After reconnaissance, three sites were selected for measured sections (Figure 3). The stratigraphic section at site 1 (Bay Village, Ohio) is 17 m thick and extends laterally about 330 m (Figure 4). The stratigraphic section at

site 2 (Bedford, Ohio), located in the Cuyahoga Valley National Park, is 16 m thick and extends laterals for about 500 m. The stratigraphic section at site 3 (Brooklyn Heights Village Park, Ohio) is 15 m thick and extends laterally for about 450 m. The locations were determined using Garmin GPS (±3 m accuracy). At each location, the stratigraphic section was described in detail, including lithology, composition, texture, and sedimentary structures [21] . Color was determined using a Munsell Color Chart [22] . Outcrop photomosaics were evaluated for stratigraphic architecture. Field work was augmented by 143 collected samples. A total of 68 samples were used to make thin sections for microfacies analysis, and 12 samples were used for SEM/EDAX analysis. Other samples were used for descriptions of sedimentary structures and trace fossils. In addition, a total of 56 paleocurrent measurements were collected using a Brunton compass, and evaluated using standard statistical methods [23] .

After evaluating the inventory of available well data, five wells in northeastern Ohio were selected (Figure 3). The basis of selection was: availability of drill core and geophysical logs from each well, and the presence of the CSM in the section. The cores are stored in the Ohio Geological Survey, H.R. Collins Core Laboratory, in Alum Creek State Park, Ohio. Details about the well locations, drilling history, core length, and core material conditions are given elsewhere [21] . In these five wells, the CSM ranged from 3.3 to 19.5 m thick. Cores were cleaned, photographed, and described using a binocular microscope. A total of 18 samples were collected to make thin sections for microfacies analysis and for SEM/EDAX analysis.

For each well, the available geophysical logs included the gamma-ray log, resistivity log, density log, and caliper log. The details about the geophysical logs and methods of analysis are given elsewhere [21] . The geophysical log pattern for the CSM is distinctive, with a notably higher gamma-ray signature, lower bulk density, and higher resistivity than either the underlying Chagrin Shale Member of the Ohio Shale, or the overlying Bedford Shale (Figure 5). All of these properties are consistent with the higher organic content of the CSM, in

comparison to the underlying and overlying units. The individual well cores could be correlated to their respective geophysical logs, both in overall thickness and such that contacts were about ±0.1 m matches [21] . At a finer scale, sandier intervals (interbedded with the shales) in well cores correspond to lower gamma-ray values, higher bulk densities, and lower resistivity values on the respective geophysical logs [21] .

Thin sections were prepared using standard techniques. Using the petrographic microscope at 25× and 40×, point counts were conducted on sandstone samples using the Gazzi-Dickinson method [24] with n > 300 counts per slide. Many core samples were chips too small for standard thin sections, and these were examined as grain mounts using the Leica MZ binocular microscope at 12.5× magnification. SEM samples were prepared using standard techniques [25] , sputter coated with gold-palladium, and examined using a Hitachi model S-2700 Scanning Electron Microscope, with an LaB₆ filament, acceleration voltage of 20 kV, beam current (spot size) of 10 A, and working distance of 8 - 12 mm. Elemental determinations used an EDAX SW 9100-60 energy dispersive spectrometer, with a bombardment time of 100 seconds, beam intensity of 20 kV, and sample tilt of 20⁰. Semi-quantitative EDAX analyses were performed using a standard method [26] .

Sandstones in the CSM are very fine- to fine-grained, poorly to moderately sorted, sub-angular to subrounded, quartz arenites or quartz wackes (QFL ratio is 92:7:1) [21] . Interbedded sandstones in the CSM represent about 11.4% of the total thickness of the unit. The sandstones contain a variety of primary sedimentary structures including massive bedding, hummocky stratification, planar lamination, ripple lamination, sole marks, fluid escape structures, load structures, and trace fossils. Both normal and inverse grading can be observed. Most of the lithics are sedimentary rock fragments (chert, shale, or siltstone clasts). Accessory minerals include biotite, muscovite, and chlorite. Cements are typically calcite and/or clay minerals. Black opaque grains in sandstones were discovered to be particulate organic matter using the SEM/EDAX [21] .

Siltstones in the CSM are very fine- to coarse-grained, moderate to well sorted, sub-angular to rounded, calcareous quartz siltstones. In the CSM, interlaminated siltstone and mudstone beds a few mm thick represent about 65% of the unit. The siltstones contain a variety of primary sedimentary structures including planar lamination, massive bedding, and micro-cross lamination. The siltstones exhibit both normal and inverse grading, in some instances as part of fining- or coarsening-upward sequences with sandstone. Burrows in the siltstones can be filled by silt- or mud-sized material.

“Shales” in the CSM include organic-rich mudstone, mudshale, claystone, and clayshale. Petrographic analysis shows that the content averages 3% sand (ranges from 0% - 5%), 37% silt (ranges from 5% - 65%), and 49% clay (ranges from 23% - 70%). Particulate organic matter (including plant debris) averages 10% (ranges from 7% - 20%); however this figure may underestimate the very finely disseminated organic component. SEM/ EDAX analysis shows that the principle clay-sized minerals are quartz, chlorite, mixed-layer clays, and illite, with a low (<1%) kaolinite content, consistent with earlier studies [11] [15] . Mud floccules are not observed, probably due to compaction and diagenesis [21] .

This study recognized 14 lithofacies, each defined as a unique combination of lithology, composition, texture, and sedimentary structures (Table 1). Each lithofacies represents a component of a particular depositional environment, and the groupings (or facies associations) and sequences of lithofacies aid in the recognition of ancient depositional environments. The CSM can be considered a rhythmically bedded sequence of mudrocks (siltstones and “shales”) interbedded with episodic deposits (tempestites, turbidites, and hyperpycnites) and hiatuses or breaks in sedimentation represented by condensed sections. Many of these features are fine-scale and can be difficult to observe in outcrop, but can be easily distinguished in polished core sections or in thin sections.

The sapropelite facies association is typically represented by laminated (lithofacies Ml) or massive (lithofacies Mm) black carbonaceous mudstone interbedded with laminated (lithofacies Cl) or massive (lithofacies Cm) dark gray calcareous claystone or marl. The interbedded character of the unit can be observed at different scales,

ranging from micro-laminae up to distinctive beds at outcrop scale. Such rhythmically interbedded pelagites have been recognized in the geologic literature as sapropelites [27] . On the macro-scale (≥10s of cm), some intervals are characterized by varying amounts of organic matter, such that more organic-rich layers average 13 cm thick and less organic-rich layers average 17 cm [21] . On the meso-scale (1 - 10 cm), bedding intervals have an outcrop appearance of rhythmically bedded (or “varved”) interbedded organic-rich black mudstones and carbonate-rich gray claystones or marls (Figure 6(A)). The stripped appearance is due to the black mudstones (Munsell Color Index N1) weathering various shades of brown (Munsell Color Index 5YR 4/4 to 10YR 4/2) interbedded with dark gray marls (Munsell Color Index N5) weathering to bluish gray (5B 7/1) or greenish gray (5G 6/1). On the micro-scale (<1 cm), closer examination using a binocular microscope (Figure 6(B)) shows micro-laminae within the sapropelite or marl intervals. Note that the textural difference between the mudstones and claystones is the presence of fine-grained particulate organic matter in the mudstones. On a micro-scale, there appears to be repeated packages of thinning-upwards, sapropelite micro-laminae within one marl layer (Figure 6(B)). SEM analysis of the clay minerals reveals both the sapropelite and marl consist of chlorite (Figure 6(C)), mixed-layer clay (Figure 6(D)) and illite (Figure 6(E)), with varying amounts of organic matter and carbonate. In summary, the nature of these “varves” appears to be alternating organic-rich sediment and carbonate-rich sediment.

The hemipelagite facies association consists of intervals of thinly interbedded quartz siltstone and sapropelite (lithofacies SSMl). An example is shown in Figure 6(F). These are interpreted as a variation of the sapropelite facies association that includes frequent episodic introduction of siliciclastic material (possibly as distal turbidites or distal hyperpycnites). Such deposits are classified in the geologic literature as hemipelagites [28] , and represent basinal environments influenced by adjacent sources of terrigenous sediment. In the CSM, hemipelagite intervals have a mean thickness of 19 cm.

The tempestite facies association consists of hummocky stratified very fine- to fine-grained sandstones (lithofacies Sh) overlain by planar laminated sandstone (lithofacies Sl), and ripple laminated sandstone (lithofacies

Sr), often terminating in wave ripplemarks (Figure 7(A)). The base of the sequence is typically erosional (Figure 7(B)), and may demonstrate groovecasts (Figure 7(C)) or flutecasts. In other cases the base of tempestites includes casts of tracks, trails, and shallow burrows made by benthic organisms (including Neonereites and Planolites) on the upper surface of the underlying mudrocks (Figure 7(D)). Some of the tempestites are amalgamated (i.e., one tempestite directly overlies another tempestite). This amalgamation forms when the upper part

of the previous tempestite was eroded by the storm event that emplaced the second tempestite. The tempestites in the CSM have a mean thickness of about 13 cm (range 4 - 21 cm), but amalgamated sequences can be up to 40 cm thick. These sequences of deposits are interpreted as storm deposits, in other words they are the results of combined flow due to erosion, re-suspension, and deposition by high energy waves in a shallow water environment [29] [30] .

The turbidite facies assemblagevconsists of massive normally graded sandstone (lithofacies Smg), planar laminated sandstone (lithofacies Sl), ripple-laminated sandstone (lithofacies Sr), laminated siltstone (lithofacies SSl) and laminated or massive mudrocks (lithofacies Ml or Mm). The base of each sequence is erosional with scour marks and tool marks (Figure 8(A)). The deposits can be recognized as turbidites, comprised of the characteristic Bouma sequences (divisions A-E), which can be abbreviated as T_abcde. In the CSM, most turbidites are incomplete Bouma sequences with the observed most common variants being: T_ab, T_abc, T_bc, T_ae, or T_be (Figure 8(B)). The average thickness of turbidites in the CSM is about 6 cm, but these range upward to 40 cm thick. The upper portion of individual turbidites often includes combined-flow ripples (Figure 8(C)), and also may include fluid escape structures. Similar deposits have been called “wave-modified turbidites” [31] . Previous studies have interpreted wave-modified turbidites are similar to those observed in the CSM as hyperpycnal-flow generated, prodelta turbidites [32] .

The hyperpycnite facies assemblage consists of thin, laterally discontinuous, sandstone or siltstone in thick sequences of shale (Figure 9(A)). In the field, weathering can make it difficult to distinguish hyperpycnites from distal turbidites or distal tempestites. However, when examined using the microscope, the characteristic properties of hyperpycnites [33] can be readily recognized. Two types of hyperpycnites are present in the CSM. Sandy hyperpycnites consist of massive, inverse- (lithofacies Smi) to normally- (lithofacies Smg) graded sandstones, while silty hyperpycnites consist of massive, inverse- (lithofacies SSmi) to normally- (lithofacies SSmg) graded siltstone. Sandy hyperpycnites may include a middle phase of planar laminated sandstone (lithofacies Sl) or ripple laminated sandstone (lithofacies Sr) (Figure 9(A)). The basal contacts are typically non-erosive. The upper and lower contacts typically grade into the “background” sapropelite deposits (Figure 9(B)). The presence of features such as escape burrows (Figure 9(B)) and fluid escape structures demonstrates that these are event layers. Most of the hyperpycnites in the CSM are <1 cm thick, but in measured sections the mean thickness was

found to be 1.7 cm, this value being artificially larger because of the difficulty of recognizing and measuring the thinner hyperpycnites in the field.

The condensed section facies assemblage consists of discrete layers of massive (lithofacies Mm) and concretionary (lithofacies Mmc) calcareous mudstones, often extensively burrowed. The concretions are vertically-flattened, discoid shapes in cross-section. Laterally, they range from individual nucleation sites (round) to irregular shapes caused by a number of coalesced nucleation sites (Figure 10(A)). The trace fossils in these layers are dominated by burrows such as Psilonichnus (Figure 10(B)), which contrast to the more typical tracks and bedding-parallel structures observed elsewhere (e.g., Figure 7(D)). The combination of carbonate cement, concretions, and vertical burrows suggests that layers were firmgrounds [34] , representing sediment packages that

became semi-consolidated due to low deposition rates or a hiatus.

Individual tempestites, and especially intervals of amalgamated tempestites, can be correlated over a distance of ≥35 km between the three outcrops (Figure 11). With less certainty, notable sandier intervals (comprised of a number of beds of tempestites and/or turbidites) can be correlated in the subsurface over a distance of up to 75 km [21] . Tempestites are the deposits of major storms in shallow-water environments [30] . These event deposits form in water depths above storm-weather wave base, but because of subsequent re-working by fair-weather waves, the deposits might not be preserved at water depths above fair-weather wave base [35] . In other words, tempestites are most commonly preserved in the offshore transition zone, at water depths typically ranging between 50 - 100 m deep (the boundary between the offshore zone and the offshore transition zone) to 10 - 30 m deep (the boundary between the offshore transition zone and the lower shoreface zone) [36] . In addition, some of the tempestites in the CSM show significantly erosive bases, the presence of intraclasts, and amalgamation. These “proximal tempestites” would be expected to form at the shallow range of water depths given above (i.e., near the boundary between the offshore transition zone and the lower shoreface zone) [37] . In summary, tempestites are shallow water storm deposits, and the ability to correlate individual sandy storm deposits over distances of 10s of km, in an overall muddy sequence, suggests an extensive, shallow water, muddy shelf depositional environment that was affected by episodic major storms.

Paleocurrent data using groove casts at site 1 are shown in Figure 12. Groove casts are bidirectional indicators, but the data are interpreted to show paleoflow to the west-southwest based on facies relationships, thickness trends, and earlier studies [19] . Paleogeographic relationships [10] suggest the major source of siliciclastic sediment was to the east-northeast, or in other words the Catskill delta facies (Figure 2).

The coarser-grained siliciclastics (sandstone and siltstone) in the CSM are event deposits: turbidites, tempestites, and hyperpycnites. These show two additional characteristics: 1) the importance of density underflows (turbidites and hyperpycnites) in transporting the coarser-grained siliciclastics into a mud-dominated marine environment, and 2) the importance of wave processes in modifying or re-working those coarser-grained siliciclastics (tempestites and wave-modified turbidites) within the marine mud-dominated environment. Our observations are consistent with recent findings that marine shelf sequences often include marine-mudstone encased, prodelta turbidite complexes [32] . Marine hyperpycnal flows are generated at river mouths during river flood events, where the river suspended sediment plume plunges to the seafloor, and the resulting hyperpycnal flow is capable of erosion, transport of sand and silt into deeper water, and deposition [38] . In the Book Cliffs (Utah), similar deposits in the Cretaceous Mancos Shale are interpreted to have been deposited at paleo-water depths of 10 - 20 m [32] . Density underflows were clearly the dominant process of transporting coarser-grained siliciclastic sediment, for example, one 14-m stratigraphic section contained 24 hyperpycnite event layers and 4 turbidite event layers.

The importance of wave modification or re-working in these deposits cannot be over-stated. Wave modification processes occur at or above storm-weather wavebase, which can be as deep as 100 m in some modern open shelf environments, but was more likely to have been significantly shallower (probably ≤50 m paleo-water depth) in these epicontinental seaway deposits. Further, the presence of proximal tempestites (amalgamated hummocky stratified sandstones) is suggestive of deposition toward the deeper end of the lower shoreface zone (possibly 10 - 30 m paleo-water depth). In addition, the presence of combined ripples in turbidites indicates that the turbidites were also deposited above storm-weather wavebase, and not deposited in basinal (below storm-weather wavebase) conditions.

The bulk of the CSM consists of carbonaceous black mudstones and mudshales interbedded with gray calcareous claystones and clayshales. The black and gray “shales” are interbedded on micro- (<1 cm), meso- (1 - 10 cm), and macro- (10s of cm) scales. The shales are also interbedded with wave modified or wave re-worked deposits, and trace fossils (predominantly bedding plane tracks and trails) are relatively common. These characteristics are not consistent with a depositional model requiring permanent water column stratification and anoxic bottom water conditions.

The observed scale differences in bedding patterns suggest multiple variables may be explaining the cyclicity. On a micro-scale (<1 cm thick), the rhythmically bedded black and gray “shales” represent variations in organic matter content. The black mudstones and mudshales contain more silt and particulate organic matter than the more carbonate-rich gray claystones and clayshales (Figure 6(B)). Rhythmic bedding on the micro-scale might, therefore, be attributed to the episodic occurrence of plankton blooms (marls) and intervening episodic flood events that generated hyperpycnal flows, and transported silt and terrigenous particulate organics (sapropelites) to the basin. The evidence would also suggest that any form of density stratification (i.e. formation of a pycnocline) would have been transient, interrupted by water column mixing inherent in wave re-working of the sediment (tempestites and wave modified turbidites). The rate of burial and anoxic conditions in the sediment may have played a key role in organic matter concentration and preservation.

Previous researchers have noted that variations in organic carbon (measured as TOC) and Fe-sulfides (measured as total sulfur or TS) in the CSM correlate to gray-scale density values [39] . The black (high TOC/low TS) layers were found to have large amounts of poorly preserved (i.e., bacterially reworked) organic matter, while the gray (low TOC/high TS) layers were found to contain smaller quantities of well-preserved organic matter of mostly marine algal origin [39] . These observations were interpreted to show that the black mudstones (with a higher silt content) represented lower bulk sedimentation rates, allowing bacterial reworking and concentration of organic matter, while the gray claystones were deposited at higher bulk sedimentation rates, allowing better preservation of organic matter and a greater flux of reactive Fe for precipitation of authigenic Fe-sulfides [39] . It is not clear from reference [39] which scale of cyclicity was evaluated, and these results are more consistent with our observations for meso-scale cyclicity described below.

On the meso-scale (1 - 10 cm thick), the “varves” evident at outcrop scale (Figure 6(A)) are packages of sapropelites-dominated or marl-dominated sediment, both containing the micro-scale rhythmicity described above, but showing greater or lesser importance of the black or gray shale end-members. Climate-induced cycles have been used to explain rhythmic bedding patterns in sapropelites. For example, Cenozoic sapropelites in the Mediterranean Sea area have been explained using wet-dry oscillations caused by precession-induced insolation variations, such that wetter conditions (northern hemisphere summer insolation maxima) favor increased continental run-off, higher nutrient flux to the basin, a reduction in evaporation and lesser bottom water formation [40] [41] .

Finally, on the macro-scale (10s of cm thick) the CSM may be interpreted as showing paleo-water depth variations. Most of the tempestites (and all of the amalgamated tempestites) are found in the middle part of the unit. This can be interpreted to show shallowing upward from the lower part of the unit to the middle part, and then deepening upward from the middle part to the top of the unit [21] . Whether or not such changes in water depth/ accommodation space were due to tectonism, such as isostasy and flexure in this foreland basin setting [42] , or were driven by Late Devonian eustatic sea level rise [43] , is beyond the scope of this paper. It has also been proposed that progressive changes upward in the CSM are a result of Late Devonian seawater chemistry variability [45] .