Kimberly Member:

The Kimberly Member was defined by Fisher and Rensberger (1972: 15) as gray, yellow-gray, pale greenish-gray, and buff tuffs well exposed near the Kimberly post office. Reported to be ∼900 ft in composite maximum thickness, its lower boundary was determined by zeolitization and hence was considered by them to be diachronous on a regional scale. However, in the type area of the member, they noted that the zeolitization boundary and the regional dip of the contact between the Kimberly and Turtle Cove members is nearly coincident so that locally the boundary may be approximately synchronous. Our examination of the contact in the type area of the member south of Kimberly demonstrated that the lithology of the tuffaceous claystones above and below the boundary is identical with the exception of zeolitic alteration. Nor is there any evidence of unconformity or other pause in deposition at this horizon. On the basis of these lithologic observations it would be reasonable to predict faunal continuity across the boundary: Rensberger's (1971, 1973, 1983) rodent zonation for these sediments suggests this is in fact the case.

The upper boundary of the Kimberly Member is an erosional surface at many localities. The upper part of the formation was removed by downcutting accompanying tectonic deformation during late John Day time. Along the west wall of the John Day valley south of Kimberly (secs. 14, 23, 26, T10S, R25E) the eroded surface of the Kimberly Member is overlain by the Columbia basalts. However, if the member is traced northward along the west wall of the valley, a prominent calcareous paleosol appears at the top of the unit (well exposed in SE;q1, NW;q1, SW;q1, NE;q1, sec. 12, T10S, R25E) that probably represents an uneroded remnant of the terminal Kimberly Member surface.

At other locations south of Kimberly, higher stratigraphic units fill an irregular topography developed on the surface of the Kimberly Member (Fisher, 1967). These are the sands, gravels, and tuffs of Merriam. Fisher mapped these higher beds together with the rocks of the Kimberly Member as a single map unit (Fisher and Rensberger, 1972: Maps 2, 3, Tjkh), but in his text he recognized the higher beds as a distinct stratigraphic unit, the Haystack Valley Member (Fisher and Rensberger, 1972: 17). This reflected Fisher's assumption that these sands and gravels correlated with coarse fluvial sediments in the Balm Creek area of Haystack Valley, as described below.

Our examination of rocks referred by Fisher to the Kimberly Member was centered on the principal exposures of the unit: the type area near Kimberly, the northern Rudio Creek drainage, and along the John Day river from Haystack Valley to Kimberly. Rocks of the Kimberly Member in the type area are uniformly gray tuffaceous sediments, predominantly siltstones, sandstones, and less frequent intraformational conglomerate, with a distinctive obsidian-shard tuff near the base of the Rudio Creek section, considered by Fisher and Rensberger (1972: 14–15) as a possible equivalent of the Deep Creek Tuff. These tuffaceous sediments display primary sedimentary structures and incised channel fills indicating local fluvial reworking together with interbedded airfall units. Burrows and rhizoliths are common at many levels, demonstrating a succession of paleosols suggestive of a gradual, episodic pattern of sediment accumulation through periodic ashfall events. The sediments within channels suggest both rapid deposition of volcaniclastic debris in intense flood events as well as more prolonged periods of channel filling and winnowing that yielded uniformly well-sorted sediments. In the Rudio Creek area, a fluvial channel within the Kimberly Member (NW;q1, NE;q1, NE;q1, sec. 27, T9S, R26E) is incised into tuffaceous gray siltstones and is filled with poorly sorted, locally derived, tuff-pebble conglomerate interbedded with fluvial tuffaceous sandstones and diatomite. These sediments were rapidly deposited with little winnowing. Fossils in these channels include larger unabraded skeletal elements of mammals, indicating minimal transport and rapid burial. On the other hand, incised channel fills situated within the member in its type area south of Kimberly contain well-sorted, clast-supported, tuff-pebble conglomerates filling the base of channel scours. The good clast sorting and rounding and the presence of only well-abraded fragments of mammal bone within these conglomerates demonstrate more prolonged fluvial processing of the channel sediments.

We did not find the Kimberly Member to be particularly fossiliferous, although occasional specimens occur in many outcrops. Much of the bone from coarse sandstones and conglomerate bodies is fragmented and strongly abraded, indicating fluvial processing. Mammal remains in the siltstones and sandstones tend to range from partially articulated limb segments to isolated teeth, mandibles, and limb elements. Complete skeletons are uncommon and we encountered none during our work. The condition of the skeletal material indicates that carcasses decayed, were scavenged, and in some instances the remnants were buried by ashfalls or by fluvial reworking of tuffaceous sediments, possibly during rainfall events following volcanic eruptions.

Gray unzeolitized tuffaceous sediments of the Kimberly Member cropping out in the type area south of Kimberly can be traced eastward into the Rudio Creek drainage and northwestward along the John Day valley from Kimberly nearly to Haystack Valley. Typical gray Kimberly Member tuffs when traced northwest from Kimberly terminate at the projected trace of the Richmond Fault along the east margin of the Balm Creek valley (Fisher and Rensberger, 1972: Map 3, E;h1, sec. 34, T8S, R25E). Kimberly Member gray tuffs occur in the upper part of outcrops southeast of the fault on the upthrown block in sec. 34; sections on the downthrown block north of the fault in Balm Creek valley contain unzeolitized gray cliff-forming tuffs that we regard as much younger in age. In apparent agreement with Rensberger (1983: 29), we recognize these gray tuffs in the Haystack Valley-Balm Creek area as part of the Haystack Valley Member, as revised in this study.

Fisher and Rensberger (1972: 17) did suggest that rocks age-equivalent to the Kimberly Member (along Rudio Creek) are represented by a zeolitized section of green beds in Haystack Valley at UCMP locality Haystack 1. This age-equivalence was said to be demonstrated by the common occurrence of similar species of entoptychine rodents (Rensberger, 1983: 29).1

Haystack Valley Member:

This member was named by Fisher and Rensberger (1972: 17) for mostly unzeolitized fluvial and lacustrine tuffaceous sediments, including airfall tuffs, interbedded with coarse sandstone and conglomerate in the Haystack and Balm Creek valleys, and along the east wall of the John Day valley south of Kimberly. In actual practice, Fisher's concept placed all John Day rocks that incorporated welded tuff-bearing conglomerates in the Haystack Valley Member. Such rocks are typically separated from the Kimberly Member by erosional unconformity south of Kimberly along the east wall of the John Day valley and include fluvially reworked tuffaceous sediments of all textural grades from clay to gravel.

Fisher and Rensberger (1972: 17) frequently refer to “welded-tuff-bearing conglomerate at the base of the Haystack Valley Member …”. Their generalized regional stratigraphic section depicts the conglomerate as the basal unit within the member (Fisher and Rensberger, 1972: fig.1), separated by erosional unconformity from the Kimberly Member below. Indeed, two of their measured sections (Fisher and Rensberger, 1972: fig. 3, nos. 5 and 8) south of Kimberly show a welded tuff conglomerate as the basal unit of the Haystack Valley Member. However, their principal measured section for Haystack Valley (Fisher and Rensberger, 1972: fig. 3, no. 10), sited at Balm Creek, shows the welded tuff conglomerate as a single horizon ∼100 ft above the base of the unit. This placement also appears in the composite lithostratigraphic column for the Kimberly and Picture Gorge quadrangles (Fisher and Rensberger, 1972: fig. 3). We became interested in the stratigraphic occurrence of welded tuff conglomerates when we encountered them at multiple levels at Balm Creek, including one prominent bed quite high in the local section. This contrasted with the member south of Kimberly along the east wall of the John Day valley where, as indicated by Fisher and Rensberger, the welded tuff conglomerate nearly always occurs at a single stratigraphic horizon at the base of the unit.

Comparing the mammalian faunas from the two areas (Balm Creek and the east wall of the John Day valley south of Kimberly) clarified the situation. The Haystack Valley Member rocks at Balm Creek carried a different and much older fauna (appendices 1-1 to 1-7) than that found in the conglomerate unit south of Kimberly (appendices 1-12 to 1-15). Although both stratigraphic units had been included in the Haystack Valley Member, they proved to be of significantly different ages. We thus became aware that welded tuff conglomerates occurred over a considerable time interval in the upper John Day beds.

These observations led us to study the lithostratigraphy, the distribution of fossil mammals, and the clast composition of the conglomerates in more detail in the two principal geographic areas mapped by Fisher and Rensberger (1972): (1) Haystack Valley, and (2) south of Kimberly along the John Day valley. We discovered that, based on the contained mammal faunas and lithostratigraphic relations, several stratigraphic units of different ages were included within the Haystack Valley Member, and could be discriminated on superpositional relations, bounding unconformities, lithofacies, and faunal content. In addition, the conglomerates in these units differed in both clast composition and size in a predictable manner (table 1). Three regionally significant unconformities of early Miocene age (EMU) are recognized in the study, and are indicated in our measured sections, from older to younger, as EMU₁, EMU₂, and EMU_3.

We describe these units first in the Haystack Valley-Balm Creek area, then from exposures along the east and west walls of the John Day valley south of Kimberly.

Revision of the Haystack Valley Member5:

Fisher and Rensberger did not describe the rocks of the Haystack Valley Member in the Balm Creek valley in detail; however, their principal stratigraphic section (Fisher and Rensberger, 1972: fig. 3, section 10) was measured in the Balm Creek drainage at the same location as our Asher Section. Because section 10 had lost Unit 5 and the upper part of Unit 4 to erosion, Fisher and Rensberger's concept of the Haystack Valley Member did not include these uppermost John Day rocks.

Fisher and Rensberger (1972: pls. 7, 8) selected the welded tuff-bearing conglomerate at the 300 ft level in our Asher section as the base of the Haystack Valley Member. The totality of exposures in the Balm Creek drainage (figs. 4–8) demonstrates the difficulty in using the appearance of a particular welded tuff-bearing conglomerate to identify the base of the Haystack Valley Member since there are multiple levels with conglomerate of this type in the Haystack Valley-Balm Creek area, and the conglomerate body of their plate 7 is not the stratigraphically lowest such lens. In fact, lithologies above and below this bed are similar and do not warrant stratigraphic separation. The location of the lower boundary of the member is further placed in doubt when, in section 10, an additional ∼100 feet of the member is shown below the welded tuff conglomerate level.

Fisher and Rensberger considered all rocks of their section 10 from the welded tuff conglomerates up to the capping basalts as the Haystack Valley Member. We suggest a more circumscribed stratigraphic definition of the Haystack Valley Member (revised) incorporating only our Units 1 and 2, thus limiting the member to the lower 435 feet of the Asher Section along Balm Creek (fig. 4). The upper boundary of the unit is demarcated by the prominent siliceous paleosol developed on the gray marker tuff (GMAT, fig. 4) of Unit 2B. Rocks below this marker bed in Units 1 and 2 are of uniform lithology and surface-weathering characteristics; the alternating resistant ledges and softer interbeds give a horizontally ribbed appearance to these outcrops along Balm Creek and were termed “ribbed tuffs” during our field study. Units 1 and 2A are zeolitized, more strongly toward the base. Importantly, monomictic welded tuff conglomerates are present at multiple levels in Unit 2A, although more concentrated from 100 to 170 feet below the gray marker tuff (GMAT) in the middle part of the section. In Unit 2B, only a single lens of polymictic conglomerate with welded tuff pebbles was observed, occurring at the base of the unit in a channel cut into Unit 2A (fig. 8, ∼2632 ft). Above Unit 2, only rare, minor welded tuff conglomerate lenses occur in the section below the coarse basal polymictic conglomerate of Unit 5 (at Section HS-1, thin granule to pebble conglomerate lenses at two levels within the lower part of Unit 3 contained welded tuff clasts). Fossil mammals in the Balm Creek drainage are entirely restricted to the revised member and do not occur above it, except for one specimen, a peccary skull, found at the base of Unit 3 (appendix 1-6).

The revised member is principally exposed along Balm Creek in sections 33–34, T8S, R25E, but does not occur south of the Richmond Fault in section 34 (Map A). It also crops out in cliffs along the west and east walls of Haystack Valley where the member is in fault contact with the Turtle Cove Member which occupies the central part of the valley.

It may prove advisable after future field study to emphasize a lithologic and biochronologic distinction between Units 1 and 2A. At this level in the Asher section there is an apparent division between green zeolitized soft-weathering claystones and siltstones of Unit 1 carrying the Entoptychus individens–Allomys tessellatus rodent assemblage, and overlying pale gray-green to yellow-gray well-indurated tuffaceous siltstones of Unit 2A lacking this assemblage.

The rocks of Unit 1 with E. individens–A. tessellatus were considered part of the Haystack Valley Member by Rensberger (1983: 90, table 16). We concur, thinking it appropriate at present to include these beds in the member because (1) the zeolitized coarse sandstones of Unit 1 at one of the localities yielding E. individens (SW;q1, NE;q1, NE;q1, SE;q1, sec. 33, T8S, R25E: probably UCMP locality Haystack 22) included a few highly weathered welded tuff pebbles, suggesting that these coarse fluvial lenses mark the first access of local streams to exposures of the Picture Gorge ignimbrite, the source of the welded tuff pebbles in these beds; and (2) a general similarity in the “ribbed tuff” lithologies of Units 1 and 2A.

Balm Creek Member:

Fisher and Rensberger included all rocks of section 10 from the welded tuff conglomerates up to the capping basalts in their Haystack Valley Member. We think it useful to recognize a marked change in the style of sedimentation within this stratigraphic sequence that occurs at the boundary between Units 2 and 3. Unit 3 in its lower part is comprised of yellow-gray tuffaceous siltstones, claystones, and very fine sandstones with thin interbeds of low-angle, cross-stratified gray fine-to-medium sandstones. The yellow-gray beds appear to be airfall tuffs occasionally reworked by lower-flow regime fluvial processes, punctuated by slightly coarser detrital gray sands and gravel representing intermittent sheet-flooding. The upper part of Unit 3 is made up of thinly laminated lacustrine tuff and claystone that in its lower part intertongues with the yellow-gray tuffs of the lower part of Unit 3. Occasional thin detrital volcaniclastic sands and airfall tuffs occur within the upper 25 feet of the lacustrine beds. The presence of quiet-water lacustrine sediments probably reflects the damming of local drainages in response to syntectonic compression that modified local topography during the development of the Balm Creek syncline.

Units 3 and 4 are lithically distinguished from the units below by their lack of zeolitization. Unit 3 weathers to smooth slopes with a soft powdery surface suggestive of montmorillonitic clay; Unit 4 is distinguished by fresh glass, including in some beds black obsidian shards.

Unit 4 is characterized by lower (beds I–IV) and upper (beds V–X) sets of stacked fining upward sequences (fig. 5, Beardog Section: Middle and Upper) of tuffaceous fluvial sandstones (some with coarse obsidian shards), separated by a massive 15-ft thick gray airfall tuff. This tuff is overprinted by a prominent paleosol (fig. 5, 425–440 ft).

The upper limit of Unit 4 is marked by a 12-ft thick terminal paleosol (fig. 5, 508–520 ft) developed on the uppermost fining-upward sequence (fig. 5, bed X). This paleosol is removed where incised by the fluvial channels of Unit 5 (fig. 6). Unit 4 lithologies are clearly distinct from the zeolitized ribbed tuffs and conglomerates of our revised Haystack Valley Member.

We designate Units 3 and 4 the Balm Creek Member, noting that these units are well exposed along the axis of the Balm Creek syncline (Map A: W;h1, SW;q1, sec. 27, T8S, R25E). Because no fossil mammals have been found in Unit 4 and only a single fossil mammal (a peccary skull) has been found in Unit 3 at its base, age is conjectural. However, the peccary (table 4, cf. Hesperhys) suggests a late or latest Arikareean age for Unit 3, which is supported by the stratigraphic placement of Units 3–4 between Units 2A (early late Arikareean) and 5 (early Hemingfordian, see below).

Importance of Unit 5 in Regional Correlation:

As mentioned previously, Fisher and Rensberger's section 10 omitted Unit 5 because at that location post-John Day erosion removed the uppermost rocks of the formation, and the resulting topographic low was later filled by basalt flows. Section 10 was measured 0.43 mi (0.7 km) southwest of the more complete Beardog Section where Unit 5 escaped post-John Day erosion. The existence of Unit 5 has important implications for regional lithostratigraphy and tectonics in the Haystack Valley-Kimberly area. Unit 5 includes coarse polymictic welded tuff-bearing conglomerates and stream-deposited sands, debris flows, lacustrine claystones, and yellow-gray tuffaceous fine-grained rocks with siliceous paleosols that correspond to a similar lithofacies association found in the highest stratigraphic unit of the upper John Day Formation along the east wall of the John Day valley south of Kimberly. It became evident that the polymictic welded tuff-bearing cobble conglomerates of Unit 5 correlated to the equally coarse polymictic welded tuff conglomerates of this high unit along the east wall of the valley south of Kimberly (table 1), and that these rocks were much younger than previously believed. Although no fossil mammals were found in Unit 5 at Balm Creek, a representative early Hemingfordian fauna (table 4; appendices 1-12 to 1-14, fauna of the Rose Creek Member, see below) occurred in the unit south of Kimberly. We eventually realized that these particularly coarse polymictic conglomeratic beds had no close temporal relationship with the monomictic welded tuff conglomerates of the revised Haystack Valley Member at Balm Creek and in Haystack Valley where ⁴⁰Ar/³⁹Ar dating of Unit 2A (table 2, and fig. 5, Beardog Section: Lower, 25–45-ft level) revealed that the revised Haystack Valley Member is a much older rock unit.

Rose Creek Member (East Wall of the John Day Valley from Kimberly to McCarty Creek):

The outcrops used by Fisher and Rensberger to identify the Haystack Valley Member south of Kimberly were considered by them to be “particularly well developed along the eastern margin of the John Day River valley (secs. 7 and 17, T10S, R26E), where [this unit] lies in well-exposed contact with the underlying Kimberly Member.” At the base of the unit they recognized a distinctive welded tuff-bearing conglomerate (Fisher and Rensberger, 1972: 17–18, fig. 3, measured section 8). The unit appears as a series of topographically high exposures not exceeding 70–100 feet in thickness along the eastern wall of the John Day valley for about 7 miles south of Kimberly post office (fig. 2). It lies with erosional and angular unconformity (EMU₂) across both Kimberly and Turtle Cove Members, as Fisher noted. The polymictic welded tuff-bearing conglomerate, which can be traced at the base of the unit from Spring Creek south to McCarty Creek (figs. 9–11, Maps C, D), is strikingly similar in clast size and composition to the polymictic conglomerate at the base of Unit 5 in the Beardog Section at Balm Creek (fig. 6, Map A).

A distinctive lithofacies association characterizes the unit south of Kimberly: polymictic welded tuff-bearing conglomerate, debris flows, coarse-shard tuffs (commonly with vesicular obsidian shards), and thin beds of dark gray airfall tuff. Interbedded fine-grained tuffaceous claystones, siltstones, and very fine sandstones are commonly olive-gray to pale yellow-gray and weather to smooth slopes with occasional thin resistant interbeds. The debris flows are lithically uniform: welded tuff pebbles derived from the Picture Gorge ignimbrite occur as well-rounded matrix-supported clasts scattered through fine-grained gray structureless tuff (fig. 12). These flows are frequently heavily overprinted by paleosol features: siliceous white rhizoliths, invertebrate burrows, devitrification, and soil brecciation. The unit is not zeolitized and often terminates in sets of stacked siliceous paleosols. This lithofacies association is also present in Unit 5 on Balm Creek (fig. 6).

An important lithologic correlation exists between the high unit south of Kimberly on the east wall of the valley and Unit 5 at Balm Creek (figs. 6, 10). Both units show erosional and angularly unconformable relations with the John Day rocks below; both have basal polymictic conglomerates of similar clast type and mean intermediate diameter for the 10 largest clasts (table 1, ;ms6 to ;ms8ø, cobble gravel of the Wentworth-Udden scale); both include similar lithologies and lithofacies above the basal conglomerate; and both represent the stratigraphically highest unit in their local areas beneath the Columbia basalts. We select the name Rose Creek Member for this unit and designate as its type area the narrow belt of nearly continuous exposures along the east wall of the John Day valley south of Kimberly (figs. 9, 10), extending from the NW;q1, sec. 8, T10S, R26E on the north, through the west half of sections 17, 20, 29, T10S, R26E, to the SW;q1, sec. 32, T8S, R26E, Mt. Misery 7.5-minute quadrangle (Maps C, D, sections 7–14). We also apply this member designation to Unit 5 of the Balm Creek syncline based on similar lithologic, lithofacies, and positional relations.

This interpretation differs from that of Fisher and Rensberger (1972), who suggested that the high unit south of Kimberly along the east wall of the John Day valley correlated with our Unit 2 and higher beds below the Columbia basalts in the Balm Creek section. Their interpretation was based in part on the presence of welded tuff conglomerate in both Unit 2 of the Balm Creek section and in the Rose Creek Member. Here we emphasize that the mean intermediate diameters and clast types of the Unit 2 conglomerates of the revised Haystack Valley Member in the Balm Creek sections differ from the Rose Creek Member conglomerates south of Kimberly as shown in table 1. Unit 2A conglomerates are monomictic, welded tuff-bearing, pebble gravels, whereas Rose Creek Member conglomerates are polymictic (welded tuff and various other clasts) pebble-to-cobble gravels considerably coarser than those of Unit 2. Furthermore, biochronologic and radiometric data discussed below show that Unit 2 at Balm Creek and the Rose Creek Member along the east wall of the John Day valley differ significantly in age by as much as 5–6 million years (table 3).

Welded-Tuff Conglomerates of the Rose Creek Member:

Clasts derived from the Picture Gorge ignimbrite, particularly those from partially to strongly welded parts of the flow, make up a significant fraction of the pebbles and cobbles in the conglomerates of the Rose Creek Member. The term ignimbrite is used here in the sense of Fisher (1966b: 6) as a general term for all rock types that are produced by dispersal of fragments from a pyroclastic flow. Fisher observed several lithologies in the two cooling units of the Picture Gorge ignimbrite, including air-fall tuff, unwelded and partially welded lapilli tuff, and densely welded tuff. Some of these types can be recognized in hand specimen in the conglomerates and are summarized in table 1. Hay (1963: 206) recorded that the “ignimbrite exhibits all gradations between nonporous, highly welded tuff and porous, friable lapilli tuff in which the particles of glass are neither flattened by compaction nor welded.”

According to Fisher (1966b), a basal air-fall tuff and unwelded tuff making up the lowest two units of the ignimbrite are rhyolitic and the higher units are dacitic in composition. The air-fall tuff and the two overlying cooling units are composed mostly of vitric shards and pumice fragments which can be readily observed in hand specimen under the binocular microscope. Lapilli include pumice and rock fragments of crystalline salic rocks and vitric welded tuff. The unwelded base of Fisher's cooling unit 1 is a friable tuff that grades in places into welded tuff and a black obsidian-like glass.

Many clast types in the Rose Creek Member conglomerates can be attributed to the Picture Gorge ignimbrite. They are derived from the densely welded or partially welded portions of the ignimbrite and can be grouped into discrete classes: (a) olive to dark brown, crystal-poor, densely welded tuff with dark pumice fiamme; (b) greenish-gray, pale yellow or red rhyolitic/dacitic welded tuff, with a few small euhedral or broken quartz and feldspar phenocrysts distributed in an aphanitic microcrystalline groundmass (without dark pumice fiamme); (c) partially to densely welded, gray to dark gray obsidian-shard tuff, often vitreous, exhibiting shard compaction and buff pumice lapilli; (d) welded obsidian-like black glass clasts (not a true obsidian but an end-member densely welded tuff described by Ross and Smith, 1960: 24, fig. 9); (e) gray lapilli tuff, commonly with brown vitric welded tuff lapilli in a gray devitrified aphanitic groundmass. Locally, clasts of the densely welded tuff display an attractive red and gray banding.

Smith (1960) observed that deformation of shards and pumice fragments is the only certain criterion for recognition of welding in tuffs that have crystallized, which is often the case in geologically older ash flows. Only the densely welded and/or crystallized units of the ignimbrite have survived as clasts in the conglomerates of the Rose Creek Member, the unwelded tuff and air-fall tuff units being too soft and friable to survive erosion and transport as cohesive clasts. Smith noted that the transition from unwelded to densely welded is accompanied by loss of pore space and progressive deformation of glass shards and pumice. The Rose Creek conglomerate clasts derived from the ignimbrite, despite their textural and compositional variety, are nearly all entirely lacking in pore space and many display deformed pumice clasts and compacted shards, indicating their derivation from the welded portions of the ignimbrite cooling units. It is important to consider whether in Miocene and older ash flows such as the Picture Gorge ignimbrite the composition of the various ignimbrite zones may have been affected long after cooling by burial and ground-water diagenesis, further obscuring the initial textural composition of the rocks (Smith, 1960: 153). However, transport of these clasts and subsequent diagenesis in Rose Creek Member drainages has not prevented the identification in hand specimen of the various ignimbrite phases.

In addition to clasts derived from the ignimbrite, the Rose Creek conglomerates also contain clasts derived from the Deep Creek tuff (Fisher, 1962) and from subjacent vitric tuffs of the John Day Formation incised by the Rose Creek Member.

Johnson Canyon Member (West Wall of the John Day Valley south of Kimberly):

Fisher's appraisal of the Kimberly and Haystack Valley Members south of Kimberly mentioned the stratigraphically high unit on the east wall (our Rose Creek Member) but did not discuss or examine in detail the beds thought by him to be of the same age on the west wall. Fisher and Rensberger (1972: fig. 3, measured section 9) presented a stratigraphic section of the west wall at the mouth of Johnson Canyon; a conglomerate near the base in that section was questionably correlated with the basal conglomerate of the Rose Creek Member (their section 8) on the east wall. Haystack Valley and Kimberly Members were mapped together as a single unit both on the west and east walls of the valley without differentiation (Fisher and Rensberger, 1972: Map 2).

The best exposures of the Kimberly and Haystack Valley Members (sensu Fisher and Rensberger) along the west wall occur in Johnson Canyon 0.6 mi (1 km) southwest of Kimberly and along the John Day river for 3 mi south of Kimberly (Maps B, C, sections 16–20). Here Kimberly gray tuffs are disconformable beneath, in fact incised by, yellow-gray tuffaceous siltstones, sandstones, and intraformational monomictic vitric tuff-pebble conglomerate (figs. 13, 14, sections 16–17, EMU₁). These yellow-gray, fluvially worked tuffaceous sediments and the subjacent gray tuffs of the Kimberly Member are evidently the rocks assigned to their Tjkh map unit (Kimberly and Haystack Valley Members) by Fisher and Rensberger (1972: Map 2). It is these yellow-gray beds incised into the Kimberly Member that we regard as a lithostratigraphic unit distinct from the Haystack Valley Member (revised) along Balm Creek, and also lithically distinct from the Rose Creek Member along the east wall.

Outcrops at the mouth of Johnson Canyon (Maps B, C, sections 18–20) constitute the thickest measured section of these yellow-gray tuffaceous rocks along the west wall of the John Day valley south of Kimberly (∼355 ft, fig. 15, from 1920 to 2278 ft). Here, these Johnson Canyon beds can be divided into an upper unit conformably superposed on a lower unit (fig. 16). The lower unit (fig. 15, ∼1920–2130 ft) consists of fluvial sequences of intraformational monomictic vitric tuff-pebble conglomerate, tuffaceous sandstone, and siltstone that fine upward, terminating in a prominent paleosol that is continuous along the north wall of the canyon for 0.25 mi (0.4 km). An enormous tuff-filled vertebrate burrow, probably excavated by a large carnivore, can be seen in the north wall of Johnson Canyon penetrating this well-developed land surface (fig. 15, section 20, 2130 ft). Calcareous cementation predominates in the lower unit. Above the lower unit is an upper unit of three superposed massive gray airfall tuffs overprinted by siliceous paleosols (fig. 15, section 20, 2130–2212 ft). These cliff-forming tuffs are succeeded by coarse obsidian-shard tuffs, each fining upward and ending in a developed paleosol horizon (fig. 15, section 20, 2212–2250 ft). The upper unit concludes with ∼25–30 ft of yellow-gray tuffaceous sandy siltstones beneath the capping basalts. The upper unit is characterized by silica cementation, especially in the paleosol horizons rich in siliceous rhizoliths.

A white vitric tuff is present in the lower unit of the Johnson Canyon beds 0.4 mi (0.6 km) south of the mouth of Johnson Canyon at UCMP locality Picture Gorge 1. Named the ATR tuff (Fremd et al., 1994), it has been dated by C.C. Swisher of Rutgers University at 22.6 ± 0.13 Ma (⁴⁰Ar/³⁹Ar). The ATR tuff can be traced from Picture Gorge 1 northward into the principal measured section of the Johnson Canyon beds along the canyon's north wall (fig. 15, Picture Gorge 1 to Johnson Canyon East). The tuff serves as a useful marker bed within Johnson Canyon and occurs at progressively lower elevations in outcrops from south to north, due to local block faulting.

We designate the upper and lower units at Johnson Canyon as the Johnson Canyon Member. These rocks can be distinguished from the Kimberly Member by the yellowish-gray color of the fine-grained tuffaceous beds, the more varied lithofacies, more numerous and deeply incised intraformational monomictic conglomerates of locally derived vitric tuff pebbles, the presence of the ATR marker tuff, and superposition on the uniformly gray Kimberly Member. When traced southward along the west wall of the John Day valley, the Johnson Canyon Member rests in erosional disconformity (EMU₁) on the gray tuffs of the Kimberly Member (fig. 13, Big Valley and UCMP V-6668 sections). A measured section at UCMP locality V-6668 shows a basal channel of the Johnson Canyon Member incised to a depth of at least 50 ft (fig. 13, 2310 ft) into the Kimberly Member gray tuffs. Similarly, at our Big Valley section, superposition of the Johnson Canyon Member on Kimberly beds is demonstrated by a sharp erosional contact between the underlying Kimberly gray tuff and a channel fill of intraformational monomictic conglomerate at the base of the Johnson Canyon Member (fig. 13, 2340 ft).

The Johnson Canyon Member is best exposed in Johnson Canyon and in isolated outcrops along the west wall of the John Day valley south of Kimberly above the uniformly gray Kimberly beds that form the base of the cliffs. We have not observed the member south of the N;h1, NE;q1, sec. 12, T10S, R25E. South of this area the unit appears to have been removed by erosion, if it was ever present. The member occurs north of Johnson Canyon in the area of the Kimberly post office and can be traced northwest of Kimberly into the SE;q1, sec. 24, T9S, R25E, and to the center, sec. 13, T9S, R25E along Bologna Creek.

When we first studied the outcrops in Johnson Canyon and other less exposed sections of upper John Day rocks south of Kimberly along the west wall, we were surprised to discover that there are no welded tuff-bearing conglomerates in these beds. The seven to eight levels of intraformational monomictic conglomerate in Johnson Canyon and nearby outcrops contain only locally derived, well-rounded, indurated vitric tuff clasts from subjacent tuffs incised by the channel scours (table 1, Johnson Canyon). Also, the clast size was much smaller on average (mean I.D. ;lt7 cm) than that for the basal conglomerate of the Rose Creek Member on the east wall (mean I.D., 12 samples, 12.9 cm, table 1). This prompted us to ask whether the outcrops attributed by Fisher to the “Haystack Valley Member” on the west and east walls might not represent two different lithostratigraphic units that differed to some degree in age. Alternatively, they might be temporally equivalent sediment prisms distinguished by lithofacies formed in different depositional environments. In the latter case the polymictic welded tuff conglomerates of the east wall and the monomictic intraformational conglomerates of the west wall might simply reflect stream access to geographically disjunct source areas, one deriving welded tuff clasts from the Picture Gorge ignimbrite, the other from a drainage network of local gullies lacking access to the ignimbrite. We looked for other evidence that might provide insight into this problem. It was the unexpected discovery of a topographically high remnant outcrop of the Rose Creek Member on the west wall of the valley that initially suggested that the Rose Creek Member and the Johnson Canyon Member might be distinct lithostratigraphic units of different ages. This was eventually confirmed by the mammal faunas from the two members (table 4) and by radiometric dating.

Relative Ages of the Johnson Canyon and Rose Creek Members:

High on the west wall of the John Day valley in the NE;q1, sec. 12, T10S, R25E, opposite the mouth of Spring Creek, is an isolated outcrop overlying several hundred feet of Kimberly gray tuff at an elevation of 2670 feet (figs. 13, 14, Dead Cow Gulch Section). The outcrop begins with 70 feet of gray-brown claystones overlain by 30–40 feet of tuffaceous sediments that include distinctive coarse obsidian-shard tuff and debris flow units heavily overprinted by paleosols. This facies association is similar to Rose Creek Member lithofacies along the east wall of the valley (figs. 9, 10) where welded tuff conglomerates and fluvial sandstones yielded an early Hemingfordian fauna (∼18.2–18.8 Ma) and were accompanied by obsidian-shard tuff and debris flow units with strong paleosol overprints. There is an evident correlation between these topographically high obsidian-shard tuff and debris flow units on the east and west walls of the valley, suggesting that the Dead Cow Gulch outcrop is an isolated outlier of the Rose Creek Member, in fact the only vestige of the member remaining along the west wall of the valley south of Kimberly.

The Rose Creek Member outlier rests on a mature calcareous paleosol at the top of the Kimberly Member that probably represents the uneroded, intact terminal surface of Kimberly deposition (fig. 13, Dead Cow Gulch Section, 2670 ft). A short distance (2000 ft) to the northeast at UCMP Locality V-6668 (fig. 13), a basal channel of the Johnson Canyon Member that deeply incises the Kimberly gray tuffs at this location is 360 ft below the high remnant outlier of the Rose Creek Member. Although these outcrops are separated by a normal fault (fig. 13, Dead Cow Gulch fault), the contrast in lithofacies and lithology in such a short distance is so marked that lateral equivalence is unlikely. The Johnson Canyon Member to the north of the Rose Creek outlier is capped by basalt flows that abut against the fault; thus, if the Rose Creek Member was superposed on the Johnson Canyon Member at an earlier time, it was removed by erosion before influx of the basalts.

At no locality south of Kimberly have we been able to observe the Rose Creek Member in superposition on the Johnson Canyon Member. However, biochronologic evidence discussed below reveals that the mammalian fauna obtained from the Johnson Canyon Member in Johnson Canyon is older than that from the Rose Creek Member, and at Sutton Mountain the Rose Creek Member overlies rocks that closely resemble the Johnson Canyon Member south of Kimberly (fig. 17).

Upper John Day Units at Sutton Mountain:

Hay (1963: fig. 2, sections 10–11) measured particularly thick sections of the upper John Day beds north of Bridge Creek along the southern face of Sutton Mountain. We studied the same area which occurs in the N;h1, sec. 33, and S;h1, sec. 29, T10S, R21E, Sutton Mountain 7.5-minute topographic quadrangle (fig. 17, appendices 1-10, 1-11). Upper John Day rocks fill a large paleovalley incised into earlier John Day beds in the center of section 33; this paleovalley fill is downdropped to the north along a west to east-trending normal fault mapped by Hay (1963: fig. 2, sec. 33, T10S, R21E). A thick, lithically homogeneous series of yellow-gray upper John Day tuffaceous sandstones, siltstones, and occasional lenses of intraformational monomictic vitric tuff-pebble gravel fills the paleovalley north of the fault, continuing upward for ∼350 ft (fig. 17, Unit D, lower sequence6). This rock unit is overlain along a sharp but conformable contact by a gray cliff-forming tuff containing small obsidian shards, heavily overprinted by multiple stacked siliceous paleosols (fig. 17, Unit D, upper sequence); this upper sequence, 25–30 ft thick, is expressed topographically as a prominent bench along the south face of Sutton Mountain (center and NW;q1, sec. 33, T10S, R21E).

Resting on Unit D, and in places incised into it with low relief, are ∼100 ft of coarse polymictic welded tuff conglomerates, medium to coarse detrital sandstones, and brown to gray-brown tuffaceous claystones and siltstones interbedded with debris flow, gray lapilli tuff, and coarse obsidian-shard tuff units: these make up a series of fluvial fining-upward volcaniclastic units, most beginning with welded tuff conglomerate, grading upward through detrital sandstones to brown claystones. These lithofacies are typical of the Rose Creek Member south of Kimberly, suggesting correlation of the member over an east-west distance of ∼32 mi (51.5 km).

In lithology, lithofacies, color, and style of sedimentation, Unit D resembles the Johnson Canyon Member south of Kimberly. The ∼350 ft of section making up the lower sequence of Unit D closely corresponds to the lower unit in Johnson Canyon (fig. 15), and the upper sequence of gray tuff overprinted by siliceous paleosols is much like the upper unit in Johnson Canyon, although not as thick. We also have collected teeth of the small chalicothere Moropus cf. oregonensis from both the lower sequence of Unit D at Sutton Mountain and from the lower unit of the Johnson Canyon Member in Johnson Canyon. Thus, we consider Unit D at Sutton Mountain the lithologic and temporal correlative of the Johnson Canyon Member.

Superposition of the Rose Creek Member on the Johnson Canyon Member could not be demonstrated in the John Day valley south of Kimberly. However, along the south face of Sutton Mountain the superposition of the Rose Creek Member on the Johnson Canyon Member (Unit D) supplies this critical stratigraphic relationship. Thus, the Johnson Canyon Member, considered older than the Rose Creek Member based on the mammalian fauna from Johnson Canyon south of Kimberly, can be documented in stratigraphic relation beneath the Rose Creek Member at Sutton Mountain, supporting the age relationship suggested independently by faunal content.

Fauna of the Haystack Valley Member (revised):

Fossil mammals occur throughout Units 1 and 2A in the Balm Creek drainage but were not found in Unit 2B. The fauna from Units 1 and 2A includes rodents, several oreodonts, two species of peccaries, a horse, a dicerathere rhinoceros, temnocyonine amphicyonids (beardogs), and rare specimens of small canids, camel, tapir, and hypertragulid (table 4; appendices 1-1 to 1-7). The stage of evolution of the fauna suggests an early late Arikareean age based on correlation with the Great Plains sequence.

Key mammalian taxa indicative of this age assignment are Miohippus, Hypertragulidae, temnocyonine beardogs, and advanced species of the rodents Entoptychus (E. individens) and Allomys (A. tessellatus), none of which has been found stratigraphically higher than the redefined member in the study area. Also useful in this regard are primitive forms of the camel Paratylopus, the oreodont Paroreodon, the tapir Nexuotapirus, and a rodent possibly representing the earliest known species of Schizodontomys (or, alternatively, a transitional form intermediate [?] between Tenudomys and Schizodontomys).

The rodent Entoptychus individens, the most advanced entoptychine reported by Rensberger (1971), occurs in the lower part of Unit 1 (fig. 4, Asher Section, 0 to 48 ft) ∼250 to 280 ft below the dated tuffaceous siltstone (fig. 5, 25–45-ft interval, ∼23.5–23.8 Ma; table 2) in Unit 2A of the Beardog Section. This rodent is common in the lowermost ∼48 feet of Unit 1 as exposed along Balm Creek; however, because the lower contact of Unit 1 with subjacent rocks was not identified anywhere in the Balm Creek drainage, we are not able to determine the proximity of this rodent zone to the base of the revised Haystack Valley Member. We did not find entoptychines above Unit 1 in the Balm Creek sections.

The fauna together with the radiometric date from Unit 2A (table 2) indicate that the revised member is much older than previously believed, certainly older than the Rose Creek Member and its fauna south of Kimberly, which previously had been correlated by Fisher and Rensberger (1972) with rocks of the revised member in the Balm Creek area.

Fauna of the Balm Creek Member:

A partial skull of a small peccary (cf. Hesperhys; appendix 1-6) was found in the lowest few feet of Unit 3 (fig. 5, 175–180 ft, Beardog Section: Lower); no additional fossil mammals are known from Unit 3, or from Units 4 and 5. The peccary indicates a late Arikareean age for the base of Unit 3, an age which probably pertains to the lithically uniform yellow-gray tuffaceous siltstones, claystones, and very fine sandstones comprising the lower half of the unit (fig. 5, 175–260 ft, Beardog Section: Lower and Middle) that produced this fossil. Because the lacustrine beds in the upper half of Unit 3 intertongue with the lower yellow-gray tuffaceous sediments, it seems likely that the entire unit is of late Arikareean age. The age of Unit 4 remains uncertain.

Fauna of the Johnson Canyon Member:

Only the lower unit of the Johnson Canyon Member contains fossil mammals; none have been found in the near vertical exposures of the cliff-forming upper unit. Fossil mammals from the lower unit are frequently found in the intraformational conglomerates as stream-abraded specimens, but scattered and partially articulated individuals also occur in the fine-grained tuffaceous siltstones and sandstones. We examined specimens from this unit at the University of California-Berkeley and at the Los Angeles County Museum of Natural History, combining these identifications with those from our own fieldwork. Comparison of the fauna of the lower unit with the Great Plains sequence suggests a late to latest Arikareean age.

Age-indicative taxa of the Johnson Canyon Member include an advanced species of the oreodont Paroreodon; the camel Paratylopus cameloides; the peccary Hesperhys; an unidentified dromomerycid and the moschid deer Parablastomeryx cf. advena; the equids Archaeohippus, cf. Kalobattipus, and an anchithere; the small chalicothere Moropus oregonensis; the rhinoceroses Diceratherium and a small species of Menoceras; the tapir Miotapirus harrisonensis; the rodents Schizodontomys and Mylagaulodon; the small canids Cynarctoides cf. luskensis and Desmocyon thomsoni; and temnocyonine amphicyonids (table 4). We have not found Miohippus, parahippine horses, hypertragulids, or entoptychine and allomyine rodents in the Johnson Canyon Member. Miohippus, Entoptychus, and Allomys are presumably extinct by the time of deposition of the member; parahippine horses first appear in the deposits of the Rose Creek Member.

Fossils collected from the lower unit at the mouth of Johnson Canyon were plotted relative to the ATR tuff (∼22.6 Ma), which serves as a useful stratigraphic datum. The mammalian fauna from below the ATR tuff does not appear to differ from the fauna collected above the tuff. However, we maintained these collections as two separate assemblages since future sampling of these rocks may indicate a difference. The assemblage below the tuff is designated here the Picture Gorge 1 local fauna; the assemblage above the tuff is named the North Wall local fauna. The North Wall local fauna includes not only the fossils from the lower unit of the member along the north wall of the canyon, but also the mammals from UCMP Locality V-6431 (Little Dike Locality) on the south side of the canyon where fossils were also found above the ATR tuff.

Almost all fossil mammals from the Johnson Canyon Member come from exposures at UCMP Picture Gorge 1 (V-6666) and Little Dike (V-6431) localities, and from the north wall outcrops (UCMP locality V-6432) at the mouth of Johnson Canyon (fig. 15; appendices 1-8, 1-9). The North Wall and Picture Gorge 1 local faunas compare with mammals collected by University of Nebraska and American Museum field parties from the Harrison Formation and Upper Harrison beds in Sioux County, Nebraska, and in adjacent Goshen and Platte Counties, Wyoming (Hunt, 1985, 1990). The Upper Harrison beds include the Eagle Crag Ash, a volcanic tuff dated by the fission track method (zircon) at 19.2 ± 0.5 Ma (Hunt et al., 1983). The Agate Ash within the Harrison Formation has recently been dated by the ⁴⁰Ar/³⁹Ar method to ∼23 Ma (sanidine, 22.9 ± 0.08; Izett and Obradovich, 2001). In our view this provides an age range for the lower unit of the Johnson Canyon Member, and accordingly we have indicated its age in table 3 as within the ∼19.2–22.6 Ma interval.

The fauna of the lower unit of the Johnson Canyon Member contains a number of taxa that indicate a younger age relative to that of the fauna from the Haystack Valley Member (revised). Johnson Canyon fossils of the oreodont Paroreodon, the camel Paratylopus, and the rodent Schizodontomys represent species more evolved than those of the same genera in the revised Haystack Valley Member along Balm Creek. The absence of Miohippus in the Johnson Canyon Member and the first appearance of dromomerycid and moschid cervoids also support this inference. Temnocyonine amphicyonids occur in both the revised Haystack Valley Member and the Johnson Canyon Member but are absent and presumed extinct in the Rose Creek Member. The presence of a small species of Menoceras, temnocyonines, the canid Cynarctoides cf. luskensis, and tapir Miotapirus harrisonensis, and the absence of advanced parahippine horses in the lower unit at Johnson Canyon indicate a fauna older than that of the Rose Creek Member on the opposite wall of the John Day valley.

The heteromyid rodent Schizodontomys is represented in the lower unit at Johnson Canyon not only by the genoholotypic species (S. greeni, UCMP 39435, Rensberger, 1973: 62), which was almost certainly collected from below the ATR tuff at Picture Gorge 1, but also by several individuals of Schizodontomys found by us above the tuff at UCMP locality V-6432. The UNSM specimens of the genus are dentally advanced and compare with a species of Schizodontomys (S. cf. harkseni, UW 1298, Munthe, 1981) belonging to a fauna from the uppermost part of the Arikaree Group (Upper Harrison beds) at Uva, Platte County, Wyoming. We suggest that the Uva fauna is one of the youngest Arikareean assemblages yet discovered in the type area of the Arikareean NALMA and, based on the dated Eagle Crag Ash, probably falls near ∼19 Ma. Thus, because of the near identity of Schizodontomys from Johnson Canyon and from Uva, we consider that the 22.6 Ma date for the ATR tuff may be a maximum age, and that the fauna from the lower unit in Johnson Canyon more likely falls somewhere within the younger part of the ∼19.2–22.6 Ma interval.

Fauna of the Rose Creek Member:

Study of mid-Cenozoic mammal faunas from the Arikaree and Hemingford Groups (late Oligocene to mid-Miocene) of the central Great Plains has resulted in a detailed biochronology calibrated by radiometric and paleomagnetic data (MacFadden and Hunt, 1998; Woodburne and Swisher, 1995; Tedford et al., 1987, 1996; Hunt, 1985, 1990). During our exploration of the Rose Creek Member on the east wall of the John Day valley south of Kimberly, we discovered fossil mammals useful in age determination at two localities. We were also able to examine specimens from this unit at the University of California-Berkeley collected in the 1960s by Lester Kent and John Rensberger. Comparison of these mammals with the Great Plains sequence suggests a precise correlation and supplements the earlier work of Rensberger using fossil rodents.

Age-indicative taxa discovered in the Rose Creek Member include the large oreodont Merycochoerus magnus; the dromomerycid Barbouromeryx; the small moschid deer Parablastomeryx; the camels Paratylopus and a larger species (cf. Protolabis); the equids Parahippus pawniensis, Archaeohippus, and an anchithere; the chalicothere Moropus oregonensis; a large beaver (cf. Hystricops); the rodents Sewellelodon and Mylagaulodon; and two new species of wolf-sized amphicyonid carnivorans (table 4; appendices 1-12 to 1-15). We have not found Miohippus, Paroreodon, temnocyonines, entoptychine and allomyine rodents, the heteromyid Schizodontomys, or tapirs in the Rose Creek Member.

These mammals are designated the Picture Gorge 36 local fauna. Almost all specimens have come from Rose Creek outcrops at UCMP locality Picture Gorge 36 (appendix 1-14) at the head of Bone Creek and UNSM localities at the head of Rose Creek (appendices 1-12, 1-13). This local fauna compares with a fauna collected by University of Nebraska and American Museum field parties in Sioux County, Nebraska (Northeast of Agate local fauna, MacFadden and Hunt, 1998: 163). The Northeast of Agate local fauna occurs within sediments of the lower part of the Runningwater Formation paleomagnetically calibrated at ∼18.2–18.8 Ma (MacFadden and Hunt, 1998: 162–163) and is stratigraphically above the Eagle Crag tuff in the Upper Harrison beds dated by fission-track (zircon) at 19.2 ± 0.5 Ma (Hunt et al., 1983). Consequently, we believe that the Picture Gorge 36 local fauna indicates an age of ∼18.2–18.8 Ma for the Rose Creek Member south of Kimberly.

The association of several Rose Creek Member taxa support an early Hemingfordian age, equivalent to that of the lower part of the Runningwater Formation in Nebraska. Most important is the presence of the oreodont Merycochoerus magnus, which is found within a sharply restricted interval in the Great Plains and Rocky Mountains, the earliest Hemingfordian (fig. 18). In the Rose Creek Member this index oreodont is associated with the dromomerycid Barbouromeryx; the small moschid Parablastomeryx schultzi; a large beaver near Hystricops; the equid Parahippus pawniensis; and a large, long-limbed species of daphoenine amphicyonid, all closely related to similar species in the early Hemingfordian faunas of western Nebraska.

The Picture Gorge 36 local fauna is the youngest fauna known from the John Day Formation in the type area south of Kimberly. It dates the Rose Creek Member and provides an upper limit on John Day deposition at ∼18 Ma for the eastern facies of the formation. The geographic extent of the Rose Creek Member also suggests, based on its presence (without fauna) in Haystack Valley (Balm Creek, Unit 5) and at Sutton Mountain, that fluvial incision and deposition were geographically widespread at the beginning of the Hemingfordian in the Kimberly, Mt. Misery, and Sutton Mountain topographic quadrangles.

Haystack Valley—Balm Creek:

Fisher (1967) mapped the structural framework of the Haystack Valley area north of the Richmond fault. Haystack Valley lies on the south margin of a structural province situated between the Blue Mountains on the north and the nearly parallel trend of the Richmond fault on the south (fig. 19). In this area John Day sediments were warped into broad southwest-northeast trending synclines (Mitchell and Kahler synclines of Fisher, 1967: fig. 2) that parallel the trend of the Blue Mountain axis. Lesser anticlinal and synclinal folds of similar trend occur immediately north of the Richmond fault and south of the Kahler syncline in the Haystack Valley area (Fisher, 1967: fig. 5). Although he did not name it, the Balm Creek syncline belongs to this group of lesser folds and was mapped by Fisher (1967: fig. 5; Fisher and Rensberger, 1972: Map 3, secs. 26–28, T8S, R25E).

The Balm Creek and Rose Creek Members, the uppermost units of the John Day Formation identified in the Haystack Valley area, are preserved along the trend of the Balm Creek syncline. Southeast of the synclinal axis these members have been largely removed by erosion (fig. 20). However, in this same area the revised Haystack Valley Member is well exposed and can be traced from the syncline to the south and east where it is truncated by the projected trace of the Richmond fault. These exposures extending from the syncline to the fault form a rugged terrain of ridges and valleys that produced the greater part of the mammalian fauna from the member. They are broken by a prominent southwest-northeast trending reverse fault (not indicated on Fisher's maps) with vertical separation of ∼20–50 ft that parallels the trend of the syncline. It extends from the NE;q1, sec. 33 through the N;h1, NW;q1, sec. 34, to the S;h1, SW;q1, SE;q1, sec. 27, T8S, R25E where it disappears beneath the Columbia basalts, which are not faulted. Regional northwest-southeast compression produced the synclinal folding and faulted the south limb of the syncline prior to eruption of the basalts, warping and offsetting rocks of the revised Haystack Valley Member and Balm Creek Member immediately northwest of the Richmond fault.

The Rose Creek Member apparently lies with angular unconformity on the Balm Creek Member along both the synclinal axis and to the south of the axis adjacent to the Richmond fault (in fig. 20 the Rose Creek Member does not occur along the cross-sectional transect). A measured section (HS-1 Section, fig. 8) in proximity to the fault shows that the upper part of the Balm Creek Member has been upwarped and removed by erosion, and the unconformity is overlain by welded-tuff conglomerate and tuffaceous units of the Rose Creek Member. Similar Rose Creek incision into the Balm Creek Member also occurred along the axis of the syncline but there the upper part of the Balm Creek Member was preserved (Beardog Section: Middle-Upper, figs. 5, 6).

Northwest of the Balm Creek syncline along the floor of Haystack Valley itself, uppermost John Day units have been removed by erosion and only the lower part of the formation (fig. 20, Tjt, Turtle Cove Member) is present (Fisher and Rensberger, 1972: Map 3). These outcrops are fossiliferous and produce a fauna that includes the large oreodont Promerycochoerus, Miohippus, hypertragulids, less advanced entoptychines, and the temnocyonine Mammacyon, indicating an age more typical of the middle of the formation above the Picture Gorge ignimbrite. On the west wall of Haystack Valley, cliff-forming exposures of upper John Day units are down-faulted against these older John Day rocks. The west wall preserves a section that includes the revised Haystack Valley Member and in places the lower part of the Balm Creek Member (fig. 20). These outcrops of uppermost John Day rocks can be traced westward to the village of Spray where prominent exposures occur north of the town.

John Day Valley South of Kimberly:

Southeast of the Richmond fault the regional structural pattern (fig. 19) is characterized by a more westerly to easterly grain of anticlines and fault trends (Fisher, 1967: fig. 2—the John Day, Middle Mountain, and Hamilton faults; Richmond and Black Butte anticlines). A structural profile based on Fisher's (1967: fig. 5) transect from Middle Mountain on the south to Kahler Basin on the north passes through the John Day valley south of Kimberly and includes the uppermost John Day rocks of this report (fig. 21). Fisher's profile shows John Day sediments filling a shallow basin ∼25 mi (∼40 km) in north-south extent. The John Day strata south of Kimberly are gently downwarped according to Fisher's interpretation; faulting is limited to the trace of the Richmond fault and is not identified south of Kimberly. Somewhat more pronounced folding is shown northwest of the Richmond fault where, at the scale of the profile, John Day strata have been warped into southwest-northeast trending synclines and anticlines.

Fisher's (1967) structural interpretation can be augmented by our field observations that identified extensive faulting south of Kimberly along both east and west walls of the John Day valley. The location of the faults is important in interpreting the lithostratigraphy of the formation south of Kimberly. From Johnson Canyon on the north to Rose Creek on the south, a distance of ∼3 mi (∼4.8 km), faulting is evidenced by vertical offset of marker units in measured sections, by the presence of springs in the valley walls, and by basalt-filled valleys approximately coincident with suspected fault trends. Faults are normal, down to the north, with vertical offsets from ∼40 to ;mt400 ft (∼12 to ;mt120 m). Fault trends are difficult to determine in several locations because of the covering basalts, but can be limited to northwest-southeast to west-east alignments. These faults apparently continue to the east into the Rudio Creek drainage where John Day exposures along the creek in secs. 25, 26, and 36, T9S, R26E are also down-faulted to the north along similar trends, preserving a thick accumulation of Kimberly Member sediments near the mouth of Rudio Creek.

Measured sections on the west wall of the John Day valley south of Kimberly (figs. 13, 15) indicate significant down-faulting of the Johnson Canyon Member and the subjacent Kimberly Member, beginning at Dead Cow Gulch in the N;h1, NE;q1, sec. 12, T10S, R25E, and continuing northward to Johnson Canyon (W;h1, NW;q1, sec. 31, T9S, R26E). The Kimberly Member appears to be downfaulted on the north at least 200 ft (60 m) at Dead Cow Gulch (fig. 13), the southernmost fault we have identified along the west wall. Immediately north of the fault the contact of the Johnson Canyon Member on Kimberly gray tuffs is present in Dead Cow Gulch at UCMP locality V-6668, and can be traced 0.8 mile (1.25 km) to the north at approximately the same elevation into the Big Valley section. If the Johnson Canyon Member is then followed farther north along the west wall, the member is apparently down-faulted ;mt400 ft (;mt120 m) between the Big Valley section and Johnson Canyon (fig. 13). This suggests to us that the considerable thickness of the Johnson Canyon Member in the vicinity of Johnson Canyon and Kimberly post office is due to preservation of the member in a half-graben or graben setting.

Measured sections along the east wall of the John Day valley indicate a similar structural setting. Although there is some variation in the elevation of the lower contact of the Rose Creek Member from its southernmost outcrops at McCarty Creek (Foree) to Rose Creek on the north, the unit lies at a consistently high topographic elevation along the east wall (fig. 10). At Rose Creek, however, the section is abruptly down-faulted with nearly 200 ft (60 m) of vertical offset (Rose Creek fault, fig. 9). One mile (1.6 km) farther north at Spring Creek (fig. 9, section 7), the Rose Creek Member has been down-dropped an additional 160 ft (48 m). Beyond Spring Creek to the north, basalt flows descend to low elevations along the east wall and the Rose Creek Member is absent; if outcrops of the Rose Creek Member were once present, they were removed by pre-basalt erosion.

The down-to-the-north offset of John Day lithostratigraphic units on both walls of the John Day valley south of Kimberly suggests that uppermost John Day rocks were preserved from erosion on down-faulted blocks within a graben or half-graben system (fig. 22) that extends roughly parallel to Fisher's (1967: fig. 2) Monument syncline. The graben includes Kimberly Member and Johnson Canyon Member outcrops (1) along the John Day river northwest of Kimberly as far as Bologna Creek, with exposures possibly extending northwest to the Richmond fault; (2) north of Kimberly to the basalt rim of the John Day valley; and (3) east and northeast an unknown distance, probably as far as Portuguese Canyon (Fisher and Rensberger, 1972: Map 3, NE corner, T9S, R26E), and including the down-faulted exposures at the mouth of Rudio Creek. Pronounced erosional relief on the John Day Formation occurs along this graben system associated with the Monument syncline, and Fisher (1967: 121) realized that this relationship suggested that local drainages first developed along these topographic lows in post-John Day, pre-Columbia basalt time.

Eruption of Columbia Flood Basalts:

Shortly after the deposition and extensional faulting of the early Hemingfordian Rose Creek Member, the first flows of the Columbia flood basalts in the region erupted from the Monument dike fissures between Kimberly, Monument, and Courtrock, and from fissures in the vicinity of Picture Gorge. These flows belong to the Picture Gorge Basalt Subgroup (PGBS) of the Columbia River Basalt recently discussed by Bailey (1989). The ages of the oldest PGBS flows are unknown since the available radiometric dates are from flows at the type section in Picture Gorge where the lower third of the PGBS is absent (Bailey, 1989). Initial potassium-argon ages on flows at the type section ranged from 14.4 to 16.1 Ma (Baksi, 1974: 432, based on basalt crushed to coarse to very coarse sand size for argon extraction); more finely crushed samples yielded ages from 14.4 to 15.9 Ma but unfortunately these ages obtained from 14 flows did not correspond to the stratigraphic succession. Watkins and Baksi (1974) averaged the ages after combining these samples and identified an age range of 14.7 ± 0.2 to 15.9 ± 0.3 Ma. Baksi (1989) later adjusted these dates, arriving at an age range from 15.1 to 16.3 Ma, in an attempt to account for loss of radiogenic ⁴⁰Ar due to inclusion of fine-grained basaltic material in the samples. It is conceivable that the earliest PGBS flows, which do not occur at Picture Gorge, may be much older.

The oldest flows of the PGBS are assigned to the Twickenham Basalt made up of three members, from oldest to youngest: the Donnely Basin, Bologna Creek, and Muleshoe Creek Members (Bailey, 1989). The thickness of the oldest flows in the study area varies considerably in conformity with their eruption on a John Day surface of high relief. Bailey's (1989) study suggests that the Twickenham Basalt flows emerging from the Monument dike fissure system migrated along the synclinal/graben topographic lows, west to Donnely Basin and Alder Mountain, eastward to Monument Mountain and Zion Scope, and as far south of Kimberly as Holmes Creek. This flow pattern closely corresponds to the trend of the Mitchell-Monument synclines of Fisher (1967: figs. 2, 4) and his 2000-ft contour drawn at the base of the PGBS.

Evidently, the earliest PGBS flows of the Twickenham Basalt were following the structurally controlled, topographically low synclinal and/or graben structures that today correspond closely to the course of the John Day River from the vicinity of Monument westward to Kimberly, Haystack Valley, and southwest to Donnely Basin, Twickenham, and Sutton Mountain. Of particular interest in support of this view is that south of Kimberly the first flows of the Twickenham Basalt extended to the south only as far as the vicinity of Holmes Creek (fig. 2). These early flows do not extend southward beyond the Rose Creek fault system because they evidently abut against the fault scarps that existed at the conclusion of John Day deposition south of Kimberly. For example, in sec. 12, T10S, R25E, PGBS flows abut the Dead Cow Gulch fault, demonstrating the existence of the fault scarp at the time of eruption of these early flows (figs. 13, 14). Bailey's (1989) data can be interpreted to suggest that in time younger flows of the Monument Mountain Basalt (above the Twickenham Basalt) overtopped the faults and extended south to Middle Mountain. Over time the irregular John Day topography was filled by later flood-basalt to create a nearly level terrain: flows of the Monument Mountain Basalt are of considerable areal extent and by this time indicate low topographic relief throughout the basin (Bailey, 1989: 79).

Bailey (1989) remarked on the unusual thickness of flows of the Twickenham Basalt near their type locality along the John Day River, and his sections show that the earliest flows coincide with the present course of the river along the North Fork from Lower Camas Creek to Monument and Kimberly, thence westward to Bologna Creek, Haystack Valley, Donnely Basin, and Twickenham. It becomes evident that the present-day North Fork of the John Day River and its continuation west to Twickenham follow this ancient structural trend occupied by the earliest PGBS flows. This could only occur if at least some degree of moderate warping and/or faulting of the PGBS has continued along this structural trend.

Further evidence in support of this line of reasoning is found in the structural setting of the PGBS from Kimberly south along the John Day River to Picture Gorge. Fisher (1967: fig. 4) indicated that the modern course of the river followed a north-south linear topographic low filled by PGBS flows from Kimberly to Middle Mountain, and several tectonic maps of this area indicate the presence at this location of a north-south trending syncline (Walker, 1977; Brown and Thayer, 1966). This synclinal axis does not extend south beyond Middle Mountain, cut off by the east-west trend of the Richmond anticline (Fisher, 1967). That the Richmond anticline was elevated and eroded toward the end of John Day time is indicated by the angular unconformity of Rose Creek Member gravels on lower John Day rocks and the close stratigraphic approach of the Rose Creek Member to the Picture Gorge marker ignimbrite north of Middle Mountain (fig. 10, measured section 14). Thus, at the time that the earliest Twickenham flows erupted from the Monument dike fissures, the Middle Mountain-Richmond anticline area was a structural and topographic high, and only later did much younger flows of the PGBS eventually onlap this trend (Bailey, 1989: fig. 7, Branson Creek section). South of the Richmond anticline a separate fissure system erupted PGBS flows in the Picture Gorge area. The Richmond anticline-Middle Mountain high appears to have separated the flows of the Monument dike system from those of the Picture Gorge system to the south, “substantiated by the many flows at Picture Gorge that cannot be correlated with the thicker sequences in the central portion of the basin” (Bailey, 1989: 82).