^11i SMITHSONIAN INSTITUTIONBUREAU OF AMERICAN ETHNOLOGYBULLETIN 104A SURVEY OF PREHISTORIC SITESIN THE REGION OF FLAGSTAFF, ARIZONA BYHAROLD S. COLTON ?cW;/^^s& \tn SMITHSONIAN INSTITUTION> , S . BUREAU OF AMERICAN ETHNOLOGYBULLETIN 104 A SURVEY OF PREHISTORIC SITESIN THE REGION OF FLAGSTAFF, ARIZONA BYHAROLD S. COLTON ,*! III! -l"^ UNITED STATESGOVERNMENT PRINTING OFFICEWASHINGTON : 1932 For sale by the Superintendent of Documents, Washington, D. C. Price 30 cents LETTER OF TRANSMITTAL Smithsonian Institution,Bureau of American Ethnology,Washington, D. C, February 17, 1931,Sir: I have the honor to transmit the accompanying manuscript,entitled " A Survey of Prehistoric Sites in the Region of Flagstaff,Arizona," by Harold S, Colton, and to recommend its publication as abvdletin of the Bureau of American Ethnology.Very respectfully yours, M. W. Stirling,Chief.Dr. Charles G. Abbot,Secretary oj the Smithsonian Institution, III CONTENTS PageIntroduction 1Geography 2Methods of an archeological survey 4Maps 6House types 8Potsherds 9Culture horizons 12Geographic distribution of cultures 13Agriculture 16Water supply 17Description of sites:Upper Rio de Flag and about Elden Spring 17Walnut Canyon 18Doney Park 22Gourd Flat 25Turkey Tanks 27Grass Tank and Pipers Crater 27Among the Big Cinder Cones 29Tolchaco 29Bonito Park 29Deadmans Flat 30Cedar Ridge and the Lava Terraces 30Big Hawk Valley 32Hulls Canyon 32Citadel 33Antelope Prairie 36W^upatki Basin 36Rodin Basin 41Number of sites 44Age of ruins 44Summary 45Authorities consulted 47Appendix 49Index 67Errata _ 69V ILLUSTRATIONS Page1 . Black and white potsherds 122. CoiTUgated potsherds 123. Walnut Canyon 204. Acropolis of Red Peak 245. Citadel region 326. Index chart to location of detail maps 707. Map No, 1. Black Point to Deadmans Flat 708. Map No. 2. Wupatki Basin and Rodin Basin 709. Map No. 3. Cinder cones of the desert 7010. Map No. 4. Flagstaff and Turkey Tank 70TEXT FIGURES1. Sample survey card 52. Distribution of decorated ware 143. Distribution of undecorated red ware 154. Walnut Canyon houses (103, 475, 476, 477) 205. Walnut Canyon (385, 386) 206. Section of cliff dwelling 217. Plan of cliflF dwelling (322) 218. Details of cliff dwelling (746) 229. First fort, Walnut Canyon... _ 2410. Old Caves Pueblo (72) 2511. Picture Canyon 2612. Earth lodge in Bonito Park (409) 3013. Houses on the terraces (449, 457, 459, 474, 479, 576, 577, 578, 652,678, 682) 3114. House (618) . 3315. The citadel (355, 357, 358) 3516. Wupatki (405) 3717. House at mouth of Antelope Canyon (625) ._ 3818. Crack-in-the-rock (537) 3919. Fort near Grand Falls (580). Map of region 4020. Fort near Grand Falls (580) . Plan and eastern elevation 4221. The smaU house kiva (521) (886) 43vu A SURVEY OF PREHISTORIC SITES IN THEREGION OF FLAGSTAFF, ARIZONA By Harold S. Colton INTRODUCTIONThe present report deals with an archeological survey of a smallportion of northern Arizona lying between the San Francisco Peaksand the Little Colorado River. While the report includes most ofthe sites in the areas drained by Walnut Creek and DeadmansWash, it purposely excludes sites on the intervening watershed.The report summarizes the results of a study extending through 10summers (1916, 1919, 1921, 1923, and 1925-1930). It aims topresent three aspects: The distinctive features of individual ruins,the relations of the early cultures with those of neighboring regions,and the relations of occupied regions with features of the externalenvironment. Such a survey can never be said to be complete, yetit is important that the present knowledge of the region be recorded.Although the region about Flagstaff has long been known as thecenter of a large ancient population, relatively little study has beengiven to the remains. This seems strange when we compare thenumber of ruins here (649 recorded) with those in other places, andappears stranger still when we consider the accessibility of the ruinsto modem routes of transportation. Casual visits by Powell (1891),Stevenson (1887), and Fewkes (1900, 1904) are the only references tothe region prior to 1916 by trained observers. Popular accountshave flowed from many pens and have appeared from time to timein magazines and newspapers (Alliot, 1914). Two reasons can befound for the lack of investigation, both having to do with theaccessibility of the region to the railroad and highway. On firstglance it would seem that the investigator was repelled by the near-ness to civilization and having traveled far preferred to go fartheron into the wilds. Be that as it may, the accessibility of the regionto investigators made it also accessible to pot hunters who early ex-ploited and nearly destroyed the cliff ruins where interesting textileand basket remains have been preserved. The relatively heavyrainfall (15-25 inches), poor building materials (lava rock), and poroussoil, all contributed to the lack of preservation of artifacts other than1 Z BUREAU OF AMERICAN ETHNOLOGY [bull. io4those made of stone or clay. Notwithstanding the early neglect,Fewkes in 1926 excavated Elden Pueblo, from which a representa-tive collection of pottery was obtained. In 1928 Dr. Byron Cummingsbegan the excavation of the Turkey Hill ruin. When the materialfrom these sites is studied much light will be cast on the prehistoricpueblo art in this peripheral area.Since the burial grounds of the ancient pueblos and small house siteshave proved so rich, what remains is rapidly falling into the hands ofthe commercial pot hunter. A careful survey is, therefore, necessaryat the present time.This paper was prepared for publication in 1923. At that timevery few sites were found in a line about 6 miles wide running eastfrom the San Francisco Mountains to the Little Colorado?the dividebetween the Walnut-Rio de Flag drainage and the Deadman drain-age. For this reason maps were not prepared. Since then dozensof sites have been discovered in that string of townships. Since theyform the basis of another paper now in preparation they are omittedfrom the present survey.The writer wishes to thank Mr. J. C. Clarke, Mr. L. F. Brady,Mr. F. C. Baxter, Mr. E. L. Burnham, Dr. Edwin Helwig, Mr.Lyndon L. Hargrave, Miss Katharine Bartlett, and particularlyhis wife, Mary-Russell F. Colton, all of whom aided him in hissearch for sites. GEOGRAPHYThe high plateau of Arizona culminates in the San Francisco vol-canic field. This volcanic region is surpassed only by the ColumbiaRiver field, and is therefore the second largest in the country, support-ing over 200 extinct cones of various sizes, shapes, and ages; thetallest cone forms the massif called the San Francisco Peaks (altitude12,700 feet) and is the landmark of north central Arizona. Shelvingeastward from the peaks the plateau dips into the valley of the LittleColorado River and rises again in a series of great steps of vari-colored cliffs called the Painted Desert, the last cliff forming the mesason which the Hopis built their pueblos. So the San Francisco Moun-tains see the sun rise over the Hopi mesas, and the Hopis watch thesun set behind the San Francisco Peaks.The structure of the earth beneath the San Francisco Mountainvolcanic field is important, if we are to understand the conditionsunder which the ancient inhabitants lived. Beneath the fields of lavaand ash the Moencopi red sandstones and shales cover wide areas andoutcrop beyond the lava fields by the Little Colorado. This depositis important because it is the only rock that holds water over a verywide area. (Moencopi is said to be a Hopi word meaning the placeof running water.) Where the Moencopi has not been stripped off in COLTON) PREHISTORIC SITES NEAR FLAGSTAFF 6the process of erosion we may look for springs. The springs north ofFlagstaff?Rodins Spring, Little Rodins, and Reisers Spring?are allin the red rocks; and there is reason to believe that under outcrops ofthe Moencopi lies the porous Kaibab limestone, a sandy limestone,much fractured and cracked?a rock which holds no water. Becauseof its hard texture, when the soft Moencopi is removed, it remains.Over large areas, therefore, the Kaibab limestone forms the surface ofthe plateau. Since it is porous, no springs are associated with it.Like other limestone rocks, clay is a by-product of its weathering anddams in the canyons and washes will hold the water for some time.Beneath the Kaibab limestone the Coconino sandstone forms a thicklayer. Although it does not outcrop extensively on the plateau, it isimportant because it forms a precipitous cliff in the canyons whereverthe intermittent streams have cut deep enough to expose it. Like theKaibab above, it is waterless. The only rocks which bear water arethe red Moencopi shales and sandstones. These exposures are limitedto the fringes of mesas and a narrow strip 3 to 6 miles wide lying westof the Little Colorado River. East of the river it is soon buried bymore recent deposits.In the summer of 1930 a new factor in the physiography of this areawas recognized. Archeological evidences have accumulated which in-dicate that Sunset Crater was active in early Pueblo time. This erup-tion occurred between 600 and 850 A. D. In this eruption Sunsetspread black basaltic sand over the land. In places near the craterthis black sand lies over 2 feet deep. As we leave the region of thecrater the layer thins out in every direction. Because of the pre-vailing westerly winds the sand layer thins out more rapidly to thewest than to the east. Black sand is found over a sector which in-cludes about half a circle with a radius of about 20 miles. The centerof this circle is Sunset Crater. The area of black sand extends fromthe base of the San Francisco Peaks to the Little Colorado River andfrom Black Point to Padre Canyon. During the 1,200 years whichhave elapsed since the ash fall the winds have gathered the sand intodunes which, in the forested region, have been anchored by vegetation.On the windy steppes the sand has drifted into the canyons and on thelee side of outcrops of rock, leaving the table-lands free.This sand cover no doubt influenced the agricultural use of the land,forming a mulch and conserving the moisture. It also must haveabsorbed most of the run-off and so influenced the springs. It isinteresting to observe that the densely populated area in early PuebloIII coincided with the area covered by the ash. In the seven or eighthundred years elapsing since the region was densely populated, thehigh winds have removed the black sand cover and deposited it in thecanyons, where much of it has been carried away into the Little 4 BUREAU OF AMERICAN ETHNOLOGY [bull. 104Colorado River, Be this as it may, the black sand is an importantfactor in the human occupancy of this region.METHODS OF AN ARCHEOLOGICAL SURVEYThe methods adopted by the Museum of Northern Arizona to itsarcheological survey were developed by the author over a period of14 years (1916-1930). As these methods are adaptable to small in-tensive surveys, it is important to record them.Four archeological surveys are actually progressing in the South-west: the Laboratory of Anthropology at Santa Fe under Dr. H. P.Mera, the University of New Mexico under Reginald G. Fisher (1929-30), Gila Pueblo under H. S. Gladwin (1928-30), and the archeologicalsurvey begun by the author and continued by the Museum of NorthernArizona. Each survey follows a different method and has a differentscope.The scope of the University of New Mexico, Laboratory of An-thropology, and Gila Pueblo surveys include the whole Southwest.On the other hand, that of the Museum of Northern Arizona is quiterestricted. It is an intensive survey. Starting from Flagstaff as acenter, an attempt is made to record all the sites in every directionradiating from that center, no matter how small and insignificant theymay be. Its ultimate aim is to cover the whole northern portion ofArizona. Its immediate aim is to study the region between Flagstaffand the Hopi villages.Of the vast area in our scheme this study is an insignificant part,but as it is the hub from which other studies arise, it is important.The region lying between the San Francisco Mountains and theLittle Colorado River forms the substance of this portion of theproject.At Flagstaff the following methods are practiced in the field andthen in the laboratory:(A) In the field.(1) When a site is discovered it is given a serial number which ispainted in white paint on the smoothest rock or, if rocks areabsent, tree trunk nearest the north side of the site. Theletters N. A. precede the number, so that the initiated willknow that it is a site recorded by the Museum or NorthernArizona. (Example: N. A. 1653.) Where trees and rocksare absent an oak stake branded with N. A. and stampedwith the site number is driven.(2) a. This number is now entered on the 3 by 5 inch survey card.(Fig.l.)6. If the site is well known and has a name that, too, is entered.c. The Museum has divided northern Arizona into drainageareas of convenient size in the belief that the small drain- coLTON] PEEHISTORIC SITES NEAR FLAGSTAFF(A) In the field?Continued.age areas are cultural units, and that the final picture willbe a mosaic built out of these small areas. At the sametime the highlands between drainage areas are, in thesemiarid west, relatively uninhabited because of the lackof water, and these become barriers to the cultural units.In this paper the following drainage units are studied:Little Colorado-Wupatki Basin, Little Colorado-KodinBasin, Antelope Wash, Deadmans Wash, Rio de Flag,Upper Walnut Creek, and Lower Walnut Creek. Theproper drainage is recorded on the card. In the laboratorythe cards are fUed according to the drainage area. No. 6 BUKBAU OF AMERICAN ETHNOLOGY [bull. io4(B) In the laboratory.(1) The site is recorded by number in a special accession book.(2) The site is recorded on a map. (PI. 4.) The township isthe unit.(3) References to previous records on the sites are recorded onthe card.(4) The field card is filed according to drainage.(5) A separate file of 8 by 11 inch folders is kept for those sites onwhich much data is had, such as photographs, plans, andnotes. These are filed according to site number.(6) The potsherds that were collected are now washed in water,then in hydrocholric acid, and dried. On each sherd theserial number of the site is written in India ink and shel-lacked, so that if sherds from two sites become mixed theyare easily separated.(7) The sherds are then placed in specially made cartons of standardsizes, 6 by 12 by 10 inches, 6 by 12 by 6 inches, 6 by 12 by2}^ inches, and cardboard boxes 6 by 10 by 1 inch. Eachbox bears a label giving the site number, drainage, potterytype, and special remarks. These boxes are arranged onshelves, which are 12 inches apart?the large boxes bythemselves on the shelf and the smaller boxes and cartonsstacked. MAPSIn a survey, maps play a large part in the record. For this paperthe maps have been compiled from several sources which have beencombined with the direct observations of the author. Although alittle more accurate than any existing maps, they are far from perfect.The sources of the maps are as follows: The township plot of theGeneral Land Ofiice surveys extending from 1885 to 1919 (2 inchesequal 1 mile), the excellent series of contour maps of the UnitedStates Forest Service, Coconino sheets, Nos. 2, 3, 4, 5, and 6 (scale1 inch to the mile), United States Land Ofiice map of Arizona (1inch equals 12 miles), and the Geological Survey topographic map ofthe Flagstaff quadrangle (1 inch equals 2 miles). The author wishesto thank those Government bureaus, the Forest Service, the GeologicalSurvey, and the General Land Office, for the use of these excellentseries.From the above sources a key map and two series of large-scalemaps have been drafted. The key map shows the main features ofthe region, the auto roads, a few of the principal ruins, and theposition of the large-scale maps. On the first series, comprisingfive sheets drawn originally 2 inches to a mile and reduced to three-fourths of an inch to a mile, most of the ruins have been plotted. coLTONj PREHISTORIC SITES NEAR FLAGSTAFF 7In the second series of four maps certain regions in which ruins arevery abundant have been shown in still larger scales.The ruins have been located on the maps by two methods. In anumber of cases the section and quarter stones have been found andidentified. This is particularly true on the terraces north and east ofthe San Francisco Peaks. In other places the ruins have been locatedby distance from a road, wash, or other important landmark, while inthe settled districts the fence lines form a useful basis. Where nolandmarks are near a few ruins have been located but approximately.If a site is near a section boundary and the stone was not discovered,the ruin may have been placed in the wrong section. However, theaverage error is not over a quarter of a mile and probably considerablyless.The most important differences between these maps and the sourcemaps are as follows: Walnut Canyon at the second fort has beencorrected by a triangulation with a reflector compass. Drainage fromAztec Seep was found to enter Hulls Canyon. The contours ofO'Neil Peak have been corrected.A third series of maps comprises the house plans. These plans havebeen prepared by two methods. Where the house lines and room linesare not very distinct the distances were paced. When they are sharpand clear the distances have been measured by tape. But a few of theplans have been prepared for publication. The larger houses and thosewhich show some special characteristic arrangement have beenselected; hundreds of others are recorded on cards and in the files.On the maps, the author has used those place names which hefound on existing maps. However, these names are so few that hehas been obliged to search for others. He has applied to the localinhabitants and when these have failed him, he has coined names.These new names usually are made from some local landmark.One change has been made which diverges from the usage on somemaps. The author considers that the Rio de Flag is a tributary ofWalnut Creek and not that Walnut Creek is a tributary of the Rio deFlag. The Rio de Flag has no channel where it enters Walnut Creekand rarely ever flows, while Walnut Creek has a wide continuous bedin the region of the junction. In fact, if one follows the stream bed,the Rio de Flag could not be found. At the same time the nameWalnut Creek should be substituted for San Francisco Wash.On the maps the author adopted a series of symbols which seem tobe suitable to this particular region. It must be understood that theinterpretation of the ruins is not final, as few excavations were made.It seems desirable to indicate on the maps the magnitude of theruin. So the author on hypothetical grounds selected the followingfive or six types to be recorded. He assumed that a condition similarto that of the Hopi existed, that the married daughters tended to live 8 BUREAU OF AMERICAN ETHNOLOGY [botl. io4 near their mothers. If, in times of peace and plenty, two daughterslived to marry, then we would have the following classification:1 to 2 rooms A one-family house.2 to 4 rooms A house occupied by a mother and one or two marrieddaughters.5 to 8 rooms A house occupied probably for three generations all inone family.9 to 16 rooms A house occupied for four or five generations by onefamily or built by a congregation of several families orclans.Pueblos are probably made up of the aggregation of several fam-ilies. The larger pueblos were perhaps even composed of several clans.The symbols for earth lodge and sherd area in most cases representsome kind of a pit house and nearly all belong to Pueblo II. Theother symbols on the maps are self-explanatory or are made clear inthe legend. HOUSE TYPESEight types of structures have been recognized in the San FranciscoMountain area.(1) The masonry dwellings from single-roomed houses of 1-story to3-story pueblos with as many as 50 rooms. These may or may notbe associated with a rectangular kiva.(2) Small masonry dwellings constructed under natural over-hanging rocks.(3) Small masonry dwellings constructed in artificial caves exca-vated in volcanic ash under a lava flow?called cavate dwellings byFewkes, 1896.(4) Masonry forts in defensive positions on rims of canyons or onthe tops of hills.(5) Masonry granaries in the open?associated with earth lodges.(6) Adobe or masonry granaries built in caves.(7) Earth lodges. Rectangular pit house (Hargrave, 1930) withan entrance through the roof and a ventilator on the east side. Earthlodges include the vestibule house (Colton, 1920) with an alcove onthe east side which is not an entrance. These may be (a) excavatedin the ground (a rectangular pit house), (b) built on the surface ofthe ground, or (c) built on the top of a low mound. All three arevarieties of a single type and belong to about the same period, nomatter on what level the floor is placed. (The problem of thevestibule house will be considered in a paper in preparation.)From a survey of the region it would seem probable that the rec-tangular kiva accompanies all ruins with which Tusayan or LittleColorado black-on-white pottery is found. It is usually representedby a depression on the east or southeast side of the building. In thelarge pueblos, such as Elden and New Caves, the kiva seems to be a coLTON) PREHISTORIC SITES NEAR FLAGSTAFF 9large room at ground level. In the small house sites the Idva hasvaried from 9 by 9 to about 12 by 16 feet. From the few excavationsmade it seems that the kiva contains a ventilator and deflector. (SeeRuin No. 521, fig. 21.)Ruin No. 887 furnishes a well-preserved kiva roof, a large pinetimber 8 inches in diameter running across the middle of the room,and supports 3 to 4 inch pine timbers which are spaced about 4 to 6inches apart. On these are laid a neat floor or split cedar shakesabout 3 feet long. On this roofing lie bunches of grass and on top alayer of clay about 4 to 6 inches thick. Placing all of this togetherwe have a fairly complete picture of the San Francisco Mountain kiva.Mr. J. C. Clarke has reported an oval kiva buried in a sand dunenear Heisers Spring. This may turn out to be a pit house whenit is excavated. In the Citadel region are some evidences of ovalkivas or pit houses?excavation alone will reveal their significance.However, whatever they are, they are rare.POTSHERDSWhen potsherds are studied from the sites in this area it will beseen that they fall into the types listed below. Where the typeshave been well studied they have been given names according tothe method proposed at the Pecos Conference in 1927, a geographicalname followed by a descriptive term. Types that have not beensufficiently well studied or whose center of development lies out ofthe area of this study and which have not yet been given a per-manent name, are preceded by an asterisk (*).(A) Wares with no decorative designs.(1) *Plain gray ware.?Granular surface due to coarse temper;throat of jar not coiled.(2) Rio-de-Flag brown.?Plain red, brownish, and brownish-blackware, smooth surface, noncorrugated, tempering materialcoarse sand. Jars and bowls.(3) Deadman gray.?Plain gray ware. Surface and paste gray,surface smooth, noncorrugated, tempering material containsfine mica. Jars and bowls.(4) Deadman fugitive red.?Plain gray ware similar to the above inpaste and surface, but with the surface painted withhematite after firing and not refired. The red paint willwash off with water. Jars.(5) Sunset red.?Plain ware, grayish paste, temper black basalticcinders, surface floated, red \vith blackish firing marks.In worn places the basaltic cinders show through the red.Bowls and jars.54563?32 2 10 BUKEAU OF AMERICAN ETHNOLOGY [botl.io4(A) Wares with no decorative designs?Continued.(6) Flagstaff red.?Plain ware, grayish paste, temper black basalticcinders or sand. Exterior usually with a heavy float or slipof red with polishing marks or striation without definite foci.Interior of bowls polished a metallic black to deep gray.Interior of jars gray. The fact that this ware frequentlyhas cinder temper and that the polishing marks are withoutdefinite foci distinguishes it from the Gila red ware of theMiddle Gila (Gladwin, 1930). Jars and bowls. Called byFewkes, 1927, Flagstaff ware.(B) Wares with decorative designs.(7) *Plain gray with a coil neck.?Same paste and temper as No. 1.Sherds from the bottom of the jar can not be distinguishedfrom bottom of No. 1 jar.(8) *Gray corrugated.?Exterior surface completely corrugated,coils indented. The many varieties have not yet beenclearly distinguished. Mostly jars. (PI. 2, (3) (4) (5) (6)(7).)(9) *Gray corrugated.?Coil partially obliterated. Mostly jars.(PI. 2.)(10) Elden corrugated.?Exterior surface may be red or yellowpaste and basaltic temper similar to Flagstaff red. Corru-gations quadrilateral in shape with rounded edges. In-terior with or without a black metalUc burnished surface.Bowls and small jars. Fewkes, 1927. Figure 219. (PI.2 (1) (2.)(11) *Basket maker black-on-white.?Black paint on a gray granularsurface similar to Nos. 1 and 7. Coarse temper showsthrough the surface. Bowls and jars. (PL 1 (1).)(12) *Pueblo I black-on-white.?Paste similar to Nos. 1, 7, and 11.Interior of bowls or outside of jars floated and polished.On the polished white surface a black design is applied ? fine lines, pendent dots, small triangles and stepped ele-ments characterized by "marked crudity of brush work."Kidder, Southwestern Archaeology, p. 75, pi. 34. Bowlsand jars. (PI. 1 (2).)(13) Deadmans black-on-white:?Gray paste, white polished slipon which a design of black broad lines, pendent dots, isolateddots, large triangles and squares. Exterior of bowlsrough. Bowls and jars. (PL 1 (3).)(14) Deadmans black-on-gray.?Gray paste, surface without a slip,similar to Nos. 3 and 4. Black design similar to No. 13applied directly on the gray surface. Bowls and jars. coiTOKl PREHISTORIC SITES NEAR FLAGSTAFF 11(B) Wares with decorative designs?Continued.(15) Walnut black-on-white corrugated.?Gray paste, corrugatedexterior, interior a white slip design similar to Deadmansblack-on-white. Bowls.(16) Tusayan black-on-white.?Gray paste, fine sand temper, float,or white slip, black designs of various geometric forms,that called "serations" being most characteristic. Sameas Kidder Proto-Kayenta, Kidder, 1924. Bowls and jars.(PI. 1 (4).)(17) Wupatki black-on-white.?Thick ware, gray paste, coarse sandtemper. White sUp. Black design similar to Kayentanegative pattern. Probably a local attempt to reproduceKayenta black-on-white. Exterior not finely finished.Bowls and jars. (PI. 1 (5).)(18) Little Colorado black-on-white.?Dark gray paste, coarsesand temper, white slip. Usually thicker than Tusayanblack-on-white. For designs see Gladwin, 1930 (a),Plate 3. Jars and bowls. (PL 1 (6).)(19) Deadmans black-on-red.?Gray paste, sand temper, red sUpinside and outside, painted with a black design. Bowlsand jars.(20) *Black-on-white sherd temper.?Gray paste, black design onwhite slip.(21) Tusayan black-on-red.?Gray paste burning pinkish or yellow-ish on the surface, sherd temper, painted inside and outsidein red on which a black design is applied. Bowls. There isa similar ware without the red paint found in Pueblo IIsites in the Upper Moencopi Wash. On these the black isapplied directly to the pink surface. This has not yet beengiven a name.(22) Tusayan -polychrome.?Called by Kidder 1924 Proto-Kayentapolychrome. A variety of Tusayan black-on-red. Graypaste burning yellowish on the surface, sherd temper.On the yellowish surface a red design is outlined in black.On the yellowish exterior of bowls a broad red line isdrawn below the rim. Bowls.(23) Kayenta polychrome.?Paste and yellowish sHp or float similarto Tusayan polychrome. Ked is used on large masses.The black is outlined with a narrow white line. On bowlsthe exterior contains a broad red line as in the Tusayanpolychrome. See Kidder, 1924. Bowls and small jars.(24) In the late sites other wares manufactured in neighboringregions are sometimes encountered, such as Jeddito black-on-white, Kayenta black-on-white, Klag-E-to black-on-yellow, and Little Colorado polychrome. 12 BUREAU OF AMERICAN ETHNOLOGY [bull, u ?CULTURE HORIZONSThe Southwestern Archeological Conference, held at Pecos, N. Mex.^in 1927 (Kidder, 1927), proposed the following chronological classifi-cation of culture periods in the Southwest.Basket Maker I, or Early Basket Maker.?A postulated (andperhaps recently discovered) stage, preagricultural, yet adumbratinglater developments.Basket Maker II, or Basket Maker.?The agricultural, atlatl-nonpottery-making stage, as described in many publications.Basket Maker III, or Late Basket Maker, or Post Basket Maker. ? The pit or slab house building, potter-making stage. (The threeBasket Maker stages were characterized by a long-headed population,which did not practice skull deformation.)Pueblo I, or Proto-Pueblo.?The first stage during which cranialdeformation was practiced, vessel neck corrugation was introduced,and villages composed of rectangular Uving rooms of true masonrywere developed. (It was generally agreed that the term pre-Pueblo,hitherto sometimes applied to this period, should be discontinued.)Pueblo II.?The stage marked by widespread geographical exten-sion of life in small vUlages; corrugation, often of elaborate technique,extended over the whole surface of cooking vessels.Pueblo III, or Great Period.?The stage of large communities,great development of the arts, and growth of intensive local speciali-zation.Pueblo IV, or Proto-Historic.?The stage characterized by contrac-tion of area occupied; by the gradual disappearance of corrugatedwares; and, in general, by decline from the preceding cultural peak.Pueblo V, or Historic.?The period from 1600 A. D. to the present.Since the Pecos Classification was published the pottery types as-sociated with the culture periods in the central area have become moredistinct. In the San Francisco Mountains culture periods correlatewith house types and pottery types as follows:Basket Maker I and Basket Maker II have not been distin-guished in the area covered by this paper.Basket Maker III is but poorly represented, three sites only beingrecorded.House type.?An earth lodge, rectangular pit house. Hargrave(1930).Pottery.?Plain gray. Basket Maker black-on-white.Pueblo I is poorly represented ; aU the known sites lie east of Flag-staff among the cinder hills.House types.?An earth lodge, a rectangular pit house. Hargrave(1930). BUREAU OF AMERICAN ETHNOLOGY BULLETIN 104 PLATE 1 Black on White Potsherds I, Basket Maker III black on white; 2, Pueblo I black on white: 3, Deadmans black on white, indexfor Pueblo II; 4, Tusayan black on white, index for Pueblo III, north of the San Francisco Peaks; "), Little Colorado black on white, index for Pueblo III, south of the San J'rancisco Peaks;6, Wupatki black on white, index for late Pueblo III (12.50-1300). This type is similar to Kayentablack on white but not so thin nor so well finished. BUREAU OF AMERICAN ETHNOLOGY BULLETIN 104 PLATE 2 ^H J^^m^t.irj6#^.'M-^^I>tt.^- 5 \W^" ^bu?.^ f6 \ j(pgj^3fe!iiwiiiii?i CHARACTERISTIC CORRUGATED POTSHERDS1, 2, Elden corrugated, Pueblo III, south of the San Francisco Peaks; 3 and 5, Pueblo IIIcorrugated, Pueblo III, north of the San Francisco Peaks; 4 and 6, partially obliteratedcorrugated, late Pueblo II and early Pueblo III; 7, corrugated, Pueblo II. coLTONl PREHISTOKIC SITES NEAR FLAGSTAFF 13Pottery.?Plain gray, plain gray with a neck coil, Pueblo I black-on-white, Rio de Flag brown.Pueblo II is very abundant everywhere.House types.? (1) Earth lodge, a rectangular pit. Hargrave (1930).(2) Earth lodge of the vestibule type, Colton (1920), under the ground,on the ground, or where the snowfall is heavy and the soil clayey,on the top of a mound. (3) Masonry granaries. (4) Masonry forts.Pottery.? (1) Drainage of Deadmans Wash, Antelope Wash:Deadmans gray, Deadmans fugitive red, Deadmans black-on-white,Deadmans black-on-gray, Deadmans black-on-red, gray corrugated.On late sites Tusayan black-on-red is found.(2) Drainage of the Rio de Flag and Walnut Creek: Rio de Flagbrown, Sunset red, Deadmans black-on-white. Walnut black-on-whitecorrugated. Gray corrugated, Deadmans black-on-red.Pueblo III.?Includes most of the sites recorded on the maps.Everywhere abundant except close to the San Francisco Peaks.House types.?Two-room to forty-room masonry pueblo. Usuallywith a rectangular kiva on the east or southeast side.Pottery.? (1) Deadmans and Antelope drainage and Little Colorado,Wupatki, and Rodin basins. Gray corrugated, often partially oblit-erated, Tusayan black-on-white, Tusayan black-on-red, Tusayanpolychrome, and in the late pueblos Wupatki black-on-white andKayenta polychrome are added to the others.(2) Rio de Flag and Walnut Creek: Gray corrugated, partiallyobliterated; Tusayan black-on-white; Little Colorado black-on-white;Tusayan black-on-red; Tusayan polychrome; Flagstaff red; Eldenredcorrugated; and Sunset red. At late sites Wupatki black-on-white,Kayenta polychrome. Little Colorado polychrome, Klag-E-to blackon yellow, and Proto-Four-mile polychrome are met with.Pueblo IV and V are not represented in this region, although threesites were occupied up to the transition into Pueblo IV: Wupatki,Old Caves, and Turkey Hill.GEOGRAPHIC DISTRIBUTION OF CULTURESThe method of studying the distribution of pottery differs fromother similar studies. Fewkes (1904) counted the number of pieces,presumably whole bowls and jars, of each type collected from aseries of sites, and has presented us with a pottery cross section ofArizona, 70 miles east of the Flagstaff region; Kidder (1917) at Pecosand Nelson (1916) in the Gallesteo Basin and Gladwin (1930) haveeach studied pottery complexes by counting potsherds and expressingthe results in percentages. The author feels that this method canbe improved upon, because some kinds of pottery are more fragilethan others; at least this is true in the San Francisco Mountainregion where the coarse brown and red ware is frequently found in 14 BUREAU OF AMERICAN ETHNOLOGY [BVLL. 104 smaller fragments than the black-and-white ware. For this reasonthe author weighs the sorted potsherds from a given site rather thancounts them. ? .'il.* Scale" '* " ^ *" ** <>'? "' 80 9o '*f(nes.DlSTRTBUTIOM Of ^ORtHCRN "WXRE IN NoRTH EASTERIiArizona. ExpsessED Ik ^o By "Weight. Figure 2.?Distribution of black on white and corrugated wares in northeastern ArizonaThe method followed was to collect every potsherd, and thenweigh the piles. The results expressed in percentage form the basisof Figures 2 and 3. colton] PREHISTORIC SITES NEAR FLAGSTAFF 15Pottery studies reveal that in Pueblo II and in Pueblo III twocultures existed in the San Francisco Mountains. The sites onWalnut Creek drainage have a different complex from those in the Distribution Op UtJDECot^Ai^o Rso Ware In IWoRTKeASTERN ARIZONA EXPRESSED 1(4 ^Bv WeimS FiGUHK 3.?Distribution of undecorated red wares in northeastern ArizonaDeadmans Wash drainage. A line running east from the main peaksin the region of the Coconino Divide to the Little Colorado seems tohave been a barrier or a frontier between Walnut Creek and Dead- 16 BUREAU OF AMERICAN ETHNOLOGY [bull. i04mans Wash complexes. In Upper Rio de Flag and Upper DeadmansWash the sites are nearly all Pueblo II. Pueblo III is absent, but inthe upper Rio de Flag, Rio de Flag brown is the utility ware, whilein Upper Deadman it is Deadmans gray or Deadmans fugitive red.In lower Deadmans and the Lower Rio de Flag, Pueblo III sites aremore striking than the Pueblo II sites. South of the line Flagstaffred is dominant, while north of the line it is absent, or nearly so.Little Colorado black-on-white south of the line, Tusayan black-on-white is the common one to the north. It is hard to see what kindof a barrier existed which prevented the mixing of two tribes, but itwas an important one. AGRICULTUREFrom the survey it appears that the eruption of Sunset Crater wasa dominant incident in the history of our area. Previous to thatevent the sites were clustered close to the peaks. After the eruptionthey spread from the peaks to the Little Colorado River.For primitive agriculture a light soil is necessary, a soil that canbe penetrated by a digging stick. Therefore we find the early sitesassociated with alluvial parks in the forest and old intercone basins.After the eruption of Sunset Crater, with the country covered witha thin layer of black sand, a new era for agriculture developed. Thisblack sand absorbed heat and at the same time acted as a mulch,conserving the water in the soil.At the present time the Hopi selects for his cornfields a thin sanddune which is traveling slowly across the land. With his diggingstick he plants the seed in the soil, which is kept moist by the mulchof sand, a foot to 18 inches deep, depending on the thickness of themulch. It is thought probable that the black sand made a similarmulch and that an area which had previously been unsuitable foragriculture became a garden spot. Because the sand cover washedand blew away, the late pueblos, like Elden, Turkey Hills, and OldCaves, are found close to the peaks, while the people of Wupatkicultivated the dunes in the shelter of an escarpment. Now that thesand cover has been largely removed, the land has gone back to thegrassy steepe that originally existed.In many places lines of stones are visible which mark the positionof the ancient fields. Lines of stones dammed little draws and caughtthe sand and moisture, as we see to-day at the second Hopi Mesa.(See wash west of Crack-in-Rock.) Parallel lines of stones wereused as windbreaks, such as can be seen to-day at Sikyatki at theFirst Hopi Mesa (or in this region southeast of Ruin 377, or Southof Grand Falls).Broken hoes are picked up on the site of the fields and a diggingstick has been found in one excavation. COLTOKJ PREHISTORIC SITES NEAR FLAGSTAFF 17Methods of agriculture as far as we can see were similar to themethods used to-day by the Hopi.WATER SUPPLYThe rainfall ranges between 40 inches on the mountains (Pearson,1921) and 25 inches at the base, and from 5 to 10 inches by theriver (Monthly Weather Review). In the region suitable for agri-culture the rainfall is between 15 and 25 inches. The porous natureof the soil renders this water unavailable for domestic purposes exceptwhen caught in basins in lava canyons or when it appears in springsin the Moencopi sandstone and shales. Certain clay basins will alsohold temporary pools. The distribution of the ruins is largely amatter of water supply. Water is scarce over the entire area, not-witiistanding the fact that the rainfall would support running streamsif the region had a different geological structure. The author acceptsthe doctrine of climate change to account for ruins in now waterlessregions. DESCRIPTION OF SITESUpper Rio de Flag and about Elden SpringNorth of Flagstaff the Rio de Flag has its sources on the southernflanks of the San Francisco Mountains. Although petroglyphsabound at the old Veit ranch on Mount Agassiz, altitude 8,100feet, the highest dwellings occur lower at Little Le Roux Spring(Cliff shelter), altitude 7,600 feet. In this region north of Flagstaffthe most notable group of ruins lies a few hundred yards north ofthe Flagstaff reservoir on the Schultz Pass road. These furnish the"best examples of earth lodges (Colton, 1920) which had been builtupon a platform or mound. The remains of other earth lodgesfringe the base of the mountains east to Elden Spring. In this regionno houses were built; the notable group west of Elden Spring in thepines described and figured by Colton (1918 a, b), were probablyplatforms on which earth lodges had been built and 125 besideNational Highway a mile south of Elden Spring (Colton, 1918),may be a similar platform.Mr. L. F. Brady, of the Mesa School, discovered pottery frag-ments 14 feet below the surface of the earth in the walls of the arroyoof the Rio de Flag associated with the stumps of buried trees. Theauthor aided him in digging out a pit-house site 4 feet below thepresent level of the alfalfa field of the county farm. The details ofthis work are left to Mr. Brady to record.North of Flagstaff the characteristic dwelling is the earth lodgeon a platform while the characteristic potsherd of the dwellers isRio de Flag brown. 18 BUREAU OF AMERICAN ETHNOLOGY [bull. io4Walnut CanyonAmong the antiquities of the Flagstaff region the cHff dwelHngs inWalnut Canyon are best known. Not only have they been longestexploited, but also they are more frequently visited than any of theantiquities of the region, being examined yearly by over 10,000people. (Map, pi. 3.) In 1906, to preserve the ruins, the Presidentset aside as a National Monument two sections and a quarter in-closing the most conspicuous remains. Being within 6 miles of thetranscontinental highway (U. S. No. 66), it is one of the most readilyaccessible national monuments to the automobile tourist. Becauseof this accessibility and the large number of visitors to the canyon,an account of it is necessary.It can not be said that Walnut Creek rises, because it rarely hasany water flowing in it. Yet it drains a very large area lying betweenthe San Francisco Peaks and Mormon Mountain 40 miles to thesouth. The waters from an occasional cloud-burst in the summer orthe melting snows in the spring flow north from the region aboutMormon Mountain, down Clark Valley, past Lake Mary to FishersTank. For many miles a cliff or escarpment rises abruptly 200 or300 feet above its eastern banks. This cliff, a fault escarpment,forms the western edge of Anderson Mesa. At Fishers Tank thechannel turns east, slicing through Anderson Mesa, cutting a rockycanyon in the Kaibab limestone and underlying Coconino sand-stone 300 to 400 feet deep; and follows a meandering course east-ward with many sharp hairpin bends, its alternately sandy andboulder-strewn bed opening out finally into a shallow channel nearWinona Station on the Santa Fe Railroad.Halfway between Fishers Tank and Winona Station the ancientpeople of the region congregated in a village. Tilling the groundin the parks to the northeast, they found water, shelter, and securityin the canyon. Water, the primal necessity, exists in several placesin natural pools and probably can be found by digging in the sandin other places. The unequal weathering of the sandy Kaibab lime-stone forms natural overhangs with flat floors. Taking advantage ofthe overhanging ledges, side and front walls are the only structuresneeded to complete a house. Security can be found on the almostinaccessible "islands" and peninsulas which occur at nearly everysharp bend in the canyon.In occupying the sunny north walls of the canyon, the bitter coldof the winter was ameliorated, and shelter from the irritating windsof the spring and early summer attained. Except for the chore ofcarrying water up the well-built trail from the pools in the canyon.Walnut Canyon must have been a pleasant place in which to live.The first fort or "Steamboat Fort" lies 3% miles southwest ofWinona Station up the canyon. (Map, fig. 9.) Here the canyon cowoN] PREHISTORIC SITES NEAR FLAGSTAFF 19 is shallow, about 100 feet deep. Houses crowd the top of a peninsula,the perpendicular walls of which form an adequate defense. Otherhouses occupy the promontories on the canyon wall, while the wideflats on the canyon floor contain foundations grouped about a largecorral-like inclosure. At the base of the fort on the west side of the "island " a kiva seems to have been built. The first fort would form agood place for investigation.The second fort, 3 miles above the first fort and a mile above theSanta Fe Railroad dam, is not so favorable for study. Here thecanyon is deeper and the remains of habitation fewer. Evidencesof long occupation are not present. About the first fort the inhabi-tants lived in houses on the surface of the ground. At the secondfort they occupied houses built under overhanging ledges. Thesecond fort presents few points of interest, notwithstanding itspicturesque situation at the junction of a large tributary canyonwith Walnut Canyon.Within the National Monument, a third fort similar to thosepreviously described occupies a peninsula jutting out from thenorthern rim of the canyon, and is surrounded by one of the fre-quent hairpin curves of Walnut Creek. Here the canyon is exceed-ingly narrow as well as deep. Nevertheless the third fort, calledpopularly the "island," lies in the thickest settled portion of thecanyon. It is the one "island" easily accessible and frequented bythe tourist. The sides of the "island" are covered with cliff dwellingsand the top crowned with a fort where traces of the defensive wallsstill remain.The fourth fort occupies the end of a peninsula jutting out fromthe south wall a quarter of a mile above the third fort. While thesouthern face is covered by dwellings, the northern face, supportinga growth of balsam, Douglas fir, and other members of the Canadianfloral zone, is without many houses. The fort is defended by a wallacross the neck of the promontory. A mile above the fourth fortlies the fifth fort. This is a peninsula or "island" joined to thesouthern wall. Both the east and west exposures of the "island"contain dwellings. However, like the second fort, it never formedthe refuge for a teeming population. Farther up the canyon penin-sulas and promontories abound, but they were not used by theprimitive inhabitants of the region as places for defense. This maybe because the cornfields were too far away; other places just aseasily defended lie nearer the arable gounds.The ruins about Walnut Canyon fall in four groups: (1) Small oneor two roomed houses on and back from the canyon rim, a quarter ofa mile or so, or on the islands or peninsulas in the canyon, (2) three orfive roomed houses similarly located, (3) cliff dwellings, and (4) forts. 20 BUREAU OF AMERICAN ETHNOLOGY [BULL. 104The positions of the dwellings on the level ground are shown on themap and are not different from others in the region. The group onthe promontory in the eastern part of the National Monument(Nos. 108, 385, 386, 475, 476) is perhaps the most interesting and 477 5$Wlllll||l|l|,1 103 i ^/nim\u\\\^i'0 Jp -ession A7Q Al> 4^76 H-Ed / Figure 4.?Walnut Canyon?Group of houses on the north rim (103, 475, 476, 477)deserves the most study. (Figs. 4 and 5.) Notwithstanding dozensof houses on the level ground, the characteristic dwelling in WalnutCanyon is the cliff shelter. (Figs. 6-8.) Whereas, the Coconinosandstone at the bottom of the canyon forms, almost everywhere,except where crushed in fault-^ ing, a precipitous wall, theS overlying hard Kaibab lime-t stone weathers characteristi-cally in a series of steps. Threerelatively softlayers are overlaidby a hard layer. The weather-ing of the three soft layers hasresulted in a series of threeledges with overhanging roofs.Here on each tier the people of the region built rooms, one afteranother in a single line; but in two places they added an outer series ofrooms (fig. 7). Although the overhanging rock made a roof unnecessaryfor the inner rooms, the outer rooms entirely beyond the overhangrequired a constructed roof. ED303 hmfa.l I 38e15 FT-Figure 5.?Walnut Canyon?Group of houses on thenorth rim (385, 386) BUREAU of! BULLETIN 104 PLATE 3 BUREAU OF AMERICAN ETHNOLOGY BULLETIN 104 PLATE 3 COX.TON] PREHISTORIC SITES NEAR FLAGSTAFF 21The construction of the partitions is peculiar. The partition wallswere built and completed before the end walls were started, so thatthe end walls were not tied to the partitions walls. Having nothing ??5TnoKe HoleWooden Linlel One Fool Figure 6.?Walnut Canyon?Generalized section of a cliff house basedparticularly on house 746. This section illustrates the way the dwellersin the canyon made use of the overhanging limestone rocks; how thewalls were constructed and the floors leveledto hold them when rain water undermined them, they fell. Thisaccounts for the almost universal absence of complete front walls.Since a large population once lived in this region, it is a strikingfact that no trace of a kiva has been found except at the first fort far ! I I I 11 322Figure 7.?Plan of house 322 which was built under an overhanging ledge of limestone.Three rooms extended beyond the line of overhang, which must have had constructedroofs. No. 322 would be a good site to restoredown the canyon. Where the ruins are thickest nothing suggestinga kiva has been observed.As the canyon is visited by so many people yearly, the author sug-gests that the present trail be closed, that Casa 103 be reconstructedinto a custodian's house and local museum and Casa 475, which con-tains two good-sized rooms, be restored. He also suggests that a 22 BUREAU OF AMERICAN ETHNOLOGY [bull. io4 trail be built from the end of the promontory opposite the "island"down to the level of the middle series of cliff dwellings. At this levela portion of cliff dwelling No. 322 should be restored. With thesechanges made, Walnut Canyon will be a much more interesting andinstructive place for itinerant tourists.Walnut Canyon was occupied principally in late Pueblo II andearly Pueblo III cultural strata.DoNBT ParkEliDEN PUEBLOOn the divide joining Sheep HiU with Elden Mountain, west of theNational Highway, in the pines, lies Elden Pueblo, No. 142, excavatedby Fewkes in 1926. (Fewkes, 1927, plan, p. 210.) This Pueblo,like several others in the neighborhood, such as Turkey Hills, was Figure 8.?Details of door and smoke hole of house 746,one of the cliff dwellings in Walnut Canyonlocated with little thought of defense. Of the main building DoctorFewkes excavated 33 rooms, leaving about 16 rooms unexcavated, sothe building contained about 49 rooms on the ground floor. Aboutthe central building are the remains of three small 1 -story houses ofthree to five rooms each. The kiva of Elden Pueblo proved to be alarge room at ground level with a low seat or banquette around theinside wall. Elden Pueblo belongs to Pueblo III culture horizon.TURKEY HILLS PUEBLOAmong the larger pueblos of this region mention must be made ofone (No. 660) lying a quarter mile south of the Winslow Road on thelower flats of the Turkey Hills. A large mound which might coverthe remains of a 3-story building with perhaps 15 basal rooms has a1-story extension of 11 rooms extending northeast. Three small 1to 4 roomed houses lie a few yards to the southeast. Between thelargest of these, a house with 4 rooms, lies a depression which excava-tion might show to be a kiva. An outcrop of lava north of the ruinwas used as permanent metates for grinding corn. COLTON) PREHISTORIC SITES NEAR FLAGSTAFF 23Although not as large as Elden Pueblo, this ruin is one of the largerones on the level ground. Being near the present settled district ofDoney Park, it is subject to vandalism and it is reported that somestone has already been removed for building purposes. In the sum-mer of 1928 Dr. Byron Cummings, with the aid of his students, beganthe excavation of this ruin. Culture horizon is Pueblo III. Pot-sherds show that this site was occupied later than Elden Pueblo.WINONA ROAD GROUPNear Locketts Tanks, east of Doney Park on a lava flow, stood agroup of small houses described by Colton (1918 a, 6). This grouphas been destroyed by the location of the new "National Highway,which has passed through the main house of the group. It furnisheda very good example of the small house ruin of the region, a house notvery different from that described by Prudden (1920) in the SanJuan region, Pueblo III. OLD CAVES PUEBLOThe Old Caves Pueblo (No. 72), plastered on the sunny southslope of a small cinder cone, overlooks Doney Park. Fewkes (1900and 1904) has described it well and figured some of the curious under-ground rooms beneath each house. The plan (fig. 10) shows thenumber of rooms and also records the rooms with undergorundchambers. Since this pueblo is unique in pueblo architecture, itshould be preserved from the pot hunters, who are mining in itsburial grounds. This hilltop should be a National Monument. Theculture horizon is late Pueblo III and very similar to Turkey Hill.PICTURE CANTON AND PORCUPINE CAVEEast of Cliffs Station the Rio de Flag crosses north of the trackand enters a lava canyon. In this canyon lies a natural tank. North-east of the tank a series of small houses of early Pueblo III clusterat the canyon rim, while other house sites are formed in the talusunder the lava cliff. (Map, fig. 9.) In one place the primitiveinhabitants mined the loose lapilli and scoria from under the lavaflow, forming a series of four chambers called locally Porcupine Cave.The last chamber still is separated by a stone and adobe wall from thethird chamber. This wall is punctured by a doorway with a woodenlintel stni in place. The wall is neatly plastered. (Since this waswritten in 1923, vandals have destroyed all the walls.)In 1919 the author excavated a number of graves in the talus eastof the cave, recovering a number of artifacts, and in 1921 he excavateda rectangular chamber in the talus which may or may not have beena kiva. Here nothing notable was found but the fireplace and acache of five stone axheads. 24 BUREAU OF AMERICAN ETHNOLOGY [bull. 104Picture Canyon (fig. 11) was named for the large number ofpetroglyphs which cover the basalt rocks; indeed, a notable array.It is a typical site of the people who made the Flagstaff red pottery.Nearly all the bowls recovered were of that type. However, a fewpieces of Little Colorado black-on-white and corrugated ware wererecovered from the burials. These burials, under the ledges of basaltbeneath the sites of the houses, were as follows : 320 FtForm Line Interval - 50 Ft.Figure 9.?First Fort in Walnut Canyon. Masonry houses are found on the canyon rim while pithouses cover the bench on the canyon floor(a) Male, adult, artifacts, nine bowls of Flagstaff red ware, ladle,shell earring, and a disturbed primary burial (227).(6) Child aged about 3, artifacts. Small olla, small bowl, black-on-white. Area shell, bottom of small Elden corrugated bowl and bluepigment, and a disturbed primary burial.(c) Child (age not determined), artifacts, small bowl, small olla ofblack-on-white. Area shell, bottom of small Elden corrugated bowland a disturbed primary burial.(d) Child 6 years, lower jaw only, artifacts, small corrugated jarcontaining the bones of three ground squirrels, 3 ladles, 4 small bowlsof Flagstaff red. BUREAU OF AMERICAN ETHNOLOGY BULLETIN 104 PLATE 4O'NEIL CRATER GROUPRuins Northeast of Flagstaff. Ariz.Sec,29.Twp.22N..R.3E. colton] PREHISTORIC SITES NEAR FLAGSTAFF 25(e) Child encased in clay, age not determined, artifacts, 2 bowls ofFlagstaff red ware, sherds of a Tiisayan black-on-white bowl.Gourd FlatTHE ACROPOLIS OP " RED " OR "o'NEIL" PEAK NEW CAVESNorth of the ranches in Gourd Flat rises the red rim of a raggedcrater. This rim is low on the east side but rises 300 feet on the west. ?_^^ <^^'^'''%. ^ %j>f^S Figure 10.?Old Caves Pueblo (72). A communal house straggling down the southern slope of a smallcinder cone. It occupies the most convenient point near Doney Park suitable for defense. The curiousunderground chambers in nearly every room hollowed out of the cinders make it unique in puebloarchitectureCrowning the rim, a fringe of ruins with a few gaps enclose the crater.While the ruins on the east and the south rim are interesting, those onthe west and north rim are remarkable. Fewkes (1904) commentedon the ragged and inhospitable lava slopes that the inhabitants choseas dwelling places. He described a few of the cavate rooms on theface of the crater, but left us without a picture of the whole.54563?32 3 26 BUREAU OF AMERICAN ETHNOLOGY [bull. 104The acropolis of Red Peak (map, pi. 4) covers the northwesternportion of the crater rim, while detached houses of 1, 2, and even11 rooms straggle down the hill to a saddle which had been leveledoff perhaps as a dance court as suggested by Powell (1891). Fewkes(1904) interpreted it as a reservoir. On the terraces at the south footof the mountain lie other houses. (See Colton 1918 6.)The inner face of the crater falls steeply away, in places precipitous,while the outer slope is more gradual. The inner face, therefore,could be easily defended, short intervals of wall being all that isnecessary. To protect the gradual outer slope, an extensive wall was Figure 11.?Picture Canyon, showing the position of Porcupine Cave (225). Petroglyphs areabundant on rocks west of the cave constructed. This wall, over 700 feet long, seems in places to havebeen breast-high, but is now in many stretches a mere retaining wallfor a terrace. At the northeast end the wall merges into a series ofdwellings which form a kind of bastion. From the middle of thelong wall and perpendicular to it runs a second wall combined with aseries of rooms. This wall forms the inner line of defense. A thirdwall into which was also combined a number of dwellings quite as inthe pueblo architecture of the region.Another unusual feature of the houses in and about this acropolisis the underground rooms which almost every house possesses? a type of cellar. This recalls the feature of Old Caves Pueblo 4 milesto the west as described by Fewkes (1904). There the rooms werehollowed out of the half-fused scoria; here at Red Peak the rooms are COLTON] PKEHISTORIC SITES NEAR FLAGSTAFF 27 excavated in the soft cinders, necessitating retaining walls. Thisfeature is unique.Although no kivas are recognized on Red Peak, yet a large building24 by 36 feet may have played a part in the ceremonies of the inhabi-tants. This large single-roomed building recalls similar rooms atZuiii, in Verde Valley (Mindeleff, 1896), and in the Tonto Basin(Fewkes, 1904), supposedly used for purposes of ceremony.We can not leave Red Peak without some reference to its watersupply. Lying as it does between two lava basins of Locketts Tankand Turkey Tanks, it is over 2 miles from the former and 3K milesfrom the latter. It is possible that the curious inclosure in the centernow filled with drifted sand is a reservoir or tank.The extensive acropolis of Red Peak is useless for any economicpurpose and with Old Caves it should be set apart as a NationalMonument.Like other sites in this region, the pottery complex consists ofPueblo II and Pueblo III. FORTRESS HILLSOn the east Gourd Flat is bounded by a row of four small volcanichills, the origin of which is not very clear. The second hill from thesouth end is crowned hj a well-built fort (Colton, 1918), while theslopes are dotted with depressions bearing potsherds of Rio de Flagbrown ware. Since this region has already been fully described,nothing more need be added except that the earth lodge sites extendnorth onto the lava flow which issued from O'Neil Peak. All thesesites belong to the culture horizon of Pueblo II.Turkey TanksBelow the junction of the Rio de Flag, Walnut Creek enters a smalllava canyon and like most of such canyons this contains pools ofpermanent water. Where water abounds ruins will be found.Turkey Tank caves in section 26 were described by Fewkes (1900).The fort and ruins near the lower tank are figured and described byColton (1918, a and b). These belong to the Pueblo III culturehorizon. Earth lodge sites Pueblo II are scattered along the hill-tops; on all of these Rio de Flag brown pottery is common. In thecanyon east of Turkey Tanks lie a few cavate dwellings.Grass Tank and Pipers CraterEast of Turkey Tanks, Walnut Creek enters a deep lava canyon,containing a few pools of water, and finally discharges on a widelava, cinder, and cedar-covered plain. In a shallow canyon.in theplain lies Grass Tank. Below Grass Tank the creek enters a limestonecanyon in the plain. In the cedars about Grass Tank ruins are quite 28 BUREAU OF AMERICAN ETHNOLOGY [bull. io4numerous; Pueblo III masonry houses and Pueblo II pit-house sitesare abundant.The most notable Pueblo III ruins in this region are as follows:(a) An eyrie little ruin of five rooms (No. 475) perched on a pre-cipitous spatter cone in section 31.(6) A 7-room house (No. 473) with kiva and small circular building4 feet in diameter. The author excavated this latter to the depth of4 feet and found the stones of the east side resting on a burial. Thishe left intact. Such small circular underground chambers are fairlycommon in connection with ruins in the cedars. The author makes nosuggestion as to their function.(c) The "Ridge Ruin" (No. 1785) in section 7 was discovered byMr. J. C. Clarke. For this region with some 30 rooms on the firstfloor, it is one of the larger pueblos. Traces of two kivas are evident.(d) A well-preserved ruin built of lava blocks covers the top of thewestern rim of the crater northeast of Pipers Ranch. It is a sitecomparable to the acropolis of Red Peak but much smaller. Thewalls, nothwithstanding their laval construction, are in a good state ofpreservation. Cavite dwellings and boulder sites scatter down thesoutheast end of the ridge.(e) One-half mile east of No. 473 on the brink of the canyon ofWalnut Creek lies an interesting ruin (No. 885) with 11 rooms on theground floor.(/) In a lava cave on the north side of the canyon below No. 885lies a well-preserved kiva (No. 886) with the roof about one-fifthintact. As it shows the structure of the kiva roof it is worth mention.It is about 9 by 9 feet with 6 feet depth. A ventilator and shafthave been exposed by Mr. J. C. Clarke. A section of the kiva canbe seen in Figure 21. Very few potsherds were found.(g) East of the kiva, in another similar cave, is a small, well-pre-served cliff dwelling (No. 887).(h) One hundred feet east of ruin No. 885 lies a bowlder site withDeadmans Flat black-on-white, fugitive red, and walnut red.Pueblo II.(i) Ruin No. 521 on the north side of Walnut Creek about a milebelow Grass Tank has about seven rooms and in 1925 the authorexcavated the kiva, 15 feet 9 inches by 12 feet 6 inches, exposing aventilator 8 inches by 9 inches and deflector 1 foot by 2 feet on theeast side. In front of the deflector lies the fireplace, 10 by 12 and 10inches deep.ij) One mile north of the Ridge Ruins on the edge of the samelava flow lies a pueblo ruin wliich we have called Two Kivas (No. 700).It is a little smaller than the Ridge Ruin but similar to it in manyways. coLTON] PREHISTORIC SITES NEAR FLAGSTAFF 29Grass Tank and Youngs Canyon have evidently furnished waterfor the inhabitants of this region. Other tanks in Walnut Creek,now dry, seem to account for the concentration of sites in certainplaces. Among the Big Cinder ConesAfter alternately passing tlii-ough limestone and lava canyons.Walnut Creek permanently enters the limestone. Where it passesits last lava flow pools of water remain. These pools constituteWalnut Tanks.About Walnut Tanks and on the west flanks of Merriam Crater liea few small pueblo ruins and earth lodge sites. They are notablebecause they were built in a treeless, waterless waste. Althoughbut few have been recorded, sites have been found wherever lookedfor between the cinder cones and the Little Colorado, but they arefew and far between. TOLCHACOThe region about Tolchaco, the old crossing or ford of the LittleColorado, much used before the bridge was built at Leupp, containssome interesting ruins. One only, the "Burned Ruin," was visitedand measured.On a red sandstone rock overhanging the river this little buildingwith 19 first-floor rooms is interesting. Having been destroyed byfire when stored with corn, the intense heat fused the clay or abodefloors, forming a kind of slag frequently called lava by the uninitiated.The ground all about is covered with carbonized ears of corn, slag,and much pottery scrap.The pottery is distinctly that of the southern complex: Eldencorrugated ware and Little Colorado black-on-white, and graycorrugated ware. It is interesting because it lies on the edge of aculture area. Although other ruins were not visited in the region,the author has been informed that they exist. Late Pueblo III.BoNiTO ParkAs the Rio de Flag and Walnut Creek carry us from the pine forestsof the mountains to the deserts along the Little Colorado near Leupp,so another similar series of sites leads us from the pines on the north-east side of the San Francisco Peaks to the Little Colorado near BlackFalls. While the first series of sites is over 40 miles long, the seriesfrom Bonito Park to the river is less than 30 miles. In the pines westof Bonito Park near Jack Smith's Tank lie scattered a large numberof sites of earth lodges. Some are vestibule houses and others arepit houses. The associated pottery, Deadmans black-on-white,Deadmans gray, Deadmans black-on-red, Deadmans fugitive red. 30 BUREAU OF AMERICAN ETHNOLOGY [buil. io4One of the earth lodge sites (No. 409) the author excavated in 1928,locating the fireplace. (For details see fig. 12.) In the summer of1930 the Museum Expedition under Lyndon L. Hargrave excavateda large number of sites in this region. This area will be treated in aseparate paper. Deadmans FlatDeadmans Flat contains two large groups of ruins, one cut by theold Grand Canyon Road and the other just east of the Tuba Roadand south of Dove Tank. Here masonry granaries are associated withearth lodge depressions (pit house). The pottery is of the character- ^-> H o I 2 3 4 5 rr-Figure 12.?Plan of earth lodge 409. The "vestibule," socalled, seems not to have been an entrance but was used as astorage alcoveistic Deadmans Flat Pueblo II complex, in which plain gray warespredominate.Although a large number of ruins have been measured and some areof fair size, no one is deemed worthy of special description.Since aU the ruins are located along the two washes that headnorth of Sugar Loaf Crater, it would seem to indicate that springsonce existed which fed a stream running out from the mountain.An ancient dam was visible in 1923 across the wash near site No. 191.Cedar Ridge and the Lava TerracesAfter crossing Deadmans Flat the Tuba City Road climbs overCedar Ridge. Off to the east lie a few scattered sites. None are colton] PREHISTORIC SITES NEAR FLAGSTAFF 31large and none are notable in any way. Continuing north, the TubaRoad drops off the Cedar Ridge lava flow on to a lower flow, the fourthterrace. Here ruins are very frequent and some are quite extensive.Nowhere else in the region can one better study the evolution of the ^'^'"|""V""|""1A\0UNPI'Sfovy; rioFtHwK --'? I? r *": ' -d ' ' ' '? tn' Wo sc?U Me scal ;;^ ^^ 474 A^s ??1 1 ? 32 BUKEAU OF AMERICAN ETHNOLOGY [bull. io4 of ruin No. 438 is an oval depression surrounded by a mound of exca-vated material. This depression has a north-south axis about 100feet long and an east-west axis about 60 feet. This resembles thebowl at Casa Grande and at Wupatki.Below the fourth terrace lies the third terrace. Although struc-tures of two or three rooms are fairly abundant, none require a specialdescription.The second terrace lies north and east of the third. Three ruinsare large enough to deserve special mention. No. 574, a mound 10feet high, 18 to 20 rooms, 2 story, built about a court; No. 578, 7rooms; No. 579, 8 rooms, all built about courts. The lower lavaterrace lies to the east of the second, third, and fourth and extendsnearly to Hulls Canyon. Although small ruins dot its surface nonerequire particular mention.Big Hawk ValleyNorth of the terraces the old Tuba Road crosses a broad lava-filled valley while the new Tuba Road skirts it on the east side. Wecall this Big Hawk Valley. Crossing it on the old road one gains theimpression that the valley is covered with ruins. However, a carefulsurvey shows about 25, all located in a band about three-quartersof a mile wide and 2 miles long. These ruins are mostly 2 to 4roomed stone houses with lateral walls embracing a southeasterncourt. A kiva could be inferred in some cases by a depression in theground in the court, while northwest or northeast another depressionseemed to represent a reservoir or tank or merely the excavation outof which the clay and stone were removed for building purposes.(Fig. 13.) Plans Nos. 678 and 682.All of these sites are covered with Pueblo III potsherds except one,No. 618 (fig. 14), which is very much like a boulder site in the lowerTonto Basin. The pottery, however, is of the characteristic PuebloII of the region. Hulls CantonBig Hawk Valley has one branch extending northeast into theCitadel fiat, but the drainage does not flow that way. The washdraining the terraces swings over the south mesa near the Citadel,then turns southeast into Hulls Canyon. Here are located 10 moresites. These ruins parallel the canyon, which is very shallow andfilled with black volcanic sand. They rest on the limestone floor asthe lava cap ends at the edge of the lower terrace a mile to the south-west. The ruins are built of limestone. A deep fissure in the lime-stone forms an interesting feature of this region. A mile southwest ofthe ruins in Hulls Canyon lies the Red House, No. 590, a 3-roomed,2-story house, the central room of which has been restored by somesheepman and roofed. It lies on the Moencopi sandstone in the BUREAU OF pt^ ?^ P*758West W cmRuins No BUREAU OF AMERICAN ETHNOLOGY BULLETIN 104 PLATE 1 colton] PEEHISTORIC SITES NEAR FLAGSTAFF 33 cedars, surrounded on three sides by lava cliffs of the lower terrace.The remains of a kiva (?) lie to the cast of the house. The presentwalls of the central room are lower than the original 2-story house buthigher than the original first floor, which must have had a very lowheadroom, judging by the sockets for the joists. The present door-way is new. It is interesting to note that the joists were made ofCottonwood. No cottonwood is found at present nearer than the ^618. riii-.-.-JJ 3i Feet-FiGUKE 14.?House 618. A Pueblo II site in Black Hawk ValleyLittle Colorado River, 15 miles away. A more careful exploration ofthe cedars in this region would no doubt reveal other sites.CitadelNortheast of Flagstaff, on the steppes forming the western portionof the Painted Desert, are two curious sinks about 100 yards across,the geologic origin of which is not clear. Both sinks seem to be onfault lines. They are on the eroded edge of a basaltic lava flow,probably of the first period of eruption of the San Francisco volcanicfield (Robinson, 1912). In each sink the lava lies on a thin bed ofMoencopi red sandstone underneath which the Kaibab limestone isexposed for 50 feet or more. About both of these sinks the earlyinhabitants of the region built dwellings. The first, Citadel sink,located in township 25 North, range 9 East, section 7, lies on the south- 34 BUREAU OF AMERICAN ETHNOLOGY [bull. io4 ern edge of a large group of ruins described in part by Fewkes (1897and 1904) and Barrett (1926). The second sink (township 25 North,range 9 East, section 19) has about it fewer ruins; some cliff dwellingsinside (about six rooms) and some small houses on the rim.Neither within nor without the sinks is water at present found.Indeed, the whole region is so waterless that the cattle and sheepcan not use the range, although it produces excellent grass. Cattle-men and sheepmen, attracted by the ruins, have made serious attemptsto develop water, but the ancient source is still a mystery.A drainage system, taking origin in the cinder cones 10 miles tothe west of the sinks, flows in shallow canyons in the lava eastward,passing through Big Hawk Valley to the north of the second sink andto the south of the Citadel sink. Turning south, the wash entersHulls Canyon, a cinder-filled valley, and diverges into DeadmansWash. From the Tuba City Road to Deadmans Wash its dry sandybed is bordered by ruins on either side and at one place near the cross-ing of the wash by the Old Tuba Road, Navaho teamsters are saidto get water by digging in the sand. This spot is called "Aztec Seep."It is possible, therefore, that this wash once contained a large under-flow of water.Again, although not visible in many places because of the talus,the water-bearing Moencopi sandstones and shales outcrop on theedge of almost all the lava caps. Here springs may have once existedas they do now under similar conditions farther southeast in the Wu-patki and Rodin Basins.Be that as it may, the only positive sources of water are a tank onthe top of Magnetic Mesa, a tank south of Ruin No. 360, and two orthree dams in the broad depression in Antelope Prairie, three-quartersof a mile east-southeast of the Citadel ruin.Rising above the northeast rim of the first sink, a little mesa of redsandstone, capped with basalt, supports an impressive ruin having30 rooms on its ground floor, called by Fewkes the Citadel. (Fig. 15.)In the center of the patio is a depression which may represent a kiva.The hillside to the east is terraced and two other sites, 357 and 358,lie at the base. The culture horizon of the Citadel, as well as all theother standing ruins in this part of the Wupatki National Monument,is that called Pueblo III. However, Pueblo II sites are common,but in most cases have been reoccupied in Pueblo III times.North and west of the Citadel (map, pi. 5) are two shallow can-yons intersecting and forming a letter T. The southwest arm receivedthe drainage from a limited area to the southwest but may have oncereceived the drainage from Aztec Seep, as the dividing line betweenthe two systems is very low. The arm called Cedar Canyon drainsto the northwest, while the northeast arm drains into a blind basinwithout any outlet. The drainage, therefore, is peculiar. colton] PREHISTORIC SITES NEAR FLAGSTAFF 35Coupled with this*peculiar lay of the land, mention must be madeof a number of fault lines; a system running north-northwest andsouth-southeast is crossed by a system running at right angles to it. ^OFT- Figure 15.?The Citadel, 355, occupying the top of a butte, the eroded remnant of an ancient lavaflow. The Citadel is a landmark on the western edge of the Painted Desert and is the center aroundwhich are grouped large numbers of other sites. It lies in the Wupatki National MonumentAt these faults the Kaibab limestone has weathered into deep fissuresand in places into little meaningless canyons.Scattered along the canyon rims and mesa tops are a large numberof ruins. Where the building material has been good, such as lime-stone or sandstone, the walls stand well. On the other hand, whenlava has furnished the material, little shows but a pile of stones. 36 BUREAU OF AMERICAN ETHNOLOGY [bull. io4Since Fewkes (1904) and Barrett (1926) described the majorruins, space forbids a repetition. However, mention must be made ofthe rectangular kivas which show particularly well at certain places,viz, Ruin No. 373 and Ruin No. 379. At Ruin No. 377 a cattleman,mistaking the kiva for a well, excavated it. Not finding water, hecontinued digging. Now a shaft 60 feet or more deep reaches thetalus of the vallej^ floor, and represents a futile attempt to get water.The major ruins of the Citadel group with their well-preservedwalls should be protected from further inroads of wind and water.They are included in the Wupatki National Monument.Antelope PrairieEast of the Citadel, a broad grass-covered limestone plain extendsto the Doney Mountain escarpment. This plain is bounded on thenorth by North Mesa and on the south by the dense cedars betweenHulls Canyon and Doney Mountain. Fourteen ruins were locatedhere, all of the familiar type of 2 to 8 roomed stone houses. Damswere found in the shallow wash that follows the foot of North Mesa.This part of the present dry range is grazed by antelope only; waterbeing too far away for stock and rainfall too light to fill artificialtanks. (Fig. 13, sites 652, 678, and 682.)Wupatki BasinA geographical region of considerable unity comprises the red rockcountry on the west side of the Little Colorado River extending fromRodins Spring and the near-by crater on the south to Black Point onthe north (map No. 2, pi. 5). We call this region the Wupatki andRodin Basins. Although the red Moencopi sandstones and shalesoutcrop in a narrow band along the Little Colorado for many miles,here they widen out. The country is wild and desolate; mesas redand bare rise above shallow canyons floored with black volcanic sand.To the east, cliffs appear above the cottonwoods along the river.The Doney Mountain escarpment and great lava-capped cliffs, halfburied in black sand dunes, bound the basin on the west.On account of the w^ater-holding quality of the Moencopi rocks,the ancient people built numerous houses. Because the buildingmaterial was good and the rainfall light, these houses have stood well.Here are found the best-preserved ruins in the whole Flagstaff region.The basin is hard to get at. Although the northern portion atCrack-in-the-Rock can be easily reached by automobile, and thesouthern portion approached on foot from Grand Falls or RodinsSpring, yet long stretches of black sand or precipitous lava cliffsblock every approach to the central and most interesting region. Thehorse, therefore, is still the preferred mode of transportation to theruins of the Wupatki Basin. colton] PREHISTOEIC SITES NEAR FLAGSTAFF 37The Wupatki Basin takes its name from the principal ruin (fig. 16),called by J. C. Clarke, Wupatki (Hopi for long house). This doubleruin, first pictured by Sitgreaves in 1853 and later described andfigured by Fewkes (1904) and called by him Ruin A, Group B, is the ;>oott. NNWwAijuer) IN ov;tuIOOft.X4/ft .