to several smaller ones gathered about it. The smaller chambers are sometimes at the same altitude as the central or principal one, and sometimes at a lower altitude. About one hundred and fifty of these chambers have been excavated. Most of them are now partly filled by the caving in of the walls and ceilings, but some of them are yet in a good state of preservation. In these chambers, and about them on the summit and sides of the cinder cone, many stone implements were found, especially metates. Some bone implements also were discovered. At the very summit of the little cone there is a plaza, inclosed by a rude wall made of volcanic cinders, the floor of which was carefully leveled. The plaza is about 45 by 75 feet in area. Here the people lived in underground houses—chambers hewn from the friable volcanic cinders. Before them, to the south, west, and north, stretched beautiful valleys, beyond which volcanic cones are seen rising amid pine forests. The people probably cultivated patches of ground in the low valleys.

About 18 miles still farther to the east of San Francisco mountain, another ruined village was discovered, built about the crater of a volcanic cone. This volcanic peak is of much greater magnitude. The crater opens to the eastward. On the south many stone dwellings have been built of the basaltic and cinder-like rooks. Between the ridge on the south and another on the northwest there is a low saddle in which other buildings have been erected, and in which a great plaza was found, much like the one previously described. But the most interesting part of this village was on the cliff which rose on the northwest side of the crater. In this cliff are many natural caves, and the caves themselves were utilized as dwellings by inclosing them in front with walls made of volcanic rocks and cinders. These cliff dwellings are placed tier above tier, in a very irregular way. In many cases natural caves were thus utilized; in other cases cavate chambers were made; that is, chambers have been excavated in the friable cinders. On the very summit of the ridge stone buildings were erected, so that this village was in part a cliff village, in part cavate, and in part the ordinary stone pueblo. The valley below, especially to the southward, was probably occupied by their gardens. In the chambers among the overhanging cliffs a great many interesting relics were found, of stone, bone, and wood, and many potsherds.

[Footnote 6: Seventh Ann. Rep. Bur. Eth., 1891, p. xix.]

It will be seen that the first group described bears a remarkably close resemblance to the cavate lodges on the Rio Verde. The lodges themselves are smaller, but the arrangement of main apartment and attached back rooms is quite similar. It will be noticed also that in the second group described village ruins are again associated on the summit of the cliff or ridge. Major Powell ascertained that these cavate lodges were occupied by the Havasupai Indians now living in Cataract canyon, who are closely related to the Walapai, and who, it is said, were driven from this region by the Spaniards.

The cavate lodges on the Rio Grande, in New Mexico, in the vicinity of the modern pueblo of Santa Clara, were also visited in 1885 by Major Powell and are thus described by him:[7]

The cliffs themselves are built of volcanic sands and ashes, and many of the strata are exceedingly light and friable. The specific gravity of some of these rocks is so low that they will float on water. Into the faces of these cliffs, in the friable and easily worked rock, many chambers have been excavated; for mile after mile the cliffs are studded with them, so that altogether there are many thousands. Sometimes a chamber or series of chambers is entered from a terrace, but usually they were excavated many feet above any landing or terrace below, so that they could be reached only by ladders. In other places artificial terraces were built by constructing retaining walls and filling the interior next to the cliffs with loose rock and sand. Very often steps were cut into the face of a cliff and a rude stairway formed by which chambers could be reached. The chambers were very irregularly arranged and very irregular in size and structure. In many cases there is a central chamber, which seems to have been a general living room for the people, back of which two, three, or more chambers somewhat smaller are found. The chambers occupied by one family are sometimes connected with those occupied by another family, so that two or three or four sets of chambers have interior communication. Usually, however, the communication from one system of chambers to another was by the outside. Many of the chambers had evidently been occupied as dwellings. They still contained fireplaces and evidences of fire; there were little caverns or shelves in which various vessels were placed, and many evidences of the handicraft of the people were left in stone, bone, horn, and wood, and in the chambers and about the sides of the cliffs potsherds are abundant. On more careful survey it was found that many chambers had been used as stables for asses, goats, and sheep. Sometimes they had been filled a few inches, or even 2 or 3 feet, with the excrement of these animals. Ears of corn and corncobs were also found in many places. Some of the chambers were evidently constructed to be used as storehouses or caches for grain. Altogether it is very evident that the cliff houses have been used in comparatively modern times; at any rate, since the people owned asses, goats, and sheep. The rock is of such a friable nature that it will not stand atmospheric degradation very long, and there is abundant evidence of this character testifying to the recent occupancy of these cavate dwellings.





Above the cliffs, on the mesas, which have already been described, evidences of more ancient ruins were found. These were pueblos built of cut stone rudely dressed. Every mesa had at least one ancient pueblo up off it, evidently far more ancient than the cavate dwellings found in the face of the cliffs. It is, then, very plain that the cavate dwellings are not of great age; that they have been occupied since the advent of the white man, and that on the summit of the cliffs there are ruins of more ancient pueblos.

[Footnote 7: Seventh Ann. Rep. Bur. Eth., op. cit., p. XXII.]

Major Powell obtained a tradition of the Santa Clara Indians, reciting three successive periods of occupancy of the cavate lodges by them, the last occurring after the Spanish conquest of New Mexico in the seventeenth century.

It will be noticed that here again the cavate lodges and village ruins are associated, although in this case the village ruins on the mesas above are said to be more ancient than the cavate lodges. A general view of a small section of cliff containing lodges is given in plate XXXI for comparison with those on the Verde. The lodges on the Rio Grande seem to have been more elaborate than those on the Verde, perhaps owing to longer occupancy; but the same arrangement of a main front room and attached back rooms, as in the cavate lodges on the Verde, was found.

As the cavate lodges of the San Francisco mountain region have been assigned to the Havasupai Indians of the Yuman stock, and those of the Rio Grande to the Santa Clara pueblo Indians of the Tanoan stock, it may be of interest to state that there is a vague tradition extant among the modern settlers of the Verde region that the cavate lodges of that region were occupied within the last three generations. This tradition was derived from an old Walapai Indian whose grandfather was alive when the cavate lodges were occupied. It was impossible to follow this tradition to its source, and it is introduced only as a suggestion. Attention is called, however, to the tradition given in the introduction to this paper with which it may be connected.





Aside from the actual labor of excavation, there was but little work expended on the Verde cavate lodges. The interiors were never plastered, so far as the writer could determine. Figure 291 shows the plan of one of the principal sets of rooms, which occurs at the point marked D on the map, plate XXV; and plate XXXII is an interior view of the principal room, drawn from a flashlight photograph. This set of rooms was excavated in a point of the cliff and extends completely through it as shown on the general plan, plate XXV. The entrance was from the west by a short passageway opening into a cove extending back some 10 feet from the face of the cliff. The first room entered measures 16 feet in length by 10 feet in width. On the floor of this room a structure resembling the piki or paper bread oven of the Tusayan Indians, was found constructed partly of fragments of old and broken metates. At the southern end of the room there is a cubby-hole about a foot in diameter, excavated at the floor level. At the eastern end of the room there is a passageway about 2 feet long leading into a smaller roughly circular room, measuring 7 feet in its longest diameter, and this in turn is connected with another almost circular room of the same size. The floors of all three of these rooms are on the same level, but the roofs of the two smaller rooms are a foot lower than that of the entrance room. At the northern end of the entrance room there is a passageway 3 feet long and 2 feet wide leading into the principal room of the set. This passageway at its southern end has a framed doorway of the type illustrated later.



The main room is roughly circular in form, measuring 16 feet in its north and south diameter and 15 feet from east to west. The roof is about 7 feet above the floor. Figure 292 shows a section from northwest to southwest (a, b, figure 291) through the small connected room adjoining on the south, and also an east arid west section (c, d, figure 291). The floor is plastered with clay wherever it was necessary in order to bring it to a level, and the coating is consequently not of uniform thickness. It is divided into sections by low ridges of clay as shown in the plan and sections; the northern section is a few inches higher than the other. Extending through the clay finish of the floor and into the rock beneath there are four pits, indicated on the plan by round spots. The largest of these, situated opposite the northern door, was a fire hole or pit about 18 inches in diameter at the floor level, of an inverted conical shape, about 10 inches in depth, and plastered inside with clay inlaid with fragments of pottery placed as closely together as their shape would permit. The other pits are smaller; one located near the southeastern corner of the room is about 6 inches in diameter and the same in depth, while the others are mere depressions in the floor, in shape like the small paint mortars used by the Pueblos.



The room, when opened, contained a deposit of bat dung and sand about 3 feet thick in the center and averaging about 2 feet thick throughout the room. This deposit exhibited a series of well-defined strata, varying from three-fourths to an inch and a half thick, caused by the respective predominance of dung or sand. No evidence of disturbance of these strata was found although careful examination was made. This deposit was cleared out and a number of small articles were found, all resting, however, directly on the floor. The articles consisted of fragments of basketry, bundles of fibers and pieces of fabrics, pieces of arrowshafts, fragments of grinding stones, three sandals of woven yucca fiber, two of them new and nearly perfect, and a number of pieces of cotton cloth, the latter scattered over the room and in several instances gummed to the floor. Only a few fragments of pottery were found in the main room, but outside in the northern passageway were the fragments of two large pieces, one an olla, the other a bowl, both buried in 3 or 4 inches of debris under a large slab fallen from the roof.

Owing to its situation this room was one of the most desirable in the whole group. The prevailing south wind blows through it at all times, and this is doubtless the reason that it was so much filled up with sand. In the center of the room the roof has fallen at a comparatively recent date from an area about 10 by 7 feet, in slabs about an inch thick, for the fragments were within 6 inches of the top of the debris. The walls are smoke-blackened to a very slight extent compared with the large room south of it.

At the northeastern and southwestern corners there are two small pockets, opening on the floor level but sunk below it, which seem to have been designed to contain water. That in the southwest corner is the larger; it is illustrated in the section, figure 293. As shown in the section and on the plan (figure 291), a low wall composed of adobe mortar and broken rock was built across the opening on the edge of the floor, perhaps to increase its capacity. This cavity would hold 15 to 20 gallons of water, a sufficient amount to supply the needs of an ordinary Indian family for three weeks or a mouth. The pocket in the northeastern corner of the room is not quite so large as the one described, and its front is not walled.



West of the main room there is a storage room, nearly circular in shape, with a diameter of about 6 feet and with a floor raised about 2 feet above that of the main room. Its roof is but 3 feet above the floor, and across its western end is a low bench a couple of inches above the floor. In the northeastern corner there is a shallow cove, also raised slightly above the main floor and connecting by a narrow opening with the outer vestibule-like rooms on the north. These northern rooms of the lodge seem to be simply enlargements of the passageway. The northern opening is a window rather than a door as it is about 10 feet above the ground and therefore could be entered only by a ladder. The opening is cut in the back of a cove in the cliff, and is 6 feet from the northern end of the main room. At half its length it has been enlarged on both sides by the excavation of niches or coves about 4 feet deep but only 2 feet high. These coves could be used only for storage on a small scale.



In the southeastern corner of the main room there is another opening leading into a low-roofed storage cist, approximating 4 feet in diameter, and this cist was in turn connected with the middle one of the three rooms first described. This opening, at the time the room was examined, was so carefully sealed and plastered that it was scarcely perceptible.

A different arrangement of rooms is shown in plan in figure 294 and in section in figure 295. This group occurs at the point marked A on the map. The entrance to the main room was through a narrow passage, 3 feet long, leading into the chamber from the face of the bluff, which at this point is vertical. The main room is oblong, measuring 17 feet one way and 10 the other. At the southern end there is a small cist and on the western side near the entrance there is another hardly a foot in diameter. North of the main room there is a small, roughly circular room with a diameter of about 6 feet. It is connected with the main room by a passage about 2 feet long. On the floor of the main room there are two low ridges of clay, similar to those already described, which divide it into three sections of nearly equal size.



East of the main room there is another of considerable size in the form of a bay or cove. It measures 13 feet by 6 feet, and its floor is 20 inches higher than that of the main room, as shown in the section (figure 295). Attached to this bay, at its northern end, is a small cist about 3 feet in diameter, and with its floor sunk to the level of the floor of the main room. East of the cove there is another cist about 4 feet in diameter and with its floor on the level of the cove. Adjoining it on the south and leading out from the southeastern corner of the cove or bay, there is a long passage leading into an almost circular room 9 feet in diameter. The back wall of this room is 33 feet from the face of the cliff. The passage leading into it is 6 feet long, 2 feet wide at the doorways, bulging slightly in the center, and its floor is on the same level as the rooms it connects; its eastern end is defined by a ridge of clay about 6 inches high.



In the eastern side of the circular room last described there is a storage cist about 3 feet wide and 2 feet deep. No fire-pit was seen in this cluster, although if the principal apartment were carefully cleaned out it is not improbable that one might be found.

A cluster of rooms somewhat resembling the last described is shown in plan in figure 296. This cluster occurs at the point marked B on the map. The main room is set back 5 feet from the face of the bluff, which is vertical at this point, and is oblong in shape, measuring 19 by 11 feet. Its roof is 7 feet above the floor in the center of the room. Attached to its southern end by a passage only a foot in length is a small room or storage cist about 5 feet in diameter. At its northeastern corner there is another room or cist similar in shape, about 7 feet in diameter, and reached by a passage 2 feet long. This small room is also connected with a long room east of the main apartment by a passage, the southern end of which was carefully sealed up and plastered, making a kind of niche of the northern end. At the southeastern corner of the room there is a small niche about 2 feet in diameter on the level of the floor.



The eastern side of the main room is not closed, but opens directly into an oblong chamber of irregular size with the roof nearly 2 feet lower and the floor a foot higher than the main room. This step in the floor is shown by the line between the rooms on the ground plan. The second room is about 6 feet wide and 20 feet long, its southern end rounding out slightly so as to form an almost circular chamber. Near the center of its eastern side there is a passageway 2 feet long leading into a circular chamber 10 feet in diameter and with its floor on the same level as the room to which it is attached. The back wall of this room is 35 feet from the face of the cliff.

A group occurring at the point marked E on the map (plate XXV) is shown in plan in figure 297. It is located in a projecting corner of the bluff and marks the eastern limit of the cavate lodges at this end of the canyon. The group consists of five rooms, and has the distinction of extending four rooms deep into the rock. The main room is set back about 13 feet from the face of the bluff, about 7 feet of this distance being occupied by a narrow passageway and the remainder by a cove. The depth from the face of the bluff to the back of the innermost chamber is 47 feet. The main room measures 16 feet in length and 11 feet in width, and its roof is less than 7 feet high in the center. Near its center and opposite the long passageway mentioned there is a fire-pit nearly 3 feet in diameter.



At the northeastern corner of the main room there is a wide opening leading into a room measuring 8 by 7 feet, with a floor raised 2 feet above that of the principal apartment. The roof of this chamber is but 4 feet above the floor. Almost the whole eastern side of this room is occupied by a wide opening leading into another room of approximately the same size and shape. The roof of this room is only 3 feet 10 inches above the floor, and the floor is raised 6 inches above that on the west. In the northeastern corner there is a short narrow passageway leading into a small circular room, the fourth of the series, having a diameter of 4 feet. The roof of this apartment is only 3 feet above the floor.



In the southeastern corner of the main room there is a narrow passageway leading into a circular chamber about 8 feet in diameter. This chamber is connected with the second room of the series described by a passageway about 2 feet long, which opens into the southeastern corner of that room. This passageway, at its northern end, is 1 feet below the room into which it opens. One of the most noticeable features about this group of rooms is the entire absence of the little nooks and pockets in the wall which are characteristic of these lodges, and which are very numerous in all the principal groups, noticeably in the group next described.



At the point marked C on the map there is an elaborate group of chambers, consisting of two groups joined together and comprising altogether eight rooms. This is shown in plan in figure 298. The rock composing the front of the main room of the southern group has recently fallen, making a pile of debris about 4 feet high. The room originally measured about 12 by 22 feet. Its eastern side is occupied by a passageway leading into an adjoining chamber and by two shallow, roughly semicircular coves, apparently the remains of former small rooms. Along the northern wall of the room there are two little nooks at the floor level, and along the southern wall there are four, one of them (shown on the plan) being dug out like a pit. The roof of the room was about 6 feet above the floor.

The passageway near the eastern side is 4 feet long, and is 3 feet wide—an unusual width. It opens into a roughly circular room, 8 feet in diameter, but with a roof only 3 feet above the floor. Along the northeastern side of this room, there are three small pockets opening on the floor level. On the southern side of the room there is a wide opening into a small attached room, roughly oblong in shape and measuring about 6 by 4 feet. Along the southern wall of this little room there are two small pockets, and at the southwestern corner the rock has been cleared out to form a low cavity in the shape of a half dome. In the northwestern corner of the room there is another wide passage to a small room attached to the main room. This passage is now carefully sealed on its southern side with a slab of stone, plastered neatly so as to be hardly perceptible from the southern side. The room into which this passage opens on the north is attached to the northeastern corner of the main apartment by a narrow passage, 1 feet wide and a foot long. It is roughly circular in shape, about 6 feet in diameter, and is the only chamber in the southern group which has no pockets or cubby-holes. Of these pockets there are no fewer than twelve in the southern group. Near the northern corner of the main room there is a doorway leading into a cove, which in turn opens into the main room of the northern group.

The main room of the northern group is setback about 9 feet from the face of the bluff, but is entered by a passageway about 3 feet long, the remainder of the distance consisting of a cove in the cliff. The room is 22 feet long and 13 feet wide and its roof is 6 feet above the floor. In the southwestern corner there is a small pocket in the wall, and in the northwestern corner two others, all on the floor level. In the eastern side, however, there is a cubby-hole nearly 2 feet in diameter and about 2 feet above the floor. This is a rare feature. The southern end of the room opens into a kind of cove, raised 2 feet above the floor of the main room, and opening at its southern end into the main room of the southern group. In the floor of this cove there is a circular pit about 18 inches in diameter (marked in the plan, figure 298). Although resembling the fire holes already described, the position of the pit under consideration precludes use for that purpose; it was probably designed to contain water. At the northeastern corner of the principal apartment there is an oblong chamber or storage cist, measuring 6 feet by 7 feet.



Connected with the main room by a passageway 2 feet long cut in its eastern wall, there is an almost circular chamber 7 feet in diameter, and this in turn connects with another chamber beyond it by a passageway 2 feet long and less than 2 feet wide. The roofs of the two chambers last mentioned are but 4 and 4 feet, respectively, above the floor, and in none of the rooms of this group, except the main apartment, are pockets or niches found. The whole group extends back about 45 feet into the bluff.

BOWLDER-MARKED SITES.

Within the limits of the region here treated there are many hundreds of sites of structures and groups of rooms now marked only by lines of water-rounded bowlders. As a rule each site was occupied by only one or two rooms, although sometimes the settlement rose to the dignity of a village of considerable size. The rooms were nearly always oblong, similar in size and ground plan to the rooms composing the village ruins already described, but differing in two essential points, viz, character of site and character of the masonry. As a rule these remains are found on and generally near the edge of a low mesa or hill overlooking some area of tillable land, but they are by no means confined to such locations, being often found directly on the bottom land, still more frequently on the banks of dry washes at the points where they emerge from the hills, and sometimes on little islands or raised areas within the wash where every spring they must have been threatened with overflow or perhaps even overflowed. An examination of many sites leads to the conclusion that permanency was not an element of much weight in their selection.

Externally these bowlder-marked sites have every appearance of great antiquity, but all the evidence obtainable in regard to them indicates that they were connected with and inhabited at the same time as the other ruins in the region in which they are found. They are so much obliterated now, however, that a careful examination fails to determine in some cases whether the site in question was or was not occupied by a room or group of rooms, and there is a notable dearth of pottery fragments such as are so abundant in the ruins already described. Excavation in a large ruin of this type, however, conducted by some ranchmen living just above Limestone creek, yielded a considerable lot of pottery, not differing in kind from the fragments found in stone ruins so far as can be judged from description alone.

In the southern part of the region here treated bowlder-marked sites are more clearly marked and more easily distinguished than in the northern part, partly perhaps because in that section the normal ground surface is smoother than in the northern section and affords a greater contrast with the site itself. Plate XXXIII shows one of these bowlder-marked sites which occurs a little below Limestone creek, on the opposite or eastern side of the river. It is typical of many in that district. It will be noticed that the bowlders are but slightly sunk into the soil, and that the surface of the ground has been so slightly disturbed that it is practically level; there is not enough dbris on the ground to raise the walls 2 feet. The illustration shows, in the middle distance, a considerable area of bottom land which the site overlooks. In plan this site shows a number of oblong rectangular rooms, the longer axes of which are not always parallel, the plan resembling very closely the smaller stone village ruins already described. It is probable that the lack of parallelism in the longer axes of the rooms is due to the same cause as in the village ruins, i.e., to the fact that the site was not all built up at one time.

The illustration represents only a part of an extensive series of wall remains. The series commences at the northern end of a mesa forming the eastern boundary of the Rio Verde and a little below a point opposite the mouth of Limestone creek. The ruins occur along the western rim of the mesa, overlooking the river and the bottom lands on the other side, and are now marked only by bowlders and a slight rise in the ground. But few lines of wall are visible, most of the ruins consisting only of a few bowlders scattered without system. From the northern end of the mesa, where the ruins commence, traces of walls can be seen extending due southward and at an angle of about 10 with the mesa edge for a distance of one-fourth of a mile. Beyond this, for half a mile or more southward, remains of single houses and small clusters occur, and these are found in less abundance to the southern edge of the mesa, where the ruin illustrated occurs. The settlement extended some distance east of the part illustrated, and also southward on the slope of the hill. Two well-marked lines of wall occur at the foot of the hill, on the flat bottom land, but the slopes of the hill are covered with bowlders and show no well-defined lines. Scattered about on the surface of the ground are some fragments of metates of coarse black basalt and some potsherds, but the latter are not abundant.

The bowlders which now mark these sites were probably obtained in the immediate vicinity of the points where they were used. The mesa on which the ruin occurs is a river terrace, constructed partly of these bowlders; they outcrop occasionally on its surface and show clearly in its sloping sides, and the washes that carry off the water falling on its surface are full of them.



In the northern end of the settlement there are faint traces of what may have been an irrigating ditch, but the topography is such that water could not be brought on top of the mesa from the river itself. At the southern end of the settlement, northeast of the point shown in the illustration, there are traces of a structure that may have been a storage reservoir. The surface of the mesa dips slightly southward, and the reservoir-like structure is placed at a point just above the head of a large wash, where a considerable part of the water that falls upon the surface of the mesa could be caught. It is possible that, commencing at the northern end of the settlement, a ditch extended completely through it, terminating in the storage reservoir at the southern end, and that this ditch was used to collect the surface water and was not connected with the river. A method of irrigation similar to this is practiced today by some of the Pueblo Indians, notably by the Hopi or Tusayan and by the Zuni. In the bottom land immediately south of the mesa, now occupied by several American families, there is a fine example of an aboriginal ditch, described later.

In the vicinity of the large ruin just above Limestone creek, previously described, the bowlder-marked sites are especially abundant. In the immediate vicinity of that ruin there are ten or more of them, and they are abundant all along the edge of the mesa forming the upper river terrace; in fact, they are found in every valley and on every point of mesa overlooking a valley containing tillable land.

It is probable that the bowlder-marked ruins are the sites of secondary and temporary structures, erected for convenience in working fields near to or overlooked by them and distant from the home pueblo. The character of the sites occupied by them and the plan of the structures themselves supports this hypothesis. That they were connected with the permanent stone villages is evident from their comparative abundance about each of the larger ones, and that they were constructed in a less substantial manner than the home pueblo is shown by the character of the remains.

It seems quite likely that only the lower course or courses of the walls of these dwellings were of bowlders, the superstructure being perhaps sometimes of earth (not adobe) but more probably often of the type known as "jacal"—upright slabs of wood plastered with mud. This method of construction was known to the ancient pueblo peoples and is used today to a considerable extent by the Mexican population of the southwest and to a less extent in some of the pueblos. No traces of this construction were found in the bowlder-marked sites, perhaps because no excavation was carried on; but it is evident that the rooms were not built of stone, and that not more than a small percentage could have been built of rammed earth or grout, as the latter, in disintegrating leaves well-defined mounds and lines of debris. It is improbable, moreover, that the structures were of brush plastered with mud, such as the Navajo hogan, as this method of construction is not well adapted to a rectangular ground plan, and if persistently applied would soon modify such a plan to a round or partially rounded one. Temporary brush structures would not require stone foundations, but structures composed of upright posts or slabs, filled in with brush and plastered with mud, and designed to last more than one farming season, would probably be placed on stone foundations, as the soil throughout most of the region in which these remains occur is very light, and a wooden structure placed directly on it would hardly survive a winter.

In the valley of the Rio Verde the profitable use of adobe at the present time is approximately limited northward by the thirty-fourth parallel, which crosses the valley a little below the mouth of Limestone creek. North of this latitude adobe is used less and less and where used requires more and more attention to keep in order, although on the high tablelands some distance farther northward it is again a suitable construction. South of the thirty-fourth parallel, however, adobe construction is well suited to the climate and in the valleys of Salt and Gila rivers it is the standard construction. Adobe construction (the use of sun-dried molded brick) was unknown to the ancient pueblo builders, but its aboriginal counterpart, rammed earth or pis construction, such as that of the well known Casa Grande ruin on Gila river, acted in much the same way under climatic influences, and it is probable that its lack of suitability precluded its use in the greater part of the Verde valley. No walls of the type of those of the Casa Grande ruin have been found in the valley of the Verde, although abundant in the valleys of the Salt and Gila rivers, but it is possible that this method of construction was used in the southern part of the Verde region for temporary structures; in the northern part of that region its use even for that purpose was not practicable.

In this connection it should be noted that all the ruins herein described are of buildings of the northern type of aboriginal pueblo architecture and seem to be connected with the north rather than the south.



IRRIGATING DITCHES AND HORTICULTURAL WORKS.

One of the finest examples of an aboriginal irrigating ditch that has come under the writer's notice occurs about 2 miles below the mouth of Limestone creek, on the opposite or eastern side of the river. At this point there is a large area of fertile bottom land, now occupied by some half dozen ranches, known locally as the Lower Verde settlement. The ditch extends across the northern and western part of this area. Plate XXXIV shows a portion of this ditch at a point about one eighth of a mile east of the river. Here the ditch is marked by a very shallow trough in the grass-covered bottom, bounded on either side by a low ridge of earth and pebbles. Plate XXXV shows the same ditch at a point about one-eighth of a mile above the last, where it was necessary to cut through a low ridge. North of this point the ditch can not be traced, but here it is about 40 feet above the river and about 10 feet above a modern (American) ditch. It is probable that the water was taken out of the river about 2 miles above this place, but the ditch was run on the sloping side of the mesa which has been recently washed out. No traces of the ditch were found east of the point shown in plate XXXIV, but as the modern acequia, which enters the valley nearly 10 feet below the ancient one, extends up the valley nearly to its head, there is no reason to suppose that the ancient ditch did not irrigate nearly the whole area of bottom land. The ancient ditch is well marked by two clearly defined lines of pebbles and small bowlders, as shown in the illustration. Probably these pebbles entered into its construction, as the modern ditch, washed out at its head and abandoned more than a year ago, shows no trace, of a similar marking.



A little west and south of the point shown in plate XXXIV the bottom land drops off by a low bench of 3 or 4 feet to a lower level or terrace, and this edge is marked for a distance of about a quarter of a mile by the remains of a stone wall or other analogous structure. This is located on the extreme edge of the upper bench and it is marked on its higher side by a very small elevation. On the outer or lower side it is more clearly visible, as the stones of which the wall was composed are scattered over the slope marking the edge of the upper bench. At irregular intervals along the wall there are distinct rectangular areas about the size of an ordinary pueblo room, i.e., about 8 by 10 and 10 by 12 feet.

In February, 1891, there was an exceptional flood in Verde river due to prolonged hard rain. The river in some places rose nearly 20 feet, and at many points washed away its banks and changed the channel. The river rose on two occasions; during its first rise it cut away a considerable section of the bank near a point known as Spanish wash, about 3 miles below Verde, exposing an ancient ditch. During its second rise it cut away still more of the bank and part of the ancient ditch exposed a few days before. The river here makes a sharp bend and flows a little north of east. The modern American ditch, which supplied all the bottom lands of the Verde west of the river, was ruined in this vicinity by the flood that uncovered the old ditch. Figure 299 is a map of the ancient ditch drawn in the field, with contours a foot apart, and showing also a section, on a somewhat larger scale, drawn through the points A, B on the map. Plate XXXVI is a view of the ditch looking westward across the point where it has been washed away, and plate XXXVII shows the eastern portion, where the ditch disappears under the bluff.

The bank of the river at this point consists of a low sandy beach, from 10 to 50 feet wide, limited on the south by a vertical bluff 10 to 12 feet high and composed of sandy alluvial soil. This bluff is the edge of the bottom land before referred to, and on top is almost flat and covered with a growth of mesquite, some of the trees reaching a diameter of more than 3 inches. The American ditch, which is shown on the map, runs along the top of the bluff skirting its edge, and is about 14 feet above the river at its ordinary stage. The edge of the bluff is shown on the map by a heavy black line. It will be observed that the ancient ditch occurs on the lower flat, about 3 feet above the river at its ordinary stage, and its remains extend over nearly 500 feet. The line, however, is not a straight one, but has several decided bends. One of these occurs at a point just west of that shown in the section. About 80 feet east of that point the ditch makes another turn southward, and about 40 feet beyond strikes the face of the bluff almost at right angles and passes under it.

About 50 feet north of the main ditch, at the point where it passes under the bluff, there are the remains of another ditch, as shown on the map. This second ditch was about a foot higher than the main structure, or about 4 feet above the river; it runs nearly parallel with it for 30 feet and then passes into the bluff with a slight turn toward the north. It is about the same size as the main ditch, but its section is more evenly rounded. Figure 300 shows this ditch in section.



As already stated, the American ditch is about 14 feet above the river, while the ancient ditch is less than 4 feet above the water. This decided difference in level indicates a marked difference in the character of the river. The destruction of the modern ditch by the flood of 1891 is not the first mishap of that kind which has befallen the settlers. The ditch immediately preceding the current one passed nearly over the center of the ancient ditch, then covered by 10 feet or more of alluvial soil, and if a ditch were placed today on the level of the ancient structure it would certainly be destroyed every spring. The water that flowed through the modern ditch was taken from the river at a point about 3 miles farther northward, or just below Verde. The water for the ancient ditch must have been taken out less than a mile above the southern end of the section shown in the map.



At first sight it would appear that the ancient ditch antedated the deposit of alluvial soil forming the bottom land at this point, and this hypothesis is supported by several facts of importance. It is said that ten years ago the bottom land, whose edge now forms the bluff referred to, extended some 25 or 30 feet farther out, and that the river then flowed in a channel some 200 or 300 feet north of the present one. Be this as it may, the bottom land now presents a fairly continuous surface, from the banks of the river to the foothills that limit the valley on the west and south, and it is certain that this bottom land extended over the place occupied by the ancient ditch; nor is it to be supposed that the ancient ditches ended abruptly at the point where they now enter the bluff. The curves in the line of the ancient ditch might indicate that it was constructed along the slope of a hill, or on an uneven surface, as a deep excavation in fairly even ground would naturally be made in a straight line.

The face of the bluff shows an even deposit of sand, without apparent stratification, except here and there a thin layer or facing of mud occurs, such as covers the bottom of the ancient ditch and also of the modern ditch. Singularly enough, however, over the ancient ditch, about 5 feet above its bottom, there is a stratum of sand and gravel, and on top, within a few inches of the surface of the ground, a thin stratum of mud. This mud stratum extends only about 8 feet horizontally and is slightly hollowed, with its lowest part over the center of the ditch. The gravel stratum also was laid down over the ditch, is tilted slightly southward and occurs in two layers, together about a foot thick. It first appears a few feet south of the point where the main ditch enters the bluff and over the ditch both layers are distinctly marked, as shown in plate XXXVIII. Both layers are clearly marked to a distance of 4 feet north of the northern side of the main ditch; here the lower layer thins out, but the upper layer continues faintly marked almost to the edge of the small ditch. At this point the gravel stratum becomes pronounced again and continues over the small ditch, almost pure gravel in places, with a decided dip westward. At a point just beyond the northern side of the small ditch the gravel layer disappears entirely.

The occurrence of this gravel in the way described seems to indicate that the ditch was built along the slope of a low hill forming the edge of the bottom land at that time, and that subsequently detritus was deposited above it and over the adjacent bottom land forming a smooth ground surface. Against this hypothesis it must be stated that no evidence whatever was found of more than a single deposit of sandy loam, although the exposures are good; but perhaps were an examination made by a competent geologist some such evidence might be developed.



There is one fact that should not be lost sight of in the discussion, viz, the very low elevation of the ditch above the river. The Verde is, as already stated, a typical mountain stream, with an exceptionally high declivity, and consequently it is rapidly lowering its bed. If, as already conjectured, the water for the ancient ditch was taken from the river but a short distance above the point where remains of the ditch are now found—and this assumption seems well supported by the character of the adjacent topography—the slight elevation of the bed of the ditch above the river would indicate that, in the first place, the ditch was located, as already suggested, along the slope of a hill, and in the second place, that the ditch was built at a period of no great antiquity. The occurrence of the high bluff under which the ditch now passes does not conflict with this suggestion, for the deposition of the material composing it and its erosion into its present form and condition may be the result of decades rather than of centuries of work by a stream like the Verde, and certainly a hundred, or at most a hundred and fifty years would suffice to accomplish it. At the present time a few floods deposit an amount of material equal to that under discussion, and if subsequently the river changed its channel, as it does at a dozen different points every spring, a few decades only would be required to cover the surface with grass and bushes, and in short, to form a bottom land similar to that now existing over the ancient ditch.

In conclusion it should be noted, in support of the hypothesis that the ditch was built before the material composing the bluff was laid down, that immediately under the ditch there is a stratum of hard adobe-like earth, quite different from the sand above it and from the material of which the bluff is composed. This stratum is shown clearly in plate XXXVIII.

The hypothesis which accords best with the evidence now in hand is that which assumes that the ditch was taken out of the river but a short distance above the point illustrated, and that it was built on the slope of a low hill, or on a nearly flat undulating bottom land, before the material composing the present bottom or river terrace was deposited, and that the ditch, while it may be of considerable antiquity, is not necessarily more than a hundred or a hundred and fifty years old; in other words, we may reach a fairly definite determination of its minimum but not of its maximum antiquity.

On the southern side of Clear creek, about a mile above its mouth, there are extensive horticultural works covering a large area of the terrace or river bench. These have already been alluded to in the description of the village ruin overlooking them, but there are several features which are worthy a more detailed description. For a distance of 2 miles east and west along the creek, and perhaps half a mile north and south, there are traces of former works pertaining to horticulture, including irrigating ditches, "reservoirs," farming outlooks, etc.

At the eastern end of these works, about 3 miles above the mouth of Clear creek, the main ditch, after running along the slope of the hill for some distance, comes out on top of the mesa or terrace nearly opposite the Morris place. The water was taken from the creek but a short distance above, hardly more than half a mile. West of the point where the ditch comes out on the mesa top, all traces of it disappear, but they are found again at various points on the terrace. Plate XXXIX shows a portion of the terrace below and opposite the rectangular ruin previously described. In the distant foreground the light line indicates a part of the ancient ditch. Plate XL shows the same ditch at a point half a mile below the last, where it rounds a knoll. In the distance is the flat-topped hill or mesa on which the rectangular ruin previously described is located. About a hundred yards southeast of this point further traces of the ditch may be seen, and connected with it at that point are a number of rectangular areas, which were cultivated patches when the ditch was in use.

The whole surface of the terrace within the limits described is covered by small water-worn bowlders scattered so thickly over it that travel is seriously impeded. In many parts of it these bowlders are arranged so as to inclose small rectangular areas, and these areas are connected with the old ditch just described. Plate XXXIX shows something of this surface character; and in the right hand portion of it may be seen some of the rows of bowlders forming the rectangular areas. The rows which occur at right angles to the ditch are much more clearly marked than those parallel to it, and the longer axes of the rectangular areas are usually also at right angles to the ditch line. On the ground these traces of inclosures can hardly be made out, but from an elevated point, such as the mesa on which, the rectangular ruin overlooking these works is located, they show very clearly and have the appearance of windrows. Traces of these horticultural works would be more numerous, and doubtless more distinct, were it not that a considerable part of the area formerly under cultivation has been picked over by the modern settlers in this region, and immense quantities of stone have been removed and used in the construction of fences. This has not been done, however, in such a manner as to leave the ground entirely bare, yet bare areas occur here and there over the surface, where doubtless once existed a part of the general scheme of horticultural works.

One such bare area occurs close to the edge of the terrace about a mile and a half above the mouth of the creek. In its center is a structure called for convenience a reservoir, although it is by no means certain that it was used as such. It occurs about 100 yards from the creek, opposite the Wingfield place, and consists of a depression surrounded by an elevated rim. It is oval, measuring 108 feet north and south and 72 feet east and west from rim to rim. The crown of the rim is 5 feet 8 inches above the bottom of the depression and about 3 feet above the ground outside. The rim is fairly continuous, except at points on the northern and southern sides, where there are slight depressions, and these depressions are further marked by extra large bowlders. At its lowest points, however, the rim is over 2 feet above the ground, which slopes away from it for some distance in every direction. Plate XLI shows the eastern side of the depression; the large tree in the middle distance is on the bank of Clear creek and below the terrace. Plate XLII shows the northern gateway or dip in the rim, looking southward across the depression. The large bowlders previously referred to can be clearly seen. A depression similar to this occurs on the opposite side of the valley, about half a mile from the river. In this case it is not marked by bowlders or stones of any description, but is smooth and rounded, corresponding to the surface of the ground in its vicinity. In the latter as in the former case, the depression occurs on a low knoll or swell in the bottom land, and the surface of the ground slopes gently away from it for some distance in every direction.



The purpose of these depressions is not at all clear, and although popularly known as reservoirs it is hardly possible that they were used as such. The capacity of the Clear creek depression is about 160,000 gallons, or when two-thirds full, which would be the limit of its working capacity, about 100,000 gallons. The minimum rate of evaporation in this region in the winter months is over 3 inches per month, rising in summer to 10 inches or more, so that in winter the loss of water stored in this depression would be about 10,000 gallons a month, while in summer it might be as high as 35,000 or even 40,000 gallons a month. It follows, therefore, that even if the reservoir were filled to its full working capacity in winter and early spring it would be impossible to hold the water for more than two months and retain enough at the end of that time to make storing worth while. It has been already stated, however, that these depressions are situated on slight knolls and that the land falls away from them in every direction. As no surface drainage could be led into them, and as there is no trace on the ground of a raised ditch discharging into them, they must have been filled, if used as reservoirs, from the rain which fell within the line that circumscribes them. The mean annual rainfall (for over seventeen years) at Verde, a few miles farther northward in the same valley, is 11.44 inches, with a maximum annual fall of 27.27 inches and a minimum of 4.80 inches. The mean annual fall (for over twenty-one years) at Fort McDowell, near the mouth of the Rio Verde, is 10-54 inches, with a maximum of 20.0 inches and a minimum of 4.94 inches.[8]

[Footnote 8: Report on Rainfall (Pacific coast and western states and territories), Signal Office U.S. War Dept., Senate Ex. Doc. 91, 50th Cong., 1st Sess., Washington, 1889; pp. 70-73 (Errata, p. 4).]

If these depressions were used as reservoirs it is a fair presumption that the bottoms were plastered with clay, so that there would be no seepage and the only loss would be by evaporation. Yet this loss, in a dry and windy climate such as that of the region here treated, would be sufficient to render impracticable a storage reservoir of a cross section and a site like the one under discussion. Most of the rainfall is in the winter months, from December to March, and it would require a fall of over 12 inches during those months to render the reservoir of any use in June; it would certainly be of no use in July and August, at the time when water is most needed, save in exceptional years with rainfall much in excess of the mean.

On the other hand, there is the hypothesis that these depressions represent house structures; but if so these structures are anomalous in this region. The contour of the ground does not support the idea of a cluster of rooms about a central court, nor does the dbris bear it out. Mr. F. H. Cushing has found depressions in the valleys of Salt and Gila rivers somewhat resembling these in form and measurement, and situated always on the outskirts of the sites of villages. Excavations were made, and as the result of these he came to the conclusion that the depressions were the remains of large council chambers, as the floors were hard, plastered with mud, and dish-shaped, with a fire-hole in the center of each; and no pottery or implements or remains of any kind were found except a number of "sitting stones." Mr. Cushing found traces of upright logs which formed the outer wall of the structure; he inferred from the absence of drainage channels that the structure was roofed, and as the ordinary method of roofing is impracticable on the scale of these structures, he supposed that a method similar to that used by the Pima Indians in roofing their granaries was employed, the roof being of a flattened dome shape and composed of grass or reeds, formed in a continuous coil and covered with earth. If the depressions under discussion, however, are the remains of structures such as these described, they form a curious anomaly in this region, for, as has been already stated, the affinities of the remains of this region are with the northern architectural types, and not at all with those of the southern.

There is a third hypothesis which, though not supported by direct evidence, seems plausible. It is that the depression of Clear creek, and perhaps also the one on the opposite side of the Verde, were thrashing floors. This hypothesis accords well with the situation of these depressions upon the tillable bottom lands, and with their relation to the other remains in their vicinity; and their depth below the surface of the ground would be accounted for, under the assumption here made of their use, by the high and almost continuous winds of the summer in this region. Perhaps the slight depressions at the northern and southern side of the oval were the gateways through which the animals which trampled the straw or the men who worked the flails passed in and out. Whether used in this way or not, these depressions would be, under the assumption that the bottom was plastered with mud, not only practicable, but even desirable thrashing floors, as the grain would be subjected during thrashing to a partial winnowing. This suggestion would also account for the comparatively clean ground surface about the depressions and for their location on slightly elevated knolls.

Scattered over the whole area formerly under cultivation along Clear creek are the remains of small, single rooms, well marked on the ground, but without any standing wall remaining. These remains are scattered indiscriminately over the terrace without system or arrangement; they are sometimes on the flat, sometimes on slight knolls. They number altogether perhaps forty or fifty. Plate XLIII shows an example which occurs on a low knoll, shown also in plate XL; it is typical of these remains. It will be noticed that the masonry was composed of river bowlders not dressed or prepared in any way, and that the dbris on the ground would raise the walls scarcely to the height of a single low story.



The location of these remains, their relation to other remains in the vicinity, and their character all support the conclusion that they were small temporary shelters or farming outlooks, occupied only during the season when the fields about them were cultivated and during the gathering of the harvest, as is the case with analogous structures used in the farming operations among the pueblos of to-day. Their number and distribution do not necessarily signify that all the terrace was under cultivation at one time, although there is a fair presumption that the larger part of it was, and the occurrence of the ditch at both the upper and the lower ends of the area strengthens this conclusion.

As it is impossible that an area so large as this should be cultivated by the inhabitants of one village, it is probable that a number of villages combined in the use of this terrace for their horticultural operations; and, reasoning from what we know to have been the case in other regions, it is further probable that this combination resulted in endless contention, and strife, and perhaps finally to the abandonment of these fields if not of this region. The rectangular ruin already illustrated is situated on a hill south of the terrace and overlooks it from that direction; on the opposite side of Clear creek, on the hill bounding the valley on the north, there are the remains of a large stone village which commanded an outlook over the terraces in question; and a little farther up the creek, on the same side and similarly situated, there was another village which also overlooked them. There were doubtless other villages and small settlements whose remains are not now clearly distinguishable, and it is quite probable that some of the inhabitants of the large villages in the vicinity, like those near Verde, hardly 3 miles northward, had a few farming houses and some land under cultivation on this terrace.

Thus it will be seen that there was no lack of cultivators for all the tillable land on the terrace, and there is no reason to suppose that the period when the land was under cultivation, and the period when the villages overlooking it were occupied, were not identical, and that the single-house remains scattered over the terrace were not built and occupied at the same period. The relation of the stone villages to the area formerly cultivated, the relation of the single-room remains to the area immediately about them, the character of the remains, and the known methods of horticulture followed by the Pueblo Indians, all support the conclusion that these remains were not only contemporaneous but also related to one another.



STRUCTURAL CHARACTERISTICS.

MASONRY AND OTHER DETAILS.

The masonry of the stone villages throughout all the region here treated is of the same type, although there are some variations. It does not compare with the fine work found on the San Juan and its tributaries, although belonging to that type—the walls being composed of two faces with rubble filling, and the interstices of the large stones being filled or chinked with spalls. This chinking is more pronounced and better done in the northern part of the region than in the south.

The rock employed depended in all cases on the immediate environments of the site of the village, the walls being composed in some cases of slabs of limestone, in other cases of river bowlders only, and in still others of both in combination. The walls of the large ruin near Limestone creek were composed of rude slabs of limestone with an intermixture of bowlders. The bowlders usually occur only in the lower part of the wall, near the ground, and in several cases, where nothing exists of the wall above the surface of the ground, the remains consist entirely of bowlders. A good example of this peculiarity of construction is shown in plate XLIV, and plate XLV shows the character of stone employed and also a section of standing wall on the western side of the village. A section of standing wall near the center of the ruin is illustrated in plate XIII. It will be noticed that some of the walls shown in this illustration are chinked, but to a very slight extent. The wall represented in plate XLV has slabs of limestone set on edge. This feature is found also in other ruins in this region, notably in those opposite Verde, though it seems to be more used in the south than in the north. An example occurring in the ruin opposite Verde is shown in plate XLVI. In this case chinking is more pronounced; the walls are from 2 to 2 feet thick, built in the ordinary way with two faces and an interior filling, but the stones are large and the filling is almost wholly adobe mortar. The two faces are tied together by extra long stones which occasionally project into the back of one or the other face.

The western cluster of the ruin last mentioned, shown on the ground plan (plate XVII), has almost all its walls still standing, and the masonry, while of the same general character as that of the main cluster, is better executed. The stones composing the walls are smaller than those in the main cluster and more uniform in size, and the interstices are carefully chinked. The chinking is distinctive in that spalls were not used, but more or less flattened river pebbles. The different color and texture of these pebbles make them stand out from the wall distinctly, giving quite an ornamental effect.



That portion of the standing wall of the ruin opposite Verde, which occurs in the saddle northeastward from the main cluster, shown on the plan in plate XVII, represents the best masonry found in this region. As elsewhere stated, this was probably the last part of the village to be built. These walls are shown in plate XLVII. It will be noticed that the stones are of very irregular shape, rendering a considerable amount of chinking necessary to produce even a fair result, and that the stones are exceptionally large. The masonry of this village is characterized by the use of stones larger than common, many of them being larger than one man can carry and some of them even larger than two men can handle.

All the larger and more important ruins of this region are constructed of limestone slabs, sometimes with bowlders. The smaller ruins, on the other hand, were built usually of river bowlders, sometimes with an intermixture of slabs of limestone and sandstone but with a decided preponderance of river bowlders. This would seem to suggest that this region was gradually populated, and that the larger structures were the last ones built. This suggestion has been already made in the discussion of the ground plans, and it is, moreover, in accord with the history of the pueblo-builders farther northward, notably that of the Hopi.

Plate XXI illustrates a type of bowlder masonry which occurs on Clear creek; plate XLVIII shows the masonry of the ruin at the mouth of the East Verde, and plate XVI shows that of a ruin at the month of Fossil creek. In all these examples the stone composing the walls was derived either from the bed of an adjacent stream or from the ground on which they were built, and was used without any preparation whatever; yet in the better examples of this type of masonry a fairly good result was obtained by a careful selection of the stones. A still ruder type of masonry sometimes found in connection with village ruins is shown in figure 290. This, however, was used only as in the example illustrated, for retaining walls to trails or terraces, or analogous structures.

In a general way it may be stated that the masonry of the village ruins of this region is much inferior to that of the San Juan region, and in its rough and unfinished surfaces, in the use of an inferior material close at hand rather than a better material a short distance away, and in the ignorance on the part of the builders of many constructive devices and expedients employed in the best examples of pueblo masonry, the work of this region may be ranked with that of the Tusayan—in other words, at the lower end of the scale.



There is but little masonry about the cavate lodges, and that is rude in character. As elsewhere stated, walled fronts are exceptional in this region, and where they occur the work was done very roughly. Figure 301 shows an example that occurs in the group of cavate lodges already described. It will be noticed that little selection has been exercised in the stones employed, and that an excess of mortar has been used to fill in the large interstices. Figure 290 (p. 221), which shows a storage cist attached to the group of cavate lodges, marked D on the map (plate XXV), exhibits the same excessive use of adobe or mud plastering. At several other points in the area shown on this map there are short walls, sometimes inside the lodges, sometimes outside. In all cases, however, they are rudely constructed and heavily plastered with mud; in short, the masonry of the cavate lodges exhibits an ignorance fully equal to that of the stone villages, while the execution is, if anything, ruder. It is singular that, notwithstanding the excessive use of mud mortar and mud plastering in the few walls that are found there, such plastering was almost never used on the walls in the interiors of the lodges, perhaps because no finer finish than the rough surface of the rock was considered desirable.



The cavate lodges seem to have been excavated without the aid of other tools than a rough maul or a piece of stone held in the hand, and such a tool is well adapted to the work, since a blow on the surface of the rock is sufficient to bring off large slabs. Notwithstanding the rude tools and methods, however, some of the work is quite neat, especially in the passageways (which are often 3 or 4 feet long and quite narrow) and in the smaller chambers. In the excavation of these chambers benches were left at convenient places along the wall and niches and cubby-holes were cut, so that in the best examples of cavate lodges the occupants, it would seem, were more comfortable, so far as regards their habitation, than the ordinary Pueblo Indian of today, and better supplied with the conveniences of that method of living. It should be stated in this connection, however, that although the group of cavate lodges gives an example of an extensive work well carried out, the successful carrying out of that work does not imply either a large population or a high degree of skill; the only thing necessary was time, and the amount of time necessary for the work is not nearly so great, in proportion to the population housed, as was required for the better types of pueblo work in the San Juan country (the village ruins of the Chaco canyon for example), and probably no more than would be required for the construction of rooms of equal size and of the rather poor grade of work found in this region.

Although no examples of interior wall-plastering were found in the group of cavate lodges described, such work has been found in neighboring lodges; and in this group plastered floors are quite common. The object of plastering the floors was to secure a fairly even surface such as the soft rock did not provide, and this was secured not by the application of layers of clay but by the use of clay here and there wherever needed to bring the surface up to a general level, and the whole surface was subsequently finished. This final finishing was sometimes omitted, and many floors are composed partly of the natural rock and partly of clay, the latter frequently in spots and areas of small size.

The floors were often divided into a number of sections by low ridges of clay, sometimes 8 inches broad. These ridges are shown on the ground plans (figures 294 to 298, and in plate XXV). Their purpose is not clear, although it can readily be seen that in such domestic operations as sorting grain they would be useful.

DOOR AND WINDOW OPENINGS.

The masonry of this region was so roughly and carelessly executed that little evidence remains in the stone villages of such details of construction as door and window openings. Destruction of the walls seems to have commenced at these openings, and while there are numerous standing walls, some with a height of over 10 feet, no perfect example of a door or window opening was found. It is probable that the methods employed were similar or analogous to those used today by the Hopi, and that the wooden lintel and stone jamb was the standard type.







In the cavate lodges window openings are not found; there is but one opening, the doorway, and this is of a pronounced and peculiar type. As a rule these doorways are wider at the top than at the bottom and there are no corners, the opening roughly approximating the shape of a pear with the smaller end downward. The upper part of the opening consists always of the naked rock, but the lower part is generally framed with slabs of sandstone. Plate XLIX shows an example that occurs in the upper tier of lodges at its eastern end. The floor of this lodge is about 2 feet above the bench from which it was entered, and this specimen fails to show a feature which is very common in this group—a line of water-worn bowlders extending from the exterior to the interior of the lodges through the doorway and arranged like stepping stones. This feature is shown in figure 302, which represents the doorway of group E, shown on the general map (plate XXV) and on the detailed plan, figure 297. Figure 303 shows a type in which the framing is extended up on one side nearly to the top, while on the other side it extends only to half the height of the opening, which above the framing is hollowed out to increase its width. This example occurs near that shown in plate XLIX, and the floor of the chamber is raised about 2 feet above the bench from which it is entered. The illustration gives a view from the interior, looking out, and the large opening on the right was caused by the comparatively recent breaking out of the wall. Figure 303 shows the doorway to the group of chambers marked E on the general map, an interior view of which is shown in figure 302. In this example the obvious object of the framing was to reduce the size of the opening, and to accomplish this the slabs were set out 10 or 12 inches from the rock forming the sides of the opening, and the intervening space was filled in with rubble. Plate XXXII, which shows the interior of the main room in group D, shows also the large doorway on the north.





It will be noticed that these doorways all conform to one general plan and that this plan required an opening considerably larger in its upper third than in the lower two-thirds of its height. This requirement seems to be the counterpart or analogue of the notched doorway, which is the standard type in the cliff ruins of Canyon de Chelly and other regions, and still very common in Tusayan (Moki). Figure 304 shows a notched doorway in Canyon de Chelly and figure 305 gives an example of the same type of opening in Tusayan. The object of this peculiar shape in the regions mentioned has been well established,[9] and there is no reason to suppose that similar conditions and a similar object would not produce a similar result here. This type of opening had its origin in the time when the pueblo builders had no means, other than blankets, of temporarily closing door openings and when all the supplies of the village were brought in on the backs of the inhabitants. In order to secure protection against cold and storm the opening was made of the smallest possible size consistent with its use, and the upper part of the opening was made larger in order to permit the introduction of back loads of faggots and other necessaries. This purpose would be almost as well served by the openings of the cavate lodges as by the notched doorway, and at the same time the smallest possible opening was exposed to the weather. The two types of openings seem simply to be two different methods of accomplishing the same purpose—one in solid rock, the other in masonry. That it was considered desirable to reduce the openings as much as possible is evident from the employment of framing slabs in the lower portions, reducing the width of that part generally to less than a foot, while the upper portions are usually 3 feet and more in width, and the absence of framing slabs in the upper part of the openings was probably due to their use as suggested; no slabs could be attached with sufficient firmness to resist the drag of a back load of wood, for example, forced between them. The strict confinement of door openings to one type suggests a short, rather than a long, occupancy of the site under discussion, a suggestion which is borne out by other details; and this unity of design renders it difficult to form a conclusion as to the relative age of the two types of openings under discussion. So far as the evidence goes, however, it supports the conclusion that the doorways of the cavate lodges were derived from a type previously developed, and that the idea has been modified and to some extent adapted to a different environment; for if the idea had been developed in the cavate lodges there would be a much greater number of variations than we find in fact. There can be no doubt, however, that the cavate lodge doorways represent an earlier type in development, if not in time, than the notched doorways of Tusayan.

[Footnote 9: A Study of Pueblo Architecture, by Victor Mindeleff: 8th. Ann. Rep. Bur. Eth. for 1886-1887; Washington, 1891, pp. 1-228.]



CHIMNEYS AND FIREPLACES.

Nowhere in the village ruins or in the cavate lodges of the lower Verde were any traces of chimneys or other artificial smoke exits found. The village ruins are too much broken down to permit definite statement of the means employed for smoke exits, but had the inhabitants employed such exits as are in use in the pueblos today some evidence of them would remain. Probably there was no other exit than the door, and perhaps trapdoors or small openings in the roofs, such as were formerly employed in the inhabited pueblos, according to their traditions. In the cavate lodges no exit other than the door was possible, and many of them are found with their walls much blackened by smoke.

The fireplaces or fire holes of the cavate lodges have already been alluded to, and one of the best examples found is illustrated in plate XXXII, and the location of a number of others is shown on the general plan. These fireplaces are located not in the center of the chamber, but near the principal doorway, and doubtless the object of this location was to facilitate the escape of the smoke. Fire holes were never located in interior rooms. The fireplace illustrated in plate XXXII has been already described (p. 227); it was excavated in the solid rock of the floor and was lined with fragments of pottery laid in mud mortar as closely as their shape would permit. A part of this pottery lining can be seen in the illustration. When the room was cleared out the fire hole was found to be about half full of fine ashes.