w WW \\x"^\ V ■\ * ^ \ W \^ EX LIBRIS ET IN MEMORIAM George Burnap A . MA1910 1885* »1938 V Cornell University Library E 78.C6F5 Prehistoric viilages, castles, and tower 3 1924 020 500 033 The original of this book is in the Cornell University Library. There are no known copyright restrictions in the United States on the use of the text. http://www.archive.org/details/cu31924020500033 SMITHSONIAN INSTITUTION BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PREHISTORIC VILLAGES, CASTLES, AND TOWERS OF SOUTHWESTERN COLORADO BY J. WALTER FEWKES '^i **fip»^?'?pp WASHINGTON GOVERNMENT PRINTING OFFICE 1919 ti Ar^S'&/ y LETTER OF TRANSMITTAL Smithsonian Institution, Bureau of Ameeican Ethnology, Washirigton, J). C, January 23, 1919. Sik: I have the honor to transmit the accompanying manuscript, entitled "Prehistoric Villages, Castles, and Towers of Southwestern Colorado," by J. Walter Fewkes, and to recommend its publication, subject, to your approval, as Btilletin 70 of this Bureau. Very respectfully, J. Waltek Fewkes, GUef. Dr. Charles D. Walcott, Secretary of the Smithsonian Institution. CONTENTS Page Introduction 9 Historical 10 Classiflcation 14 Villages 16 Rectangular ruins of the pure typo 16 Surouaro 16 Goodman Point Ruin 17 JohnBon Ruin 18 Bug Mesa Ruin 19 Mitchell Spring Ruin 19 Mud Spring (Burldiardt) Ruin 20 Ruin with semicircular core C2 Wolley Ranch Ruin 22 Blanchard Ruin 23 Ruins at Aztec Spring •. 23 Great open-air ruins south and southwest of Dove Creek post office . 28 Squaw Point Ruin. 28 Acmen Ruin 29 Oak Spring House 29 Ruin in Ruin Canyon 30 Cannonball Ruin 30 Circular ruins with peripheral compartments 31 Wood Canyon Ruins 32 Butte Ruin 32 Emerson Ruin 33 Escalante Ruin 36 Cliff-dwellings 37 Cliff-dwellings in Sand Canyon 38 Double cliff-house 38 Scaffold in Sand Canyon 38 Unit-type houses in caves 39 Cliff-houses in Lost Canyon 40 Great houses and towers 40 Masonry 40 Structure of towers 42 Hovenweep district 44 Ruin Canyon 44 Square Tower Canyon 45 Classification of ruins in Square Tower Canyon 46 Hovenweep House (Ruin 1) 46 Hovenweep Castle 47 Western section of Hovenweep Castle 47 Eastern section of Hovenweep Castle 48 Ruin 3 48 Ruin 4 49 Ruin 5 49 Ruin 6 49 5 b CONTENTS Claasification— Continued. Great houses and towers — Continued. Hovenweep district — Continued. Page Eroded bowlder house (Riiin 7) 49 Twin Towers (Ruin 8) 50 Ruin 9 50 Unit-type House (Ruin 10) 50 Stronghold House (Ruin 11) 51 Ruins in Holly Canyon 52 Ruin A, Great House, Hackberry Castle 52 Towers[Oand D] 52 Holly House , 53 Ruins in Hackberry Canyon 53 Horseshoe House 53 Towers in the Main Yellow Jacket Canyon 54 Davis Tower 55 Lion (Littrell) Tower 55 McLean Basin 55 Tower in Sand Canyon 57 Towers in Road (Wickyup) Canyon 57 Towers of the Mancos 58 Holmes Tower 58 Towers on the Mancos River below the bridge. 59 Tower A 59 Tower B 59 Megalithic and slab house ruins at McElmo Bluff 60 Grass Mesa Cemetery 64 Reservoirs 64 Pictographs 65 Minor antiquities 66 Historic remains 68 Conclusions 68 Index 77 ILLUSTRATIONS PLATES 1. a, Butte Ruin. 6, Aztec Spring Ruin, c, Surouaro, Yellow Jacket Spring Ruin. 2. a, Blanchard Ruin. 5, Blanchard Ruin, Mound 2. c, Surouaro, Yellow Jacket Spring Ruin. 3. a, Acmen Ruin. 6, Mud Spring Ruin. 4. a, Building on rock pinnacle, near Stone Arch, Sand Canyon. 6, Stone Arch, Sand Canyon. 5. a, Tower in Sand Canyon, b, Unit-type House in Sand Canyon. 6. a, Stone Arch House, Sand Canyon. 6, Cliff-house, showing broken comer. 7. a, Scaffold in Sand Canyon. 6, Storage cist in Mancos Valley, c, Pictographs near Unit-type House in cave. 8. Double cliff-dwelling, Sand Canyon. 9. a. Cliff -dwelling under Horseshoe Ruin. 6, Cliff-dwelling, Ruin Canyon. 10. a, Kiva of cliff ruin, Lost Canyon. 6, Cliff ruin, Lost Canyon. 11. a, Square Tower in Square Tower Canyon. 6, Tower in McLean Basin, c, Ruin in Hill Canyon, Utah. 12. Head of South Fork, Square Tower Canyon. 13. North Fork of Square Tower Canyon, looking west. 14. a, Hovenweep House and Hovenweep Castle, from the south. 6, Hovenweep Castle, from the west, c, Hovenweep Castle, from the south. 15. a, West end of Twin Tower, showing small cliff-house. 6, Twin Towers, Square Tower Canyon, from the south, c. Tower 4, junction of North and South Forks, Square Tower Canyon. 16. a, Hovenweep Castle, with Sleeping Ute Mountain, South Fork, Square Tower Canyon. 6, Entrance to South Fork, Squaie Tower Canyon. 17. Stronghold House, Square Tower Canyon. 18. a, Head of Holly Canyon. 6, South side of Hovenweep Castle, Square Tower Canyon. 19. a, Holly Canyon group, from the east. 6, Great House at head of HoUy Can- yon, from the north, c. Unit-type Ruin, from the east. 20. a, Great House at head of Holly Canyon, from the south. 5, Ruin B at head of Holly Canyon, from the west, c, Great House at head of Holly Canyon. 21. a, Great House, Holly Canyon. 6, Stronghold House and Twin Towers, Square Tower Canyon. 22. a, Hovenweep Castle. &, Southern part of Cannonball Ruin, McElmo Canyon. 23. a, Square tower with rounded comers, Holly Canyon. 6, Holly Tower in Holly Canyon, c. Horseshoe House. 24. a, Horseshoe Ruin. 6, Bowlder Castle, Road (Wickyup) Canyon. 25. a, Closed doorway in Bowlder Castle, Road (Wickyup) Canyon, b, Broken- down round tower, Square Tower Canyon. 26. a, North side of tower. Square Tower Canyon. 6, D-shaped tower near Davis ranch. Yellow Jacket Canyon, c. Model of towers in McLean Basin. 27. Round tower and D-shaped tower in McLean Basin. 28. a, D-shaped tower in McLean Basin, showing cross section of wall. 6, Round tower in McLean Basin, showing standing stone slab. 29. a, Holmes Tower, Mancos Canyon. 6, Lion Tower, Yellow Jacket Canyon. 8 ILLTJSTEATIONS 30. a, Tower above cavate storehouses, Mancos Canyon, below bridge, h, Tower on mesa between eroded cliffs and bridge over Mancos Canyon, on Cortez Ship- rock Road. 31. a, Tower above cavate storehouses, Mancos Canyon, below bridge. 5, Eroded shale formation in which are small walled cavate storehouses. 32. a, Reservoir near Picket corral, showing retaining wall. 6, Kiva, Unit-type House, Square Tower Canyon. 33. Kctographs, Yellow Jacket Canyon. TEXT FIGURES Page 1. Ground plan of Aztec Spring Ruin 26 2. Ground plan of Wood Canyon Ruin 32 3. Metes and bounds of Emerson Ruin 34 4. Schematic ground plan of Emerson Ruin 35 5. Ground plan of Unit-type House in cave 39 6. Square Tower Canyon 45 7. Ground plan of Hovenweep House 46 8. Ground plan of Hovenweep Castle 47 9. Ground plan of Twin Towers 50 10. Ground plan of Unit-type House 51 11. Holly Canyon Ruins 52 12. Horseshoe (Hackberry) Canyon 53 13. Ground plan of Horseshoe House 54 14. Ground plan of Davis Ruin 55 15. Ground plan of Lion House 55 16. Ground plan of ruin with towers in McLean Basin 56 17. Doorway in Round Tower, McLean Basin 57 18. MegaUthic stone inclosure, McElmo Bluff 61 PREHISTORIC VILLAGES, CASTLES, AND TOWERS OF SOUTHWESTERN COLORADO By J. Walter Fewkes INTEODUCTION The science of archeology has contributed to our knowledge some of the most fascinating chapters in culture history, for it has brought to Hght, from the night of the past, periods of human development hitherto unrecorded. As the paleontologist through his method has revealed faimas whose hke were formerly unknown to the natujrahst, the archeologist by the use of the same method of research has resurrected extinct phases of culture that have attained a high development and declined before recorded history began. No achievements in American anthropology are more striking than those that, from a study of human buUdings and artifacts antedating the historic period, reveal the existence of an advanced prehistoric culture of man in America. The evidences of a phase of culture that had developed and was on the decline before the interior of North America was explored by Europeans are nowhere better shown than in southwestern Colorado, New Mexico, Arizona, and Utah, the domain of the Chfif-dwellers, or the cradle of the Pueblos. There flourished on what is now called the Mesa Verde National Park, in prehistoric times, a characteristic culture unlike that of any region in the United States. This culture reached its apogee and declined before the historic epoch, but did not perish before it had left an influence extending over a wide territory, which persisted into modern times. Through the researches of archeologists the nature of this culture is now emerging into full view; but much material yet remains awaiting investigation before it can be adequately mxderstood. The purpose of this article is to call attention to new observations bearing upon its interpretation made by the author, under the auspices of the Bureau of American Eth- nology, on brief trips to Colorado and Utah in 1917 and 1918. The peculiar chfl-dwellings and open-air villages of the Mesa Verde are here shown to be typical of those found over a region many miles in extent. They indicate a distinct culture area, which is easily distinguished from others where similar buildings do not exist, but 9 10 BUREAU OF AMEKICAK ETHNOLOGY [boll, to not as readily separated from that of adjacent regions where the buUdings are superficially similar but structurally different. In order to distinguish it from its neighbors and determine its horizon, we must become familiar with certain architectural characteristics. As our knowledge of the character of buildings in this area is incom- plete, the intention of the author is to define the several different types of buildings that characterize it. When, in 1915, there was brought to light on the Mesa Verde National Park, Colorado, the mysterious structure, Sun Temple, the author recalled well-known descriptions of towers and other related buildings that have been recorded from other localities in south- western Colorado and Utah. The pubhshed descriptions of these structures did not seem to him adequate for comparisons, and he planned an examination of these great houses and towers, hoping to gather new data that would shed some hght on his interpretation of Sun Temple. During the field work in 1917, thanks to an' allotment from the Bureau of American Ethnology for that purpose, he under- took a reconnoissance in the McElmo district, where similar buildings are found and where he believed cultural relatives of the former inhabitants of Mesa Verde once lived. In 1918 he extended his field work stiU farther. He investigated ruins as far as the western tribu- taries of the YeUow Jacket Canyon, penetrating a short distance beyond the Colorado border into Utah. The object of the following pages is to make known the more important results of this visit, and interpret the evidence they present as a contribution to our knowledge of the extension in prehistoric times of the Mesa Verde culture area. HISTOKICAL Attention was first publicly called, about 40 years ago (1875-1877), by Messrs. Jackson,^ Holmes, Morgan, and others, to some of the ruins here considered. It is difficult to identify all of the ruins mentioned or described by these pioneers. Their "Plovenweep Castle" is supposed to lie in about the center of the district here considered, possibly on Square Tower (Ruin) Canyon, although the large castellated buUding ^ in HoUy Canyon would also fulfill con- ditions equally well. Their "Pueblo" may have been situated on the McElmo near the mouth of Yellow Jacket Canyon. Early writers rather vaguely refer to a cluster of castles and towers as situated some distance from the "Burial Place," which is readily identified on the promontory at the mouth of the McElmo, as prob- ably those in Square Tower (Ruin) Canyon, but the cluster may be 1 Ancient Ruins in Souihwostcrn Colorado. Hopt. U. S. Gcol. Surv. Terr. (Haydcn Survey) for 1874, Wasliington, 1870. 2 The situation of a spring near Ilovenwcop Castle indicates that the Great House may be the Hoven- wcep Castle of early writers. FEWKES] PEEHISTOEIC VILLAGBSy CASTLES, AITD TOWERS 11 either at Square Tower or HoUy Canyon, both of which are about the same distance from this site. As "Pueblo" is not indicated on the map accompanying the Hayden report, the sites of rock shelters "some 7 nules from 'Pueblo' and 3 miles from the McElmo" remain doubtful. The author retains the name "Hovenweep Castle" for the ruin in Square Tower Canyon. In his accoxmt of ruins in the region visited, Prof. W. H. Holmes ' considers several other ruins, as "the triple-waUed tower" (here called Mud Spring village, p. 20), ruins at Aztec Spring (p. 23), cliff-dwellings and towers of the San Juan and Mancos, the "slab cysts" or burial places on the Dolores, and the promontory at the junction of the Hovenweep and McElmo (p. 60). The best- preserved towers and castellated buildings which his article considers occur on the San Juan and Mancos Canyons, districts on the periphery of the region covered by this account. These pioneer reports of Jackson and Holmes not only called attention to a new archeological field, but also introduced to the archeologist several new types of prehistoric American architecture of which nothing was previously known. They have been repeatedly quoted and are still constantly referred to by writers on southwestern archeology. Although Jackson made many photographs of the castles and towers of the Hovenweep, none of these were published in his reports, possibly because halftone methods of reproduction were then un- known. The illustrations that appear in the text of early reports are mainly reproductions of sketches. These reports, in which the discovery of the tower type of architecture and its adjacent cliff- dwellings were announced, should thus rightly rank as the first important steps in the scientific investigations of the stone-house builders of this district of our Southwest; although the allied "Casas Grandes" or great houses of the Chaco had been described a few years before by Gregg, Stimpson, and others. We have, in addition to these pioneer reports, several magazine articles of about the same date, the material for which was largely drawn from them. One of the most important newspaper articles of that date was written by Mr. Ernest IngersoU, pubMshed in the New York Tribune, and another, of anonymous authorship, is to be found in the Century Magazine for the year 1877. New forms of towers and castellated buildings were added in these accounts to those of the earlier authors. One of the most important contributions to the antiquities of the region about Mesa Verde was made by the veteran ethnologist, Mor- gan, who published notes contributed by Mr. Mitchell on a cluster of 1 Report on the ancient ruins of Southwestern Colorado. Tenth Ann. Rept. U. S. Geol. Surv. Terr. (Hayden Survey) for 1876, Washington, 1879. 12 BUREAU OF AMEEICAW ETHNOLOGY [edll. 70 moimds near Ms ranch. As no name was given this village it is here called the Mitchell Spring Village. Morgan likewise mentions the ruin at Mud Spring and a tower in the ruin near his spring. Professor Newberry was the first author to affix the name Surouaro to a ruia situated at the head of the Yellow Jacket Canyon. Several of these ruins were described and figured by Mr, Warren K. Moorehead as "The Great Euins of Upper McElmo Creek" in the Illustrated American for July 9, 1892, the sixth of a series of articles under a general title "In search of a Lost Race." He gives descrip- tions of a "cave shelter" foimd near Twin Towers, Square Tower in "Euin Canyon," a buildiag (Hovenweep Castle), and the tower at the junction of the North and South Forks of Euin Canyon. This paper is, ajCcompanied by a map of Ruin Canyon by Mr. Cowen. In Moorehead's discussion of these remains, individual towers and other ruins are -designated by capital letters, A-V, to some of which are also affixed the names "Hollow Boulder," "Twin Towers," "Square Tower," etc. Details of structure and measurements of the more striking buildings and a discussion of certain features of structure, some of which will be considered later under individual rutas, are likewise given. The most important general article yet published on the prehis- toric remains of the region here considered is by Dr. T. Mitchell Prudden,^ who also mentions several of the ruins here treated. His most important contribution is a description of what he calls the "unit type," which he recognized as a fimdamental structural feature in the pueblos of this region. He also showed that the kiva in Montezuma Valley villages is identical with that of cliff-dwellings in the Mesa Verde, and emphasized, as an important feature, the union of the tower and the pueblo, a characteristic of the highest form of pueblo architecture. Doctor Prudden has followed his comprehensive paper above mentioned with an account '' of the excavation of one of the mounds at Mitchell Spring in which he adds to oiu- Icnowledge of the structure of his "unit tj'-pe." In "A Further Study of Prehistoric Small House Euins in the San Juan Watershed,' ' ^ Doctor Prudden has furnished important addi- tional data which shows the imiformity of the unit type over a large area of the San Juan drainage. The following among other prehistoric remains in the district mentioned or described by Doctor Prudden are covered by the author's reconnoissance: 1 The Prehistoric Ruins of the San Juan Watershed in Utah, Ari.:ona, Colorado, and New Mexico., Amgr. Anthrop., n. s. vol. v, no. 2, 1903. 2The Circular ICivas of Small Ruins in the San Juan Watershed. Amer. Anthrop., n. s. vol. xvi, no. 1, 1914. 8 Memoirs Amer. Anthrop. Asso., vol. v, no. 1, 1918. FEWKES] PEBHISTORIC VILLAGES, CASTLES, AND TOWEES 13 1. Ruins at Dolores Bend (Esoalante Ruin). 2. WoUey Ranch Ruin. 3. Burkhardt Ruin (Mud Spring Village). 4. Goodman Point Ruin. 5. Unnamed ruin west of Goodman Lake. 6. Ruin at junction of McElmo and Yellow Jacket. 7. Groxlp on Yellow Jacket nearly opposite mouth of Dawson Canyon (Davis or Littrell Tower). 8. Surouaro. 9. Cannonball Ruin. , ' 10. Towers and buildings of Ruin and Bridge Canyons.-'' 11. Pierson Spring Ruin. 12. Bug Spring Ruins. The following towers can be identified from his figures: ' 1. "Square building opposite mouth of Dawson Creek." Prud- den, pi. xviii, fig. 2. (This building is not square, but semicircular.) 2. Cannonball Ruin. Prudden, pi. xxi [xxii]. 3. "Small tower-like structure ... at the head of Ruin Canyon, in the Yellow Jacket group." Prudden, pi. xxiii, fig. 2. (This building is not in Ruin Canyon, but in Holly Canyon.) 4. "Tower . . . about the head of Ruin Canyon." Prudden, pi. xxiii, fig. 1. (This is the most eastern of the Twin Towers, but not about the head of the canyon.) 5. Sand Canyon Tower. Prudden, pi. xxiv, fig. 2. Although mainly devoted to descriptions of the clifF-houses of the Mesa Verde, Baron G. Nordenskiold's "CliflF Dwellers of the Mesa Verde" discusses in so broad a manner the relationship of pueblo ruins and cliff-houses that no student can overlook this epoch-making work. In fact, Nordenskiold laid the foundations for subsequent students of pueblo morphology, although some of his comparisons and generalizations were premature because based on imperfect observations which have been superseded by later investigations. The partial excavation of the excellent ruin at the head of Cannon- ball Canyon by S. G. Morley ^ sheds considerable light on the mor- phology of prehistoric buildings in the McElmo district. Unfortu- nately no attempt was made by him to repair the walls of this ruin for permanent preservation, but it is not too late still to prevent their further destruction and preserve them for future students and visitors. Morley's description of the buildings is accompanied by good photographs and a groimd plan. He brought to light in this ruin examples of the characteristic imit-type kiva. 1 Amer. Anthrop., n. s. vol. v, no. 2j 1903. sThe Excavation of tlie Cannonball Euins in Southwestern Colorado. Amer. Anthrop., ii. s. vol. x, no. 4, 1908. 14 BUREAU OF AMERICAN ETHNOLOGY [bull, n The latest work on the McEhno Ruins, one part of which has already- appeared, is a joint contribution by Morley and Eadder.' In this publication accurate dimensions and sites of ruins in the McElmo and Square Ruin Canyons are given, with other instructive data. Morley and Kidder have designated the ruins by Arabic numbers, and in a few instances by names. The author has preserved these numbers so far as possible in his accoimt. The following ruins in Ruin Canyon and neighboring district cov- ered by this reconnoissance are described by Morley and Kidder: No. 1. Wickyup Canyon, Ruin 1 and Ruin 2, "Boulder Castle." No. 2. Two towers in Ruin Canyon: I'', near the mouth; I'', Towers on or near forks, No. 1 [Hovenweep Pueblo], No. 2 [Hovenweep Castle.] No. 3. [Square Tower.] No. 4. [Oval Tower.] No. 5. [Tower.] No. 6. [6.] No. 7. [Boulder Chff-house.] No. 8. Twin Towers. No. 9. [9.] No. 10. [Unit-type House.] No. 11. Gibraltar House and ruin. [Stronghold House.] No. 12. [12.] The pueblos and cave dwellings of the "Pivotal group" (those on or near the promontory at the jimction of the McElmo and Yellow Jacket Canyons) were also studied by the authors. Alm ost the whole article by Morley and Kidder, which the editor announces wiU be completed in a future number of "El Palacio," is devoted to descriptions of buildings ^ in Ruin and Road (Wickyup) Canyons and the ruins of the "Pivotal group" at the base of a promontory between the jxmction of the Yellow Jacket and McElmo. CLASSIFICATION In the classification by Morley and Kidder and the majority of writers, sites rather than structural features are adopted as a basis,- although all recognized that large cliff-dweUings hke Cliff Palace are practically pueblos built in caves. In the following classification more attention is directed to differences in structure than to situation, notwithstanding the latter is convenient for descriptive purposes. 1. Villages or clusters of houses, each having the form of the pure pueblo type. The essential featm-e of the pm-e type is a compact 1 The Archaeology of McElmo Canyon, Colorado. El Palacio, vol. Iv, no. 4, Santa Fe, 1917. 2 The dimensions of buildings and lowers given in this article are welcome additions to our knowledge, but from lack of ground plans one can not fully determine the arrangement of rooms designated in indi- vidual ruins. FEWKEs] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 15 pueblo, containing one or more unit types, circular kivas of character- istic form, surrounded by rectangular rooms. These units, single or consolidated, may be grouped in clusters, as Mitchell Spring or Aztec Spring Kuins; the clusters may be fused into a large build- ing, as at Aztec or in the community buildings on Chaco Canyon. 2. Cliff-houses. These morphologically belong to the same pure type as pueblos; their sites in natural caves are insufficient to sepa- rate them from open-sky buildings. 3. Towers and great houses. These buildings occur united to cliff-dweUings or pueblos, but more often they are isolated. 4. Rooms with walls made of megaliths or small stone slabs set on edge. In reports on the excavation of Far View liouse ' on the Mesa Verde, the author called attention to clusters of mounds indicating ruined buildings in the neighborhood of Mummy Lake, a little more than 4 miles from Spruce-tree House. This cluster he considers a village; Far View House, excavated from one of the moimds, is regarded as a prehistoric pueblo of the pxn'e type. The forms of other buildings covered by the remaining moimds of the Mummy Lake site are unknown, but it is probable that they will be found to resemble Far View House, or that all members of the village have similar forms. This grouping of small pueblos into villages at Mummy Lake on the Mesa Verde is also a distinctive feature of ruins in the Montezuma Valley and McEhno district. In these villages one or more of the component houses may be larger and more conspicuous, dominating all the others, as at Goodman Point, or at Aztec Spring. The houses composing the village at Mud Spring were about the same size, but at WoUey Eanch Ruin only one mound remains, evidently the largest, the smaller having disappeared. The third group, towers and great houses, includes buildings of oval, circular, semicircular, and rectangular shapes. Morphologically speaking, they do not present structural features of pueblos, for they are not terraced, neither have they specialized circular ceremonial rooms, kivas with vaulted roofs surrounded by rectangular rooms, or other essential features of the pueblo type. The group contains build- ings which are sometimes consolidated with cliff-houses and pueblos, but are often independent of them. In this type are included castel- lated biuldings in the Mancos, YeUow Jacket, McElmo, and the numerous northern tributary canyons of the San Juan. 1 A Prehistoric Mosa Verde Pueblo and its People. Smithson. Eept. for 191S, pp. 461-488, 1917. Far View House — a Pure Type of Pueblo Euin. Art and Archaeology, vol. vi, no. 3, 1917. 16 BUREAU OF AMEEICAN ETHNOLOGY [bull, to Villages eeotangular euins of the pure type As the word is used in this report, a village is a cluster of houses separated from each other, each building constructed on the same plan, viz, a circular ceremonial room or kiva with miural banquettes and pilasters for the support of a vaulted roof, inclosed in rectangular rooms. When there is one kiva and surrounding angular rooms we adopt the name "unit type." When, as in the larger mounds, there are indications of several kivas or unit types consolidated — the size being in direct proportion to the number — ^we speak of the building as belonging to the "pure type." Doctor Prudden, who first pointed out the characteristics of the "unit type,"' has shown its wide dis- tribution in the McElmo district. The Mummy Lake village has 16 mounds indicating houses. Far View House, one of these houses, is made up of an aggregation of four unit types and hence belongs to the author's ','pure type." While villages similar to the Mummy Lake group, in the valleys near Mesa Verde, have individual variations, the essential f eatiires are the same, as will appear in thefoUowing descriptions of Surouaro, and ruins at Goodman Point, Mud Spring, Aztec Spring, and Mitchell Spring. Commonly, in these villages, one mound predominates in size over the others, and while rectangular in form, has generally circular depressions on the surface, recalling conditions at Far View mound before excavation. These mounds indicate large buildings in blocks, made up of many unit forms of the pure type, united into compact structures. One large dominant member of the village recalls those ruins where the village is consohdated into one community pueblo. The separation of mounds in the village and their concentration in the community house may be of chronological importance, although the relative age of the simple and composite forms can not at present be determined; but it is important to recognize that the units of con- struction in villages and community buildings are identical. SUROUABO The cluster of mounds formerly called Surouaro, now known as Yellow Jacket Spring Ruin, is situated near the head of the canyon of the same name to the left of the Monticello road, 14 miles west of Dolores. This village (pis. 1, c; 2, c) contains both large and small houses of the pure pueblo type, covering an area somewhat less than the Mummy Lake group, on the Mesa Verde. The arrangement of mounds in clusters naturally recalls the Galisteo and Jemez districts. New 1 The situation of the cemetery, one of the characters of Prudden's "unit type," appears constant in one-Mva buildings, but is variable in the pure type, and, as shown in the author's application of the unit type to the crowded condition in Spruce-tree House and other cliH-houses, does not occur in the same position as in pueblos of the pure type open to the sky. PEWKEs] PKEHISTOKIC VILLAGES, CASTLES, AND TOWERS 17 Mexico, where, however, the arrangement of the mounds and the structure of each is different. The individual houses in a Mesa Verde or Yellow Jacket village were not so grouped as to inclose a rectangu- lar court, but were irregularly distributed with intervals of consider- able size between them.' The largest mound in the Surouaro village, shown in plate 1, c, corresponds with the so-called "Upper House" of Aztec Spring Ruin, but is much larger than Far View or any other single mound in the Mummiy Lake village. Surouaro was one of the first ruins in this region described by American explorers, attention having been first called to it by Pro- fessor Newberry,^ whose description follows: "Surouaro is the name of a ruined town which must have once contained a population of several thousands. The name is said to be of Indian (Utah) origin, and to signify desolation, and certainly no better could have been selected. . . . The houses are, many of them, large, and all built of stone, hammer dressed on the exposed faces. Fragments of pot- tery are exceedingly common, though like the buildings, showing great age. . . . The remains of metates (corn mills) are abundant about the ruins. The ruins of several large reservoirs, built of masonry, may be seen at Surouaro, and there are traces of acequias which led to them, through '^hich water was brought, perhaps from a great distance. " Goodman Point Ruin This ruin is a cluster of small mounds surrounding larger ones, recalUng the arrangement at Aztec Spring. They naturally faU into two groups which from their direction or relation to the adja- cent spring may be called the south and north sections. The most important mound of the south section, Block A, meas- ures 74 feet on the north, 79 feet on the south, and 76 feet on the west side. This large mound corresponds morphologically to the "Upper House" at Aztec Spring (fig. 1, A). About it there are arranged at intervals, mainly on the north and east sides, other smaller mounds generally indicating rectangular buildings. The southeast angle of the largest is connected by a low wall with one of the smaller mounds, forming an enclosure called a court, whose northern border is the rim of the canyon just above the spring. A determination of the detailed architectural features of the buUding 1 In bis valuable study, Pueblo Euins of the Galisteo Basin, New Mexico (Anthrop. Papers of tbe Araer. Mus. Nat. Hist., vol. xv, pt. 1, 1914), Mr. Nelson figures (Plan I, £) an embedded circular kiva in what he calls the "historic part" of the Galisteo Euin, but does not state how he distinguishes the historic from the prehistoric part of this building. The other Mvas at Galisteo are few In number and not embedded, but situated outside the house masses as in historic pueblos. ' Report of the exploring expedition from Santa Pe, New Mexico, to the Junction of the Grand and Green Elvers of the Great Colorado of the West in 1859, under the command of Capt. J. N. Macomb, p. 88, Wash- ington, 1876. 108852 °^19— Bull. 70 2 18 BTJEEAU OF AMERICAN BTHHOLOGY tBULL. 70 buried under Block A is not possible, as none of its walls stand above the mass of fallen stones, but it is evident, from circular depressions and fragments of straight walls that appear over the surface of the mound, that the rooms were of two kinds, rectangular forms, or dwellings, and circular chambers, or kivas. This mound resembles Far View House on the Mesa Verde before excavation. A large circular depression, 56 feet in diameter, is situated in the midst of the largest mounds. A unique feature of this depression, recognized and described by Doctor Prudden, are four piles of stones, regularly arranged on the floor. The author adopts the suggestion that this area was once roofed and served as a central circular kiva, necessitating a roof of such dimensions that four masonry pillars served for its support. The mound measures about 15 feet in height, and has large trees growing on its surface, offering evidence of a considerable age. Several other rooms are indicated by circular surface depressions, but their relation to the rectangular rooms can be determined only by excavation. Johnson Ruin This ruin, to which the author was conducted by Mr. C. K. Davis, is about 4 miles west of the Goodman Point Kuiu near Mr. John- son's ranch house, in section 12, township 36, range 18. It is said to be situated at the head of Sand Canyon, a tributary of the McEhno, and is one of the largest ruLas visited. The remains of former houses skirt the rim of the canyon head for fuUy half a mile, forming a con- tinuous series of mounds ia which can be traced towers, great houses, and other types of buildings, and numerous depressions indicating sunken kivas. The walls of these buildings were, however, so tum- bled down that little now remains above ground save piles of stones in which tops of buried walls may stiU be detected, but not without some difficulty. In a cave under the "mesa rim" there is a small cliff-house in the walls of which extremities of the original wooden rafters still remain in place. In an open clearing, about 3 miles south and west of Mr. J. W. Fulk's house, Kenaraye post office, there is a small ruin of rectangu- lar form, the ground plan of which shows two rectangular sections of different sizes, joiued at one angle. The largest section measures approximately 20 by 50 feet. It consists of low rooms surrounding two circular depressions, possibly kiVas. Although constructed on a small scale, this section renunds one of the Upper House of Aztec Spring Euin. The smaller section, which also has a rectangular form, has remains of high rooms on opposite sides and low walls on the remaining sides. In the enclosed area there is a circular depres- sion or reservoir, corresponding with the reservoir of the Lower House at Aztec Spring Ruin. FEWKEs] PEEHISTOEIC VILLAGES, CASTLES, AND TO WEES 19 Buo Mesa Ruin The author was guided by Mr. H. S. Merchant to a village ruin, one of the largest visited, situated a few nules from his ranch house. This village is about 10 miles due south of the store at the head of Dove Creek, and consists of several large mounds, each about 500 feet long, arranged parallel to each other, and numerous isolated smaller mounds. Nor far from this large ruin there is a prehistoric reservoir estimated as covering about 4 acres. Many circular depres- sions, indicated kivas, and hnes of stones showed tops of buried rec- tangular rooms. Excavations in a small mound near this ruin were conducted by Doctor Prudden.' The canyon which heads near the corral on the road to Merchant's house revealed no evidence of prehistoric dwellings. Mitchell Spring Ruin This ruin takes its name from the earliest known description of it by Morgan,^ which was compiled from notes by Mr. Mitchell, one of the early settlers in Montezuma Valley. Morgan's account is as follows : "Near Mr. Mitchell's ranch, and within a space of less than a mile square, are the ruins of nine pueblo houses of moderate size. They are built of sandstone intermixed with cobblestone and adobe mortar. They are now in a very ruinous condition, without standing waJls in any part of them above the rubbish. The largest of the number is marked No. 1 in the plan, figure 44, of which the outline of the original structure is still discernible. It is 94 feet in length and 47 feet in depth, and shows the remains of a stone wall in front inclosing a small court about 15 feet wide. The mass of material over some parts of this structure is 10 or 12 feet deep. There are, no doubt, rooms with a portion of the walls still standing covered with rubbish, the removal of which would reveal a considerable portion of the original ground plan." The author paid a short visit to the Mitchell Spring village and by means of Morgan's sketch map was able to identify without difficulty the nine mounds and tower he represents. The village at Mitchell Spring differs from that at Mud Spring and at Aztec Spring mainly in the small size and diffuse distribution of the component moimds and an absence of any one mound larger than the remainder. It had, however, a round tower, but unlike that at Mud Spring village, this structure is not united to one of the houses. The addition of towers to pueblos, as pointed out by Doctor Prudden^ several years ago, marks the highest development of pueblo architecture as shown 1 Memoirs Amer. Anthrop. Asso., vol. v, no. 1, 1918. 2 Houses and House-life olthe American Aborigines. Cent. N. Amer. Ethn., vol. iv, pp. 189-190, 1831. s Prudden excavated a unit-type ruin from one of the Mitchell Spring mounds. (Amer. Anthrop., vol. xvi, no. 1, 1914.) 20 BTJEBAU OF AMERICAN ETHNOLOGY [edli.. to not only in open-air villages but also in some of the large cliff pueblos, like Cliff Palace. Isolated towers are as a rule earlier in construction. The unit-type mound uncovered by Doctor Prudden is one of the most instructive examples of this type in Montezmna Canyon, but the author in subsequent pages wiU call attention to the existence of the same type in Square Tower Canyon. All of these pueblos probably have kivas of the pure type, practically the same in structure as Par View House on the Mesa Verde National Park. Mud Spring (Buekhaedt) Buin The collection of mounds (pi. 3, 6), sometimes called Burkhardt Ruin, situated at Mud Spring, belongs to the McEImo series. This ruin, in which is the " triple-waUed tower" of Holmes, for uniformity with Mitchell Spring Ruin and Aztec Spring Ruin, is named after a neighboring spring. Like these, it is a cluster of mounds forming a village which covers a considerable area. The arroyo on which it is situated opens into the McElmo, and is about 7 mUes southwest from Cortez, at a point where the road enters the McElmo Canyon. The extension of the area covered by the Mud Spring mounds is east-west, the largest mounds being those on the east. These latter are separated from the remainder, or those on the west, by a shallow, narrow gulch. There are two towers united to the western section overlooking the spring, the following description of one of which, with a sketch of the ground plan, is given by Holmes.* "The circular structtu-es or towers have been built, in the usual manner, of roughly hewn stone, and rank among the very best specimens of this ancient architecture. The great tower is especially noticeable ... In dimensions it is almost identical with the great tower of the Rio Mancos. The walls are traceable nearly all the way round, and the space between the two outer ones, which is about 5 feet in width, contains 14 apartments or ceUs. The walls about one of these cells are still standing to the height of 12 feet; but the interior can not be examined on account of the rubbish which fill s it to the top. No openings are noticeable in the circular walls, but doorways seem to have been made to commimicate between the apart- ments; one is preserved at d . . . This tower stands back about 100 feet from the edge of the mesa near the border of the village. The smaller tower, 6, stands forward on a point that overlooks the shallow gulch; it is 15 feet in diameter; the walls are 3^ feet thick and 5 feet high on the outside. Beneath this ruin, in a little side gulch, are the remains of a wall 12 feet high and 20 inches thick . . . The apartments number nearly a hundred, and seem, generally, to have been rectangular. They are not, however, of uniform size, and certainly not arranged in regular order." I Op. cit., pp. 398-399. FEWKEs] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 21 Morgan ' gires the following description of the same ruin which seems to the author to be the Mud Creek village: "Four miles westerly [from Mitchell ranch], near the ranch of Mr. Shirt, are the ruins of another large stone pueblo, together with an Indian cemetery, where each grave is marked by a border of flat stones set level with the ground in the form of a parallelogram 8 feet by 4 feet. Near the cluster of nine pueblos shown in the figure are found strewn on the ground numerous fragments of pottery of high grade in the ornamentation, and small arrowheads of flint, quartz, and chalcedony dehcately formed, and smaU knife blades with convex and serrated edges in considerable numbers. "This is an immense ruin with small portions of the waUs still standing, particularly of the round tower of stone of three concentric walls, incorporated in the structure, and a few chambers in the north end of the main building. The round tower is still standing nearly to the height of the first story. In its present condition it was impossible to make a ground plan showing the several chambers, or to determine with certainty which side was the front of the struc- ture, assimiing that it was constructed in the terraced form . . . The Round Tower is the most singular feature in this structure. While it resembles the ordinary estufa, common to all these structures, it differs from them in having three concentric walls. No doorways are visible in the portion still standing, consequently it must have been entered through the roof, in which respect it agrees with the ordinary estufa. The inner chamber is about 20 feet in diameter, and the spaces between the. encircling walls are about 2 feet each; thewaUs are about 2 feet in thickness, and were laid up mainly with stones about 4 inches square, and, for the most part, in courses. There is a similar round tower, having but two concentric walls,, at the head of the McElmo Canyon, and near the ranch of Mr. Mitchell [Mitchell Ruin]." Aa the name Mud Spring is locally known to the natives, especially to employees of livery stables and garages, the ruin is here called Mud Spring. The tower and the other circular buildings are united to other rooms as in similar groups of mounds. The presence of surface depressions, thought to indicate circular kivas,' shows that the Mud Spring mounds are remains of a viUage of the same type as the Mxunmy Lake group, but with towers united to the largest mounds. The time the author coifld give to his visit to the Mud Spring Ruin (pi. 3, h) was too limited to survey it, but he noticed in addition to the two circular buildings already recorded, a large mound situated on the west side of the gulch, and numerous small mounds on the east 1 Op. cit., p. 190. 2 Although the kivas of Mud Spring Ruin have not been excavated there is little doubt from surface indications that they belong to the unit type. 22 BUEEAU OP AMERICAN ETHNOLOGY [bull. 7Q side of the same, each apparently with, a central depression like a kiva. All these mounds have been more or less mutilated by indiscruniaate digging, but many mounds, still untouched, remain to be excavated before we can form an adequate conception of the group. The "triple-walled tower" is now in such a condition that the author could not determine whether it was formerly circular or D-shaped; the "small tower" is in even worse condition and its previous form could not be made out. The Mud Spring mounds cover a much larger area than descriptions or ground plans thus far published would indicate. Originally Mud Spring Ruin consisted of a cluster of pueblos of various sizes, each probably with a circular kiva and rectangular rooms, combined with one or more towers at present too much dilapidated to determine architectural details without excavations. Like the other clusters of pueblos in the McEhno and Montezimia VaUey, the cemetery near Mud Spring Ruin has suffered considerably from pothunters, but there still remain many standing walls that are well preserved. Ruin with Semicircular Core This ruin is situated on the San Juan about 3 miles below the sandy bed of the mouth of the Montezuma, on a bluff 50 feet above the river. The ground plan by Jackson ^ indicates a building shaped like a trapezoid, 158 feet on the northeast side, 120 on the southeast, and 32 on the northwest side. The southwest side is broken mid- way by a reentering area at the rim of the bluff over the river. In the center of this trapezoidal structure there is represented a series of rooms arranged hke those of Horseshoe House, but com- posed of a half-circular chamber surrounded by seven rooms between two concentric circular walls. Thus far the homology to Horseshoe House is close but beyond this series of rooms, following out the trapezoidal form, at least five other rooms appear on the ground plan. The position of these recalls the walls arranged around the tower at Mud Spring village. In other words, the ruin resembles Horseshoe House, but has in addition rectangular rooms, added on three sides, forming an angular building. So far as the author's information goes, no other ruin of exactly this type, which recalls Sun Temple, has been described by other observers. I WoLLEY Ranch Ruin WoUey Ranch Ruin, situated 10 miles south of Dolores, is one of the largest mounds near Cortez. There are evidences of the former existence of a cluster of mounds at this , place, only one of which now remains. This is covered with bushes, rendering it difficult to trace the bounding walls. 1 Tenth Ann. Rept. U. S. Geol. Surv. Terr. (Hayden Survey) lor 1876, pi. xlvili, fig. 2, 1879. FEWKEs] PEEHISTOBIC VILLAGES, CASTLES, AHD TOWEES 23 Blanohard Ruin Several years ago private parties constructed at Manitou, near Colorado Springs, a cliff-dwelling on the combined plan of Spruce- tree House and Cliff Palace. The rocks used for that purpose were transported from a large mound on the Blanchard ranch near Leba- non, in the Montezuma Valley, at the head of Hartman's draw, about 6 miles south of Dolores. Two mounds (pi. 2, a, I), about three-quarters of a mUe apart, are all that now remain of a consider- able village; the other smaller mounds, reported by pioneer settlers, have long since been leveled by cultivation. As both of these mounds have been extensively dug iato to obtain stones, the walls that remain standing show much mutilation. The present condition of the largest Blanchard mound, as seen from its southwest angle, is shown in plate 2, h. About half of the mound, now covered with a growth of bushes, still remains entire, exposing walls of fine masonry, on its south side. The rooms in the buried buildings are hard to make out on account of this covering of vegetation and accumu- lated debris; but the central depressions, supposed to be kivas, almost always present in the middle of mounds in this district, show that the structure of Blanchard Ruin follows the pure type. Ruins at Aztec Spring The mounds at Aztec Spring (pi. 1, 6), situated on the eastern flank of Ute Mountain, at a site looking across the valley to the west end of Mesa Verde, were described forty years ago by W. W. Jackson ' and Prof. W. H. Holmes.^ The descriptions given by both these pioneers are quoted at length for the reason that subsequent authors have added little from direct observation since that time, notwithstandiag they have been constantly referred to and the illustrations reproduced. As a result of a short visit, the author is able to add the few fol- lowing notes on the Aztec Spring mounds. The ruin is a village consisting of a cluster of unit pueblos of the pure type in various stages of consolidation. J^o excavations were made, but the surface indications point to the conclusion that the diflferent mounds indi- cate that these pueblos have different shapes and sizes. The author's observations differ in several unimportant partic- ulars from those of previous writers, and while it is not his intention to describe in detail the Aztec Spring village he .will call attention to certain features it shares with other villages in the Montezuma VaUey. > Kept. U. S. Geol. Survey Terr. (Hayden Survey) for 1874, Washington, 1876. s Op. cit. 24 BUREAU OF AMERICAN ETHNOLOGY tBui-t. 70 The best, almost the only accounts of this village are the follow- ing taken from the descriptions by Jackson and Holmes published ill 1877. Mr. Jackson gives the following description:^ ''Itninediately adjoining the spring, on the right, as we face it f jftjm below, is the ruin of a great massive structure [Upper House ?] of some kind, about 100 feet square in exterior dimensions; a portion only of the wall upon the northern face remaining in its original posi- tion. The debris of the ruin now forms a great mound of crumbling rock, from 12 to 20 feet in height, overgrown with artemisia, but showing clearly, however, its rectangular structure, adjusted approx- imately to the four points of the compass. Inside this square is a circle, about 60 feet in diameter, deeply depressed in the center. The space between the square and the circle appeared, upon a hasty examination, to have been filled in sohdly with a sort of rubble- masonry. Cross-walls were noticed in two places; but whether they were to strengthen the walls or divided apartments cotild only be conjectured. That portion of the outer wall remaining standing is some 40 feet in length and 15 in height. The stones were dressed to a uniform size and finish. Upon the same level as this ruin, and extending back some distance, were grouped line after line of foundations and mounds, the great mass of which is of stone but not one remaining upon another . . . Below the above group, some 200, yards distant, and communicating by indistinct ^nes of dSiris, is; another great wall, inclosing a space of about 200 feet square [Lower' House?] . . . This better preserved portion is some 50 feet in length, 7 or 8 feet in height, and 20 feet thick, the two exterior sur- face of well-dressed and evenly laid courses, and the center packed' in solidly with rubble-masonry, looking entirely different from those; rooms which had been fiUed with debris, though it is difficult toi assign any reason for its being so massively constructed . . . Thei town built about this spring is nearly a square mile in extent, the^ larger and more enduring buildings in the center, while all about, are scattered and grouped the remnants of smaller structures, com-- prising the suburbs.' The description by Professor Holmes ' is more detailed and accompanied by a ground plan, and is quoted below: "The site of the spring I found, but without the least appearance of water. The depression formerly occupied by it is near the center of a large mass of ruins, similar to the group [Mud Spring village] last described, but paving a rectangular instead of a circular building as the chief and central structure. This I have called the upper house in the plate, and a large walled enclosure a little lower on the slope I have for the sake of distinction called the lower house. . _ . _ _ . ' 1" - ^ - z r= ........■■ » f Op. cit., pp. 377-378, ' Op. cit., p. 400. ffEWKEs] PEEHISTOEIC VILLAGES, CAS1LES, AITD TOWEES 25 "These ruins form the most imposing pile of masonry yet [1875] found in Colorado. The whole group covers an area about 480,000 square feet, and has an average depth of from 3 to 4 feet. This would give in the vicinity of 1,500,000 solid feet of stonework. The. stone used is chiefly of the fossiliferous limestone that outcrop, along the base of the Mesa Verde a mile or more away, and its trans- portation to this place has doubtless been a great work for a people so totally without facilities. "The upper house is rectangular, measuring 80 feet by 100 feet, and is built with the cardinal points to within a few degrees. The pile is from 12 to 15 feet in height, and its massiveness suggests an original height at least twice as great. The plan is somewhat difficult to make out on account of the very great quantity of debris. "The walls seem to have been double, with a space 7 feet be- tween; a number of cross walls at regular intervals indicate that this space has been divided into apartments, as seen ia the plan. "The walls are 26 inches thick, and are built of roughly dressed stones, which were probably laid in mortar, as in other cases. "The enclosed space, which is somewhat depressed, has two lines of debris, probably the remains of partition-walls, separating it into three apartments, a, h, c [note]. Enclosing this great house is a network of fallen walls, so completely reduced that none of the stones seem to remain in place; and I am at a loss to determine whether they mark the site of a cluster of irregular apartments, having low, loosely built walls, or whether they are the remains of some imposing adobe structure built after the manner of the ruined pueblos of the Rio Chaco. "Two well-defined circular enclosures or estufas [kivas] are situated in the midst' of the southern wing of the ruin. The upper one, A^ is on the opposite side of the spring from the great house, is 60 feet in diameter, and is surrounded by a low stone wall. West of the house is a small open court, which seems to have had a gateway opening out to the west, through the surrounding walls. "The lower house is 200 feet in length by 180 in width, and its walls vary 15 degrees from the cardinal points. The northern wall, a, is double and contains a row of eight apartments about 7 fefet in width by 24 in length. The walls of the other sides are low, arid seem to have served simply to enclose the great court, near the center of which is a large walled depression (estufa B). " The number of bliildings that composed the Aztec Spring village (fig. 1) when it was inhabited can not be exactly estimated, but as indicated by the largest mound, the most important block of rooms exceeds in size any at Mitchell Spring Euin. While this village also covered more ground than that at Mud Spring, it shows no evidence of added towers, a prominent feature of the largest mound of the 26 BTJREAU or AMERICAN ETHNOLOGY [BnLI.. 70 latter. Two sections (fig. 1, A, B) may be distinguislied in the arrangement of mounds in the village; one may be known as the western and the other as the eastern division. . The highest and most conspicuous mound of the western section (A) is referred to by Professor Holmes as the "Upper House." Surface characteristics now indicate that this is the remains of a compact rectangular buUding, with circular kivas and domiciliary rooms of different shapes, the arrangement of which can not be de- I O I Fio. 1.— Ground plan ol Azteo Spring Ruin. termined without extensive excavations. The plan of this pueblo published by Holmes ^ shows two large and one small depression, indicating peripheral rectangular chambers surrounded by walls of rectangular rooms. The author interprets the depressions, K, as kivas, but supposes that they were not rectangular as figured by Holmes, but circular, surrounded on all four sides by square secular chambers, the "Upper House " being formed by the consolidation of several units of the pure pueblo type. Although Aztec Spring Ruin is now much mutilated 1 Op. oit., pi. xl. FEWKES] PKEHISTORIC VILLAGES, CASTLES, AND TOWERS 27 and its walls difficult to trace, the surface indications, aided by com- parative studies of the rooms, show that Holmes' "a," "h," and "c," now shown by depressions, are circular, subterranean kivas. They are the same kind of chambers as the circular depressions in the mounds on the south side of the spring. The height of the mound called "Upper House" indicates that the building had more {han one story on the west and north sides, and that a series of rooms one story high with accompanying circular depressions existed on the east side. The "Upper House" is only one of several pueblos composing the western cluster of the Aztec Spring village. Its proximity to the source of water may in part account for its predominant size, but there are evidences of several other mounds (E-H) in its neighbor- hood, also remains of pueblos. Those on the north (C) and west sides (E-H) are small and separated from it by intervals sometimes called courts. The most extensive accumulation of rooms next the "Upper House" is situated across the draw in which the spring lies, south of the "Upper House" cluster already considered. The aggregation of houses near the "Upper House " is mainly composed of low rectangular buildings among which are recognized scattered circular depressions indicating kivas. The largest of these buildings is indicated by the mound on the south rim of the draw, where we can make out remains of a number of circular depressions or kivas (K), as if several unit forms fused together; on the north and west sides of the spring there are small, low mounds, unconnected', also suggesting several similar unit forms. The most densely populated part of the village at Aztec Spring, as indicated by the size of the mounds clustered on the rim around the head of the draw, is above the spring, on the northwest and south sides. There remains to be mentioned the eastern annex (B) of the Aztec Spring village, the most striking remains of which is a rectangular inclosure called "Lower House," situated east of the spring and lower down the draw, or at a lower level than the section already considered. The type of this structure, which undoubtedly belonged to the same village, is different from that already described. -It resembles a reservoir rather than a kiva, inclosed by a low rectan- gular wall, with rows of rooms on the north side. The court of the "Lower House" measures 218 feet. The waU on the east, south, and west sides is only a few feet high and is narrow; that on the north is broader and higher, evidently the remains of rooms, over- looking the inclosed area. Perhaps the most enigmatical structures in the vicinity of Aztec Spring village are situated on a low mesa south of the mounds, a few hundred feet away. These are circular depressions without accom- panying mounds, one of which was excavated a few years ago to the 28 BUEEAtr OF AMERICAN ETHNOLOGY tsotl.. lo depth of 12 feet; on the south there was discovered a well-inade wall of a circular opening, now visible, by which there was a communica- tion through a horizontal tunnel with the open air. The author was informed that this tunnel is artificial and that one of the workmen crawled through it to its opening in the side of a bank many yards distant. No attempt was made to get the exact dimensions of the com- ponent houses at Aztec Spring, as the walls are now concealed in the mounds, and measurements can only be approximations if obtained from surface indications without excavation. The sketch plan here introduced (fig. 1) is schematic, but although not claimed as accurate, may serve to convey a better idea of the relation of the two great structures and their annexed buildings than any previously advanced. The author saw no ruined prehistoric village in the Montezuma Valley that so stirred his enthusiasm to properly excavate and repair as that at Aztec Spring,* notwithstanding it has been considerably dug over for commercial purposes. Great Open-Air Ruins South and Southwest of Dove Cbebk Post Ofmcb In the region south and southwest of Dove Creek there are several large pueblo ruins, indicated by moimds formed of trimmed stone, eolean sand, and clay from plastering, which have certain charac- ters in common. Each mound is a large heap of stones (pi. 3, a) near which is a depression or reservoir, with smaller heaps which in different ruins show the small buildings of the imit type. These clusters or villages are somewhat modified in form by the configura- tion of the mesa surface. The larger have rectangular forms regu- larly disposed in blocks with passageways between them or are without any definite arrangement. Squaw Point Ruin This large ruin, which has been described by Doctor Prudden as Squaw Point Kuin and as Pierson Lake Ruin, was visited by the author, who has little to add to this description. One of the small heaps of stone or mounds has been excavated and its structure found to conform with the definition of the unit type. The subter- ranean communication between one of the rectangular rooms and the kiva could be well seen at the time of the author's visit and recalls the feature pointed out by him ia some of the kivas of Spruce- tree House. The large reservoir and the great ruin are noteworthy features of the Squaw Point settlement. It seems to the author that the large block of buildings is simply a congeries of unit types the structure of one of which is indicated by 1 Mr. Van Kloecfc, of Denver, has oflorod this ruin to the Public Parks Service tor permanent preser- vation. It is proposed to rename it the Yucca House National Monument. BEWKEs] PEEHISTOKIC VILLAGES, OASTLES, AND TOWERS 29 the small buildings excavated by Doctor Prudden, and that structiir- aUy there is the same condition in it as in the pueblo ruins of Monte- zuma Valley, a conclusion to which the several artifacts mentioned and figured by Doctor Prudden also point. The same holds true of Bug Point Euin, a few mUes away, also excavated and described by Doctor Prudden. Here also excavation of a small mound shows the unit type, and while no one has yet opened the larger mound or pueblo, superficial evidences indicate that it also is a complex of many unit types joined together. Until more facts are available the relative age of the small unit types as com- pared to the large pueblo can not be definitely stated, but there is little reason to doubt that they are contemporaneous, and nothing to support the behef that they do not indicate the same culture. AcMEN Ruin Following the Old Bluff Road and leaving it about 5 miles west of Acmen post office, one comes to a low canyon beyond Pigge ranch. The heaps of stone or large mounds cover an area of about 10 acres, the largest being about 15 feet high. East of this is a circular depression surroimded by stones, indicating either a reservoir or a ruined building. The top of the highest mound (pi. 3, a) — no waUs stand above the surface — is depressed hke mounds of the Mummy Lake group on the Mesa Verde. This depression probably indicates a circular kiva embedded in square w^s, the masonry of which so far as can be judged superficially is not very fine. There are many smaller mounds in the vicinity and evidences of cemeteries on the south, east, and west sides, where there are evidences of desultory digging; fragments of pottery are numerous. These moimds indicate a considerable village which would well repay excavation, as shown by the numerous specimens of corrugated, black and white, and red pottery in the Pigge collection, made in a small mound near the Pigge ranch. The specimens in this collection present few features different from those indicated by the fragments of pottery picked up on the larger moimds a mile west of the site where they were excavated. They are the same as shards from the mounds in the McElmo region. Oak Spring House About 15 miles southwest of Dove Creek on Monument Canyon there is a good spring called Oak Spring, near which are several piles of stones indicating former buildings, the largest of which, about a quarter of a mile away, has a central depression with surrounding walls now covered with rock or buried in soil or blown sand. Very large pinon trees grow on top of the highest walls of this ruin, the 30 BUREAU OF AMERICAN ETHNOLOGY [btill. 70 general features of which recall those at Bug Spring, though their size is considerably less. In the surface of rock above the spring there are numerous potholes of small size. One of these, 4 feet deep and about 18 feet in diameter, is almost perfectly circular and has some signs of having been deepened artificially. It holds water much of the time and was undoubtedly a source of water supply to the aborigines, as it now is to stock in that neighborhood. Ruin in Buin Canyon One of the large rim-rock ruins may be seen on the left bank of Ruia Canyon in fuU view from the Old Bluff Road. The ruin is an immense pile of stones perched on the very edge of the rim, with no walls standing above the surface. The most striking feature of this ruin is the cliff-house below, the walls and entrance into which are visible from the road (pi. 9, b). It is readily accessible and one of the largest in the country. On either side of the Old Bluff Road from Ruin Canyon to the "Aztec Reservoir" small piles of stone mark the sites of many former buildings of the one-house type which can readily be seen, especially in the sagebrush clearings as the road descends to the Picket corral, the reservoirs, and the McEhno Canyon. Cannonball Ruin One of the most instructive ruins of the McEImo Canyon region is situated at the head of Cannonball Canyon, a short distance across the mesa north of the McElmo, at a point nearly opposite the store. This ruin is made up of two separate pueblos facing each other, one of which is known as the northern, the other as the southern pueblo (pi. 22, 6). Both show castellated chambers and towers, one of which is situated at the bottom of the canyon. The southern pueblo was excavated a few years ago by Mr. S. G. Morley, who published an excellent plan and a good description of it, and made several sug- gestions regarding additions of new rooms to the kivas which are valuable. Its walls were not protected and are rapidly deteriorating. This pueblo, as pointed out by Mr. Morley,* has 29 secular rooms arranged with httle regularity, and 7 circular kivas, belonging to the vaulted-roofed variety. It is a fine example of a composite pueblo of the pure type, in which there are several large kivas. Morley has pointed out a possible sequence in the addition of the different kivaa to a preexisting tower and offers an explanation of the chronological steps by which he thinks the aggregation of rooms was brought about. Occasionally we find inserted in the walls of those houses large arti- ficially worked or uncut flat stones, such as the author has mentioned as existing in the walls of the northwest comer of the court of Far View House. This Cyclopean form of masonry is primitive and may 1 Amer. Anthrop., n. s. vol. X, no. i, pp. 596-610, 1908. FEWKES] PKEHISTOEIC VILLAGES, CASTLES, AND TOWERS 31 be looked upon, as a survival of a ruder and more archaic condition best shown in the Montezuma Mesa ruins farther west, a good ex- ample of which was described by Jackson.* CIROULAB RUINS WITH PERIPHERAL COMPARTMENTS It has long been recognized that circular ruins in the Southwest differ from rectangular ruins, not only in shape but also in struc- tural features, as relative position and character of kivas. The rela- tion of the ceremonial chambers to the houses, no less than the external forms of the two, at first sight appear to separate them from the pure type.^ They are more numerous and probably more ancient, as their relative abundance imphes. These circular ruins, in which group is included certain modifica- tions where the curve of one side is replaced (generally on the south) by a straight wall or chord, have several concentric walls; again, they take the form of simple towers with one row of encircling compart- ments, or they may have a double wall with inclosed compartments. Many representations of semicircular ruins were found in the region here considered, some of which are of considerable size. The simplest form is well illustrated by the D-shaped building. Horseshoe House, in Hackberry Canyon, a ruin which wiU be considered later in this article. Other examples occur in the Yellow Jacket, and there are several, as Butte Euin, Emerson, and Escalante Ruins, in the neigh- borhood of Dolores. In contrast to the village type consisting of a number of pueblos clustered together, but separated from each other, where the growth takes place mainly through the union of components, the circular ruin in enlarging its size apparently did so by the addition of new compartments peripherally or like additional rings in exogenous trees. Judging from their frequency, the center of distribution of the circular type lies somewhere in the San Juan culture area. This type does not occur in the Gila Valley or its tributaries, where we have an architectural zone denoting that a people somewhat different in culture from the Pueblos exists, but occurs throughout the " Cen- tral Zone," so called, extending across New Mexico from Colorado as far south as Zuni. Many additional observations remain to be made before we can adequately define the group known as the circular type and the extent of the area over which it is distributed. The following examples of this type have been studied by the author: 1 Op. oit., pp. 428-429. 2 It is premature to declare that the kivas in circular ruins do not belong to the raulted-rooled type simply Iroffl w^t gf obseryatjpfl to that effect. In Penasco Blanco and other ruins of the Chaco Canyon group, as shgwn in ground plaas, they appear to be embedded in secular rooms. Additional studies of the architectural features §f e^guJar pii^blps we desirable. 32 BUREAU OF AMERICAN ETHNOLOGY [BULL. 70 Wood Canyon Ruins Keports were brought to the author of large ruins on the rim of Wood Canyon, about 4 miles south of Yellow Jacket post office, in October, 1918, when he had almost finished the season's work. Two ruins of size were examined, one of which, situated in the open sage- brush clearing, belongs to the village type composed of large and small rectangular mounds. The other is composed of small circular or semicircular buUdings with a surrounding wall. The form of this latter (fig. 2) would seem to place it in a subgroup or village type. Fig. 2.— Ground plan ol Wood Canyon Euin. Approach to the inclosed circular mounds was debarred by a high bluff of a canyon on one side and by a low defensive curved wall (E), some of the stones of which are large, almost megahths, on the side of the mesa. From fragmentary sections of the bmied walls of one of these circular mounds {A, B), which appear on the surface, it would seem that the buildings were like towers (C, D). This is one of the few known examples of circular buildings in an area protected by a curved wall. In the cliffs below Wood Canyon Ruin is a cliff- dwelling {G li, J) remarkable mainly in its site. Butte Ruin The so-called Butte Ruin, situated in Lost Canyon, 5 mUes east of Dolores, belongs to the circular type. It crowns a low elevation, s^ieep on the west side, sloping more gradually on tho east, and sur- ffEWKEs] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 33 rounded by oidtivated fields. The view froBi its top looking toward Ute Mountain and the Mesa Verde plateau is particularly extensive. The butte is forested by a few spruces growing at the base and extending up the sides, which are replaced at the summit by a thick growth of sage and other bushes which cover the mound, rendering it difficult to make out the ground plan of the ruin on its top. From what appears on the surface ii would seem that this ruin was a circular or semicircular buHduig about 60 feet in diameter, the walls rising about 10 feet high. Like other circular mounds it shows a well-marked depression in the middle, from which radiate walls or indications of waUed compartments. Like the majority of the buUdings of the circular form, the walls on one side have fallen, suggesting that a low straight wall, possibly with rectangular rooms, was annexed to this side. In the neighborhood of Butte Kuin there is another hOl crowned with a pile of stones, probably a round building of smaller size and with more dilg,pidated walls. Old cedar beams project in places out of the mounds. The cliff-houses below the largest of these mounds show weU- made walls with a few rafters and beams. There are pictographs on the cliff a short distance away. Emerson Ruin This ruia crowns a low hill about 3 miles south of Dolores (fig. 3). The form of the mound is semicircular with a depression in the middle around which can be traced radiating partitions suggesting compartments. Its outer wall on the south side, as in so many other examples of this type, has fallen, and the indications are that here the wall was straight, or like that on the south side of Horse- shoe Ruin. The author's attention was first called to this ruin by Mr. Gordon Parker, supervisor of the Montezuma Forest Reserve, it having been discovered by Mr. J. W. Emerson, one of his rangers. The circular or semicircular form (fig. 4) of the mound indicates at once that it does not belong to the same type as Far View House; the central depression is surrounded by a series of compartments separated by radiating walls like the circular ruins in the pueblo region to the south. Mr. Emerson's report, which follows, points out the main features of this remarkable ruin.' "The letter referring to the circular ruin near Dolores was prepared by Mr. Emerson, the discoverer of this ruin, and was transmitted to the Smithsonian Institution as part of a phase of cooperative work with the Forest Service, by Mr. Gordon Parker, superintendent of the Montezuma Forest Reserve. 108852°— 19— Bull. 70- 3 34 BUREAU OF AMEKICAN ETHNOLOGY [BULL. 70 DoLOEBS, Colorado, July 7, 1917. In August, 1916, I visited Mesa Verde National Park. While there Doctor Fewkes inquired in regard to ruins in the vicinity of the Big Bend of the Dolores River. He informed me that the log of two old Spanish explorers of 1775 described a ruin near the bend of the Dolores River as of great value. Later, during October, 1916, 1 visited a number of ruins in this vicinity, including the one which (for the want of a better name) I have mapped and named Sun Dial Palace. Later, last fall, I again visited these ruins with Mr. R. W. Williamson, of Dolores, Colorado. ( W^MMay. i Hd. i Feh.25,1881 \ IfESERVeiR I EMERSON RUIN Chas.A.Kincf. Ed. May 5,1883. SecM Filed on. All of section filed on, but since last date on map so not paienied. Scale 4in..'lTni. VanK^aOE. Fig. 3. — Metes and bounds of Emerson Ruin. (After Emerson.) On July 5, 1917, I again visited these ruins, which I have designated as Reservoir Group and Sun Dial Palace.' For location and status of land on which they lie see map of sec. 17, T. 37 N., R. 15 W., N. M. P. M. [fig. 3]. While examining Sun Dial Palace I noted the "D-shaped construction, also that the south wall of the building ran due east and west." Also please note the regu- larity of wall bearings from the approximate center of the elliptical center chamber. I also noted that a shadow cast by the sun apparently coincides with some of these walls at different hours during the day. This last gave suggestion to the name. <■ Also SCO detailed map of construction of Sun Dial Palace [fig. 4]. FEWKEs] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 35 Also please note that the first tier of rooms around the middle chamber does not show a complete set of bearings but seems to suggest that these regular bearings were obtained from observation and study of a master builder. The result of his study was built as the next circular room tier was added. The two missing rooms on the western side of the building seem to suggest that this building was never completed, and also bear out my theory of an outward building of room tiers from the middle chamber. On the ground this buUdiag is fully completed on the south side and forms a due east and west line. An error in mapping the elliptical middle chamber has given the south side an incomplete appearance. I believe that the excavation and study of this ruin will recall something of value, as Father Escalante wrote in his log in 1775. Respectfully submitted. (Signed) J. Waed Emekson, Forest Ranger. W-«- f / 1 — /- j— -X \ \ 4 \ *-E. S Fig. 4. — Schematic ground plan of Emerson Euin. (After Emerson.) A personal examination of the remains of this buUding leads the author to the conclusion that while it belongs to the circular group, with a ground plan resembling Horseshoe House, and while the central part had a wall completely circular, the outer concentric curved walls did not complete their course on the south side, but ended in straight walls comparable with the partitions separating compartments. The author identifies another ruin as that mentioned by the Catholic fathers in 1775. 36 BUREAU OF AMEEICAN ETHNOLOGY [bdli,. to ESCALANTE EuiN The name Escalante Euin, given to the first ruin recorded by a white man in Colorado, is situated about 3 miles from Dolores on top of a low hill to the right of the Monticello Road, just beyond where it diverges from the road to Cortez. The outline of the pile of stones suggests a D-shaped or semicircular house with a central depression surrounded by rooms separated by radiating partitions. The wall on the south or east sides was probably straight, rendering the form not greatly unlike the other ruins on hiUtops in the neigh- borhood of Dolores. This is supposed to be the ruin to which reference is made in the following quotation from an article in Science : ' "There is in the Congressional Library, among the documents collected by Peter Force, a manuscript diary of early exploration in New Mexico, Colorado, and Utah, dated 1776, written by two Cath- olic priests, Father Silvester Velez Escalante and Father Francisco Atanacio Dominguez. This diary is valuable to students of arche- ology, as it contains the first reference to a prehistoric ruin in the con- fines of the present State of Colorado, although the mention is too brief for positive identification of the ruin.^ While the context indi- cates its approximate site, there are at this place at least two large ruins, either of which might be that referred to. I have no doubt which one of these two ruins was indicated by these early explorers, but my interest in this ruin is both archeological and historical. Our knowledge of the structure of these ruins is at the present day almost as imperfect as it was a century and a half ago. "The route followed by the writers of the diary was possibly an Indian pathway, and is now called the Old Spa];iish Trail. After entering Colorado it ran from near the present site of Mancos to the Dolores. On the fourteenth day from Santa Fe, we find the following entry: 'En la vanda austral del Vio [Rio] sobre un alto, huvo anti- quam (te) una Poblacion pequena, de la misma forma q" las de los Indios el Nuevo Mexico, segun manifieran las Ruinas q^ de invento registramos.' - "By tracing the trip day by day, up to that' time, it appears that the ruin referred to by these early fathers was situated somewhere near the bend of the Dolores River, or not far from the present town Dolores, Colo. The above quotation indicates that the ruin was a small settlement, and situated on a hill, on the south side of the river or trail, but it did not differ greatly from the ruined settlements of the Indians of New Mexico with which the writers were familiar, and had already described. " 1 Fewkss, J. W., The First Pueblo Huin in Colorado Mentioned in Spanish Documents. Science vol. XLTI, Sept. 14, 1917. sDiarloyDoreoterodelasnuevasdeaoubrimlentosdetlerrasalosr'bosN. N. OE. OE. del Nuevo Mexico por los E. E. P. P. Fr. Silvester Velez Escalante, Pr. Francisco Atanacio Dominguez, 1776. ( Vide Sen. Ex, Doc. 33d Congress, No. 78, pt. 3, pp. 119-127.) mwkes] prehistoric villages, castles, and towers 37 Cliff-Dwellings There are numerous cliff-houses in this district, but while, as a rule, they are much smaller than the magnificent examples in the Mesa Verde, they are built on the same architectural lines as their more pretentious relatives. Both large and small have circular subterranean kivas, similarly constructed to those of Spruce-tree House, and have mural pilasters (to support a vaulted roof, now de- stroyed), ventilators, and deflectors. There are also many rooms in cliffs, possibly used for storage or for some other unknown purposes, but too small for habitations. It is significant that these are identical so far as their size is concerned with the "ledge houses," near Spruce-tree House, indicating similar or identical uses. The kivas of cliff-dweUings of size in the region considered have the same structural features as those of adjacent ruins, but very little resemblance, save in site, to those of cliff-dwellings in southern Ari- zona, as in the Sierra Ancha or Verde Valley, the structure of which resembles adjacent pueblos. The absence in the McElmo region of very large cliff-houses is due partly but not wholly to geological conditions, the immense caves of the Mesa Verde not being duplicated in the tributaries of the McElmo; but wherever caverns do occur, as in Sand Canyon, we commonly find diminutive representatives. While differences in geological features may account for the size of these prehistoric buildings,- the nature of the site or its size is not all important.' Here and there one sees from the road through the McElmo Canyon a few small cliff-houses, and if he penetrates some of the tributaries, he finds many others. The canyon is dominated by the Ute Moun- tain on the south, but on the north are numerous eroded cliffs in which are many caves affording good opportunities for the con- struction of cliff-houses. These buildings do not differ save in size from the cliff-houses of the Mesa Verde. Their kivas resemble the vaulted variety and the masonry is identical. Although the existence of cliff-dweUings in the tributaries of the McElmo has long been known, the characteristic circular kivas which occur in the Mesa Verde had not been recognized previous to the present report. The relative age of the pueblos and great towers and the same structures in caves can not be decided by the data at hand, but the indications are that they were contemporary. On account of the similarity in structure of the McElmo chff- dwellings to those on Mesa Verde, only a few examples from the 1 Attention may bo called to the fact that often we find very commodious caves without correspondingly largo cliff-houseSj even in the Mesa Verde, 38 BUREAU OF AMERICAN ETHNOLOGY [btill. to former region are here considered. It may be worthy of note that while McElmo cM-dwellings are generally accompanied by large open-air pueblos and towers or great houses on the cliffs above, in the Mesa Verde open-air buUdings^ are generally situated some distance from the clifif-dwellings. CLIFF-DWELLINGS IN SAND CANTON Several small cliff-houses occur in Sand Canyon, one of the northern tributaries of the McElmo. Stone Arch House, here figured (pi. 6, a), so called from the eroded cliff (pi. 4, h) near by. It is situated in the cliff, about a mile from where the canyon enters the McElmo Canyon near Battle Kock. Abundant pinon trees and a few scrubby cedars grow in the low mounds of the talus below the ruin, near which, on top of a neighboring rock pinnacle, stOl stand the well-constructed walls of a small house (pi. 4, a). DOUBLE CLIFF-HOUSE The formerly unnamed cliff-house shown in plate 8^ is one of the best preserved in Sand Canyon. It consists of an upper and a lower house, the former situated far back in the cave, the latter on a pro- jecting terrace below. Unfortunately it is impossible to introduce an extended description of this building as it was not entered by the author's party, but from a distance the walls exhibit fine masonry. It is imique.in having double buildings on different levels, an arrange- ment not rare in a few examples of cliff-dweUings on the Mesa Verde. As shown in plate 8, the character of the rock on which the lower house stands is harder than that above in which the cave has been eroded. The upper house is wholly protected by the roof ^ of the cave and occupies its entire floor. The lower house shows from a distance at least two rooms, the front wall of one having fallen. From a distance the walls of both the lower and the upper house seem to be well preserved, although many of the component stones have fallen to the base of the chff. SCAFFOLD IN SAND CANTON One of the cliffs bordering Sand Canyon has an inaccessible cave in which is an artificial platform or lookout shown in plate 7, a. Although this structure is not as well preserved as the scaffold in the neighborhood of Scaffold House in Laguna (Sosi) Canyon, on the Navaho National Monument, it seems to have had a similar purpose. ^ Sun Temple, however, is a seeming exception and follows the McElmo rule of proximity; several large cim-dwelllngs occur under the cliff on which this mysterious building stands. 2 Taken from a point across the canyon, the only one from which both houses can be included in the same photograph. ' For a good example of cliff-houses at different levels, see Cliff-Dwellings in Pewkes Canyon, Mesa Verde National Park, Holmes Anniversary Volume. PEWKES] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 39 It is constructed of logs reaching from one side of the cave to the other supporting a floor of flat stones and adobe. Its elevated situation would necessitate for entrance either holes cut in the cliffs or ladders. UNIT-TYPE HOUSES IN CAVES In subsequent pages the author will describe a ruin called the Unit-type House, situated in the open on the north rim of Square Tower Canyon. A similar 'type of imit-type house is foimd in a cave in Sand Canyon. The reader's attention may first be called to the definition of a imit type, which is a buUding composed of a, circular kiva, with mural banquettes and pedestals supporting a vaulted roof, with ventilator, reflector, and generally a ceremonial opening near a. central fire hole in the floor. This kiva (fig. 5) is generally embedded in or surrounded by rectangular rooms. The single-unit type has one kiva with several surrounding rooms; the so- called pure type is composed of these units united. In an almost inaccessible cave (pi. 5, h) in Sand Canyon a few miles from the McElmo road near the scaffold already mentioned there is a cliff ruin, so far as known the first described single-unit house in a cave. It covers the whole floor of the cave (fig. 5) and its walls are considerably dilapidated, but the kiva shows this instruc- tive condition : The walls are double, one inside the other, with two sets of pedestals, the outer of which are very much blackened with smoke of constant fires; the inner fresh and im tarnished, evidently of late construction. A similar double-walled kiva known as "Kiva A" exists in Spruce-tree House, as described in the author's account of that ruin.* On the perpendicular wall of the precipice at the right hand of the ruin in the cave above mentioned are several pictographs shown in plate 7, c. 1 Antiquities of the Mesa Vorde National Park: Spruce-treo House. Bull. 41, Bur. Amer. Ethn., 1909. 40 BUREAU OP AMEMCAN ETHNOLOGY tBCLL. 7d The rectangular rooms about the kiva are in places excavated out of the cliffs, but show standing walls on the front. These were not, however, constructed with the same care as those of the kiva. The cliff-house in Hackberry Canyon (pi. 9, a) is one of the most instructive. It lies below Horseshoe House and appears to be a sec- ond example of a unit-type kiva and surrounding rooms. The cliff-dwelling in Ruin Canyon* visible across the canyon from the Old Bluff City Road is well preserved. On the rim of the canyon are piles of stone indicating a very large pueblo, with surface circu- lar depressions indicating unit-type houses. CLIFF-HOUSES IN LOST CANYON Lost Canyon, a southern tributary of the Dolores River, contains instructive cli£f-houses to which my attention was called by Mr. Gordon Parker, superintendent of the Montezuma Forest Reserve, who has kindly allowed me to use the accompanying photographs. This cliff-house (pi. 10, a, I) belongs to the true Mesa Verde type and shows comparatively good preservation of its walls, some of the beams being in place. It is most easily approached from Mancos. There are small cliff-houses in the same canyon not far from Dolo- res, but these are smaller and their walls very poorly preserved. An interesting feature of these cliff-houses in Lost Canyon is that they mark the northern horizon of cliff-dwellings of the Mesa Verde type, having kivas similarly constructed. Great Houses and Towers Great houses and towers differ from pueblos of the pure type but may often be combined with them, forming composite houses ar- ranged in clusters called villages. Castles and towers may be iso- lated structures without additional chambers, or may have many annexed rooms which are rectangular, round, or semicircular in form. Semicircular towers surrounded by concentric curved walls connected by radial partitions forming compartments are shown in Horseshoe Ruin, to which attention has been called in preceding pages, and possibly in the circular or semicircular ruins on hilltops near Dolores. MASONRY The masonry of the great house and tower type (pi. 11, a, h) varies in excellence, not only in different examples but also in dif- ferent portions of the same building. Some of the walls contain some of the best-constructed masonry north of Mexico; others (see pi. 6, i) are crudely made. In the Great House of the HoUy 1 The name Ruin Canyon, often applied also to Square Tower Canyon, is retained (or this canyon. FEWKEs] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 41 group, where the walls show superior construction, the lowest courses of rock are larger than those above, but in Horenweep Castle smaU stones are found below those of larger size; the Kound Tower in McLean Basin shows small and large stones introduced for orna- mentation. The ambitious constructors of several towers have built the foun- dations of these towers on bowlders sloping at a considerable angle, and it is a source of wonder that these walls have stood for so many years without sliding from their bases. Although so well constructed in many instances, the courses were weak from their want of binding to the remaining waU. As a consequence many corners have fallen, leaving the remaining walls intact. The builders often failed to tie in the partitions to the outer walls, by which failure they lost a brace and have sprung away from their attachment. In a general way we may recognize masonry of two varieties. 1. That in which horizontal courses are obscure or absent. This has resulted from the use of stones of different sizes, the intervals between which are filled in with masses of adobe. These stones are little fashioned, or dressed only on one side, that forming the face of the wall. 2. That constructed of horizontal courses, constituting by far the larger number of these buildings. Each course of this masonry is made of well-dressed stones, carefully pecked, and of the same size. In this horizontal masonry the thickness of stones used may vary in different courses (pi. 11, 6). They may be alternately narrow or thick, or layers of thick stones may be separated by one or more layers of tabular or thin stones. This method of alternation may be so regular as to please the eye and thus become decorative, a mode of decoration that reached a h%h development in the Chaco Ruins. The stones ia the horizontal style of masonry are equal in size through- out the whole building in some cases, and show not only care in choice of stones but also in dressing them to the same regulation size. In these cases the joints fit so accurately that chinking has not been found necessary and a minimmn use of adobe was required. The inner walls of kivas are much better constructed than the outer walls of the same or of the walls about them. The masonry here is regular horizontal. The sides, lintels, and thresholds of door- ways are among the finest examples of construction. With the exception of walls sheltered by overhanging cliffs, the plastering has completely disappeared, but there is no reason to doubt that the interiors of aU the great houses and towers were formerly plastered. It is instructive to compare the masonry of the great houses and towers of the Mancos with that of the towers in HiU Canyon (pi. 1 1 , c) in Utah, the most northern extension of these two types. In Eight Mile Ruin, one of the largest of these buildings in Hill Canyon, we have 42 BUREAU OF AMERICAN ETHNOLOGY l^^'^^- ™ a circular tower with, annexed great houses, all constructed of well- dressed stones, the masonry in the walls showing on one side of the tower. No excavations, however, have yet been undertaken in Hill Canyon Kuins, and it is not known whether the unit type of kiva is found there, but the combination of great houses and towers is evi- dent from the ground plans elsewhere published.' The feature of the towers ia Hill Canyon is the clustering into groups, somewhat recalling the condition in Cannonball Euin, where, however, they are united. In the Eight Mile Euin one of the towers is separated from the remaining houses. Several towers have accompanying circular depressions with surrounding mounds. This association can well be seen in Holmes Tower on the Mancos Canyon and in Davis Tower and one or two others on the Yellow Jacket. These depressions, sometimes called reservoirs, have never been excavatedj- but from what is known of rooms accompanying towers in the western section of Hovenweep Castle it may be that they indicate kivas. Some towers have no sunken area in the immediate vicinity, especially those mounted on rocky points or perched on bowlders. At Cannonball Ruin there are several kivas side by side in one section and towering above them is a massive walled tower and other rooms. STHUCTUEE OF TOWERS None of the towers examined have evidences of mural pilasters to support a roof or recesses in the walls as in vaulted-roofed kivas. They are sometimes two stories high, the rafters and flooring resting on ledges of the inner wall. Lateral entrances are common and windows are absent.^ While the author has found no ruin of the same ground plan as Sun Temple on the Mesa Verde, D-shaped towers or great houses from several localities distantly recall this mysterious building, and there may be an identity in use between Sun Temple and the massive- walled structures of the.McElmo and Yellow Jacket; what that use was has not thus far been determined.' If they were constructed for observatories we can not account for the square tower in the South Fork of Square Tower Canyon, from which one can not even look down the canyon, much less in other directions, hemmed in as it is 1 Smithson. Misc. Colls., vol. 68, no. 1, 1917. 2 Our knowledge of the entrances into kivas of the vaulted-roofed type is not all that could be desired. Kiva D of Spruce-tree House has, a passageway opening through the floor of an adjacent room, and Kiva A of Cliff Palace has the same feature. Doctor Pruddcn has found lateral entrances from kivas into adjoin- ing rooms in his unit-type pueblo. The majority of cliff-dwellers' kivas show no evidence of lateral en- trances, 3 Mr. Jackson, op. cit., p. 415, regarded it likely that the towers were "lookouts or places of refuge for the sheep herders who brought their sheep or goats up here to graze, just as the Navajos used to and as the Utes do at the present time." This explanation is impossible, for there is no evidence that the builders of the towers had either sheep or goats, the Navajos and the Utos obtaining both from the Spaniards. PEWKES] PREHISTOEIC VILLAGES, CASTLES, AND TOWERS 43 by cliffs. Isolated towers are often too small for defense; and they sliow no signs of habitation. Are they granaries for storage of corn or places for rites and cere- monies? Do they combine several functions — observation, defense, and storage of food? Thus far in studies of more than 30 towers and great houses not one has been found so well preserved that enough remains to determine its use, and yet their walls are among the best in all southwestern ruins. Some future archeologist may find objects in towers that will demonstrate their function, but from our present knowledge no theory of their use yet suggested is satisfactory. It is impossible from the data available to determine the century in which the towers and great houses of the region were constructed. Thus far a few were seen with great trees growing in them, but none with roofs; the state of preservation of the walls does not point to a great age. Several writers have regarded them as occupied subse- quently to the Spanish conquest, while others have ascribed to them a very remote antiquity. It can hardly be questioned that the cliff- dweUers, and by inference their kindred, the tower builders, were superior in their arts to modern Pueblos. It is important to determine first of all the forms of these towers; whether their ground plans are circular, oval, square, rectangular, or semicircular. The northern wall of many is uniformly curved and the last to fall, which might lead to the belief that the southern side, generally straight, was poorly made, but one can not determine that by direct observation, since the latter has fallen. As a matter of fact the south wall was generally low and straight, over 50 per cent of the "round" towers being semicircular, D-shaped, or some modifica- tion of that form ; but we also have square and rectangular towers. It is also important to determine whether these had single or multiple chambers and the arrangement of the rooms in relation to them. This is especially desirable in towers with concentric compartments. It is also instructive to know more of the association of towers with pueblos and cliff-dwellings or to analyze component archi- tectural features. The tower type often occurs without appended rooms. At Cliff Palace and Square Tower House it is united with a pueblo village under cliffs; in Mud Spring Kuin it has a like relation to rooms of a pueblo in the open. Has its function changed by that union? What use did the tower serve when isolated and had it the same use when united with other kinds of rooms in cliff-dwellings and pueblos ? No writer on the prehistoric towers of Colorado and Utah has emphasized the fact that a large number of these buildings are semicircular or D-shaped, but it has been taken for granted that the fallen wall on the south side was cxured, rendering the tower circular 44 BTJREAT; of AMEMOAN ethnology [bull, to or oval.^ In most cases this wall was the straight side of a D-shaped tower. Doctor Prudden, who first recognized the importance of a union of towers with other types of architecture in the McElmo district, says: '' "Towers of various forms and heights occasion- ally form a part of composite ruins of various tjrpes." He says also: "Several of the houses are modified by the introduction of a round tower." And again: "At the head of a short canyon north of the Alkali, which I have called Jackson Canyon . . . each building consists of an irregular mass of rooms about 200 feet long, with low towers among them." As our studies are morphological, dealing with forms rather than sites of towers, Httle attention need be paid to their situation on bowlders, in cliffs, or at the bottoms of canyons. The majority of the castellated ruins considered in the following pages are in the proposed Hovenweep National Monument, but there are others in the miain Yellow Jacket and its other tributaries. HOVENWEEP DISTEICT The name Hovenweep ("Deserted Valley") is an old one in the nomenclature of the canyons of southwestern Colorado and formerly (1877) was apphed to the canyon now called the Yellow -Jacket, but at present is limited to one of the tributaries. The name is here used to designate an area situated jtist over the Colorado State line, in Utah, part of which it is hoped will later be reserved from the pubhc domain and made a monument to be called Hovenweep National Monument. The ruined castles and towers in this district are marvelously well preserved, consideriag their age and imperfect masonry. We can de- termine their original appearance with no difficulty and use them in reconstructing the possible forms of more dilapidated ruios, now piles of debris. The best castles and towers known to the authofare local- ized in three canyons: (1) Square Tower Canyon, (2) HoUy Canyon, (3) Hackberry Canyon. There are, of course, other castles and towers ia the Yellow Jacket-McElmo region, but there is no locality where so many different forms appear in equal numbers in a small area. Rttin Canyon The Old Bluff Boad from Dolores diverges southward from that to MonticeUo at Sandstone post office and passes a pile of rocks visible from the road on the Ruin Canyon long before it reaches Square Tower Canyon (fig. 6) . This large ruin is situated on the east rim and under it in the side of the cliff are fairly well-preserved cliff-houses. 1 The tower figured by Prudden (Amer. Anthrop., n. s. vol. y, no. 2, pi. xviii, fig. 2) as a "round tower" is really somidrcular, as shown in the ground plan (flg. 14) hero published. 2 Ibid., pp. 241, 263, 273. FEWKBs] PKEHISTOEIG VILLAGES, CASTLES, AND TOWERS 45 Other ruios with high standing walls were reported in Euin Canyon but were not visited. The duplication of names of canyons in this district is misleading. Names like Ruin Canyon are naturally applied to canyons in which there are ruins. When the author learned at Dolores of Ruin Canyon, he supposed it was a tributary of the Yellow Jacket or McEhno, but while the canyon known to cowboys at Dolores by this name has large rums on its rim, it is not the "Rum Canyon" to which attention is now directed. The duplication of names has led me to retain the name Ruin Canyon for one and to suggest the name Square Tower Canyon for the other. After leaving Ruin Canyon the Old Bluff Road takes a southerly course, passing through the cedars until a sagebrush clearing replaces the "timber," where it crosses two well-preserved Indian reservoirs, or bare surfaces of rock, dipping south, the southern border having as 8 Twin Towers squahe rotvff! ca/vvo/^ Fig. 6. — Square Tower Canyon. a retaining wall a low ridge of earth to hold back the water. The retaining wall of the second reservoir has been built up by stockmen and, when the author was there, contained considerable water. Cross- ing the second reservoir a trail turns east or to the left and follows the road to Keeley Camp, near which are the "Keeley Towers." At present an automobile can approach within a mile of these ruins. Square Towek Canton To reach the Square Tower Canyon (pis. 11-17) one returns to the reservoir on the Bluff Road and continues east about 3 miles farther, where a signboard on the left hand indicates the turn off to Square Tower Canyon. Followiog the new direction about southeast the great buildings are visible a mile away. An automobile can go to the very head of this canyon and a camp can be made within a few feet of Hovenweep House. If the visitor approaches Square Tower Canyon from the McElmo, he passes throTogh Wickyup Canyon, where there are two towers on the summits of elevated buttes, not far from the junction of the canyon and the Yellow Jacket. 46 BUREAU OF AMEKICAN ETHNOLOGY IBULIi. ra The castles and towers in Square Tower Canyon hare been known for many years and haye been repeatedly photographed.^ Several descriptions of these ruins have been printed, but no satis- factory studies of their structure have been published. They are recognized as prehistoric and are generally thought to have been inhabited contemporaneously with the chfE-dwellers of the Mesa Verde, being built in the same style of architecture. Classification of Buins in Squabe Tower Canton The ruins in Square Tower Canyon are classified for convenience in description as follows: (1) Ruins which have indications of inclosed circular kivas, with mural pilasters and banquettes, and closely approximated surround- ing rooms. To this class belong ruins 1, 2, and 10. Of these, Unit- type Ruin (No. 10) has only one kiva and belongs to the simplest or unit form of the pm-e type. Ruins 1 and 2 have two or more kivas and are form,ed by a imion of several units, combined with towers and great houses. (2) Ruins, the main feature of which is absence of a cir- cular kiva. The Twin Towers be- long to this second or " great house " type. The few cliff-dweUiags in this canyon are small, generally without kivas, resembliug storage cists rather than domiciles. HovENWEBP House (Ruin 1) This ruin (fig. 7), the largest in the canyon, is situated at the head of the South Fork. Although many of its walls have fallen, there stiU remains a semicircular great house (B, 0, D) with high walls con- spicuous for some distance. The ruin is a pueblo of rectangular form belonging to the pure type, showing circular depressions iden- tified as kivas (E), embedded in collections of square and rectangular rooms, and massive walled buildings (E) on the south side. The standing walls of the ruin are remains of a conspicuous D-shaped tower (B, 0, D), which is multichambered. Its straight wall measures 23 feet, the curved wall 56 feet, and its highest wall, which is on the northeast corner, is 15 feet high. At the northwest angle of the ruin (A) there stand remains of high walls which indi- riG. 7.— Ground plan of Hovenweep House. 1 Among tho older photograplis seen by the aattior are those of W. H. Jackson, prints of which are on exhibition in the State Historical Museum at Denver, Colo. FEWKEs] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 47 cate that corner of a rectangular pueblo. Hovenweep House (pi. 14, a) was the largest building in this canyon, but with the exception of the addition of a semicircular tower or great house, does not differ greatly from a pueblo like Far View House on the Mesa Verde. The piles of stone and earth indicating rooms below justify the conjec- ture that when the fallen debris is removed the unfallen walls will stiU rise several feet above their rocky foundations. If properly excavated, Hovenweep House would be an instructive building, but in its present condition, while very picturesque, its structure is difficult to determine. Fig. 8. — Ground plan of Hovenweep Castle. Hovenweep Castle This ruin (pis. 14, h, c; 18, 6), like the preceding,^ has circular kivas compactly embedded in rectangular rooms arranged about them, indi- cating the piu-e type of pueblos. The massive-walled semicircular towers and great houses are combined with square rooms and kivas, indicating that it is distiaguished by two sections, an eastern and a western, which, united, impart to the whole the shape of a letter L (fig. 8). WESTERN SECTION OF HOVENWEEP CASTLE The western section (fig. 8, A-D, M) of Hovenweep Castle is made up of five rooms, the most western of which, M, is semicircular, while A, B, C, and D are rectangular. Room A is almost square, one of its walls forming the straight wall of the south side of the semicircular 48 BUREAU OF AMEEICAN ETHNOLOGY [buli.. 70 tower, M. At the union its walls are not tied into the masonry of the circular wall of the tower, as naay be seen in the illustration, plate 14, 6, implying that it was constructed later. There is an entrance into A from the south or cliff side, and a passageway from A to Room B, which latter opens by a doorway into Room 0. All rectangular rooms of the western section communicate with each other, but none except A seem to have had an external entrance. The photograph of the south wall of the west section of the ruin (pi. 14, c) shows small portholes in the second story and narrow slits in the tower walls. The lower courses of masonry are formed of tlrnmer stones than the rows above, but smaller stones compose the courses at the top of the wall. A view of the north wall of the western section (pi. 22, a) shows the tower and rooms united to it. There is no kiva in the western section. EASTERN SECTION OF HOVENWEEP CASTLE The longest dimension of the western section (pis. 12, 14, c) is ■approximately east-west; that of the eastern is nearly north-south. 'The eastern section (fig. 8, E-L), like the western, has a tower (i), which is situated between two circular depressions or kivas {K). On the north and south ends the eastern section is flanked by rectangular rooms. Those at the north end were better constructed, and even now stand as high as the walls of the western tower. The views show that their corners are not as well preserved as their faces, which is due to defects in masonry, as lack of bonding. Although much d6bris has accumulated around the kivas, especially in their cavities, it is evident that these ceremonial rooms were formerly one storied, and practically subterranean on accoimt of the surrounding rooms. Several fragments of walls projecting above the acciunulated debris indicate rooms at the junction of the eastern and western sections of the ruin, but their form and arrangement at that point are not evident and can be determined only by excavation. The inner kiva walls show evidences of mural pilasters and banquettes like those of cliff dwellings and other pure pueblo types. Ruin 3 The square tower (pi. 11, a), standing on a large angular rock in the canyon below Hovenweep Castle, is a remarkable example of prehistoric masonry so situated that it is shut in by cliffs, rendering the outlook limited. Several published photographs of this tower give the impression that it stands in the open and was an outlook, but that this is hardly the case will be seen from a general view looking west up the South Fork. BEWKES] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 49 Ruin 4 This ruin is a small tower situated in. a commanding position on the point of the mesa where the canyon forks. The section of the wall still standing indicates a circular form, the north side of which has fallen; the part still intact, or that on the south side, exhibits good masonry about 8 feet high (pi. 15, c). Ruin 5 The walls of the north segment of a tower stand on a large angular block of stone rising from a ledge above the arroyo, or bed of the canyon, below Kuin 4, on the South Fork. What appears to have been a doorway opens on its north side ; this opening is defended by a wall, remains of a former protected passageway into the tower. On the perpendicular cliff of the precipice near Ruin 5 and below the point on which Ruin 4 stands there are several almost illegible pictographs, below which are rather obscure evidences of a building, the features of which oan be determined only by excavation. Instructive features of Tower No. 5 are two parallel walls, one on each side of the doorway, hke those of the circular towers on the promontory at the junction of the Yellow Jacket and McElmo. Other towers on the canyon rim show defensive walls, as in Ruin 9, con- structed about their entrances from corners of the buildings to the mesa rim, effectually preventing passage. Morley and Kidder have suggested that the walled recess in the cliff below Ruin 9 was probably "built to prevent access from below. This feature is found in the floor entrances of a building near the Great House of the Holly group. Ruin 6 Tliis ruin is a small tower whose curved walls are so broken down that the form is not evident. It is situated in the base of the talus at the head of the South Fork (pi. 26, a). Eboded Bowldee House (Ruin 7) This house, more remarkable from its site than its structure, was constructed in an eroded cave of a bowlder halfway down the talus of the cliff. The front walls are somewhat broken down, but others built in the rear of the cave stUl remain intact. On the top of the bowlder is the debris of fallen walls, suggesting a former tower, but not much remains in place to determine its outlines. Where the walls are protected the mortar shows impressions of hiunan hands and at one place there are the indentations of a corncob used by the plasterers to press the mortar between the layers of stone. There were formerly at least two rooms in the rear of the cave, the front walls of which have fallen and are strewn down the talus to the bottom of the canyon. 108852°— 19— Bull. 70 il 50 BUREAU OF AMEEICAN ETHNOLOGY [BULL. 70 Fig. 9.— Ground plan oi Twin Towers. Twin Towers (Ruin 8) The so-called Twin Towers, which seen together from certain points appear as one ruin (pi. 15, a, I), rank among the most impressive buildings in Square Tower Canyon. They stand on the south side of the canyon on a rock isolated by a cleft from the adjoining cliff. The larger (fig. 9, A-E) has an oval ground plan and a doorway in the southwest corner; the smaller (F, G, E, 7) is horseshoe shaped with a doorway in the east wall, which is straight. The arrangement of rooms is seen in figure 9. Small walled-up caves are found below the foundation on the northwest base of the larger room. Ruin The ground plan of this ruin is rectangular in form, 19 feet 6 inches long by 10 feet wide. The standing walls measure 11 feet in altitude. It is situated on the south rim at the mouth of the South Fork, just above Kuin 7, a few feet back from' the cliff. A doorway opening in the middle of its north wall was formerly made difficult of entrance by walls, now fallen, extending from the northeast and northwest angles to the edge of the cliff. The masonry throughout is rough; projecting ends of rafters indicate a building two stories high. There are peepholes with plastered surfaces through the southeast and west walls, which suggest ports. A short distance east of the building is a circle of stones reminding the author of a shrine. TJnit-Typb House (Ruin 10) This pueblo (pi. 19, c), from a comparative point of view, is one of the most interesting ruins in the Hovenweep, and is situated on the very edge of the canyon on the North Fork not far from where it begins. It is the simplest form of prehistoric pueblo, or the unit ^ of a pure type, made up of a centrally placed circular ceremonial room (fig. 10, K) embedded in- rectangular rooms, six in number {A-F). The resulting or external form is rectangular, oriented about due north and south; the southern side, which formerly rose from the edge of the canyon, being much broken down and its masonry precipitated over the cliff. The central kiva (fig. 10) is made of exceptionally fine masonry and shpws by what remains that it had mural banquettes, and pilasters to support the roof, with othjer features like a typical kiva of the 1 The "unit type" was first recognized by Doctor Prudden in his illuminating studies of the pueblos ot the San Juan Basin. The author was the first to point out its existence in cliS-houses of the same area. FEWKES] PEBHISTOKIC VILLAGES, CASTLES, AND TOWEES 51 Mesa Verde cliff-houses. A side entrance opens in one corner into a small room (fig. 10, G) in which ceremonial objects may have been formerly stored (pi. 32, 6) . The kiva of Unit-type House is architecturally the same as those ■with vaulted roofs at Spruce-tree House, Chff Palace, and Far View House on the Mesa Verde. A similar structure, according to Prudden,^ occurs at Mitchell Spring Ruin in the Montezuma Valley, and near the Picket corral. The same type was found by Morley^ at the Cannonball Euin and by Kidder ^ in a kiva on Montezuma Creek in Utah, where clusters of mounds would appear to be composed of Fig, 10. — Ground plan ol Unit-type House. single or composite ruins of this type. This smaJl pueblo was prob- ably inhabited by one social unit, and may be regarded as the first stage of a compound pueblo. Stronghold House (Rtjin 11) Ruin 1 1 is composed of a cluster of several smaU, buildings, one of which is situated on the north edge of the mesa somewhat east of Ruin 10 (pi. 25, 1) ; another, called by Morley and Kidder Gibraltar House, formerly of considerable size, was buUt on the sloping surface of an angular bowlder (pi. 17, 21, 6). Although many walls have fallen, enough remains to render it a picturesque ruin, attractive to the visitor - and instructive to the archeologist, by whom it has been classed as a tower. This building from the east appears to be a square tower, but it is in reality composed of several rooms perched on an inaccessible rock. • Circular Kivas in San Juan Watershed. Amer. Antlirop., n. s. vol. 16, no. 1, 1914. 2 Excavation of the Cannonball Euins in southwestern Colorado. Amer. Anthrop., n. s. vol. x, no. 4, 1908. ' Explorations in southeastern Utah. Amer. Joum. Archseol., 2d ser., vol. xiv, no. 3, 1910. 52 BtTREAtr OF AMEKICAW ETHNOLOGY Ruins in Holly Canyon [BDLL. '0 The towers in Holly Canyon (fig. 11) are in about the same con- dition of preservation as those in Square Tower Canyon. They cluster about the head of a small canyon (pi. 18, a) and may be approached on foot along the mesa above Keeley Camp, about a mile distant. Two of the Holly ruias belong to the tower type and were built on fallen bowlders. One of these has two rooms on the ground floor. (Pis. 19, a, I; 20 a, c.) Tig. U.— Holly Canyon Kuins. RUIN A, GKEAT HOUSE, HACKBEKBY CASTLE Ruin A (pi. 21, a), the largest building of the group, which stands on the edge of the canyon, is rectangular in form, measuring 31 by 9 feet, and is 20 feet high (fig. 11, A). Evidences of two rooms appear on the ground plan, one of which is 14 feet long, the other 12 feet inside measurement. The partition separatiag the two rooms is not tied into the outer walls, an almost constant feature in ancient masonry. The ends of the rafters are still seen in the wall at a level 12 feet above the base. Fallen stones have accumulated in the rooms to a considerable depth, and the tops of the remaining wall, where the mortar is washed out, will tumble in a short time. Ruin B (pi. 20, i), situated a short distance north of Riun A, also stands on the canyon rim. The north wall is entire, but the south wall has fallen. What remains indicates that the ruin was about square, with corners on the north side rounded, imparting to it a semicircular form. The entrance into this room may have been through the floor. TOWERS [O AND D] These towers (pi. 23, a, 6) show some of the finest masonry known in this region, being constructed on fallen bowlders which their PEWKEs] PEEHISTOEIC VILLAGES, CASTLES, AND TOWBKS 53 foundations almost completely cover. HoUy Tower (pi. 23, h) measures 16 feet bigli and 21 feet in diameter. It is 7 feet wide, its top rising to a height level with that of the mesa on which stand build- ings already considered. One of the two rooms of this tower is narrower and wider than the other, shown in an offset as if con- structed at a different time. Its foundations are 17 feet long by 8 feet wide, the highest wall measuring, at the southeast corner, 12 feet 8 inches. There is a fine doorway, wide above and narrow below, in the north wall. The approach at present is difficult on account of the height of the rock on which it stands, but there are evidences of former footholes. HOLLY HOUSE Several broken-down walls, some of which are over G feet high, situated east of Ruin A, appear to belong to a pueblo of considerable size (fig. 11, E, F), but the large foundation rock on which it is situated has settled, its top having separated from the edge of the canyon, so that the corner of the building (F) is out of plumb. The walls on the adjoiniag cliff are also much broken down, although several sections of them rise a few feet above the general surface. The cause of this change in level of the base may have been an earth- quake or the settling or shding of the bowlder on the talus down the hill. The united buUding appears to have been a pueblo of rectan- gxilar form. Its walls are so broken down that it was not possible to determine its exact dimensions. Ruins in Hackbeery Canyon HORSESHOE HOUSE Horse Shoe House B\5 The large building in Hackberry Canyon, one of the terminal spurs of Bridge Canyon, a mile northeast of the cluster in Holly Canyon, is particularly instructive from the fact that surrounding the remains of a circular tower, for two-thirds of its circumference, is a concentric wall with compartments separated by radial partitions (fig. 12, 1). Horseshoe House (pi. 23, c) stands on the north edge of the canyon (fig. 12, 1), having its straight wall on the south side, as is usually the case, the well-preserved north side being curved. The northeastern corner still stands several feet high. The southeastern corner formerly rested on a projecting rock, which recalls the cornerstone of Sim Temple. The masonry of most of the southern segment of the enclosed circular room or tower has fallen down the cliff. Fig. 12.— Horseslioe (Hackberry) Canyon. 54 BUREAU OF AMEEICAN ETHNOLOGY [bull. There does not appear to have been a doorway on the south side, and there is not space for rooms on this side on account of the nearness to the edge of the chil. While the form (fig. 13) of Horseshoe Euin re- calls that of Sun Temple, in details of room structure it is widely divergent. The length of the south wall, or that connecting the two ends of the horseshoe, is 30 feet, its width 27 feet; the highest wall on the northwest side is 12 feet. Figure 13 shows the arrangement of the rooms and the mutilation of the south wall of the ruin. The dis- tance between the outer and inner concentric walls averages 4 feet; the circular room is 17 feet in diameter. In the same cluster as Horseshoe Ruin (pi. 24, a) there is another well-made tower (fig. 12, 4), constructed on a point at the entrance to the canyon, and below it in a cave are well-preserved walls of a cliff- dwelling. Fig. 13.— Ground plan o£ Horseshoe House. A short distance due north of Horseshoe House, at the head of a small canyon, a tributary of Bridge Canyon, there are two large pueblos and a round tower. The pueblos are mentioned by Prudden, who gives a ground plan which indicates an extensive settlement. TOWERS IN THE MAIN YELLOW JACKET CANYON Of the several towers and great houses of the main Yellow Jacket Canyon two may suffice to show their resemblance to those in Square Tower Canyon. The two towers considered belong to the D-shaped variety, the straight wall, as is almost always the case, being on the south side. FEWKES] PREHISTORIC VILLAGES, CASTLES, AND TOWERS 55 Davis Tower Mr. C. K. Davis, who lives not far from the Yellow Jacket Spring, conducted the author to a tower of semicircular ground plan (fig. 14) near his ranch. This ruin (pi. 26, h), is situated on a rocky ridge on top of the talus halfway down h to the bottom of the canyon, on * its right side. Lion (Littiiell) Towee ' This tower (pi. 29, I) is built on a bowlder situated in Yellow Jacket Canyon a mile from Mr. Littrell's ranch and about 5 miles south of the Yellow Jacket post office; approximately 20 miles from Dolores, Colorado. Its ground plan (fig. 15) is D-shaped, the lower story being divided by ^"'- "-Q™™-! p'^"^ "^ ^ayis Eum. partitions into four rooms. The wall of the middle room seems to be double, or to have been reenforced. It measures 40 feet on the straight side, the highest wall being about 25 feet above the base. The foundations rest on the irregular smiace of a bowlder to which it conforms. M LEAN BASIN McLean Basin is 3 miles from the Old Bluff City Road near Picket corral, 32 miles from Dolores. It has been a favorite wintering place for stock and is well known to herdsmen. One can approach the ruin from the road to Bluff City and the towers here referred to are easily reached by a trail down the mesa to the high- est terrace. There are said to be several ruins in the McLean Basin, the two tow- ers (pis. 26, c, 27, 28, a, I) visited being placed in an exceptional position in Fia.is-Ground plan of Lion House. reference to surrounding rooms. One of these towers is circular, the other D-shaped or semi- circular in ground plan (fig. 16, A, B). Previously to the author's study of the southwestern towers two forms of these structures were recognized; the square or rectangular, and the circular or oval. It is now known that several of the towers 1 This tower Is reputed to bo the home of a mountain lion, henco the name Lion House. 56 BUEEAtr OF AMERICAN ETHNOLOGY [boll. 10 previously described as circular are ra reality D-shaped, and this form is probably more common than the circular. The rectangular bmlding in the McLean Basin has a circular tower dpi. 28, I) on the southwest angle and a D-shaped tower (pi. 28, a) on the northeast. They resemble two turrets rising above the remaining walls that form the sides of the rectangles. These towers average about 12 feet high, and are well constructed, while low connecting walls of coarse masonry rise slightly above the surface. It would appear from the amount of debris that the remaining walls indicate a row of buildings, one story high, with circular subterranean kivas, but this can not be accm-ately determined without excavation of the ruin. Outside of the rectangle, however, there are at least Fig. 16. — C round plan of rain with towers in McLean Casiu. • two circular areas, possibly kiva pits. The rectangular building measTU-es about 50 feet square. The ground on which the buildings formerly stood slopes to the south, and back of it on the north rises a low perpendicular bluff which effectually shelters it in that direction. The union of a circular and a semicircular tower with a rectangular ruin is a feature not common in the McElmo- Yellow Jacket region, but appears in Hovenweep Castle, elsewhere described. Lower down the sides of the basin and near by are many indications of walls of buildings. The pottery in the neighborhood belongs to the same black and white types commonly found in the Hovenweep and Mesa Verde areas. Except for their peculiar relation to the rectangular building the McLean towers do not differ essentially from others, which fEWKEs] PREHISTORIC VILLAGES, CASTLES, ASTD TOWERS 57 leads to the inference that they were used contemporaneously and for the same purpose. There is a well-made doorway (fig. 17) in the Roimd Tower. TOWER IN SAND CANYON Sand Canyon, which opens iato McElmo Canyon near Battle Eock, has several types of prehistoric ruins, viz, towers, clifif-houses, and large rim-rock pueblos. The tower type of architecture represented by the example here figured (pi. 5, a) is isolated from other forms of bmldings. This tower is figured by Doctor Prudden, who mentions another in the neighborhood which the author did not visit. TOWERS IN EOAD (WICKTUP) CANYON The nomenclature of the northern canyons of the McElmo has considerably changed in the last 40 years. What we now call the YeUow Jacket was formerly known through its entire course as the Hovenweep. A small canyon opening near its mouth, now known as Road Canyon, was formerly called the Wickyup. The Old Bluff City Road from Dolores, Colorado, to Bluff City, Utah, divides into two branches a short distance before it descends into the McElmo, its left branchpassing throughRoad Canyon, the right bank of which follows the YeUow Jacket, which the traveler fords a short distance above its jimc- tion with the McElmo. Wickyup Canyon may be called picturesque, its cliffs being worn into fantastic shapes by water and sand. It has im- portant antiquities, among the most striking of which are two towers (pi. 24, I), crowning the tops of low buttes or hills. The waJls of these towers are well constructed, one being a simple structvu-e with a single room, the other having appended rectangular rooms ex- tending toward the northwest, some distance along a ridge of rocks. An examination of these two towers, which are about one-quarter of a mile apart, shows that they belong to the same type as the simple forms of those above mentioned, and as the entrance to Square Tower Canyon is not far away, they probably be- long to the same series. The first of the towers, called "Bowlder Castle," is situated a few himdred feet east of the road, from which it. is easily seen. This ruin is rectangular in shape and rises from a basal mass of debris indicating broken-down waUs of rooms. At a level with c^ig^ Fig. 17.— Doorway in Round Tower, McLean Basin. 58 BUEEAU OF AMEEICAN ETHNOLOGY [BnLL. 70 the top of this debris on its southern side stands a well-constructed tower with well-made doorway, the threshold and lintel of which are smooth stones, whose edges project slightly from the surface of the waU. One remarkable feature of this tower is that the doorway has been walled up with rude secondary masonry (pi. 25, a). The south wall of this building has tumbled over, as is usually the case, but the north wall rises several feet above the base. The masonry of the second tower is also broken down on the south side, but the standing remains of the north wall, which is circular, are over 10 feet high. The indications are that the ground plan of this building was oval in shape and that it inclined inward sHghtly from founda- tion to apex. Scattered over the surface are the remnants of fallen walls, and near it there is a well-marked depression, not tmlike those found in unit-type mounds, indicating kivas. TOWEES OF THE MANGOS The author's examination of the towers in the region considered embraced likewise a few in the Mancos Canyon and valley. In all essential features the Mancos towers resemble those of Mesa Verde, the McElmo, and the Yellow Jacket Canyons, and were evidently built by the same people who constructed the towers on Nava,ho Canyon and elsewhere on the Mesa Verde National Park. A brief reference to two or three of these Mancos Kiver towers may suffice to poiat out their general structure. Holmes Tower One of the towers figured by Holmes in 1877 is still among the best preserved in this region and can be visited by following up the Mancos Canyon from the west about 10 miles from where the Cortez road crosses the Mancos Uiver before going on to Ship Rock. There is at this point a bridge and near the crossing an industrial farm of the Ute Reservation where accommodations were . obtained. The Mancos Valley widens after leaving the canyon, the southern side of Mesa Verde appearing as a series of high mesas separated by canyons. In the neighborhood of the western end of Mesa Verde are lofty buttes, one called Chimney Rock, another the Ute Woman. This valley and the canyons extending into the Mesa Verde contain numerous piles of stone indicative of buildings of rectangular shape with numerous circular depressions. No cluster of mounds like those in Montezuma VaUey was seen, but about 40 sites of buildings were distributed at intervals. None of these have standing walls above ground. Following up the Mancos Canyon in a wagon about 9 miles an arroyo was encountered and from there horses were taken and the river crossed to its south bank, above which, on the shelving terrace, is the Holmes Tower, visible many miles down the canyon. This PEWKEs] PREHISTOEIC VILLAGES, CSSTLES, AND TOWEES 59 tower (pi. 29, a) is in much the same condition as when sketched by Holmes over 40 years ago. It is circular in form, about 10 feet in diameter, and about 16 feet high, with a broken window on the north side. The sky line is irregular. It is one of the best preserved towers, but not as high or as well constructed as some of the Hoven- weep specimens. Accompanyiag this tower on the north there are mounds indicative of rooms and two circular saucer-like" depressions. Excavations revealing a few human bones, including a well-worn human skull, have been made in a burial place southeast of the tower, where the surface is covered with fragments of pottery. Except in size Holmes Tower do.es not differ from others already described, but, like them, is connected with rectangular rooms. Farther up the Mancos Canyon there are other towers, one of which, Great Tower, is mentioned by Holmes in his report. On the way up the canyon, perhaps two-thirds of the distance from the bridge to the Holmes Tower, midway ia the alluvial plain and on the right bank of Mancos Creek, stands a circular ruin which conforms to Hohnes's description of Great Tower but is too poorly preserved to be positively identified. All that now remains of this building is a large pile of rocks with a central depression, but no signs of radiating partitions, although such may have existed when it was constructed and for many years after it began to fall into ruin. TOWBKS ON THE MaNCOS RiVER BeLOW THE BkIDGE There are two towers sitaated on the south side of the Mancos below the bridge on the Ship-rock Road, one about 6, the other 7 miles distant. The walls of the first of these (pi. 30, 6) are visible for some distance and are about 6 feet high, evidently very much broken down on the south and east sides. Its shape is round and there is a pile of stones indicating rooms on the east side separated from the tower by a depression. It would be a valuable contribution to our knowl- edge of these ruins if some one would determine the nature of these pits, which can hardly be regarded as reservoirs, but suggest kivas. The tower (pi. 31, a) situated farther down the Mancos River has a more commanding position than Tower A and is conspicuous because it stands on a projecting precipice, below the rim of which are walled- up artificial caves. These caves have apparently never been entered by white men; the walls of masonry are unbroken and there arc square openings, windows or doorways, which can be made out long before reaching the place. 60 BUREAU OF AMBKICAN ETHNOLOGY [snLL. 70 This tower (pi. 30, a) is almost perfectly round, about 10 feet in diameter, and stands at least 6 feet high. The south wall has fallen. In the pile of rocks on that side may be readily seen the top of a straight wall reaching to the edge of the cliff , as if for protection, but no other fallen walls may now be seen in the neighborhood. The face of the cliff below this tower (pis. 7, S; 31, &) is almost perpen- dicular, the component strata of soft shale alternating with harder rocksj the foriner well fitted for artificial excavations. The author was not impressed with the idea that any considerable number of troglodytic inhabitants dwelt in the small cliff rooms (pi. 31, 6)^ dug in it. Farther on there are other caves the walls of the entrance to which are still in sight. It is true the surface of the cliff may have been eroded and fallen ia the time since they were abandoned. They appeared to be storage cists rather than inhabited rooms. Along the vaJley by the side of the road down the Mancos from the bridge to the ruins many heaps of stone were noticed in the valley but none of these were extensive or had walls standiag above ground. Nor were they arranged in clusters as is common in the Montezuma Valley. On top of these heaps were foxind large fragments of slag in which was embedded charred corn, indicating a great fixe. StmUar slag also with burnt com has often been found by the author on the floor of excavated rooms. Megalithic and Slab House Euins at McElmo Bluff The ruined walls, on the bluff situated at the junction of the McElmo and Yellow Jacket Canyons are archeologicaUy instructive. As the mesa between the two canyons narrows in a promontory, about 100 feet in altitude, its configuration reminds one of the East Mesa of the Hopi. It is inaccessible on three sides, but on the fourth, where the width of the mesa is contracted, there are remains of a low zigzag wall, extending from one side to the other. At the western base of this promontory, on the ledge higher than the river, there are artificial walls built on bowlders in the sides of which shallow caves are eroded and near by them circular depressions. There are likewise remains of a small pueblo with walls much broken down and across the river the ruins of a community house, one of the largest in the district. The exceptional character of the ruins on top of this promontory has been mentioned or described by several visitors, as Holmes, Jackson, and Morley and Kidder, and various conjectTires have been made as to their character and^-'^^:'"^ \-^, .--\.> a, STONE ARCH HOUSE, SAND CANYON b, CLIFF-HOUSE, SHOWING BROKEN CORNER (Photographs by Jacob Wirsula) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 7 a, SCAFFOLD IN SAND CANYON b, STORAGE CIST IN MANCOS c, PICTOGRAPHS NEAR UNIT-TYPE HOUSE IN VALLEY CAVE (Photographs by T. G. Lemmon) 2 O >- z < < i C3 ,- »^ LU ^ BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 9 a, CLIFF-DWELLING UNDER HORSESHOE RUIN b, CLIFF-DWELLING, RUIN CANYON (Photographs by Jacoh Wirsula) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 10 a, KIVA OF CLIFF RUIN, LOST CANYON ^^MBB^Wip^sy*'( ^^ 5^^Z^^rC^?j^^^illM K MaK.*jgq^M & ^^M HMp»a ^^H Krr.\ "2SnK^ . . fcjw'" fc^^i^a^MlMl^BI IfcaHlV".^ ,,I^HL ^^-^ ^ J^HRf^*^ R * cv ■•*» gl^T iHliPflll^H 1 '^fe'wfc hHH ^^H ^T^^'^l^gl HriHIIH b, CLIFF RUIN, LOST CANYON (Photographs by Gordon Parker) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 11 a, SQUARE TOWER IN SQUARE TOWER CANYON b TOWER IN McLEAN BASIN c, RUIN IN HILL CANYON, UTAH (Photographs by T. G. Lemmon) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 14 a, HOVENWEEP HOUSE AND HOVENWEEP CASTLE, FROM THE SOUTH b, HOVENWEEP CASTLE, FROM THE WEST .^^'^■■ ^m- m'l ir:f^f. '^m.'-r-.t^M^ i.P \- ■ -* ' ■'*»■■ v.. V c, HOVENWEEP CASTLE, FROM THE SOUTH (Photographs by Jacob Wirsula) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 15 a, WEST END OF TWIN TOWER, SHOWING SMALL CLIFF-HOUSE (Photograph by J. Walter Fewkes) b, TWIN TOWERS, SQUARE TOWER CANYON, FROM THE SOUTH (Photograph by Jacob Wirsula) c, TOWER 4, JUNCTION OF NORTH AND SOUTH FORKS, SQUARE TOWER CANYON (Photograph by Jacob Whsula) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 16 a, HOVENWEEP CASTLE, WITH SLEEPING UTE MOUNTAIN, SOUTH FORK, SQUARE TOWER CANYON b, ENTRANCE TO SOUTH FORK, SQUARE TOWER CANYON (I'hotographs by Geo. L. Beam. Courtesy of the Denver & Rio Grande Railroad) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 17 STRONGHOLD HOUSE, SQUARE TOWER CANYON (Photograph by Geo. L. Beam. Courtesy of the Denver & Rio Grande EaUroad) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 18 a, HEAD OF HOLLY CANYON 'tji<^M' b, SOUTH SIDE OF HOVENWEEP CASTLE, SQUARE TOWER CANYON (Photographs by Geo. L. Beam. Courtesy of the Denver & Eio Grande Railroad) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 19 a, HOLLY CANYON GROUP, FROM THE EAST (Photograph by Jacob Wiisula) b, GREAT HOUSE AT HEAD OF HOLLY CANYON, FROM THE NORTH (Photograph by T. G. Lemmon) 0, UNIT-TYPE RUIN, FROM THE EAST (Photograph by T. G. Lemmon) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 20 a, GREAT HOUSE AT HEAD OF HOLLY CANYON, FROM THE SOUTH b, RUIN B AT HEAD OF HOLLY CANYON, FROM THE WEST c, GREAT HOUSE AT HEAD OF HOLLY CANYON (Photographs by Jacob Wirsula) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 21 a, GREAT HOUSE, HOLLY CANYON b, STRONGHOLD HOUSE AND TWIN TOWERS, SQUARE TOWER CANYON (Photographs by Geo. L. Beam. Courtesy of the Denver & Hio Grande Kailroad) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 22 a, HOVENWEEP CASTLE b, SOUTHERN PART OF CANNONBALL RUIN, McELMO CANYON (Photographs by T. G. Lermnou) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 23 a, SQUARE TOWER WITH ROUNDED CORNERS, HOLLY CANYON (Photograph by Jacob 'VVirsiLla) b, HOLLY TOWER IN HOLLY CANYON (Photograph by Jacob Wirsula) c, HORSESHOE HOUSE (Photograph by T. G. Lemmon) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 24 a, HORSESHOE RUIN (Photograph by Jacob "Wirsiila) j|y"'«. b, BOWLDER CASTLE, ROAD (WICKYUP) CANYON (Photograph by T. G. Lemmou) BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 25 a, CLOSED DOORWAY IN BOWLDER CASTLE, ROAD (WICKYUP) CANYON (Photograph by J. Walter Fewkes) b, BROKEN-DOWN ROUND TOWER, SQUARE TOWER CANYON fPhotosraph by Jacob Wirsula) BUREAU OF AMERICAN ETHNOLOGY ULLETIN 70 PLATE 25 a, NORTH SIDE OF TOWER, SQUARE TOWER CANYON (Photograph by Jacob Wirsula) b, D-SHAPED TOWER NEAR DAVIS RANCH, YELLOW JACKET CANYON (Photograph by Jacob Wirsula) c, MODEL OF TOWERS IN McLEAN BASIN (Photograph by De Lancoy Gill) ^fcto Mi Z o H co C5 Z Q 2 < en < < -J CD to O z o < < z „ < UJ m BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 30 ii^. a, TOWER ABOVE CAVATE STOREHOUSES, MANCOS CANYON BELOW BRIDGE ^ ■^ b, TOWER ON MESA BETWEEN ERODED CLIFFS AND BRIDGE OVER MANCOS CANYON ON CORTEZ SHIP-ROCK ROAD (Photographs by T. G. Lemmon) 8*i '•» A Vi> '> , v ■x;' BUREAU OF AMERICAN ETHNOLOGY BULLETIN 70 PLATE 32 a, RESERVOIR NEAR PICKET CORRAL, SHOWING RETAINING WALL b, KIVA, UNIT-TYPE HOUSE, SQUARE TOWER CANYON (Photographs by T. G. Leramon^