Devil’s Graveyard Formation (New) 
Eocene and Oligocene Age 
Trans-Pecos Texas 

JamesB. Stevens 
Margaret S. Stevens 
John A. Wilson 

Texas Memorial Museum 
Bulletin 32 
The University ofTexas at Austin 

© 1984 by Texas Memorial Museum 
TheUniversityofTexas atAustin All rights reserved Published September 24,1984Printed in the UnitedStatesofAmerica 
ISSN 0082-3074-32 
A portion of the museum’s general operatingfunds for this fiscal year has been providedthrough a grant from the Institute ofMuseum Services, a federal agency that offers generaloperating support to the nation’s museums. 
The Bulletin is an irregularly published series of technical monographs deriving primarily from research done onTexasMemorialMuseum collections and projects. 
•
TexasMemorialMuseum TheUniversityofTexasatAustin •2400Trinity •Austin,TX78705 
CONTENTS Abstract 1 Introduction 1 Abbreviations 2 Acknowledgments 3 lithostratigraphy, Buck Hill Group 3 Devil’s GraveyardFormation(new) 4 Lower and middle members 4 BanderaMesaMember(new) 5 Marker beds, lower member 5 Basal Tertiary conglomerate 5 Variegated beds 6 Lunch Locality sandstone 6 Strawberry tuff 6 Marker beds, middle member 7 Titanothere channels and Purple Bench tuff 7 Repeats 7 Marker beds, Bandera Mesa member (new) 7 Skyline channels 7 Cotterchannels 19 Upper breccia-conglomerate 19 UpperpartofBanderaMesaMember(new) 19 Summary 20 References 20 
ILLUSTRATIONS 
Figure 	Page
1. 	IndexmapofTexas andPresidio andBrewsterCounties 8 
2. 	Sketch map ofTrans-Pecos Texas with volcanic centers 9 
3. 	Geologic map 10 
4. 	Diagrammatic cross section, Devil’s GraveyardFormation(new) 12 
5. 	Diagrammatic stratigraphic section 13 
6. 	Measuredsection,lowermember,lowerboundarystratotype 14 
7. 	Measuredsection,lowerandmiddlemembers 15 
8. 	Measured section, southern part Bandera Mesa Member (new), North Fork Alamo de Cesario Creek 16 
9. 	Measured section, middle and Bandera Mesa Members (new), South Fork Alamo de Cesario Creek 17 
10. Measuredsection,BanderaMesaMember(new),upperboundarystratotype.... 18 

DEVIL'S GRAVEYARD FORMATION (NEW)
EOCENE AND OLIGOCENE AGE 
TRANS-PECOS TEXAS 

James B. Stevens 1 
Margaret S. Stevens2 
John A. Wilson 3 

ABSTRACT 
The Devil’s Graveyard Formation (new, Eocene and Oligocene) is described as that part of the Buck Hill Group above the Cretaceous and beneath the Mitchell Mesa Rhyolite or the Yellow conglomerate of Moon (1953). It replaces the terminology used by Moon (1953) for the lower part of the Buck Hill Group,“Buck Hill Group undifferentiated,” as well as “Pruett-Duff” (Erickson, 1953, and Stevens, 1978) and “Pruett and Duff undivided” (Barnes, 1979, and Henry and Duex, 1981). The names “Pruett” and “Duff” apply to the northern area in the southern part of the Davis Mountains where their type sections are located. The lithology of the Devil’s GraveyardFormation is predominately clastic and sufficientlydifferent from that of both the Pruett and Duff formations to warrant its new name. The Buck Hill Group in the Agua Fria-Green Valley area consists of the Devil’s Graveyard Formation, the Mitchell Mesa Rhyolite, the Tascotal Mesa Formation, and the Rawls Basalt. The Devil’s Graveyard Formation is divided into lower and middle unnamed members and an upper Bandera Mesa Member, each separated byconcentrations of channel conglomerates. Locallyuseful marker beds that are associated with vertebrate fossil localities are described. The areas of exposure are in west-central Brewster and east-central Presidio 
counties, Texas. 
INTRODUCTION A thick body of Eocene and early Oligocene volcani­clastic fluvial and lacustrine sedimentary rock is exposed in the Big Bend country of Trans-Pecos 
Texas west of Texas highway 118, in west-central Brewster and adjacent Presidio counties (fig. 1). The area includes contiguous parts of the Agua Fria,Tascotal Mesa, Jordan Gap, and Buck Hill fifteen-minute topographic quadrangle maps (fig. 2) where sediments of the Pruett and Duff formations, Buck Hill Volcanic Group (Goldich and Elms, 1949) are exposed. The area lies south and west of outcrops of the Cottonwood Spring Basalt, the unit that defines thetop ofthePruettandbase oftheDuffformations 
(Goldich and Elms, 1949), and embraces an undiffer­entiated stratigraphic section either placed within the Buck Hill Volcanic Group by Moon (1953), regarded as Pruett-Duff by Erickson (1953) and Stevens (1978), or called “Pruett and Duff undivided” byBarnes (1979) and Henry and Duex (1981).
The study area receives about 12-14 in. (30.5-35.6cm) of precipitation annually, hence is sparsely vege­tated. Although Pleistocene stream and sediment deposits mantle the older rocks in certain areas, much of the Eocene and early Oligocene deposits are well exposed. Among the present investigators, Wilson and 
M. Stevens have found abundant vertebrate fossils as well as nonmarine invertebrates in these deposits.Furthermore, the writers have all found these sedi­mentary rocks to contain an extensive radiometricallyand biostratigraphically datable record of initiation and progressive development of volcanic and post-Laramide tectonic activity for the Trans-Pecos area. In view ofthis and because the lithology in the Devil’s Graveyard area is distinctive we have decided to name and describe a new unit, herein designated as the Devil’s Graveyard Formation, to clarify stratigraphicand biostratigraphic relationships of these rocks and 
to help clarify the regional geology and facilitate correlations. 
'Assistant Professor of Geology,Departmentof Geology,Lamar University,Beaumont, Texas. 
2 Instructor of Geology, Lamar University of Orange County, Orange, Texas, and Lamar University, Port Arthur, Texas,
andResearchAssociate, TexasMemorialMuseum,The UniversityofTexasatAustin. 
3 Professor Emeritus, Department of Geological Sciences, and Research Associate, Texas Memorial Museum, The Uni­
versity of Texasat Austin. 
TexasMemorialMuseumBulletin No.32 
The best exposure of the Eocene rocks of the Devil’s Graveyard Formation occurs along Alamo de Cesario Creek (fig. 3), a major tributary of TerlinguaCreek, in an area called the Devil’s Graveyard (Moon,
1953:pi. 1). The Devil’s Graveyard, approximately6 by 3 mi (9.7 by 4.8 km) or about 18 square mi (29 square km), lies within a graben and is bounded on the northwest by a highly irregular line of pediment-capped cliffs and steep slopes 115 to 295 ft (35 to 
90 m) high. 
Eocene and Oligocene rocks closely related to those exposed in the Devil’s Graveyard occur discon­tinuously in an area over 43.5 mi (70 km) long alongthe Presidio-Brewster county line from Mitchell Mesa 
on the north to the Rio Grande on the south, and almost 18.6 mi (30 km) wide measured from Buck Hill on the east to the face of Bandera Mesa on the west. Areas of notable exposure of sediments directlyrelated to the Eocene rocks and making up most of the Tertiary deposits in the Devil’s Graveyard include the area approximately 1.8-3.1 mi (3-5 km) south-southeast of the Graveyard at and near the base of the northern and western sides of Agua Fria Moun­tain (fig. 3), an important local landmark. The follow­ing areas are defined according to direction and dis­tance away from the Devil’s Graveyard. Early Oli­gocene sediments are best exposed along Needle and Dogiecreeks, 10.8mi(17.5km)northwest,aswellas along drainages cutting the eastern face of Bandera Mesa, particularly near Puerto Potrillo, 21 mi (34 km)northwest. Oligocene rocks are also exposed around adjacent outlying buttes such as Red Hill, Church Mountain, and McKinney Mountain (fig. 3). 
The volcanic rocks of Trans-Pecos Texas have received considerable study during the last 25 years,and their distribution and composition havebeen the subject of numerous theses, dissertations, and publi­cations. The volcanic centers (fig. 2) are now known (Gorski, 1970; Parker, 1976; Cepeda, 1977, 1979).Much of the formal stratigraphic nomenclature of this region has grown by the naming of flow rocks, some of which define the top and bottom of sedimentaryunits, but until lately little attention has been given
the sediments themselves. Exceptions include studyof volcaniclastic deposits in the Castolon area, BigBend National Park (Stevens, 1969,M. Stevens et al.,1969); of sediments within the Vieja Group north­west of Presidio (Walton, 1972; 1975; 1978; 1979);of the Tascotal Formation (Walton, 1978; 1979);andof sediments of the Rawls Formation (Robinson,1976). Schiebout (1974) dealt only with older, pre­volcanic Tertiary deposits. One reason why emphasishas been placed on the volcanic rocks is that some are excellent marker-beds with extensive lateral conti­nuity and have been used to subdivide the strati­graphic section. For example, as originally defined,the Duff Formation of the Buck Hill Group “. . . is chiefly rhyolitic tuff with minor breccia and conglom­erate. The tuff lies on the Cottonwood Spring Ba­salt and is overlain by the Mitchell Mesa rhyolite. . .” (Goldich and Elms, 1949, p. 1159). Somewhat differ­ently, Maxwell et al. (1967) named flows within the Chisos Formation, Big Bend National Park, as formal 
members but left the intercalated sediments unnamed. The techniques of defining a lithologic unit only on the basis of bracketing flows or by namingonly flows within the section make it difficult to direct attention to the sedimentary units without a long descriptive phase, or to correlate these units accurately when the flows are absent. 
We therefore believe that the vertebrate fossils and the radiometric dates now known for volcaniclastic sediments of the Buck Hill Group clarify the strati­
graphy of the extensive sediments below the Mitchell Mesa Rhyolite in the areas of the Jordan Gap, Buck Hill, Tascotal Mesa, and Agua Fria quadrangles where the Cottonwood Spring Basalt is absent. 
In the absence of useful igneous marker beds beneath the Mitchell Mesa Rhyolite to the south of the present distribution of the Cottonwood SpringBasalt (fig. 4), it is necessary to establish the relative positions of geographically separated fossil sites on the basis of sedimentary stratigraphy. A new forma­tion with two unnamed lower members and a named upper member is here formally proposed. Additional informal units below the rank of member but useful in local correlation are described. 
Tracing of the sedimentary units and measuring of stratigraphic sections was done mainly by James B. Stevens and Margaret S. Stevens. Petrography and sedi­mentologic interpretation were the responsibility of 
J.B. Stevens except where otherwise noted. The fos­sils were discovered and collected primarily by R.H. Rainey, M. S. Stevens, and J.A. Wilson, and identified by Wilson unless otherwise noted. A brief lithologicdescription of each marker bed is given for purposesof field identification. Attempts were made to relate each major collecting locality to marker beds (fig. 5).
However, the vertical and horizontal extensions of the volume of rock partly occupied by the vertebrate faunas do not necessarily coincide with and are not to be interpreted as coincident with the lithologic unit. 
Enlarged aerial photographs 27 square in. (68.6 square cm) were used as base maps for field work. Fossil localities were plotted on these photographs,whichhave beenfiled inthearchivesoftheVertebrate Paleontology Laboratory, Texas Memorial Museum,The University of Texas at Austin. Geology is in partmodified from Goldich and Elms (1949), Seward (1950), Moon (1953), and Erickson (1953). Strati­graphic sections were measured with Brunton com­pass and steel tape. 
ABBREVIATIONS 
ACSN American Commission on Stratigraphic Nomenclature ISSC International Subcommission on Stratigraphic Classification lUGS International Union of Geological Sciences K-Ar potassium-argonMS measured section(s) my millions ofyearsTMM Texas Memorial Museum 
1984 Devil’s Graveyard Formation—Lithostratigraphy 3 
ACKNOWLEDGMENTS 

This investigation was made possible by the financial support for field work from the Vertebrate Paleontol­ogy Laboratory, Texas Memorial Museum, The Uni­versity of Texas at Austin and from Faculty Research Grants awarded to Margaret Skeels Stevens by Lamar University, Beaumont, Texas. Additional aid, primar­ily in theformofsalaryfor JohnA.Wilsonprior to 1976, was received from the Geology Foundation,Department of Geological Sciences, The University of Texas at Austin. The Owen-Coates Fund of the Geol­ogy Foundation defrayed part of the publication expenses. Thin sections and K-Ar dates were made available through the facilities of the Department of Geological Sciences and the Geology Foundation. The Bureau of Economic Geology furnished financial aid for part of one summer and the drafting of a manuscript geologic map of the Jordan Gap Quad­
rangle. We are grateful for this continued support.
Mr. and Mrs. Billy Pat McKinney, lessees, and J.H.Burton and Macon Richmond, owners of the AguaFria Ranch, provided us with invaluable assistance inthe form of shelters and legendary western hospitality.The upper reaches of arroyo Alamo de Cesario extend into the ranch of Dr. Walter W. Dalquest of Midwest­ern State University, Wichita Falls, Texas, who gen­erously permitted us to collect vertebrate fossils from his land. We are grateful for access to the San Jacinto Ranch, M-Ranch, Coffee Cup Ranch, and the Mont­gomery ranches in the Jordan Gap and Crystal Creek area, owned or leased by Clegg and D’Ette Fowlkes,who also most graciously provided shelter. Sincere thanks also go to Earl Hammond who allowed us access to the Rancho Mortillo, and Duncan Cooperwho allowed us to collect on El Rancho Triste (for­merly 3-Bar Ranch) in the Hen Egg Mountain area. 
Robert H. Rainey, chief preparator at the Verte­brate Paleontology Laboratory, assisted during some of the field seasons and was co-discoverer with John 
A.Wilson of theWhistler Squat local fauna. In 1971,Dr. Ann Marie Forsten, Sally Rogers Scanlon, and Richard P. McCulloh were able field assistants. Mar­
garet S. Stevens discovered numerous important fossil sites including Boneanza, Serendipity, Titanothere hill, Skyline, and significant localities at Red Hill,near the old Cotter Ranch headquarters, and Mont­gomery bone bed. J.A.Wilson made an important
discovery at what has been dubbed “Purple Bench.” 
Dr. Wilson is grateful to his colleagues in the Department of Geological Sciences, The Universityof Texas at Austin, for helpful discussions. Specialthanks are given to Dr. S. E.Clabaugh and Dr. Donnie Parker for help with the igneous rocks, and Dr. Fred 
W.DowellforK-Ar dates.J.B.Stevenshashadseveral valuable discussions with Dr. A. W. Walton, Depart­ment of Geology, University of Kansas, concerningthe diagenesis of certain sedimentary rocks in and around the study area, and the origin of depositionalepisodes.
An extremely important and parallel study on potassium-argon dating of rocks of the Trans-Pecos volcanic field has been progressing under the direc­tionofDr. FredW.Dowell(1979)oftheDepartmentof Geological Sciences, The University of Texas at 
Austin. A summary and evaluation of some 81 perti­nent K-Ar dates is given by him, but not all of them were determined at The University of Texas. Other K-Ar dates indirectly applicable to this study are given by Gilliland and Clark (1979), Daily (1979),Parker (1978), Parker and McDowell (1979), HenryandMcDowell(1982),Keller etal.(1982), andMaugerand McDowell (1982). Data from the above sources, as well as dates commissioned from Geochron Labo­ratories during the progress of this study, have been incorporated into the diagrammatic section (fig. 5).All potassium-argon dates in this report have been recalculated using the tables of Dalrymple (1979)according to the recommendation of the lUGS. 
The authors are grateful to Drs. R. J. Emry, C. D. Henry and Amos Salvador for reading the manuscriptand making helpful suggestions. Editing, typesetting,and layout were done by Maria de la Luz Martinez. 
LITHOSTRATIGRAPHY 
BUCK HILL GROUP 

The Buck Hill Group was formally defined byGoldich and Elms (1949) and contained, in ascendingorder, the PruettFormation whichincludesinitsupperpart the Crossen Trachyte, Sheep Canyon Basalt, and Potato Hill Andesite; the latter three units were mem­bers. The Cottonwood Springs Basalt was described as a formation overlying the Potato Hill Andesite Mem­ber of the Pruett Formation. The Duff Formation overlies the Cottonwood Springs Basalt, and the latter in turn is overlain by the Mitchell Mesa Rhyo­lite. AnunnamedunitwasincludedabovetheMitchell Mesa, and Goldich and Seward previously (1948) gave the name “Tascotal Formation” to the tuffs above the Mitchell Mesa Rhyolite and “Rawls Basalt” to the basalt above the Tascotal Formation. 
McAnulty (1955) elevated the igneous members of the Pruett Formation to formations and restricted thePruett Formation to the tuffaceous sediments below 
the Crossen Trachyte and above the Cretaceous. Because none of the igneous flow units below the Mitchell Mesa Rhyolite extend southward beyond the west-central part of the Buck Hill Quadrangle, or east-central part of the Jordan Gap Quadrangles, it provedimpossible for Moon (1953) and Erickson (1953) to map subdivisions of the Buck Hill Group below the Mitchell MesaRhyolite. Moon (1953) used “Buck Hill Volcanic Series” without any formal subdivisions,and Erickson (1953) attempted a breakdown using“Pruett Formation” as a lower unit and “Duff Form­ation as an upper unit, with a transitionalunit mapped as “Pruett-Duff Tuff” and discussed as “Pruett-Duff.” In the explanation for his map, Erickson shows the contact between the Pruett and Duff Formations as an arbitrary, roughly north-south dashed line, with Pruett-Duff indicated by a blending of the distin­guishing colors. No line of contact between Pruett and Duff formations is shown on the map. Instead, areas of blended color indicate where Pruett-Duff meets the Pruett and Duff (formal) Formations. The areas of contact along indefinite lines do not reflect the attitude of the beds. 
Earlier papers on the vertebrate fossils from the 
Agua Fria Quadrangle by Wood (1972, 1973), Wilson 
No. 32
Texas Memorial Museum Bulletin 
(1972, 1974, 1977), and Wilson and Szalay (1976)have referred to the fossils as having been collected from the Pruett Formation. However, as work pro­gressed it became impossible to prove that the sedi­ments from which the specimens had come were, in fact, Pruett Formation. Some vertebrate fossils have been found at and near the base of the Buck Hill Group as mapped by Moon (1953). Others have been collected from sediments mapped by Erickson (1953) as “Pruett-Duff tuffs” and still others from Erickson’s Duff Formation. Parts of all of these units contain the same Whistler Squat local fauna. Like Erickson and Moon, we have found that a boundary separatingPruett from Duff Formation is not mappable south of the limit of the Cottonwood Spring Basalt. For this reason we propose a new unit. 
Devil’s GraveyardFormation*(new) 
The name Devil’s GraveyardFormation is given to the predominantly volcaniclastic rocks within the Buck Hill Group, from Cretaceous rocks to the base of the overlying Mitchell Mesa Rhyolite, or to the base of the informal Yellow conglomerate (Moon, 1953), and south of an arbitrary vertical cut-off defined by the southernmost outcrops of the igneous members of the Pruett Formation as defined by Goldich and Elms (1949). The upper boundary of the Devil’s GraveyardFormation is marked by the base of the Mitchell Mesa Rhyolite, well exposed along Bandera Mesa, McKinney Mountain, and elsewhere in the Jordan Gap Quad­rangle. At the southeast end of Bandera Mesa, Tasco­tal Mesa Quadrangle, a conglomerate that is correla­tiveandlithologically similartoMoon’s(1953)Yellow conglomerate appears below the Mitchell Mesa Rhyo­lite (fig. 4). South and east of that part of Bandera Mesa in the Agua Fria Quadrangle, the Mitchell Mesa Rhyolite thins and becomes discontinuous as channel fillings within the Yellow conglomerate. The Yellow conglomerate was described and informally named byMoon (1953) who mapped it as part of the Buck Hill “Series.” Erickson (1953) described and mapped the lateral continuation of this unitas part of the Tascotal Formation. Where it underlies the Mitchell Mesa Rhyolite it was designated Duff Formation. The Yellowconglomerateislithologically distinctive, unre­lated to rocks of the Devil’s Graveyard Formation,and makes the most easily recognizable bounding unit for the Devil’s Graveyard Formation where it is pres­ent. It is presently intended that the northern bound­ary of the Devil’s Graveyard formation be defined by a line drawn from the east face of Bandera Mesa, 2 mi (3.2 km) northeast of Puerto Potrillo, to hill 4026, 
to Turney Peak, and then southeast to pass south of hill 4140 (west edge of Buck Hill Quadrangle, the southernmost outcrop of Cottonwood Spring Basalt),and then northeast to the divide south of Boat Moun­tain and Crossen Mesa. 
The type area (ISSC, 1976) is the Devil’s Grave­yard on the Agua Fria Quadrangle and its westward extension in the headwaters of Alamo de Cesario Creek and including the south face of Bandera Mesa on the adjacent (west) Tascotal Mesa Quadrangle. No two measured sections in this area are exactly the same and no single section is more typical than any 
other. The measured section shown diagrammaticallyin figure 6 (MS 15, fig. 3) includes the lowest beds found resting on Cretaceous rocks and is designated a lower boundary stratotype. Because it can be shown (fig. 4 and Wilson et al., 1979) that the Mitchell Mesa Rhyolite and Yellow conglomerate he on progres­sively older rocks to the south toward the Solitario uplift (fig. 4), the highest units of the Devil’s Grave­yard Formation are found to the north in the south-central and central portions of the Jordan Gap Quad­rangle. The northernmost section of the upper part of the Devil’s Graveyard Formation is below elevation 4357 just south of Puerto Potrillo. This section (fig.
10, shown as MS 19 in fig. 3) is designated an upperboundary stratotype.
Measured sections 16, 17, and 18 (figs. 7,8, 9) on figure 3 are diagrammatic representations of sections measured by the Stevens’ in the type area and are designated as reference sections for the formation. Figure 7 (MS 16, fig. 3) shows the lower, but not low­est, and middle parts of the formation; figure 8 (MS
17, fig. 3) shows the middle part of the formation on the North Fork, Alamo de Cesario Creek; figure 9 (MS 18, fig. 3) is, in effect, a continuation of the sec­tion shown in figure 8 and shows a different aspectof the upper unit, as compared to the upper boundary stratotype (fig. 10). Correlation of measured sections,admittedly not precise within the type area of the Devil’s Graveyard Formation, suggests that, conserva­tively, 1550 ft (472 m) of rocks belonging to the for­mation are exposed there. This isby no means a max­imum thickness of the formation, since much of the rock exposed at the upper boundary stratotype (fig.10) is not included in this thickness, as aresult of the 
tilting and beveling.
Lower and middle members, Devil’s GraveyardFormation.-The lower part of the Devil’s GraveyardFormation is not formally subdivided, but it has been convenient to divide it into two informal members,
the lower and middle members (fig. 5). The two are separated by a disconformity that can be traced throughout the type area and for some distance to the north along the eastern side of the known out­crop area of the Devil’s Graveyard Formation (fig. 3).In the eastern area, the disconformity lies above the 
Strawberry tuff (figs. 5,6, 7). In the southwest, the disconformity lies on the same marker bed, a distinc­tive brick-red mudstone to sandstone. The near coin­cidence of a marker bed with a traceable disconfor­mity makes recognition of the lower and middle mem­bers practical within and near the type area; but litho­logic distinction in the area of contact depends on subtle aspects, including patterns and style of sedi­mentation, patterns of diagenetic alteration, and petrography of the conglomerates. Thus formal status is inappropriate.
The measured sections can be used as examples of 
the rocks referred to the lower and middle members. 
The measured section of figure 6 (MS 15,fig.3) is rep­
resentative ofallbutthehighestpart ofthelowermem­
ber. The lower 150 ft (45.7 m) of measured section 
*Name approved for publication by Geologic Names Com mittee, U.S. Geological Survey. 
1984 Devil’s Graveyard Formation-Bandera Mesa Member 5 
16 (fig. 7) are in the lower member, and the suc­ceeding approximately 500 ft (152.4 m) to the base of the Bandera Mesa Member are measured in the middlemember. 
Bandera MesaMember *(new) 
The name Bandera Mesa Member is given to the rocks in the Devil’s Graveyard Formation extending from the base of the distinctive red-to-gray (rare) sandstone-and conglomerate-filled channels exposed at the topsof the cliffs at the northern margin of the Devil’s Graveyard, and in the canyon of the North Fork,Alamo de Cesario Creek, to the base of the Mitchell Mesa Rhyolite or Yellow conglomerate. A major sys­tem of channels informally called the Skyline chan­nels occurs at the base of the Bandera Mesa Member. 
Thus, the boundary is placed at the base of the lowest of these ledge-forming channels at any particular sec­tion. Where present, the channel fillings are invariably a major influence on the topography of the area. Examples of the channels are shown in figures 7,8,and 9. The arbitrary cut-off established for the north­ern boundary of the Devil’s Graveyard Formation is intendedto apply to the Bandera Mesa Member. 
The area extending southeast along the front of Bandera Mesa in the south-central part of the Jordan Gap Quadrangle from Puerto Potrillo to Red Hill,and then east across Green Valley to McKinneyMountain, is designated a type area (ISSC, 1976)for the Bandera Mesa Member. The measured section shown in figure 8 (MS 17, fig. 3) to 413 ft (126 m)above the zero point is designated as a lower bound­ary stratotype, and the section shown in figure 10 (MS 19, fig. 3) is designated as an upper boundary stratotype. 
The thickness of the Bandera Mesa Member mea­sured at the upper boundary stratotype is 789.3 ft 
(240.6 m). Approximately 6.5 mi (10.5 km) south­east along the front of Bandera Mesa, just north of Middle Canyon, Erickson (1953) measured 764 ft 
(232.9 m), and 5 mi (8.1 km) farther southeast,just north of Smuggler’s Gap, the Stevens’ measured (unfigured) 810 ft (246.9 m) in the Bandera Mesa Member. These sections are not laterally equivalentbecause of the northward tilting, and because of erosion of Devil’s Graveyard Formation prior to the extrusion of the Mitchell Mesa Rhyolite. None of these sections includes the lowest part of the Member, nor is precise correlation of the lower boundary stra­totype possible. J.Stevens considers 985 ft (300 m)to be a conservative estimate of the thickness of the Bandera Mesa Member. 
The basal part of the Bandera Mesa Member con­sists of pink to brick-red fluvial sedimentary rocks whenever exposed; the pink to dark brown, coarser sedimentary rocks exposed in the type area are also of fluvial origin. The southern part of the Bandera Mesa Member, as exposed in the face of Bandera Mesa, from North Canyon about 0.8 mi (1.3 km)south of the northern boundary of the Tascotal Mesa Quadrangle to the northwest side of the west­ern end of the Devil’s Graveyard, is strikingly dif­ferent. Much of the differenceresults from diagenetic 
processes (Henry and Duex, 1981), giving rise to white, gray, and pink rocks as hematite becomes markedly less dominant. It is also noteworthy that lacustrine sediments, including gypsiferous clays and limestones, form a prominent part of the southern exposures, while rocks in the more northern expo­sures are characteristic of steeper, smaller alluvial fans. Additionally, there are three sets of thin basalt flows occurring at 534 ft (162.8 m) above the base of the section shown diagrammatically in figure 5. The “basalts,” possibly mugearites or hawaiites in view of the work by Parker (1976) and Barker (1977), are found only in the northwest face of Bandera Mesa. 
Potassium-argon dates ranging from 43.9 to 49.7 my (fig. 5) on biotite and tuffaceous rocks in the lower part of the Devil’s Graveyard Formation agreeclosely with dates given in recent work by Mauger(1981), Capps (1981), Bockhoven (1981), and Maugerand McDowell (1982), who demonstrated the age and 
style of volcanism in several areas in central Chihua­hua. That area may have been a possible source for at least some of the large amount of volcanically-derivedsediment in the lower and middle members of the Devil’s Graveyard Formation. Certainly, the streams that deposited these units flowed from that generaldirection. Work by Cepeda (1977, 1979), Cameron et al. (1982), McDowell (1979), and Walton (1978,1979a, 1979b) support the hypothesis that volcanic activity in and near the Chinati-Infiernito caldera complex in West Texas was the source of, and stronglyinfluenced, the deposition of the upper northern part
of the Bandera Mesa Member. 
Marker beds, lowermember 
A number of thin or discontinuous units within the Devil’s Graveyard Formation are locally useful strati­graphic marker beds (ACSN, 1972) but do not merit member status. They are, however, important in de­tailed correlation, documenting the relative strati­graphic positions of fossil localities. Shown diagram­matically in figure 5 and on various of the enclosed measuredsections, thesemarker beds willbediscussed in stratigraphic order as informal units. A more de­tailed lithologic description of these beds is given on each measured section. 
Basal Tertiary conglomerate. Junction facies. -The 
lowest marker bed, directly overlying Cretaceous 
rocks, is called the Junction facies of the Basal 
Tertiary conglomerate. It consists of small pebble
sandy to large medium pebble conglomerate made up
of very well-rounded limestone and subangular to 
subrounded chert clasts overlain by and in some places
interlayered with more extensive, very coarse to fine 
feldspathic calclithites and calclithites. To the south 
and west, where the conglomerates are generally clast­
supported, some units up to two meters thickare well-
sorted and imbricated, although major normal grad­
ing over minor reverse grading in units 0.5-1.5 m 
(1.6-4.9 ft) thick is more common. Flow direction, 
as determined from both imbrication and abundant 
0.5-1 m (1.6-3.2 ft) trough cross-bedding sets in the 
*Name approved for publication by Geologic Names Com mittee, U.S. Geological Survey. 
Texas Memorial Museum Bulletin No. 32 
sandstones, is to the east or slightly southeast. The 
thickness of 55 ft (16.8 m) shown in figure 6 must be nearly a maximum; thicknesses of 25-30 ft (8-9 m) are more common. This bed pinches out against pre­existing topography developed on Cretaceous rocks to the northeast of the Devil’s Graveyard and is thin or absent (e.g., fig. 7) in the northwestern part of the type area (Tascotal Mesa Quadrangle). At three local­ities, one near the base of the measured section shown in figure 6, large numbers of micromammal teeth and a few isolated jaw fragments have been col­lected from conglomeratic, very coarse sandstones within the Junction facies. 
Variegated beds.--The smectitic claystones, mud-stones, and minor sandstone or conglomerate channel deposits that overlie the Junction facies or correlative material or rest directly on Cretaceous rocks are infor­mally called the Variegated beds. They are brightlycolored-red, orange, purple, light gray, or green. Light-gray layers are nearly pure clay, while the more brightly colored units show substantial amounts of admixed gypsum, quartz, potassium, and sodic felds­par. Some of these minerals are diagenetic, but claysin various shades of red, purple, and orange are sandy.Where the Variegated beds are of clay they show a characteristic “whaleback” topography (see Moon,1953,pi.3,fig. 1).Athicknessof108.5ft(33.1m)is shown in figure 6, and this is believed to be a near maximum. The Variegated beds are usually thinner in the eastern and northeasternpart of the Devil’s Grave­yard and are often made up of minor channel-fillingmuddy sandstone. Elsewhere, the Variegated beds are believed to be partially lacustrine. Variation in thick­ness is produced by underlying topography, minor erosion that preceded or accompanied deposition of overlying tuffaceous, biotite-rich sandstones, and growth faulting with maximum displacements of 30 to 42 ft (9 to 13 m). A welded tuff sometimes accompanied by two or three much-altered thinner layers that may be air-fall tuffs, occurs within these sandstones and clays. The maximum known thickness of the weldedtuffimmediately underlyingthe Whistler Squat fossil quarry is 20 in (51 cm). The welded tuff is discontinuously exposed from the southwestern to the northeastern corners of the type area of the Devil’s Graveyard Formation. A sample of the tuff gave K-Ar dates of 48.6 and 45.8 my (recalculated to 49.7 and 
46.9, fig. 5).
Lunch Locality sandstone. .—The sandstones asso­ciated with this tuff originally had a high content of glass shards, now altered to orange clinoptilolite.Where this sand is coarser and better washed, it be­comes a zeolitic, volcanic, quartz-bearing, calclithic,sanidine arkose to feldspathic calclithite sufficientlydistinct so that, although occurring only as minor lenticular channel fillings, it serves as a local strati­graphic field marker bed dubbed the “Lunch Local­ity” sandstone. At Whistler Squat quarry, Lunch Locality sandstone (capping hill 3395, Moon, 1953,PL 1) occurs 15-20 ft (5-6 m) higher than the welded 
tuff discussed above. Whistler Squat quarry is about 1,300yds(1,189m) eastofWhistlerSquatandacross a fault. The conglomerate capping the butte at Whis­tler Squat is a Titanothere channel filling whereas the 
conglomerate above the quarry site is a Lunch Local­ity and is considerably lower in the section. A verysimilar, more lithic sandstone occurs as minor lenses within and above a conglomerate directly overlying the Variegated beds or resting on Cretaceous rocks. Thus,there are two Basal Tertiary conglomerates rather than one. The Lunch Locality conglomerate is similar to the conglomerate of the Junction facies but con­tains one to three percent very well-rounded silicic volcanic pebbles. It is best distinguished from Junc­tion facies by the clinoptilolitic sandstones associated with it. The upper conglomeratic unit is informallycalled the Northeast facies of the Basal Tertiary con­glomerate and is the lowest Tertiary conglomeratenortheast of the Devil’s Graveyard. Together with interlayered sandstones, it can be up to 30 ft (9 m)thick, but is normally half of this. The textures and primary sedimentary structures of the Northeast facies are identical with those of the Junction facies. Both are sheet-like bodies of sandstone and conglom­erate. Vertebrate fossils belonging to the Whistler Squat local fauna occur in tuffaceous sandstones above and below the Northeast facies, but the onlyfossils hitherto found in the conglomerate are largesilicified logs up to 4 ft (1.2 m) in diameter and some 75 ft (22.9 m) long.
A third conglomerate, informally called the Car Park conglomerate, has approximately 20 percentsilicic volcanic pebbles and occurs above the North­east facies as a basal filling in channels cut in some places well into the Variegated beds. To the northeast of the Devil’s Graveyard where the Lower member is thickest, Car Park conglomerate channels are separatedfrom the Northeast facies by an interval of lacustrine and palludal sediments with some minor channels filled by sandstone similar to Lunch Locality sand­stone. On the southwestern side of the Devil’s Grave­yard are found places where Junction facies, North­east facies, and Car Park conglomerate are stratigraphi­cally superimposed with only minor amounts of inter-lensed sandstone. The lithic portion of the sandstones becomes progressively more volcanic upward.
Strawberry tuff.-Themarkerbedinformallycalled the Strawberry tuff has been mentioned previously as being at or near the top of the Lower member, Devil’s Graveyard Formation. It is present at the top of the section shown in figure 6. In figure 7 the Strawberrytuff is comparatively thick and is 152 ft (46.3 m)above the base of the formation. Clinoptilolite, a white altered glass, occurs in a slightly meanderingchannel exposed in relief on the Strawberry tuff in the central part of the Devil’s Graveyard. The Straw­berry tuff is recognizable by its lateral persistence,striking brick-red color, and its tendency to form led­ges. It is 2 to 20 ft (0.6 to 6 m) thick and, althoughoccasionally coarser-grained, is usually a zeolitized (clinoptilolite and analcime) sandy mudstone or mud-stone. The analcime is an exception to the accurate generalization of diagenesis in the Devil’s GraveyardFormation shown by Henry and Duex (1981, fig. 4).Some fossiliferous, pink, suspended-load channel deposits (Galloway, 1977), 25 to 30 ft (7.6 to 9.1 m) thick are exposed at two localities (fig. SC, Serendi­pity, and SD, Unio Cliff) in the northeastern part of 
1984 Devil’s GraveyardFormation-Markerbeds 7 
the Devil’s Graveyard where most of the Lower mem­ber above the Strawberry tuff is preserved. These are referred to informally as the “Serendipity channels,”although all exposures may be parts of a single channel. 
Marker beds, middle member Titanothere channelsandPurpleBenchtuff.—Thebase of the Middle member of the Devil’s Graveyard Form­ation is marked by a disconformity having as much as 95ft(29 m)ofreliefinsomeplacesinthenortheastern parts of the Devil’s Graveyard. Channel-filling con­glomerate(see Moon, 1953,fig. 3), zeoliticvolcarenite to zeolitized fine vitric arenite-filled channels at and slightly above this disconformity, are termed Titan­othere channels because at one locality limb bones,teeth, jaws, and skulls of these animals have been found. Titanothere channel conglomerates in at least three locations are directly overlain by orange, altered (clinoptilolitic), biotite-rich vitric arenites which caused serious stratigraphic confusion in the earlystages of this study because of theirsimilarity to zeo­litized sandstones associated with the Northeast facies conglomerate. Titanothere conglomerate is usually of no great lateral extent, but one channel can be traced for about four mi (6.4 km). Thicknesses are highlyvariable but rarely greater than 10 ft (3m). Titan­othere conglomerates are moderately to well sorted,grain-supported, small to large medium, pebble con­glomerates in which normal grading is common, reverse grading rare. Flow direction to the east/north­
east is indicated by imbrication, cross-bedding in sandstones, and channel orientation. The well-rounded pebbles are subequally of limestone, silici­fied limestone, and silicic volcanic rocks with minor amount of chert and green silicic quartzarenite.Hematite in the silica, zeolite, and calcite cements at most exposures lends a red color, but on weathered surfaces the channel appears dark brown. This color­
ation is in marked contrast to that characterizing the older conglomerates which appear a yellowish gray,but is common in the remainder of the Devil’s Grave­yard Formation. 
In the western part of the Devil’s Graveyard where the buried topography at the base of the Middle mem­ber is subdued, a striking purple tuff, a minor ledge-former called the Purple Bench tuff, can be traced out of a Titanothere conglomerate channel. The tuff is exposed in the banks of Alamo de Cesario Creek close to the eastern edge of the southeastern part of the Tascotal Mesa Quadrangle for 3 mi (4.8 km) and into the area beween the North and South forks of Alamo de Cesario. A large concentration of vertebrate fossils has been found in a nodular, yellow, zeolitized vitric arenite approximately 1 to 3 ft (.3 to .9 m)above the Purple Bench tuff. The Purple Bench tuff is probably an altered airfall ash layer, although it may have been slightly reworked; its lower part is another unusually pure clinoptilolite. The PurpleBench tuff varies from about 8 in (20 cm) to 2 ft (60 cm) thick, and its base defines the boundarybetween the lower and middle members. 
Repeats..—The middle part of the middle member is a thick section of sedimentary tuff well exposed as cliffs (see Moon, 1953, fig. 2) forming the northern 
boundary of the Devil’s Graveyard. The tuffs are light gray, cream, pink, or brilliant red. There is a series of three units collectively called the “Repeats,” each consisting of a lower, redder part, and a grayer upper part, and each about 20 to 25 ft (6 to 8 m) thick. The lighter portion is often coarser; in some places it is a cross-bedded tuffaceous sand, whereas the under­lying red portion is usually mudstone. In a few placesthe red portion shows very small ripple and con­torted ripple cross stratification. Usually only two,and sometimes only one, of these repeats are exposed,and it is often difficult to know which one is present.The two repeats are exposed at approximately 350 ft (107 m) and at 450 ft (137 m) above the base of the measured section pictured in figure 8. The lower of these repeats forms the base of the measured section in figure 9. 
Markerbeds, Bandera Mesa Member (new) 
Skyline channels..—A series of brick-red to dark-red­brown to gray sandstone, or deeply-incised conglom­erate and sandstone-filled channels, determines the position of the top of the line of irregular cliffs (fig. 7) that form the northern margin of the Devil’s Graveyard, from about 1.8 mi (2.9 km) east of the Devil’s Backbone, a prominent trachyte dike, to 3.9 mi (6.3 km) west toward the western end of the Devil’s Graveyard. These channels constitute the next informal lithozone, called the Skyline channels. Fur­ther examples of these channels are exposed in the pink and red walls of the canyon of the North Fork of Alamo de Cesario (fig. 8). The minor disconformitywith which the slightly sinuous, incised channels are associated marks the base of the Bandera Mesa Mem­ber, Devil’s Graveyard Formation, and must lie close to an Eocene-Oligocene boundary. The conglomerates are mostly but not always clast-supported, and the sorting varies from well-sorted and well-imbricated to poorly sorted. The clasts range in size from small peb­bles to cobbles and boulders, with some boulders of locally derived, pink tuffaceous sandstone reaching60 cm (1.97 ft) in intermediate diameter. Some depositional units are as much as 3 m (9.8 ft) thick,but thicknesses of one-half to one-third this are more common. The conglomerate and sandstone that fills someoftheincisedchannelstotalsmorethan 100ft (30 m). Flow direction, judged mainly by tracingindividual channels and noting cross bedding, was to the east. Clasts are well rounded and usually silicic 
(trachytic) volcanic rock fragments with subequalamounts, approximating a sum of 30 to 70 percentof silicified limestone and limestone pebbles. Mafic,vesicular, and dark-colored volcanic rock fragments are present in amounts of 5 to 27 percent. The pres­ence of this lithology is distinctive, inasmuch as mafic clasts are rare in the Devil’s Graveyard Formation except in the Cotter channels, which are discussed below. Except for the previously noted color, the vol­carenites associated with the Skyline channels are not distinctive. A Duchesnean ? vertebrate fauna has been recovered from the Skyline channels. Also, numerous mammal tracks, gastropods, and impressions of palmfronds have been found along bedding planes of the Skyline channel sandstones. 
Texas Memorial Museum Bulletin No. 32 
Figure 1.-Map of Texas with Presidio and Brewster counties shaded. Enlarged map shows location of principal fossiliferous areas; Big Bend National Park shaded. 
1984 Devil’s Graveyard Formation-Figure 2 9 
Figure 2.-Sketch map of Trans-Pecos Texas and adjacent Mexico showing extent of igneous and volcaniclastic rocks, known and inferred volcanic centers, and the location of geologic quadrangles. AF, Agua Fria Quadrangle (Moon, 1953); BH, Buck Hill Quadrangle (Goldich and Elms, 1949); BM, Bofecillos Mountains Quadrangle (McKnight, 1970); CM, Cathedral Mountain Quadrangle (McAnulty, 1955); JG, Jordan Gap Quadrangle (Seward, 1950); TM, Tascotal Mesa Quadrangle (Erickson, 1963).Modified from Gorski (1970). 
10 Texas MemorialMuseum Bulletin No. 32 1984 Devil’s GraveyardFormation-Figure 3 11 
Figure
measured


sections. 3.-Geologic

map
of 
parts
of 
theAguaFria,

Tascotal

Mesa,
and

Jordan 

Gap


quadrangles

to
show 

areas 

of

exposure

of
Devil’s
Graveyard




Formation 

(new)and
location 

of 
TexasMemorialMuseumBulletin No. 32 
Figure 4.-Fence diagram of Buck Hill Group from the Southern Davis Mountains to the west side of Agua Fria Mountain and South Lajitas Mesa to Tule Mountain. The diagram was not drawn to scale and most faulting was removed. Positions of samplesfor potassium-argon dates shown as K-Ar 1 through 7: (all dates corrected to 1976 lUGS standards) K-Ar 1, Alamo Creek Basalt,40,44,39,43my;K-Ar2,tuffdirectlyunderlyingWhistlerSquatQuarry (TMM41372),49,46my;K-Ar 3,micaceous tuff above Whistler Squat Quarry and below Strawberry tuff, 43 my; K-Ar 4, upper basalt on Bandera Mesa, 33 my; K-Ar 5,Mitchell Mesa Rhyolite, average of 18 dates 32.28 ± .6 my; K-Ar 6, Basalt (Rawls 2, Bofecillos Mountains), 28 my; K-Ar 7, ash beneath Skyline channel at Tepee Canyon (TMM 41578), 42 my. 
1984 Devil’s GraveyardFormation-Figure 5 13 
Figure 5.-Diagrammatic stratigraphic section of the Buck Hill Group, Devil’s Graveyard-Bandera Mesa area, Brewster and Preside Couties, Texas, to show approximate relative thickness and stratigraphic position of fossil localities, local faunas, and North American land mammal ages. Radiometric dates (in millions of years) are from McDowell (1979) except the followingwhich are from Geochron Laboratories; 42.7 (biotite-bearing ash, just below Skyline channel); 33.0 (whole rock basalt, lowest basalt on Bandera Mesa). The number in parentheses following the date signifies the number of samples. All dates on this diagramhavebeen correctedtoconformtolUGSrecommendation SubcommisiononGeochronologyandwillthereforediffer from those given in Wilson (1980), Wilson and Schiebout (1981), Wilson and Schiebout (1984). Letters A through J refer to stratigraphicpositionofoneormorefossillocalities. A.Junction,41443;.6mieastofJunction,41444; B.WhistlerSquatno. 2 
quarry, 41372; C. Serendipity, 41745; D. Unio Cliff, 41784; E. Titanothere Hill, 41723; F. Purple Bench, 41672; G. Skyline,Tepee Canyon, 41578, H. Horseshoe stone coral, 41853; I. Red table, 41965; J. Red Hill, 41781. 
TexasMemorialMuseumBulletin No.32 
Figure


6.—Measured

section 

of
lower


member,
boundary
stratotype,

and 
of
Devil’s
Graveyard



Formation

(new).This 
is

measured

section 

15 
onfigure
3. 
1984 Devil’s GraveyardFormation—Figure 7 15 
Figure 7.-Measured section, lower and middle members, Devil’s Graveyard Formation (new). This is measured section 16 on figure 3. 
16 Texas Memorial Museum Bulletin No. 32 

Figure
Formation





(new).8.-Measured
is
This section 
of

measured

North

section 

Fork
17 of
onfigure
3. Alamo 
de
Cesario
Creek
showing

the

lithology

of 
the

southern 

part
of
the

Bandera

Mesa

Member 

(new)
of
the

Devil’s
Graveyard 

1984 Devil’s Graveyard Formation-Figure 9 17 
Figure 9.-Measured section of middle member and Bandera Mesa Member of Devil’s Graveyard Formation (new) on South Fork of Alamo de Cesario Creek. This is section 18 on figure 3. 
TexasMemorialMuseumBulletin No. 32 
Quadrangle,Figure 
Presidio10.—Measured

County,section,
Texas.upper

is
This boundary
section 

19 
stratotype

on
of
figure
3. 
Bandera

Mesa
Member

(new),
Devil’s
Graveyard

Formation

(new)
at
southeast 

corner 

of
Puerto
Potrillo,
Jordan

Gap 
1984 Devil’s Graveyard Formation-Cotter channels 19 
Cotter channels. —No well-exposed section imme­diately above the Skyline channels was available for study to clarify the stratigraphic relationship between the Skyline channels and the overlying sediments. A set of channels occurs stratigraphically above the Sky­line channels near the old Cotter Ranch headquarters on Needle Creek (Erickson, 1953, pi. 1), and is infor­mally called theCotter channels.Thebestexposed partsof these sinuous to highly sinuous channels are point-bar deposits encountered along Needle, Dogie, Crystal,and Musgrave creeks, on the Tascotal Mesa and AguaFria Quadrangles. The strongly-graded conglomerates are usually not clast supported and are poorly to moderately well-sorted, medium to very small pebbledeposits. They have little lateral continuity and depo­sitional units of conglomerate or sand conglomerates are rarely thicker than 0.5 m (19.6 in) in the pointbars and are more commonly half that figure. The small exposures prevent precise estimates of point-bar thickness, but 22 to 34 ft (7 to 10 m) is prob­able in some places. The streams depositing the sedi­ments filling the Cotter channels flowed generallyeastward. Labile pebbles oflocally derived tuffaceous sand-and-mudstone and calcitic nodules of tuffaceous fine sandstone are abundant and constitute one-half to two-thirds of the pebbles in given exposures, and more in the eastern exposures of the Cotter channels. The durable pebbles are rounded to well-rounded and consist of subequal amounts of limestone, silicified limestone, and silicic (trachytic) volcanic pebbles,witharelative increaseinthe numberofvolcanicpeb­bles upward in the section. Sandstones in the point-bar deposits are medium to fine, except basally, and are volcarenites laterally and vertically gradationalwith zeolitized (clinoptilolite) and montmorillonized vitric arenites. Stratigraphically higher channels,apparently gradational with the Cotter channels, are seen along the east face of Bandera Mesa and are simi­lar to the low sinuosity channels characteristic of depositing streams for most of the Devil’s GraveyardFormation. The Cotter channels yield both vertebrate 
and invertebrate fossils. Upper breccia-conglomerate.,-Goldich and Elms 
(1949, p. 1146) named and described this unit, situ­ated 60 ft (18.3 m) below the top of the Pruett Formation, and stated that it covered many squaremiles and (by implication) that it extends west from the area of Boat Mountain and Whirlwind Spring in the north-central part of the Buck Hill Quadrangle,into the Jordan Gap Quadrangle (Goldich and Seward,1948). The present writers observe that in the Jordan Gap Quadrangle north of McKinney Mountain and west to the base of Antelope Mesa the Upper breccia­conglomerate is exposed as a broad sheet of hematitic,
zeolitized, angular to subrounded conglomerateresistant to erosion, and weathering dark brown to nearly black. Thickness of Upper breccia-conglomeratevaries from 15 in. to 11.5 ft (0.4 to 3.5 m). Its strati­graphic position is shown on figure 5. 
The Upper breccia-conglomerate of Goldich and Elms (1949) provides a valuable stratigraphic marker bed from the northern part of the Buck Hill Quad­rangle at Boat Mountain westward to the Jordan Gap Quadrangle near McKinney Mountain. At Boat 

Mountain and at Whirlwind Spring it occurs in the upperpart ofthePruettFormation(GoldichandElms,1949, p. 1148). McAnulty (1955, p. 545) reportedfossil bones now identified as early Oligocene from the upper breccia at Cottonwood tank just west of Boat Mountain. However, Seward (1950) did not map any Pruett Formation where the Upper breccia­conglomerate is exposed on the Jordan Gap Quad­rangle and must thereforehaveincludeditwithinthe Duff Formation. Early Oligocene fossils are also pres­ent in the lower part of the Bandera Mesa Member which was mapped by Seward (1950) as Duff Forma­tion. This is made more understandable by the east-west cross section of Henry and Duex (1981) in 
which they show that their clinoptilolite and mont­morillonite diagenetic zone extends from the Duff Formation down into the western biostratigraphicequivalent of the Pruett Formation. 
From Puerto Potrillo southeastward, the Upperbreccia-conglomerate is represented by basal fillingsand lateral bar deposits in straight to slighly sinuous channels which indicate an eastward flow direction. The southernmost of these channels is known from a set of large down-dropped blocks at the base of the southern and western sidesofRedHill.Conglomeratic,altered, vitric volcarenites 168 ft (51.2 m) above the baseofthemeasuredsectionatPuertoPotrillo(fig. 10) are lateral to the top of one of these channels. The stratigraphically highest fossils recovered thus far from the Devil’s Graveyard Formation come from 60 to 80 ft (18 to 24 m) below the Upper breccia­conglomerate, Bandera Mesa Member, at Red Hill,Montgomery tank area (Montgomery bone bed) and at the base of the southwestern face of McKinneyMountain. 
Upper part of the Bandera Mesa Member (new).This part of the Bandera Mesa Member is known to the-authors only from exposures in the face of Bandera Mesa, on McKinney Mountain, and along Puerto Potrillo. The bulk of this unit, as with other parts, is made up of fine to very fine calcitic, zeolitized vitric arenites, with minor 1 to 3 ft (30 to 91 cm) thick calcitic, zeolitic volcarenites. The boundary between the clinoptilolite-opal-montmorillonite and clinoptilo­lite-montmorillonite diagenetic facies of Henry and Duex (1981) occurs close to North Canyon, at the northern boundary of the Tascotal Mesa Quadrangle,where the color change previously mentioned takes place. Depositionally, the principal differences can be seen by contrasting sediments northwest of South Canyon with those to the southeast. As Wilson et al. (1979) and Stevens (1979) noted, the southern partof the section has been thinned by pre-Mitchell Mesa Rhyolite and pre-Yellow conglomerate erosion. Fan­
glomerates in the original sense are absent with one,
possiblyYellowconglomerate-related, exception.Con­
glomerates contain rounded to well rounded, medium 
pebbles, although they are usually not well sorted. 
Higher up, limestone and silicified limestone pebbles
become important or even dominant. Lake deposits
including limestone units up to 20 ft (6 m) thick are 
found low in the upper part of the Bandera Mesa 
Member and interlayered with the highest Cotter 
channels, at and near Smugglers Gap. Discontinuous, 

Texas Memorial Museum Bulletin 	No. 32 
thin basalt layers recur high in the section from justnorth of South Canyon to the point about 1 mi (1.6km) from the southeast end of Bandera Mesa where basalt flows were beveled by erosion prior to extru­sion of the Mitchell Mesa Rhyolite.
Stevens’ (1979) description of an upper informal member of the “Pruett-Duff” unit applies mainly to 
the upper part of the Bandera Mesa Member north of South Canyon. There are few conglomerates above the Upper breccia-conglomerate, but near the top,there is a major fanglomerate, represented in figure10 by a 40-ft (12.2-m) interval starting 601 ft 
(183.2 m) above the base of the section. This very coarse fanglomerate is also exposed high on McKinney Mountain. 

SUMMARY 

A thick section of volcaniclastic sediments is named the Devil’s Graveyard Formation. Vertebrate fossils of the early late Eocene Whistler Squat local fauna occur in the basal conglomerate and in the lower member of the formation. Late Eocene vertebrate fossils, the Serendipity local fauna, were found in the middle member and a small vertebrate fauna, the Sky­line local fauna, was discovered in red conglomeratesand is probably of Duchesnean age. The upper or 
Bandera Mesa Member yielded a Chadronian fauna named the Coffee Cup local fauna. The Pruett and Duff formations of the Buck Hill Group were bounded by lava flows, the stratigraphic positions of which were not possible to locate in the area of expo­sure of the Devil’s Graveyard Formation. Lower mid­
dle and upper stratotype sections and the strati­graphic position of important vertebrate fossil local­ities were given. 

REFERENCES 
American Commission on Stratigraphic Nomenclature (ACSN), 1972. Code of stratigraphic nomencla­ture. Amer. Assoc. Petroleum Geologists, Tulsa,Okla., pp. 1-21, 2nd printing.
Barker, R. S., 1977. Northern Trans-Pecos magmaticprovince: introduction and comparison with the Kenya rift. Geol. Soc. Amer. Bull. 88:1421­
1427. Barnes, V.E., Project Director, 1979. Atlas of Texas Emory Peak-Presidio Sheet. Bur. Econ. Geol.,Univ. Texas—Austin. 
Bockoven, N. T., 1981. Tertiary stratigraphy of the Sierra Del Gallego area of Chihuahua with com­parisons to the Pena Blanca uranium district. In Amer. Assoc. Petroleum Geologists, Studies in Geology (13), Uranium in volcanic and volcani­clastic rocks, 181-187. 
Cameron, Maryellen, K.L. Cameron, and J.C. Cepeda,1982. Geochemistry of Oligocene volcanic rocksfrom the Chinati Mountains, West Texas. Geol. Soc. Amer., South Central Section, Abstracts with
Programs, 16th Ann. Mtng., p. 106. 
Capps,R.C., 1981.GeologyoftheRanchoElPapaloteArea, Chihuahua, Mexico. In Amer. Assoc. Petro­leum Geologists, Studies in Geology (13), Urani­um in volcanic and volcaniclastic rocks, 243-264.
Cepeda, J.C., 1977. Geology and geochemistry of the igneous rocks of the Chinati Mountains, Presidio County, Texas. Unpubl. Ph.D. dissertation, Univ. Texas—Austin. 154 pp. 
, 1979. The Chinati Mountains caldera, Presi­dio County, Texas. In Cenozoic geology of the Trans-Pecos volcanic field of Texas. Bur. Econ. Geol., Univ. Texas—Austin, Guidebook 19:106­
125. 
Daily, 	Michael, 1979. Age relations in alkaline rocks from the Big Bend region, Texas. In Cenozoic 
geology oftheTrans-PecosvolcanicfieldofTexas. Bur. Econ. Geol., Univ. Texas—Austin, Guide­book 19:92-96. 
Dalrymple, G.8., 1979. Critical tables for conversion of K-Ar ages from old to new constants. Geology7:558-560. Erickson, R.L., 1953. Stratigraphy and petrology of the Tascotal Mesa Quadrangle, Texas. Geol. Soc. Amer. Bull. 64(12):1353-1386. 
Galloway, W.E., 1977. Catahoula Formation of the 
Texas coastal plain: depositional systems, com­position, structural development, ground water flow history, and uranium distribution. Bur. Econ. Geol., Univ. Texas—Austin, Rept. Invest. 
87. 56 pp.
Gilliland, M.W., andH.C. Clark, 1979. Paleomagnetismof early Tertiary volcanics of the Big Bend, Texas. In Cenozoic geology of the Trans-Pecos volcanic field of Texas. Bur. Econ. Geol., Univ. Texas— 
Austin, Guidebook 19:48-54. 
Goldich, S. S., and M.A. Elms, 1949. Stratigraphy and petrography of the Buck Hill Quadrangle, Texas. Geol. Soc. Amer. Bull. 60:1133-1183. 
Goldich, S. S., and C. L. Seward, 1948. Green Valley
-
Paradise Valley field trip. West Texas Geol. Soc. Field Trip, Oct. 29-31. 
Gorski, D.E., 1970. Geology and trace transition ele­ment variation of the MitrePeak area, Trans-Pecos Texas. Unpubl. master’s thesis, Univ. Texas— Austin. 200 pp. 
Henry, C.D., and T.W. Duex, 1981. Uranium in dia­genesis of the Pruett, Duff, and Tascotal forma­tions, Trans-Pecos Texas. In Amer. Assoc. Petroleum Geologists, Studies in Geology (13),Uranium in volcanic and volcaniclastic rocks,
167-180. 

1984 Devil’s Graveyard Formation-References 	21 
Henry, C.D., and F.W. McDowell, 1982. Timing, dis­tribution, and estimates of volumes of silicic vol­canism in Trans-Pecos Texas. Geol. Soc. Amer.,South Central Section, Abstracts withPrograms,16thAnn. Mtng. p. 113. 
International Subcommission on Stratigraphic Classi­fication (ISSC), 1976. International stratigraphicguide, ISSC of International Union of Geol. Sciences (lUGS) Commission on Stratigraphy.
H.D. Hedberg, ed., John Wiley and Sons, New York. 200 pp.

Keller, P.C., N.T. Bockoven, and F.W. McDowell,1982. Tertiary volcanic history of the Sierra del Gallego area, Chihuahua, Mexico. Geol. Soc. Amer. Bull. 93:303-314. Mauger, R. L., 1981. Geology and petrology of the central part of the Calera-del Nido block, Chi­huahua, Mexico. In Amer. Assoc. Petroleum Geologists, Studies in Geology (13), Uranium in volcanic and volcaniclastic rocks, 181-187. Mauger,R.L.,andF.W. McDowell, 1982.K-Ardatingof igneous rocks in central Chihuahua, Mexico. Geol. Soc. Amer., South Central Section,AbstractswithPrograms, 16thAnn.Mtng.p. 116. 
Maxwell, R. A., J.T. Lonsdale, R. T. Hazzard, and J. A. Wilson, 1967.GeologyofBigBendNationalPark. Univ. Texas Publ. 6711. 320pp.
McAnulty, W.N., 1955. Geology of the Cathedral Mountain Quadrangle, Brewster County, Texas. Geol. Soc. Amer. Bull. 66:531-578. 
McDowell, F.W., 1979. Potassium-argon dating in the Trans-Pecos Texas volcanic field. In Cenozoic geologyoftheTrans-Pecos volcanicfieldofTexas. Bur. Econ. Geol., Univ. Texas—Austin, Guide­book 19:10-18. 
McKnight, J.F., 1970. Geologic map of Boffecillos 
Mountain area, Trans-Pecos Texas. Bur. Econ. Geol., Univ. Texas—Austin, Geol. Quad. Map. 37,with text. 
Moon, C.G., 1953. Geologyof AguaFria Quadrangle,Brewster County, Texas. Geol. Soc. Amer. Bull. 61:151-196. 
Parker, D.F., 1976. Petrology and eruptive history of anOligocene trachyticshieldvolcano, nearAlpine,Texas. Unpubl. Ph.D. dissertation, Univ. Texas— Austin. 184 pp. 
, 1978.ThePaisanovolcano:stratigraphy,age,and petrogenesis. In Cenozoic Geology of the Trans-Pecos volcanic field of Texas, Bur. Econ. 
Geol., Univ. Texas—Austin, Guidebook 19: 97-105.
Parker, D.F., and F.W. McDowell, 1979. Stratigraphyand K-Ar geochronology of Oligocene volcanic rocks, Davis and Barrilla Mountains, Trans-Pecos 
Texas. Geol. Soc. Amer. Bull. Pt. 1, 90:1100 1110. Robinson, 8.R., 1976. Stratigraphy and environment 
of deposition of member 9 of the Rawls Forma­tion, Presidio County, Texas. Unpubl. master’s thesis, Univ. Houston, 151 pp. 
Schiebout, 	J.A., 1974. Vertebrate Paleontology and Paleoecology of the Paleocene Black Peaks For­mation, Big Bend National Park, Texas. Texas Memorial Mus. Bull. 24. 88 pp. 
Seward, C.L., 1950.GeologyoftheJordanGapQuad­rangle, Texas. Unpubl. master’s thesis, Texas A&M Univ. 72 pp.Stevens, J.8., 1969.GeologyoftheCastolonarea,BigBend National Park, Brewster County, Texas. Unpubl. Ph.D. dissertation, Univ. Texas—Austin. 129 pp. , 1979. Eocene-Oligocene volcaniclastic sedi­ments, West-Central Brewster County and adja­cent Presidio County, Trans-Pecos Texas. In Cenozoic geology of the Trans-Pecos volcanic field of Texas, Bur. Econ. Geol., Univ. Texas— Austin, Guidebook 19:150-156. Stevens, M.S., J.B. Stevens, and M.R. Dawson, 1969. New early Miocene Formation and vertebrate local fauna, Big Bend National Park, Brewster County, Texas. Texas Memorial Mus. Pearce-Sellards Ser. 15. 53 pp.

Walton, A.W., 1972. Sedimentary petrology and zeo­litic diagenesis of the Vieja Group (Eocene-Oligo­cene) Presidio County, Texas. Unpubl. Ph.D. dis­sertation. Univ. Texas—Austin. 264 pp. , 1975. Zeolitic diagenesis in Oligocene vol­canic sediments, Trans-Pecos Texas. Geol. Soc. Amer. Bull. 86:615-624. 
, 1979. Volcanic sediment apron in the Tasco­tal Formation (Oligocene?), Trans-Pecos Texas. 
J. Sed. Pet. 49; 303-314. , 1979
b.Sedimentology and diagenesis of the 
Tascotal Formation: a brief summary. In Ceno­zoic geology of the Trans-Pecos volcanic field of Texas. Bur. Econ. Geol., Univ. Texas—Austin,Guidebook 19:157-171. 
Wilson, J.A., 1972. Vertebrate biostratigraphy of Trans-Pecos Texas and northern Mexico, in the geologic framework of the Chihuahua tectonic 
belt. West Texas Geol. Soc., R.K. DeFord Sym­
posium, pp. 157-166. 
, 1974. Early Tertiary vertebrate faunas ViejaGroup and Buck Hill Group, Trans-Pecos Texas: Protoceratidae, Camelidae, Hypertragulidae. Tex­as Memorial Mus. Bull. 23. 34 pp. 
, 1977. Early Tertiary vertebrate faunas BigBend area Trans-Pecos Texas: Brontotheriidae. Texas Memorial Mus. Pearce-Sellards Ser. 25. 15 pp. 
Wilson, J.A., J.B. Stevens, and M.S. Stevens, 1979. 
New cross section from the southern Davis Moun­tains to northeast Solitario. In Cenozoic Geologyof the Trans-Pecos volcanic field of Texas. Bur. Econ. Geol., Univ. Texas—Austin, Guidebook 19: 147-149. 
Wilson, J.A., and F.S. Szalay, 1976. New adapid pri­mate of European affinities from Texas. Folia Primat. 25(4):294-312. 
Wood, A.E., 1972. An Eocene hystricognathous ro­dent from Texas: its significance in interpreta­tions of continental drift. Science 175: 1250­1251. 
, 1973. Eocene rodents, Pruett Formation,southwest Texas; their pertinence to the originof the South American Caviomorpha. Texas 
Memorial Mus. Pearce-Sellards Ser. 20. 41 pp.