Geology of the Knife River area, North Dakota

Benson, William Edward · 1953 · Report

The Knife River area, consisting of six 15-minute quadrangles, includes the lower half of the Knife River valley in west-central North Dakota.

Public-domain full text preserved in the Mountain Man Mining Library. Original source: pubs.usgs.gov.

GEOLOGY OF THE KXIFS RIVER AREA NORTH DAKOTA By William E. Benson U. S. GEOLOGICAL SURVEY 'S FILE REPORT

This report Is preliminary and has not been ed:.ted oi' reviewed for conformity with U. S. Geological Survey standards and nomenclature

CONTENTS Abstract

r Introduction Location and extent of the area. Purpose and scope of the report Previous investigations Field work and acknowledgments Geography General setting Topography Drainage and water supply Climate Vegetation Culture and accessibility Stratigraphy . Advance summary I Subsurface rook Rooks not exposed in North Dakota Late Cretaceous and early Paleocene rocks Rooks exposed in Knife Mver area Historical Background Fort Union formation General History

Sentinel Butte shale Previous correlations

Results of current investigations 11 Golden Valley formation Previous descriptions Results of present work Tertiary system Paleocene series Fort Union formation - Tongue River member.. Distribution and composition Sand and sand stone / Shale and clay Lignite beds ' Clinker Silicified wood Silioified sandstone and shale

Measured sections Thickness Relation to adjacent formations Fossils Interpretations Eocene eeries Golden Valley formation Name and definition

Page Distribution Composition Lower member Local variations of "marker bed 11 kaolin olays "Marker bed" in Fort Union formation.. Upper member Oxidation and bleaching of upper member Oxidation of other formations Measured sections Fossils ' , Relation to adjacent formations Interpretation Younger formations not present in the Knife River area . Tertiary System Oligooene series "White River formation General Description Little Badlands and Chalky Buttes areas.. Other previously described localities ... New localities Miooene (?) or Pliocene-(?) aeries Gravels on an old erosion surfaoe

Page Surficial deposits Quaternary system

Pleistocene series Wisconsin stage Till (Qvrt) Distribution and topographic expression Composition Color and depth of oxidation Solution and redeposition of calcium carbonate "Stratified till" / Pebble-orientation studies Differentiation of tills Krem moraine Valley fills (Qsd) First fill, Qsd-L Second fill, Qsd2 ' Distr ibution Strati graphic relations Interpretation Third fill, Qsd3 Distribution and Interpretation Stratigraphic relations

Page Late Wisconsin gravels Outwash (Qwo) loe-contact deposits (Qio and Qio') Qic Qic 1 Missouri River gravels (Qrag) Early Mpconsin gravels Older glacial gravels (Qwg) Gravel on older cut terrace? (Qtg)... Older pand and gravel, undifferentiated f Cut terraces (Ot and Qlt) Summary and correlation of Pleistocene deposits Recent series Alluvium (Qma and Qal) Eolian sand and eilt Residual silica deposits Geomorphology and glacial geology of southwestern North Dakota Land forms Origin of the Missouri River trench General setting Hypothesis of glacial diversion Date of the diversion

Page Pre-diversion drainage of North Dakota Other Pleistocene diverrion valleys

Major regional diversion valleys Date of the diversion Minor diversion valleys 1B1 Wisconsin drift sheets of southwestern North Dakota. Previous correlation? Results of present work Stratigraphio evidence Evidence from the drift borders Early "Wisconsin drifts - lowan and Tazewell t lowan drift border Granite boulders southwest of the lowan border Tazewell drift border Late Wisconsin drift - Mankato Mankato drift border Land forms in the Knife River area General Statement Diversion valleys Major trenches South Fork and Elm Creek trenches Goodman Creek and Golden Valley trenches... Beulah trench Minor trenohes and notched divides

Page Abandoned " Drainage pattern of th- :fe River area Aligned drainage Hypothesis of re.-lcnal tilt. Hypotheses involvir- wind action Hypothesis of structural control Capture of barbed tritutarios ' Thickness of fill in tr- ajor valleys Mature topography of h-sdwater areas

Mass-wasting proce?-c e?

Subsidance ",

Soil creep '""" Earthflow """"

Slump blocks Ro national slinrp? ' Rookslide slvttf "

Structure .

General statement "

Structure contour map Folds """" " Faults ,, . Structures of Paleooene ae TV Deformed beds at Garrison Bam Subsurface evidence of Paleocene warping

Page Episodes of tertiary deformation in southwent North Dakota Geologic History Eoonomio Geology ,,,,... Sand and gravel Early "Wisconsin gravel deposits (Qwg,Qsg,Qt£,Qsd-,) Deposits of the intermediate fill (Q?d2) Deposits of late "Wisconsin outwash (QwOjQsdj) loe-oontact deposits (Qic, Qic 1 ) Missouri River gravels (Qmg) Ceramic clay Oil and gas possibilities ' Coal beds Physical properties Chemical composition ... .H Distribution and correlation of beds Beulah-Zap bed Coal beds above the1 Beulah-Zap bed Sohoolhouse bed ..; Twin Buttes bed Alamo Bluff bed Sohaffner bed 25? Coal beds below the Beulah-Zap bed Beulah area

Page Spaer bed Hazen"ET bed

Subsurface beds Hazen area HatonV bd Ha ten "B" bed Star bed Stanton area Coal Creek bed Stanton bed Knoop bo d 257 f Hancook bod 258 Looal bed 50 feet above the Stanton bed... Garrison area 259 Wolf Greek bed 259 Garrison Creek bed 260 Kruokenberg bed 260 Possible thick local bed' 260 Blackwater - Emmett area 261 Bed"CC" Bed"DD" Garrison Crek bed 262 Looal bed 110 feet above Garrison Creek bed 263

Page Correlations vdth beds in adjoining coal field? Mining anc development 2G4 Bibliography Illustrate ons Appendix - Log of Kelley-Lrutz 4'l well

ILLUSTRATIONS Pages Fig. 1 Index map showing location, of Knife River area and adjacent coal fields. in folder v Fig. 2 Sketch map of quadrangles of Knife River area Fig. 3 Physiographic provinces of North Dakota 274 / Fig. 4 Stratigraphic correlation chart Fig. 5 Strati graphic cros? section from Glendive Montana to Bismarck, Worth Dakota Fig. 6 Percentages of erratic stones in till 277 - 283 v Fig. 7 Composition of pebbles of gravel deposits... 284 - 290 Fig. 8 Rose diagrams of pebble orientations in till in folder v/ / Fig. 9 A. Relations between Qsdg and till sheets in northwest part of Hazen quadrangle B. Relations of Pleistocene deposits in Knife River valley near Eeulah 291 " Fig, 10 Cross section of "40 foot terrace" of Missouri River near site of Fort Clark. 292 / Fig. 11 Former drainage of North Dakota as inferred by Todd Fig. 12 Pre-diversion drainage of North Dakota in folder Fig. 13 Pleistocene diversion valleys mapped by Leonard and Alden. . Fig. 14 Structure map of North Dakota Figs. 16 - 20 Measured coal sections in folder Fig. 15 - Broncho quadrangle v Fig. 16 - Medicine Butte quadrangle Fig. 17 - Golden Valley quadrangle s Fig. 18 - Eeulah quadrangle Fig. 19 - Hazen quadrangle Fig. 20 - Stanton quadrangle / if, lf-33. !s

Plate 1 Geologic map of Knife River w-Stt-p'Vrf L4t.we.P- area - in 7 parts

Plate 2 Geologic map o"f Southwestern North Fakota in folder ' Plate 3 Drift borders and diversion valleys in folder / Plate 4 Pleistocene features of Knife River area... in folder i/ Plate 5 k Structure map of Knife River area in folder / Plate 6 A. Badlands of Little Missouri River B. Culture and accessibility Plate 7 A. Tongue River member of Fort Union formation B. Concretion from Fort Union formation Plate 8 Fossil wood from Fort Union formation.' Plate 9 A, Lower member of Golden Valley formation near Hebron, Morton Co B. Golden Valley formation in Lone Butte, McKenzie Co.

Plate 10 A. Upper member of Golden Valley formation near type locality B..Upper member of Golden Valley formation on Medicine Butte Plate 11 Salvinia preauriculata Plate 12 A. Golden Valley - Fort Union contact near Dickinson B. Golden Valley - White River contact in western Stark Co. Plate 13 Golden Valley - "White River unconformity in Little Badlands, Stark Co.

Page Plate 14 A. Gypsum crystal from Golden Valley formation B. Gravel-capped planation surface near Coffin Rutte Plate 15 A. Two tills separated by pebble band B. Two tills in valley of Kinneman Creek Plate 16 A. Lens of marl in till B. Distal edge of Krem moraine Plate 17 A. Crc-sbedded outwash fill, Qsd B. Qsd£ on soil tone, Beulah gravel pit % Plate 18 Soil zone, Qsd-, and Qsd2 at Beulah gravel pit Plate 19 Fine-grained sand and silt of Qsd£ Plate 20 A. Deltaio bedding in Qed near Golden Valley B. Qsdj near Riverdale Plate 21 Till on ioe-oontact gravel faoies of Qpd Plate 22 A. Small kame in Hazen quadrangle B. Largest erratic in Knife River area Plate 23 Detailed view of ice-contact.gravel in Beulah trench Plate 24 A. Missouri River gravel, Qmg, on till B. Alluvium on Missouri River gravel Plate 25 A. Earthflow in Broncho quadrangle B. Closeup of faulted Fort Union beds Plate 26 Rookslide slump in Beulah quadrangle Plate 27 Panorama of deformed Fort'Union beds at spillway excavation of Garrison Dam

Page Plate 28 A. Beulah-Zap coal bod at Dakota Star mine B. Beulah-Zap coal bed overlain by till

Table 1 Coal analyses i-n foldrr Table 2 Coal production of North Dakota Table 3 Coal production of Mercer County in 1950

The Knit'e River aroa, consisting of six l-rr.iriuto qu'dr-T.glos, 3 r.c~-ur. DD the lower half of the Kr.i fe River jvalley in west central North Dakota. Th-s area, in the center of the Will iston Basin, is underlain by the Tongue River member of the Fort Union formation (Paleocene) and the Golden Valley fnr~.a- t(.tion (Eocene). The Tongue River includes beds equivalent to the Sentinel /Butto shale the Golden Valley fomvtion, v/hich receives its first detailed description in this report, consists of two members, a lower ner.ber of f.ry to w' 1te sandy kaolin clay and an upper member of cross-bedded micaceous Gnd- s tone o ,

Pro-Tonguo River rocks that crop out in southwestern North Dakota i-.cludo the Ludlow raeir.ber of the Fort Union formation, the Carmonball marine foration (Paleocene) and the Hell Creek, Fox Hills, and Piorre formations, all Uppo- Cretaceous. Post-Golden Valley rocks include the UMte Fiver fomr t.ion (CJ i- 7 cene) and prcvels on an old planr.tion surface tht may be Miocene or Pliocene. lP-1 deposits include glacial and fluvi'-'.l deposits of Pleistocene ags and alluvium, dune sand, residual silica, and landslide blocks of Recent apo . Three apes of glacial deposits can be differentiated, Inrpely on the basis of three fills, separated by unconformities, in the Knife River valley. All t v.ree °re of Wisconsin ape and probably represent the lowan, Tazewoll, and M°nk"to subs tapes. Deposits of the Car"/ subs tap, e have not been identified either in the Knife River area or elsewhere in southern North Drkota 0 lowan glacial deposits form the outermost drift border in North Dakota. Southwest of this border are a few scattered granite boulders that are residual from the erosion of either the T-Thite River formation or a pre-Visconsin till. The Tazewell drift border cannot be followed in southern North Dakota. The Manknto drift border can be traced in a general way from the South Dakota State line northwest ''.cross the Missouri River and thrmigh the middle of the Knife River arv. The major land forms of southwestern North Dakota ?.re: (l) high hutr.es '] th'-t stand above (?) a gravel-capped pin nation surface and (3) a gently-rolling upland; below the upland surface are (4-) remnants of a brod valley stage of j erosion into which (f) modern valleys have been cut J The broad valley profiles' of many streams continue east across the Missouri River trench and are part of a '"ormer drainage system th-.t flowed into Hudson Bay. Crossing the divides are j (0 large trenches, formed when the former northeast-flowing streams were dammeby the glacier and diverted to the southeast. The largest diversion valley is ] occupied by the Missouri River; another diversion system, now largely abandoned extends from the Killdeer Mountains southwest to the mouth of Porcupine Creek in Sioux County. By analogy with South Dakota, most of the large diversion

valleys are thought to have been cut in Illinoian timeo ! Numerous diversion valleys of Illinoian to late Fisconsin age cut across

*he divides. Other Pleistocene land forms include ground and end moraines,

knmes, and terraces. Land forms of Recent age iriude dunes, alluvial terraces [ floodplains, and several types of landslide blocks. One type of landslide, j! called rockslide slump, has not previously been described.

if Drainage is well adjusted to the structure, most of the streams flowing ; . down the axes of small synclineso The bedrock formations have been gently folded into small domes and synclines that interrupt a gentle northward regional dip into the Williston Basin. Three episodes of deformation affected southwestern North Dakota in Tertiary tine: (l) intra-Paleocene, involving warping and minor faulting; (2) post- j Eocene, involving uplift and tilting; (3) Oligocone, involving uplift and gentl folding. j Mineral resources i?->clude ceramic clny, sand and p.ravel and lignite co°l. Th.e Knife River area is the largest lignite-producing district in the United ' States 4

INTRODUCTION Location .arid extent of the area Tho Knife River area is in west-central North Dakota, about

rtO miles northwest of Bionerck, end consists of six 15-minute vjundr'iMfiles, covering about 1,270 square miles. The Knife River rising west of the town of Manning in western North DaVota, flows eastward to -the town of Stanton, where it empties into the Missouri River. Its short course of about 90 miles is all in the soft Tertiary rock of the Grent Plains, and, in consequence, the stream is usually a muddy one, heevily loaded with fine sand and silt. The area covered by this report takes in the lower 4A miles of the Knife River and a short segment of the Missouri River. / Mercer County, the leading lignite-producing county of the United States, is nearly all within the K.-dfe River area, which includes also parts of Dunn, KcLean, Oliver end Stark cov_".tf. 5. Thp I ,-o-' t.: -r. .M* s.'.u? r*fci *.r.d i'~ rI?~Icn zo other coel bearing areas mapped by the U. 3. Geological Survey are shown in fig. 1, The six quadrangles that comprise the Knife River area are grouped in two tiers and form a capital "I" lying on its side (fig. 2). The four quadrangles of the northernltier are Stanton, rlazen, Beulah, and Golden Valley, end extend from 101° 15' to 102° 15' west longitude and from 4-7° 15 f to 4-7° 30* north latitude. .The two quadrangles of the southern tier are Medicine Butte and Broncho, and extend from 101° 4-5' to 102° 15 f west longitude and from 4-7° 00' to 47° 15' north latitude.*

current index maps s v:cv only tho nnrr.es "Erorcho" and "Gol-'-er. Valley." The other nargs on these r.aps are: A.A (l'rcdic5-e Butte), Kasrr.er (Belah), Krem (Hazen), and Deapolis (S tan ton) . Twe nnnes used in this report were formally adopted in 19/.£ and 19/+9 as the result of field work by both the Topographic and Geologic Divisions of the U. S. Geological Survey. In order to fill in a small unmappe'd area between the Fort Berthold Indian Reservation, the Minot area, and the six quadrangles just named, the Knife River area was extended north to include the southeast corner of the Blackwater and the southern- edge of the Ermett quadrangles (see fig. 2 ). Purpose and scope of the report The Knife River area was mapped as part of the Department of the Interior's program for development of the Missouri River B--sin. The main objectives of the project were to study the Tertiary and

Pleistocene, stratigraphy of the area and to determine the types and amounts of valuable mineral resources present. The Bureau of Reclamation's proposed Broncho Dam and Knife River Irrigation Project lie within the mapped area, and basic geologic data were desired to aid both in the construction program in the future development of the region. Also, accurate geologic maps of the coal-bearing formations were needed to aid in land classification and in estimating total reserves of the area because the opening of Truax-Traer's "Dakota Star Mine" rorth of Haren in 194A and the modernization of equipment in the mi">es

at Beulah and Zap skyrocketed Mercer County into the lead anong the lignite-producing counties of the country* The detilsd napping was supported b;- reconnaissance over much of the southwest psrt of ilqrth Dakota, ar.d the results of this reconnaissance will be sunr-rtrized in this report. Previo'is irvos 11 ?& 11 ons The geology of the Knife River area had not previously been mapped in detail, although the coal beds in certain pnrts of the area have been examined by a number of workers. W:ldr (1905, pp. 34-35) and Sir.ith (1903, pp. 19-24) described briefly scr<e of the ligr.ite outcrops along and rear tho i-'isso - :ri River in the Stanton quadrangle; Bauer and Herald (1921) r-apned the lignite beds.in the cart of the Fort / Berthold Indian Reservation thr t :s included in the northern parts cf the 3e-ilnh ancl Golden Valley quach-anlos; end Leonard, O'Vococr; and Dove (1925 ; pp. 11-122 ?.nd 325-131) ~-pped the outcrop of the Belah-Tap coal bed in the vicinity of Be'ilah, and describe outcrops of otheV beds near Hr.zen and Stanton. General str- J:i£rap;,ic studies in western North Dakota have beer, i-ade by Leonard (IT'-", 1°0°, 19n9 and 192?.), Then ind Dobbin (1924), Kline (1942), Sealer (1942), ::srnen (1943), ?-r.d Ercv/r. (lr-iS). The Pleistocene geology is discussed briefly by Todcl (19?i I-eonard (1916 and 1916), and Alden (1932). Clapp, Eabcock, and Loon-t-rd (1 -90£) studied so.Te of the clrys in the K/ife River area in a pper dealing with structural and ceranic cleys of \'orth Dakota.

Vood'.s (1904) reconnaissance of part of west-central North Dakota includes most of '-the Knife River area, and this paper is outstanding for the nur.ber of good observations that must have, been made in a very short time. Wood was one of the best observers and geologic thinkers who have worked in North Dakota. He was careful to differentiate fact from inference, and, although one may disagree with some of his interpretations, his observations and descriptions of outcrops are excellent, and his reasoning is sound. Field work and acknowledgments Field investigations were conducted in the,summers of 194& through 1949 in consequence of an allotment of funds from the U. S. Department of Interior's Missouri River Basin appropriation. I was in charge of these investigations and was assisted in 194& by R. E. Basile and R. B. Colton, in 1947 by R. B. Colton and F. Stugard, Jr., and in 194& by R. M. Lindvall, H. S. Mayberry and J. R. Scurlock. Except for the Stanton quadrangle, which has been mapped topographically, planimetric base maps for the Knife River area were compiled from General Land Office township plats and aerial photographs. The plats were used as horizontal control, and the drainage and culture were taken from the photographs. I compiled the base for the Broncho quadrangle; the Topographic Division of

the Geological Survey compiled the base raps for the rest of the aren. Except for the Stanton quadrangle, latitude and longitude lines are approximate*, and subject to correction. The field manning was done directly on aerial photographs at a scala of about 1:20,000, and the information ws s transferred to the brsc nv-ps with the rid of a vertical sketchmnstr. Primary vertical control was obtained from Coast and Geodetic Survey bench mirks, from the Stanton topographic rniadrangle, and by spirit leveling. Bench marks established by those methods were used e.s a basis for surveying with plane table and with a Paulin altimeter. The altimeter traverses were more rapid, and were considered sufficiently accurate for most of the work. , I wish to thank the various members of the Knife River party for their fine assistance and cooperation during the field studieso

Thanks are due also to many of the local residents of Torth Dakota, who aided the work in many ways, from supplying critical ii'rma tion on wells and mines to supplying tractors to aid our mired trucks. The cooperation of the following persons and organi7ations in supplying maps, well logs, .and other pertinent information is greatly appreciated: J. M. Hughes, Land Commissioner of the Northern Pacific Railway Company, U. S. Engineers, Garrison District; U. S, Bureau of Reclamation, Bismarck Office; and Dr. Wilson M. Laird, State Geologist of North Dakota. I wish to acknowledge especially the aid and cooperation of Professor R. F. Flint of Yale University and R. W. Brown of the Geological Survey.

Flint, who was in charge of regional Pleistocene studies in South Dakota, visited the area several times and discussed the field

relations of the placil deposits. Brovm, in addition to inti fying the fossil plants, visited the Knife River party for several days each suraner from 1947 through 19A.9 and worked with 113 on the general stratigraphic relations of the Tertiary deposits.

Fi. sketch map of thu physiographic subdivision of Korth Dakota, a hows that tn Knife River area lies entirely within the tLaciated portion of the Missouri r'lateau which in turn is a section of the Groat Plains Province (Fennemon, pa pea 2?6 to 277). "ho- Missouri Plu-teau occupies the western haif of north uokota and is divided inLo glaciated and unj/laciated portions. The eastern half of the State is occupied by the Western Lake section of the Central Lowlands province. The boundary between these two provinces is the Missouri Escarpment, which lies 30 to 90 miles t east of the Missouri River, and trends north-northwest from South Dakota to the Canadian border. The name "Goteau du Missouri" hs been applied to this escarpnent and to the hi$i ridge of morainic hills just wost of it but Andrews and Lemke (Andrews, 1939, 50; Lsmkc, manuscript in proparatior) quote a decision by the U. S. Board of Geographic Names to the effect that the name "Goteau du Missouri" should be applied to the part of the Missouri Plateau th.-.t lies between the escarpment and the Missouri River. This loaves the eccarp-uent itself without a nane, and I prefer the usage of (r/29, pp. 10-11) who .o.veL> the following description: "On the west cm border of the Drift Prairie rises an escarpnent which is evon mor-j abrupt than that w'.nch bounds the prairie on the east. This feature is the eastern ede of the Missouri Plateau and is known as the Missouri escarpment. This escarpuent trends northwest and pasces near and to the west of Crosby, Kcninnre, !'inot , Garriiifjton, and Jamestown. It is in general 300 to 600 feet above the lower plain to the est. The plateau, vrnich stretches westward from this escarpuont to aii i beyond the western border of the state, occupies fully hi.lf of the state and is a characteristic portion of the Or at riair.s."

East of th. J I'isaouri i'.rcarp i-?nt, the "..'estrn Laks section of Central Lowlands province is divided into the Drift Prairie and the Red River Valley. The Red River Valley in the oxtron.e e's.Ft r>m part of the state ic the floor of glacial Lake Agassiz an.} has virtually no r-'liof oxcopt for a few beach ri i; Tos of sar.d and gravel. The western boun iary of this old lake floor is the low .and indistinct Fer.-bina 'Escarpment, from which the Drift Prairie rJ jres gradually v/estv;ard. Tho Drift Prairie consists :>f a bedrock surface buried by 50 LO several htm.ired feet of late V/i scons in laci-'Q. drift, mor.tlygrouiid moraine. The ro'jjid surface has very

low relief, and the few hills that rise a.txjv/3 the r;ener s\irt'ace have bedrock cores. Drainage is poorly integrated, ,-jnd ana 11 lakes and ponds occupy shallow depressions in the, surface of the drift. 1 few Inrrr li'kes, such as Devils Lake caid Stump Lake, probably occupy former stream channels, now partly buried by the thick drift. Mantling the surface of the Kdssouri >Lscarpmer.t and cap pin r the edf.e of the Missouri Plateau is a large mora:nic belt, 5 to 20 railes wide, that has been called the "Altarnont nioro.5ne." Bedrock out0 crops are v-'Ty few and the illusion is ere ited that the whole hei-'ht of the escarpnent is due to the accumulation of a huge end mora:!ne. Gevc'.ral v/order;-:, hov.'cv(ir, have noted that bedrock cix>ps out in some of the highest hills in the i\oraine and alonf. the slos of the escarpment (How.ird, Gott, an'i Lindvall, l'?6, pp. 120-1205, -vnd Tnwnsend 'ud Jonke, 1951, pp. &*5-9A8) :m\ it now seoras certain tl t the io.Ai,ion uf oiie e.o.rpnit is control U-d by th buried bedrock topofrapi.y. The bedrock escarpuent, in turn,

probably controlled the location of the moraine by retarding the rate of advance of the glacier and keeping the edge of the ice

at or near the same position for many years. The evidence of bedrock control of both the Missouri Escarpment and its morainic cap has been discussed recently by Towns end and Jenke and is not treated in this report. Townsend proposes the name "Max moraine" to replace the old term "Altamont," because the moraine in North Dakota is not demonstrably continuous with the Altamont moraine at its type locality in eastern South Dakota; in fact, it is not even certain whether the two moraines are of the same age, I agree with Townsend,.and in this report I shall / use the term Max moraine. The edge of the Missouri Plateau must have been dissected into a narrow tract of badlands before it was covered by the Pleistocene ice sheets, for although bedrock crops out in some of the higher hills, Townsend reports that some drill holes have

penetrated up to 300 to 400 feet of drift before encountering bedrock. Southwest of the Max moraine the level uplands at the Coteau du Missouri are blanketed by drift 10 to 30 feet thick, but bedrock is exposed in nearly all of the small valleys. South and west of the Missouri River, the Missouri Plateau is a gently rolling upland ranging in altitude from 1300 to 2,800 feet above sea level. Numerous buttes and mesas stand above the general level of this upland, among which are Sentinel Butte, Bullion Butte, and H. T. Butte, the last named being the hip.hst

point in North Dakota, with an altitude of 3>468 feet. Into the rolling upland of the Missouri Plateau have been carved the mature valleys of the Cannonball, Heart, Knife, and Little Missouri rivers, all of which are tributaries to the Missouri River. Narrow tracts of badlands border each of these streams along some parts of its course, but none of the badland areas is comparable in size or in depth of dissection to the badlands of the Little Missouri Rivor. iThese badlands have a local relief of 200 to 600 feet and flank the river in a belt 2 to 10 miles wide throughout most of its course in North Dakota (pi. 6A). The Missouri River itself occupies a great trench-like valley f 200 to 600 feet deep and 1 to 3 miles wide. In contrast to the valleys of its tributaries, the Missouri trench is incised sharply into the surface of the Plateau with little or no regard for the regional slope to the northeast. The valleys of the Knife, Hart, and Cannonball rivers can be seen from many miles away because the land slopes gently down toward these valley floors, but in many places one can be within 4 or 5 miles of the Missouri River and see no sign whatsoever of the valley. In some places the surface of the Plateau actually slopes away from the trench of the Missouri. This topographic discrepancy between the Missouri River valley and the valleys of its tributary streams is due to the unusual origin of the Missouri River trench, and will be discussed in connection with the glacial history.

Although the northeastern psrt of the Missouri Plateau in Horth Dakota has quite obviously been modified by glacier.-;, the boundary between the glaciated and unrliciated areas is toporraphi<Jally indistinct. Southwest of the Missouri River the drift cover thins and the glacierc- had little or no effect on the topography. The actual drift border is marked only by the limit of erratic stones brought down fron Canada by the ice sheets. Toporraphy The Knife River area is v.'ithir. the glaciated portion of the Missouri Plateau but it has surface features typical of both the glaciated and unglaciated portions. The uplands in the northern and eastern parts of the area are / covered by glacial drift 5 to 20 feet thick with typical ground moraine topography. The Krejn moraine, a sirall recessional moraine with a local relief of about 20 feet, caps the divide between the Knife- and Missouri River drainages in the north'ern parts of the Golden. Valley, Beulah, and Hazen quadrangles. Bedrock in this part of the area is exposed chiefly in the stream valleys and in the narrow badland tracts that border the Missouri River. There are no prominent buttes or mesas and the nearly flat, till-covered uplands merge imperceptibly v;ith the gently sloping sides of most of the smaller valleys. Southwest of Beulah the gliciai drift thins, becomes patchy, and finally disappears except for the few erratic boulders of .granite and limestone that renain on the surface of the ground. The topography £ee:;s to have been r.odified very little by the ice sr:eets 0 iz

The interstrea-n divides consist of fently rolling rj*ass-covered uplands that f-~rn.de without sharp break into the maturely dissected slopes of the valleys. A more youthful topography is characteristic of the valleys themselves and the areas immediately border in ,r them. Bordering the Knife Rivor and eczne of its major tributaries are small tracts of badl'jidse The rr.cst prominent landmark in the Knife River area is Xeiicir.e Butte, a conical hill th.it stands about 200 feet above the surrounding uplands. Medicine Butte is in the v/est-central part of the quadrangle that bears its name and is clearly visible for many railes from the north, east, and west. To the south the upland fra;ually /' rises to a high point about 3 miles from the butte. The crest of this topo?raphic hirh area actually is a few feet higher than the top of Medicine Butte and effectively blocks a view of the butte from the south. The hifjiest point in the Knife River area is in the Golden Valley quadrangle in sec. 22, T. 146 N., R. 90 W. T! is point, which is occupied by the United States Coast and Geodetic Survey's Triangulation Station "Miller," is about 2,420 feet above s-a level (determined by Paulinaltimcter). The lowest point in the area is in the southeastern corner of the Stanton quadrangle where the flood plain of th-5 Missouri River is about 1,6?0 feat above sea level. The maximum relief, therefore, is about 750 feet, although local relief in most parts of the area is not more than 300 or 400 feetc

The most striking topographic feature of the Knife River area is the series of through valleys or trenches that cross the area from northwest tosoutheast with little regard for the major drainage pattern. Although these valleys are lrge, sore of them as large as the valley of the Knife River itself, they contain only small streams, or, in some segments, ho streams at all. Like the Missouri trench they are believed to have been cut by waters flowing at or near the margin of an ice sheet. The largest of these valleys are in the southern part of the area in the Broncho and Medicine Butte quadrangles and today are occupied by the South Fork of the Knife River and Elm Creek. These two valleys continue southeast out of the Knife River area, cross into the Heart River drainage basin, and come together just east of the town of Glen

Ulln about 15 miles south of the Medicine Butte quadrangle (see fig.13) They are a small segment of a long Pleistocene drainage course that extends from the Yollowstone River across to the Little Missouri River and from there to the Knife, Heart, and Cannonball River drainages, finally joining the Missouri River near Fort Yates in southern North Dakota. Other large trenches in the Knife River area are: The Goodman Creek trench, which connects the Little Missouri River drainage with Spring Creek in the Golden Valley quadrangle; the Golden Valley trench, which crosses the high area between Spring Creek and the Knife River in the Golden Valley, Beulah, and Medicine Butte quadrangles; and H

the Beuiah trench, which, connects the Missouri River with Spring Creek and the Knife River in the Ben lab. and Kasen quadrangles. The Beulah trench splits into two parts about 5 iniles north of the tovn of Beulah, One branch trends southwest and joins Spring Creek about 3 niiles east of the town of Zap; the other branch trends east to join Antelope Creek and then southeast to join the Knife River near Hazen. All of thess through valleys or trenches have wide, nearly flat floors and are partly filled with unknown thicknesses of alluvial and colluvi-al material. Drainage and -vjiter supply The Knife Rivor area is entirely within the drainare b.sin of the Missouri River. Runoff flows directly into the Missouri / River in the northern and eastern pi.rts of the area and into the Missouri River by way of the Knife River and its tributaries in the rest of the area, with the exception of a small trajt in the-northeastern part of "the Golden Valley quadrangle. This tract is drained by Hans Creek, which rises ne:-j* the Dunn County - Mercer County line, and flows northwest through the Goodinan Creek trench to the Little Missouri River. The Missouri River forms a snail part of the northern boundary of the area and flows south, through the Stanton quadrangle. The floodplain of the Missouri in this quadrangle is 2 to 3 niilss wide and has a gradient of about 1.6 feet p.sr mile. During the early pioneer d~ys the Missouri Riv-jr w is- an important route of travel between east and west and was ar. important source of water for domestic and livestock use. The Le-.ris and Clark .Expedition followed, the Missouri iiiver valley and the -site of Fort Mndan,

their first winter's camp, is in the southeastern part of the Stanton quadrangle. With the development of railroads and hjphways the Missouri River Valley was virtually abandoned as a route of travel, but it is still an important source of water supply for towns, ranches, and farms that are located along its course. In the Knife River area the town of Riverdale, which overlooks the Garrison Dam site, obtains its water from the Missouri River. The Knife River enters the area in the western part of the Broncho quadrangle and flows east and northeast to its confluence with the Missouri River near the town of Stanton. The flood plain of the Knife River is 1 to 2 miles wide and has a relief of about / 10 to 15 feet. Above its junction with Spring Creek, about 1 mile west of Beulah, the gradient of the Knife River flood plain is about 7 feet per mile; from Beulah downstream to the mouth the gradient is about 4 feet per mile. All of the major tributaries of the Knife except Spring Creek enter from the south. From west to east these major tributaries are: the South Fork of the Knife River, Willow Creek, Elm Creek, Beaver Creek, Brush Creek, Otter Creek, and Kinneman Creek. The South Fork of the Knife River and Elm Creek are underfit streams that occupy two of the trenches that cross the area. The rest of the tributaries appear to be normal in that they fit their valleys. Spring Creek enters the area in the southwestern part of the Golden Valley quadrangle and flows east across that quadrangle and the Beulah quadrangle to join the Knife River one mile west of

Beulah. Its flocdplain is about one cille vide and has a gradient of 12 feet per mile. Spring Creek's only l-.rge tributary in the area is Gooeb.an Creek, which rises in the western part of Kercer County and flows southeast to join Spring Creek near the town of Golden Valley. Both Spring Creek and the Knife River rise in the Great Plains and are therefore subject to great seasonal varjabions of the North Dakota climate* They are fed by enoui springs to insure a perennial flow, but the sunnier flow is so small that neither stream is used as a water supply for any of the srall towns in the area. With the exception of Riverdale all the towns and ranches in the Knife River area derive their water supply frora wells. A few / of the ranches situated along the bottomlands have shallow wells in the alluvial fills of the valleys, but for the most part both municipal and domestic wells tap aquifers in the Fort Union formation (Paleocene}. The well at Truax-Traer's Dakota Star Mine north of Hazen goes through the Fort Union, Cannonball (Paleocene), and Hell Creek (Cretaceous) formations and gets part of its water frora the Fox Hills sandstone (Cretaceous). The aquifers in the Fort Union foriaation are lignite beds arid some of the sandstone beds, As a general rule the lignite aquifers are better and give a stronger flov; o-f water. Water from the sandstone aquifers has a better taste i. / than that from the lignite beds but vit is also harder and is. therefore not preferred. V.'ater supply for livestock is derived in part from wells and in part from springs that issue from lignite outcrops*

In the central part of the Broncho quadrangle are three artesian veils; two are in the bottom of the Knife River valley and one is in --£ Tiller cf ZIr - fv;rv. vei;, ZcvrV;-; ir. x/, '.'. ~/.'y'.> R. 9Ckr ., is about 300 fe?t deep and its collar is about 150 feet above the Knife River. This well is not flowing but the head of water cones within a few feet of the surface. The head of water in this small artesian basin, therefore, is about 150 feet above the flood plain of the Knife River. Were'it not for this fourth well, the three flowing

wells might be catalogued in Fuller's narrow valley type of artesian system (1908, p. 39). But the presence of a head of water almost 150 feet above the Knife River floodplain at a distance of 2 miles from the river suggests that this is truly a small artesian basin. From the structure contour map shown in pi. 5 it can be inferred that the valley of the Knife River is about 150 feet lower structurally than the southern boundary of the area. The strata continue to rise to the south so that it is quite possible to get enough artesian head between the Knife River Valley and a point about 20 miles south of the area to account for these wells. Climate North Dakota, in the center of the great interior plains of North America, has a continental climate. Winters ere long and severe; summers, rather short and hot. The mean annual temper-'ture for the state from the years 1892 to 1949 is 39.6 degrees F., but this gives little indication of the extrere range of temperature. Winter temperatures are frequently below -20°) sur-mer temperatures are frequently above 100 .. The highest

temperature ever recorded by a U. S. Weather Bureau station is 124° F. at Medora in Billings County on Septmeber 3, 1921. The lowest official temperature on record Is -60° at Parshall in Mountrail County on February 15, 193&, but an unofficial low of -74° F. was recorded by the Lewis and Clark Expedition at Fort Mandan in the Stanton quidrangle on December 16, 1304. The average annual precipitation for the Gtate is between 1? and 18 inches, more than half of which falls during the swrimer months. The highest precipitation is in the southeastern part of the State in the Red River Valley, where the average is more than 22 inches per year. This average decreases to the west to a low of less than 14 inches at the Montana-North t Dakota State line. The average annual precipitation in the Knife River area is about 16 inches per year. Since 1940 average precipitation in North Dakota has been above or less than one inch below norral and the crops by and large have been good. The most severe year in North Dakota was 1936. The average precipitation for this year was only 8.83 inches and practically no crops were harvested. February of that year was the coldest month on record, and July the hottest month; the temperatures ranged from -60° F. at Parshall on February 15 to 121° F. at Steele in Kidder County on July 6. The growing season (the period between the latest and the earliest killing frosts) ranges from 110 to 120 days. Although this is a comparatively short season, the growth of crops is favored ttf the length of the days in the summer months. On June 21, the longest day of the yoar, the sun shines for about 16 hours.

Crop growth is also favored by two other factors, the large number of clear days during the growing season, and the fact that more than

half the annual rainfall comes during the summer months. Most of the summer precipitation occurs as local thunder showers GO that althoufji the average precipitation for the Stato may be high, certain local areas can have crop failures during a year when the average precipitation is above normal. The success or failure of wheat and othr pr -in crops depends not only on the location of the local thunder showers, but also on their timing. If showers fall when grain is sprouting and when it is filling out, a successful crop may be harvested even though the / precipitation is below average. If showers do not fall at these critical periods the crop may fail even though the annual average is above normal 0 Vegetation The Knife River area has a prairie type vegetation. Native grasses cover most of the uncultivated uplands and slop3s, -.r. r,rees are confined to the valley bottoms and to protected gullies where the water supply is more plentiful. Boxeldor, cottonwood, green ash, and white elm are the moot abundant types of trees, and of these only the green ash is coirunonly found away from springs or the bottoms of stream valleys. Small groves of green ash are found on the north end northeast sides of some hills where evaporation is slower than it is on slopes more directly exposed to the hot summer sun. Cottonwood is abundant along the bottoms of the major streams. Apparently it is limited to areas whnra the water table is within a few feet of the surface.

Rocky Mountain juniper, falsely called cedar, is quite con-non in the badlands of the Little Missouri River, but in the Knife River area the only n.'itive conifer is a low-creeping juniper, (ponderosa pine does riot occur natura-kly but has been rro'.vn successfully in some windbreaks.) The creeping juniper prefers steep, partially bare slopes and is abundant on the sandy clay outcrops of the basal member of the Golden Valley formation (Eocene). Bur Oak was found in some of the gullies adjacent to the Missouri River but appears to be absent over most of the area. Quaking aspen occurs in small groves usually at the heads of cullies near the Knife and Missouri Rivers. Several small varieties of willow border most of the streams that have water during the summer months, but the only species thit grows to the size of a small tree is the peach leaf willow. In the latter part of the 19th century many of the early settlers planted trees as windbreaks and shade around their homesteads; but the types of trees were ill chosen to stand the rigors of the climate 0 And the settlers did not realize the need for cultivating their rroves. As a result nearly all of these early plants soon died out. Today, however, improved nvithods of cultivating and a wiser selection of the types of tree has shown that windbreaks and shade belts can be successfully rov/n around the farms and ranch houses, Russian olive has been imported and is a major constituent of most windbreaks. ' Native fruits include will plum, western chokecherry. buffalo berry or bull berry, and service berry. Buffalo berry is most abundant in the stream valleys but is common also on some parts of the upland.

The following list of trees and shrubs is not complete but shows the corron types identified in the Knife River area: Conifers:

(l) Ground Juniper (Juniperus sp.?) Deciduous trees and shrubs: (1) Black Sage (Artemsia tridentata) (2) Boxelder (Acer negundo) (3) Serviceberry (AnelancMer florida) (4.) Dogwood (Cornus sp.) (5) Black Hawthorn (Cretaefnis (6) Green Ash (Fraxinus pennsylvanl ca lancnol n (?) Cuaklng Aspen (Populus tremlo3des) ' (8) Plains Cottonwood (Populus 8ar/?entii) (9) Wild Plum (Primus amricana) (10) Western Chokecherry (Pninus virp.iniana var. demiasa) (11) Wild Rose (Rosa hurr.il is) (12) Peachlenf Vfillow (Sallx PTnypdaloides) (13) Silver Buffaloberry (Shephrrdia arp.entea) (14.) Nnnnyberry (Viburnun lentrp.o) (15) White Elm (TTlnus anericana) (16) Bur Oak (uercus macrocarpa) Cultivated trees I (1) Maple (Acer sp.) (2) Russian Olive (Klaeaflnns ang.ustifoHs) (3) Vies tern Yellow Pine (Pinus ponderosa)

Native grasses cover i.io~t of the region that is not under cultivtic and arc on of the chief natural resources. The rnot ahur. i;urit types are t,he blue sterasj wo stern "wheat f.rass, slenaer wheat grass, r;u:ick prase, cord grass, wild rye, the hlua joints, fraima.3, buffalo rasr, and prairie June grass. The blue joints, sand graur, and cord granges are the principal hay crops on the moist bottom lands. Over-pauturinfj during uhe dry su:.::.er months has killed out some of the clirax --rasses in racily of the pastures v;iLh the ros-ilt of permarient injury to tr.icts. ILvery possible effort should be made to preserve the native gras-jes; they are naturally adapted to the conditions and it is improbable that introduced jonas can take their places satisfactorily. The vegetal cover has a pronounced effect on the rite of runoff and evaporation in this area. In wooded areas , the thick forest litter and humus soil retain moisture that falls as roin or snow, and the shade o£. the woods slows evaporation. In contrast, the prairie grasses that cover most of the Knife River area form a thick matted sod that is fairly impervious, preventing much of the rainfall from soaking in, and increasing the runoff from heavy showers. ThJs sod also holds the moisture from light rainfalls near the surface '-.-here evaporation is relatively rapid. The ground water table, therefore, is relatively deep; in fact most of the water tables are ctu'illy perched, in that they are confined to aquifers above and below which the beds are dry. The breaking up of the tough impermeable sod by cultivation has further lowered the water table in many places. Although the rainfall is absorbed more readily in the cultivated soil, evaporation is also more rapid and the new vegetation makes a heavy

demand upon the ground water. The result is lowered water tables, increased drying up of springs and streams, and a need for deeper ' f wells Culture and accessibility George Mcriwether Lewis and William Clark were not the first white men to see the Missouri River in North Dakota, but the diary of the expedition under their command is the first detailed account we have of tne region. They reached what is now North Dakota jn the fall of the year 1804, and built Fort Kandan, their first winter camp, on the floodplain of the Missouri River a few miles downstream from the mouth of the Knife River (sec. 14, T. 144,N., R. 84 This / stretch of the Missouri valley was at that time inhabited by four smaller tribes of Indians, the Mandans, the Ahnahaways (also called Wattasoons and Shoes), the Minnetarees of the Willows, and the Minnetarees proper a tribe of the Fall Nation. These tribes built lodges of wood, mud, and skins, and the sites of most of their villages are still easy to find, especially with the aid of air photographs. The locations of all but one of the villages occupied in 1804 were found and have been shown on the Stanton quadrangle. The large village north of the Knife River in sec. 21, T. 145 N., 84 W., was occupied by the Minnatarees, a tribe of about 450 braves. The village in the northern part of sec. 33 was inhabited by Minnetarees of the Willows, and the village on the hill near the center of the same section was that of the Ahnahaways, a tribe of about 300 people. The village in sec. 16, T. 144 N., R. 84 W., was one of two Mandan villages;

the other K-mdan villa f.e was on the north side of the Missouri nearthe expedition 1 s camp, and the site of this village was not seen. /: / The village near old Fort Clark in sec. Ik was probably also a Man dan

village that had been abandoned prior to 18Gfv . The two Kandan villages had a total population of about 500, all that was left of the tribe that a few decades earlier had totaled nearly 2,000 and had occupied : nine villages farther down the Missouri near the modern cities of Bismarck and Mandan. Disease and war had reduced the Kandans in ' number, and they had moved north to be near the other tribes for mutual protection against the Sioux. It was during the stay at ! Fort Kandan that the expedition obtained the services, as an interprets f of Jacques Charbonnoau, a French trapper, who lived with the Ahnahaways' with his two Indian wives. One of the wives was Sacajawea

(Sha-ka-ja-way-ha), who was later to earn a pl'ice in American history as a member of the expedition. [ , The Knife River area west of the Missouri was peopled, in l#Gk, : by the numerous and warlike Sioux nation. The Sioux were nomadic, and the sites of their villages are difficult to find, as they consist today of tepee rings, that is stones arranged in circles to act as anchors for the bottoms of the tepees. Most of these rinps have been obscured by the works of the white men, but a few can still be seen on those level parts of the upland that hve not been farmed. In the Broncho quadrangle, in sees 1 and 15, T. 1/+2 N., R. 90 V.'., are small fr,rassed-over pits d\i£ into residual concentrations of dark brown to

black flint. These pits are the work of the Sioux, who used the flint for weapons and other"implements. Evidently the Indians did net fashion their tools--around the pits, because, although broken pieces of flint are numerous, no half-formed or broken artifacts can be found in or around the diggings. Apparently the flint was dug,hand cobbed, and the best pieces taken elsewhere for manufacturing. The present population of the Knife Rivar area is abo'it 11,000. According to the 1950 census figure, Mercer County alone has 8,686 people, over half of whom live on ranches or farms. Beulah is the largest town, with a population of 1,501; Hazen, with a population of 1,230, is the only other town of more than 1,000 people. Prior to 1900 most of the towns in the Knife River aro.i wers along the Missouri and Knife rivers, with a few scattered in the upland regions. The small settlements on the Missouri uiver were important shipping centers for grain, anJ the village of Kren in -he uplands nsar the center vf the Hazon quaiririJle wis a ~.T.ll milling center. In the erly part of the 20th century the Northern Pacific Railway Company put a branch line across the Knife River area west to the town of Killdeer. Villages not adjacent to the railroad were gradually abandoned and the population shifted to the valleys of the Knife River and Spring Creek, the course followed by the railroad, A few miles south of the Knife River area another east-west string of towns follows U. G. Highway 10 and the main line of the Northern Pacific Railway Company,

Kost of the Knife Rivsr area is easily accessiblor/ 1 .'i-j.T.r/;- lie. The majority of the section lines have been left open :~s rirht-cf- w.-iys, and many of those have been graded and partly graveled. ?vo Ctite hirhwr.ys cross the area. Highway 49, an all-weather gravel road, nins north near the eastern edge of the Medicine Putts Quadrangle and connects Beulah v/ith U. S. 10 about 15 miles co,;th of the area. Midway 25 runs east-west across the central part of the area, following nearly the same route as the Killdeer Branch of the Morthern Pacific Railway Company except in the Gtanton qu?,dranf*le whero it turns south toward Hanover. This highway is paved in the Stanto/i quadrangle and is an all-weather gravel road in the rest of the area. / Several of the county-serviced roads in the northern tier of quadrangles have been heavily graveled ani are passable in moc-t weather. In the Broncho and Medicine Butte quadrangles, Hi/rhway 49 is the only all-weather road. Some of the other roads have been partly Traveled but some sections of them aro inp-issable after heavy rains. All surface transportation is subject to the vagaries of

the severe winters in North Dakota, and both the roads and the railroad may b>? blocked for many days after a severe blizzard. In the exceptionally severe winter of early 1949, the towns of the Knife niver valley were isolated several times for periods as long as 2 1/2 weks, during which the only method of communication and supply was by air. Stock ranching and dry farming are the chief occupations in this fertile but submarginal land. The 16-inch average annual rainfall makes dry farming a- gamble in the Knife River area. In yiars of average or above average rainfall the crops are for the most part excellent,

for the soils formed on the Fort T.Tnion fonr.-.tion -ir en Ue rlaci °.l .if:; it r. are '-oo-l. Durir:~ ~ry ye ir3 virtually no crorn are lv.rvosted, ;::i' -iurJrif the xtrcTicT*rlruht of the 10?0'c even the ar-t-ire lind cucc\ur.->d to the rifos of tho w-r-thor. The little liveSook v h -:t wr. k"pt dur.1.3 t'vjse ;:-!rr- Vr-d to r- fed by hay th"t VIE iT,"vortcd or thot had beon stored. N'3W irr.rovj 1 tochnir'uec o 4' dr;,r rrij.r.p; h'ivo improve-o the yield of crops enough that- n\or-t :f u,he f-.rr.ers c.in nuke a successful livir:.;; if only about 1/2 of the ya;\~s have above averare rainfall. It therefore see;.:s likely that birrinr .1 drastic change in climate, this area v.lll rerrain a conlr'r.ed rinchii-.f, and farra'.n region.

STRATIGRAPHY The strata underlying the Knife river area un-i adjacent parLs

of Xorth Oakota are described in five general categories: (l) the subsurface rocks nob exposed in the State but knov.n from drilling are 'briefly mentioned; (2) the late Cretaceous and early Faleoccne rocks knov.n from nearby areas; (3) the Paleocene and Cocene rocks exposed v.ithin the area; (4) younger Tertiary formations not present in the TAnife iuiver area; (5) surficial deposits. Advance Summary The Knife River area is just southeast of the center of the .."illiston Ba:iin, a shallow structural basin occupying most of v/est- / era Ilorth Jakota. The greater part of the "..illiston Basin is directly underlain by the Tongue River member of the Fort Union formation (late Paleocene). Early Paleocene and late Cretaceous rocks crop out around the margins of the basin in the southern and south/estern parts of t-he State; Exicene and Cli "ooene beds cap rome of the Vnghpr nnt-t.s underlie small synclinal areas in the central part of the oasin. Deep wells drilled to explore the possibi3.it.ies for oil and gas in the Vji3.li.ston Basin show that strata- ranging in age from Cambrian to late Cretaceous underlie but do not crop out in the basin. Only two bedrock units, the Tongue River member of the Fort Union formation (Paleocene) and the Golden Va3.1ey formation (Eocene) crop out within the Knifo River area. This classification of the Tertiary bods differ from that corunon3y accepted for the area. .'>iring the mapping it was found that beds former' called L'entinol ?utte shale (1'ocer.e) by C-gr ar.i other (1

ifl

contjtin a Talcocene florc- an.l i:rtorf:;_nj;er '..1th the Ton-'jiie River member of the Fort Union fom.-rbion. They have therefore beon included in the Tonrue River, and the Paleocene-Zocene boundary has been shifted upward several hundred feet in the section. The strata above this revised Tongue River member contain a wa catch (ertrly Eocene) flora and for these I have proposed the name Golden Valley formation (Beuscn and Laird, 1947, and Benson, 1949). Mantling the bedrock in the Kr.if e River area are thin glacial deposits of " isconsin age. At least three substages are represented, the oldest being the lov/an and the youngest probably the llankato substage. Deposits of the Gary cubotage may be missing. The border of t the late YJiaconsin (ankato) drift is rather indistinct but trends northwestward across the area through the Golden Valley, Broncho, and riedicine Bitte quadrangles. Soutliv/est of this line the drift is very thin and patchy arid in most places consists of boulders scattered on the beirock surface, Northeast of this border the till is from 5 to 20 feet thick on the upland surfaces, and bedrock is exposed chiefly in the sides of large valleys and in the bottoms of small ones. Pleistocene deposits other than till include glaciofluvial deposits, valley fills related to the glacial deposits, raid isolated gravel deposits of indeterainat-e origin* The stratiraphic succession of the valley fills and racial deposits is tho main evidence that three glacial substoes are represented in the Knife.River area. Recent deposits include windblown sand, landslide blocks, residual deposits of s.ilicified boulders, and two aes of alluvium, local'thick deposits of colluviuni, especially in some of the larger

valleys,have been mapped with the alluvium. Subsurface rocks '-; Rocks not exposed £n North Dakota The IVilliston Basin contains a thick series of strata, ranging in age from Cambrian (Deadwood formation) to late Cretaceous (iliobrara formation), that do not crop out anywhere in the State. These recks t' are not dealt with in this report, but have been described from drill loss by Ehlers (1943, pp. l6l£-l620) and by Laird (1944). A preliminary log of the Kelley-Plymouth Leutz //I well, drilled in the Iledicine Butte quadrangle in 1950, is given in the appendix. The present search for oil in the Williston Basin will certainly / yield a great deal more information on these buried strata within the next few years. Late Cretaceous and early Paleocene rocks The oldest rocks exposed in the Williston Basin of ilorth Dakota

are the Pierre shale and Fox Hills sandstone of the Montana group (late Cretaceous). The Pierre shale is a gray to black marine shale composed largely of montmorillonitio clay with local thin beds of bentonite. It is about 930 feet thick in the eastern part of North Dakota and thickens to about 2400 feet in the western part of the State. Under the Knife River area it is about 1200 to 1300 feet thick. In South Dakota the Pierre shale has been divided into eight members by Crandell (1950, pp. 2337-2346), but these members have not as yet been distinguished in North Dakota. Overlying the Pierre shale with a transitional contact is the Fox Hills 'sandstone, a marine formation consisting of gray and green

shaly sandstone and sandy shale that characteristically weathers to a rusty brown. Many of the sandstone beds are glauconitic. The Fox Hills sandstone ranges in thickness from about 60 to 35 feet along the eist flank of the Cedar Creek anticline in Bowman County, to nearly 300 feet thick along the missouri River in Sioux, I-'orton and r.Tj;ions bounties. Along the Cedar Creek anticline the upper part of the Fox Hills consists of a light gray sandstone known as the Colgate sandstone member. The Colgate member contains thin lignites and some fossil plants, but it also contains the fossil Halyrienites major, which is supposedly indicative of brackish water. In Sioux and *jnmons counties a light gray sandstone, the probable equivalent of the Colgate member, is locally present at the top of the Fox Hills formation, and appears to interfinger with a sequence of banded shale and sandstone containing marine fossils. The Fox Hills sandstone is disconformably overlain by the Hell Creek formation, the uppermost Cretaceous fonnation in Ilorth -jakota. The Fox Hills and Kell Creek'formations appear conformable structurally but the upper surface of the Fox Hills was eroded and channelled prior to the deposition of the Hell Creek. I have seen this disconfor::iity at Glendive, Montana, near Ilarmarth in Bowman County, Ilorth Dakota, and also along the Cannonball uiver in Sioux County, North Dakota. Qnith (Smith, V.'endell, personal communication) reports it east of the Missouri River in Eromons County and Jensen (Jensen, F.S., personal communication) has found it near Jordan, ilontana. The Hell Creek was formerly a member of the Lance formation, and was raised to formationrank .after it was discovered that the

Cretaceous-Tertiary boundary lies in the middle of the old Lance formation in North Dakota and Montana (Dorf, Erling, 1940 and Brown, R.L., vS 1938). The Hell Creek fonnation of North Dakota is now thought to be the equivalent of the type Lance formation in Carbon County, Wyoming. Except for the local Breien member in Uorton and Sioux counties, North Dakota, the Hell Creek formation consists of non-marine calcareous gray sandstones and brown to black carbonaceous and bentonitio clays. Thin lignite beds are fairly common in the lower third of the formation but the upper part contains no persistent beds of carbonaceous clay or lignite. Dinosaur bones are common and are easily found wherever tho Hell Creek crops out, and these, together with a typical late Cret taoeous flora (Brown, R.L., personal communication) are the evidence for the age of the Hell Creek. In south-central North Dakota a thin fopsiliferous marine sandstone interfingers with the lower part of the Hell Creek formation. This sandstone, the Breien member, contains a braokish-water fauna that includes Halymenites and Qstrea glabra, and records a local readvance of the sea into southern North Dakota. According to Laird and Mitchell (1942, pp 14-15) the Breien member lies about 20 feet above the base of the Hell Creek formation and grades both above and below into typical non-marine Hell Creek beds. The Cretaceous-Tertiary boundary in North Dakota ip marked by no Unconformity and by no sharp lithologic break. The basal Fort Union fonnation overlies the Hell Creek wi th a contact that is conformable at every place where I have seen it. In some places the colors of the two formations contrast sharply, but elsewhere even this color contrast i? absent. Typically the Fort Union is a lighter and yellower

gray than the more sombre Hell Creek but at some localities this color contrast is reversed and the Hell Creek is actually lighter than the basal Fort Union. According to Brown (Brown, R.L., personal communication) the most reliable lithologic criteria for mapping this contact are the persistent lignite beds that typify the Ludlow member of tho Fort Union formation. The dying out of the dinosaur? and the appearance of Tertiary elements in the flora both seem to take place just beneath these first persistent lignite?. Tho Fort Union formation (Paleocene), directly underlies most of the Williston Basin and the upper part of this formation accountsfbr most of the bedrock outcrops in the Knife River area. The Fort Union t formation in North Dakota has been divided into three members, the Ludlow lignitio member, the Tongue River member, and the Sentinel Butte shale member. The Ludlow lignitio member is the lowest member of the formation and doee not orop out in the Knife River area It is the approximate equivalent of the Tullook and Lebo members of the Fort Union formation in central Montana (See fig. 4 ). The Ludlow member consists of yellow-gray sandstones, light to dark gray shaly olays, and thin beds of lignite; it resembles the overlying Tongue River member except for a generally darker color and thinner and.less persistent lignite bodC' In many places in southwestern North Dakota it is impossible to draw a sharp boundary between the Ludlow and the Tongue River members. The Ludlow member interfingers eastward with the Cannonball marine formation. The Tullook and Ludlow members of the Fort Union, together with the Cannonball formation, were formerly designated as members of the Lance forma34

tion but the recent work of Prown (1933), Dorf (1940), and Fox anr Ros (1942), has shown that all of these are Paleocene and the Tullock ar.d

-U Ludlow are now considered members of ; Fort Union fonr.ation. The Cannonball marine formation was deposited in a shallow sea tha occupied central North Dakota during early Paleocone tim*1 . It intertongues with and overlie? the Ludlow member of the Fort Union formation in south-central North Dakota. Two tongues of the Cannonhall formation have been identified in the Ludlow exposures along the little Missouri River near Uarmarth (Hares, 1928, p.24, and Brown, 1948a, p,127l). At its type locality along the Cannonball River in south-central North Dakota, the formation consists of 250 to 300 feet of light to dark

brownish gray shaly sandstone and dark gray marine shales that contain an abundant fauna of pelecypods, gastropods, foraminifera, and sharks' teeth. The formation thins to the west, is concealed beneath glacial drift to the northeast, and until recently was thought to be restricted to southern and central North Dakota. Recent work by Lemke and Prown, however, has shown that the Cannonball persists into northern North Dakota and covers a large area in the eastern part of the Minot region (Brown, R.L., and Lemke, R.., 1948, and Lemke, R. W., 1950). Stanton (1920,pp 10-15) originally thought that the megascopic fauna of the Cannonball formation was a reliot of the Fox Hills fauna and indicated (M'Vr) a late Cretaceous age. But FOX and Ross show that the age of the ; megascopic "fauna can be Paleocene as well as late Cretaceous. They state also that the foraminifera are more closely related to the fauna of Paleooone Midway formation of the Gulf Coast region than thoy are to

any knov.Ti Cretaceous fau'ta. The Cannonball formation cror)s out "irsouri River as far north as the tovm of V/nshburn, about 10 miles

east of the Knifo River area. The raisinp; of the Cannonball to the rank of a formation, v.-hilr leaving the Tullock and Ludlow as members of thn Fort Union formation, rorults in the anomalous situation in which tho Fort Union formation both overlies and underlie." the Cannonball formation at its tyne locality. To me it would so em more logical to retain the Cannonball as a marine member of tho Fort I'nion fonuation, .just a? it wu? a marine member of the Lance formation, but in this renort I shall follow tne official usar.e of the Geological Survey, which derifnates the f Cannonball as a separate formation. Rocks exposed in the Knife River area

The sedimentary rocks exoored in the Knife River area comoriro tho Tongue River member of the Fort Union formation (Palrocono) and the olden Valley formation (early Eocene\ hero designated as a formation for tho first time. This ronre<-ents a revision in the strati rranhio section formerly accepted for this region, and, in order to clarify the terminology, the history of the naming; and correlation of the formations will bo discursed before the formationFare described in detail. Historical background Previous correlation of the various formations and members of tho late Cretaceous and early Tertiary strata in western North Dakota k are f.ivon in fif( .J£ The details of some of those correlations will now be discussed.

FORT UN'I ON F03VATTON General hi story

The Fort Union formation was first described by Meek and Haydon (1862, p. 432), who called it the Fort Union or Great Lignitic croup of bed?. Thp type locality, near the mouth of the Yellow?tone River in North Dakota, consists of beds now correlated with the Sentinel A Buttp shale (Brown, 1948*, O.1270), but Meek and Hayden included in this group all the beds overlying the Fox Hills sandstone and underlying the Wind River deposits in "Wyoming. Several years later Hayden (1869, pp. 89-92) extended the Lignitic group to include beds along the Rocky Mountain front as far south as Fenver, and possibly as far as the Raton coal field in southern Colorado. In 1876 Hayden replaced the term Lignitic group with the name Laramie groun (1876, pp. 20-27 and 40-46), and in 1878 (p.IV) he stated that the old Lignitio group evidently included beds of at least three different ages, namely, Wasatch on Fort Union on Laramie, By 1900 the correlation of the Laramie group had become so confused that Hatchor (1903) redefined Upper Cretaceous bed"; in eastern Vfyoming as the Lance Creek beds, later called the Lanoe Formation. Hatohor defined the Lance as the sombre-colored beds, containing ceratopsian dinosaur bones, overlying the Fox Hills sandstone and underlying the Fort Union formation. He considered the Lance to be the equivalent of type Laramie and to be Upper Cretaceous. According to Brown (personal communication) and Dorf (1940) Hatcher's type Lanco in "Wyoming is still valid formation and is of late Cretaceous age. When later workers ex37

tended the name Lance to the north and northeast of its type locality, they included beds that are now known to be Paleooene. In Montana tho Lance included the Hell Creek and Tullock members. Traced eastward tho Tullock intertongued with and was replaced by the Ludlow lignitio member, which in turn gave way to the east to the Cannonball marine member. Overlying tho Lance was the Fort Union formation, which in central Montana consisted of the Lebo and Tongue River members and in western North Dakota of the Tongue River and Sentinel Butte shale members. As stated above, by 1940 the great controversy about the Lance formation had been settled to the patisfaotion of nearly all the geologists who had worked into area and now it is/generally agreed that the Cretaceous-Tertiary boundary is at the top of the Hell Creek beds in Montana and the Dakotas and at the top of the original type Lance in eastern Wyoming. The Tullock and Ludlow have now been made members of the Fort Union formation, and the Cannonball has been raised to the rank of a formation. The Fort Union formation as now defined includes all tho continental beds of Paleocene age in western North Dakota and Montana. In central Montana it consists of the Tullook, Lebo, and Tongue River members. Traced eastward (fig. 4 ) the Tullock and Lebo merge and grade into the Ludlow lignitic member and in western North Dakota the Fort Union formation consists of the dark Ludlow member, light-colored Tongue River member, and, at the top, the controversial Sentinel Butte shale member. Sentinel Butte shale Previous correlations The Sentinel Butte ehale was first described by Leonard and Smith (1909, p.16) as a member of the Fort Union formation in western North

Thorn and bobbin (1924) called it Fort Union (?) but suggested that it correlates with tho so-called intermediate coal group of northern Wyoming, and with the Kincsbury conglomerate that overlaps the granite core of the Bighorn Mountain?. Thin overlap they regarded a? tho bane of the Wasatch and suggested, therefore, that the Sentinel "iitte ir of " anatch age also. Although this correlation was only tentative and did not have supporting paleontologic evidence, the Wasatoh ape of the Sontinel Putte shale has largely been accepted by later worker", including Seagor and others (1942) and Honnen (1943). Seager and others clarified the early Tertiary formation? of western North Dakota as follows (1942, pp. 1414-1415): Eocene series Wapatch formation Unnamed member Sentinel Butte shale member Paleocene series Fort Union formation Tongue River member Cannonball member - Ludlow member They stated also (p.1417) that the Sentinel Butte shale extends east from its type locality near Sentinel Butte and is the surface rock in moat of Dunn County and part of Mercer County - in other words, it underlies a large part of the Knife River area. They did not, however, describe the lithologies of the Tongue River and Sentinel Butte members in those counties, nor do they discuss the nature of the .contact between them. Hennen (1943) attempted a reconnaissance correlation of all the Tertiary formations in the western part of North Dakota and reaffirmed that the Sentinel Putte shale, which he too regards as Wasatoh, extends east to the Knife River area. He relied, however, neither on paleontology nor on lithologic differences between formations, but on the

assumed great lateral persistence of sand and lignite beds, and arsurr.motion not confirmed by the detailed mapping in the Knife River area. Hennon relied particularly on the assumed continuity of a silioified bed suopored to contain volcanic ash and silicified plant remains. This is his "marker sandstone 21", which he place? near the top of the Tongue River member. Brown and I together have examined this "marker sandstone" in the vicinity of Sentinel Butte and Medora, and we believe it is a silicified carbonaceous shale, probably formed along the margin of a coal swamp. we found no evidence of volcanic ash in the unit. Hennen 1 s belief that there is only one such bed in this part of the stratigraphic column is contradicted both by the earlier work of Hares (1928,pp.34-36) and by the detailed mapping in the Knife River area, where six similar silicified beds were found at six different stratigraphic horizons. Most of the objections to using silicified shales as marker horizons over wide areas were discussed by Brown (i948a, pp.1263-1269}; but Hennen (1948) eo strongly defended his correlations that I v.as led to recheck sorr.e of the field relations in the badlands near Medora. Individual beds, as well as the boundary between the Tongue River and Sentinel Butte members, are easy to trace in this badland area, and it is obvious that Kennen has "jumped section 11 in crossing the Little Missouri River. East of the river in southern Billings County, the silicified bed that he calls "marker sandstone 21" near the entrance of Roosevelt Park and near Sully Springs railroad station is actually in the Sentinel Butte shale. Vi'ert of the river in Golden Valley County the silicified bed that he also calls "marker sandstone 21", is 100 feet lower strixtigraphically and is in the Tongue River member. The Sully Springs silica bod actually

door extend vert into Cclco- Vrll*-- ? ? . r-r" Vv ? -*cl-.'.--:; :-;r.o:;r of Sentinel Rutte -hale northeast of Flattop Rutte. Elsev.'hore wort of the river, the Sentinel Rutte shale and, of course, tho Sully Springs silica bed, have been removed by erosion. Inasmuch as s correlations in central North Dakota are based on his identification across ereas of poor exposure, of "marker sandstone 21", I think it unlikely that his Tongue Hivr-Sentinel Butte contact in and around the Knife riivcr area is at the same stratigraphic horizon as the Tongue River-Snntinel Rutte contact near Medora and Sentinel Butte. Results of current investigations The mapping and study of the formations in the Knife Siver area in- / dicated that the beds above the Cannonball formation and below Seafar's "unnamed member" comprise but one lithologic and paleontologic unit. Their flora, according to Brown (personal communication), is a typical Tongue River assemblage. Trie ref ore, either the Sentinel Butte shale is absent from the area or it is Paleocene in age and inseparable from the Tongue River member. This inability to distinguish the Sentinel Butte shale in the Knife River area led Brown to reexaraine the whole problem, and he concluded that the Sentinel Butte shale is, as originally defined, an upper member of the Paleocene Fort Union formation (I948a, p.1272). At various times during the summers of 1947 through 1949 Brown and together examined the Paleocene and Eocene formations in western North Eakota in an attempt "to determine what happens to the Tongue River- . 5sentinel Butte contact east of Sentinel Butte and the Little Missouri "iver*. We reached the following tentative conclusions:

(1) The contact between the Tongue River and Sentinel Butte whale members of the Fort Union formation is essentially a color boundary, with little lithologic difference between the two member?. (2) This contact cannot be traced directly east because it dips in (- ' - " that direction into the "Wil 1 i pton Pacin and in concealed by younrer formations. It con, however, be traced along the Little Missouri River north and south from the type locality of the Fentinel Butte shale near Modora. To the south, the Sentinel Butte shale can be identified as far as the Marmarth coal field (Hares, 1928), beyond which area erosion has

(r

removed all the late Paleocene beds. To the north, the color contact can be followed, at or near the same stratigraphic horizon, as far as southern WcKenzie County, where the dip into the Williston Basin carries it below f the floor of the Little Missouri Valley. (3) Beds representing the approximate stratigraphic horizon of the Tongue River-Sentinel Butte contact reappear at the surface on the east aide of the Williston Basin in eastern McKenzie, northeastern Dunn and western Mercer counties. In this area, however, there is no color change. The section as a whole is dark, resembling the type Sentinel Butte shale; but it also contains numerous light beds that resemble the Tongue River. (4) The eastward darkening of the section is probably due to eastward thinning of the Fort Union formation, especially the Tongue River member. Near Medora the combined thickness of the Tongue River and Sentinel Butte members is between 1,000 and 1,500 feet and sand comprises about half of the section. In Mercer County, the thickness of the Tongue River-Sentinel Butte beds is probably less than 800 feet, and the section is 60 to 65% gray shale. Also, as the total volume of sediments decreases, the relative abundance of carbonaceous material increases,

causing a darkening of the color. It is not surprising that the color contrart betv/een the Sentinel Putte shale member and the Tongue Rivermember doe? not perrist as far east as the Knife River area. (5) The Sentinel Butte shale, therefore, is ir.appable as a separate mentor of the Fort Union formation only near its type locality in w-rten North Dakota. To the east it appears to intertongue, both laterally and vertically, with the Tongue River member. We therefor suggest that tho name "Sentinel Putte" be used only in western North Dakota; and that bods of equivalent age in the central part of the state be included in the Tongue River member of the Fort Union formation. GOLDEN VALLEY FORMATION Previous descriptions / In numerous isolated localities in western forth Dakota the Tongue River and/or Sentinel Putte shale member of the Fort Union formation is overlain by beds of light-colored clay and sand that have received little attention since 1906 when they"were firrt described by Leonard, Clapp and Pabcock in the Fourth Biennial report of the North Dakota Geological Survey. The author?"of this report described numerous outcrops of white sandy fire clays overlying the Laramie formation (which then included the Fort Union formation) on the interstream divides of west-central North Dakota. Leonard said of these clays (p.88): "These white clays, whiofe cover an area of approximately 4,000 square miles, lie at an elevation of from 2,450 to 2,600 feet above sea level and are confined to the tops of the higher ridges and divides. They have a maximum thickness of about 150 feet. "The fire clays are remarkably uniform over tho entire area, not only in appearance but in chemical composition as shown by analyses of samples from many different localities. Their white color makes them conspicuous wherever they are exnosed."

About 25 outcrop areas of these clays were examined in detail and chemical analyses and detailed ceramic tests were made on numerous samnles. Concerning the stratifranhic position of these clays the authors said only that they are of Tertiary age and overlie the Laramie formation. In the Chalky Puttns area in Slope County, somr beds of Vthite Rivrr

sandstone were included in these "light colored Tertiary clay."

In 1911 Leonard again described these clays briefly and considered them as upper Fort Union (1911, p.535). He stated that they are well exposed near Diokinson and Gladstone and several miles north of Hebron. In the same report (p.534) Leonard described the Fort Union formation along the Little Missouri River in Billings County as consisting of two members, subsequently called the Tongue River and Sentinel Butte shale members by later workers. Leonard did not state what he believes to be the relationship between the dark Sentinel Butte shale and the lightcolored upper unit near Diokinson. The only other paper that describes these light beds is the discussion by Seager and others (1942, p.1416), who considered thorn

"Wasatch (Eocene). They state: "The Wasatoh formation of Eocene age is represented in North Dakota by two members. The younger is an unnamed light-colored unit, well exposed north of Hebron, North Dakota. The lower member is the j Sentinel Butte shale, a dark bentoniticVuit whose type locality is south of the town of Sentinel Butte, North Dakota." Seager 1 s only evidence of a Wasatoh age for the "unnamed member", is the fact that it overlies the Sentinel Butte shale; and, as stated above, the Sentinel Butte shale is no longer regarded as "Wasatch in

Remits of orerent work In June 1947 the Xnifa River field party discovered specimens of the floating fern Salvinia preauriculata Perry near the base of "ur.r.ar.ed ze-lrcr" in the hill? north of Eecron. is considered by Brown to be diagnostic of the lower v.ocene and it therefore appears that the "unnamed member'1 is actually Eocene and the only Eocene formation so far recognized in North Dakota. After the discovery of Salvinia the "unnamed member" was studied in detail in the Knife River area and mapped in reconnaissance in the rest of southwestern North Dakota. It was proposed in 1947 (Penpon and Laird, 1947, pp. 1166-1167) to call these Eocene strata the Golden Valley formation and this name has been tentatively adopted pending the detailed description and definition given in this report. Because of his great interest in the stratigraphy of the State, I originally invited Dr. "Wilson M. Laird, State Geologist of North Dakota to join me in the study and Dr. Laird spent a woek with me during 1947, contributing many valuable ideas in the early stages of this -work. Unfortunately thp pressure of other duties made it necessary for Dr. Laird to withdraw from this study. In subsequent years H. V. Brown of the Geological Survey spent several days each summer with me and his groat knowledge and experience in Tertiary stratigraphy were of inestimable value in the reconnaissance study of the Eocene strata. Tertiary system

PALKOCENE SEKIES Fort Union formation - Tongue River member Distribution and composition For reasons already discussed, all the Paleooene bed? between the

top of the Cannonball and the bottom of the Golden Valley fornations have been included in the Tongue River member of the Fort Union formation. Beds stratigraphically equivalent to the Sentinel Butte shale : probably comprise the upper part of this Paleocene section, but they j cannot be distinguished either lithologically or paleontolofically from : y>e cc -a ts e - ' c e J-'-CT-O'' A *-oe c-N .c - a , "-J f T*"1 7' IK t, ( The Tongue River member directly underlies most of the Knifv r\ivr area and is well exposed in the dissected areas bordering the Knife River, Spring Creek, and the Missouri River. In this area the member consists of beds of non-marine sand and sandstone, silt, shaly clay, r and lignite. Thin beds of lenticular limestone concretions are common especially in the sand beds or at the contact between a bed of sand and a bed of ehaly clay. Contacts between beds, especially between sand and clay, are usuall; completely gradational both laterally and vertically, although a few exposure? show sharp erosional contacts at the base of a sand bed. Then fore, correlation of the beds over areas of poor exposure is difficult and uncertain at best. t ! Sand and sandstone Beds of sand and soft sandstone comprise about 30 to 35 per cent of the Tongue River member in the Knife River area. They range in thickness from a few inches up to 50 or 60 feet, and are evenly distributed throughout the reotion, except that the upper 100 feet of the member, which tends to have more and thicker sand beds than tho lower part of the seotion. Despite the great thickness of some of these bods, they have little lateral persistence. Certain beds of sandstone 20 or 30

feet thiok pinch rapidly and die out in a few hundred feet, so that in many plaoos it is difficult to matoh accurately sand beds in sections measured less than half a mile apart. The color of the sand beds ranges from buff and light yellow gray to medium gray with the various colors distributed uniformly in the section. The lighter-colored beds resemble the sands of the Tongue River member in eastern Montana, except that they seem to contain le ss white mica. The darker sand beds resemble the sands of the Sentinel Ritte shale near its type locality. Most of the sand beds are fine-to medium-grained j a few are coarse grained, and two beds near the base of the member in the southeast part t of the Stanton quadrangle are conglomeratic, the "pebbles" consisting of clayballs. Tho sorting is variable. ?ome beds are well-sorted and consist chiefly of sand-sire particles; others are very poorly sorted and contain large percentages of silt and clay. There are all gradations between clayey sand and sandy clay. Bedding is more easily seen in the beds that have been indurated to sandstone. Although most of the bedding is even and horizontal, ripple bedding is common in many of the fine-grained beds. A few of the coarse beds show out-and-fill stratification, with the forcsets having a general easterly dip, Megasoopically -She sand grains appear to be mostly quartz, with some grains of angular oaloite, a few pieces of fresh to weathered feldspar and a suite of heavy minerals that do not constitute more than 1 or 2% of the grains. No microscopic study was made of these heavy minerals but in the Heart Putte quadrangle, about 35 miles southeast of the Knife Hivr area, Ti?dale indicates that these heavy minerals 7/ero probably

derived fron metamorohic and ineoMP rocks (1941,'po.30-31). Tisdalo f-jrther states that whil pcrr.e of the rrains show little round Inr and are probably in thir firt cycle of sedimentation, other? are >t*llrour.de-d and orcbably record a second or third -ne rat ion of sec'ir-.er.- tation (pi. 7D). Some of the sand beds are weakly cemented by calcium carbonate or iron oxide to form a crumbly sandstone. The iron oxide cement in r>robably secondary, the result of th wathorin/r of a sandstone whose original cement was a ferruginous calcite. In a few bd.s the volume of the calcite cement ir> greater than than of the sand prrainp. Vany of the pand beds have little or no calcite cement but do have a considerable admixture of clay and harden to a compact soft pandptone on surface exnoaure. btill other beds are composed chiefly of fine-trained quartz and are not indurated at all. Many of th pandton-0 bds are made up of a verier of large cnl-areo\i concretions. Some there concretions are nearly pure limestone and, in nlacep, conta-'n fossils of plant* or fresh-water gatroDodp. Mopt of the uncementeri sand bdp weather lip-.ht yellow to yellowpray. The indurated bedp weather light <ray to dnrV brown, deoendinp; largely on the iron content of th oementinr. material. Some of the sandstone is spotted with small balls of limonite, which rtain the surface a yallo-wish bro'ATi. A few of the<;e balls contain cores of unweathered iron sulnhide, t>robably marcasite. Seams of limonite arc

common both in the sandstones and in the shales. The source area of the Port Union formation wan probably to the wept, in the Rocky Mountain area in northern "rtyomin,"-, and Montana.

Tisdale (1941, pp. 30-31) found that some of the garnrt frapnrnts in the sandstones have magnetite inclusions similar to garnets in thr mrtamorphic rocks exposed in the Black Hillg and suggests that the lack Kills may have been a source for some of the Fort Union sedimpnts. He does not believe, however, nor do T, that the Black Hill? supplied all the sediment? of the Fort Union formation in North Dakota. First of all ' the volume of the Fort Union appears to be too great to have coin from ' so small a source area. Second, in the Knife area the sandstones with cut-and-fill type crossbedding show foresetting to the east, indicating that the major streams probably came from the west rather than the south Shale and clay f Beds of shale, shaly clay, and clay make up about 60 or 65 per cent of the Tongue River member in the Knife River area and rare in thickness from a few inches to several tens of feet. They range in color from light to dark gray and brown, depending largely on the carbonaceous

content. f Most of the shales and shaly olays are silty to sandy, but some ' beds of pure plastic clay occur, especially just above or below a bed of lignite. The bedding is horizontal, and some of the silty shales show fine laminations. The hard dry surfaces of the clay show varying amounts of cracking, and this together with the gray color' has led many workers to call these beds "bentonite" and "bentonitic clay." The terms, bentonito, bentoniti olay, and volcanic ash have been applied very loosely to many of the Tor tiary sediments in the Great Plains. Any clay that cracks on outcrop is said to be bentonitio; any white sandstone is said to contain volcanic

ash, Most of the beds reported to contain this volcanic material have

not been investigated in the laboratory, which would seen to be essential before the term? can be safely used. In the Xnife River area I found no true bentonite and only a few beds that are probably bentonitic clay. The results of petrographic tests of the drill corns from the Broncho dam site which were furnished by H, W. Kirchen of the Bureau of Reolamation indicate that most of the clay beds are composed of a mixture of montmorillonite type, illite type, and kaolinite type clays, and that the illite type and kaolinite type clays in general predominate over the montmorillonite type. One or two beds of dark gray clay do appear to swell up to 100 percent and these may actually r be bentonitio. The other beds, however, appear to swell very little and the cracking on the surface is almost certainly due entirely to the drying out of the beds.

Many of the silty and sandy clay beds appear to have some calcium carbonate cement and nearly all of the clnys, with or without this cement, "bako to a hard surface in the summer sunlight. This induration, however, is a surface phenomenon and is temporary; when wet the olays regain their original plasticity. Lignite beds Reds of lignite range in thickness from a few inches up to many feet and are the most persistent beds in the Tongue River member. Kven the lignites, however, are unreliable as marker beds across areas of poor exposure, for their lateral persistence is extremely variable. For example, the Twin Buttes lignite bed in the western part of tho Medicine Butte. quadrangle is a thin impure lignite that only locally exceeds

2 feet in thickness, yet it persists over about 100 square miles in this quadrangle and is easily identifiable by a thin clay parting above the middle of the bed. In contrast to this other beds of lignite 5 or 6 feet thick pinch out into clay or sand within a few hundred feet. This rapid pinching of the coal bed doe? not aprtear to be due to later channeling but rather is a facies change. The thick Beulah-Zap lignite bed crops out in over two-thirds of the Knife River area and is the best horizon marker in the Tongue Rivermember of the Fort Union formation The appearance and physical properties of the lignite are described under the section on Koonomic Geology. Clinker-- r Many of the thicker beds of lignite have burned along their outcrops and the fires have fused and baked the overlying sediments to form a red clinker, or "pseudo-scoria." Where the overlying beds were sandy or where the fires were not excessively hot, the alteration involved baking with little effect on the original sedimentary textures. Where the beds were clay and where the fires had a good draft, some of the strata havo been fused and resemble the slag from a blast furnace or the real scoria of a lava flow. In fact, the first English description of the clinker beds of North Dakota appears to be in the diary of the Lewis and Clark expedition, where these rocks were called pumice stone, implying that they were of volcanic origin. After they have been baked and clinkared, the shales and sandstones are much more resistant than most of the other beds in the Fort Union formation. Many benches and small isolated buttes owe their existence to their protective cappings of clinker.

The fires that "burned the lignite and baked the overlying sediments have boen ascribed to spontaneous combustion, lightning, and prairie fires, but the latter two causes seem totally inadequate to explain tho great amount of burning that ha? taken place from time to time. That lignite can catch fire by spontaneous combustion was demonstrated in the summer of 1949 when a pile of lignite, stored experimentally near the Garrison Dam in the Stanton quadrangle, started to smolder vrithout benefit of lightning or any other external fire. The burning of tho lignite beds proceed? inward from the outcrop for varying distances depending on tho thickness and character of the overburden and also upon the thickness and quality of the lignite itself, t Thin bed?; burn, but in general the combustion of these beds is slow and generates little heat. The clinker thus produced is usually thinnor and lighter in color than the clinker produced by the burning of a thick bed Vhen a thick bed of good quality burns the volume of the bed decreases to about 20% of the original volume. This causes tho overlying strata to slump and crack, creating new paths to supply oxygen to the fire below. If the bed crops out on a steep slope and the overburden is thick these cracks may not extend to the surface of the ground above, and burning will stop a few feet back from the outcrop. If the slope on which the lignite crops out is gentle and the overburden is thin the , collapse and cracking can continue to supply air to the fires for as much as a quarter to a half a mile back from the outcrop, and thr bed will burn out thoroughly for this distance. Once the bed is set on fire it can smolder and burn for many years. The manner in which these* lignite beds burn is well phown by the so onllod burning coal min nort

wort of Anidon in Slot)** County, ?outhwestern North Dakota. Here, in Sec. 14, T. 135N., ft. 102V.., th Harmon lignite bod is actively "burning today and apparently ha? been burning continuously since before 1911 wnn Kar<s manned the Varmarth lignite fild. Hares report" that this bed WRP burning with intense heat and that large snowballs rolled into the opening of one of the vent? were converted almost immediately to steam (Hares, 1928, p.51), The burning of the lignite beds and the formation of the clinker has evidently gone on ever since the Fort !!nion formation was exposed to weathering and erosion. Pieces of clinker are found in the oldest glacial deposits in the Knife River area, and as. these deposit? are r thought to be of lowan ag-e it indicates that the beds were extensively burned prior to "Wisconsin time. On the other hand, in the northern. r>art of the Beulah quadrangle in many parts of T. 146N., R.88W., till of late Wisconsin (Mankato) age has been clinkered by the burning of the Beulah- Zap bed. Some hand specimen? of this clinker show in themrelvr? two generations of burning, for the baked till includes pieces of older clinker. Har? (l n28, p,F?) shows that the VThite Hiver formation in southwest North Dakota and eastern Montana contains a few fragment? of ancient clinker, and I have found two small pieces of what appear to be clinker in the VJhite River sandstone on West Rainy Butte in Slope County, Thus, the burning of the lignite.bds seems to have started in th'1 Tertiary period as soon as the streams carved deep enough to expose the Fort Union formation, and has continued through to the present. Silicified wood-- Silicified stumps and lop-r occur at many strati graphic hori rons

in the Knife River area. Usually these stumps are found at the top of a bed of coal or carbonaceous shale but locally they are in sandstone beds Most of the fossil wood retains enough carbonaceous material to color it dark gray to black. This dark color, however, can be seen only on a fresh break, for on exposure it bleaches rapidly to light yellow or white. In general the original cell structure of the wood has been well preserved, and is easily seen with a hand lens. Although the Fort Union '

formation contains abundant fossil leaves of deciduous trees, all of the wood found in the Knife River area was of a coniferous type. This is ' probably due to the fact that the deciduous trees grew on slightly higher and dryer ground. From this position their leaves could be washed into

the basins of accumulation, but the wood of dead'or dyin trees would

quickly rot away. The coniferous trees, on the other hand, were probably swamp dwellers, similar to the bald cypress of today, and when these tree died or were blown over during a storm, many of them were submerged or were buried quickly below the water table where decay was inhibited. cording to Brown (personal communication) many of these trees were pro- j [ bably of the genus Metasequoia, but some of them may have been close re-

natives of the modern bald cypress. At one locality in the Proncho

quadrangle R. B. Colton discovered a fossil stump that had knobs or knees similar to those found on the bald cypress. E It is not possible to say definitely when the wood was silicified E but it was probably not long after the deposition of the sediments. The acid waters underlying the coal swamps of Fort Union time probably had a higher content of dissolved silica than the ground waters of any succeed ing geologic epoch. Also, it seems unlikely that the original cellulose

of the trees could have withstood the pressure of many hundred? of feet of overlying sediments without destroying much of the cell ptructuro. Carbonized log? and stumps that comprise parts of the coal bods have been crushed and flattened, but most of the- silicified wood in the Fort Union formation rhov.rs excellent preservation of the cell structure.". --Silicified whale and sandstone-- Besides replacing the wood, silica has invaded, cemented, and partly or completely replaced cthrr local bds in the Tongue ivar member. In general, threo' types of silicifid beds can be differentiated, but they are probably part of a gradational series and there is no rh.aro

/ dividing line between them. The most striking type of silicified bd is probably the silicified sandstone or fine-grained quartzite. Thane beds consist of very fine-drained quartz in a matrix of microcrystall ino silica. The whole rock is gray and homogeneous in appearance; it is very hard and breaks into sharp angular fragments, when hit with a hamner, Local bed? of this silicified sarsd rtcne occur in many parts of North Dakota and at various strati graphic horizons in all the. formations of continental origin. Plocks of this silicified sandstone or ouartzito are extremely resistant to both chemical and mechanical weathering and persist on the surface long after the weaker strata above and below them have been eroded away. In Sioux County, southern North Dakota, I have seen large blocks of this sandstone more than five miles from the outcrop of their parent bed, which here is in the Hell Creek formnMon. These blocks do not appear to have been transported laterally but rather seem to have been let down vertically more than 200 fee.t from their

original position The second type of silicified bed is silioified shale. The silioified shale is similar to the sandstone except that it tends to be yellower, splits along definite bedding planes, and shows few or no quartz grains under the hand lens or under the microscope. Prior to silioification these beds were apparently gray shale. Commonly, the silicified shale contains many plant impressions, nearly all of which appear to have been rushes or other 1 swamp dwelling plants. Numerous cylindrical holes pass completely through blocks of the silicified

shale and appear to record the position of former roots or stems that were not silicified and that have been destroyed by weathering. Boulders of silicified sandstone and shale are not ar common in the Knife River area as they are in some other parts of western "North Dakota. A group of hills in the southeastern part of the Medicine Butte quadrangle and extending south beyond the boundary of this quadrangle is capped by residual boulders of silioified sandstone and shale, but this is the only large concentration known in the area. These residual blocks are probably let down from a silioified bed in the Golden Valley formation. The third type'of silicified rock is actually a variety of silicified shale,* but the original bed, instead of being a gray shale, was a black carbonaceous bed probably a lignitio shale around the margin of a coal ewamp. This is the most common type of silicified rock in the Knife River area and occurs in at leapt 6 different ptratigraphic horizons within the area. These silicified carbonaceous shales range in thickness from about inch to 8 inches and are very thin bedded and

platy. Like the fossil wood these beds bleach white on weathered outcrops but the fresh material is dark gray to black and appears to contain a great deal of residual carbonaceous material. Locally, silicifiec stumps and log? rise out of these old soil zones. Apparently the silici fied black shales grade laterally into silicified gray shale, silicifiec sandstone, or into a dark, dense chert. Residual concentrations of thif chert were formerly dug by the Sioux Indians in Sees. 1 and 15, T. 142N, 2. 90W. Hennen's "marker sandstone 21" (1943, pp.1569-1570), which he believed to be a bed of volcanic ash that persisted over most of western North Dakota, appears to consist of a number of local silicified carf bonaceous shales. These silioified zones are very helpful in detailed local correlations of the strata but there is no reason to suppose that any one of them persists over many miles, or that, once it has pinched out in a given direction, a silicified zone a few miles farther on ia a continuation of the first bed. The silicified sandstones and shales have been described by Tisdal (1941, pp.13-14) and in slightly more detail by Hares (1928, pp.34-36). ' During his mapping of the Heart Butte quadrangle Tisdale did not find any of these silicified rocks overlain by other bedrock strata, and postulated that they might be due to upward moving ground water which '"i is depositing silica at the present day surface. This seems to me a rather unlikely explanation. First, there is little other evidence to show that the present ground waters are carrying much silica in solutio or that they are depositing it at any particular horizon. Second, although these silicified beds apparently do not crop out in the Heart '

Butte quadrangle they may be seen in place, overlain by other Tertiary strata, in many parts of western North Dakota, and they have been penetrated by some water wells. Hares cites numerous ocourrenoes of these silicified beds in North and South Dakota, and in many of these place? the silioified beds are exposed in plaoe on a hillside. Where they have been encountered in wells the silioified beds do not appear to be quite as hard as they are on the surface, and it is ouite possible that Tindale is correct to the extent that the beds are case-hardened

by surface weathering. The siliceous cement, however, appears to have been introduced into the bed long before its exposure from the present surface. r It is impossible to say when these beds were silioified. It certainly happened before the beds were carved into their present topo-

graphy and-before the roots and stems of the swamp plants had decayed and rotted away. My own opinion is that the silicification probably took plaoe during the early Tertiary shortly after the deposition of the sediments. Measured sections The following partial sections of the Tongue Kiver member of the Fort Union formation are typical of its lithology and appearance in the Knife River area: Section of the Tongue River member of the Fort Union formation measured in the east half of sec. 14, T. 144N., K. 84Vf. The base of the section is the bottom of the bluff of the Missouri Kiver Valley near the Lewis-Clark monument. Top of section is capped by residual of clinker of the Stanton coal bed. On adjacent hills this clinker

crops out about 15 feet higher stratigraphioally than the top of this seotion. Foot 1. Clay, silty, light gray, soft. 2. Lignite, black, impure 3. Clay, dark gray, silty 4. Unexpoaed

5. Shale, gray, sandy, interbedded with sand, light gray with clay binder. 8.0 approx, 6. Silt, gray, weathers light yellow 2,5 7. Lignite and carbonaceous clay 8. Clay, dark gray, plastic 9. Clay, yellow-brown, soft, silty ,2.8 10. Clay, brown, carbonaceous 11. Shale, light gray, silty 12. Shale, gray-brown, carbonaceous 13. Clay, shaly, yellow-gray, with iron oxide seams 9.2 14. Lignite 15. Clay, light gray 16. Clay, shaly, light gray to brown, carbonaceous with many plant remains 17. Shale, light gray with yellow silt bands 18. Lignite, impure 19. Shale, gray 20. Unexposed 21. Limestone and sandstone concretionary horizon 2?. Sand, clayey, weathers hard with drab yellow-gray color 3.0 approx.

23. Shale, brown carbonaceous 24. Clay, gray, plastic, non-silty 25. Shale, black 26. Sand, and andy clay, drab gray with iron oxide seams

2B. Clay, brown carbonaceous 29. Shale, light gray " 30. Limestone concretion 31. Clay, shaly, drab yellow- gray 32. Lignito 33. Clay, medium gray, carbonaceous f 34. Sand, hard, clayey, with concretionary zones, stainnd yollow by iron oxido 35. Limestone concretions 36. Shale, drab yellow-gray, fopiliferous near thn base 37. Lignite, impure 36. Shale, gray* silty, slightly sandy 39. Poorly exposed unit, mostly yollow silt and sandy phale 40. Sandtono, fine grained, ripple bddod anrl cross bedded, liHt fray. Harder ledgoforming beds at the ton. 41. Shale, carbonaceous, pandy, with frosh water invertebrate fossils including CampoloTna and ViviparuR 27. Shale, drab yellow- gray with iron oxide scams 10. annrox. 20. anprox. 42 Unexnosod t Bottom of section at floodplain of Missouri River, altitude about 1680 i : Total thickness

Section of Fort Union formation measured in thr- northeast quarter soo. 1, T. 143?!,, R Saw. 1. Till C ?.. Lignite eulah Zao bed Feet 3. approx, 6.4 ) ( 3. Shale, gray, wall bedded with nilt and ) ( sand laminae

2.8 ) ( t ) f 4. Sand, fine-grained, micaceous vlth car- ) ( bonaceous laminae 3.1 ) ( 5. Lignite 3.0 ) 6. Ilnexnosed 7. Shale, brown, carbonaceous with abundant foppil wood 2s 8. Clay, shaly, gray, soft 4 / 9. Sand, fine-grained, soft, gray, with silt laminae 10. Clay, shaly, gray, with laminae of silt, Local tone of concretions about 5 feet above base 11. Clay, dark gray plastic 1 / 12. Lignite 13. Shale, brown carbonaceous 14. Clay, dark greenish-gray, very plastic, swells on exposure, may be bentonitic 15. Shale, light brown to gray, carbonaceous, many fossil leaves including Osmunda 16. Lignite and lignitic clay 17. Clay, dark gray, very soft 1. Shale, gray to light brown, carbonaceous abundant leaf fragments 19. Lignite 20. Clav, carbonaceous TVulah- £ap bed

21. Shale, silty, intarbedded with sand, fine-drained, Laminated; all stained light yellow by limonite 22. Lignite - the Spaor bed 2,6 23. Shale, light brown, carbonaceous 24. Unexposed 25. Clay, shaly, hard, greeni sh-p.ray with limonito concretions 26. Claystone, light gray 27. Clay, gray, silty 28. Sand, gray-brown, clayey 2 / Base of exposure Section of the Tongue River member of the Fort Union formation measured in the northwest quarter of Sec. 5', T. 142H., R. 89Vf., and the southwestern corner of Sec. 30, T. 143N., R. 89W. Top concealed Feet 1. Sandrtone, light gray, fine-grained 2. Clay, shaly, gray 3. Sand, gray, clayey 4. Clay, gray, silty and sandy 5. Shale, brown, carbonaceous 6. Shale, dark gray, silty 7. Sand, and sandstone, clayey, light yellow-gray 8. Clay, shaly, gray to brown 9. Lignite and carbonaceous clay, -Schoolhouse bed 10. Shale, brown carbonaceous 11. Unexpo?ed 13. Sand, gray, fine-grained

13. Clay, dark gray, plastic 14. Clay, dark gray, carbonaceous 15. Silt, soft pray with sand laminae 16. Sand , and sandstone, fine -grained , thin bedded, clayey, with calcareous concretions 17. Sand, dark brovm carbonaceous IB. Clay, gray, sandy 19. Clay, dark gray, eilty 20. Shale, brown carbonaceous with lignitic streaks 21. Clay, '"sandy, silty, pray 22. Sand and sandstone 23. Clays tone, gray, silty to sandy, very hard 24. Clay, blaok, lignitic 25. Clayey sand to sandy clay, gray 26. Shale, brown, carbonaceous, with many fossil 27. Clay, gray 20. Sand, clayey, gray 29. Clay, silty to sandy 30. Shale, brown carbonaceous

31. Sand gray, clayey, interbedded with silty clay 32. Unexposed 37. Clinker of Beulah-Zap coal bed 8 approx. 34. Unexposed 35. Shale, gray to brown 36. Shale, dark brown, carbonaceous 37. Lignite, impure

38. Clay, gTay brown carbonaceous Section of the upper part of the Tongue River member of tho Fort I'nion formation measured in the northern part of Sec. 1, T. 142N., X. 91V,. Top-Golden Valley Formation 1 . Unexposod; contact botv/een Golden Vn] ley and Fort Union forraations in this interval 6 /- 2. Sane, liht pray, soft, with local calcareous concretions 7. Sand, very f ine- grained , clayey, light f.ray to buff ' ' 4. Lignite, with interboddod bro\vn carbonaceous shale t 5. Sand . dark gray 6. Clay, silty, dark gray 7. Clay, black li.-;nitic 8. Clay, shaly, dark brown to £ray , carbonaceous 9. Clay, dark £ray 10. Unexooped 11. Sandstone, medium to light yellow gray, calcarous, with limestone concretion lenses 12. Clay, gray plastic with limonite concretions at tho base 13. Clay, medium gray, soft 14. Clay, lifht gray, stained white on the surface 15. Clay, medium to dark f.rny, silty, very nlastio. 16. Unexposod 17. Clay, gray, shaly

IS. Shalo, brown carbonaceous 13. Lignite, - Twin Butter bed 20 Clay, brown carbonaceous Thickness No one exDosure or area shows tho complete section of the River member. If the beds havo been correlated correctly between many partial suctions measured in different parts of the area, a total of aboub 550 to 575 feet of the mombor is exoosod. Inasmuch as thr Cannon-

ball marine formation crops out on the Missouri River iust to the east of the Stanton quadrangle it would arroear at first glance that this fi£- : . ure represents nearly tho total thickness of the Ton.ue River member in this part of North Dakota, certainly the top of the Cannonball cannot

be very far below the base of the section measured in the Stanton quad- I rangle. However, the core of a hole drilled at the Oarrison Dam site £a-rris:*n indicates that the >lijwipr part of the Tongue River member is 175 to 200

feet thicker than it is in the southeastern part of tho Stanton quad- t. rankle. The thickening of the Tongue River strata in this direction t i ' may be duo either to marine of flap or to the sinking of a local basin j during "deposition. At any rate, this extra 175 feet plus tho 575 feet

indicated by exposures in the Knife River area give a total thickness : of about 750 feet for the Tongue River in the northern part of the .E Stanton quadranglo. Data from deep wo] Is drilled in other part of the E Knife River area are unreliable but appear to indicate that the Tongue River member is at least 750 or 800 feet thick over most of the area. Relation to adjacent formations : In central Korth Dakota the Tongue iHver member of the Fort TTnion !

formation overlaps and is essentially conformable with the Cannonball forrcation, as can be seen in exposures a few miles southeast of the Stanton quadrangle. The upper contact with the Golden Valley formation appears to be conformable in all but a few exposures, where the basal sand of the Golden Valley fills in small channels in the top of the Tongue River. Fossils Leaf impressions are common or even abundant in many of the shale of the Fort Union formation and are found sparingly in some of the san stone beds and in some of the limestone concretions. A single floral assemblage is persistent throughout the formation in the Knife area, t and is, according to Brown, a typical Tongue River flora. Brown has identified the following types from specimens collected during this surveyj Coniferous trees: Glyptostrobus sp. Thuja sp. Metasequoia sp. Deciduous trees: Cercidiphyllum arcticun (Heer) Brown Aralia notata Leuereux Viburnum antiquum (Dewberry) Holliok SapindTTs" grandifolius Ward , tJlmus sp. Querous penhaloe PlataniTs sp. Ferns: Osmunda sp. Onoclea sp.

Dryopteris sp. Water plants: Paranymphea Eouisetum ' Al DO

Freshwater gastropods and peleoypods are found locally throughout the section. In a very few local beds these shells are so abundant that form small ledges of ooquina. The following types, typical of tl Fort Union formation, have been tentatively identified from collectior made in the Knife Rivor area: Goniobasis nebrasoensis Meek and Hayden Campeloma sp. , Unio sp Corula sp. Viviparus trochiformis Meek and Haydn Viviparus sp. , Although many of the badland exposures were searched carefully, ; remains of fossil mammals were found in the Knife River area. The on" bones noted were a few vertebrae of the reptile Champsosaurus and son pieces of turtle shell. Interpretation The transitional contact between the Tongue River member of tho Fort Union formation and the underlying Cannonball marine formation suggests that the Cannonball sea became shallow and was gradually re-' placed by a low-lying coastal-plain swamp. The Fort Union formation ; is often spoien of loosely as being of fluvial origin, but this implie deposition by aggrading streams on wide flood plains an environment j that vould not be likely to produce a formation like the Fort Union. The lon,, horizontal bands of larainated shale and the even thin-bedde" sandstones of the Fort TTnion record deposition in quiet waters or ; water? that were flowing only gently. Tho extensive lignite bds she that great swamps persisted for long periods of time. The fossil woe appears to represent trees that had much the same habitat as the bale cypress of the Gulf Coast today and the only common vertebrate fossi]

found in the Fort Union of North Dakota are Champ s osaurus t a crocodilelike reptile, and one or two species of turtle. Those various facts indicate that the Fort Union fonration was laid down as a coastal-plain deposit marginal to the Cnnnonball sea. Much of the country probably resembled the mouth of the Amazon Pviver in Prax'il. The great development of the lignite beds and the prub-tropical floral assemblage ini dicate that the climate was moist and very warm. EOCENE SERIES

Golden Valley formation The Golden Valley formtion here receives its first detailed description. Because all previous references to the beds that comprise r this now formation have been brief, tho following descriptions and

measured sections will not be confined to the Knife River area, but will embrace tho outcrop area of the Golden Valley formation in south-

' wester North Dakota. Name and Definition The Golden Valloy formation is defined as the strata of Eocene age in western North Dakota that overlie the Tongue River and Sentinel Butte shale members of the Paleocene Fort Union formation and that arc unconformably overlain by the Oligocene Vfhite River formation. The nar "c was chosen because of the- excellent exposures of the formation in the vicinity of the town of Golden Valley in Mercer County. No one sectio; shows the entire formation, but exposures in sees. 32 and 33, T. 144N. R. 90V*. and in sees. 2 and 5, T. 143N., K. 90\V. south and southwest of Golden Valley show tho character of both the upper and lower members o" the fonrjxtion. The choice of this aroa as a type locality can be critJ"

cized on the ground? that the basal contact is not well cxpored, and, a? the formation caps the hills, the upper limit is not defined. Hov.-- ever, the thickest and most extensive remnants of the formation are pre* served in this area. In other areas where White River beds ovorlie the Golden Valley fonr/ition, p re-White River erosion har removed mort or al/ of the thick upper member of the Golden Valley formation. In tho arras where the basal contact, is better exposed, the remnantr of the formation are thin and patchy.

The Golden Valley formation as here defined includes most of Leonard's "light colored Tertiary clays" (1906, p.88) and most of Seaee "unnamed member" of the V.'asatch (1942, pp. 1414-1422). f Distribution j All tho known outcrops of the Golden Valley formation are in North Dakota and it seems probable that the formation ir confined to this ; State. To the south, east, and west, older strata rise around the mar-, gins of the V'illisten Basin and any equivalent of the Golden Valley forrmation has been stripped by erosion. Remnants of the Golden Valley formation are preserved on the interstream divides and in small synclinal basins over a large part of soutlwestern North Dakota. The distribution of these remnants can be seen

on the geologic map pi. 2. rost of the areas were mapt>ed during tho study of the Golden Vnlley formation in the runners of 1947 through E The outcrops in Oliver County were maopod by V. D. Johnson, Jr. of the Fuels Branch of the Geological Purvey, and some of tho outcroos in the Fort Berthold Indian Reservation in Northern Dunn County were

corralled from recent mapping by Ellis Gordon and Robert Dingman of the Y/ater Resources Division of the Geological Survey. The largest outorop areas of the Golden Valley formation are in tho eastern part of Dunn and western part of Mercer counties, in northwest Morton County, and in central and northeast Stark County. Many small isolated outcrops are scattered over these counties and also in McKenzie, Billings, Hettinger, Slope and Oliver counties. The southwestern limit of known outcrops of the formation is in the Little Badlands area in Stark and Hettinger counties. The easternmost outcrop is in southern Oliver County. In the Knife River area the Golden Valley formation underlies / large parts of the Broncho and Golden Valley quadrangles and a few small tracts in the Medicine, Butte and Beulah quadrangles. It has been removed by erosion from the Hazen and Stanton quadrangles. Composition The Golden Valley formation is easily divisible into two members, the lower of which is so distinctive in lithology lateral persistence, and prominence of outcrop-that it can be used as a marker bed over the entire outcrop area of the formation. --Lower member-- The lower member of the Golden Valley formation consists chiefly of Kaolinitic clays and shales Ab to 35 feet thick. The shales aro light to dark gray and weather very light gray to very light purplish

gray. They are typically eilty and locally contain fine sand, but in some exposures they are very plastic and nearly free of coarse material.

Locally they contain numerous fossil plants. The clays are very light gray to white, tough, and are slightly sandy to very sandy. Certain parts of the clay bed? are mottled and stained a bright yel] ow-orano by iron oxide, a brighter staining than is present in any other beds either in the Golden Valley formation or in the upper part of the Fort Union formation. The staining seems to come from flmall pellets of iron oxide that are in turn relicts of siderite pellet?. Thin layer? of iron OX ' oxide,-crystals of gypsum are common in both the shales and clay? (pi.14. /i In gross aspect most outcrops of this member consist of throe units: (l) a basal unit of light purplish-gray shale; (2) a middle of tough white sandy clay, the middle or top part of which is stained yellow; and (3) an upper unit of light gray to purplish shale similar to the basal unit. This sequence is, of course, not invariable but it

is typical of at least 2/3 of the outcrops of the member (pi.9). ; Locally a thin, discontinuous unit of brown, micaceous, carbormceo** /

;' shale and fine-grained sand underlies the "marker-bed"/Kaol ins and api pears to belong -with the Golden Valley formation rather than with the ' underlying Fort Union formation. This unit is well developed in a sec- " ' tion taken about three miles northwest of Hebron in sec. 23, T 140N., R 91V/., where the Golden Valley formation caps a few isolated buttes.

Northeast of this locality the "marker-bed" lies directly on the

Earnisoh lignite, and, in the pit of the Hebron Prick Company, a sandy: phase of the "marker-bed" channels into this lignite. This channel

contact is taken to be the Pal eocene-Eocene boundary. In the locality just cited, however, about 14 feet of silty, micaceous sand and rhalo lie between the Harnisch lignite and the faxolin clays, and I have tentatively placed this unit in the Golden Valley formation.

The description of the "unnamed member" by Seaper and others (194< p.1416) is evidently intended to apply to the marker bed shales and sai clays. They state: "Large, bright yellow, calcareous sandstone concretion? not exceed-3 ing six feet in diameter, imbedded in an ashy white matrix are characteristic of this formation. The concretionary zone is overlain by a bed that apoears to contain a volcanic ash, ana it in turn is overlain by more than 15 feet of yellow bentonitic clay. In some localities hard plish gray bench forming sandstone lie? several feet above the bentonire.. layer." The "yellow bentonitic clay" and "purplish sandstone" are probably the yellow stained clay and a sandy facies of the overlying carbonaceous shales of the lower member. 1 have examined carefully many outcrops the Golden Valley formation end found no b°d- that am bentonitic that contain glass shards. f Several types of laboratory tests indicate that the light colored ' clays and shales of the lower member consist chiefly of kaolin clay mixed with varying amounts of silica, either in the form of opal or a discrete quartz grains. (l) The many chemical analyses and ceramic t ' made by Leonard, Clapp and Babcock (1906 pp. 132-191) show that these clays and shales are typical kaolin type vdth an admixture of silica.. I ". (2) Thermal dehydration tests made by Mips E. C. Fisher of the r-eo-

logical Survey indicate that most of the clay is kaolinite with some. ! small percentages of halloysite or endellite. The thermal dehydratiovvcurves show also that there is a large amount of material that is ine and cannot be identified by this. (3) Thin sections of the white sc

; clays "show that 'most of the clay has a low birefringence typical of kaolinite. Thin p-hr injure of a mineral with rli-htly higher : birefringence may be an iron montmorillinite. These stringers occur

along small fractures and were probably formed during some post-Eoceno period of weathering. Lignites up to two or three feet thick are present locally but have no lateral persistance except for one thin carbonaceous zone at the top of the member. This carbonaceous or lignitic zone is only a few inches thick in most localities but is Dresent in nearly all of th outrot)* that were examined. In places it thickens to as much as 6 foot but in general it is very lenticular and is not thought to contain valuable reserves of coal. In the Knife River area this bed has been mapped as tho Alamo Bluff Bed wherever it is more than two feet thick. Local variations of "marker bed" kaolin clays / Although in general the'lower member of the Golden Valley forrr.tior is much the same over its whole outcrop area, it does show three types of facies change, all local, but all recurring in enough places to brake them noteworthy. These facies changes are: (1) Locally the entire sequence of kaolpnitio clays and shales grades rapidly into a white kaolpnitio sand. A few samples of this sand were examined under binocular and petrographic micro scones and consist chiefly of the following minerals: Angular to subangular ouartz, 60 to 65; angular oaloite, about 30 fresh to slip-htly weathrd fold-

spar, 2 to 3$; heavy minerals, 2 to 3/£. The heavy minerals were not studied in detail but appear to be similar to those found in the Fort Union sands. Garnet, tourmaline and kyanite were particularly common in the samples examined under the petrographic microscope. r (2) Another variation or facies change in the lower member is tho

local pinching out of tho white sandy olays so that in some outcroor

the entire member consists of very light gray to light purplish-gray shales. In most places this variation is extremely local and outcrops within a quarter of a mile of the abarrent section show the typical three-fold division. (3) In Stark County, south of Diokinson and also between Gladstone and Taylor, the "marker bed" shows a third type of lateral variation. In these areas the yellow staining of the white clays is scarcely noticeable on outcrop; the whole member is white to very light gray and appears to contain substantial amounts of free silica, especially near the top. Locally this free silica has indurated both shale and sandstone to form hard silicified beds like those described in the Fort Union formation. Siliceous blocks from the Golden Valley formation litter the slopes of Davis Butte, a few miles North of Diokinson.

"Marker bed" in Fort Union formation I know of only one bed in North Dakota that can be mistaken for the lower member of the Golden Valley formation. This bed which ia just above the base of the Tongue River member and 50 to 75 foot belo* the Harmon lignite, drops out in southwestern North Dakota in Bowman and Slope counties and is almost identical in appearance to the lowei member of the Golden Valley formation. It ie well exposed in the Meo' oine Pole Hills southwest of Bowman and oan be traced north from thig locality to a point about 13 miles east of Amidon. Over much of thig distance the bed has a silicified zone at the top and the yollow stai ing is les? pronounced than in the "marker-bed" of the Golden Vrxlley formation. One of the best exposures of this bed in in the valley of

Deep Creek in the southeast corner of sec. 30, T. 135V., R. 102'V. in Slope County, Although it was not described by Hares it would sem to be an excellent local marker horizon for stratigraphic work in the Marmarth area. A light gray/colin bed in about the same strati graphic // oosition crop? out at the hare Anarchist Rutte in Karding Counry northwest South Dakota. --Upper member- - The upper member of the Golden Valley formation is as a rule poorly

exposed. It consists largely of coarse to fine-grained, yellow to gray micaceous sand and silt with minor amounts of gray shale and silty clay. The maximum thickness observed is about 150 feet and is in the / type area south of the town of Golden Valley. The total thicknes-s of the formation in this area is about 180 feet and this is the thickest section observed anywhere in the state. In most other outcrop areas erosion has spared less than 100 feet of the Golden Valley formation. Shales, silts, and clays comprise about 25 to 30/£ of the member and are most abundant in the lower part. Some of these beds, especially the silts, have a greenish cast not found in the Fort Union beds, others of th*n, especially the dark gray and brown carbonaceous shales, are identical in appearance to the Fort Union shales. Beds of fine-tocoarso-grained micaceous sand and calcareous sandstone comprise 70 to 75% of the member as a whole and are thickest and most persistent in the upper part. Individual beds are as little as 6 inches and as much as 50 feet thick. In general tho sands and sandstones of the Golden Valley formation resemble those of the Fort Union formation except for the following features: (l) the beds of the Golden Valley formation are more highly cross-bedded,

exhibiting both cut-and-fil1-type and deltaic-type cross-bedding. (2) The coarre-grained bedp are locally conglomeratic, the pebbles consisting mostly of clayball? and reworked fossil wood, with a few wellrounded fraents of quartz and chert. As noted ai<ove, the only "pebbles" in the Fort Union formation are a few clay balls. (2) The Golden Valley sands tend to be better sorted and have beds with large percentages of clay, (4) In the Knife River area, at least, there is much more mica in the sand? of the Golden Valley formation than in the Fort Union sands. In the extreme western part of North Dakota the sands of the Tongue River member of the Fort Union formation becomes more micaceous, and there is less distinction between them and the Golden Valley formation. Coarse cross-bedded brown sandstone with mica flakes up to 1/8 inch in diameter locally comprepos most of the upper member of tho Golden Valley fonnation. Medicine Butte in the Knife River aroa is capped by about 80 feet of this sandstone (see pl.lOB ). Other prominent buttes capped by this sandstone are the Blue Buttes in McrCenzio County which were formerly thought to be capped by the White River fonnation (Leonard, 1922, p.227 and Nevin, 1946, p.4). Black Butte, in Hettinger County, is composed partly of this sandstone, but is probably oapped by White River beds. Lignite beds are locally orerent, but do not have tho thickness or lateral persistance of the Fort Union coals. The Shaffner bed in the Broncho quadrangle is in a local shaly facies of the upper member of tho Golden Valley formation. --Oxidation and bleaching of upper member --

Around the margin of the Little Badlands syncline in wertern .Stark County the upper member change? its appearance. The outcrops around the south and east sides of this basin resemble outcrops of the upper member elsewhdYe in the State, consisting mostly of yollow to £ray micaceous sands and silts intorbedded with thin carbonaceous shales. Around the northern and western edos of the syncline, however, the upper member has boon bleached and oxidized and appears in many outcrops as white to lijr,ht p;reen and pinkish pray micaceous sands, some of which are stained bright yellow by iron oxide. These colors aro similar to the colors of the marker bed in the lower member, but the lithology is typical of the upper mombfer. The gradation between oxidized and unoxidized phases of the upper member can be seen in the northwest quarter of sec. 14, T. 138N., R. 98W. Hre, the "marker bed" crops out on the east bank of a small north-flowinp stream, and dios southward into the syncline at an anle of about 5 depress. Overlying the "marker bed" are brown carbonaceous shales and yellowgray fine-drained sand and sandstone beds. Followed to the south the "marker-bed" dips below stream level and the color of the unner beds gradually changes to light greenish pray and 'while; with yellow mottlings. ,The oxidized pha?e of the upper member of the Oold en Valley formation is well exposed also in sec. 15, T. 138N., R. 98Vrf., and at White Butte in tho northwest corner of sec. 32, T. 139N., R. 97Vi. At both localities the Hold en Valley formation is overlain by white conglomeratic sandstone of White River formation, and at Whito tto this

basal White River sandstone channels 15 to 20 feet into the Golden Valley formation (see pi, 13 ). Diagnostic Eocene fossils were found Just below this unconformity at White Butte. This oxidation of the Golden Valley formation evidently records period of deep weathering that preceded deposition of the tohite Kivor formation. Oxidation of other formations Southwest of the Little Badlands in Slope County, the White River formation in H. T. Butte and the Chalky Buttes is underlain by 30 to 50 feet of white, yellow, and pink sands interbedded with variegated green, pink and yellow shale?. Although these beds greatly resemble the / oxidised upper member of the Golden Valley formation in Stark County, they are not underlain by the distinctive Kaolin clay of the marker-bed member and all the fossils I found in them are typical Fort Union species. It is my opinion that these beds are probably Fort Union strata, oxidized in pro-White River time, just as were the strata of the Golden Valley formation in the Little Badlands area. This opinion is further supported by tracing the White River formation south toward the Black Hills. Evidently the Black Hills area was domed in the post-Golden Valley pre- White River interval, and the land was deeply eroded and stripped. The amount of this erosion increased toward the center of the uplift, and the White River formation lies on progressively older beds as one goes south toward the Black Hills. Regardless of .their age, the bed? imimmediately underlying the "White River in northwestern South Dakota have in many places been bleached and oxidized to various shades of green, pink, yellow and light gray.

--Measured Sections-- The following detailed sections show the typical lithology and general appearance of the Golden Valley formation in south.-estern North Dakota. Because this is a new formation that has not heretofore been described in detail, I have included several sections measured outside the Knife ttiver area. Section of the Golden Valley formation measured ME 4-, sec. 2, T. 143N., R. 90W., Broncho quadrangle, Mercer County. This section' is in the type area of the formation. Upper member Feet 1. Sandstone, gray to brown, fine-to coarse-grained micaceous, conglomeratic 64 approx. 2. Sand, gray to buff, weathering yellow-brown, fineto coarse-grained with conglomeratic zones containing claybalis and fossil wood; highly crossbedded; similar to unit 1 but not cemented 3. Clay, dark gray, toward top becomes silty and contains thin beds of fine sand; large gypsum crystals at the top of this unit 4. Lignite, impure, interbedded with dark brown lignitio shale 5. Clay, brownish-gray to dark gray interbedded with sand, very fine-grained, micaceous, contains gypsum crystals

6. Sand, fine-grained, micaceous, carbonaceous 7. Clay, dark to medium gray with iron oxide seams 8. Unexposed 9. Sand, very fine-grained, micaceous, poorly exposed, partly cemented to calcareous sandstone Lower member 10. Shale, light to dark gray with brownish gray carbonaceous layers and a few thin stringers of lignitic olay, yellow rones stained by iron oxide 6,3

11. Clay, marsive, tough, light zray to whito, mottled and stained yrl low-orange on weathered surface, small limonite pellets, slight sandy in lower two feet 12. Shale, light to medium fray, slir.htly publish, carbonaceous, numerous plant remain?, pome yellow of iron oxide Bare not exnoced Section of the Golden Valley formation measured ME,, sec. 8, T. 144N., ri. 90W., Golden Valley quadrangle, I/order County. Upper member Feet band, fine -grained micaceous, extensively ripplebedded and cemented into concretionary sandstone, minor amounts of interbodded gray- green shale Shale, greenish- gray with some minor sandstone / Lower member 2, Lignitic clay 0.? Clay, light gray, slightly carbonaceous , shaly, with minor amounts of iron staining Clay, tough, white, sandy, mottled bright yellow by iron oxido, numerous iron oxide concretions Shale, medium gray, fairly well bedded, numerous plant fossils including Metaseouoia, Platanus , and Sapindujs Base of formation concealed Section measured in sec. 1, T. 142N., H. 91V*. , Broncho Quadranrl, Dunn County. The *lower part of the section was measured along the center of the north lino. The upper part was measured in the southwest quarter of the northwest quarxer of the section Goldon Vul loy formation Upper member Sandy, fine-trained , micaceous, pray to buff, oapped by calcaroqus pand stone

-Shale, alternating brown anc gray, carbonaceous Lignite, impure Clay, brown, carbonaceous £haie, light gray ' Lignite, the Shaffner bed Clay, brown, carbonaceous 1*0 Unexposed Sand, fine-grained, soft, micaceous 10. Sandstone, fine-grained, ripple-bedded, calcareous 11. Sand, fine-grained, micaceous Lower member 12. Clay, black, lignitic. Alamo Rluff lignite zone 13. Clay, light gray to white, blocky, silty 14. Shale, light brown, carbonaceous with fossil plants weather? light purplish ' 15. Clay, black, lignitic 0,5 1C. Sand, clayey, kaolinitic, interbedded with sand, dark brown, carbonaceous.weathers liht pinkish gray 17. Clay, brovm, carbonaceous 18. Unexposed, contact between Golden Valley and Fort Union formations somewhere in this interval 6. approx.

Fort Union formation,* Tongue River-Sentinel Butte member 19. Sand, light gray, crossbedded with local sandstone ooncretions 20. Sand, very fine-grained, clayey, interbedded with soft gray silt 21. Shale, dark brown, carbonaceous with thin beds of lignite 2,9 22. Sand, fine-grained, dark gray 1,0 23. Clay, dark gray, silty Base of exoosure

Section of the Golden Valley foriration measured SE*-, SE- sec. 10, T. 142N'., K. 88Yii., Medicine Butte Quadrangle, Mercer County. Till Golden Valloy formation 1'pper rrmbcr Sandstone, very fine-grained, silty, micaceous, ripple-bedded Shale, medium gray, silty, weather? yellow-buff Lower member Lignite, soft to hard - Alamo Bluff bed Shale, lif.ht to medium gray, slightly silty, upper part weathers light purplish-fray, large gypsum cryr.tals , Clay, shaly, brownish-gray, weather? li~ht gray, heavy deposits of iron oxide along joints and bedding pianos Clay, white, slightly sandy, extensive yelloworange mottling, iron oxide in concretions and along joints 5,7 fc. Shalo, medium gray to dark brown, carbonaceous, ' Bilty, weathers purplish-gray Section measured SE, SE- Sec. 23, T.140N., R. 91W., northeastern Stark Crunty. Golden Valley formation Upper member Sandstone, fine-grained, micaceous 3,9 Silt, clayey, laminated, micaceous, yellowish-gray to brown Clay anc shale, gray to reddish brown, carbonaceous, interbedded with impure lignite

Sane, fine-drained, very micaceous, with weathered iron oxide concretions Clay, silty grading upward into silt, light to medium gray Lower member Shale, dark brov.Ti, carbonaceous Clay, medium gray, weathers dark purplish-gray Clay, light gray to white, poorly bedded, with reddish-yellow mottling by iron oxide Shale, hard, gray, carbonaceous, micaceous 10. Shale, dark brown to black with lignitic laminae 11. Shale, gray, silty with plant fossils Golden Valley formation

Local basal unit 12. Sand, very fine-grained, micaceous, crossbedded with minor amounts of interbedqed silty, carbonaceous clay, grades laterally into carbonaceous shale Fort Union formation, Tongue River member 13. Lignite, with stringers of brown carbonaceous shale - The Harnisch bed 14. Shale, dark gray to brown, carbonaceous 15. Lignite 16. Shale, gray, carbonaceous with plant fossils 17. Lignite 18. Shalo, carbonaceous ' 19. Shale, dark gray to brown, carbonaceous, sandy at the top , Section measure SE-J, r>E-};- sec. 3, T. 140N., tt. 90VV. ,

Golden Valley formation Upper member Feet Sand, light gray to buff, fine-grained, micaceous Sandstone, crossbedded, light gray, calcareous 3. approx. Sand, light gray to buff, fine-grained, silty croflsbedded, numerous concretions 7. approx. Lower member Clay, light gray, plastic Clay, very dark, carbonaceous with lignite fragments, contains abundant leaves including Salvinia, Metaseouoia and Nymp hea Shale, slightly silty, light brownish-gray Shale, medium gray to brown, laminated f Clay, light gray, non-silty, poorly bedded 3,0 Shale, dark brown 10. Clay, light gray to brown 11. Clay, silty to sandy, light gray with abundant leaves including Cercidiphyllum 54 12. Unexposed; contact between Golden Valley and Fort Union.formation} either in or at the base of this unit

5. approx. Fort Union formation, Tongue River-Sentinel Putte member 13. Lignite - the'Harnisch bed >'/I ' 4 /

The top of this' lignite was located by an auer hole base of the section. This section of the lower member represents one of the local variations or facies changes of this bed. The middle unit of sandy kaolin with pronounced yellow staining is absent at this exposure and has been replaced by a light brownish-gray to gray carbonaceous clays. The yellow stained bed is present, however, in the exposures one-half mile to the west.

Section measured at White %tte, four miles southeast of South Heart, NF.Y roc. 32, T. 130V., :?. 97W. , Stark County White ttivor formation Fo'-h Clay, sv/ellin~, probably bcntonitic, pray- brown very light gray, extremely elastic Siltstone and claystone, vrry randy with siliceous cement, thinly bedded, canped by coarse conglomeratic sandstone 3. Silt-tone, hard, clayey, pinkish-white Sand, weakly cemented by calcium carbonate, white, co&rse to very coarre-raineri with franules of quartz and pebbles of chert, quart zite and andesite pornhyry, unper part grades laterally into sandy clay and micaceous clayey sand Channel unconformity f Golden Valley formation, upper member Clay, greenish- gray , silty, micaceous, laminae of very fine-drained micaceous sand, stained yellow by iron oxide 0 to 8. feet Clay, liht pray to white, numerous iron oxide concretions and yellow iron staining, 2 inch bed of carbonaceous clay near the top Shale, silty, micaceous, dark f,ray, grading laterally into yellow-p;reen and brown Base of exposure Section of the Golden Valley formation on Lone Butte , 7 miles northeast of Grassy Butte, SK-7- sec. 33, T. 147N., R. 98W. , McKenzie County. Golden Valley formation Upper member band, lir.ht gray, silry, micaceous ?. Shale, gray-brown, silty Shalo, gray, carbonaceous

Shale, gray-brovm, silty Sand, fine-grained , gray, micaceous "with local large concretions 11,7 6. . Shale, light gray, silty, sandy Shale, dark brown, carbonaceous Clay, shaly, light gray 5,6 Lignite and carbonaceous shale 10 10. Sanu, fino-grainea, yellow to gray, micaceous 11. Shale, brown, carbonaceous 12. Shale, light gray, silty, micaceous 13. Shale, carbonaceous ' 14. Clay, ahaly, light gray to brown, eilty t 15. Shale, brown, carbonaceous 16. Clay, light yellow-gray Lower member 1Y. Shale, light gray to gray-grown, carbonaceous 18. Shale, dark purplish-gray, yellow stained at the base 19. Clay, light gray to white, very sandy, some yellow staining 20. Sand and sandstone, white, kaolinitio, highly orossbedded 21. Clay, dark gray-brown, very hard 22. Shale, light gray 23. Sandstone, silicified with plant impressions 24. Sand and sandstone, clayey, fine-grained, orossbedded 25. Shalo, dark gray to brown, carbonaceous Fort Union formation, Sentinel Butte shale member

26. Lignite, locally clinkered

27. Shale, dark gray

28. Sand and sandstone, silty, thin bedded Several .hundred feet of Fort Union strata are exposed belov/- the base of this unit.

Fossils The only animal fossils collected from the Golden Valley formation consist of one Garpike scale and a few crushed gastropods from the lower member, and a fragment of a small mammalian leg bone from the upper member. Plant remains, however, are common especially in the purplishgray shales of the lower member. The following types have been identified: Lower member: Salvinia preauriculata t Berry Sapindus pyandifolius, Ward Cercidiphyllum arcticum, (Heer) Brown Betula sp. ' Celastrus sp. Gorylus sp. Hicoria sp. Juglans sp. Salpichlaena sp. Equisetum sp. Metasequoia sp. Sequoia (?) Ginkpp sp. Nymphea sp. Upper member: Salvinia preauriculata, Berry Ouercus castaneopsis Lesquereux Cereidiphyllum arcticumj (Heer) Brown

This flora, according; to Prown, is an Eocene assemblage. Plants in general evolve less rapidly than animals. As a result most plant species are fairly long ranging and difficult to use in age determination unless a fairly complete floral assemblage is collected. In the northern Great Plains and Rocky Mountain front region, hov/over, the base of the Eocene (V/asatch) is marked by the sudden aopearance of the floating fern, Salvinia oreauriculata Berry, (peo plate 11 ). The leaves of the Salvinia are small and oval to heart shaped, and are easily recognized by their peculiar pattern of anastomosing veinlets and by a peculiar puckering that fives many ppfcimens the appearance of being punctate. In describing the Kingsbury conglomerate, Provm says this about Salvinia (l948b, pp. 1169-1170): "...at localities north, south, and east of Kingsbury Ridge, where the conglomerate fingers out among fossiliferous carbonaceous shales or sandy clays, the latter contain an Eocene species of floating fern, Salvinia preauriculata Perry I have collected thousands of fossil plant specimens from hundreds of localities in the Paleocene terrane of the Rocky Mountains and Plains, but I have never found a Paleocene Salvinia. In 1941, however, I found a Cretaceous species(plate 1, figure 11) in the "Laramie" formation (Lance) east of Craig, Colorado. The absence of Salvinia from the Paleocene record of the Rocky Mountain region is, therefore, somewhat anomalous. Its simultaneous, re-appearance at about the same stratigraphic levels nearly everywhere in V/yoming, Montana, and North Dakota is unexplained, but, in consequence, Salvinia preaurioulata serves, at least locally and tentatively, as an Eocene index species"." Rrown a]so advises (personal communication) that in most localities in Montana and Wyoming he has been able to find Eocene vertebrates in bed? that contain the Oalvinla or that intrfinrr with the Salvinia-boaring beds. In southwestern North Dakota Salvinia has been positively

identified from 6 outcrops and tentatively identified in a seventh. Two of the outcrops are in the upper member of the Golden Valley fornuxtion; the other five, in the lower member. Because much of the v£> rk on the Golden Valley formation was of reconnaissance type, many outcrop.i were not examined for fossil content and Falvinia will probably be found at many other localities. The location of the outcrops that yielded Savinia are shown on the on the map on plate 2. The localities where Salvinia has been identified a lower member are:(l) in northwest Morton County, SE-4-, sec. 3, and (2) NEj 880. 11, T. 140N., R. 90W.; (3) in Oliver County, NWJ-, sec.27, T. 141N., R. 86W.; and (4) in southwest Mercer County, NE-J- sec. 29, T. 143N., R. 90W. A possible fifth locality is in southern McKenzie County t about 4 miles southeast of the Grassy Butte post office, sec, 29, T. 145X. , H. 981ft. A few fragmentary plants were collected from the markerbed in this area and some of the fragments seem to be Salvinia. All 5 occurrences of Salvinia in the lower member of the Golden Valley formation are in the upper 1/3 of this member. The absence of Salvinia in the white sandy olay and the lower purplish shale of this member may be due to imperfect collecting, or it may be due to a time lag in the migration of the species. The sudden appearance of Salvinia in beds of Wasatch age in the Rocky Mountain region is evidently due to migration, inasmuch as no Paleocene species are known froa this region. If this migration was west to east, Salvinia may not have reached North Dakota until shortly after the beginning of

the Eocene, that is until part of the Golden Valley formation had already been deposited. The occurrences of alvinia in the upper member of the -olden Valley formation are in Stark County, SE-, sec. 4, T. 137N., R. 96V*'., and NW-*- sec. 32, T. 139N., R. 97W. At the latter locality Salvinia is found in a thin bed of black carbonaceous shale that contains also specimens of Quercus castaneopsis, Lesquereux. Quercus castaneopsis is, according to Brown, common in the Green River formation of middle Eocene age. Inasmuch as Salvinia is also found in the middle Eocene, it is possible that the upper member of the Golden Valley formation may be middle Eocene, the time equivalent of the Green River formation, Relation to adjacent formations

The Golden Valley formation is essentially conformable on the Tongue Kiver and Sentinel Butte members of the Fort Union formation. Where the lower member consists of white kaolinitic sand there was some erosion and channeling of the Fort Union beds. Where the lower member consists chiefly of clay and shale the contact appears to be conformable, even gradational (pi. 12A). The upper contact of the Golden Valley formation is not exposed in the Knife River area. It is visible on Antelope Butte? in eastern Stark County and along the northern margin of the Little Pad Lands synoline in western Stark County. In both localities the Whito River beds are disconformable on the Golden Valley formation. At White Butte in sec, 32, T. 139N., R. 97W., conglomerates of the V/hito River formation channel into the oxidized upper member of the Golden

Valley formation wi th a visible disconformity of about 20 feet (pi.13). th .YJV>i-h Rivr formation present, in North Dakot-a t.ho Golden Valley forration is either greatly thinned or is missing altogether. This thinning of the formation is apparently due to pre- Oligocene erosion. Interpretation It is difficult to say exactly under what conditions the Golden Valley formation was deposited. The upper sandy member seems to have been deposited by streams flowing over a low flat plain, possibly a coastal plain. The few lenticular beds of shale and lignite indicate that at times parts of this plain were swampy, but the general scarcity of lignite and greater grain size of the sediments, as compared with the Fort Union formation, indicate that the streams were flovring more rapidly than in Fort Union time and that the margin of the sea was probably farther east. The environment of the lower member is harder The snitf of mnrra]<: in t.he sandy facie* of this member ana the lack of weathering of these minerals indicates that the kaolin was not weathered in situ but was transported from some source area. Crossbedding in both members dips generally to the east, indicating that this source area was to the west. The Jtaolin seem to have been transported as Xaolin and deposited as one continuous or nearly continuous blanket over southwestern North Dakota. This suggests deposition in a broad, extremely shallow freshwater lake. The water must have been shallow so that the streams could build their natural levees and carry sand for pome distance out into the lake. Fivenjo,

it is still difficult to understand how suoh a large blanket of relatively pure kaolin could be brought in by streams without having that kaolin mixed with montmorillonite clays or with large amounts of other materials. YOUNGER FORMATIONS NOT PRESENT IN THE KNIFE RIVER AREA Tertiary System OLIGGCENE SERIES White River formation General Description The White River formation almost certainly extended over the MV-: Knife riiver area at one time, but the/ have long since been removed , by erosion. However, the lithology and distribution of the White J-i 6Hiver formation in southwestern North Dakota is- important in deciphering the late Tertiary and Quaternary history of the State in general and the Knife River area in particular. In South Dakota beds originally called the Yhite River formation are now called the Chadron and TVule formations of the White River group. These beds underlie large areas in western South Dakota and are well exposed in the Padlands National Monument east of the Black Hills. In North Dakota the Oligocene bed? have not yet been divided into separate formations and are known collectively as the VThite 3iver formation rather than the "White River group. The White River formation has largely been eroded from most of the southwestern part of North Dakota and is preserved only where it caps high buttes or underlies some of the smaller synclines in the Mlliston Basin. Present-day knowledge of the distribution and litholofic charac93

teristics of the Vnhite River formation in North Dakota combines the work of many geologist?. Detailed descriptions of the lithology and paleontology of tho formation are given by Douglas (1904), Quirke (1918), and Leonard (1922). Other localities at which the V<Mte Hiver formation crops out were described briefly by Doyd (1914), Hares (1928), Seager (1942), and Powers (1945). During the reconnaissance study of the Golden Valley formation I visited all the previously recorded outcrops of the V/hite Kiver formation in southwestern North Dakota and in addition identified the formation on the tops of four other buttes. Little Badlands and Chalky Buttes areas / The largest remnant of the Vvhite Kiver formation in North Dakota is preserved in the Little Badlands syncline southwest of Dickenson in Stark County. Here the formation underlies about 65 square miles and is about 200 feet thick. It can readily be divided into three members or units. The lowest of these units is probably the equivalent of the Chadron formation in South Dakota; the upper two, the equivalent of the Brule formation. The same three units are recognizable in the Chalky Buttes syncline-in Slope County, where they have a total thickness of about 320 to 350 feet. The lower unit consists of light gray to white fine to coarsegrained sandstone interbedded with light colored calcareous and siliceous olayp. The sandstone beds contain numerous pockets of stringers of well rounded pebble to cobble gravel. This unit is 40 to 60 feet thick in irhe Little Badlands and 80 to 145 feet thick in the Chalky Buttes. In both areas it is overlain by a dark gray swelling clay that resembles bentonite. Thermal dehydration tests on one sample QA

of thin clay collected from the Little Badlands area indicate that it is a hydromica clone to the composition of illite. In both areas the sandstone is weakly cemented by calcite and breaks down readily on the surface of weathered outcrops. The pebbles contained in the conglomeratic beds consist chiefly of well-rounded cherts and brown, purple, and pink quartzites. These quartzites closely resemble the pre-Cambrian bolt series of west central Montana a.nd were probably derived from these rocks. in addition to the quartzites and cherts, which comprise 80 to 95% of the pebbles, the gravels contain numerous well-rounded fragments of brown to red andesite-porphyry similar to the lava flowvS in and around Yellowstone Park. Several pebbles of limestone were picked up in the Little Badlands area and one of these contained a well-preserved Spirifer, suggesting that this pebble was derived from the Edison limestone in the Rooky Mountain region. On the western edge of the Little Badlands syncline I found two well-rounded boulders of pink granite weathering out of the conglomeratic beds. These two boulders, very likely, were derived from the Black Hills. Vi'ood (1904, pp. 116- 117) noted several granite boulders in and around the town of Dickinson in Stark County. He ascribes these boulders to glacial origin, but the proximity of Dickinson to the Little Badlands syncline suggests that they may be residual from the White River formation.

No diagnostic fossils have been collected from the lower unit in either the Chalky Buttes or the Little Badlands, although a few fragments of large teeth are probably titanothere.

The middle unit of the VShite Hiver formation consists predominantly of clay that locally has enough lime content to be classified a? a marl or impure limestone. This unit also contains beds indurated by silica tc form rough ledgy outcrops. In the Little Badlands these beds are predominantly pink and are about 50 feet thick. In the Chalky Euttes the Bamo beds are predominantly preen with minor tinpps of pink and are about 60 feet thick. This middle unit of the White River formation contains an abundant vertebrate fauna, which Douglas identifies a? Middle Oligocene, and which he correlates with the "Oieodon beds" of South Dakota. The upper unit of the White River formation consists of greenish gray shales and clays interbedded with and overlain by green sandstone

that contains pebbles and* angular fragments of marl and silicified clay, / probably derived from the middle White Hiver beds. The upper beds are about 70 feet thick in the Little Badlands and about 125 fet thick in the Chalky Buttes. Locally they contain bone fragments that Douglas identified as rhinoceros. Douglas (1904, p.288) suggests that the "White River formation was derived from the Black Hills area, but the lithologies of the conglomerat' indicate that the bulk of the material probably came from the Hooky Mountains to the west. The eastward dipping foresets of many of the sandstone beds also suggests that the streams depositing the White Rivor beds came from the west. As indicated elsewhere the Black Hills area was differentially uplifted before deposition of the White River sediments and some of the formation was undoubtedly derived from this source area. The amount, however, was probably small compared to the volume of materials of western provenance.

Other oreviously described localities Quirke (1918) measure-dabout 400 feet of White River sediments in the Killdeer Mountains in western Dunn County. Ke divided this sequence of bed? into seven units which he describes a? green crumbly calcareous clays interbedded with marly limestones and green conglomeratic sandstones. Powers (1945) suggest? that Quirke's seven units represent the lower and middle White River beds of the Chalky Buttes area. Capping many of the high buttes in southwestern North Dakota is a thick, coarse-grained sandstone that is locally conglomeratic. This sandstone is very hard and appears to have a siliceous cement. Except for its cementation this sandstone lithologically resembles the softer sandstones and conglomerates in the Little Badlands and the Chalky Buttes. Hares describes this sandstone (1928, p.40) and correctly correlates it with the fihite Hiver formation in the Chalky Buttes. He mentions it as capping H. T., Bullion, Flattop, Sentinel, Camel's Humo, and East and West Rainy Buttes. Lloyd, (1914, p. 11) states that three buttes in southwestern Grant County (now known as Coffin Puttes) are capped by the White River formation.- Leonard (1922) states that the "White River formation also caps Blue Buttes in McKenzie County and occurs as a loose gravel cap on roraelow hills about 9 miles southwest of Bowcan. Nevin (1946, p.4) also mentions the sandstone cap of Blue Buttes in McKenzie County and calls it White River. Seager (1942, pp.1415-1416) and Powers (1945, p.1192) note that

White River beds cap Young Wan 1 s Butte and Antelope Buttea in eastern Stark County, and some low buttes in southern Stark and northern Hettinrer counties near the tov/n of Lefor. I have visited all the above localities and agree with the identification of the Yxhite diver formation in all except Blue Butter. These are capped by the coarse micacaeous sandrtone of the upper member of the Golden Valley formation. Since the preparation of this report. Miller Hansen of the North Dakota Ceoloical Furvry has been investigating thr lithrlory and distribution of the White River formation in North Dakota and advises that it caps Black Butte in Hettinger County. New localities As a by-product of the study of the Golden valley formation, four new Vi'hite rtiver localities were added to the geologic map of North Dakota. These are: (l) The Medicine Pole Hills in the northern part of T. 130N, R. 104W., Bowman County. Haras (1928, p.41) describes the Medicine Pole Hills as being underlain by the fort Union formation and capped by residual deposits derived from the rthite Hiver formation. R. W. Frown and N. M. Denson and I visited this locality in the summer of 1950, and found that the gravel cap of the hills is 15 to 25 feet thick and contains abundant bones that show no evidence of being rounded or eroded by streams, concluded, therefore, that these gravels have not been reworked and are probably the equivalent of the lower White Hiver beds in the Chalky Buttes and Little Badlands. The collection of vertebrate fossils from the Medicine Pole Hills is the best assemblage from the lower Whito River beds in North Dakota. The following types were identified by

C. L. flaee'n; Leptd'rteryx sp., Hyaenodon, Dinictis, Hyracodon, and

Poebrotheriu'm. In addition there are numerous large fragment? of Tl tanothere teeth. According to Gazjrfn, this fauna indicates a lov;er Oligioene age, and these bads are the probable equivalent of the Chadron formation in South Dakota. (2) Whetstone Buttes in northwest Adams County, which are capped by 80 to 100 feot of hard conglomeratic sandstone similar to that of the Sentinel flattop and H. T. Buttes. (3) Wolf Butte, also in Adams County, and also capped by about 65 to 80 feet of sandstone that contains a few pebbles of chert and OTjartzit-9. / (4) Shepard Butte in eastern Hettinger County, which is canned by an undetermined thickness of sandstone and marly limestone resomblin/r the White River beds near Lefor. Strati graphic relations The White River beds unconformably overlie all older formations. , In west-central North Dakota they channel into and overlie the Ooldon Valley formation. From here on south they overlie progressively older formations until in and around the Black Hills in South Dakota they , rest unconformably on the Cretaceous pure shale. This indicates that t the Black Hills dome was uplifted in pre-Oligocene time, and the strati- ' graphic evidence from North Dakota suggests that this uplift was post- , Golden Valley (Wasatch), pre-Chadron (Lower Oligocene). The V*hite River formation in North Dakota is not overlain by any younger deposits.

MIOCKKE (?) OR PLIOCENE (?) SERIES (/ravels on an old erosion surfuco 7n northr-artorn "cntana ar rp.nanr of o hifh-level r>lrvnation surface capped by a conglomerate known as the Flaxville gravels. Vertebrate fossils collected from these gravels by Collier and Thorn indicate that this formation is late Wiooone or early Pliocene (1918, pp. 180-182). Later collections mace by various other geologists indicate that the Flaxville /travels may bo as young as middle or late Pliocene (Colton, R. H., U.?. Geological Survey, personal coronunication). Hurnoll (i960, p.58) visited several Flaxville localities ini.'ontana, and in one of these he found two gravels separated by a soil zone. He therefore suggests that the Flaxville gravel may include deposits of two aes, one middle to late Miocene and the other early to micdle Pliocene. Until the present work, no deposits in Xorth Dakota have been correlated even tentatively with the Flaxville gravels. The gently rolling upland- of the w°trn part of th° Stat** wore thought to ha-"-? been developed in late Tertiary time, but no definite age could bo assigned to them. Tn southwestern Grant County, T. 131 and 132., R. 90V/., are three high buttes, known ar Coffin Puttes, that are capped by the White River formation. Lloyd (1914, p.ll) describes the White River in these buttes as consisting of: ",. about 50 feet of calcareous arkosic sandstone overlying a marly limestone, both of which are referred to the formation on faunal and lithologic evidence. 11 The sandstones, which are 50 to 65 feet thick contain scattered pebbles and lenses of conglomerate. Tho lower parts of the buttes are com-

' posed of the Fort Union formation. The upland above which these

butter, ,-tand conr.istr, :n the r.ort/rvrr,. oart o*" T. 13v' '., H. v . of a cirsected planAtion surface capp~c by $s\'-'. i'et of I.-r, :-o consolidated gravel. The pebbles in the ,rravol are similar co t. in the Yihite iver conflomorato?. It ?on;s probabln that t/.* - vm? doi'.r :. - "d durinr n 1 atfi T°rtinry orosion rsyclp, tr''- nob"''--- derived from tho oror.ion of the V.hite River formation la th:- r.- Although no fosilp v;erc- found, the topographic and rtratipr:,.. position of those -ravels sur.'Ortr; that they were deoositeri .:" ., xhe saje tine a? Plaxville gravels in Montana. Similar deposits of gravel derived frorr. tho V.>.-i r,r 'ive.- !' v. apparently cap some of the higher hill? north'of Hautinor i . -. R. 95 and 96V/., but I did not have time to examine or map th -e . A picture of the dissected olanation surface nar Ooffi : " or. rl. 14 P.

Surflclal deposits Quaternary system Much of the Knife River area is mantled by surficial deposits, Pleistocene to Recent in age. The Pleistocene deposits include till, glacio-fluvial deposits, and fluvial deposits. Recent deposit- include alluvium, eolian sand and silt, residual silica, and land.lica debris. Over the country as a v;hole, the dividing line between "uhe Pleistocene and Recent epoch is difficult to establish, either physically or philosophically; in fact, Flint (1946 pp.205-208) has quosticr.td ' whether the Pleistocene epoch is ended and yhether there is r.eod c-c justification for the term "Recent". Although T am sympathetic "30 the idea that we are probably still in the Pleistocene epoch (if rhit

epoch be defined 'as one in which the distribution and volunc- of ! glacier ice is greater than in the "normal" geologic past) I think also that in glaciated areas "Recent" is a convenient term for deposits that post-date the last glacial advance. In this report, therefore, I have followed the current usage of the Oeological Survey end havo included in the Recent Series all deposits younger than those of the : Wisconsin Glacial Stage, his usage will be intelligible v/hether : "Recent" is eventually retained as an epoch, demoted to th r.nV OT" an interglacial stage, or denoted still farther to the rank of an inter glacial sxibstage of the Wisconsin stage.

PLEISTOCENE SERIES The subdivision of the Pleistocene epoch used in this report i.- as follows: Manlcato sub stage Wisconsin Gary substar;e glacial stage Tazev.-ell substae lov.-an pubtare Pleistocene Sanpamon interglacial stage Illinoian plaoial stage epoch Yarmouth inter glacial stap-e Kanpan glacial sta~e Aftonian inter glacial stae -v Nebraskan glacial stage Wisconsin stage All the Pleistocene deposits in the Knife River area have beer. assigned to the Wisconsin glacial state. There' if indirect °vid f>ncc of a pre-Wisconsin glacial advance, but no deposits related to this earlier advance have been found. The Wisconsin deposits represent at least three substages. The first two are probably lowan and Taz-- well (early "Wisconsin); the third is probably Mankato (late: Wisconsin). The Gary substage is apparently not represented in the Knife River area. Till (Qwt) Three major till sheets all of Wisconsin age, are present in the Knife River area. All have essentially the same appearance and composition, and their differentiation is possible only in the valleyr, r where they are interbedced with three fluvial and glacio-fluvial fills. On the uplands, where there are no interbedded stratified deposit*-, the tills are inseparable. A fourth till sheet young-er than the othr three, is exposed in

two IOOM 1 in -" -stern r>a-t, c? trv* area. Thir till is r.hin and diFcon.tin.ucus and probably reorients a local reacvance of tho Mankato ice sheet rather-than a separate substage. Distribution and topographic expression The 1ft te V.'i scon's in drift border (probably the I'.'ankato border) trends northwest through the Medicine Butte, Broncho, and C'oldon Valley quadrangles (see pi. 4 ). In general, northeast of this border the late Wisconsin till is extensively preserved coverinf over half of the area, and its surface IF characterized by numerous undrained depressions and unintegrated drainage. Seuthwe.c t of the border, on the other hand, the early VJisconsin till is thin and / patchy, covering less than 10 per cent of the area, and its surface has few undrained depressions. The actual boundary between these two types of topography,,however, is indistinct and difficult to map ex- ] cent in a funeral v/ay. It is characteristic of many of the large erratic boulders of granite to be at the center of small closed depressions, that appear to be not a part of the original glacial topography, but the result of the combined work of deflation and livestock. Cattle have a reat fondness for colleoting around these boulders and rubbing their heads and sides on them. Their hooves destroy the sod cover, and in the* wet season, sink down into the mud, leaving a rough broken surface. As the soil dries out it is easily picked up by the wind and it does not take more than a few score years to form a considerable degression around a popular boulder. Around some of the boulders is a rin that

at first glance appears to be polished by wind action. This rin fr. is probably due to the rubbing of horns by the cattle. This r.cthod of producing polished boulders in closed depressions around them is discussed in more detail by Pretz (1946 pp.260-262), who ascribed scr.e cf the tc the work cf risen. --Composition

The till consists of a poorly sorted mixture of sand, silt and clay in which are imbedded larger fragments of several kinds of rocl All the till in the area is highly calcareous. The matrix of the till seems to have been derived to a great extent from local rocks. The percentages of sand, silt and clay vary widely sometimes within the limits of a single outcrop. ~or example a road cut on the western edge of sec.12, T. 142N., R. 89V?. f shows till with a clay-silt matrix grading inabout 15 yards into till

'/ has a matrix of sand with only minor amounts of silt and clay, tions within a given till sheet, therefore, are probably as great differences between tills, and in that case mechanical analyses be of no use in differentiating till sheets. The fragments in the till range in size from granules to boulder" They are angular to round, with the majority being sub-angular and £ub'r~ round. These stones are of both local and foreign origin, with the

foreign stones slightly in the majority. The local stones are chieflw 1 sandstone, gray clay chert, fossil wood, and fragments of lignite ar [ f limonite. The foreign stones consist largely of Paleozoic limestone " t and dolomite from the Lake "Winnipeg area, and Pierre shale granite,

basalt, quartzite, ohert, and vein quartz from oisher parts of '.'a.~.i- toba. In addition, the till contains also a few pebbles of cuartzite and brown chert from the Rocky Mountain region. These were probably brought by streams into North Dakota or southern Canada and have "been reworked into thQ till.

Pebble counts of the stone? of foreign origin were made at numerous localities in the Knife River area in an attempt tc find some differences between the till sheets. The results of some of these counts are shown graphically in fig. 6 , and they indicate tha£ all the till5of the area have about the same types of erratic stones in about the same proportions. Limestone and dolomite are the nost abundant type, contituting 65% to 70% of all the erratic pebbles. Granite is next, constituting about 15% and all other types together make up the remaining 15% to 20%. To anyone who has examined tho erratic boulders that dot the surface of the land the high percentag-e. of limestone and dolomite is surprising at first; 90% to 95% of these boulders are granite and granite gneiss, and only 2% or are 1 limestone. Presumably the limestone was more easily broken in-co ' , small fragments than was the granite so that limeptone and dolomite predominate in the pebble size,while granite, which was plucked as large blocks, predominates in the oobble- and boulder sizes. ! The largest erratic boulder is in the Stanton quadrangle in the NE, SE}, see. 2, T. 146N., R. 84V*. (see pi. 22B ). This boulder whicA is composed of&ieiseic granite cut by two pegmatite veins, is about 22.5 feet long, 17.5 feet wide, and projects 3 to 7 feet above tho

ground. This large boulder is in the center of a closed dor-r- - i rn and its great size ia not apparent until the observer is very clore to it. --Color and depth of oxidation Unwenthered till in light to medium blue-gray, but thi<- color is seen in only a few outcrops. Apparently, because of oxidation by vadoj or ground water, most of the till outcrops are liht tan to yellowbrown in color. The depth of this oxidation is highly variable and bears little or no relation to the ae of the till. In some places blue oolor is preserved in early Wisconsin till within two or three feet of the surface. In other places 20 to 30 feet of Mankato till / appear to be thoroughly oxidized. Apparently, local differences in the position of the water table and local differences in the permeability of the till have much more influence on the depth of oxidation than does the age. Therefore, depth of oxidation, while possibly useful in differentiating Wisconsin from pre-V>isconsin tills is of no value in attempting to differentiate the substages of the Wic O onsin, in the Knife River a Solution and redeposition of calcium carbonate-- j In climates more humid than North Dakota calcium carbonate is commonly leached from the upper part of a deposit and is precipitated in the lower part of the soil profile. In more arid climates, the carbonate moves upward and is precipitated on the surface as caliche. In western North Dakota, both processes appear to be operating. 'Upward migration of the calcium carbonate has produced a coating of

caliche on tho bottoms of most of the pebbles on tho surface of the till. Downward migration has resulted in a faint lime-enriched zone about one foot thick jurt below the humified or "A horizon" of the modern soil. Evidently the balance between upward and downward migration of calcium carbonate is nearly perfect, because no exposures that were tested had been completely leached of lime; each was calcareous right to the grass roots. Hov.'ard (1947, p.1204) invertigated the possibility of using the amount of caliche on the under sides of those pebbles as a criterion of a,;e difference of tills in North Dakota but found that variations due to local factors were apparently greater than the variations due to age differences of the tills. In the Knife Kiver area the depth of secondary lime accumulation in the soil is somewhat variable, but in general the lime zone of the soil on late Wisconsin till is as thick and as well developed as that on early Wisconsin till. -Stratified till" Although in most outcrops the till is an unsorted heterogeneous mixture, in a few localities it is crudely sorted and partially stratified. Lenses of fiilt and fine- to coarse-grained sand alternate with lentils of unsorted till. The fact that the lenses of stratified material alternate with lenses of unsorted till and the fact that in some places outcrops of the stratified till can be traced laterally into exposures of completely unsorted till indicates that tho environment of deposition of tho stratified material was probably subglacial. The sorting and stratification of the sand and silt on the other hand, /Off'

indicates the action of water. A third significant factor is that all the outcrops of thf stratified till thus far soon h/ive ben in valley bottoms. It appears that suoglaoial melt waters flowing at the base of the ice were concentrated in tho pro-existing valleys. . Where the volume and velocity were great enough, these waters wore able to stratify and sort the glacial drift as it was being deposited by the ice. Hood examples of this stratified till oan be seen in the NK; , of sec. 20, T. 143N., R. 91W. in the Broncho quadrangle, and in the Svv|- of sec. 35, T. 145N., R. 85"n. in the Stanton quadrangle. The exposure in the Broncho quadrangle is about 150 yards west of a church and is in a man made gully in the south side of a section-line road. It shows about 8 feet of interfingering and interbedded till, fine sand, silt and gravel overlain by loops. The exposure in the Stanton Quadrangle is in a cut bank on the north side of the Knife Hiver and shows 10 to 12 feet of poorly sorted and stratified glacial drift overlying fort Union sandstone. One of the largest exposures of "stratified till" in western North Dakota is about 9 miles west of the Knife River area in the eastern half of *ec. 3, T. 142N., R. 93W. in eastern Dunn County. Here the Knife Hiver has cut into the south bank of the valley and has exposed 50 to 60 feet of interbedded lentils of till, silt, fineto coarse sand, and gravel. --Pebble-orientation studies Holmes (1941) showed that in undisturbed till the pebbles tend

to lie v/ith their long; axe? parallel to the direction of ice movement. Accordingly, a statistical study of pebble-orientations in a till can be used, like striations, to indicate the direction of movement of the glacier. In the Knife River area, pebble-orientation studies were made at eleven localities to see whether the direction of ice movement was signifioantly different for the several till sheets. The method used is essentially a modified version of the one described by Holmes (1941, pp. 1307-1308) and was suggested to me by H. E. Simpson Jr., of the Geological Survey. The results of the studies, shown by rose diagram (fig. 8 ) are / somewhat disappointing. Several of the diagrams show no strong preferred orientation; and in those that do have a strong orientation, the direction varies from southwest to southeast with no apparent correlation between the direction of movement and the age of the till. Variations within a given till sheet are apparently greater than any average difference between the sheets. In general the orientation diagrams seem to indicate that all the ice sheets came [ from the north and northeast, and deviations from this trend are nrobftbly the result of local touograt>hio control of ice movement. Localities 10 and 11 are in the Krem moraine, and, as would be expected, show orientations about at right angles to the trend of the moraine. Therefore, although the studies fit the hypothesis that the long axes of pebbles parallel the direction of ice movement, they are of no help in differentiating till sheets in the Knife River area.

Differentiation of tills As can be seen from the above descriptions, the several tills of the Knife River area are similar in appearance and composition, and cannot be separated by any physical or chemical tests thus far tried. The late Wisconsin drift has been less eroded than the earlier drifts, and this difference can be used in mapping the approximate limits of the late Wisconsin (Mankato) till. This topographic boundary, however, is vague in many parts of the area, and topography alone is not considered adequate evidence for inferring the existence of more than one till sheet. There is, however, amole stratipraohic oroof that more than one age of till is present in the Knife River area. In the vaLleys of the Knife River and some of its tributaries are three fluvial and glaoio-fluvial fills separated by major unconformities. Several exposures show two tills, underlying and associated with the second and third fills. Also, although the base of the first fill is not exposed, this fill is composed chiefly of glacial outwash and indicates the existence of still another glaciation. This older glacial unit is probably lowan and was the most extensive of all the Wisconsin drifts in North Dakota. The two younger tills are probably Tazewell and Mankato. The exposures that show the critical relations between the tills and fills are described in the section on the valley fills, Qsd. A fourth till is exposed at two localities in the eastern part of the Knife River area. The first locality, a gravel oit in SE-, W.1 sec. 5, T. 145N., R. 85W., shows 1 to 1.5 feet of till conformably overlying the late Wisconsin outta?h gravel, Qwo. The second locality

is also a gravel pit, this one in ME.}, NS-1 , sec. 2, T. 145N., R. Here about tvo feet of light buff till overlie ice-contact gravel that is a part of the valley fill, Qsdj. Tn both exoosures the till is thin and in neither is there any evidence of a period of erosion or noil formation between the till and the underlying gravel. Therefore, this till is interpreted as recording a local readvance of the late Wisconsin (Mankato) ice sheet rather than as a separate substage. Tn addition to the exposures of till interbedded vdth the waterlaid deposits, two exposures show two tills in direct contact. The first of these is in the NE-, NE-J- sec. 28, T. 146N.. R. 88W. , in the northern part of the Beulah quadrangle. Here, at the edge of the upland just east of the Reulah trench, a roadcut exposes a dark clayrich till overlying a lighter sandier till (see pl.lSA ). At the top of the lower till is a band of pebbles, evidently concentrated by the erosion of this till before its burial by the younger deposit. The second exposure is in the SEj, SKy sec. 12, T. 144N., R. 86V,'., and is in the valley of Kinneman Creek, .just south of State Highway 25 (pl,15B ). A cut bank on the east side of the creek exposes about 15 feet of relatively unoxidized blue-gray till that seems to have a break in the middle. At this break thre is a concentration of boulders and a dying out of many of the joints present in tho lower half of the outcrop. The evidence is not clear-cut, but two tills

are probably represented. Although both of these exposures demonstrate the existance of more than one till sheet in the Knife River area there is no means of correlating them with each other or of mapping these till sheets separately.

Two bed? of till separated by a thin bed of soft marl 'contains' fresh-water gastropods are exposed in a road cut on the south Fide of a small valley in NE-J-, SE sec. 30, T. 146N., R. 89V<. , (see pi. 16A). However, exposure? on the north side of the same valley fail to show either the marl bed or any evidence of two tills. The marl was probably deposited in a small pond, formed during the retreat of the late Wisconsin ice margin, and was buried by till deposited during a local readvance of the same ice. Krem moraine (Ctr) A small lobate recessional moraine composed chiefly of till crosses the northern part of the Knife River area in the "Golden Valley, Beulah, and Hazen quadrangles. The eastern end of this moraine is a few miles north of the abandoned villa0e of Krem, and in this report the moraine will be informally called the Krem moraine. The moraine is differentiated from the rest of the till solely on the basis of topography, and therefore, on the geologic map (pi. l) the boundary of moraine has been shown by hachures rather than by a conventional geologic contact (t)l. 16B). The Krem noraine is local and-cannot be traced much beyond the limits of the Knife River area. Its eastern end is in the northeast part of the Haren Quadrangle. Northeast of this area the moraine becomes indistinct and merges imperceptibly with the till-oovered up-

land. The 'moraine dies out also in the northern part of the Golden Valley quadrangle, but, after a gap of about 3 or 4 miles, a faint extension of the moraine lies partly in the northwest quarter of the Golden Valley Quadrangle and partly outside the area in the Fort

Perthold Indian Reservation. Northwest of this outlier the r.oraine, if ever present, has been so modified by erosion that it cannot be identified. Throughout most of its length, the Krem moraine follows the crest of the upland between the Knife and Missouri rivers. In tho central , part of the Beulah quadrangle it crosses) and 'drops into] the deep Beulah trench; and in the eastern part of the caice quadranfole, it crosses a broad topographic sag at the head/aters of Antelope Creek. Where it crosses these topographic lows the moraine consists partly of till and partly of ice-contact stratified drift. Elsewhere it consists almost entirely of till. There is no reason to separate the Krem moraine in ae from

the late-Wisconsin till south of it. Outwash from the moraine merges without apparent break with outwash and fluvial deposits associated with the late-Viisconsin till in the Knife River valley. Also, both the east-west trend of the moraine, which lies at about 45° to the general trend of the drift borders in North Dakota, and the dying out of the moraine in the Hazen and Golden Valley quadrangles suggest that the moraine is merely a local detail in the general picture of late-Wisconsin glaciation. The Krem moraine therefore does not seem to represent the drift border of any Wisconsin substage, rather it records only the readvance of a small local lobe of the ioe sheet. Valley fills (Qsd) In tho valleys of the Knife and J-'missouri rivers and in some of their major tributaries are reraiants of fluvial and glacio-fluvial deposits of silt, sand and gravel. The largest remnants of these

fills are preserved a? terraces in the Knife rtiver valley downrtreafrom Beulah. Upstream from Beulah patches of the fills are fewer and smaller. Three different fills have been identified, and, where possible, those have been mapped separately as Qsd, Qsd and Qsd~. YJhere it was not possible to differentiate the fills they have been j,:>,lO mapped under the general head 'Qsd. Qsdand Qsd? are of early Wisconsin ae and probat.ly represent the lowan and Tazewell substager. Qsdg is late Vvic COnsin and probably represents the Mankato substage. First fill, Qsdi-- Qsd is the oldest of the three fills and has been identified only in the Knife River valley near Beulah. It.consists of outwash sand and gravel with cut-and-fill type bedding that is foreset down the valley of the Knife River. The gravel beds contain stones of both local and foreign origin, with the foreign stones predominating by about 3 to 1. Many of the coarse gravels are partly cemented and stained orange by deposits of iron oxide. The sand beds, which are very much like the sands of .Qs and Qsdg, are fine-to coarse-grained and have little silt content. Characteristically they are gray and oxidize to buff on surface exposure.

The following sections show tho lithology and stratigraohic relations of Qsd. : (l) Beulah gravel pit. This exposure is on the south side of the Knife River in NE-J, sec. 36, T. 144N., R. 88Y,1 . The section given below was taken at the eastern end of the workings approximately one-tenth of a mile east of the rodeo grounds (pi?. 17 and 18). Top of exposure - approximately 32 feet above Knife River floodolain,

Feet Spoil pile from f.ravel pit 2.02. Colluvixrn v/ith modern scil rol'ilo on to Laminated silt fine sand, some lino enrichment especial ly in tho s ilt bed? in the UDor ?. feot.Qfid . Humifind oolian rand Folian sand with faint lime-enriched zone Eolian rand, lijrht brownish gray Coarse, crossbedded sand and gravel, outvash tyt>e, limestone and granites more abundant than iron oxido and r.andr.tone pebble, gravel beds are iron stained. Qsd Bottom of exposure Traced westward through an abandoned part of the avel pit to a cut at the north edge of the rodeo grounds, the boundaries betwofln units become less sharp. Qsd£ grades from laminated fine sand and silt into laminated fino to medium grain sand similar to the sand beds of Qpd-. The eolian sand thins and the humus zone becomes faint and finally disappears. Although the humus zone is very striking in the outcrop the soil profile developed on the eolian sand is not a strong one as is shovm by the very thin lime zone beneath the humified layer. The contact between Csd, and the overlying eolian sand appears to be conformable, suggesting that the original top of the old out- *ash is propervod in this exposure. If this is true, Qsd never rose higher than about 20 feet above what is now the modern floodplain of the Knife River.

(2) Section Line exoorure. This ext>OFure, approximately mile ea?t of the Deulah gravel pit f is in a gully along the east edge of *ec. 36, T. 144N., R. 88W. TOD - About 65 feet above Knife River floodplain Poorly pxpopod unit; mortly f inr-grained sand and gilt with some bed? of medium- grainnd gray sand Laninat.pd finn sand and silt with a few thin beds of medium- grained sand Rand, mdi un-to coarre-trained , with few gravel lenses and some beds of fine sand,silt Conformity Till, medium blue-gray, sandy Unconformity - ' Qsdx Fine-grained sand and silt, thin-bedded, with some bods of laminated silt and clay Coarre-grained sand outwash sand and gravel Base of exposure, a few feet above the floodplain of the Knife River The till (unit 4) has the same stratigranhio position as the eolian sand in the Brulah gravel pit. Exposures are poor between the Section Line exposure and the Beulah gravel pit so that it ia impossible to tell the exact relationship between the till and the soil zone. I interpret the till as being slightly younger than the noil zono, which was either not developed in the Section Line exoosure or was removed prior to the deposition of the till. East of the Section Line exposure Osdi is directly overlain by Osd with an unconformable contact that ranges from about 15 feet above the Knife River Hoodplain tov below river level.

(3) KeoFh Ranch exposure. The Keofh 3ar.=h erasure is in a gully on the face of the terrace on the s ou' h ?i of the Knife River, in S7,:v, sec. 8, T. 143N., R. 88Y<. Thi? r~-sure show?? essentially the same stratirrauhic section as "crs? iki-ion Line exposure . Top of exposure P Eolian sand Qsd 2 fnossibly Q; 1 ? Sand, medium-trained with a few pebbles Conformity fi Till, medium to dark blue-r.ray, clayey t Unconformity Laminated clay and silt, dark pray, beddinr neath the till slightly crumpled Coarse-brained sand and nobble fravel , starae-i yellow by iron oxide Medium-to coarse-brained sand with stringer? of / ( cranual to pebble gravel; crossbedded Very coarse gravel with boulders up to two in diameter Interbedded fine-to coarse-grained sand SLTC pebble gravel Base of exposure, about 10 feet above the Xnif* river floodplain. On the west side of the mouth of this gully Fort Union strata unconformably underlie Qsd, but the contact diTr* to the east and the base of Qsd is not exposed in the gully it*?lf . There is no direct connection between the Keogh Ranch emr-zre and the Section

Line exposure ar.d th-? correlation of the units is baced entirely, on

their physical 're semblance ana similar strati grannie succession. In both exposures the till is unoxidized, suggesting that the overlying sand (Qsd2) war? deposited .coon after the till. iThe ptratiraphic relations of Qsd can be summarized as follows: (1) The bas of Qsd, is not exposed except at the west end of the Keogh Ranch exposure, where it rests unconfonriably on Fort Union strata. However, inasmuch as Osd is an outwash deposit and is separated from overlying deposits by an unconformity or by a soil zone, it presumably overlies till at least in the stratigraphic time scale if not in actual physical contact. (2) The top of Qsd is exposed at several localities. In one exposure, the Beulah gravel pit, it is overlain conformably by eolian sand on whioh is dvplooed a veak oil orofil. Tn all oth exposures, the overlying deposits, either till or a younger fill (Qsdg) lie on the eroded surface of Qsd-,. Because of this erosion and because of the weak soil profile, I believe the interval following deposition of Qsd, was long enough to be considered as an interglacial sub stare of the Wisconsin.

Second fill, Qsd2 is the second or intermediate fill in the Knife River valley and volumetrically wathe largest of the fills. It is predominantly fino-rain and consists largely of locally derived materials with only email percentages of glacial debris (fig. 7), Beds of fine-grained sand and silt make up over half of this fill and vary in appearance from place to place. In como outcrops,

they are li~ht rray, o-'.ly fairly v.-ell sorted, and greatly reremble much of the modern alluvium of th floodplain deposits. In other outcrops they are tan to yellow, well ported ana laminated, and look much like the finer grained beds of the Fort Union formation. In many exposures the uonr 2 to 4 feet of the silts have been so churned up by frost action and soil creep that the lamination? have been obscured, leaving a nearly structureless tan silt that looks like loerr. In this structureless material, however, are small polygonal fragment? that still preserve the original laminations, showing that the material is not of eolian origin but is a part of Qsdg. Interbecded and inter finder ing v/ith the fine-drained sand and t silt are bedr of medium-to coarse-brained sand and lenses of r>pebble gravel. The oebbles of the gravel beds consist of 60 to 80 per cent iron oxide, chert, and sandstone derived from th local bedrock", which readily, distinguishes these gravels from the outwash gravels of Qsd and Qpdj. The ,',ray sands, on the other hand, are identical mr8-- scopically with the sand of the other fills. -Distribution Qsdo is well developed and its relations to other Pleistocene deposits are best shown in the southeast part of the Beulah quadrangle. In this area the top of the fill is at Irast 70 feet above and the base is below the floodplains of the Knife River and Spring Creek. The following exposures show both the lithology and stratiraphic relations of Qsd2: (l) Three exposures, the Keo.h Hanch exposure, the ection Line exposure, and the "eulah prnvel pit show the contact between Osd2

the underlying till and older fill, Qsd, . These exposures have already been described. (2) At the west end of the main, street of Beulah, Qsd£ is exposed in a road out on the wept pid* of a north-south £ravl road. In thip exposure, Qsd£ consists of yellow-brown laminated fine-grained sand and silt that greatly resembles certain beds of the Fort Union formation except that it contains a few pebbles of granite and limestone. About 25 yards north of this exposure and about 20 yards east of the gravel road, an excavation made in 1950 showed the fine-grained sand and silt unconformably overlain by till, which was in turn overlain j by the outwash fill, Qsd,. The contact between Qsdg and the till dips east into a small .valley in the western half of 500. 25, T. 144N., R. 88V/. Apparently Osd2 had been eroded from this valley prior to the deposition of the younger till and Qsdj (fig. 9P). (3) An exposure in the face of the terrace on the north side of Spring Creek in NW-J, SE- aeo. 20, T. 144N., R. 88W. The lower part of the exposure consists of Fort Union sandstone and shale unconformahly overlain by 8 to 22 f*et of laminated fine-to medium-grained sand with gravel lenses and thin clay seams. About 80 percent of the pebbles in the gravel lenses are from the Fort Union formation and this fill is believed to be Qsdg. On the eroded surface of Qsdg is 5 to 17 feet of partly oxided till and against the sloping surface of the till is deposited the outwash sand and gravel of Qsdj. (4) On the line between sees. 19 and 20, T. 144N., R. 88W. an abandoned dugway in the face of the terrace south of Spring CrAek exposes the following section:

Top of exposure, 30 feet aVove floodolain of Spring Creek. Feet, Spoil material Eolian sand, dark brown, humified Sand and gravel, poorly sorted Unconformity Till, sandy, tan to buff, oxidized Unconformity Sand and silt, laminated, tan to brown, Qsdg Unit 4 is late Wisconsin till (here completely oxidized), and unit 3 is a gravel skim deposited by Spring Creek as it planed across the top of the till. About 100 yards south of the exposure, the top of the terrace is slightly higher and an auger hole penetrated about 8 feet of medium-grained gray sand and gravel, which is probably a remnant of the outwarh fill, Qsd. (5) In the Zap branch of the Beulah trench large rerjiants of Qsdg, which here consist almost entirely of fine sand and silt, are preserved in the northern part of T. 144N., R. 88W. The tops of these remnants are up to 50 or 55 feet above the prerent floor of the valley. Till caps some of these remnants and small patches of till overlying Qsdg can be found all the way down to the present / valley floor, showing: that Qsdg had been extensively eroded before this till was deposited. Overlying the till and the firs sand of Qsdg is an outwash terrace (Qwo) whose top is 15 to 25 feet above the present valley floor. This outwanh is the equivalent of Qsds in Spring Creek and the Knife River valley.

The distribution and lithology of Qsdg outside the area jiear

Beulah can be summarized as follows: (1) West of Beulah, Qsd£ becomes ooarser grained and consists largely of fine-to medium-grained gray sand, which is difficult to separate from the sands of the other fills. In the valley of Spring Creek, remnant? of Qsd2, overlain by till and Qsd , have been identified as far west as Dodge in the Golden Valley quadrangle. The best exposures of Qsc and the overlying till are in railroad cuts in sec. 18, T. 144N., R. 90W. In the valley of the Knife River, remnants of the Pleistocene fills are smaller and more widely scattered as one goes upstream from Beulah. Most of these remnants consist of medium-grained gray sand and have been mapped as Qsd, undifferentiated. Some of these sand deposits in the Medicine Butte quadrangle are almost certainly Qsd£ for scattered over their surfaces are boulders and cobbles of granite. These stones may have been deposited on top of the sand directly by ioe or may have been rafted on the icebergs in the lake that almost certainly existed in the Knife River valley upstream from the margin of the late Wisconsin ice sheet. A fill lithologically like and probably equivalent to Qsdg apparently underlies at least the east-west segment of the Golden Valley trench. This fill is exposed only in the northern part of sec. 25, T. 144N., R. 90W., where a north-flowing stream has cut below the alluvium and colluvium that form the present floor of the trench. (2) East of Beulah the Pleistocene fills form a nearly continuous torraoo on the south side of the Knife River valley. In

certain isolated outcrop? the fine-grained panes and silts of Qsdp unconformably underlie and can be distinguished from the coarser outwash gravels of Qsd,. Most of the terrace, however, consists of fine-to coarse-brained sand with minor amounts of gravel and has been* mapped Qsd, undifferentiated. Much of the surface material of tho terrace has been reworked into sand dunes, especially in the Hazen and Stanton quadrangles, and it is impossible to tell the character of the underlying fill in these areas. On the north side of the Knife River valley, east of Beulah, the fills are less continuous in outcrop but are more easily distinguished. Jurt west of Hazen in the lower part of the valley of Antelope Creek, several pxr>osur<5 show the fine-prainpd sends of / Qsdg unconformably overlain by the outwash gravels of Qsdg. One exposure, a road cut on the south side of State Highway 25, near a junction with a gravel road, in the NE-J, NWj, sec. 11, T. 144N., R. 87V., shows the following section: Top of exposure Feet Coarse sand and gravel, QscU Unconformity Fine-grained sand and silt, laminated, tan to buff Osd2 5- Till, clayey, tan to brown, oxidized. 3 JL Base of exposure. In 1950 this exoosure had slumped in so badly that the till was no longer visible without diin or aug,ering at the base of uhe road ditch. (3) In the Imissouri Valley downstream from the mouth of the Knifo River a fine-grained sand and silt donosit that is probably remnant

of Q8 d 2 is found, in seos. .17. 18, 20 and 21, T. 144X., R. 84V.

The tops of these remnants are 70 to 80 feet above the present Missouri floodplain but these tops are erosional and give no indication of the original altitude of the top of the fill. The base of the fill is below the floodplain of the Missouri River. (4) Upstream from the mouth of the Knife River Qsd2 was not identified within the walls of the Missouri trench. In several of the small tributaries of the Missouri River, however, are patches of a fine-grained fill similar in lithology and stratipraphic position to Qsdp in the Knife River valley. These patches of probable Qsd£ crop out in the following localities: (a) In the Stanton Quadrangle in SE-ir sec. 36, T. 146N., / R. 85VJ., a small excavation near the head of a north-flowing coulee exposes about 6 feet of light yellow to gray-brown fine-grained sand and silt, similar tr the fine-grained facies of Qsd£ in the Knife River Valley. An additional 6 feet of this material was penetrated by an auger hole. Further augering indicated that the deposit probably extends southwest into the northern part of sec. 1, T. 145N., R. 85W., and underlies the divide between the north-flowing coulee and Elm Creek, a small tributary of the Knife River. This divide is low and broad, and appears to have been enlarged by waters diverted from the Missouri trench south into the Knife River. A few small kamrs and patches of till are found near the floor of the divide and granite boulders dot the surface of the fine-grained fill, indicating that both the enlarging of the divide and the deposition of the fill antedated the last glaciation. Thus the fine-grained fill occupies

the same stratigraohic position as Qscio in the Knife Kivor valley. The top of the fill is at an altitude of about 1890 feet, or about 200 feet above the modern floodplain of the Missouri River. InAsrnuch as the base of Qsd2 is known to extend below the modern floodplains of the streams in other parts of the Knife area, it appears that Qsd2 was at least 200 feet thick in this part of the Missouri trench. (b) In the northwestern part of the Stanton quadrangle in NE>;, SW*-, sec. 1, T. 146N., R. 85V/., a cut on the south side of a gravel road leading to the abandoned site of Mannhaven exposes 10 to 15 feet of till overlying 15 to 20 feet of medium-to fine-grained brown and gray sand with some beds of laminated tan silt. This fill r is correlated with Qsd£ on the basis of similar lithology and similar stratigraphic position. The top of the fine sand is at an altitude of about 1,900 feet or about 200 feet above the present floodplain of the missouri River. (c) In the northern part of the Hazen quadrangle in T. 146N. f R. 86VY., are several deposits of fine-grained sand and silt that are probably Qsd£. In two localities the fine fill is overlain by late Wisconsin till. The first of these localities, a road cut on the west side of a deep coulee in NE-, NE--, sec. 17., shows the following section: Top of exposure Feet Outwash sand and gravel 2~g Till, sandy, yellow-brown to light gray Outwash gravel 1-$ Sand, fine-grained, tan to yellow, calcareous, probably Qsdg

H

Base of exposure The IP: feet of gravel between the till and the fine-f,rainad fill is interpreted as outwash laid down in advance of the late Wisconsin ice snoet. The second locality is in NT, sec. 18, and consists of several exposures in a deep coulee that leads north to the Missouri iliver. These exposures show two tills that are in direct contact at the sides of the coulee but that are separated by 10 to 20 feet of finegrained sand near the center of the coulee (see fig. 9A ) Both the upper and lower contacts of the fine-grained fill are erosional. The lower contact is marked by a boulder pavement developed on the top f of the underlying till. The upper contact is nearly flat but locally shows that a few of the beds of the fill were eroded prior to deposition of the upper till. In one exposure, just west of a white adobe house, the upper 18 inches of the fill are enriched in calcium

carbonate; this may indicate a period of soil formation prior to deposition of the upper till, or it may indicate merely precipitation of carbonate from ground water. "Where the two tills are in direct contact, the boundary between them is not marked by any concentration of stones and is difficult to pick. The upper part of the older till is well oxidized to a yellow-brown, whereas the lower part of the younger till is only partly oxidized and still preserves "islands" of blue-gray color. The contact between these color differences, however, is gradational and not easy to identify except by careful inspection.

Strati graphic relations The strati graphic relations of as indicated by the sections just described and by the sections described in connection with Qsd-i are as follows: (l) Qndg lies unconformably on Qsd and the Fort Union formation. In the Keogh Ranch and 'Section Line exposures, Qsdo separated from Qsd- by till with no apparent erosional break between this till and Osdgexposures in sec. ,18, T. 146N., R. 86W. , however, Gsd2 R penarated from an underlying till by- a boulder pavement, indicating at leant a short interval of erosion. The main unconformity, however, seems to be between Qsd and the overlying till rather than between this till and Qsd2 and this unconformity is thou/ht to represent an t inter glacial substage of the Wisconsin sta (T,e. (2) The upper contact of Osd2 is erosional in nearly every exposure. Till and Qsd unconformably overlie Qsdg at altitudes ranging from below the floodplain of the Knife River to about 70 feet above the river near Peulah and 200 feet above the Missouri River in the Stanton quadranple. This indicates that Qsd£ was extensively dissected prior to the last ice advance and the deposition of the till and outwash. The extent of this dissection suggests that the interval recorded was considerably longer than the interval between Qsd and Qsdg* Interpretation The origin of 0?d2 i R no altogether clear. Although it contains some reworked glacial material, the bulk of this fill was derived from the local bedrock of the area. Dosnite fact, Qsd_ peems to be olosely associated, at least in time, with the till that underlies it and separates it from the older outwash fill,

It seems likely, therefore, that Qsd£ ifl an irr-vash deposit formed during the wastage of the ioe sheet that deposited the undor- / lying till. A lobe of this ioe sheet probably extended across the Missouri River trench at some point downstream from the mouth of the Knife River, causing both streams to be ponded and to agfrade their valleys. The fine-grained eand and silt beds of Qsdp indicate deposition / in water that was standing or flowing very slowly. The lenses of coarse-grained sand and gravel, on the other hand, indicate deposition by running water. This alternation of fluvial and fluvio-lacuotrine conditions was probably caused by fluctuations in the height of the ioe dam in the Missouri trench. r Third fill, Qsd3~ Qsdj is the youngest and most extensively preserved Pleistocene fill in the Knife River valley. Like Qsd it consists principally of glacial outwash and most of the stones are of foreign origin. It differs from Qsd, in being less well size sorted and in lacking cutand-fill type oross-bedding. A few exposures of Qsdj show some

cross-bedding but in most places the beds are essentially parallel. "Where Qndj consists of interbedded sand and gravel it is easily distinguished from the underlying Qsd£. A large part of Osdj, however, consists of medium-to coarse-grained gray sand with only a few stringers of pebble gravel and this facies is almost impossible to tell fromtho sand faoies of Qsdg. Some of the gravel beds contain cobbles and small boulders of granite, suggesting that this fill was deposited fairly close to the edge of the ice.

The thickest deposit? of Qsdg are in the Stanton quadrangle noar the mouth of tho Knifo River. Here the baso of tho fill is bolow the modern floodplain of the river and tho top is 100 to 110 feet above the floodplain. The profile of the upper surface of Qsdg io gentler than that of the modern floodplain, so that west of Stanton it converges on the floodplain and Qsd 3 gradually thins. , At Peulah the too of Qsd-? is about 60 feet above tho floodplaln of ' O the Knife River; and at Dodpe it is about 25 feet above the floodplain of Spring Creek. - Distribution and Interpretation- In the valley of the Knife River, Qsdg hee been identified from the mouth of the river at Stanton as\far west as the mouth of Brush Creek in the Medicine Butte quadrangle. West of this point the Pleistocene fills consist chiefly of medium-grained sand and cannot be differentiated. As indicated in the discussion of some of these sand deposits are probably the intermidate fill, Qsd,. Others are probably Qsdj, because the western part of the Knife River is beyond the late Wisconsin drift border, and in this area Qsdg probably loses its glacial character and is an inwash fill similar to Qsd2 In the lower 4 miles of the valley of Brush Creek Qsdj consists of parallel-bedded coarse-grained sand and gravel 25 to 50 feet thick. These deposits differ from Qsdj in other parts of the area in that they are better sorted and contain only small amounts of nilt and olay. Just why the fill in this valloy is better sorted than the resJb of Osd, is not clear, but the fact that it in makes the fravel

deposits in Brush Creek one of the better sources of concrete aggregate in the Knife River area. In the valley of Spring Creek Osdj has been identified as far wort as Podge and probably extends still farther upstream. About 2 miles northeast of Golden Valley, in the southern part of sec. 12, T. l44N. f R. 90V., is a deposit of very coarse sand and gravel that has been mapped as Qsd . The top of this deposit is 55 to 60 feet above the floodplain of Spring Creek, or about 20 feet higher than ' ' the general top of Qsd 3 in this vicinity. The bedding is foresrt almost due south into the valley of Spring Creek (see pi .20A) The deposit probably represents either a small kame built directly afrainjr ice or a delta built out into water ponded in the valley of Spring Crook. Auger holes around the margin of this deposit indicate that it lies in part on bedrock and in part on the fine-grained sand of Qsd 2 . Between Hazen and the mouth of the Knife River at Stanton the character of Qsdj is different on opposite sides of the valley. As indicated in the discussion of Qsd2, the terrace on the south side of the valley consists largely of medium-grained gray sand that has been mapped as Qsd, undif ferentiated. Although some of this sand may be part of the intermediate fill, Qsd2 a large part of it belongs in Qsdg. In contrast, the deposits of Qsd~ on the north side of the valley consist largely of coarse-grained sand and gravel, and those deposits overlie,, and pinch out against remnants of the cut torrace, Qt. This terrace pre sumably cut while the Knife River was flowing - at the .top of Qfld,, is not developed on the south sido of tho valley.

The differences between the north and routh side? of the valley may be explained by postulatin/r that in this part of the valley Qsdg represents a kame terrace rather than a wall to wall outvm~>! fill. The pennral fineness of grain size and the parallel beddinr of Qsdj on the south side of the valley do not fit the usual concept of an ice-contact deposit. However, the valley of the Knife Kivor downstream from Hazen is broad and open. If there were a roridual tongue of ice in the northern part of the vallfy, the southern part could easily have been filled with a deposit of comparatively finegrained outwash that would show no ice-contact features except at its northern ira rgin whore they could easily have been romovcd by later erosion. In most exposures in the Missouri River valley, Qsd consists

of parallel-bedded glacial sand and gravel that is similar both in appearance and in composition to its counterpart in the Knife iver Valley. The. largest .sinrle -deposit of Osd in the Knife River area is in the Missouri vnlley and underlies a large terrace-like flat in the northeastern part of T. 145N., R. 84VJ. Although in most of the Knife River valley (except possibly for the segment between liazen and Ttanton) Qsdj represents a wall-to-wall outwnsh fill, in the Missouri River trench this fill is a kame terrace. The evidence is as follows: (l) In several localities the modiurn-to coarse-grained sands of Qsd3 are interbodded with layers of poorly sorted very coarsegrained gravel containing slabs of lignite and boulders of granite

up to 2 feet in diameter, suggesting deposition close to the glacier. Tho so coarre beds are well exposed in tho gravel pits in sec. 21, T. 146\'., H. 84W. f (see pi.203). In Wt], sec. 16, T. 145V. f R. 84W. , are smll gravel donor its thftt appear to preserve tho constructional tor>o/,rar>hy of small kanor and crevasse fillings. (3) Tn thp northeast corner of sec. 2., T. 145V., I*. 84Yi., i r a gravel nit exposing 25 to 30 foot of medium-to coarse- ;~rain sand and gravel with tho collapsed bodding tyoical of ice-contact deposits (reo pi. 21). A few exposures plus some auger holes east of the pit

indicate' that this ice-contact travel interfingers with and in Tart / underlies the parallel-bedded gravel that is more characteristic of Csdj. At the south end of tho pit the ice-contact gravels are ovorlain by a bed of till 1 to 2 feet thick, indicating a local roadvance of tho ice across this area after Qsd~ had been deposited. The ice front, therefore, was probably never very far from the Missouri trench while Qsd,, was being deposited. (4) In the center of the Missouri tronch is a gravel formation, slightly younger than Q.sdj, that has been mapped as Missouri River gravel, Qmg. Tho base of this gravel is slightly below the modern floodplain of the Missouri River and rest? on unweathered till and local deposits of lacustrine sand and clay. As. indicated in a s>;b- poounnt section of this report, Qmg was deposited after the ioe had vacated the Missouri River tronch, but before it had retreated beyond tho Max moraine. Very likely Cm.< antedates also the roadvanoo of tho

ice that locally deposited the till on top of Qsdj. If, therefore, Qs.d- was deposited as a wall-to-wall outwash fill, it -was completely eroded from the center of the Missouri trench during the comparatively short interval between its deposition and the final retreat of the ice. Such rapid erosion, while not impossible, ie difficult to imagine at a time when streams would norually be overloaded and would bo tending to aggrade their valleys. Also, there is no evidence of this rapid erosion in the Knife River valley. If, however, Osd is a kame terrace this difficulty is eliminated. The center of the Missouri trench could have been cleared ouickly and completely by the molting of the stagnant ice, leaving the space,that was soon filled by the Missouri River gravels. Stratigraphio relations- Qsd- is closely associated with, o and in places directly overlies a till sheet. Where this till is absent, Qsdj is unconformable on Qsd.2 and older deposits. The interval between Qsdo and Qsdj was long enough to allow the erosion of more than half of Qsdg from the valleys of the Knife and Missouri rivers. Qsdg grades laterally into deposits mapped.as late Wisconsin outwash, Qwo. Qwo and Qsd are similar in appearance and como position and they have been mapped as separate units chiefly on the basis of geographic distribution and association with other Pleistocene deposits. Qsdg is the time-stratirraphic ecmivalent of the ice-contaot deposits, Qic. Qsdj is slightly olc>r than the Missouri River gravels, Qmg, and the youngest icp-contaot deposits, QIC', and is

unconformably overlain by the latter deposits. All there unit?,

however, are. closely associated in time and belong- to a sinple substage of the Wisconsin. The eroded upper surface of Qsd is locally overlain by dune o sand, Qds, and post-glacial alluvium, Qal. In most localities, however, Osdj is not overlain by any younger deposits. Late Wisconsin gravels essentially c onteruooraneous with Qrdg Other than the outwash fill, Qsd,, gravel deposits of late Aisconsin age are confined to the northeastern part of the Knife J<ivor area, within the late Wisconsin (Mankato ) drift border. These deposits have been mapped under the headings of outwash (Qwo), icer contact deposits (Qic and Qic 1 ) and Missouri River gravel (Qmf). Outwash (Qwo)-- In the northern part of the Knife River aroa, especially in the Beulah and Hazen quadrangles, are numeroup deposit? of late Wisconsin outwash (Qwo.). This outwash consists of moderately well-rorted , parallel-bedded sand and gravel in which most of the pebbles arp of Canadian provenance.(fig. 7). Laterally the outwash grades into the valley fill, Qdj; and, except for the fact that the outwash is generally coarser-grained and includes some beds of cobble gravel, the two deposits are similar both in composition and in general appearance. They have been e, differentiated on the map for the following reason: The deposits mapped as Qwo consist entirely of glacial material; they are all clearly related to the late Wisconsin till and many can be traced directly into the Krem moraine. In contrast, Csd., cannot bedistin135

guished in many exuosures from the older fills of the Knife River valley. Moreover, although sd? consists chiefly of outwash, it contain? also seme inwarh material derived from Knife River valley wert of the late Wisconsin drift border. Therefore, all the fills in the Knife River valley have been mapped under the general heading Qsd, and have been differentiated only where their stratigraphic relations are clear. The northern edge of the Knife Hiver valley hap been chosen as the arbitrary dividing line, north of which the late Wisconsin glacio-fluvial deposits are mapped Qwo, south of which, as Qsdj. Between the Knife Hiver and the Krem moraine, outwash is found r both in the valleys and on the flatter Darts of the unlands adjacent to the moraine. Several of the larger valleys have outwash trains that can be traced from the moraine south to the Knif*3 River. Other valleys contain only small isolated denosits of the outwash. In many valleyr, the smaller ones in particular, are remnants of a cut terrace, Qt. The profile of this terrace coincides with the tops of Qwo and Qsd,, and the terrace was evidently cut while the streams were flowing at the top of the late Wisconsin outwash gravels. Nearly all the remnants of the cut bench are capped by deposits of outwashtype gravel ranging in thickness from one inch to several feet. For convenience, the map symbol, Qt, has oeen restricted to those benches whose gravel cap is less than 3 feet thick; all thicker deposits havo been mapped as Qwo or Qsd.,. North of the Krem moraine most of the deposits mapped as Owo cap remnants of the out terrace, Qt, which flank? many of th small

stream? and is graded to a fill (probably Qsdj) in the Missouri trench. Strictly speaking, there gravel? are not true outwash, for they could not have been deposited until after the ice-front had retreated north of the Yissouri River. They must be the result of the reworking of till by surface stream? durinr the recession of the ice sheet. Nevertheless, these gravel? are identical in composition, appearance, and mode of occurrence with the true outwash gravels south of the Krem moraine, and they have been included in the category Qwo. --Ice-contact deposits (Qic and Qic') In the northern and eastern parts of the Knife River area are f numerous isolated deposits of stratified drift that show the poor sorting and the steeply dipping or jumbled bedding characteristic of ice-contact denosits (see pi.22A ). Most of these deposits consist of medium-to coarse-grained sand and gravel , but a few of them are fine-'grained and consist mostly of silt and fine-drained sand. Some of these ice-contact deposits are isolated mounds or kames, Others appear to have been built between ice and the sides of small valleys and could be considered rudimentary kame terraces. Still others have very little topographic expression and would probably have been overlooked except that they are exposed in road cuts or in the cut banks of streams. One deposit, in sec. 20, T. 145N., R. 88W. , is. expressed topographically as a low sinuous ridge and this deposit is probably a small rsker. - Qic - The deposits mapped as Qio are essentially contemporaneous with the outwash (Qwo) and the valley fill Qsdg. Y.nny

of the GO deposits are associated with tho Krern moraine. The largest of these is a flat-topped kameor kame terrace at the north edge of the Beulah trench in sees. 1, 2, and 1? T. 145N., R. 88V/. A This feature was clearly formed after tho ice-front had' partly retreated from the moraine but while stagnant ice filled part of tho trench. Meltwater from the ice flowed southwest through two small valleys in sec. 1 and deposited sand and gravel around a block of ice in the northeast part of sec. 11. Later the ice melted, leaving a large closed depression or kettle-hole. PI. 23 shows the typical lithology of the kame. Qic 1 - The deposits mapped as Qic 1 appear to be slightly r younger than those mapped as Qio. They are confined to the Stanton quadrangle and are related to a late readvance of the ice across the Missouri River trench in that area. They crop out at the following localities* (l) la the southeast part of the Stanton quadrangle in sees. 22, 26, and 27, T. 144N., R. 84V/., are two nor thwe fit-trending ridges composed of very poorly sorted boulder gravel The ridges are about J- mile long, 100 to 200 yards wide, and 25 to 30 feet high. The elongate shape and bouldery nature of the deposits suggest that thoy may be small end moraines, but they oould also be interpreted as kames. These moraines or kames are set on and rise above the "40 foot terrace" that flanks the Missouri River floodplaine. As will

be shown later, the terrace is capped by post-glacial alluvium (Ojna) "but is underlain in part by the Missouri River gravels (Qmg) and lato Wisconsin till. Exposures around the ridges are poor, but seem to

indicate that the boulder gravel? lie on the upper surface of the V.missouri River gravels (Qm) and are overlapped by the alluvium (Czna). (2) In seo. 11, T. 145N., H. 84Vi., are several small kames composed of stratified coarse-grained sand and pebule gravel. There kames rise above and apparently lie on the upper surface of Qsd. It is possible that somn of the other small kames in thn Stanton £" quadrangle, such as those in sec. 1, T. 145N. f R. 8#Y>'. , may be the same age as the boulder ridges and kames just iepcribed and should therefore have been mapped as Qic'. However, in the absence of direct stratigraphic evidence either for or against this correlation f they have been arbitrarily designated as Qic. The boulder ridges and kames mapped as Oic' apparently overlie and are therefore younger than both the valley fill, Csdjj, and the Missouri River gravels, Cmg. It has already been noted (p. Ill) that a'thin till sheet occupies the same stratigranhic position as these ridges and kames. Therefore the till and the gravels of Cic* are probably of the same age and are related to a late readvance of the ice-front in the eastempart of the Knife River area. This roadvance, whose approximate limit is shown on pi. 4, vras apparently later than the readvance to the Krem moraine, for, as already noted, the deposits of the Krem moraine seem to be contemporaneous with at least the upper parts of Qwo and Qsd,. --Missouri River gravels (Qmg) In the Missouri River valley is a broad terrace, whoso top is

35 to 40 feet above the prerent floodplain of thn rivor. In the valleyr of the Knife Rivor and its tributaries the equivalent terrace is composed entirely of allnviun, but in the 40-foot terrace of the Missouri valley, alluvium is merely the unnermort deposit and lie? on the eroded surface of bodrock, till, and a seaunco of gravels unlike any other gravels in the Knife Riv<;r area. For convonienco there gravels are referred to in this report as "Missouri .Rivor ravols", but it is not intended that they be designated as a formation.or given a formal name. The Missouri River travels are restricted to the 40-foot terrace in tho Missouri trench, no similar or equivalent gravrls having been t found in any other valley in the area. The character and stratigraphic relations of the gravels are well shown in gravel nits and exposures in the face of the terrace near old Fort Clark in T. 144\'., R. 84Vn., in the face of the terrace on the east side of the Missouri in sees. 15 and 22, T. 145N. , R. 84"W., and in the excavations made

in connection with the> building of Harrison dam. The Missouri River gravels range in thickness from a few inches to about 40 feet. They consist of well-sorted, parallel-bedded sand and moderately well-rounded pebble gravel. In composition tho Missouri River gravels differ from all Pleistocene deposits of the area. Pebble counts that exclude stones derived from the local bedrock show that only 50 to 60 per cent of the pebbles are.of Canadian provenance; the rest are of western derivation. The western stones consist of red, purple and brown quartzites (probaoly derived, ultimately from the pre-Cambrian Holt formation),

fray, brown and yellow cherts, and a few fra--r.?"-~~ of mors agate that probably cams from the Yellow.ctono Riv-r cr*ina;Te basin. of the wertorn p toner ar rub-round to round ar.~ ~ay be in their roono or third cycle of sedimentation. Th" Cr.-Han r toner are mostly limestone and ranite, and ar sub-re-"- sMb-ar.pular . about thn same degree of roundness a? is four.d ~.i the outwash or lower contact of the t'i crouri Rivr rr-?el.! ir erorional. Tn most outcrop? , the p ravels lie on th sli~ht.1v channeled surface of the latp Vic C onsin till; in a few outcroo 1-, till is missing and the gravels rept directly on th*3 Fort ' Tn\or. formation. The amount

r of erosion of the till is not £reat, however, s.5 the unconforni ty bctv/een till and travel probably does not repr"se-t a rr;at lencth of time. There is, unfortunately, no outcrop that exposes the contact between the Missouri gravels and the fill, Qpd,. but presumably, the travels overlie the fill. The upper contact of the Miypouri Kiver -rivels is also erosional except in sees. 22, 26, and 27, T. 144N. , R. 54*., where the travels are overlain by the boulder ridges, Qic' . Sxocsures are poor, but in general the two cenosits anppar to be confcmable. Klcewhore, the only deposit overlying: the Missouri Kiver rrivtv> is the oost-racial alluvium Oma . Prior to the deposition of the alluvium, the surface of the travels was extensively eroded. In POTT.?" places, especially near the mouths of tributary streams, erosion mmpletely removed both the gravels and the underlying till, and it these localities the 40-foot terrace consists entirely of the Ift-er alluvial fill, ( ma .

In the suimner of 1949 the excavations in the 40-foot terrace on the wept side of the Missouri River at Garrison Dam afforded an unusually good set of exposures, showing both th character and . rtrati rar-hic- relations of the Missouri River pravols. In the socalled intake channel, just north of the edge of the Stanton quadran/.lo, the unconforraitiep between the travel and the overlying alluvium and between the gravel and the underlying till were well displayed (pee pi. 24 ). In two places in the intake channel the Missouri River rraveIs are separated from the underlying till by thin discontinuous deposits of dark gray lacustrine clay and mediumgrained yellow sand, deposits not exposed elsewhere in the area. It has already been stated that probably the Missouri trench* was filled with stagnant ice during the recession of the late Wisconsin glacier. As the stagnant ioe melted, it split into blocks between which were small ponds connected by winding streams. The lacustrine clays and yellow sands were probably deposited in those ponds and streams. The critical features of the Missouri River gravels can be summarized as follows: The large numbers of pebbles of western provenance indicate that while these gravels were being deposited the Missouri River was flowing freely, unimpeded by ice or glacial debris. However, the large numbers of glacially derived ptones indicate that the ice front had not retreated very far to the northeast and was still supplying outwash to the Missouri River. Therefore, the Missouri River gravels were deposited during the recession of the late Visconsin ice, just

of thf rarr.e township arn outwash. --ruvel on older cut terracfr ( r tr)-- In thp western and northwestern part? of the rbncho Quadrangle arr- remnants of a cut or strath terrace capped by deposits of fluvial sand and gravel. In '1V142N., K. 91..., the bench remnant? with their cappings of i'' gravel range from 60 to 100 feet above the present floodolain of the

Knife Kiver. Probably npt all arc remnants of the same strath but nevertheless all appear to oe of the same general ape. In the northwest corner of the Proncho Quadrangle another group of gravel deposits manped as Qtg cap remnants of a cut terrace flank- / ing a small tributary of Spring Creek. The gradient of the terrace i? much gentler than that of the present-day stream, so that the two profiles, which are nearly coincident near the headwaters of the of the stream in sec. 5, T. 143N., ft. 91W. t diverge rapidly downstream and in seer. 34 and 35, T. 144N., '.i. 91V*. , the deposits of Ots are 40 to 45 feet above the prerent valley floor. In contrast with the older glacial gravels, which contain 70 , to 90 percent pebbles of Canadian derivation, the gravels of Qtg are 50 to 60 per cent of local origin (see fig. 6 ). Thus, the gravels of Qtg are similar lithologically to the gravel beds in the intermediate fill (Qsdg) in the Knife Rivrr valley estintj; that the two deposits may be related. It ir possible that streams flowing on the top of the fill Qsdj> planed across the bedrock and nubs-pquently deposited additional fravel on top of this bnnh. Under this hypothesis th cut terrace could have ben formed after

the valleys had been cut to their present depth. It is also possible that the bench remnant? are relics of an old erosion surface that antedates the dissection of the valleys and that the fill Qsdo rone and capped thse remnants with sand and gravel. The field evidence is consistent with either hypothesis. Older -sand and gravel undi fferentiated (Qs~)-- In contrast -with the glacial deoositr of Qwg and the fluvial deposits of Otg the denosits ranped as Osg are indeterminate in origin as well a? in correlation. They are so few in number and so small in size that no statistical count war made of thoir pebbles, but a few rough estimates were made and indicate that r.rmo of thorn contain 60 to 70 por cent stones of glacial origin while Others contain 60 to 70 per cent atones of local derivation. Thus the map heading Qjl£ s probably a scrap-basket group that includes /ravels of both glacial and fluvial origin. The glacial deposits may be remnants of kames or outwash; the fluvial deposits may be remnants of gravels graded to Qsd£ in the Knife River valley. The deposit? in sec. 32, T. 142N.. R. 89\S,, and in sec. 5, T. 141N., R. 89'A., are very likely of glacial origin as they contain a large proportion of granite and limestone oebbls and cobbles, some of whir-h aprmar to preserve glacial soles and a few of which seem to have faint remnants of Rtriation?. Cut terraces (Qt and Qlt) Flankin;- the valleys of Knife Hiver and many of thr rmallor rtreams are remnants of stream-cut benches at various altitudes above the modorn floodplains. The highest of the there benches or

strath terraces apparently fall on a single profile and have ben mapped as the terrace, Qt. Th lower benches do not fit any one profile, but have all been grouped in the general heading, Qlt. Although a? a general rule erosional features do not belong; in tho discussion of stratigraphy, these terraces are FO intirnatoly associated with the Pleistocene deposits of the area that they will bo discussed at this time. The higher cut terrace, Ot, IF well developed in the northern and eastern parts of the Knife River area. It is absent or at least cannot be identified in the southwestern part of tho area beyond the late Wisconsin drift border. Tho terrace is cut on both brdrcck and t late Wisconsin till. The downstream slope of tho terrace profile is less than the gradient of the modern floodplains. Therefore, traced upstream tho profiles of the terrace and the floodplain converge; traced downstream they diverge. Thus Ot is 25 to 30 feet above Spring Creek near Golden Valley, 50 to 60 feet above the Knife River

at Beulah, and 100 to L10 feet above the Knife and Missouri rivers at Stanton. Ae stated in the descriptions of the outwash, Qwo, and the fill, Qsdg, the long profile of Qt coincides with the tops of the higher remnants of these glacio-fluvial deposits. Also, nearly all the bench remnants are capped with at least a skim of these deposits. It is therefore difficult in many localities to tell where Qsdg or Qwo stops and Qt begins. For convenience in mapping, Qt has been limited to thore bench remnants capped with less than 3 feet of gravel, and all' thicker gravel caps have boen mapprd as Qwo or Qrdv.

it The coincidence in the. profiler of Qt and the tons of Owo arc Qsd, suggests a genetic connection between the terrace ar.r! there glacial deoositr. Anparently, aftr having doo*-! ted Owo and dj the streams flowed on thp tops of these fills long enough to cut laterally into thf valley walls and develop a strath across both till and bedrock. Potv.-een the higher cut terrace Qt arid tho modern floodnlain of the Knife River are remnants of oth'-r.Fbrath terrace r. that have nn mapped a- "lower cut terracr" (Qlt). In most of the Knife rtiver area those lower surfaces do not fall alcnr any definite profile and they probably represent non-paired, strath terraces, cut r while the Knife River was dissecting the late Wisconsin fill, Qsds. because of tho upstream convergence of the terrace profiler and the presfnt stream profiler, it was not always possible to distinguish the higher cut terrace, Qt, from the lower cut terraces, Olt, in the Golden Valley quadrangle. Therefore strath terraces in this quadrangle were mapped as Qt, although ome of them are probably correlative to tho lower cut terracor.

"

Summary and correlation of Pleistocene deposits The Knife River area is partly covered by three ages of glacial deposits, whose relations can be summarized as follows: (1) The tills are lithologically similar and cannot be differentiated except where they are interbedded with stratifiea deposits. (2) "In the valleys of the Knife and Missouri rivor." are 3 fills (Qsd). The base of each fill is belov/ tho modern floodplain of the

streams, indicating that the valleys have been cut at least to the

depths of their present floor? before any of there fill? were deposited. The firrt fill (Qsd., ) is crossbr.dded outv/ash sand and gravel. There, are no exposures showing it in contact with an underlying till, but 'it? character and str& tif.raohic relations in-

dicate that it" must reco'rd a plarial advance. The second fill (Qsdp) is comparatively fi no-rrai nd , consistinp mortly of silt and fine-to medi un-grai ned sand derived from the local bedrock. Tt is interpreted as a f luvic-lacustr i ne inwush fill, laid down when a lobe of ice blocked the Vissouri Hiver downstream from the mouth of the Knif River. This fil] overlies and is closely associated with a till sheet. The third fill (Csd) consists mostly of racial outwash. In most of the Knife River valle.y thir fill v.as "wall-to-wall", but in the Missouri River trench it was a Icane terrncc. This fill immediately overlies the third or youngest major till sheet, and both till and fill are confined to tho northern and eastern parts of the Knife Kiver area. (?) The three fills ar separated by pronounced unconformities. The interval between th" first and second fills was characterized by a fairly small amount of erosion and wealc soil-development. Tho interval between the second and third fill? was characterized by more ,ext°nsive erosion , which dissected the thick fill, Osd ? and removed the bulk of the two older till (<l) A fourth fill (On;-) i c restricted to the Missouri River trench, and- consist** partly of outwash and partly of stones derived from the wet. It anpoars to hav beon doosited arinst the sice*- of the kame terrace of Osdv sliortly after the stagnant ice hac . . v r. from the trench, and it doer not record another facial episode.

(5) A fourth till with associated ice-contact deposits is present in the extreno eastern part of the area. There deposits ovrli loth the third fill (Csdg) and the Missouri ?<iver travels (Qrn.), out tru-rr is no unconformity at their bare and they probally rpr*'rent merely n readvance of the ic thr t deporited the third till and the third fill (Qsd 3 ). (o) Other fluvial and rlacio-fluvial dT>esits in the Knife reiver area are the lateral ar.r tine enuivalentr of the three ir.aier fill? Qsd-j , Qsd2 and Q?<?.v* Thr-> dcnorit-- of ice-contact stratified drift (Qic) a no outv;arh (Qv/r) in the northern Tiart of the area arc easily phovm to be th tin enuivalent of the fill, Th r irolatr-d rmnantr of fluvial and plaeio-fluvi al denorit? in the in th'1 pouthwortrrn part of t.ho Knife River area (Qtg, Q\vf, and Qr) cannot be correlated definitely but are pmbahly the em:ival*'nt of the intermediate fill Ord 0 . All three generations of Pleistocene dfir?onit in the Knife River area are believed to be of Wisconsin a/.-.e. Qsd and Q?dg and their associated deposits are early Vkis.ccn?in, lovmn and Tazewell respectively. O.sdr, is late Msconrin and is probably I.'ankato. These correlations are based partly on a comoarison of the stratifraphic sequence of thn Knife ttiver deposits vn th the nonufmce of dated deposits in South Dakota, Iowa and Nebraska, and partly on the tracing of Wisconsin drift borders from the North Dakota-South Dakota State line north to tho Knife Kivr. The field evidence and hhe rearoriinr can be summarized as follov;s:

(l) Crandell (1951, pp. 148-149) and Warren (1949, p.1926) have independently concluded that in outh Dakota the Missouri :<iver trench was established during the lllinoian glacial stage. They alro agree, a? does Flint (1951, manuscript in preparation) that the older't drift in the trench is Lovan ana that the base of this drift in close to the present floor of the valley. Thus, Csti]_ occupies a topographic position similar to that of the lowan drift in oouth Dakota. (2} The outermost drift border in North Dakota has been traced from just west of the Knife River area south to the South Dakota State line, where it joins the lowan drift border as mapped by Flint (1951., op.cit.). Although this drift has been dissected and largely removed, its remnants are conformable with the present topography (i.e. they post-date the dissection that followed the cutting of the Missouri River trench) and arc found in the bottoms of some of the valleys. Thus drift of lov/an age is present in North Dakota, and has the same topographic and stratigraphic position as Osd in the Knife Hivor area. (3) Until a few years ago it was generally assumed that the iowan-Tazev.-ell interval was the longest of the Wisconsin interglacial substages. Hecent work in touth Dakota, Nebraska, and Kansas, however, has shown that this interval wus relatively short and was characterized by minor erosion and weak soil formation (Leonard, / ' 1950, p.1481; schultz et al., 1951, p.7; Frye, 1951, p.406; and Flint, 1951, manuscript in preoarafcion). This fits the drscription of tho interval betwrn Osd-, and Qsc n ne Knifp Kiver area. The

same recent work .ha? shown that the Tazev/ell-Cary interval v/as relatively long and was characterized by more extensive erosion and strong -soil-development (the Brady soil-forming interval in Kansas). This fits the description of the interval between Osdg nnd Qsoig in the Knife lUver area. Therefore, Csdg and its associated till sheet are probably of Tazowell age. (4) Mint (1951, op.cit.) has found that in north-central South Dakota the border of the Gary drift is crossed ana overlapped by the border of the younger iVanxato drift. in southern North Dakota, this I.'ankato drift border ir ciff'cult to follow, out can be traced in a general way from the South Dakota State lino northwest to the Knife t River area, where it coincides with the limit of the third till and the valley fill, Qpd,. I have found no evidence of any major drift border northeast of this one, either in the Knife Hiver or anywhere elre in south-central North Dakota. Certainly there is no evidence, either stratiraphic or topographic to support the old idea that the Mankato drift border is marked by the "Altamont" (now the Wax) moraine It seems probable, therefore, that in southern North Dakota the Gary drift border has been overlapped and obscured by the Mankato drift and that all the glacial deposits associated with QscU in the Knife

Rivor area should be assigned to the f-'ankato substage.

RECENT SERIES As used in this report, the term "Hocent" includes all post- Mankato deposits. These deposits include alluvium, eolian sand, landslide blocks, and deposits of residual silica. Thick deposits of colluvium cover some of the slopes, eroecially the sides of some of the larger valleys, but these have not been differentiated on the map. The landslide blocks are discussed under "Geomorphology" and arc described in this section. A Alluvium (Qma and Qal) Post-glacial stream deposits oover the floors of all the valleys in the Knife River area. Theso deposits consist mostly of tan to gray fine-grained sand and silt with a few beds of coarse-'grained sand and a very few lenses of oebble to cobble gravel. The alluvium in the valley of the Missouri River is all fine-grained, the coarse sand and gravel being restricted to the valleys of the Knife River, Spring Creek and a few of their tributaries. Some beds of the alluvium are well sorted and consist of thinly-bedded silt or fine sand; others are very poorly sorted and consist of a mixture of particles ranging in size from clay to fine- or medium-grained eand. Two ages of alluvium can be differentiated locally in tho Knife River area. The younger of these is represented by the modern floodplains of the streams; the older, by a terrace flanking the floodplains. In all the valleys except that of the Missouri River this terrace is composed entirely of alluvium. In the Missouri River valley, however, the olr'er alluvium is merely the "caprock" of the terrace and In unconformably underlain by several Pleistocene de152

posits, notably the Missouri River gravels (Qmg). Therefore, because it is both lithologically distinctive and topographically promimmt., 'the terrace in the Missouri River valley has been mapped as separate geologic units, inhere the top of the terrace is underlain by the Missouri Kiver gravels covered by less than 3 feet of alluvium, th'e terrace has been mapped as Qmg; where the alluvial cover is more than 3 feet thick, it has been mapped as the "alluvial terrace", Qma,(fip;. 10). Elsewhere in the area the terrace has neither the topographic prominence nor the distinctive composition that it has in the Missouri River trench, and in many oart? of the valleys it is im-

possible or impractical to map the boundary between the two ages of alluvium. The reasons are as follows: (l) The profile of the top of the terrace has a-gentler gradient than the modern floodplain, so that traced upstream the terrace and floodplain gradually converge. Thus the terrace, which is 35 to 40 feet above the Missouri River floodplain, is 15 to 20 feet above the Knife River floodplain in the Broncho and Medicine Butte quadrangles, is 5 to 10 feet above the f loodplains of streams like Willow Creek and Elm Creek in the middle parts of their courses, and is coincident with the modern floodplains in the headwater areas. (2) In several of the through valleys or trenches that contain no large streams, the alluvial floor, which is obviously being augmented by current slopewash, is not dissected and is apparently graded to the alluvial terrace rather than the modern floodplain of the Knife River. (3) Even in valleys like those of Willow Creek and Elm Creek where the terrace

of older alluvium is readily distinguished, the scale of the mapping makes it impractical to separate the small fragments of the terrace from the modern floodplain. Therefore, outside the Missouri River trenoh both ages of alluvium have been napped as Qal, In the valleys of tho Knifo River and Spring Creek, where the remnants of the older alluvium are topographically distinct and large enough to show on tho map, they have been separated from the modern floodplain by a hachured line. In the smaller valleys even this differentiation was not attempted. Eolian sand and silt (Qds) / The Great Plains is a country of high winds and little rain and the effectiveness of wind action is apparent to anyone who has spent much time in the area. It is not surprising therefore that most of the Knife River area has a thin veneer of wind-blown material, This is especially true of those parts of the area that are southeast of the major stream valleys, where there is an abundant source of loose, fine-grained material easily picked up by the winds. On the geologic maps (pi. l) the eolian material has been shown only where it is more than 5 feet thick or where it has dune topography. Nearly all of the deposits mapped as Qds consist of fine- to medium-grained sand and their surfaces have typical dune topography. A few consist of very fine-grained sand and silt (loess) and havo smooth upper surfaces that show no signs of dune topography. These loess deposits are minor in comparison with the deposits of dune eand and are largely restricted to the bluffs overlooking the Missouri River.

In most deposits the dune sand is fine- to coarse-grained with minor amounts of very coarse sand, granules, and a few pebbles. Pebbles up to one inch in diameter are ouite common in many of the thinner eclian deposits. Some of these pebble-bearing dunes are downslo'pe from exposures of till or gravel, and the pebbles evidently rolled or crept downhill while the dunes were being formed. Other pebbles, however, are found on top of some of the dunes on the terrace near Stanton where there is no hill from which these pebbles could have come. The only logical conclusion is that the wind must have rolled thorn onto the surface of the dune. In color, the dune sand ranges from lifht yellow to dark brown depending on the amount of included carbonaceous material. In general'the thin deposits have more humified matter and are darker than the thick deposits. The largest areas of dune sand are in the Stanton and

Hazen Quadrangles and are southeast of large remnants of the Pleistocene fills Qsd. The largest individual dunes are in the southeast part of the rfazen quadrangle in seas. 23, 24, 25, and 26, T. 144N. f R. 86V*. These high dunes form the southern end of a largo deposit of eolian sand that caps the terrace of Qsd and extends south across the till and bedrock of the valley wall of the Knife River. Tho high dunes have a local relief of 40 to 50 feet and are more than 100 feot higher than the terrace to the north. Undoubtedly they have a bedrock core, but no exposures of this core could be found in the blowout areas. These hiph dunes are grassed over, and stabilized except in the eastern parts of pecs. 24 and 25, where about 3/4 of a square mile is underlain by actively shifting sand that ie migrating; southeast and encroaching on the valley of Kinne155

man Creek (called Sand Creek by many of the local residents). Two line? of evidence suggest that most of the dunes v;ere deposited by northwert winds. First, as already mentioned, most of the largo patches of dune sand lie southeast of arear? underlain by tho Pleistocene sand fills, Qsd. Second, aerial photographs show that the dune topography has a definite northwest-southeast grain or alignment. Today, although strong winds blow from nearly every quarter, weather records indicate that the strongest storm winds are still from the northwest, and the dunes would still be actively forming were not most of the loose sand grassed over. Most of the dunes probably were formed during the "climatic optimum", about 4,000 to 6,000 years ago, when the lima'te of central United States was warmer and probably drier than it is today. However, although this date is plausible, it is difficult to prove. That the dunes are poet-Pleistocene is indicated by the fact that they overlie all the glacial deposits in tho area. That the dunes are not a product of the modern climate is shown by the fact that most of the dunes are grassed-over and stabilized. Also, on the surfaces of some of the dunes in sees. 12 and 13, T. 145'!., R. 84W., I found bits of pottery and other Indian artifacts, indicating that the dunes had been stabilized prior to that particular occupation. Therefore, the stratigraphy indicates only that the dunes are post- Pleistooene and pre-modern climate, but the warm, dry climate of the "climatic optimum" would have been a favorable time for eolian action and it is reasonable to suppose that most of the dunes were made at that time. 1F6

Residual silica depos its (Qrs) In the, southern par"t of the 1'edicine Putte quadrangle several of the hip-her hill? aro caoped by a rubble composed of broken blocks of silicifie.d sandstone and shale. Similar rubble car>s hills .iut south of tho ouadranglr. There silica blocks v/ore probably derived from a bed in the lower part of the Oolden Valley formation and have been let down to their present position by weathering and erosion. Thepf> deposit." have been differentiated on the map becaure they are a potential source of riprap for facing the upstream sides of dams.

& GEOMORPHYLOGY AND GLACIAL GEOLOGY OF SOUTHWESTERN N03TH DAKOTA Prior to 1946 the late Tertiary and Quaternary history of southwestern North Dakota had received little attention and was vory poorly known. "What little war known was bared on the good though sketchy reconnaissance work of Leonard (1912, 1916a, 1916b) and Todd (1914, 1923). Therefore, in order to provide a background for the study of Knife River area I spent a part of each summer making a reconnaissance study of the glacial drift and the land forms between the Knife River area and the North Dakota-South Dakota State line. The results of this reconnairpance are summarized in this section of the report. , In this report, southwestern North Dakota is defined as that part of the State that lies west of the Missouri River and south of the latitude of the mouth of the Little Missouri River. North of this latitude the geomorphology and glacial geology are being studied by A. D. Howard, of the Geological Survey.

Land forms The higher parts of the interstream divides in southwestern North Dakota consist of a broad gently-rolling upland above which stand a few buttes capped by the Oligocene White River formation. Surrounding at least one group of these buttes are remnants of a planation surface capped with several feet of coarse gravel. This gravel-capped surface appears to be slightly higher than, and therefore probably older than, the broad upland. Near the edges of the broad upland the surface slopes gently down toward the valleys of all the major streams except that of the Missouri River. These gentle slopes outline broad

valleys whose floors have been entrenched to form the raorf1 narrow comparatively etoep sided valley? that today contain th major streams. On either side of these streams is a dissected bolt composed partly of badlands and partly of more gentle grassed over plopes. Thepr* dissected areas have been carved from the gently sloping surfaces of the old broad valley profilrp. The belt of badlands bordering the Little Missouri River is much larger and much more spectacular than that adjacent to any other stream. Thip badland area hnr a special history and will be described later. Although most of these land forms are the result of erosion during the late Pleistocene and Recent epochs, some of thm are relicts of the Tertiary Period. The high buttes capped by the White River formation must have had their beginnings during the period of orosion that followed the deposition of those sediment?. Thus the butte tops themselves represent a surface of deposition inherited from the Oligocene pooh and the dissection of the butte slopes started either in late Oligocene or in early Pliocene, as soon as the streams had out below the base of the White River sediments. The gravel-capped planation surface that borders Coffin Buttes in southwestern Grant County has already been described on p. 100. Foth in lithology and in mode of occurrence, the gravels resemble the F]axville gravels of eastern Montana and they are probably of nearly the same age. Inasmuch as the Flaxville gravels contain vertebrate fossils, dated as late Miocene to middle Pliocene, the gravels near Coffin Puttes have tentatively been classified as Miocene (?) or Pliocene ( 7 The broad uplands are'apparently slightly younger than the plana159

tion surface gravel?. If these gravels are actually the equivalent of tho Flaxville gravels, then the broad upland was formed either in late Pliocene or early Pleistocene time. In addition to the land forms just described there are a number of through valleys or trenches that cut across the divides between the major streams. All these trenches are in the glaciated area; most of them contain small streams or no streams at all with the exception of the trench that extends from Garrison Dam south to tho State line. This trench contains the master stream of the area, the Missouri River. The origin and history of the Missouri River trenoh is ono of the key factors ""in the Ouaternary history of western North Dakota and must be understood before some of the other features can be appreciated. Origin of the Mi ssouri River trench General setting The major streams in North Dakota are shown on thesketoh map in fig. 3, It can be seen at a glance that the drainage pattern southwest of tho Missouri River bears no resemblence to the pattern northeact of that river. Southwest of the Missouri River the streams flow north and east and would appear to be part of a master drainage flowing northeast into Canada except that they have been beheaded by the Missouri River. Northeast of the Missouri the streams wander and the general pattern is hap-hazard. All of these streams in eastern North Dakota have been deranged by the deposits of the ico pheetr. Some parts of their courses seem to be remnants of pre-glacial channels while other part? are clearly now and have been established since

tho glacial drift was laid dovm. Just east of the Missouri River, the Coteau du Missouri is crossed by several large valleys or trenches that today contain no Ciajor streams. These valleys are nearly collinear with the valleys of the western tributaries of the Missouri Kiver and probably represent the former courses of these rivers. The floors of the abandoned valleys hang 125 to 175 feet above the present Missouri River and therefore seem to be extensions of the broad valley profiles of the western streams. From the Garrison Dam south to the State line the Missouri River trenoh bears little resemblence either to its tributaries from the west or to the streams east of the Coteau. The broad valley profile r that characterizes the valleys of its western tributaries is lacking in the Missouri River trench, which is incised sharply into the upland surface of the Great Plains. In pome places the upland surface slopes toward the trenoh for several miles on either side; but in many other places the upland actually slopes away from the trenoh. Evidently the Missouri River was not established in southern North Dakota until after the broad valley profiles of the other streams .had been out. The anomalous relation between the Missouri River trench and the rest of the topography in southern North Dakota is identical to the features in South Dakota described by Flint (1949) suggesting that in both states the Missouri River has had a similar history. Hypothesis of glacial diversion That the present drainage pattern in North and South Dakota

**®Rulted from the diversion of streams from their fonr.er courses by

the action of a glacier was first suggested by 0, K. Y/arrun. Later this idea wos adopted and explored by J. S. Todd, state geologirt of South Dakota. Todd's ideas on the glacial geology of both North and South Dakota has been published in several journals and his conclusions about the Missouri River were summarized in a short paper in 1917. Briefly Todd concluded that at some time in tho past the streams in western North and South Dakota had continued east across the site of-the present.Missouri River trench. The streams in North Dakota and northern South Dakota were part of a master system that drained into Hudson's ay; the streams in central and southern South Dakota turned south through the broad valley of the James River and f entered what is now the Missouri River in southern South Dakota. The advance of the glacier blocked the downstream segments of these old valleys, and the streams were forced to turn southeast along the margin of the ice. By the time the ice had retreated this new course was so deeply entrenched that the old valleys were not reocoupied. This new southeast-trend ing valley was further deepened and is tho present Missouri River trench. Since the work of Todd no one except Leonard (l916a) has seriously questioned the idea that the Missouri River tronch is of glacial I, origin. Recently Flint (1949.) studied the origin, of the Missouri River in more detail and has tracrd out the former drainage courses of the rivers in South Dakota. Concerning the method of the diversion Flint stated (p. 69):

"The James lobe of one of the Pleistocene ice sheets entered eastern South Dakota from the ::ortheapt, as inferred from the provenance of erratic boulders. Its average thickness is believed to have been at least several hundred feet, whoreas the general relief of the country invaded is inferred to have been little more than 200 feet. Henco the ice completely buried even high points on the Missouri Plateau. Flowing south-westward, this ice lobe blocked all

e.lemons of the ancient east-west drainage as far west as

the sight of the present Missouri River. Because of the ' ice blockade, the steepest remaining component of the former eastward elope of the land became a slope toward the southeast. The streams became ponded; the main valleys filled v.ith water that gradually backed up into the tributaries. In time each pond rose high enough to spill over the lowest part of the interfluve that separated it from the major valleys southeast of it. Most of these low points ocourred between the heads of two opposed tributaries. Eaoh temporary spillway thus formed became intrenched by the overflow. In effect the water flowed up one former tributary and down another opposed tributary. As the glaoial blockade was shrinking it failed to olear the ancient valleys until after the temporary diversion routes across the interfluves had become so deeply intrenched that the diverted waters were unable to return to their former paths. The floors of tho ancient valleys, partly filled with drift, were, left standing somewhat above the new profile of the diverted water-the initial profile of the present Missouri River." Thus, according to Flint's idea the Missouri River trench is a patchwork, formed from segments of small valleys that already existed. In plaoes where there was no pre-existing lower outlet short segments of the trench were probably carved into the uplands by an ice marginal stream, but by and large the diverted waters followed the already out valleys of the small tributary streams. In southern North Dakota as far upstream as Garrison Dam the Missouri River trench is similar in every way to the trench in South Dakota, and probably originated in the manner outlined by Flint. Between Garrison Dam and the mouth of the Little Missouri River, however,

the Missouri flows eastward through a valley whose sides slope gently tov/ard the river and appear to outline the broad valley profile characteristic of the pre-diversion valXeys in North and South Dakota. In this part of its course, therefore, the Missouri River

is apparently in one of the old valleys that was part of the former Hudson Bay drainage system. Date of the diversion Flint (1949, p.7l) pointed out that early Wisconsin (lowan) drift was deposited in the Mipsouri River trenoh after it had been out to about it* present depth. This indicated only that the diversion of the drainage and the deepening of the trenoh must have taken place in some pre-VVisoonsin time. ' More recently, Warren (1949) and Crandell (1951, pp.148-149) have independently discovered evidence suggesting that the drainage was diverted to its present pattern during the Illinoian glacial stage. Their evidence can be briefly summarized as follows: (1) Early Wisconsin (lowan and Tazewell) drift occurs down in the Missouri trench well below the profile of the pre-diversion drainage. This agrees with Flint's conclusion that the trench must be pre-V.risoonsin. r i v (2) On the upl'ands wetr of the Missouri trench are deposits of coarse sand and gravel containing vertebrate fossils of probable late Kansan age. These gravels are topographically above and are therefore older than the broad valley profiles of the Cheyenne and Bad rivers; and, as has already been stated the broad valley profiles °f all the streams antedate the cutting of the Missouri River trenoh. -

(3) The establishment of the Missouri River trenoh, therefore, is post-Kansan and pre-V/iscons in; and, by a process of elimination, we are left with Illinoian as the only glacial stage when this oould havo taken place. Following the same process of elimination the cutting of the broad valley profiles was post-Kansan, pre-Illinoian and therefore must have taken place during the Yarmouth interglaoial stage; and the deepening of the Missouri trench to about its present depth must have taken place during the Sangamon inter glacial stage. ' In southern North Dakota the evidence is not conclusive but seems to indicate that the Missouri River trench was formed at about the same time as in South Dakota. This evidence is as follows: (1) Drift of lowan age occurs down near the bottoms of the present valleys, well below the pre-diversion Fttream profiles. Thus the trenoh must have originated prior to the Wisconsin stage. (2) Just ae in South Dakota, the cutting of the broad valley profiles of the eastward-flowing streams antedates the formation of the Missouri River trenoh. Therefore the trench in both North and South

Dakota seems Uo have been out at the same stage in the geomorphic evolution of the landscape. It seems reasonable to assume that the broad valley profiles in both states were cut at the same time and, consequently that the Missouri River trench war, also cut simultaneously on both sides of the North Dakota-South Dakota State line. Therefore, if the Illinoian age of the trenoh is valid in South Dakota, probably it applies to North Dakota as well. Pre-diversion drainage of North Dakota Although it has been known for a number of years that the present -

eastward-flowing tributaries of the Missouri River in North Dakota once crossed tbo site of the Missouri trenoh and continued east and northeast into Canada, no detailed tracing of the abandoned courses east of the Missouri River has ever been attempted. Todd had suggested the approximate positions of SCIHD of the pre-diversion valleys but most of Todd 1 a work was in South Dakota and he did little or no actual tracing of the abandoned valleys north of the State line. My own reconnaissance east of the Missouri River in southern North Dakota was too brief to permit tracing of the abandoned channels in the detail comparable to Flint 1 s work in South Dakota. Nevertheless, I was able to locate the approximate positions of many of the larger abandoned valleys and from them reconstruct some of the pre-diversion drainage pattern of North Dakota. This reconstructed drainage is shown in fig. 12. The individual parts of this abandoned drainage will now be briefly described; the numbers in the following discussion correspond to those on the map. ! Former drainage in southern Emmons County. - In southern Emmons County is an abandoned valley that extends from'the tov/n of Linton south to the North Dakota-South Dakota State line. This valley is the northward extention of Flint's Mound City trench (1949, p.6l) and seems to be a part of a master drainage that included the present Grand, Moreau, and Cheyenne rivers. Beaver Creek, which now flows west past Linton to the Missouri River, seems to have been a tributary to this northern part of the Mound City trenoh. 2, Long Lake trench (the former course of the Cannonball River).- The present-day Cannonball River enters the Missouri River about 30

miles north of the North Dakota-South Dakota State line at the junctions of Norton, Sioux and Emmon? counties. Four miles north of the mouth of the Cannonball River is a large abandoned valley or trenoh that leads northeast from the Missouri trench through northwest Emmons County across southeast Burleigh County and into Kidder County, where it is buried and obscured by the moraines of late Wisconsin drift, at the Burleigh-Kidder County line this trench is occupied by Long Lake and wilJL be referred to in this report as the Long Lake trenoh, Todd suggested long ago that the Long Lake trenoh represents the former course of the Cannonball River (1914, p.266). Leonard (1916, p.296) objected to this idea on the ground that the mouth of the trenoh is 4 miles north of the present mouth of the Cannonball River. The method that Flint outlines for the formation of the Missouri River trenoh, however, negates this objection. The 4-mile segment of the Missouri River trenoh between the mouths of the Cannonball River and the Long Lake trenoh,is a segment of the pre- % diversion Cannonboll River, with the present-day drainage flowing opposite to the direction of the former drainage. East of Long Lake the trench has been partially to completely buried by thick glacial drift and is difficult to follow. Apparently it continues northeast finally leaving the Coteau du Missouri at a low point in the Missouri escarpment at the railroad siding of Goldwin in Stutsman County. Lake Etta and Lake Isabel in Kidder County and Chase Lake in Stutsman County seem to be on the old drainage course. 3. Abandoned course of the Heart River - From its headwaters in southeastern Billings County the Heart River flown east and southoapt

acrorc tark and Grant countier and enters Norton County in the south'.v<Drt part of T. 13GI-., K, 84/>. Here the river turn? abruptly north by northeast for about 15 milrs to T. 138N., R. 38V.., v.'here it again resumes a more easterly route. Th*3 val ley of the Heart above and belov.- this 15-mile segment i? typical of the pre-diverrion valley? in North Dakota. The uplands slope gently tov.'ard the river for several miles and the river itrelf is cut into the old broad valley wrose fleer was 125 to 150 fert above the modern flocdplain. In the 15-mile segment just mentioned, however, the Heart flow? in a youthful vallry whose floor is 400 to 500 feet below the edge of the uolandp. The broad vrflley profile is abrent and the upland does not slope toward the river; rather it ends abruptly in a tract of badlands, 1 to 2 miles wide, that separates them from the valley floor. Leading southeast from the sharp bend and collinear with the upstream portion of the Heart Kivcr is a wide trfmch that crosses from the drainage of the Heart to the drainage of the Cnnnonball Rivor, joining the latter stream noar Preien. The upland slopes toward thir trnnch, and th topofranhy of the tronch walls is much more rubdupd than the badlands bordering the "15-mile pegment". The floor of the trench ip belov; the altitude of the broad valley profile, and the trench itself is obviously a part of the Pleistocene drainage system that extend? all the way from the little *-'missouri Kiver in Dunn County to the Cannonball Hivor in ioux County (pi. 3 ). However, I suggest that the trench also represents the former (prr- 113,iaoian courro of- the Heart River and that tho

Heart and Cannonball rivers once wore confluent near Rreien. If this hypothesis is correct, the nature part of the Heart kiver valley below tho youthful 15-mile segment was originally cut by Sweetbriar Creek and one of its tributaries. Then, for reasons that are not clear, this tributary worked its way headward and captured the Heart Hivor. Perhaps ice dammed the lower part of Sweetbriar Creek, causing ponding and an eventual overflow to tho south and southwest along the present course of the Heart River. The overflowing waters may have cut the spillway notch deep enough to permit the tributary of Sweetbriar Crock to complete the job and capture the Heart River. Whatever the cause, this capture clearly 'occurred after the streams had cut below their old broad valley profiles, and probably after the establishment of the Missouri River and entrenchment of that river to about its present altitude. 4. Brittin. Trench. A few miles southeast of Bismarck is a broad flat terrace about 100 feet above the Missouri River. This terrace can be traced east into a large trench about 2 miles wide, that trends east across southern Burleigh County and joins Long Lake trench at the town of Moffit. The railroad siding known as Brittin is about in the center of this trench, which will therefore be called the Brittin trench. According to my reconstruction of the old drainage, the Brittin trench is the eastward continuation of an old valley that carried the combined waters of Sweetbriar, Square Butte, Burnt, and possibly Apple creeks. One discrepant feature of the reconstruction is that the Brittin trench is nearly twice as wide as the western part of

Long Lake trench; yet Lonr Lake trenoh was nrrumably carrying the combined Heart and Cannonba! 1 river?. Either the reconstruction is incorrect or there was sorra extraneous factor that cuasod the Brittin trench to be abnormally wide. 5. Apple Creek valley. North of the Brittin trench and subparallel with it is Apple Creek, a southwest-flowing tributary to the ?missouri River. The mouth of Apple Creek is nearly opposite the mouth of the Heart River at Mandan; and Todd (1914, p. 266) postulated that Apple Creek valley represents the pre-diversion course of the Heart River. Todd was undoubtedly influenced by Leonard's report that deep wells in this valley passed through about 200 feet of glacial silt and sand before reaching bedrock (1912, p.44). / However, this 200-foot thickness of outwash puts the bedrock floor at much too low an altitude to have been a part of the pre-diversion drainage system. The deep cutting and subsequent filling must have taken place after the establishment and deepening of the Missouri River trench. It is, of course, possible that some higher profile of Apple Creek valley was the extension of a river that combined Sweetbriar, Square Eutte, and Burnt creeks (Todd's pre-diversion "Hear't River"), but this hypothesis leaves no major stream for the Brittin trench. It seems more probable that Apple Creek was just another tributary to the Brittin trench drainage. Whether it flowV' ed west in its present valley as far as Bisnaron or whether it turned southeast at MoKenzie and joined the Brittin trenoh near Moffit is not clear from the present topograohy. 6. Former courses of the Knife and "Missouri" rivers. Todd

(1914, p.266) recognized that the east-west segment of the J/.iseouri River trench upstream frorn Garrison Dam was a part of the pre-divorsion drainage system. Ho inferred also that the Knife River once turned at Staaton, and flowed north along tho site of the present Missouri iiiver trench to join the old "Missouri" at the site of Garrison Darn (Old KorfStevenson). The combined drainage of the two streams then continued northeast through the valley of Snake Creek, across the Coteau du Missouri, and joined the Souris River somewhere east of J-'inot. Andrews (1939, p.62) outlined the course across the Coteau in more detail. He suggested that the ancestral "Missouri" Rier flowed through the Snake Creek outlet, joined what he calls the Turtle Creek spillway of glacial Lake Souris, and continued northeast through the spillway to join the Souri? River near Velva. My reconnaissance study, aided by the many newly-published topographic maps of the area, indicates the following ir.odif ications of the courses proposed by Todd and Andrews: (a) The Missouri River trench between Garrison !>am and the mouth of the Knife River shows no evidence of having been a part of the ancient drainage. (b) East of the Missouri River and in line with Knife Rivor valley is a broad, very shallow trench that can be traced as far east as Falkirk, beyond which it is obscured by glacial drift. The ancient Knife River followed either this trench or else the site of the Missouri River trench as far as the mouth of Painted Woods Cro*k, and thence north to join the shallow trench east of Falkirk. From this point on the old course is buried by thick drift, but seems to

have continued east, then northeast nast Pickardville and M and to have crorsed the Missouri escaroniont near Lincoln Valley. (c) The ancestral "ivisscuri" river probably flowed northeast through Snake Creok to a point about 15 miles east of the town of Garrison, v/here ib turned southeast, flowed through Turtle Lake, past ta>rcer, and joined the ancestral Knife River near Pickardville. Although buried by thick glacial drift, this route from the head of NriAko Creek to Pickardville is fairly easy to follow both in the field and on topographic map?. Northeast of Lincoln Valley the course of the combined Knife and "Missouri" rivers can not be followed with any-assurance. The

stream may have continued northeast to join the ancient valley now occupied by Devil's Lake am' Stump Lake (probably an old course of the Sheyenae River) or it may have turned northwest to ioin the Souris River in McHenry County. 7. ormor course of the Little I.'missouri River. As can be seen on the map (fig. 12) the old course of the Little I'.'isrouri River is mostly in northwestern North Dakota, in the area being studied by Howard, and will be outlined here only for the sake of completeness. The history and dating of the diversion of the Little Missouri River, however, is important to the general history of southwestern North Dakota and will be discussed more fully. ' ho Little Missouri River enters North Dakota in the southwest corner of ohe State arid flows north to southern McKenzie County, where it turns sharply eart and finally emoties into the Missouri Hivor near Elbowoodn. Loading northeast from the sham bend, and in line with the upstream part of the Little Missouri valley, is a

lar<:e trench that joins the I.'missouri River near Wesson. This trench is occupied in part by Cherry Creek and in part by Tobacco Garden Greek and will be called in this report the Tobacco Garden trench. Upstream from it? junction with the Tobacco Gardon trench the valley of the Little Missouri is flanked by a v.ddo strath terrace, about 150 feet above the river near the State line and 250 feet above the river in McKenzie County. Downstream from tho junction this terrace is absent. The altitude and gradient of the terrace coincide fairly well with the floor of the Tobacco Garden trench, and it seems fairly clear that this trench represents the former course of the Little Missouri River. It seems probable alro that in pre-Illinoian (l) time the Little i'missouri 6ontinuod north across the site of the Missouri River trench and joined the ancient Yellowstone somewhere near the Canadian border. / The Tobacco Garden tronch has been recognized as the former valley of the Little Missouri River by many workers. It was first noted by Wilder (1903, p. 16) and was more fully exnlored by Leonard (1916, op.300-304). Leonard also recognized that the segment of the Little Missouri River below the junction with the trench coincides in part with the outermost glacial drift border, and postulated, therefore, that the diversion from the Tobacco Garden Creek route was caused by the damming action of the ice. Two lines of evidence suggest that the diversion of the Little Missouri River from the Tobacco Garden trench did not occur until early Wisconsin time. Firrt, Howard (personal communication) stator that the bedrock floor of the northern part of the Tobacco Garden

trench IF graded to a baselevel at or slightly below the altitude of the present Missouri River. It i? possible that Tobacco Garden Crock alone could have accomplished this grading, hut it is more likely that the Little Missouri Hiver roaintained its course through tho Tobacco Harden trench until the Missouri River had deepened its valley to about its present altitude. As hac already been stated, the ma3or deepening of the Missouri River trench probably took place during the San~amon intrflacial stage, and the diversion of the Little Missouri River was therefore post-Sangamon. Second, the drift border that coincide? with new course cf the Little Missouri River is early V/isconsin, probably lowan; and it so cms very likely that diversion from the Tobacco Garden trench was accomplished by f the ice sheet that deposited this drift. The lowan date for the diversion also provides an exnlanation of the fact that the badlands bordering the Little Missouri River are more extensive and more spectaular than those bordering other rivers in North Dakota. By lowan tine all other streams had excavated their valleys to about their present depths, but the floor of the Little Missouri River at that tiir.e is represented by the high strath terrace and the floor of the Tobacco Harden trench. The present mouth of the Little Missouri River near Elbowoods is more than 200 feet lower than the old mouth near Nesson, so that after the diversion the Little Missouri had a steeuer fradient. Thug the river was rejuvenated and cut deeply into its old valley floor, leaving this floor as a strath terrace. The rapid deepening of tho master stream valley steepened tho gradient of all the - snail tributaries, enabling thorn to dissect th \vnlls of the Little 17*

Missouri River valley into badlands. This dissection ha? continued from early V.isco-.- ~zo the present tire. Badlands undoubtedly bordered other riT-r? in southwestern North Dakota shortly after they deepened their valleys in response to thf deepening of the Missouri River trench; but that dissection took place longor ago and since then nc*z of the badland ridrer and spurs hive been rounded by weathering and mass-wasting. pihor Pleistocene diversion valleys Ilajor regional diversion valleys In South Dakota the only large valley that appears to be the result of glacial diversion is that of the Missouri River itself. In North Dakota, however, west of the Missouri River trench are many trenches that cross divides between the present streams. The largest of these trench.- fora an anastamoring valley system that extends from the Killdeer ?iountains in Dunn County southeast across Dunn, Stark, Morton, Grant, and Sioux counties, finally joining the Missouri River a few miles above Fort Yates. For convenience I have .given individual names to various segments of this valley system (pi.3), cut all these segments are part of one large feature and sho-ld be regarded as a unit. Similar trenches connect the valley of the Little Missouri River in western North Dakota with the valley of the Yellowstone River in eastern Montana. These through valleys or trenohes were first recognized and mapped in reconnaissance by Leonard (1916). Later, Alden (1932) mapped similar features in eastern Montana

and showed that they connected with the valleys Leonard had discovered in western North Dakota. Although Leonard doubted that the Missouri Rivor trench orifinatad as the rerult of glacial diversion, he clearly recofnized the systorr. of throu-h valleys as the product of water? diverted from their former courses by the margin of an ice sheet (1916,on.295-300). Concerning there valleys he stated (p.299): "Pleirtocene valley of "irsouri and Yellcv/stone Rivers "But while the Mirsouri Uiver probably occupied its present valley for a oonriderable time prior to the Glacial period, the icc-shcet, when it invacod the region, blocked the valleys of both the Missouri and Yellowstone rivers and also tho pro-glacial valley of the Little Missouri, forcing theso streams to seek new channels. Lakes were formed in the, valleys of the Ycilowstone and Little Missouri rivers, the water rising until it overflowed the divide between the latter and the Knife Hiver south of the Killdeer Mountains. The combined waters of the three rivers flowed east across Dunn County and southeast across Norton to tho mouth of the Cannon Ball river. The valley thus formed crosses the divide between the Knife and Heart rivers and also that between the Heart and the Cannon Ball. The leng-th of this Pleistocene valley of the Yellowstone and I'iisrouri rivers from tho head of the Knife to the mouth of the Cannon Ball is 155 miles." The course of this former Pleistocene drainage as mapped by Leonard and Alden is shown in fig. 13. My reconnaissance map of the same trenches is shown in pi. 3. A comparison of the two maps shows that I disagree with Leonard's mapping of this former drainage at only three places. These are: (l) Porcupine Creek trench. Leonard thought that in southern Morton County the old Pleistocene drainage turned east at the town of Timmer, and entered the Cannonbal1 River near Breien, 17G

and from there continued east to the Missouri. My reconnaissance rhowed that another branch of this trench trends south to Shields, then turns east, crosses the divide between the Cannonball River .and Porcupine Creek, and continue? east through the latter valley to join the *'irsouri River about four miles north of Fort Yates. After the ice had cleared southern North Dakota as far as the prosonj mouth of the Cannonball River, probably all the drainage from the nortmvert joined the Cannonba]1 at Broien, as suggested by Leonard. (2) Killdeer trench. As can be seen from the map (fie. 13) and from his description of the diversion drainage, Leonard believed that the waters of the Yellowstone, Missouri and Little Missouri rivers were diverted to a course south of the Killdeer

Mountains and entered the headwaters of the Knife River valley about six miles southeast of Grassy Putte in McKenzio County. Alden accepted Leonard's idea and postulated that the waters left the valley of the Little Missouri River in southern McKenzie County by way of Bicycle Creek. I have examined the uplands in southern McKenzie and northern Billings counties and have found no evidence of any major drainage channels in this area. Howard (personal communication) has also visited this area and has independently come to the same conclusion. On the other hand, I did find a valley in western Dunn County that probably represents the former diversion path of the Yellowstone and Little Missouri rivers. This valley, which I have called the Killdeer trench, starts

Just east of the Xilldeer Mountains at tho low divide between Jim's Creek and the drainage of Spring Creek. The valley trends southeast part the to'.vn of Xilldeer, joins and crosses the valley of Spring Creek at Lake Ilo about two miles west of Dunn Centor, and continues south, entering the valley of the Knife River about a mile east of Fmerson.' Tho diverted water? of the Yellowstone and Little Missouri rivers, therefore, probably never flowed south of the Killdeer Mountains. Early in the Wipeonsin stage the waters probably flowed down the valley of the Little Missouri to the mouth of Jim ? s Creek east of the Killdeer Mountains, and from there spilled over into the diversion val leys to the southeast. This system of diversion valley?, however, is believed to have been out during some pro-Wisconsin glacial stags; and, as has already been pointed out, the east-west segment of the Little Missouri River north of the Killdeer Mountains did not exist at that tiir.o. During the earlier glaoial advance the Little Missouri River probably continued northeast through the Tobaoco Garden trench to a point

about five miles, northeast of Watford City. From here it was diverted southeast through the valley of Cherry Creek and continued in this direction until it was able to spill over into the drainage of Spring Creek somewhere east of the Killdeer Mountains. (3) Emerson trench. Leonard* s map shovrs that after entering

and crossing tho -valley of'the Knife River tho Pleistocene drainage followed three separate courses tov/ard the southeast. To Leonard* s

three branches should bo added a fourth - a largo r,hallow trench that loads south from the valley of the Knife River at Emerson, crosses tho head/ators of Deep Creek and finally turn? east to join the Doep Greek trench along the Dunn-Stark County lino. These four branch valleys leadin.- south from the Knife River were not used cirr.t&anoously. The altitude of their floors is progressively lower from west to east, and they were probably used one at a time, starting; with tho trench south of Emerson and ending with tho Elm Creek trench. DA IT. OF THE 11 VERSION It rooms, well established" that the abandoned valley system just described was formed during the Pleistocene epoch whan the streams of southwestern North Dakota and eastern Montana were diverted from their normal courses by a glacier. The mechanism of tho diversion was exactly the same as the one which caused the formation of the Missouri River trenoh. As a matter of fact, the only difference between the Missouri Rivor trench and this valley system is that the UiLssouri trench was successful in holding the diverted waters after the retreat of the ice sheet. The date of this diversion cannot be definitely established. Leonard and Alden both considered that tho waters of the Little Missouri and Yellowstone rivers wore originally diverted by tho pro-Wisconsin ice sheet that deposited ~the outermost drift in southwestern North Dakota. According to Leonard this drift was

Kansan; according to Alden, Illinoian (?) or Iov:ant At tho time o*' Aldon' s work the I ov.-c.n vrar. regarded as a soparnte pro-Wisconsin glacial stage J'y reconnaissance maoping of the glacial features indicates that tho outermost drift is of early Wisconsin (lowan) age. Tho diversion system, however, I believe to be pre-V,!isoonsin, probably Illinoian. The evidence, scanty and inconclusive, is a? follows: (1) Since tho cutting of this valley system, mass-wasting has rounded the sharp spurs, and small tributaries have cut fairly wide straths graded to the floors of the various member trencher. In contrast, the sides of the Peulah trench in the Knife River / area are youthful and are drained by steep-sided gullies; yet there is strati graphic evidence that the Beulah trench must be as old as early Wisconsin. (2) Small patches of early Wisconsin (lowan or Tazewell) drift lie on the maturely dissected sides of these valleys. This also suggests that the diversion valleys are of pre-Wisconsin age. (3) Both Crandell and Warren (op. oit.) show that the topography of western South Dakota has been inverted since Kansan time. Gravels of late Karfsan or early Yarmouth age now cap

divides between the major streams. It seems likely that western North Dakota has had a similar history, and that valleys of Kansan age would have been well dissected and largely destroyed by this port-Kansan erosion. But the diversion valleys under discussion are very muoh a Dart of the present topography. IdO

I therefore surfest that this diversion system is oost-Kansan, pro-Wisconsin, or in other words, Illinoian. Probably the system was established Trhile th'3 Tllinoian ice was near its maximum. Later, as the ice rstreatod, a second system war established and maintained, and eventually becarr.e the Virsouri River trench. Although the Killdeer-to-Fort Yatas diversion system was abandoned after the establishment of the Missouri ftiver trench, it was undoubtedly re-used and its bedrock floor was probably Ipwered during at least the rv,o early 'Wisconsin advances of the ice. How much of the deeneuin occurod during the original diversion and how much durinr subsequent occupations is impossible r to tell. Minor diversion valleys Between the I owan drift border and the Missouri River trench are numerous isolated diversion valleys or trenches that trend northwest-southeast across the divides. These range in size from trenches one to two miles wide and 15 to 20 miles long to small notched divides, 50 to 130 yards wide and loss than a quarter of a nilo long. All these diversion valleys, largo and small, originated in the same way as the Jissouri River trench, except that the smaller ones wore cut by smaller amounts of water and woro used for shorter periods of time. Most of these smaller diversion valleys are too small to be shown on the reconnaissance map. Those that croso the Knife River area are- described in detail in a subseouent section of the reoort.

Wisconsin drift sheets of southwestern North Dakota Previous Correlations The presence of glacial drift south and v/ost of the Missouri River trench in North Dakota has been knovm for many years, but most previous workers have classified it as pre-V.'i scons in or earliest Vi'isconsin (lowan). Leonard (l916b) thought that the so-called "Altamont" moraine on the Cotoau du Missouri marks the limit of Wisconsin glaciation and that all the drift -southwest of this moraine is pro-Wisconsin, probably Kansan. Alden (1932, pp.75-78) accepted the "Altamont" moraine as marking the Wisconsin drift border, but considered the older drift to the southwest to be of either Illinoian or lowan age. 'Though he gave two possible ages for the older drift, Alden clearly preferred to think of it a? lowan, which he regarded as a pro- Wisconsin glacial stage. In coinpliling the Glacial Map of North America, Flint and others (1945) ohose to emphasize the lowan rather than the pre-Wisoonsin facet of Alden 1 s correlation, and on the glacial map all drift southwest of the "Altamont 11 moraine is shown as lowan (early Wisconsin). Townsend and Jenke (l94l) pointed out that the correlation of the "Altamont 11 moraine of northern North Dakota with the type Altamont moraine in eastern South Dakota is just guesswork and is based neither on the stratigraphy of the deposits nor on the detailed mapping of the moraines through the intervening areas.- Therefore, they proposed that the moraine in northern North Dakota be renamed the Max Moraine. They also showed that the Max moraine owes its position and development to the bedrock high of the Missouri escarpmcntj andthoy

suggested that this moraine may not mark the teminus of any flncial advance. Recently Flint has studied the glacial deposit? of South Dakota and has concluded that deposits of all four substages of the Yiisconsin are present in that State (1951, report in preparation). Flint generously mace available his map of the drift borders and this was a great help in my attempt to correlate the drift sheets in southwestern North Dakota. Results of present work The present study of the glacial geology of routhm North Dakota indicates that three agp s of drift are present routhvmst of the Missouri River and that all three belong to the Wisconsin stnge. The two older drifts are early Wisconsin (lowan and Tazewell respectively) and the third drift is late Wisconsin, probably Mankato. The evidence for these conclusions comes partly from the strati graphic relations of the deposits in the Knife River area and partly from the tracing of the drift borders between the Knife River area and the North Dakota-South Dakota State line. STRATI GRAPHIC EVIDENCE The Pleistocene deposits of the Knife River area have been described. The factors that bear on their dating and correlation can be summarized as follows: , (l) Three different ages of deposits are differentiated. All three are conformable with the topography that postdates tho deepening of the Missouri River trench and are therefore post-Illinoian.

(2) The two oldar drifts have similar tonograohic and aro separated by a minor unconformity. Roth, therefore, are assigned to tho early V.i rconrin. (') The third drift in uch ie~s eroded than tho other tv/o and is sonaratod fron th olc! er deposits by a marked unconformity. This unconformity probably represents tho mid-Viisoonsin (Tazewcll-Cary) intorval and the third drift, therefore is thought to be lato Msconsin. '(4) The border of the late Yi scons.in drift crosses thr Knife River area. Northeast of its margin I have found no evidence, either topographic or s trati graphic, that would indicate a drift border.

Quite to the contrary, the young outwash fill in the Knife Hiver valley (Qsd.,) and the outy/ash in valleys that drain the i.lax moraine o are both graded to the same fill in the Missouri Hiver trench, suggesting that the young drift in tho Knife area and the surface drift of the !l'>ax moraine are of the rair.e general age. Thus all the drift northeast of tho late Wisconsin border seems to be of the same age. (f>) Flint (1951, report in preparation) has shown that the Gary drift has been overlapped by tho Wankato in northern South Dakota, and 1 have found no features that suggest the Gary drift reomorges in southern North Dakota. Therefore, the single late Viisconsin drift in southern North Dakota is probably Mankato.' EVIDENCE FRO?.' THE DRIFT PORDKRS Flint (1951, op.cit.) has shown that all four "Wisconsin drifts are present in South Dakota, but that only three can be identified

at tho North Dakota-South Dakota State line. Near Java, in V.;alworth County, the Mankato drift crosses the Gary border and the Cary drift is overlaoped from here to the State line 25 mile? farther north. Therefore, at tho Stato line there are only three Wisconsin drifts, the lowan, the Tazewell, and tho Mankato. In North Dakota, unfortunately, the drift borders cannot be traced with as much certainty as they can in South Dakota, at least in reconnaissance. There are two reason? for this: (l) In Sonth Dakota the only VVisconsin ice to cross the Missouri Kiver trench was the lowan; the other three drift borders lie east of the Missouri Hiver. In North Dakota, however, all the known Wiscons in drift borders lie west of the trench. This is an important fact b'ecause the Missouri River trench acted as a baffle and removed much of the sediment load of any ice sheet that crossed it. Thus, in many places in southern North Dakota even the late Wisconsin drift consists only of scattered boulders lying- on the eroded bedrock surface. The differentiation of such thin drifts is very difficult. (2) In South Dakota the various Wisconsin till sheets are separated by deposits of loess up to several feet thick, and in many exposures the "Wisconsin tills can be identified by their stratigraohic positions. The "Peorian" (lowan plus Tazewell) loess is very persistent, especially east of the Missouri rtiver, and is very helpful in separating early from late "Wisconsin tills (Flint, 1951, op.cit.). In North Dakota these loess sheet? thin and virtually die out 10 to 15 miles north of the State line, and the thin drifts west of the Missouri River trench are in most places not separated by any interveninf deposit.

thel"ss, despite tho difficulties in tracing the drift bordorr, the follovrinp; oneral conclusion? can bo drawn: (1) The lov/an drift horror i? marked by the south/ost limit of erratic stones. It can be trac°c without any largo £apr from the South Lake t a State line northv/-t to- the Kill deer Mountains. (2) Th Tazewell drift border is too faint to follow in North Dakota. (3) The -ankato drift border cannot be traced continuously, but can be followed in a general way from tho South Dakota State lino northwest to the Knife River area, where it coincides with the margin of the third or youngest drift. (4) The Gary drift has not been identified in southern North Eakota. Early Wisconsin drifts - lov/an and Tazewell The early "Wisconsin drift sheets crop out in a northwest-trending belt 20 to 30 miles wide in southern North Dakota. Over most of this area the presence of former ice sheets is attested only by the presence of erratic boulders, mostly of granite, with a few of limestone and dolomite. A very few patches of till and ice-contact stratified drift are scattered over this area but these are very ttinor and do not cover more than one per cent of the total area. Several email deposits of till were found in the bottom of valleys that are graded to the floor of the Missouri River trench. The abof till over most of the area is due in part to erosion of ihe and in part to nondepositicn. Judging from tho patchoc of till

aftor the stagnant tonrue of ice in the ' ' isso'jri trench n leaving Qsd as a kane terrace. A ?hort time later the ice readvanced across thf Missouri trench in the Stanton 01; ad ran pi and deposited the boulder travels (QicM on top of the >'irsouri Hiver gravels near Fort Clark". This IP the last Pleistocene ovent recorded in the Knife Hiver area. Early Vi'irconsi n gravel deposits In the southwest cart of the Knife River area are small isolated natchar of sand and gravel that can not be correlated dfinitely with any one till rheet or valley fill. Sor.e of thse deposits are definitely of glacial origin, pome are fluvial origin and still other? are indeterminate. I believe them to be of early V.isconsin are because; (l) they lie beyond the late v'.isconsin O-'ankato) drift border and (2) they have been more extensively eroded than the late Wisconsin deposits to the northeast. On the map there fravelr have been divided into three catafories based partly on their litholoy and partly on their occurrence. Older The deposits mapped as Owe cannot be correlated definitely with either of the early Wisconsin till sheets but are almost certainly of glacial origin and are associated vdth one of the tills. These deposits are small and comparatively few in number and no attempt was made to discriminate between outwash and ice-contact deposits. Probably the two deposits in seer. 14, and 26, T. 143N., n. 91rt., are kames; in contrast probably the small patches in seer. 8, and 17,

that are preserved, and fro;,- the younor till deposits to the northeast, probably the early V<iscort?in drift was deposited as a very thin and patchy blanket. IGtVAN DRIFT BGHD&R The outermost drift border in southern North Dakota is dravm along the southwest limit of erratic stones. Although the stones are not everywhere abundant, they are numerous enouph in mort places to permit the drift border to be traced fairly accurately. This drift border was first notd by Wood (1904), who manped a segment of it in southern Dunn and northern ?tark counties. Later, Leonard (1916 and 1919) maoped this drift border in reconnaanco from the South Dakota State line northwest to the Montana State line. The drift borders as mapped by Wood and Leonard have been shown on pi. 3. Leonard's version, it can be seen, was extremely generalized; apparently he drew a line tangent to the outermost lobes of erratics and made no attempt to show any of the details. Western North Dakota today is far more accessible than it was when Leonard examined the glacial deposits, and I was able in a relatively short time to map the drift border from the South Dakota State line to the Killdeer Mountains. As can bo seen on the map (pi.3) this line is not, as Leonard had supposed, nearly straight, but rather is very sinuous. In general it shows that the ico front had lobes that extondd up the larger valleys and reentrants where it imninged on higher ground. This drift border crosses the North Dakota-South 1'akobn -tnte

line in the southwestern part of T. 12'N., K. 82'<-., a:;d at thin point it coincide? v/i th Flint's lowan drift border. Partly because of this coincidence and partly because the drift northeast of this "border shows the same dopree of weathering and erosion as the lowan drift in bouth Dakota, the outorn.ort drift in T.'orth I-'akota har been correlated with thf Jowan racial subrtage. r,i<AN;TK. ROT FT DIOR" FOUTIMET op ??rt: TOW.*T DRIFT CRDKR In most places in southern North Dakota the limit of erratic stones is fairly well defined and the drift border can "be ciravrn v/i th moderate confidence. Ther are, however, a few franito bowlders one to several mils beyond and across inter stream divides from the general drift border, and I do not believe that there should be included in the lowan drift. It is possible that these out-lying boulders are relicts of some pro-Wisconsin glaciation. But it has already been noted that the lower part of the rthite River formation, contains a fnw scattered moulders of granite (p. 95) and it is equally possible that these anomalous franiter have been let down onto the present surface by the erosion of the overlying White Rivrjr pediments. The localities whor there anomalous boulders have been found have been shown on the map (pi. 3); the numbers on the map correspond to those in the following discussion: (l) Boulders in western I'orton County. In the northern part of T. 130M. , :i. 90Vi. t are two occurrences of granite boulder-? 8 to 10 miles south and wrrt of the Tovan drift border. The drift border in this area is fairly well defined and skirts the edfe of a highland area west of Glen Ulln and south of Hebron. The outlyinp138

boulder? arc in the highland area and ar'7- ir.or than 100 fot higher than tho ode cf the lov.an drift. The first of these occurrences is Questionable; it consists of two rr.nite boulders near each ether in a road ditch in a small valley roc. 14. If there boulders wore not throv.-n from a true'/-', they could be explained by ice rafting; in a lake wort of the edf-o of the Towan loo. The second occurrence, however, cannot be accounted for either by ice-raftinf or by the work of man. In the northeast corner of sec. 14 if? a gravel deposit that capr the divide between Heart Butte Croek and one of its tributaries. The fravol consists r-rincinnlly of poorly rscytd angular fragments of chert, iron oxide and sandstone derived from the Fort Union formation, but included also are several cobbles cf white sandstone and two sir-ill boulders of granite The sandstone cobble? v:ere derived from the *'hite rtiver forrra tion, the nearest outcropr of which are even today only 10 mile? awey. The Oranites must have been derived either from the hite lUver formation or from the erosion of a glacial deposit. If they are of glacial origin there pranites must represent some pre-ttisconpin stage, for the topography of this area has been inverted since this gravel deposit was laid down.

(2)Bouldor of fjanodiorite in eastern Stark County. In FWvf, sec. 22, T. 139N., R. 92., is a single boulder granodiorite or possibly diorite. This boulder is about 8 miles south of the Towan drift border and on the opposite side of the divide between the and Heart rivers. Stones of similar lithology are found in

the facial drift but have net been noted in the fthite ivor formation . (3) Gravel deposits near Gladstone. Through most of it course in btark and to or ton counties the Heart Hiver is flanked by a terrace 40 to 60 feet above the flood->lain and capped vrith poorly sorted sand and gravel. hast of the Trwnn drift border ths re gravels contain nobbles of granite and limestone obviously derived from glneial drift. Yjost of the drift border thn gravel? consist entirely of strnes derived from Fort Union and Vshite River formations, except at the town of Gladstone where several granite boulderr have boen dug from gravel pits on the terrace. Severn! othr rranite erratics were noted in the valleys of thp Kpart and Gren rivers just northwer-t of Gladstone. These granites, however, are probably derived locally from the erosion of the hjfher gravel deposits in this area. Southwest of the Heart River at Gladstone are deposits of sand and gravel up to 200 feet above the present river. These deposits, unlike those capping the terrace, consist principally of well sorted coarse-grained sand with a few stringers of rounded pebbles. The pebbles are principally ouartz, quartzite, and vari-colored chert with minor number of red-brown andesite porphyry. Similar dnosits of sand and gravel extend for some distance up the valley of Green 3iver in T. 140N., R. 95X, but are not found in the valloy of the Heart Kiver upstream from Gladstone. Thfsn high level sand and gravel deposits were first notd by " ood (1904, p. 117), and were later described by Leonard (l r<16, p. 530). Both Leonard and Vn stated that the deposits contained large {Tanite boulders, and

inferred therefore that a lobe of the ice sheet had crosred the Heart River at Gladstone. Concerning these gravel deposits Leonard stated: (1916, p. 530): "That a lobe of the ice sheet crossed the Heart River at Oiadrtono is shown by the presence of thick deposits of drift travels on the unland one to two miles south of the Heart and at nn elevation of between 100 ar.d 200 foot above the rivor level. In places the travel and sand have a thickness of at least 90 feet, and the deposit contains a number of good-sized granite boulders. A well defined gravel ridge marks the edge of the drift for throe or four miles in this area south of the Heart River at Gladstone. This ridro rises 30 to 40 feet above the surface on either side and falls away rather abruptly on the south, while the north slope is more gradual." Alden also visited the gravel deposits south of Gladstone, and accurately described them as follows (1932, p. 78): r "In 1921, when the Gladstone deposit was visited by the present writer, sand and gravel were being taken from a small pit near the top of the north slope. South of this is a larger excavation to which a spur track formerly lead from the railway. These pits expose 10 to 30 feet of stratified cross-bedded sand and gravel, the pebbles in whr" rh more closely resemble the bench gravel along the Yellowstone River than the glacial drift. These pebble? range from less than an inch to three inches in diameter and consist princioally of quartz, quartzite, and chert, with some agate and some dense dark greenish crystalline rocks, diorite, and prophyry. None of the stratified material, so far as is noted, is certainly glacial drift. There were about a dozen boulders of gray and pink granite gathered in a pile on the bottom of the pit, and these are probably of glacial derivation, but they may have rolled down from the surface during excavation of sand and gravel. The deposit appears to can the ridge for some distance to the south and east. It is possible that this is an erosional remnant of a late Tertiary or early Pleistocene gravel bench similar to the benches of the Yellowstone ;<ivrr. If the granite boulders were actually interbedded with the sand and gravel however as is intimated by Leonard, it is probably a glacial deposit. No glacial drift was seen by the present writer on a traverse extending about nine miles south of Gladstone and thence east and north to Hichardton nor along the main road and railway between these two places."

Although as Alden stated, the pebbles in th-; Gladstone deposits are cimilor to pebbles in .jravels along the Yellowstone Kivrr, they rescn.ble even more the gravels of the k'<hite Kiver formation in North Dakota, Also, the beds of coarse sand in the Gladstone deposits could have bpen derived from the lower third of the A'hite Hiver formation, but could not have come from any othrr local source. therefore, it seems probable that the Gladstone dPOOS its wer made by the rosion nr.c rodeposition of r-materials from the V'hitc -ivor formation. Inasmuch as the V-'hite Hiver is known to contain at loastfcvo granite boulders, not even the granite boulders in the Gladstone deposits ,need bo of glacial derivation. The reason why the Gladstone gravels were deposited is, not apnarent. These thick sands are apparently restricted to the valley of the Green River and to the valley of the Heart Hiver just below the mouth of the Green. The easiest way to explain such local aggradation in this part of the world is to invoke an ice darn. However, i have examined the area northeast of these deposits and have found no glacial deposits between Gladstone anc the lowan drift border, 10 miles to the north. The edge of the lowan drift lies north of the divide between the Knife and Hrart river drainages and shows no tendency to be lobate across this divide toward Gladstone, therefore, if the Gladstone deposits wero caused by glacial action, this action probably took place during some pre-'ATi scons in advance, and all other traces of this glaciation have now been removed. It is my opinion, however, that the Gladstone deoosits are not of glaoial origin, because, of all the materials in those denorits,

only the granites are even possibly of glacial derivation, and these oould equally well have cone from the Vihite River bed P. (<0 Granite boulder? near Dickinson. Wood (1904, pp.116-117) reported several granite boulders on the hillsides in and around Lirkinson and several more from the terrace of the Heart River west of Dickinson, "Wood thought that these erratics had been rafted up the Heart River valley by bergs calving from the hypothetical Gladstone lobe of the glacier, but I think it more probable that they are relicts of the "hite River formation, (5) Granite boulders in southwestern Dunn County. In T. 142N., H. 95 and 96'. f are several granite boulders on the dissected north edge of the upland divide between the Heart and Knife river drainages. These boulders are about 3 miles southwest of i;he lowan drift border. There is no local evidence whether they are of glacial origin or were derived from the V\hite River formation. TAZF/A'KLL DiUFT BORDER Flint (1951, report in preparation) has identified and traced the approximate limits of the Tazewoll drift in north-control South Dakota. According to Flint, the Tazewell border is impossible to trace accurately; it apparently lios east of the Missouri River everywhere except in the northern part of the State, where is possibly is west of the river. Extending from northeastern Cor son County, South Dakota into southeastern Sioux County, North Dakota IP a ridge of poorly ported bouldor gravol that in probably of inio origin and that may bo tho feeblo terminal moraine of the ice sheet.

Traced northwest into Sioux County, North Dakota, this morainic ridge dies out, and frorr. hare on to the northwest I could find no suggestion of the Tazewell drift border. Therefore, the "best that can be said of the patchy drift that lies between the lowan and the Mankato drift borders is that it probably includes deposits of both lowan and Tazewell age, but that these deposits oannot be differentiated. Late Vii scon sin drift - Mankato Drift of late Wisconsin (Mankato) age covers most of eastern and northern North Dakota. VJest of the Missouri River this drift crops out in a belt, 25 to 30 miles wide in the Knife River area, f that narrows to the south, finally ending about 15 miles north of the South Dakota State line. Lithologically the Mankato drift is similar to the lowan and Tazewell drifts, and can be separated from them only on the basis of stratigraphy and differences in topography. In Mercer County the tfankato drift is much less eroded than the early Wisconsin drifts and covers most of the gently rolling uplands in the northern and eastern parts of the Knife River area. Farther south, however, in Norton and Sioux countior, even the tfankato drift has been extensively eroded and over much of its outcrop arra is reprerented only by scattered granite boulders. MANKATO Did FT BOHL'fcR Flint"1 s map shows that the Mankato drift border enters North Dakota in southeast Eraons County, in T. 129N., R. 76W. In this

area the Mankato drift is separated from th early Msconrin drift by an intervening blanket of loes.% and the J'ankato drift border is drawn at the soutrr.rest limit of till or erratic stone? on ton of this loess. The drift; border can be traced from th<? State line northwest to V/est field, whor" it entors and is lost in the dissected topography -chat borders the Missouri Hiver trench. Between the dangling end of the Mankato drift border near testfield and the lato Wisconsin drift border in the Knife iiiver area are five localities in Sioux, Morton and Oliver counties where deposits or topographic forms suggest presence of a drift border, one of these locali-ies by itself would justify the drawing of a drift border, but the fact that they fall on a nearly straight line connecting- the Mankato drift border near West fie Id wi th the late Wisconsin drift border in the Medicine Butte cuadranfle mokes it seem probable that the.e five localities are actually rearorentive of the Mankato drift border. These localities have been lettered A through E on the map on ol. 3. Localities A and B are in northern Sioux County. Locality A is in sec. 19, T. 133N., 3. SOW. and is on the upland between the Cannonball River and Porcupine Creek. At this point numerous glacial stones overlie at least six feet of coarse-brained sandy loess* This loess is probably the equivalent of the pre-Mankato loess farther southeast and overlying glacial drift is of late Wisconsin age. Locality B, in sec. 7, T. 133N., K. 81W, consists of several exposures .on tho north edgo of the bluffs of the Cannonball Rivor.

These bluffs, which are composed of sandstones and shales of the Hell Creek: formation, are caused by about eight feet of sand/loess, Drobably tho same loop? ar at locality A. The loess is separated from the Hell Creek formation by a one to two foot thickness of very bouldery till, but no glacial stones were found on top of the loess. Auin, if this ie the pre-Mnnkato loe.**, the underlying glacial drift is early Wisconsin and this locality lies outside the late Wisconsin drift border. The ?.'ankato drift border, therefore, has been drav/n between A and P. Locality C is in southern Morton County a few miles south of Barner. Thir locality is not a single outcrop or group of outv

crops but a zone several miles long. In this zone there is a northwest-trending line, northeast of which granite boulders &rv two or ttree times as numerous as they are southwest of it. Also, northeast of this line several patches of till and a fev.r small dames were noted, whereas to the southwest the drift seems to consist entirely of scattered boulders. Locality D is in Morton County about six miles east of Almont. Extending north from sec. 11, T. 137N., 8. 85W., to the central part of T. 138N., R. 85W., is a series of long linear ridges comprised ohiefly of boulder gravel. These ridges probably represent a poorly developed Mankato end moraine. Locality E is in southwestern Oliver County and consist? of a series of small trenches that form a line extending from the eastern part of T. 141N., R. 87W., northwest into the eastern part of the Medicine Butto quadrangle in seo. 25, T. 142N., R. 88W. These

trenohos range in size from a tenth of a mile lonr and 50 yard? wide to nearly tv.ro miles long- and a quarter of a mile wide. They all trend southwest and cut through the divide? between Beaver Creek and Otter Creek and between Peaver Creek and "rush Creek. Apparently the f'ankato ice stood at this position lonp enourh for mflltwuterr to cut the se notches ar thry spilled over into the drainage of Paaver Creek. Al.?o, northeast of this series of trenches the till cover is thicker and more persistent than it is to the southwest. This series of trenches and the distribution of the till was first noted and correctly interpreted in 1"949" by V.. D. Johnson Jr. of the Geological Survey, who pointed them out to me in the field. f In the Knife Hiver area the la'ankato drift border is located partly by differences in the topography and thickness of the tills and partly by the southwest limit of exposures of till overlying the intermediate fill, Qsdg. In the Medicine Butte quadrangle, the Mankato till is much thicker and more extensive than the older drifts and tho Mankato drift border is easily traced from the eastern ed-e of the auadrangle to a point about a mile east of Medicine Putte. Here at locality F (pi. 3) the drift border is marked by a few small icecontact deposits and also by a trench cut by meltwaters that spilled over from the drainage of Paaver Creek and flowed northinto the Knife River valley. Northwest of locality V the Mankato drift border is difficult trace in detail. It apparently extends up the Knife Kiver vallry r far wost as locality G (pi. 3), which is in the Proncho quadranf1

in tho SE:}, sec. 2, T. 142N., R. 90W. At thin locality a deoonit of fin*3 - to medium-grained sand (probably Osd2) IP overlain by several boulders and cobbles of granite. V/est of locality G there is no evidence of Mankato drift in the valley of the Knife River. It is possible that the drift border should be drawn far enough gouth to include the large patch of till that caps the upland in the northwest part of the Broncho quadrangle, but there i no good evidence either for or against this hypothesis. The drift border has been arbitrarily drawn so that it makes a reentrant around this upland and then extends west of the valley of Spring Creek.

f In the Golden Valley quadrangle the Mankato drift extends up the valley of Spring Creek at least as far as locality H and probably as far as Dodge. Locality H, a railroad cut in the east half of sec. 18, T. 144N., R. 90'A'., shows Mankato till lying unconformably oniho intermediate fill, Qsd. West of this locality the Mankato outwash fill, Qsdg, extends at least to Dodge, but could not be identified west of the Golden Valley quadrangle. Therefore, the Mankato drift border has been drawn just west of Dodge, at which point it leaves the Knife River area. No attempt was made to trace the Mankato drift border northwest of the Knife River'area. It probably continues in a general northwest course and crosses the lower part of the Little Missouri River in eastern Dunn County.

Land forms in the Knife River area General statement The land forms of the Knife River area are the result of martswasting, fluvial erosion, and glaciation. The major features have already been described in the general discussion of southwestern North Dakota, except that the Oligocene-capped buttes and the late Tertiary planation surface are not present in the Knife River area. The interstream areas consist largely of a rolling upland with gentle slopes, locally covered with drift. Standing above the ur>- land are a few monadnocks, among which are Medicine Butte and the hith hills to the south. The upland seems to be a part of the broad gently-rolling upland surface that comprises much of southwestern North Dakota. Land forms that are the direct result of glacial deoosition include the ground moraine in the northern part of the area, the Krem end moraine, and several mounds of ice-oontaot gravel; features that are the result of glacio-fluvial erosion and deposition include diversion valleys and the higher terraces; and features that arc the result of post-glacial erosion and deposition include the floodplains and lower alluvial terraces, sand dunes, and several types of forms produced by mass-wasting. Of the various geomorphic features just mentioned, the moraines, the terraces, the floodplains, and the sand dunes have been describee elsewhere in the report and will not be discussed here.

Diversion valleys Both major and minor divide? in the Knife River area arr crossed by through valleys or trenches, similar to those already described in other parts of the State. The diversion valleys of the Knife River area are divided into two groups. Group No. 1, "major tranches", includes the larger through valleys that cross divides between the major streams and were cut largely by the diverted waters of the principal streams. The floors of these large valleys are covered by'post-glacial alluvium. Group NO. 2, "minor trenches", includes all the smaller diversion valleys that connect the valleys of the smaller tributaries. Most of the smaller diversion valleys were f cut by glacial meltwater. Their floors are not covered "by alluvium and they have boen shown on both the geologic man (pi. l) and on the Pleistocene map (pi. 4 ) by hachured lines. MAJOR TRENCHFS jouth Fork and Elm Creek trenches The largest through valleys or trenches in the Knife River area are in the Broncho and Medicine Butts quadrangles and are occupied by the South Fork of the Knife River and Elm Creek. These two trenches are a part of the long system of diversion valleys that extends from the Killdeer Mountains to Fort Yates (pi. 3 ) One mile east of Glen UUen in Morton County they merge and become the Muddy Creek trench, which continues southeast to the Heart River valley. Both the South Fork and Elm Creek trenches have broad, nearly flat alluvial floors lw to 2*7 miles wide, and both are cut 200 to 300 feot below the surrounding uplands. The \jplands slopr toward 2CO

the trenches, suggesting that these trenches are the sites of previously existing valleys, which were deepened and v/widened by the diverted waters. The present divides in the trenches are a few miles south of the Knife River area, and from the divides the alluvial floors slope gently southeast toward the Heart River and gently northwest tcv/ard the Knife River. However, for several miles north of the divide in. the South Fork trench, the tributary streams enter the trench at right angles or are barbed to it, suggesting that at one time the divide was farther north than it is today. The old divide was proba;.ly somewhere in the p. outhwest corner of the Rroncho quadrangle, near the middle of T. 141N., R. The divide in the'Elm Greek trench may or may not have shifted; the pattern of the tributary streams is equivocal and could be interpreted in several ways. Thick alluvial deposits fill -both the South Fork and r-lm Creek trenches, especially in the divide areas, but the exact thickness of the fills and tho altitude of the bedrock divides are unknown. Well data aro scanty, for tho water table in the trenches is high and most farm wells are lesp than 25 feet deep. Two wo Us, however, give minimum figures for the thickness of fill. Tho first well, in the Northern Pacific Railway Company 1 s stockyards at Hebron in northwest Morton County, penetrated bedrock at a depth of. about 70 feet. But even though the we-11 is on the floodplain of tho South Fork of the Knife River, it is two miles west of where that stream joins 'the South Fork trench, and the fill in the trench is certainly

thicker. The second well, in sec. 10, T. 141N., H. 89YV., 1 8 in the El IT. Creek trench about 5 miles north of the divide area. According to the driller this well penetrated "... 123 feot of ouicksand" before the sides caved in and drilling v.as abandoned. The evidence from the well? suggests a minimum thickness of 100 fent for the alluvial fill in the South Fork and Elm Creek trencher. As indicated in the discussion of the Killdeer to Fort Yatns diversion valleys (p.175) the South Fork and Flm Creek trenches were probably cub during a pra-V'isconsin glacial s-tage (lllinoian and were reoccupied and possibly deepened during Wisconsin time. Goodman Creek and Golden Valley trenches

/ The Goodman Creek and Golden Valley trenches are in the western part of the Knife River area and extend from the drainage of the Little Missouri River south to the valley of the Knife Kiver in the Medicine Butte quadrangle. Although the two trenches are unlike in appearance their geography suggest? that they were originally out at the same time and that their later histories have been different. The Goodman Creek trench extends, the Little Missouri I?iver /i southeast into the Golden Valley quadrangle and joins the valley of Spring Creek in T. 144N., R. 90ft. The northwest part of the trench is occupied by Hans Creek, a northwest-flowing tributary of the Little Missouri River; the southeast part, by Goodman Creek, a tributary of Spring Creek. Except for the 3-mile segment Just north of Spring Crook, tho uplandr on either side slope gently toward the trench, indicating that tho tronch follows the course of a pre202

existing valley. The 3-mile segment north of Spring Creek has steen v/alls, and the upland on the east side slooes away from the trench. This segment, therefore, may be an ice-marginal channel; and, prior to the cutting of the trench, Goodnan Creek may havo flowed through the low sag 2 miles north of Golden Valley, joining; Spring Creek several miles east of its present mouth. Patches of late Wisconsin till and stratified drift are scattered along the sides of the Goodman Creek trench, indicating that the trench was cut prior to late Wisconsin time. The Golden Valley trench appears to be a continuation of the Goodman Creek trench. Its western end joins the valley of Soring Creek opposite the mouth of Goodran Creek. From here it extends east about 7 miles into the highland that forms the divide between the Knife River and Spring; Creek. Here it turns south to oin the Knife River valley in the northwest part of the Medicine Putte quadrangle. The divide in the floor of the trench is in sec. 29, T. 144V ., R. 89V?., and is 125 to 150 feet above the flood plains of forin Creek and the Knife River. From the divide the trench floor si ones west towards Spring Creek wi th a gradient of about 18 feet uer mile, and south tov:ard the Knife River with a gradient of about 20 feet to the mile. These gradients are much greater than those of any other trenches in the Knife River area. The east-west segment of the trench is to 3/4 of a mile wide and is drained by two streams that start in the highland south of the trench and flow north across the trench and through the low hills that divide Golden Valley trench from Spring Creek. The smaller of

"

the tv.-o streams crosses the trench in sec. 25, T. 144N., R. and has cut a narrow coulee 20 to 25 feet into the floor of the tronch. Exposures in the coulee show that the floor of the trench is underlain by fine-drained alluvial fill that; ir probably tho equivalent of the : Jleipbocf-ne fill, Qsd 2 (Tazowell). The contact betv/een this Pleistocene fill and the overlying Recent alluvium is impossible tc determine accurately. The presence of the old alluvial fill (Qsdo) in the trench

provide an explanation to tho problem of why the small streams flow across rather than along the floor of the trench. Originally the small streams draining the area flowed north across the site of the trench to Spring Creek. After the initial cutting of the trench the small streams must have turned west down the floor of the trench and joined Spring Creek about 2 miles west of Golden Valley. Later, when the fine-grained fill, Qsd rose to an altitude slightly above the present floor of the trench, the streams spilled out to the north through their former valleys. Once they were reestablished in these northerly courses, the streams had gradients steeper than the west-sloping floor of the Golden Valley trench, and subreokuent erosion merely tended to deepen and maintain the new valleys across the trench. The north-south segment of the trench is comparatively narrow, being only about mile v:ide, no streams are incised below its alluvial floor and the drainage flows south into the Knife Rivrr. The thickness of alluvial fill (including Qsd) in the Golden

Valley trench is unknown, but in the divide area it rr.urt be 100 feet or more. The east-we st segment of the trench v/as obviously cut by ponded waters, and the bedrock floor at the wo.st end of the trench is at or below the altitude of Spring Creek. Therefore, un- ITS tho small streams flov/inp; west along the original floor *,ho trench accomplished an exceptional amount of excavation, t'r.o bocrock lip of the divide rr.urt be at or below tho altitude of Spring Creek at the west end of the trench; otherwise, the waters that cut the trench would have had no gradient. The Golden Valley trench must have cut at or very close to the cargin of an ice sheet. Otherwise the diverted waters that cut the trench'would have used the lower outlet to the east, arround the nose of the divide between Sprint Creek and the Knife Kivor. If the fine-grained fill in the Golden Valley trench has been correctly correlated with the intermediate fill, Qsd£, the trench must have been cut prior to Tazewell time. The Goodman Creek and Golden Valley trenches together seem to be a diversion route for the Little Missouri Hiver, and were probably out during the same glacial stage. Therefore, if the Golden Valley trench is pre- Tazewell, the Goodman Creek trench is pre-Tazewell also* The absence of the fill, Qsd, in the Goodman Creek trench is anomolous; but may be due to the following factors: (l) During the deposition of Qsdg the small streams of the Golden Valley trench required a steeper gradient than Goodman Creek and built their fill correspondingly higher. Later, during the erosion of Qsdp, Goodman Creek was able to remove large amounts of the fill whereas the small

stream?; of the Golden Valley trench "barely touched it. (2) I-uring the late Wisconsin glacial stage the Goodman Creek tronch was probably reused by the diverted Little Missouri River, but there is no evidence that this water flowed through the Golden Valley trench. As soon as the ice front had retreated as far east as Peulah the diverted water could flow from the Goodrnan Creek trench east down Spring Creek to the valley of the Knife River, and thence went to the north end of the Elm Creek trench. Still later, while the ice front stood at the Krem moraine, the diverted Little Missouri could flow from the Goodman Creek trenoh down the valleys of Spring Creek and the Knife River to the Missouri River trench at Stanton. Peulah trench

t The Beulah trench is sharply incised across the divide between the Knife and Missouri River drainages and was evidently cut by waters diverted from the Missouri River. The northern end of the trench is in sees. 5 and 6, T. 146N., R. 88"W., and joins the valley of Beaver Creek about 2 miles west of where the creek joins the Missouri River. From here the trench extends southeast to a point about 4 miles north of Beulah, where it divides into two branches. The Zap branch trends southwest to the valley of Spring; Creek about 3 miles east of Zap; the Hazen branch trends almost due east to the valley of Antelope Creek about 3 miles west of Hazen. The floor of the trench is 1/2 to 3/4 mile wide and is more than 200 feet below the surround : n£ uplands. The Be-ulah trench is crossed by the Krem moraine, which now forms the divide "between the Knife and Missouri watersheds. Viithin the

walls of the trench the moraine has not been modified by subsequent erosion, indicating that the Beulah trench was not reusod as an outlet for the Missouri River after deposition of the Krem moraine. North of the Krem moraine the Beulah trench drains into the Missouri Hiver via Beaver Creek. South of the moraine the main trenoh drains into Antelope Creek through the Hazcn branch. The Zap branch drains into Spring Creek but carries only the waters that drain directly into it from the surrounding uplands. The uplands flanking the trench slope north and south toward the Knife and Missouri rivers, but they also slope east toward a topographic low at the headwaters of Antelope Creek. Thus, to the wost the uplands slope toward the trench; to the east, away from it. When the trench was cut, therefore, the ice front stood very close to the eastern edge; otherwise the diverted waters would have flowed across the low area at the headwaters of Antelope Creek. The walls of the Beulah trenoh have been dissected into a fairly rugged topography, much more youthful than the topography bordering the South Fork and Elm Creek trenches to the southwest. The eulah trench and both of its branches are filled with unknown .thicknesses of post-glacial alluvium. North of the Krem moraine the only Pleistocene deposit in the trench is a large triangular-shaped patch of late Wisconsin outwash in sees. 21 and 28, T. 146N., R. .88W. South of the moraine, however, both the main trench and the Zap branch contain extensive remnants of late V\isconsin outwash (Qvvo) unconformably overlying the early Vi'is2(3?

consin fill (Qsdg).. The remnants of the fill Qsd2 are especially well preserved in the Zap branch whore their eroded tops rise up to 55 or 60 feet above the present alluvial floor. The deposits of late Wisconsin outwash, are remnants of an outwaph train, extend inr, from the Krem moraine to the valley of Spring Creek. They were deposited on tho eroded surface of the early Wisconsin fill, Post-glacial erosion has dissected the outwash fill, leaving; it as a terrace 15 to 20 feet above the modern alluvium. f In contrast to the Zap branch, the Hazen branch contains very few Pleistocene deposits. At its eastern end, where it joins tho valley of Antelope Creek, are a few remnants of late Wisconsin outwash ana one deposit that appears to be of ice-contact origin. / Elsewhere the only Pleistocene deposits in the branch area few very small patches of till and one small deposit of dirty gravel that was mapped as outwash but could possibly be of post-glacial origin. Deposits of Qsdg, so numerous in the Zap branch, are missing in the Hazen branch. The presence and distribution of the fill Qsdg, indicator that the main trunk and at least the Zap branch of the Beulah trench had been cut by late Tazewell time. On the other hand the youthful topography of the walls suggests that the trench is not pra-Wisoonsin. Therefore, the Peulah trench and the Zap branch were probably out '[ either during the lowan substage or the early part of the Tazewoll substage. The paucity of glacial deposits in the Hazen branch is difficult to explain. The absence of the early Wisconsin fill, Qsd2 , suggests that the Haton. branch was not cut until post-Tazewell time. Possibly

the branch was established during the inid-Yiicconsin erosion interval when a tributary of Antolopo Crook worked itn way hradwn.rd and capturod the drainage of tho Beulah trench. The scarcity of late Wisconsin outwnsh, however, requires another explanation. The few patches of till along its sides indicate that the TIazen branch was cut prior to late "Wisconsin time and was in existence while the outwash train was boing deposited in the Zao branch. Also, the Hazen branch obviously provides a lower outlet route than the Zap branch, and would have been used by the meltwaters unless it had been blocked, either by a fill or by ice. A sedimentary fill seems out of the question; there are no remnants of such a fill, and the small post-glacial streams could scarcely havo removed a fill without leaving some trace. If the branch was blocked by stagnant ice, it is surprising that there are no kames or kame terraces along its side?. Nevertheless, the Hazen branch must have been blocked by something, and the absence of deposits suggests that the blocking agent was ice rather than a sedimentary fill, MINOR TRENCHES AND NOTCHED DIVIDES The smaller trenches in the Knife Kiver area transect the divides between tributary streams. Most of these trenches trend south or southwest and most wore cut by meltwater with little or no help from the diverted waters of the larger streams. The floor? of most trenches are on bedrock except those near the headwaters of "Willow Creek in the Hrcncho quadrangle. The more important of the smaller trenches are as follows:

1. In the southern part of the Broncho quadrangle the wide alluvial flats near the headwaters of Willow Creek are connected with the Elm Creek trench to the east and the South Fork trench to the west by small trenches partly-to completely floored with post-glacial alluvium. The small trenches were probably cut by meltwater that flowed southwest from the ice into the South Fork trench. Like the Elm Creek and South Fork trenches, these small trenches were probably cut during a pre-Wisconsin glacial stago and were again occupied by meltwaterp of the Wisconsin ice sheets. 2. In the eastern part of the Medicine Butte quadrangle in sees. 19 and 30, T. 142N., R. 87W., are three small notches in the divide between Brush Creek and a tributary of Beaver Greek. The notches are part of the group of minor trenches that transect the divides between Beaver and Brush creeks and between Beaver and Otter creeks in southwest Oliver County (p.196). All these trenches were out in late Wisconsin time by meltwater that flowed southwest into the drainage of Beaver Creek. Once in the Beaver Creek drainage, the waters flowed northwest toward the Knife River valley. 3. About 1 mile east of Medicine Butte is a trench, mile long and 200 to 300 feet deep, that connects a tributary of Beaver Creek with a tributary of the Knife River. The trench was out by the same meltwaterp that notched the divides in Oliver County. As these waters flowed northwest down Beaver Creek they were ponded by the edge of the late Wisconsin ice sheet and were forced to rise ar.d spill over the divide east of Medicine Butte. Onoe across this divide, the waters continued west, into the Knife River valley, then west "to the Elm Creek trench.

4. In the southern part of the Beulah Quadrangle north and west of the Zap branch of the Peulah trench is a series of rrr.all trenches and notched divides that record? the course of a late "Wisconsin maltwater stream. The course extends eart from pec. 35, T. 145N. f R. 88V., and then south and joinr a snail tributary of Spring Creek in sec. 7, T. 144N., R. 881\. Ice must have blocked the Zap branch while the meltwater was cutting this course; otherwise the waters would have drained into the Zap branch through the valley in seo. 8, T. l44N. f R. 88W. The patch of outwash in sees. 7 and 8, T. 144N., R. 88VY is nearly 80 feet higher than the outwash train in the Zap branch, and was deposited by the meltwatcr stream whose course has just been outlined. ' 5.In the western part of the Stanton quadrangle, in sec. 36, T. 146N., R. 85W., and sec. 1, T. 145N., R. 85W., is a trench that crosses the divide between a small tributary of the Missouri Rivor and a tributary of the Knife River called Elm Creek (not the Elm Creek that occupies the large trench in the southern part of the area).

The north end of the trench is underlain by the early Wisconsin fill, Qsdp. The trench may have been first cut by meltwater and later partly filled by or it may have been cut when the Missouri River, flowing on the top of early Wisconsin fill, spilled over the divide into the headwaters of Elm Creek. Abandoned valley segments In addition to the trenches that cross divides, there are small to large abandoned valley segments that diverge from the modern valleys only to join them again farther downstream. These abandoned

valleys formerly carried the drainage of the modern valley stream; sorr.e v.-ere abandoned because of glacial diversion, others because of suoeroosition of the streams across bedrock spurs. 1. In the Broncho quadrangle about ]j- mile south of the proposed site for the Broncho Dam is an abandoned valley segment or channel connecting the Knife River valley in sec. 17, T. 142N., with the valley of H-lm Crop'* in nee. 9 of the same township. Test holes drilled by the Bureau of Reclamation indicate thnt the channel is filled with about 115 feet of fine-to medium-.rained fray sand, either th intermediate fill, Csdg/or the third fill, Q?d 3 . The history of this abandoned channel seen? to have been as follows: In early Wisconsin time the Knife Kiver agraced its valley un to f 50 or 60 feet above the present floodplain. During the dissection of the fill the Knife Hivar v;as superposed across a bedrock" sour and cut a new channel 115 feet deep. Later the Knife River removed the ill! from the main valley and resumed itr old course. Still later the new channel war filled with the gray sand encountered in the drill holes. The main valley of the Knife River was undoubtedly filled also, but subsequent erosion removed the gray sand from the main valley and did not remove it from the abandoned channel. In the western part of the Medicine Butte quadrangle in sees. 21, 28 and 29, T. 143N1 ., R. 89V<., is another abandoned valley s.ement of the Knife Rivpr, similar to the one south of Proncho Dam except that the brrirock floor was cut only about 40 foot below the former land surface. Thf> floor of this channel consists partly of Fort Union formation, partly of till, and partly of {rray rand. The

channel is probably the result of superposition of the Knife River from the surface of a Pleistocene fill onto a bedrock spur. Evidently the Knife River did not maintain this course long enough to erode the Fort Union as deeply as in the channel south of the Broncho Bam. 3. In the eastern part of the Medicine Butte quadrangle a major tributary of Beaver Greek abandoned a two mile segment of its former valley. The abandoned segment extends from the eastern half of sec. 22, T. 142N., R. 88 ., across sec. 23 and into the northwest corner of pec. 25. The abandoned vnlley is comparatively broad, is filled with alluvium, and has gently-sloping grass-covered walls. The present valley parallels the old course aVout -? mile to the south, and hap a narrow floor and s teei> walls. The diversion of the stream from the segment of its old valley was probably caused by an ice block. 4. In the Golden Valley quadrangle, a low sag extends from the valley of Spring Creek, near Dodge east for about 4 miles, to rejoin Spring Creek near the mouth of Goodman Creek. Tho sag is filled with deposits of the fills Qsd£ and Qsd. Between the ends of the sag the present valley of Spring Creek is narrow and its walls aro " steeper than they are in the rest of the area. The sag probably represents the former valle y of Spring Creek, The diversion was probably oaused by glacier ice- either an active lobe or a stagnant block - that forced Spring Creek out of its old valley and into its present course. The diversion occurred prior to the deposition of the early "Wisconsin fill Qsdg and is therefore pre-Tazev/ell, proUibly

lowan. 5. In the southern part of the Hazen quadrangle, in sees. 22 and 23, T. 144N., R. 87V,., the floodplain of the Knife River narrows to less than J; mile in width. Bedrock is exposed in both the northwest and southeast valley walls and is overlain in the southeast wall by f.ray sand (Qsd). The contact between the sand and bedrock dips southeast away from the present valley and disanpears below the floodplain in sec. 26. ADparently, the former course of the Knife River was through sees. 26 and 25, and the present course is the result of superposition from the fill across a bedrock spur. 6. In the eastern part of the Hazen quadrangle in sees. 24 and 25, T. 145N., R. 86W.. and in sec. 30, T. 145N., R. 85V/. f are too abandoned valley segments of Coal Creek, one north and one south of the present stream. The floors of these "in-and-out" channels are 30 to 45 fe-et above the present stream. Wiether the channels represent the former valley of Coal Creek or whether they were temporary diversion channels formed during some glacial advance could not be determined. 7. In the Stanton quadrangle east of the Missouri River and sub-parallel to it is a broad topographic sag about 8 miles long and 2 miles wide. Both ends of the sag join the Missouri River trench; the northern end about 3 miles south of Riverdale, and the southern end about 3 miles northeast of Stanton. The sag is largely filled by deposits of stratified drift (Qsd3 ) and post-glacial alluvium. The size and geographic position of the sag suggest that it was once the course of the Missouri River. It may mark the original Missouri trench formed by the Illinoian (?) diversion. The sa r was

probably blocked by ice during some post-Tllinoian glaciation, and the Missouri River v;as forced to cut its present channel. Drainage pattern of the Knife River area ALIGNED DRAINAGE Many of the tributary streams of Spring Creek and the Knife River have sub-parallel courses that trend northwest or southeast. The tendency for small streams in the Great Plains to be aliened has been noted by numerous workers and several hypotheses have been

suggested to explain the phenomenon. Hypothesis of regional tilt Pierce (1936, pp.72-73) noted an alignment of the northwestflowing tributaries in Rosebud and Custer counties, eastern Montana, Other features in the area led Pierce to conclude that the region had been tilted to the northwest and that this tilt favored the development of the northwest-flowing streams. Hypotheses involving wind action Russell (1929) noted a strong tendency among the small streams of western South Dakota to be aligned northwest-southeast. From Russell*s description and from an examination of aerial photographs of Haakon County, South Dakota, I judge.that the aligned drainage of South Dakota is different from that of eastern Montana. In Montana the alignment is orude, merely a tendency for the northwestflowing streams to be parallel or sub-parallel. In South Dakota, on the other hand, the alignment is remarkably good and affects the southwest-flowing as well as the northwest-flowing tributaries.

Moreover, the streams, besides being aligned, are nearly straight and are spaced at regular intervals. Russell analyzed several possible explanations for the aligned drainage of western South Dakota and Nebraska and concluded that structural control, regional tilting, or erosion along the strike of differentially resistant beds were all unlikely hypotheses* He suggested that wind action was th probable cause and postulated that western South Dakota had once been covered by a series of southeast-trending longitudinal dunes. Small streams flowing in the interdune depressions incised their drainage into the soft bedrock and have maintained their courses even after the removal of the dune sand. ' Baker (1948) agreed with Russell that the aligned drainage in western South Dakota was probably caused by the wind, but proposed an alternative mechanism which he called the yardang process. Baker thought that the prevailing strong winds from the northwest excavated elongate deflation hollows whose long axes trended northwest-southeast, and that the streams had gradually integraced the closed depressions into linear valleys. Crandell (1951, pp.120-130) studied aligned drainage in the vicinity of Pierre, South Dakota, and concluded that both Russell's and Baker's mechanisms may have operated in that area.

of structural control Neither wind action nor regional tilt is a likely explanation for the drainage pattern of tho Xnifo River area. First, the pattern of tho Knife River drainage does not resemble that of western South Dakota. Although many of the tributaries of t'ra Knifo River and Spring Creek aro somewhat aligned, they aro neither straight nor regularly spaced, ana they do not head in uplands that show a striking northwest-southeast pattern of eolian features. Second, although the alignment in the Kr.ifo River area superficially resembles that described by Pierco, it differs in that southeast-flowing as well as northwest- floe-/ ing streams show tho alignment. Third, there are numerous exceptions to the aligned drainage. 'Several good-sized tributary streams flow southwest, at right angles to f the alignment, and any explanation of the drainage pattern must account & for these exceptions. The structure-contour map (pi, 5) shows a definite coincidence bot*:oon drainage pattern and structure in the Knifo River aroa. Both the rralor streams and their tributaries appear to be flowing along the axes of synclines. The largest of the synclines trend east-west and are followed by the Knife River and Spring Creek. Most of the aicaller n crossfolda synclines trend northwest and are followed by the major tributary streams such as Beaver Creek, and Brush Creek in the Medicino Butte quadrangle. A few synclines trend southwest and are followed by tho noxcoptionsn to the aligned drainage pattern, notably Schaffner Creek in the northwest part of the Broncho quadrangle and a large tributary of Antelope Creek in the central part of tho Hazen quadrangle.,

Even the Pleistocene diversion valleys have a fairly consistent relation to the structural axos 0 Those diversion valleys that follow tho courses of pre-existing streams, for example the South Fork and Elm Creek trenches, follow the axes of synclines. Those trenches that aro thought to beice-marginal and that do not follow a pre-oxistino valley, for example the Beulah and Golden Valley trenches, cut across the structure. The northern part of the Goodiaan Creek trench follows a pretrench valley and also follows a syncline; the southern 3 mile segmsr.t of this trench may be ice-marginal and'this segment cuts across a sir.all structural dome. The coincidence of drainage and structure in the Knife River area

is not perfect, but in general the agreement is remarkably close. I suggest, therefore, that the drainage- pattern represents an adjustment of streams to the small bedrock structureso The adjustment protcibly started in the latter part of tho Tertiary period, when the Great Plains changed from an area of deposition to one of erosion. As tho streams, both large and small, began to carve valleys they encountered beds of varying resistance in the gently folded Tertiary rocks. "When jt downcutting stream met the top of one of the more resistant beds - in most cases a 'sandstone - tho stream tended to migrate laterally down dip in accordance with Gilbert's principle of monoclinal shifting. The lateral migration continued until the stream reached the axis of a synclinoo Hare the stream was trapped and the position of its valley stablisod.

CAPTURE OF BARBED TRIBUTARIES Several of the tributary streams of the Knife River area flcv? southwest along the axes of synclines. These streams aro barbed with respect to the major oast-flowing drainage. They are also at a disadvantage with respect to southeast-flow ing streams, which .join the Knife Rivor farther east and have steeper gradients* As a result, the headwaters of tho barbed -tributaries are subject to capture by the south:~ct flcwing streams. Small scale piracy of this sort is talcing place todny at two localities - at the headwators of Schaffnor Creek in the Broncho quadrangle, and at the headwaters of Coal Creek in the Hason quadrangle (Plo 1). Schaffner Creek heads in the upland in the northern part of the Broncho quadrangle and flows southwest down the axis of a small synclinc to join the Knife River in sec. 15, T. 1A2N., R. 91W. The headwaters of Schaffner Creek are actually closer to the Knife River at Broncho damsite than they are to the mouth of Schaffner Creek. Moreover, the Knife Rivor is 4.0 to 50 feet lower at the damsite than at its junction with Schaffner Creeko Therefore, the southeast-flowing tributaries of tho Knife River have steeper gradients than Schaffnsr Creek and will ultimately bohead that stream, A small part of this capture has already been effoctod in the NE*, seo. 30, T. L43N, 90W 0 , where a southeastflowing stream has worked its way headward and has captured a 1 mile segment of a tributary of Schaffner Creek. A similar situation exists in the central part of the Hason quadrangle about 2 miles northeast of the Dakota Star Mine, The major stream of tho area flows southwest along the axis of a synclino and Joins Antelope Creek in tho western part of the quadrangle<> This ' 219

southwest-flowing stream has a much gentler gradient than Coal Creek, which heads in the same area and flows southeast into the Knife River near the eastern edge of the Hazen quadrangle. Coal Creek has already captured part of the west-flowing drainage, in sees. 10 and 15, T. 14-5., R 0 86W 0 , and will in the near future completely behead the tributary of Antelopa Creek* Thickness of fill in the major valleys The bedrock floors of major valleys of the Knife River area have been explored by drill holes at only 3 localities, Garrison dam on the Missouri River, Broncho damsita on the Knife River, and Beulah on the Knife Rivero At Garrison dam numerous holes drilled by the U. S. 'Army Engineers show that the bedrock floor of the Missouri River is 100 to 125 feet below the modern floodplain. Except for a few closed depressions (potholss?) the floor has little relief and seems to be about as wide as the combined widths of the modern floodplain and 4-0 foot alluvial terrace. Excavations in the terrace show that the fill in this part of the trench consists largely of till and stratified drift of Pleistocene age. Beneath the floodplain, however, an unknown thickness of post-glacial alluvium overlies the Pleistocene deposits. The contact between alluvium and drift is hard to determine from drill hole data, but the alluvium may be as much as 75 feet thick* At the site of Broncho dam on the Knife River holes drilled by the U. S. Bureau of Reclamation show that across most of the valley the average depth to bedrock is about 115 feet, but near the center the bedrock surface drops about 30 feet and the fill is about 14.5 foot thick 0

No similar doep channel was notod in the drilling at Garrison dir. or at Beulah, and the presence of thds inner channel at Broncho is thcroforo .anomalous. It is probably a local scour channel. The fill overlying the bedrock floor at Broncho damsite consists of Pleistocene fluvial and glacio-fluvial deposits overlain by Recent alluvium* The contact is hard to pick from the drill hole data, but the Recent deposits are probably not more than 4-0 feet thick, A few holes have been drilled in the Knife River valley rr.ar Boulah. Two of these were exploratory water wells; the rest wore drilled by the Knife River Coal Company in search of buried coal beds. The rather sketchy data from those holes indicate that the bedrock floor of the vr.lley is 100 to 110 feet below the floodplain. No deeper channel was ciscoverod, f but the holes are too far apart to eliminate such a "possibility. Mature topography of headwater areas Several of the larger tributaries of the Knife River flov in valleys that are broader and more mature near their headwater areas than farther downstream. The best examples of this typo of valley *ro Willow Creek in the Broncho quadrangle, Kinncsian Creek in the southern -"rt of the Hazen and Stanton quadrangles, and an unnamed tributary of Antelope Creek in the central part of the Hazen quadrangle. All three streams have the following features in common: l) The lower parts of their valleys have comparatively steep sides and narrow floodplains, 2) Tha upstream parts of their valleys have gently sloping walls and broad alluvial floors one-oighth to one-half mile wide, 3) Tha profiles of the alluvial floors appear to coincide with the alluvial terrace in the Knife River valley and with the 4,0-foot terrace in the Missouri River valley.

The significance of the nature topography in the hoaciv'ter aro"= i.: not clear, and the data at hand are not sufficient to *u3tify any dofir.lt: conclusions. The following hypothesis seems plausible, but is probably only one of several possible explanations: The headwater areas may record an erosion cycle during which mature

topography was develope'd over the whole area. Uplift and rGJuvsr-i-icr. have enabled the streams to carve a more youthful topography clo23 to the larger valleys, but dissection has not yet reached the headwatoi areas and there the old topography still persists. The mature topography may have been produced during the development of the transverse brood valley profiles of the major streams, or may post date this stage. Mass-wasting process v / The picking up and transporting of surficial materials by wind, water, and ice are known as erosion. Surficial material also moves downslope under the dominant influence of gravity aided by the lubricating action of water or ice included in the material. This downslopa movement is called mass-wasting. Sharpe (1938) reviewed <°.ll the previous literature on mass-vasting and suggested a new classification. The subdivisions of Sharps's classification are based, first on the rate of movement of the material, second on the amount of included water, and third on the type of material involved,, Sharpe's primary subdivisions, based on the rate of movement, are as follows; I. Slow flovage phenomena: rock creep, talus creep, soil creep, rock glacier creep, and solifluction. II. Rapid flowago phenomena: earthflow, mud flow, and dobrisavalanchoo "

Ill, Sliding: slump, debris slide, debris fall, rr.okclide p.r.d rock fallo Subsidanco: the sinking of around over mines, caves, etc. Five types of mass-wasting, embracing all four of Sharpens estop, or If-3, have been noted in the Knife Piver ares* These are! subsidence, soil creep, earth flow, and two types of slump,, SUBSIDANCE Sharpe (1938, p. S3) defines subsidence "mover.© rt in wMch t>oro is no free slide and the surface material is displaced vertically downward with little or no hori/ontal component," TMs type of rrovaTrrt occurs principally over areas where there has been subsurface removal of material, either natural or artificial. In the Knife Piver area subf sidance has occurred over underground mines, where a coal bed h°s been removed and the roof has collapsed. The largest area of subsidance is over the underground mine 2 to 3 miles north of Beulih. In tMs area more than lr square miles are pockmarked by pits, 10 to yardc across and roughly circular In plan, marking places where the thick Peulah- Zap coal bed has been mined out underground. The entire roof of the mine has not collapsed, just certain parts of it and between tv.e pits are zones of uncollapsed strata. SOIL CREEP Soil creep is probably the most common and certainly one of the most widely recognized types of mass-wasting. It is the dominant mass-wasting process in all parts of the Knife F.iver area except the small tracts of badlands adjacent to the large rivers, where most of tho loos** is removed by alopewash. Material moved by soil creep forms tMck deposi'ta of colluyium at the base of some stoop valley wallso

EARTHFLOW An earthflow consists of a mass of sand, silt, arid clny, with or without small amounts of coarcar rock debris, and fairly wall snturn ted with water. It is the slowest moving of Sharps T s rapid flcwage phenomena (l°3#, p. 50), An aroa affected by an earthf! ow can be divided into 3 parts: At the head is a zone of small slump blocks; in the canter Is a zone of budded and fissured material; and at the toe is a mass of huranocky material that is the earthflow proper. Failure by earth new

is a common type'of mass-wasting in the Knife "River aroa especially on steep sod-covered slopes (pi, 25A) SLUMP BLOCKS Sharpe (193, 64) defines landslides as "the perceptible downward f sliding or falling of a relatively dry mass of earch, rock, or a mixture of the two," Of Sharps 1 s various types of landslide the only one in which the material moves more or less as a unit with little jumbling or mixing is the slump. Sharpe's definition of a slump 1s as follows (p £5)* "...the downward slipping of a mass of rock or unconsnT1 dated

material of any si.7©, moving as a unit or as several subsidiary units, usually with backward rotation on a more or less horizontal axis parallel to the cliff or slope from which it descends. H The surface of failure and movement of these slump blocks 1s typically curved and is concave up. In tho Piorro area of South Dakota Crndell (l°5l, p. 232 and 233) observed that the rotational typa of slump 1s vnry common but that there are also large blocks of mterial that have moved

essentially as a unit and show no rotational effects. He pointod out that a strict intorprotation of 3harpe f s definitions would leave thccc blocks unclassified, and proposed that they be included under the toni Oslump.° Rot.''.tin -.r-.l slir-r.3 "Normal" or rotational-type slump blocks are common in the Missouri River trench in places where the river has recently undercut the stoop high valley walls. In some pieces the material has had a froe fall and should probably be classified as a debris-fall (Sharpe, 1933, p 0 75). Rockslide slumps Sharpe defines a rockslide as (p 0 76): n .. 0 the downward, usually rapid movement of newly detached segments of the bedrock sliding on bedding, joint, or fault surfaces or on any other plane of separation." Although it is not an essential part of his definition Sharpo's pictures and descriptions of individual rockslides show that he considers this type of movement to involve jumbling and breaking up of the ma tori-1 0

In many of the small tracts of badlands in the Knife River area are examples of a type of landslide block that does not fit into Sharpo's classification. These blocks have moved downslope essentially as a unit with little or no jumbling of the material within the block, and by vising Crandell ! s definition can be classified as a variety of slump 0 These blocks have slid partly on bedding planes and partly over a previously existing topographic surface and, therefore, they also partly fit Sharped definition of a rockslide 0 In the absence of a more suitable term, I shall call these blocks "rockslide slumps n

A rcckslide clur.p is a landslide in which a block of bedrock detaches itself frcr* a hill ar.d slides as a unit down and away from its original position across a previously existing surface. The end product is a "block in which the "bidding or other internal structures have remained undisturbed "but which has moved downslope and has been rotated away from the parent hillside. In all the rockslido slumps of the Knife River area the block that moved was originally the nose or end of a badland spur. The surface on which the block slid was partly a bedding plane at the base of the spur and partly a slopowash-covered pediment that surrounded the spur. Apparently the block*first moved on the bedding piano and slid out onto the pediment. The bedrock of the spur probably had a slight original dip in the direction of ultimate movement; otherwise the sliding could not start on the bedding plane. Apparently this original dip need not be great, for rockslide slumps have been observed where the bedrock dips at 1° or Iess 0 The rockslido slumps in the Knife River area are thought to develop in the following manner: l) The edges of the upland or the sides of a comparatively youthful vclloy are dissected into badlands 0 Long narrow spurja develop between small stoop-sided ullios. As erosion continues, many of the badland spurs bocoma bordored by pedimont-liko planation surfaces covered with a thin sandy deposit of slopewash* 2) Continued erosion narrows the spurs As long as the crest line of any spur is not <ieeply notched no landslide occuroj gradually, however, tho crest line

is differentially lowered until a notch is cut nearly as dc;r as the base of the spur 0 3) Aftor the end of the spur has been thus isolated, it is able to detach itself and rove downslope as a unit block, Tho block moves first on a bed of saturated clay at or roar its base. Then, as it leaves Its original position, it rides out acrocs the pedir.ent and is tilted away from the main mass of the badlands. The dip of the strata in the slump block approximately coincides with the slope of the underlying pediment surfaceo

There are several good examples of rockslide slumps in the Kn'fe River area but only one of these shows anything but the end product, thpt is a block tilted away from the parent mass and resting on a pediment some distance away from the parent spur. The exception is'in the NH-£, SV-i, sec. 34, T. 147N 0 , R. 87*T., where a block about 150 feet long and 4.0 to 50 feet high slid northwest over a pediment surface for a distance of about 100 yards. The beds in the block dip about 10° to the northwest and have been dropped about 25 to 30 feet below their original position,, Subsequent erosion has cut below the old pediment surf ace, *and the base of the slump block resting on the surface of the old pediment is now well exposed in cross section. Two photographs of this slump appear in pi. 26, Detailed examination showed that between the block and the undisturbed bedrock below the pediment is a zone, 1 to l£ feet thick, of Jumbled sand, silt and clay, which probably represents the saturated material on which the block slido The sand was probably derived from the slopewash that covered the podimont and the clay from the saturated cliy bad that was the initial surface of movement. Unfortunately this exposure soon will be covered by waters impounded behind Garrison Dam and there be no further opportunity to study this particular rockslide slur.p.

STRUCTURE General sta torrent Most of western North Dakota is in tho Williston Basin, a large shallow structural basin whoso only definite bouncb.ry is on its southwest flank whore it terminates against the Cedar Creek anticline. Tho general configuration of the Williston Basin in North Dakota can be soen on Ballard's structure contour map (fig. 1/0, which has generalised contours drawn on top of the Dakota sandstone. This nap shows the cantor of the Williston Basin to be about 50 miles west of the Knife River area, but contours drawn on top of the Tertiary beds will probably show a center farther east 0 The lowest altitudes of the base of the Goldon Valley formation are in the western part of the Broncho quadrangle in the Knife River valley, in the southwestern part of the Golden Valloy f quadrangle in the valley of Spring Creek, and just northwest of the Knife River area near the town of Elbowoods* These points, all of which are in the bottoms of small synclines, indicate thnt the center of the , Williston Basin as contoured in the Tertiary strata is probably near ;, r ' the northwestern part of the Knife River area 0

c - ' . '

The general location of the Williston Basin imposes a northward*' y ; regional dip of 10 to 20 feet to the mile on most of the strata in '

Y: southern North Dakota. However, superimposed on the regional structure are numerous small domes and synclines that interrupt and in places reverse the regional dip. In no part of the basin are the beds actually flat-lying, although they may appear so in the individual outcrop* Structuro contour mat) A structure mpp of the Knife River area is shown in plo The contours are drawn on the base of the Be\ilah-Zap coal bed, which Underlies and is intermittently exposod over about two thirds of tho nron 22S

Where tho Be-ilah-7ap hod is not exposed bociuse of burial, erosion, or stratigraphic "pinchout", elevations were read on other bods and were corrected to the horizon of the Beulah-Zap hed 0 The accuracy of the contours varies greatly in different parts of the area 0 In the places where Ihe Beulah-?Ap bed is well exposed the contours are v/ell control-od. Elsewhere, however, the accuracy degenerates because of poor exposures, uncertain correlation of coal beds, or because of known variations in tratipraphic intervals between beds. Folds The structure map of the Knife River area shows that the strata are gently folded into snail domes and synclines. Neglecting the effect of the small local structures, in the southern part of the 'area the beds dip north to northwest at an average of about 15 feet to the mile; in the northern part of the area they are nearly flat with perhaps a slight dip to the west The larger local structures are synclines and elongated domes whose axes strike nearly due easto The synclines are followed by the valleys of Knife River and Spring Creek. Smaller folds trend in various directions but the majority seem to strike about north 30° to 45° west. The maximum closure on the smaller domes seems to be ?bout 1+0 to 50 feeto Faults Small superficial faults due to the local slumping are common in the badland areas near the major streams but these have not been shown on the map 0

Two possible faults arc shown on the geologic r.cp of the Broncho quadrangle, but even these may be due to local slumping. The first locality is in sees, 25 and 26, T. 11N 0 , R. 90tf. Here a few poor exrof on the northeast side of a small ridge show the Fort Union formation eli 15° to A5 NE Whether the dips were produced by landsliding or whether .they actually reflect a small fault in the bedrock could not be determined o The second locality is in the NW, NW, sec 0 2, T 0 12M., R. o where the beds of the Golden Valley formation have been dropped about 10 to 15 feet on the west side of a small fault. As in the case of the firs'o fault, it was not possible to determine whether this was a local slump block or whether it was truly a bedrock fault, A third possible fault has been shown on the Stanton quadrangle rap, in the southern part of sec. 11, T. 144N., R. S£U 0 Here the clinker of the Stantcn coal bed hps been dropped on the south sidd of a hinge fault in which the amount of displacement increases to the west. Again, it was not possible to rule out large scale slumping as a cause for the displacement. Structures of Pal, eocene age Deformed beds at Garrison Darn In the southern part of sec. 4, T. H6N 0 , R. 84.U., the excavation

for the spillway of Garrison Dam temporarily exposed in 194-9 a zono of faulted. and folded Fort Union beds overlain unconformably by flat lying Fort Union strata. The cause for the deformation is not apparent but it clearly took place within the Paloocene epoch.

The deformed zone is 1,000 to 1,200 feet wide; its length w.'is impossible to determine but probably does not exceed 2,000 footo Although some of the faults go deeper, the intensely defamed strata aro only 35 to 4-0 feet thick. At the edges of the zone the beds are gently foldod and offset by faults that strike 50 W. to M. 60 \-J. Near the middle of the zone, the beds are not folded but are so highly fractured and faulted thnt this part of the zone consists of a jumble of blocks of clay, siltstone, and lignite with no apparent preferred orientation. The intensely deformed strcta rest directly on a 2% foot lignite bed that is very gently folded and cut by only 2 or 3 faults that have vertical displacements of 6 inches to 2 feet. The deformed strata and the underlying bed of lignite are shown in plates 25A1 and 27. Closely spaced drill holes indicate that below the highly faulted zone the Fort Union formation has been warped into a shallow basin that has a closure of about £0 feet in 1/4. miloo The long axis of this basin is northeast, nearly at right angles to tha strike of the faults. The basin persists at least to a depth of 200 feet, which is the limit of the drill holoso The upper boundary of the deformed zone is an erosional unconformity, above which are £lat-lying silty sandstone beds of the Fort Union formation. -The maximum relief of the unconformity is about 6 foet a The silty sandstones-are in turn overlain by till with no crumpled beds or other evidence of ice-shove at the contact Tha exposures clearly indicate that, whatever the cause of the . folding and faulting, the strata were deformed during the Paleocone epoch. This automatically rules, out any mechanism involving ico-shove a

'Two possible causes of the deformation, neither of wine 1- "3 cr.t'.roly satisfactory, aro as follows: (1) Local landslidjng along the margins cf a Paleccer.e stream chsnnol, There is no direct evidence in favor of this hypothesis cr.d there 'ire several objections to it. First, no exposures in the spill*'ay cut snov any evidence of a deep channel in the Fort Union bods. Sccon-'l, the movement of the beds sons to have boon fron both sides tcvard the confer, where the fracturing and fn.ultinr, ~o::t intense. Third, although the intense deformation affected only about AO feat of strata, several of the faults extend deeper and cut the underlying 2>f foot lignite bod (2) Compressional stresses related to the formation of the underlying synclinal basin. This hypothesis assumes that the basin originated during the Paleocene epoch, but primary cause for the forr,r.ticn of the basin may have been either local sinking duo to dlfferar.tiil compaction or I'olding due to horizontal compression. As the basin sr.nk the beds on the limbs were slightly compressed toward the center. lelow the

£2 foot lignite bed the strata were confined by the weight of overlying beds and were gently folded. Above the 2-J- foot lirdta the strata were not confined and were probably saturated, so th.it the;- vera easily pushed toward the center of the basin. Actually the novener.t of the beds may have been due partly to compression and partly to covr-ii? sliding under tho influence of gravity<> One obvious weakness in this hypothesis is that the strike of the folds and faults is nearly at right armies the long axis of the basin. Nevertheless, it seems crore likely -th"t the deformation was related to the basin-producing forces than to Ic-crJl landsliding o

Subsurface evidor.ce of Paleocene warping The Cannonball marine formtion crops out in the Missouri Rivoi- trench as far north and vest as the Vashburn quadrangle, which adjoins tho car L.-;rn edge of the Stanton quadrangle. At the town of Washburn, about 10 r.ilo east of tho Stanton quadrangle, the upper contact of the C.innonbn?". fc;:.i- tion has an altitude of 1,74,0 foot. At the eastern edge of tho Stanton quadrangle the top of the Cannonball is below the flood'plain of the Missouri River, which has an altitude of 1,635 feet. At Garrison Dam at the north edge of the Stanton quadrangle the core from a drill hole on the weat sido of the Missouri River shows the top of the Cannonball at 1,430 feet above sea level. There is little chance that the top of the Cannonball fo mat ion, was misidentified in the drill hole, for the core was in'excellent concLL~ tion and the top of the Cennonball was picked independently by Co R. Golder and G. W. Prescott of the U. S. Engineers and R. W. Brown and myself of the Geological Survey,, Thus, between Vashburn and Garrison Dam, a horizontal distance of about 30 miles, the top of the Cannonball drops about 300 foot* The structure map of the Knife River area supplemented by reconnaissance observations in the Vashburn quadrangle indicate that the structural relief of the Fort Union beds between Washburn and Garrison does not exceed 125 to 130 feet. Therefore, there is a discrepancy botveen the t dip of the top of the Cannonball formation and the dip of the coal beds in the Fort Union formation that amounts to 175 feet in a distance of about 30 miles. The discrepancy could be due to marine offlap, and tho lew or Fort Union beds at Garrison Dam may be the time equivalent of the upper r Cahnonball bods at T>rashburn. Tho discrepancy could p.lso be due to

tho sinking of a basin during Fort Union tire, with the area armour.d Garrison Dan sinking faster than that armour.d Washbum, The latter hypotheci3 is supported by the following facts: (1) Ballp.rd's structure nap of tho Williston Basin (fig. 14) shews about 3nO foet of relief on the top of the Dckota sandstone between Vashburn and Garrison Dan. This coincides vith the dip of the Canncr-Vcll Fort Union contact and suggests that part of this dip had been acquired before the coal beds of the Fort Union formation were deposited. (2) In the Knife River valley in the eastern part of the Hazon quadrangle the interval between the Beulah Zap coal bod and the Stanton coal bod is 90 to 100 foet. In tho Missouri Rivor valley in the northeastern part of the Beulah quadrangle the interval bctweon the Beulah Zap coal bed and the Garrison Crock coal bed (which is either the samo as the Stanton bed or within 10 feet of the same stratigraphic horizon) is about 165 feet. This indicates a 65"foot northward thickening of the beds between the Beulah-7ap and Stanton-Garrison Creek beds, and is most easily explained by assuming that the northern part of the Knife River area was sinking more rapidly than the southern part during the deposition of the Fort Union formation. (3) The thickness and character of the Fort Union formation indicatea that basin sinking must have occurred during deposition of the bcd3 0 The Fort Union formation is thought to have beon deposited in coastal plain swamps, yet the forma.tion is between 500 and 1,000 foot thick. Therefore, North Dakota must have been slowly sinking during tho Palcocoro epoch, and it is unlikely that the amount or rate of this sinking would have bcon constant over the whole

It scams possible, therefore, that rainy of tho subsurface structures voro developed during the Paleocone epoch. If this is truo, many of the smaller surface structures may be reflections of larger subsurface structures and may increase rather th£n die out with donth.

Episodes of Tert.1?ry doforrqtion in southwest North Dakota The relationship between the formations exposed in southwestern North Dakota indicates at least 3 episodes of tilting or local warpinj in early Tertiary time. The first episode occurred during the Paleocene epoch. It is recorded by the deforced Fort Union strata at Garrison Dam and is inferred frcr.i other data. The deformation consisted of local warping and basin settling accompanied by minor folding and faulting* The second episode was post-i'asatch (early Eocene) and pro-Oligoceno and is recorded by the unconformity at the baso of the Uuite River formt tion. The deformation seems to have been c':iefly differential uplift and tilting away from the Black Kills dome, because the './hito River formation lies on successively older formations when traced from North Dakota south toward the Black Hills 0 The third episode occurred after the Oligocone epoch but before the development of the Miocene (?) or Pliocene (?) planation surfaceo The deformation consisted of the folding of all Tertiary formations into snail domes and synclinal basins. Evidence of this episode is most clearly seen in the synclines of the Little Badlands in Stark County and of the Chalky Buttes in Slope County. In these areas White River bods involved in the folding now have dips up to £° or 5°. The structural historyof the latter part of the Tertiary period and

of the Quaternary period includes several pulses of regional uplift, but there is no evidence that any of these pulses was accompanied by folding or involved differontial uplifto

GEOLOGIC HISTORY The geologic history of southwestern North DnVota in general and the Xnifo River area in particular as read from surface rock starts in late Cretaceous time and continues to the present. The major events are summarized as follows: (1) In late Cretaceous time most of North Dakota was covered by marine waters. In these waters were deposited first Pierro shale, then, as the sea gradually shallowed, the Fox Hill sandstone. Further shallowing of the sea gave rise to the brackish water Colgate sandstone member of the Fox Hills formation. (2) Retreat of the late Cretaceous sea and erosion of the upper surface of the Fox Hil!b sandstone. (3) Deposition in latest Cretaceous time of the fluvial Hell Creek formation,, The Hell Creek formation was deposited partly in swamps but mostly on floodplains<> During this time the sea advanced briefly into southern North Dakota and the brackish water Breien rsmbe: of the Hell Creek formation was deposited. (4.) A great change in the flora and fauna of the region occurred at the end of Hell Creek deposition. This change marks the end of the Cretaceous and the beginning of the Tertiary period, but was not accompanied in North Dakota by any major physical event. (5) Deposition, in low-lying coastal plain swamps, of the various members of the Fort Union formation (Paieocene). During early Fort Union time the sea advanced, probably from the northeast, across central North Dakota. In the sea was deuosited the

Cannonball marine formation, which intertongucs with tho lower p.'rt of tho Fort Union formation. This is tho last advance of tho sa across North Dakota. (6) An interval at the end of the Paleocena epoch during which western North Dakota remained low in altitude but received little or no deposits. During the interval the source areas to tho west were deeply weathered and deposits of kaolin clay were formed., (?) Erosion of- the kaolin clays to the west and the rodoposition of these clays in North Dakota to form the lower member of the Golden Valley formation (early Eocene). This was followed without apparent break by the fluvially deposited upper sandy member of the Golden Valley formation. (8) A period of uplift and erosion. The uplift appears to have been both regional and differential the Black Hills dome rose more* than the area north and east of it. After the uplift erosion removed large amounts of the Golden Valley and Fort Union formations from North Dakota 0 (9) Fluvial deposition of the Oligocene White Rivar forrr/ition. Large amounts of reworked volcanic ash in this formation indicate volcanism in the mountains to tho west. (10) A period of folding and warping followed or accompanied by uplift and erosion. Much of the White River formation was ror.ovcd from southwestern Ncrth Dakota and a planation surface capped by 3 to 5 feet of gravol was developed. This planation surface is

probably the equivalent of tho FlaxvilJLe plain in eastern Montana and if so was probably cut during the Pliocono opoch.

(11) Uplift and dissection of the gravel-capped planation surface and the development of the broad, gently rolling plain that today forms the uplands of the interstream divides. (12) Dissection of the broad upland surface to form the gently sloping broad valley profiles that flank the rmjor stream valleys. By analogy with events in South Dakota this dissection probably occurred in the early Pleistocene, possibly during the Yarmouth interglacial stage, (13) An advance of some pre-Wisconsin ice sheet (lllinoian?) southwest across the site of the Missouri River trench* Tho ico sheet diverted the waters of the northeast-flowing streams, turned them southeast, and new channels wero cut at or close to the- r.ni-~-: of the ice. The major sets of diversion valleys were formed: Tha first set now abandoned, extends from the Killdeer Mountains in Dunn County southeast to the Missouri River a feu miles north of Fort Yatss<> The second set was incised deeply enough to hold the drainage after the retreat of tho ice and became tho Missouri Rivcr trench, (14.) Deepening of the Missouri River tronch and all its tributary valleys to a depth as great or greater than the modern floodplains. This erosion probably took place during the Sangamon interglacial stage* (lp) Advance of the lowan ico sheet as far southwest as the outermost drift border in North Dakota 0 Till, ice-contact deposits and the outwash fill, Qsd? were deposited in the Knifo River arc~ 0

(16) A short interval of erosion and weak soil-profile development following the rotreat of the lowan ico (l?) Advance of the Tazowoll icej deposition of till and of the fine-grained fill, Qsd2 (18) Hotreat of the Tazowell ice, followed by a period of erosion during which tho thick fill, Qod2 wa3 dissoctcd to a doptli below that of the modern floodplains. During this erosion interval the bedrock floors of the Missouri trench and its tributaries nay have been lowered*. (19) The advance of tho Gary ice sheet, an event not recorded in southwestern North Dakota (20) The advance of the Mankato ice sheet across all but tho southwestern part of the Knife River area; deposition of till., stratified drift, and the valley fill, Qsdj building of tho Srora moraineo As the Mankato ice retreated it left a stagnant body of ice in the Missouri River trench causing deposition of a largo kamo terrace (21) Melting of the ico in the Missouri trench and tha deposition of the combined fluvial and glacic-fluvial fill, Qmg*. (22) Readvance of the Mankato ice sheet across tho Missouri trench in the Stanton quadrangle; deposition of till and ica-contact gravel, Qic ! 0 (23) Final retreat of the Mankato loo sheet, followed by an interval of erosion during which the Pleistocene deposits i'ero dissected to depths below that of the modern floodplain*

(24.) A period of aggradation during which tho streams deposited fine-grained sand and silt and aggradod their courses up to and above the altitudes of the modern floodplainc*. (25) Gradual dissection of the older alluvium and development of the modern floodolains of tho strccins

ECONOMIC GEOLOGY The mineral resources of the Knifo Rivor area aro sand and gravel, ceramic clay, and lignite coal. Of theso only the lignite is being extensively e:<ploited. Oil and gas are being produced from other parts of the Uilliston Basin in North Dakota and the producing formations extend under the Knife River area 0 Up to the present tiir.a, however, the only deep well- drilled in the area was a dry holoo Sand and prsvol Nearly all the deposits of sand and gravel in the Knife "River area are of glacial origin. All the larger deposits are in the northern part of the area; in fact most*of the southern part of the area is so barren of gravel that outcrops of the clinker or "pseudoscoria* are the chief source of road metal. Even in the northern part of the area most of the sand and gravel deposits are small and the materials are poorly sorted;' they have a high silt content and are used principally for road natal 0 A few small deposits are fairly clean and aro used for concrete abrogate by local residents. Most of the deposits in the Knife River area con-tain some undersirable materials. In addition to silt, these materials are stringers of reworked and weathered coal, pebbles of weak shale, pebbles of chert, and iron oxide. The various Pleistocene fluvial and glacic fluvial deposits aro characteristically different from ona another, so that any discussion of the sand and gravel resources of the area must logically bo based on the same subdivisions as those shown on the geologic mapo

Early Wisconsin gravel dopes-its (Qwg, Qsg, Qtg, With the exception of tho outwash fill Qsd-, the early Wisconsin deposits of sand and gravel are few in number and small in size. They are 4' confined to the southwestern part of the area beyond tho border of the Mankato drift. The deposits of undotarmined origin, Qsg, and the older glacial deposits, Qwg, are for the most part small in size and very poorly sorted. They contain largo amounts of silt and clay and are only used locally for road metal. The deposits of Qtg aro a little larger and .somewhat better sorted, but these contain a high percentage of chert and flint pebbles, making them unsuitable for some concrete aggregate. The deposits of the old outwash fill, Qsd-,, aro confirod to the Knife River valley near Beulaho In general the gravels'of Qsd~ arc fairly clean and free of silt. The chief fine-grained impurities are stringers of powdered and weathered lignite and secondary deposits of fine-grained iron oxide. Both of these impurities can probably bo removed by washing. A high proportion of the pebbles consist of lir.estcno and dolomite, which may restrict the use of this gravel to certain types of aggregate. A deposit of Qsd is boing mined at the Beulah gravel pit and is used both for road metal and for concrete aggregate. Deposits of the intermediate fill, (Qsd2) Tho intermediate fill, Qsd2, consists largely of modiuia- to finegrainod sand and silt. In only a few outcrops does Qsd2 contain beds of sand and gravel, and even in theso places there is usually a great deal of intorbodded fine sand and silt. There are no gravel pits in

this fill, and there is little chance of discovering any exploitable grr.vol beds in ito Deposits of late Wisconsin outwash (Owo, Qsd-) Deposits of late Wisconsin (l-'ankato) outwash are cordon in the northern and eastern parts of the Knife River area and there Pre mnnero'is p,r~vcl pits in these deposits. Nearly c.ll of the deposits, however, are poorly sorted and contain small to large amounts of silt and clay 0 They aro mined principally for road metal and would require washing before they would be suitable for concrete aggregate. An exception to this genera]l-'cy is found in the outwnsh terrace of QscL in Brush Croek between socs. 9 and 23, T. H3N., R. 8SW. where the deposits consist of r.odiuir.- to coarse-grained sand and gravel, comparatively woll sorted, and with vr-ry little admixed clay or silt. The pebbles are 60 to 65 percent pranito,

20 to 25 percent limestone and dolomite, and 10 to 15 percent basalt quartzito, sandstone, and iron oxide 0 The deposits are large and extensive and constitute one of the best potential sources of sand and gravel in the Knife River area. Ice contact deposits (Qic, Qic 1 ) Several of the large kames in or near the Krem moraine in the northern part of the Knife River area are ninod for sand and gravel. The material in most of the kames is fairly woll sorted and contains no more silt than most of the outwash deposits. The karaes, however, contain a higher percentage of shale pebbles (derived from the Pierra formation) than the outwash deposits.

Missouri River gravels (Qmg) The Missouri River gravels underlie parts of the 4-0-foot terrace in the Missouri River trench and consist of coarsG-gr-inod sand and ravol dorived partly from Canada and partly from the watersheds of the Yellc'..7- stone and Little Missouri rivers. In general these extensive deposits are better sorted and cleaner than any other gravel deposits in tho are-o They contain only small percentages of silt and clay and virtually no powdered coal or limonite. Many of the pebbles consist of chert and agates,

undesirable for certain types of concrato aggregate. However, sir.-.lr gravels are mined from the Missouri and Heart River valleys near Mandan, North Dakota and are used as aggregate after washing and screening. If the Missouri River gravels in the Stanton quadrangle are suitable for aggregate, they are the largest potential source of this commodity in thj Knife River area. Corn-ric cly Some of the beds of plastic clay in the Tongue River member of the Fort Union formation are probably suitable for the manufacture of brick and tile. The beds, however, are discontinuous and much detailed testing , would be necessary to determine their extent and potential uses. The lower member of the Golden Valley formation consists chiefly of kaolinitic clay mixed with varying amounts of quartz and mica. Tho clay ' beds are remarkably uniform both in appearance and composition over tho entire outcrop area of the formation. They are not being exploited in the Knife River area, but are being mined in Morion County by the Hebron Brick Co. and in Stark County by tho Dickinson Pressed and Fire Brick Co. and are used in the manufacture of brick, structural tile, and pottoryo

The fourth biennial report of the llorth Dakota Geological Survey csrl:: with the clay resources of the State and summarizes the results of c.otill'jL. corairdc tests made on the "o.olight-colored Tertiary clays 57 - the Golcbn Valley formation of this report (Leonard, Clapp, and Babcock, 1906, pp. 132-190), The tests showed that nearly every sampled outcrop of tho Golden Valley kaolin beds contains clay suitable for some ceramic use. Most of the sandy clays are suitable for fire brick; the silty to nonsilty clays are suitable for brick and tile; and a few beds of plaitic clay are suitable for pottery. Therefore the lower member of the Golden Valley formation is a potential source of ceramic clay whenever the formation is present and the Golden Valley-Fort Union contact can be used to estimate tho extent and location of deposits of ceramic clry in southuos': North Dakota. Oil and gas possibilities Until the spring of 1951 North Dakota was one of the marginal areas that had not produced any oil or gas. Paleozoic and Mesozoic rocks th<°t have yielded oil in Wycoing, Montana, and Canada were known to undDrlio much of the State, but the few scattered wildcat wolls had produced nothing but salt water. In March, 1951, the Amerada Petroleum Corporation discovered oil near Tioga, a small town about 25 miles east of Williston. The discovery well is on the north end of the Nesson anticline, the largest

known structural.'dome in the Williston Basin a The producing horizon is

apparently in strata of Devonian ago, but later holes in the sair.a area havo encountered oil in the Madison limestone (Mississippian) as wello Tho discovery of oil in the Tioga well has stimulated exploration in other parti of the Williston Basin, and more producing fields will probably bo discovered in the future. 2/6

In addition to tho structural traps, which so far aro the only "o'-r. for tho wildcat wells in North Dakota, there is a possibility that oil in tho Paleosoic formations may occur in roofs or on stratigraphic traps. Futuro drilling will probably determine whether any of those non- structural traps exist in the Williston Basin, The snail structures that are superimposed on the gentle regional dip of the bods in the Knife River area are probably too small to warrant much optimism for oil discoveries unless they reflect larger buried

structures. Drilling in other parts of the Willioton Pasin in the nar future will undoubtedly indicate whether some of the snail structures increase with depth. If they do, the small dones of the Knife Hiver area may well be the targets for wildcat wells. In the summer, of 1950 the Plymouth Oil Conpary drilled a well, tha Kelley-Leutz # 1 well, in the northern part of sec. 2£, To 1£2N., R. 89"'., Mercer County. The hole started in the Fort Union formation about 10 feet below the base of the Golden Valley formation, and was drilled 12,526 fe.:t ; ending in rocks of Cambrian age* The well was abandoned as a dry hole. Detailed information of tho formations encountered in this hole is not available, but a driller's log with the approximate tops of some of the 'formations is given in the appendix. The structure contour map of the Knife River area (pi. 5 ) shows that the Kelley-Leutz v/ell is not on one t/- of the larger structures in the area so that this particular well should not be the cause of too much pessimism regarding future possibilities in tho Coal bodi The coal beds are the chlof resource of the Knife Piver aroa and aro being extensively mined. All of the valuable beds aro in the Tongue RiVer "4* '

member of the Fort Union formation; a few arnll rods occur in the Goldon 2/7

Valley formation but are not persistent or thick enough to warrant development. Physical properties All of the coal in the Knife Hiver area is of lignitic rank and is similar to the lignites described from other parts of TTorth Dakota (see fig. l). The fresh lignite is tough and compact, Xoct of it appears black in gross aspect, but the powder or streak is dark brown. Soir.o of the fresh Lignite has a woody or fibrous texture and a dull lustre. Other fresh lignite is compact, black, and has a vitreous lustre. Some of tho more fibrous layers are quite soft and porous and resemble charcoal. Carbonized logs and plant stems are common in most of the becb. Some of these have a woody texture; others are massive and shiny. Th2 black shiny lignites in general break easier than tha woody fibrous varieties,, On exposure to the air the lignite loses considerable part of its high moi'sture contents This causes shrinking and cracking and tha coal soon degenerates into fine black powder. This characteristic makes it almost impossible to ship lignite in open cars to distant markets 0 The tendency of the lignite to weather to fine powder also nakes it difficult in soma places to estimate the true thickness of the beds, because the pressure of the overlying beds squeezes the powder out into the zone of soil creop<> This causes an apparent thinning of the lignite bed at the outcrop: in many places a bed 4. to 5 feet thick will appear on the weathered outcrop to be only 2 or 3 feet a This lignite contains several types of impurities, the most coirmon of which is silt or clay that forms ash when the coal is burned. Other impurities present in minor amounts are iron sulfide (probably marcacite),

sulfur, and gypsum. The lattor two are probably formed by tho disir-t-o.ration of the iron sulfidcu Parts of sor.e of the lignite beds have been silicified to form a hard, non-conbustible black substance which the miners call "bone." Chemical composition Numerous samples of lignite have been collected from various r.j'noa in the Kni re River area and have been analyzed in the chemical laboratory of the U. S. Bureau of Mines at Pittsburg, Pennsylvania. A representative sample of these analyses is given in Table 1. For the purposes of cor.- parison, a few analyses of coals from nearby fields have been included, as well as one analysis of Pittsburgh bituminous coal. Distribution and correlation of beds BEULAH-ZAP BED The most important cpal bed in the Knife River area is the Beulah- Zap bed. Mercer County is currently the largest lignite-producing county in the United States and all the larger mines are in this bod; in the calendar year 1950 the three large mines near Zap, Boulah, and Hascn produced 1,18,893 short tons. The bed was named and mappod near iho towns of Beulah and Zap by Leonard and Dovo (1925, pp. 125-130). Except whero it has boen removed by erosion, tho Beulah-7ap bed underlies the Medicine Butte quadrangle, all but the southwest part of the Broncho quadrangle, the eastern part of the Goldon Valley quadrangle, all the Boulah quadrangle, and the western half of tho Hazon quadrangle 0 To the west in the Broncho and Goldon Valloy quadrangles the bod thins and probably pinches out. To the south and southeast it extends for unknown distances into Morton and Oliver counties. To the east it thins 2/0

and probably disappears in the eastern part of the Ha sen quadrangle B To tho northeast it thins and may or may not extend into tho Minot area; if it does it is probably Andrews 1 bed B, 90 feet above tho Kintor bed To the north and northwest it extends into tho southern part of the Fort Borthold Indian Reservation, where it was mapped as bod 2 by Pishol and bed EE by Bauor and Horold, bed mapped as "EE* naar the mouth of the Little Missouri Rivor in Dunn County is probably not the same as bod in the extreme eastern part of the Fort Berthold Indian Reservation, uhero it correlates with tho Beulah-Zap bed 0 In the Knife River area the Beulah-7,ap bod is less than 5 foot thic!: only in the Broncho quadrangle; elsewhero it is much thicker, reaching a maximum of 25 feet in the canter of the Beulah quadrangle. Measured sections of the Boulah-Zap bod are shown in figs* 15 - 20, 0 .The Beulah-Zap bed is the most persistent in thickness and quality of all coal beds in the Knife River area and it is the best key horizon in the Fort Union formation. Even where outcrops are poor, the bod has been extensively burned and the clinker is readily traceable . Tho ' correlation of the other coal bods is trrach more difficult Much of the 'area consists of smooth grass-covorod slopos with few or no outcrops* Because of tha great variability in tho lateral persistence of tho beds correlation across these grass-covorcd alopos is difficult and uncertain at best. In soir.o placoo tho accuracy of correlation is coed; in othorc, little more than a reasonable guass. All beds that have boon correlated, howovor, aro probably closo to tho sair.o stratigraphic horison. 7or plo, the Schoolhouco bed in tho southoi-n part of the Knifo River

is actually several beds, all close to the same horizon, but separated by intervening areas of no coal, COAL BEDS ABOVE THE BEULAH-Z&P BED , The principal beds stratigraphically above the Beulah-Zap bed are the Schoolhouse and Twin Buttes beds in the Fort Union formation, and the Alamo Bluff and Schaffner beds in the overlying Golden Valley formation. All these beds are restricted to the southwestern part of the area. Sohoolhouse bed The Schoolhouse bed was named for an .exposure in a small mine near a rural school, in the southern part of sec. 27, T. H2N., R. 89W 0 The bed persists over most of the Broncho and Medicine Butte quadrangles, and has been tentatively identified in the southern parts of the Golden t Valley and Beulah quadrangles. It persists southeast of the Knife River area and may correlate with the thick bed whose burned out portions form the clinker cop of the hills near Glen Ullen in Morton County. The Schoolhouse bed is 4-5 to 50 feet above the Beulah-Zap bed in the eastern part of the Medicine Butte quadrangle. The interval increases westward to about 80 feet near Broncho damsite and 100 feet in the western part of the Broncho quadrangle o In the vicinity of Broncho damsite the Schoolhouse bed splits into 2 parts which gradually diverge westward,, In the vicinity of the South Fork of the Knife River, they are separated ky 25 to 30 feet of sandy shale and clay0 Only in the southeastern part of the Medicine Butte quadrangle is the bed consistently thick and of good qualityo

Twin Buttes bed The Twin Buttes, named for exposures near a pair of small conical buttes in sec. 28, T, 143 N., R. 90W,, is 130 to 150 feet above the Beulah-Zap bed. Over most of the Broncho and Medicine Butte quadrangles the Twin Buttes bed is a thin impure lignite 1 to 3 feet thick with a persistent clay parting near the middle 0 Locally, as in the vicinities of Medicine Butte and Schaffner Creek, it thickens and is a bed of good quality 5 to 7 feet thick. In the southern and eastern parts of the Medicine Butte quadrangle the Twin Buttes bed grades laterally in a very short distance into brown carbonaceous shale. The Twin Buttes bed persists north in the southern part of the Golden Valley quadrangle where it is as much as 6 or 7 feet and is locally mined ' Alamo Bluff bed The kaolin clay beds of the lower member of the Golden Valley forma- / tion are characteristically capped by a thin impure lignite or carbonaceous shale, which locally thickens to a coal bed 2 to 6 feet thick. In the Broncho quadrangle this has been called the Alamo Bluff bed* The Alamo Bluff bed was named for an exposure in a small escarpment near the ruins of an adobe house, facetiously called "the Alamo" by the Knife River party, in the northeast corner of sec. 28, T. H3No, R, 90W 0 . The "3d is too variable in thickness and quality to permit an accurate estimate of its reserves, but it probably has little or no commercial value. The local small pits 2 to 3 miles northeast of Dodge in the Golden Valley Quadrangle are probably in the Alamo Bluff bed, but these pits were 8*\imped in and a section of the bed could not be obtained.

Schaffner bed The Schaffner bed, named for exposures neap Sohaffner Creek, is a small lenticular bed of coal in the upper member of the Golden Valley formation. It occurs about 30 feet above the Alamo Bluff bed and is best developed in the northwestern part of the Bronoho quadrangle. As its average thiokness is only 2 to 2J- feet, it probably has no commercial COAL BEDS BELOW THE BEULAH-ZAP BED Thick coal beds below the Beulah-Zap bed are exposed only in the northeastern part of the Knife River area. Some of these are fairly persistent; others which are locally thick and of good quality, lense out in short distances. Therefore, correlation across areas of poor exposures is difficult and many beds have been given local names, even though detailed drilling might show that they correlate with other beds a few miles away 0 The following discussion will be organized on the basis of local geography. Eeulah area Several coal beds crop out below the Beulah-Zap bed in the vicinity of Beulah. Of these only the Spaer bed and the Hazen B bed appear to persist over any great area 0 Spaer bed The Spaer bed, about 60 feet below the Beulah-Zap bed, persists over parts of the Medicine Butte, Beulah and Hazen quadrangles. The Spaer bod was named from exposures at the Spaer ranch in the NE, sec 0 12, T. 143N., R In this part of the Knife River valley the bed is 2 4 feet thick and is easily identified by a thin cap of silioified

carbonaceous shale 0 Silioified shale also caps the Spaer bod Just northwest of tho Dakota Collieries 1 strip mine near Zap; but east of the mine the Spaer bed has no silioified cap, whereas two local beds, 'ono 35 feet above and one 30 feet below the Spaer, are overlain by a silioified carbonaceous shale. The Spaer bed was not observed in the Knife River valley east of Beulah. In- the Zap and Hazen branches of the Beulah trench a bed 55 to 60 feet below the Beulah-Zap bed has tentatively been correlated with the Spaer bed. HazenlBnbed The Ha2en°B"bed, named for exposures near the town of Hazen, crops out on the north side of the Knife River valley as far West as-the town of Beulah. Between Beulah and Hazen it is 2 to 5 feet thick and about 110 to 115 feet below the base of the Beulah-Zap bed. It is quite variable both in thickness and the presence of clay partings. In addition to the Spaer and Hazen "Bbeda there are 3 or 4 local beds that do not persist laterally for more than a mile or two0 Subsurface beds Drill holes and an old mine shaft at Beulah reveal the existence of thick beds that underlie the area but that do not crop out at the sur- Both of these beds were formerly mined underground by the old Beulah Coal Company; the access shaft was near the present power house at the end of the town0 The upper of the two beds is about 10 feet thick and is 180 to 190 below the Beulah-Zap bed. In the shaft the base of thia bed has an

altitude of 1,751 feet, putting it about £0 feet below the Knife River floodplain. Although this bed ia nearly 10 feet thick at the shaft,drill holes indicate that it underlies less than 2 square miles on- the north aide of the Knife River and is not present on the south side. The lower bed is at an altitude of 1,633 feet, or about 200 to 210 feot below the Beulah-Zap bed. Drill holes show that this bed is 8 to 10 feet thick and is present on both north and south sides of the Knife River valley. flapcm area Hazen "A"bed The lowest bed that crops out near Kazan is called the Hazon Tfbedo It is about 4 to 5 feet thick, and lies 155 to 160 feet below the Beulah- Zap bed. The Hazen'&" bed was traceable only on the north side of the river, and extends from the Gallagher Mine miles northwest of Hazen (locality 357) east to seo. -31, H5N., R. 86W. On the south side of the Knife River the bed exposed at locality 370 in sec. 12, T, LUN-, R. 86W. is probably the Hazen TV" bed. Hazen "bed The Hazen IT bed, about 4,0 feet above the Hazen bed, was identified on both sides of the Knife River valley. It is L, to 6 feet thick and of good quality. East of Hazen poor exposures on both sides of the Knife River valley prevent an accurate tracing and correlation of this bed, it seems to be at about the same horizon as the Coal Creek bed. bed A poorly exposed bed 25 to 4-0 feet above the Hazen bed and 75 to '5 feet below the Beulah-Zap bed bae been named the Star bedo The bed penetrated by he water well at the Dakota Star Mine, where it is

about 10 feet thick. Except for this drill hole the Star bed is exposed only at two other places. Its thickness is apparently quite variable , t and its latoral persistence questionable. It is possibly the lateral equivalent of the Stanton bed to the easto Stanton area The Beulah Zap coal bed pinches out north of the Knife River somewhere between Hazen and Stanton. From Hazen east, therefore, it is more difficult to relate the beds to the horizon of the Beulah-Zap bed. Several coal beds were mapped in the southern part of the Stanton quadrangle, but of these only the Stanton bed can be definitely traced over all of the area 0 Coal Creek bed The Coal Creek bed, named for exposures along the creek of the same name, was traced on the north side of the Knife River valley in the eastern part of the Hazen and the western part of the Stanton quadrangles. It is probably present also on the south side of the river, but except in the valley of Klnneman Creek it is buried by the late Pleistocene glacial fluvial deposits. Where it has been mapped, the Coal Creek bed varies from about 2& to 1+ feet in thickness. It is 35 to 4.0 feet below the Stanton bed and may be the lateral equivalent of the Hazenbedo Stanton bed The Stanton bed was named for its development in the high hills 4 to 6 e miles south of Stanton. The bed has been traced from the eastern part of the Hazon quadrangle apross the southern half of the Stanton quadrangle<> Although it is laterally persistent the thickness and quality of this are quite variable. It is thickest in the high hills a few miles south of Stanton, being 8 to 11 feet thick in the vicinity of the Kamin

Mine in sec, 13, T 0 144N., R. 85W. Vest of Stanton in the Knife River valley the Stanton bed is 4 to 8 feet thick and has a thick roof of soft yellow-gray eand. East of the Missouri River the Stanton bed is 4- to 6 feet thick and of good quality. It has been traced to the eastern margin of the Knife River area, and apparently persists for several miles eastward, into the Washburn area. The Stanton bed is very close to the same strntigraphic horizon as the Garrison Creek bed a few miles to the north. It was not possible, however, to definitely correlate these two beds. On the west side of the Missouri River the Stanton bed pinches out to the north and the Garrison Creek pinches out to the south, and there is a gap between them in sees. 5 and 8, T, H5N., R 84W 0 On the east side of the Missouri River the east-central part of the Stanton quadrangle is covered with thick deposits of till and stratified drift, and the beds cannot be traced across this covered area without the aid of drill holes. Knoop bed The Knoop bed was named for exposures on the Knoop Ranch on the west side of the Missouri River in sees. 8, 9, 16, and 17, T. 145N., R. 85W 0 At its type locality the Knoop bed consists of two beds or "benches" of coal separated by 3 to 4- feet of shaly clay. Traced to the north the upper thine and dies out, but the lower bed persists and can be traced as as sec 0 13, T. I$H* 9 R. 85W 0 Over most of the area in which the' bed is exposed the Stanton bed is either missing or its covered by drift so that it is not possible to measure the interval between them. The best estimate is that the Knoop bed is 85 to 95 foot below the

Stanton bed near the Knife River, but this interval appears to increase northward. In eec 0 30, To K6N., R. 84W., the Knoop bed is 95 to 100 feet below the Garrison Creek bed, and at its northernmost outcrop in seo. 13, T. lA&J., R. 85W., it is 110 to 115 feet below the Garrison Creek bed. If this interval continues to increase to the north it is possible that the Knoop bed may correlate with the Wolf Creek coal bed, which crops out near river level at the site of Mannhaven, Hancock bed In the bluffs on the east side of the Missouri River in sees. 26 and 35 T, H5N., R. 84W., are numerous abandoned adits and small pits in a bed that appears to be 6 to 9 feet thick. The bed could not be traced beyond these two sections,., One of the old adits in the northwest corner of sec, 35 was known as the Hancock Mine, and from this the bed takes its name. The Hancock bed is 150 to 160 feet below the Stanton bed, which crops out higher in the bluffs. If the 85 to 95 foot interval between the Stanton and Knoop beds on the west side of the Missouri River is correct, the Hancock bed is probably not the equivalent of the Knoop bed across the river. If these two beds do not correlate, then the Knoop:

bed does not persist as far east as the east side of the Missouri River trench in the southern part of the Stanton quadrangle 0 local bed 50 feet above the Stanton bed. In the high hills north of the Missouri River trench in the southeastern part of the Stanton quadrangle is a lignite bed 2 to 3 feet thick and 50 to 55 feet above the Stanton bed. This bed could not be traced of this part of the Knife River area and could not be correlated with other bed.

Garrison area Three coal beds have been mapped in the northern part of the Stanton quadrangle near Garrison'Dam. Two of these, the Wolf Creek and Garrison Creek beds, can be traced northward into the Minot area where they were named and mapped by Andrews (1939, pp. 70-72). Wolf Creek bed In the northern part of the Stanton quadrangle, the Wolf Creek coal

'bed lies at or slightly below the altitude of the Missouri River, and the only natural outcrop of this bed is in the cutbank of the river about one mile south of the former town of Mannhaven, The only other exposures of this bed in the area are in the abandoned mine at Mannhaven and in the

excavations at Garrison Dam. The bed at Mannhaven described by Wilder (1905, p. 37) and Smith (1908, p. 23) is the Wolf Creek coal bed. Excavations on the western side of the Missouri River at Garrison Dam show that the Wolf Creek coal bed consists of two beds each 5 to 6 feet thick and separated by several feet of shaly clay. The upper split is the one that was''formerly mined at Mannhaven. On the east side of the Missouri River d excavations and drill holes show that the two splits come together to form one bed about 11 feet thick with only a minor shale parting of 1 to 2 inches in the middle. Northward the two splits of the bed diverge again and near the mouth of Wolf Creek in the Minot area are about 20 feet apart. Andrews (1939, pp. 69-70) called the upper split the Wolf Creek coal bed and the lower split a local coal bed which is 23 feet below the Wolf Creek bed. B Smith (1908) suggested that the Wolf Creek and the Hancock beds were probably one and the same; but inasmuch as several miles separate the known outcrops of these beds and inasmuch as it is known that beds 8 to 10 feet thick can pinch out in less than half a mile, all that can

safely said is that the two beds appear to be near the same stratigraphic horizon. Garrison Creek bed The Garrison Creek bed can be traced from the type locality in the Minot area south into the Knife River area. In the Stanton quadrangle the Garrison Creek bed is exposed in the west abutment of Garrison Dam, where it consists of two beds 35 and 2.2 feet thick, separated by nearly 3 feet of gray clay. Traced to the south the interval between the beds Increases until, at locality 393, sec. 2/., T. 14.611., R. 85W 0 , they are about 15 feet apart. South of this locality both beds thin rapidly and apparently pinch out before reaching the northern boundary of T. 1A5N., R. 84W 0 On the east side of the Missouri Fiver near Garrison Dam tlie splits of the Garrison Creek are much closer together and the shale parting is a few inches to 2 feet thick. The Garrison Creek bed could not be traced south of sec. 22, T. H6N., R. 84W. Kruckenberg bed The Kruckenberg was named for exposures on the Kruckenberg farm in °eo. 2, T. 14,6N., R. 85W 0 It is about 50 feet above the Garrison Creek bed was mapped on both sides of the Missouri River trench in the northern of the Stanton quadrangle. West of the river the bed is 2 to 2% feet east of the river, 3 to 4 feet thick. It could not be traced north °£ the Knife River area. p°33ible thick local bed In the summer of 194-9 numerous water wells were drilled in the n°rthern and eastern parts of the Knife River area. Two of those wells Apparently penetrated a very thick coal bed. One, in the NWi, sec a 2, '. U6N., R 0 85W 0 reported a lignite bed 36 feet thick about 60 feet below

the Garrison Creek bed. The other well, about 1/2 mile north in the Garrison quadrangle in sec. 35, T, H7N., R 0 85W. reported about 14. feet of coal at the same horizon*, Two miles to the east in sees. 1 and 12, T, M6N 0 , R. 85W., the approximate stratigraphic horizon of this thick lignite bed Is marked only by two to three feet of brown carbonaceous shale. Thus if the well logs are reasonably accurate this very thick lignite bed pinches out in less than 2 miles. ftla ckwa ter-ETnme tt area In order to fill in a gap between the maps of the Port Berthold Indian Reservation, the Minot area, and the six quadrangles of the Knife River area, the maps of the Beulah and Ha sen quadrangles were extended north to include the southeast corner of the HLackwater and the southern edge of the Emmett quadrangles. This added area together with the northern parts of the Beulah and Hazen quadrangles will be called for convenience the Blaokwater-Einmett area. Several ooal beds, including the Beulah-Zap bed were mapped in this area. In the western part of the Blackwater- Emniett area, the eastern end of the Fort Berthold Indian Reservation is underlain by three coal beds, the Beulah-Zap bed (bed ffEE ttof Bauer and The two latter beds were originally mapped by Bauer and Herold and the current mapping is in essential agreement their earlier work Three beds of coal, the Garrison Creek bed, a local bed, and the top bed, were mapped in the eastern part of the Blaokwater-ETnmett , Of these only the Beulah-Zap beds correlate with the beds of the s Berthold Indian Reservation,,

Bed "CC" Bed "CC 0 is 125 to HO feet below the Beulah-7ap bed and can be traced intermittently across most of the northwest corner of the Beulah quadrangle. It is 5 to 6 feet thick and appears to be of uniformly good quality. Many of the small pits in which Bauer and Herold were able to measure sections of this bed are now filled in and grassed over, so that the information on this bed is more complete in their older report than in this report. East of the edge of the Indian Reservation in the Beulah quadrangle is a zone about 4- miles long in which there are few or no bedrock outcrops. Bed "CC" apparently does not persist across this area , *nd it could not be identified in the northeastern part of the Beulah quadrangleo ' I Bed "DD" In the northern part of the Beulah quadrangle Bauer and Herold have apparently mapped two different lignites as bed "DD", In the center of sec. 12, T. H&N., R. 89W. they found'some blocks of lignite in a swampy area and correlated this with their bed nDDn farther to the northwest "(1921, p. 170). This locality (which is not shown on the Beulah quadrangle) is only about 50 feet below the clinker of the Beulah-Zap bed, whereas ia T. H6N., R. 88W., the bed mapped "DD" is 80 to 90 feet below the BQulah-7ap bed. It is this lower bed that has been shown as WDD" on the quadrangle Creek bed The Garrison Creek bed was traced by Andrews between Garrison and Blackwater-Emmett arefiu In the southwest corner of the Emmett

quadrangle it consists at locality 331 of two beds of coal 4..3 and 2.5 feet thick separated by 1.5 feet'of gray clay. Traced to the west both parts of the bed thin, and at locality 327 in the southeast corner of the Blackwater quadrangle the Garrison Creek bed consists of two beds each less than 2 feet thick separated by 10 to 12 feet of clay. The bed continues to thin to the west and pinches out altogether in sec. 32, T. 147N., R. 87tf The Garrison Creek bed in the Blackwater-Eramett area is 160 to 165 feet below the Beulah-7np bed, or about 20 to 30 feet below the stratigraphio horizon of Bed "CC" in the Fort Berthold Reservation* Local bed 110 feet above Garrison Creek bed In the eastern part of the Blackwater-Emroett area is a local bed, f 2 to 3& feet thick, 45 to 50 feet below the Beulah-Zap bed, and 110 to 115 feet above the Garrison Creek bed. This bed could not be traced out of this small area, and is probablylocal. Correlations with beds in adjoining coal fielda Four different surveys of ooal beds meet in the Blackwater-Emmett area These arei the maps of the Knife River area, Andrews 1 map of the Minot area (1939), Bauer and Herold's map of the Fort Berthold Indian Reservation south of the Missouri River (1921), and Pishel's map of the Fort Berthold Indian Reservation north of the Missouri River (1912)> Correlation of the coal beds among these four surveys is difficult partly because some of the beds appear to pinch in this area and partly because Bauer and Herold made little attempt to correlate their mapping with that °f Pishel. As a result of the mapping of the Knife River area, the following correlations seem probablet

Pinhel identified 3 coal beds north of the Missouri River and called them 1A, 2, and Of these, bed 1A probably consists of two different lignites. East of a large segment of a terrace that Pishal calls Armstrong Plata his bed 1A is the Garrison Creek bed and is about 160 feet below bed 2. West of Armstrong Flats, bed 1A is only 125 to 130 feet below bed 2 and is probably the equivalent of Bauer and Herold's bed "CC". Pishel's bed 2 is the Beulah-Zap bed of the Knife River area, Bauer and Herold ! s bed "EE n , and possibly Andrews 1 bed "B". Pishel ! a bed 3 has not been identified in any of the other areas. Andrews 1 suggested that the Wolf Creek coal bed was Pishel's bed 1A, but this is incorrect. The Wolf Creek bed goes below the level f of the Missouri River Just east of the Blackwater-Emmett area and does

not orcp out in either the Fort Berthold Indian Reservation or,the Knife River area. Andrews 1 Minter bed apparently does not persist southwest out of the Minot area and is not represented by any beda in the other areas. Bauer and Herold's bed "DD" does not persist to the north or east and was not mapped by Pishel. Mining and development The mining of the lignite bods in the Knife River area probably started before the beginning of the 20th century when the early settlers dug coal for their own use. By about 1910 several small underground mines Vere in operation mining coal near Beulah and Zap, but the first large

aine in the area seems to have been the Lucky Strike Mine at Zap, which VQS opened in June, 1918. This mine is now abandoned. By 1925 the old Beulah Coal Company (now the Knife River Coal Company) had abandoned its attempt to mine the beds below the level of the Knife

River at Beulah and had started work on the now large underground mine in the Beulah-7ap bed about 2 miles north of town. At about this same time the large atrip mine southeast of Zap vas started. The newest large mine , -3 in the area was Truax-Traer's Dakota Star Mine, about 5 miles north of Hazen. This mine, Which is the largest producer in the area,was begun in 1944* ftnd its production has made Moroer County the /number one'lignite-

producing county of North Dakota and of the United States*, Table which was compiled from the annual report of the coal mine inspector of North Dakota shows the production of the various mines in Mercer County for the year 1950, All of the mines listed in this table take their coal from the Beulah-Zap bed. The only mine active in recent t years that produced from any other bed was the Kamin Mine south of Stanton, which rained the Stanton coal bed. This operation was abandoned in 194.3. Some of the other beds in the area are rained privately for use by the individual ranchers and farmers. All of the production from the numerous small mines in the Knife River aroa is purchased locally for domestic use. A small percentage of the output of the three large mines is bought by local consumers, but most of their production is for commercial users. The output of the Knife River Coal Company at Beulah is used largely in the power plant of the Montana-Dakota Utilities Company at Beulah. The output from Dakota Collieries 1 strip mine at Zap and the Dakota Star Mine near Ha?,en is shipped to Bismarok and to ttTie larger cities of eastern North Dakota and Kinnesota<>

Recently the Bureau of Mines has conducted experiments in the storing of lignite excavated during the construction of the Garrison Dam. A pile of lignite dumped on the ground as a result of normal mining operations will rapidly weather and air slake to a fine black powder several feet into the pile. Also the inner part of the pile is apt to oatch on fire by spontaneous combustion. The Bureau 1 s experiments showed, however, that if the lignite is broken up into small lumps and is compacted with a sheep ! s foot roller every time one or two feet is added to the pile that the danger of spontaneous combustion is materially reducecU The continued mining development of the thick lignite beds of North Dakota seems assured for some time to come. In addition to the present uses, which are almost sure to continue, It is likely that North Dakota lignite will soon be in demand for use in the manufacture of synthetic petroleum products 0

BIBLIOGRAPHY Alden, Vr . C., 1932, Physiography and Glacial Geology of Eastern Montana and adjacent areas I U. S. Geol. Survey Prof* Paper Andrews, D. A., 1939, Geology and Co-il reserves of the Minot region North Dakota i U. S. Geol. Survey Bull. 906-B. . Baker, C. L., 194.8, The Pennington-IIaakon county central boundary area with general discussion of its surroundings i South Dakota Geol. Survey Kept. Inv. 61, 40 PP Ballard, N., 1942, Regional Geology of Dakota Basin: Am. AGSOC O Petroleum Geologists Bull, vol. 26, pp. 1551-1 584 Bauer, C. M. and Herald, F. A., 1921, Lignite in the western part of the Fort Berthold Indian Reservation south of the Missouri River, North Dakota l U. S. Geol. Survey Bullo 726-D. Benson, W. E. end Laird, W. M., 1947, The Eocene in North Dakota: Geolo Soc. America Bull, vol. 58, pp. 1166-1167. 1949, Golden Valley Formation of North Dakota: Geol. Soc. America Bull., vol a 60, pp. 1873-1874. and Colder, C. R., 1950, Paleocene deformation in North Dakota: Geol. Soc. America Bull., vol. 61, p. 1444- Biddle, Nicholas and Alien, Paul, 1814, History of the Expedition under the command of captains Lowis and Clark. Reprinted in 3 volumes, A. S. Barnes and Co., New York, 1904. Bftz, J Harlen, 1946, Bison-poU shed boulders on the Albertan Great Plains: Jour. Geology, vol. 54, pp. 260-262. Brown, R. W., 1938, The Cretaceous-Eocene boundary in Montana and North Dakota: Washington Acad. Sci. Jour., vol. 28, pp. 421-422. - - - -1948a, Correlation of the Sentinel Butte Shale in western North Dakota: Am. Assoc. Petroleum Geologists Bull., volo 32, pp. 1265-1274. 1948b, Age of the Kingsbury Conglomerate in Eocene: Geol. Soc. America Bull., vol. 49,'pp. 1165-1172.

Brown, R. W. end Lemke, R. V., 1Q48, Newly discovered outcrops of the Cannoriball fomv-tion in North Dakotai Science, vol. 107, pp. 624-625. Clapp, C. H., Babcock, E. J., and Leonard, A. G., 1°06, Stratigraphy and economic geology of North Dakota clays: North Dakota Geol. Survey 4th biennial rept., pp. 63-24,3. 1907, The clays of North Dakota: Econ. Geology, vol. 2, pp. 551-564Collier, A. J., 1918, The Nesson anticline, Williams County, North Dakota: U. S. Geol. Survey Bull. 691-G, pp. 211-217. and Thorn, V. T., 191B, The Flaxville Gravel and its relation to other terrace gravels of the northern Great Plains: U. S. Geol. Survey Prof. Paper 108-J, pp. 179-184. Crandell, D. R., 1950, Revision of the Pierre Shale in central South Dakota: Am. Assoc. Petroleum Geologists Bull., vol. 34, ppo 2337-2346. 1951i Geology of parts of Hughes and Stanley counties, South Dakota. Unpublished PhD dissertation, on file at the Library, Yale University, New Haven, Conn. 298 pp. Dorf, Erling, 1940, Relationship between floras of type Lance and Fort Union formations: Geol. Soc. America Bull., vol. 51 pp. 213-236. Douglas, Earl, 1908-1909, Geological Reconnaissance in North Dakota, Montana, and Idaho, with notes on Mesozoic and Cenezoic geology: Annals of Carnegie Museum, vol. V, pp. 211-288,, Dove, L. P., 1922, The geology and structure of the east side of the Nesson anticline: North Dakota Univ. Quart. Jour., vol. 12, pp. 240-249o Ehlers, A., 1943, Williston basin wildcat teat, Oliver County, North Dakota: Am. Assoc. Petroleum Geologists Bull., vol. 27, pp. 1618-1622. Fenneman, N. M., 1928, Physiographic divisions of the United States: Assoc. Am. Geographers A-nals, vol. 13, pp. 264-353 Flint, R. F. and others, 1945, Glacial Map of North America! Geol. Soc. America 1947, Glacial geology and the Pleistocene epoch, Wiley & Sons, New York, 589 pages.

Flint, R. F., 19/.9a, Leaching of carbonates in glacial drift and loess as a basis for age correlation Jour. Geol. vol. 57, pp. 297-303. - - - - - lQ49b, Pleistocene drainage diversions in South Dakota: Geograf. Ann., pp. 74o - - - - - Pleistocene Geology of Eastern South Dakota U. 3. Geol. Survey Prof. Paper - in preparation. Fox, S. K. Jr. and Ross, R. J. Jr., 1942, For-imini feral evidence for the Midway (Paleocene) age of the Cannonhnll formation in North Dakota! Jour. Paleontology, vol. 1, PF Frye, J. C., 1951, Soil-forming intervals evidenced in the Pleistocene. Soil Science, vol. 71, no. C f pp. 403 408 - . . and Leonard, A. C., 1951, Stratigraphy of the late Pleistocene loesses of Kansas: Jour .Geology, vol. 59, pp. 287-305. Fuller, M. L., 1908, Summary of the controlling factors of artesian wolls: U. S. Geol. Survey Bull. 319. Hancock, E. T., 1921, The New Selem lignite field Morton County, North Dakota: U. S. Geol. Survey Bull. 726-A 0 Hares, C. J., 1928, Geology and lignite resources of the Marmarth field, southwestern North Dakota i U. S. Gool. Survey Bull.. 775. Hatcher, J. B., 1903 Relative age of the Lance (Ceratops)beds of Converse County, Wyoming, the Judith River bnds of Montana, end the Belly River beds of Canada t /toorican Geologist, vol. 31, pp. 369-375. Hayden, F. V e , 1869, Preliminary field report (third ani:u'j] ) of U.S. Geol. Survey of Colorado and Hew Mexico, 15 pp. - 1876,. U. S. Geol. Surv. Terr 0 , 8th Ann 0 Rept. 515 pp 1878, U. S. Geol. Survey Terr. Mon. 7, Part 2. Hennen, R. V., 194-3, Tertiary geology and oil and gas prospects in the Dakota bsinj Am. Assoc. Petroleum Geologists Bull., volo 27, pp. 1567-1594.

Hennen, R 0 V., 1948, Discussion, Correlation of Sentinel Butte Shale in western North Dakota: Am. Assoc. Petroleum Geologists Bull., vol. 32, pp 1823-1824o~ Holmos, C. 0., 1941, Till Fabric: Geol. Soc. America Bull., vol. 52, pp. 1299-1354. Howard, A. D 0 , 194-6, Caliche in glacial chronology: Geol. Soc. America Bull., vol. 57, p. 1204. Gott, G. B., and Lindvall, R. y., 1046, Lake Wisconsin terminal moraine in northwestern Horth Dakota: Geol. Soc. America Bull 0 , vol. 57, pp. 1204-1205. Kline, V. H., 194-2, Stratigraphy of North Dakota: Am. Assoc. Petroleum Geologists Bull., vol. 26, pp. 336-379* Krumbein and PettiJohn, 1938, Manual of Sedimentary Petrography: Apple ton Century Press, 54-9 pp. Laird, W. M., 1944, Stratigraphy and structure of North Dakota: North Dakota Geol. Survey Bull. 18, 10 pp and Mitchell, H, H., 1942, The geology of the southern part of Korton County, North Dakota: North Dakota Geolo Survey Bull. 14-, 41 pp Lemke, R. W., 1950, Geologic map of the Velva quadrangle, North Dakotai U. So Geol. Survey open file report. Leonard, A. B 0 , 1951, Stratigraphic Zonation of the Peoria loess in Kansas: Jour Geol., volo 59, pp. 323-332. 1$50, Stratigraphic Zonation of the Peoria loess in Kansas: Geol. Soc. America Bull., vol. 61, p0 1481. Leonard, A. G., 1904, Topographic features and poological formations of North Dakota: North Dakota Geol. Survey 3rd annual report, pp a 127-177, - - Clapp, C. H., and Babcock, E. J., 1906, North Dakota Geol. Survey 4th Biennial Rept., 324 ppo - - - - -1908, The geology of southwestern North Dakota with special reference to coili North Dakota, Geol. Survey, 5th Biennial Rept, ppo 27-114*

Leonard, A. G., 1909, Cretaceous and Tertinry forrrtions of wer.torn North Dakota! Jour. Geology, vol. 19, pp. 507-54.7* - - - - - an$ Smith, C. D., 1909, The Sentinel Butto lignite field, North Dakota and Montana: U. S. Geol. Survey Bull. 341, pp. 15-35o - - - - - 1922, The Vhite River formation in North Dakotai Univ. of North Dakota Cunrt. Jour., vol. 12, pp. 218-228. - - - - - 1912, Bismarck Folio, U. S. Geol. Survey Geologic Atlas of U. S. Folio 181, Field Edit., 58 pp. 19l6a, Pleistocene drainage changes in wnstern North Dakotat Geol. Soc. America Bull., vol. 27, pp. 295-304* 19l6b, The pro-Wisconsin drift of North Dakota: Jour* Geology, volo 24., pp 521-532. Babcock, E. J. and Dove, L. P., 1925, The lignite deposits of North Dakotai North Dakota Geol. Survey Bull. Meek, F 0 B. and Hayden, F. V., 1862, Descriptions of new Lower Silurian, (Primordial) Jurassic, Cretaceous, and Tertinry Fossils, collected in Nebraska, by the exploring Expedition under the command of Capt. Wm. F. Reynolds, U. S. Top Engrs; with some remarks on the rocks from which they were obtained: Phil. Acad. Nat. Sci. Proc., vol. 13, pp a 415-4-63. Nevin, C., 1946, The Keene Dome, northeast KcKen?ie County, North Dakota: North Dakota Geol, Survey Bull. 21, Part I. Pierce, W. G., 1936, The Rosebud coal field, Rosebud and Custer counties, Montana! U, S. Geol. Survey Bull. 847-B. Pishel, M. A., Lignite in the Fort Berthold Indian Reservation, North Dakota; U.S. Geol. Survey Bull. 471-C, 1912. Powers, W. E., 1945, White River formation of North Dakotai Geol. Soc, America Bull., vol. 56, p. 1192, Quirke, To T,, 1918, Geology of the Kildeer Mountains, North Dakota 0 Jour. Geology, vol. 26, pp. 255-271* Ruhe, R. V., 1950, Graphic analysis of drift topographies I Am. Jour. Soi., vol. 248, pp. 435-443*

Russell, L. S., 1950, Trans. Royal Soc. Canada, vol. 14, rer. 3. Russell, V. L., 1929, Drainage alignment in the western Great Plains Jour. Geology, vol. 37, pp. 249-255. Schultr, C. B. t Lueninghoenpr, G. C., and Frankforter, W.D., 1951, A graphic resume of the Pleistocene of Nebraska (With notes on the fossil mammalian remains): Univ. of Nebraska State Museum, vol. 3, no. 6, 41 pp. Sealer, 0. A., and others, 1942, Discussion of "Stratigraphy of North Dakota" by V. H. Klinei Am. Assoc. Petroleum Geologists Pull., vol. 26, pp. 1414-1423. Sharpe, C, F. S. f 1938, Landslides and related phenomena: Columbia Univ, Press, New York, N. Y., 137 pp. Simpson, H. E., 1929, Geology and ground water resources of North Dakota: U. S. Geol. Survey V.ater-Supply Paoer 698. Simppon, H. F.., Jr., 1942, The Pleistocene Geology of Garrison Quadrangle, North Dakota: Unpublished Master's Thesis, Univ. of Illinois, 55 pp. Smith, C. D., 1908, The Washburn lignite fieldi U. S. Geol. Survey Bull. 381, pp. 19-29. - . - 1908, The Fort Berthold Indian Reservation lignite field. North Dakota: U. S. Geol. Survey Bull. 381, pp. 30-39. Btanton, T, W., 1920, The fauna of the Cannonball marine member of the Lance formation: U. S. Geol. Survey Prof. Pep er 128-A, 64 pp. Thorn, W. T. and Dobbin, C. E., 1924, Stratigraphy of Cretaceous-Eocene transition beds in eastern Montana and thn Dakotast Geol Soc. America Bull, vol. 35, pp. 481-505. B Tisdale, E. F.., 1941, Geolt>py of the Heart Butte ouadrangle: North Dakota Geol. Survey Bull. 13.

Todd, J. E., 1914, The Pleistocene history of the Missouri Rivert - Science, vol. 239," pp. 263-274. 1923, Is the channel of the Missouri 3iver through North Dakota of Tertiary origin?: Geol Soc. America Bull. vol. 34, pp. 469-493.

Tovmsend, H. C., 1950, Dof oration of Fort Union formation near Lign-'to, North Dakota: An. AOGOC. Petroleum Geologists Bull, vol. 3£j pp 0 1552-156£o and Jenke, A. L., 1951, The problem nf the origin of the Xax Kor~.ine of North Dakota and Cnr.'la: Am. Jour. Seio, vol. 2/,9, pp 0 S/,2-358. T ' ::-.rror., C. ?.., 19/,°, Prohnhle Hlinoiin .ie part of tho y.lsso-iri River: Geol, Soc, America B-ullo volo 60, p. 1°2' Wilder, ?. A., 1905, The Ligr.ite of Morth Dnkota and its relation to irrig.ticr. : U. S. Geol. Survey Water-Supply Paper 117. 13S3, On the existence of a deposit in n ".na ?r.d northwestern llorth Dakota that is pos3 . . qnivnlart with the Green River group: Am. Jour. 3ci., 3rd series, vol. 25, pp. /J.1-U£. Wood, L. H., 1904., Report on the region betveen the Northern Pacific railroad and Missmiri River. Its topography, clir.to, vegetation, irrigation posjil ilities and co rj.l deposits: IJorth Dakota Geol. Survey 3rd annual report, pp. £1-125. 272a

°00 N 2. Sketch map of Knife River area showing position of quadrangles.

to

.?- --' Dfwt Missouri Plateau ' -

'

'

in . 3. Sketch map .howinc physiographic ,Hvi ions of -h Dakota,

Montana

formation for mat ion Sandstone shale vo c- CM Realized strati--;/ i. . tween Gl- I action of Upper Cretaceous and Tertiary formations in--, and Bi' : mn.rck, North Dakota.

n : r-l V) f- r - CM

Percent Percent ro

CO ro CO W (D O ro ro O D

'J

p

O ) J ) ; - n J

j ? ) Granitoid rocks Limestone and Dolomite Gneiss and Schist Igneous Chert Quortzite Vein Quartz Pierre shale l- it JL-J

:-( l-J M (D O ro V-O MD

C") J

' ; ) j

"1

")

' -

r 0 Granitoid rocks Limestone and Dolomite Gneiss and Schist Basic igneous Chert Quarizite Vein Quartz Pierre shale ON O O

P e r c e n t

ro cw

en cr> ->j oo O O O O '.)

O O O Percent ro

en 0"> -J Cn o o o o o o o o . M 1?

CO CD O ro vn H3 ro

H

J

]

f ' ' ;- - J Granitoid rocks Limestone and Dolomite Gneiss and Schist Basic Igneous Chert Quartzite Vein Quartz Pierre shale £

w CD O vn

VO H ! j

- Granito'd rocks Limestone and Dolomite Gnci c,s and Schist Busic igneous Chert Quartzite Vein Quartz Pierre shole wQ O P dBo

Percent Percent . fO COo CO ro

e ro en f

ro'oJ -P en ->j oo 5OO-OOOOOO

J

Granitoid t rocks Limestone and Dolomite Gneiss and Schist Basic Igneous Chert Quartzite Vein Quartz' Pierre shale

ro oj CT> -j oo O O O O O O

"J O Va

Ci in CDo O

? tr) O -J

J

G Tl h " ' .!

"[" :" "

-

, j Granitoid rocks Limestone and Dolomite Gneiss and Schist Basic Igneous Chert Quartzite Vein Quartz Pierre shcle S G3 O O O Ha 0)p*

ro '~O H Co cn vn

Co Co

]

In )

.

O ) t. ! Per j -r ) c e n ; t

) CJ ? D ) Granitoid rocks Limestone and Dolomite Gneiss and Schist Basic Igneous Chert Quartzite Vein Cuortz f-'pierre shale -o K

W C'M CO0 O Co r3 Co

CO CO sv )

]i"i

.]

' J F J

U3 e r ( J ; e n -

H t ? r -

]

: Granitoid rocks Limestone and Dolomite Gneiss and Schist Basic Igneous Chert Quartz! te Vein Quart i Pierre shale

Percent Percent O3 D ,

., J r

) ' C O ) J -t

fe r - ;

) , J O ) ) Granitoid rocks Limestone and Dolomite Gneiss ' and Schist Basic Igneous

Chert Quartzite Vein Quartz pierre shale

ro oj

o o o o o

->l CO o o o CO h-1 HJ CO CO

r , . j

' )

-

Granitoid rocks Limestone and Dolomite Gneiss and Schist Basic Igneous Chert Quartzite Vein Quartz Pierre shale

'J- '£! 09S ci3H fM (ll) Q

o N 0§ . a

CD o Q ' o Q a £ Q o : w ex 01" ;

! O

- i' 00 ft>

0 o

c:

I'J -4- -'. O .'00 O ' J r 4

Co Cx.' Cm

Fig. 7 continued Canadian P rovenonce O 40

o QQ 4.- .ocai ! ! e (1) QsdJ NEi, NWi, sec. 36, To LNo, R, O Canadian Provenance Western Loco! QJ 30 OJ Q. pp. n

A:

O "C" 'C; C)

1, 0 cr J Q

' i "s1

f-. ' 5 '. " ( - 1i_ (. 3d c: -5" COo

: '

' ' o

vy m ' 0) o -

%J IO , ( 7 <Jj

, Gq '

r

t

! ! j t '

: ' ,

j 1 t ' CO ? D , w CO

', ! i r : . ,] i LJ (2) Qtgj NV/i, SWi, seo. 10, To H2ST., R, 91W

Fig. 7 continued Provenonce Canadian Western (3) Qsd25'SVi, Ntfi, sec, U, T. Local R. Provenance Canadian Western Local

30

0o Grcnitc.jd t t

1 '.

-! '"

g o S

J o o Metomcrp rocks

;

It)

' S t: ° ' S s- : § ,

' rj ,

. . r : -,

rn 1 r i " Andesite

" . t;

Cq

£Z3

n in

O o

O " " U) , BWt, sec. 23, T. l/OHo, R. Saf.

Percent

r\j

Po Oj

(Ji CT> o o o o o o o Percent

cr O O O O O O

jfj r co

; COw CO S ro e CO

r

f ? g

] D n r" n D D 3 ; ! "! ; ., t .-' , :

; ' , r ' : ,

; ; !

Grcr.itcid roots

and Dolomite Metamorphic

1

rocks

I Q ' ex " BJSIC igneous ; a5 Ve/'n Quartz

$

o Pierre shale

-

Andesite o

Be/f Quartz ite

' 0)

. , 3 w C/)er/

-. g A

Sandstone 1 r t 0 /ron Ox/de

Chert j

,'

] :

: - ! n" r "" " . ' :

, j ; Granitoid ro<:ks and Dolomite

Metamorphic o rocks ' " Basic Lgnecus

Vein Quartz ' "° !

Pierre shale

Andesi.te o Se/ Quartzite . :. -<J t Sondsfone

/ron Ox/de fg Cnerf )

O3 P*TFP"rtenf

fO OJ

OJ (J> o o o o o o o percent

oj

O O O O O m f oT

o

I D ;r- " ( 0) ? p -3 e - D ? p 1" B j

! D

r

Ji

Grinir-id foc'f-S

./TleS o"c w ond OQiomite , Metamorphic ' c/j rccAs § S. , o

r CL BJS/'C Igneous D

V'' s /r Quor's ID ® o , ; o M pierre snole

e Se/ Quartzite

o Sandstone

/rcn Ox/c/e

j ;

D : :

;

D : B : : D ; D ; " Granitoid rocks Limestone and Dolomite Metamorphic rocks Basic igneous Vein Quartz Pierre shale j

Andes! te Belt Quartzite Chert Sands tone Iron Oxide Chert oo Q. Q-

O

§ : i

j£o

0 1 El

Fir:.. 7 continued provenance Canadian Western Local 5O vAv 3n AJ ir>

O ; ; a

and Dolomite

; ;

! . o

Kj '

'Qj''O' W

: § : ? ' 1 g

o

: o Ew' VJ

.y

'

' XJ ' ! t_, ' if)

Xj !

'

£i 0 ' S -2?

jc D:O

w CQ''Q.

OQ o to '

'

! ; : [ : ! , . .:

, .. ., : ' '

' ,.. ,

, : '

! . , , e=i era I ; 'ra III ! (9) Qicj SVv, SWi, sec. 1, T. U5N., Ro 8&U Canadian Provenance Western Local

30

20 ; 0 '

i ,U N -3J

I o !

;

'. 2 i Q , 1 - o £ ' s , ' ;

! ri : ( ; ]

] :' ! ' i :

- : 1 ca '_J i-n

! :

1! " ' ! ! f ! ! j j

(10) Qmgj SWi, SEi, sec. 26, T. 144N 0 , R

continued P rovenonce Canadian 40 J

w a- 20 n

ot u O "1 c° V)

U : c e if. ! - ' : !

, I J O

j .u j

' u i § w i j to

O V}o en - h~4 ! , :

t0 o 0 s

w

Q:

' Western j , t CO

"5 '

; )

t :

i! o O

QQ " ; ! f-- ! ! t; c; O CO ] f- - Local O.

cr

qj

-i: o

H ' jr=t (U) Qing; SWi, SW£, seo. 32, T 0 H7N 0 , R.

Fig. 9 W -. o*/A ft A QvvT

/7\ p<lvtffntfnf M A A A.. Cross section showing relations between Pleistocene deposits in NEl/4, NEl/4, sec. 18, T. 146N. f R. 86W. NW B. Idealized cross section of Knife River valley at west edge of Beulah.

ro vo ro J70Q trr KFig. 10. Cross section of "4.0 foot terrcce" of Missouri River about 1 r.ile northwest of Fort Clark, sees, 26 and 27, T. U4Hi R. 8-J., showing relations of Missouri River gravel (Qmg) to older Pleistocene deposits and to Rocent alluvium (Qxaa) on the terrace.

rv>

VuJ ,oO Fig, 11. Former drcdure of North Dakota as informed by Todd.

ro vo -p- ,oo

Fig. 13. Pleistocene diversion valleys as mapped by Leonard and Alden.

NORTH DAKOTA Fig. l/:.. CkMoiMlired structure i.yo of Torth D:.l:otn , coivbourecl on to of After Erll.-rd (194-) o

-/.-it ion jakota Ct&r *iri, Sec. 21, T. U5 N., P.. ' Saith 1'dne, ,., Sec. 23, T. U4 N'., 85 " Knife River i-oal "In. Bevlah, I'mercer fo. Kamisch '.'.lr.* t a "J

J Norton Co. lucky Strike Mne, Sec. 26, T. 144 K., 89 ''" uercer o. Renublic !!in '( kota Collieries)

Sec. 24, T, 144 1!., ? 89VMercer Steven Bros, Coal Co, Garrison, "<$!jan Co,

Sec. 19, T, 14* II., -4 .. , Colstrip, .-Montana Fittsburph cVOIjj r 2VCu v O,*.t "

Lab. !!o. Du.2574 A-68042 A-45379 A-45831 A-45833 --/ A-45933 A-46852 "oin of Analysis A D A D A D A D A D A D A D A D A D Proxiirflte I.'moisture "

Volatile ."atter ' 27.3 Fixed Carbon Ash

U?.tir.ate Hydrogen 6:3 Carbon :itro5en Oxygen S'llphur 1.3' reatir.5 Value Talcries 3,823 6,104 6,709 3,961 6,144 4,356 6,439 7,039 4,272 6,439 7,039 4,400 6,467 6,928 4,123 6,178 6,639 4,939 6,633 7,322 7,335 7,7SO 3,490 B. T. U. 6,930 11,020 11,980 12,076 7,130 11,060 12,120 7,840 11,680 12,670 7,690 11,590 12,760 7,920 12*470 7,430 11,120 11,950 8,890 11,940 13,180 13,650 14,00015,290 ?'-!"'3 of Analysis: A-as received; C-nois£ure-free; D-Moisture-and ash-fre. A/lyses by U. S. Burraa of L2js, Pittsburgh Laboratory

OQo 900 'Vi1?'? r'ic'? / 6T9'Z1Z'Z ± ?sb'v;> z

'"'auerc Coa? 1'ir.e :>eick .yO'.-.l lline ;:akot:i Collieries /Dakota Star : ±ne r,l'.ier Coal I line Grishkov;oky Coal "Ane Knife River Coal 1 line Co, Krause Coal i-I3.no Link Coal '. 'ine "it telr tea At ! in* loxle;/ fc .ri(V;;-cv.% l.ir.e .alker Coal . ire f'loi o Jrift "trip vStrip Strip Strip Drift Strip Strip Strip Slope Strip A Mress BeuT- ail /;ap Zap Hebron Bei\lah Beulah Zap Beulah Kazen Bt.O.ali Glci: \J3.1ri Zap Tons Produced 2,000 191,143 575,455 / 2,343 41,295 704 / . output Value 5, So.". 00 <,2 :'-v.OO 47/i, 2 j A. 1,236, 65-Jt3Jc 625. co 6,693.30 831,230Kol 1,7-1.70 23:. 50 1,399.00 T.ible 3. Lr.iite Pro lucti.on of i'orce.* ('onty for fiscal ;*ear 1950

' o? 'Ift.r - T.c-'ts !'o. 1 ' oil Vvrccr ''oimty I'orth jkot'i 13O-150 193-05 1150-14'SC I46o-io00 ljCf-1620 1703-3430 340-3940 394C-3950 3950-220 423t-4690 4690-4300 4300-4o30 4"3C-5040 504C-5110 511C-5140 17;--5230 527--5360 53''0-5540 5',-5o30 57-5730 5 70-5950 shale, gray lj.;ito . 7ort Union, Cannonball and Hell Creek forr.iationo e inter -bedded J sandstone black li:itic shale shale, bentonitic uj.aciv xxiu. t/j-C shale rjray shale black shale rjray shale Top of' Dakota ss shale C. saud stringers sand sandj" sliale Top of ; .morrison ' Pierre shale, Kiobrara fa., Bent on ;':rou;j shale w/sand stringers Top of Sundance 5C94 outilti sandy shale shale sand shale chale dense ray limestone red viiu ;.;recu shale dense tan ILnestone Joi'.se jjiuk '.lolomite .r;rcen cl- gray dense IL.iestoii roJ shale

w

P 2 -V O o o O-'r ,

o o ll ° P r J

o ;a 0 © bcf f ° O CO Cf 5" 0 f J 3 M o o V>J S; ro

ro

p £° cf xi o7 ° cro : S S' CD J.1 p. o

o C-J M UJ

-Ti

. ' , j o j i r f ( ( i w ' ; -J . o o o o o o o ' '"J p TJ H :J n ri P .1 o

5~' C"f

HJ 'J (D O :j f" to P ! *-i c/ o P a o Ki C'j 3 P° J.J r1 *"j

'3 CJ* O C., O fj p,

O p, £ M O q : i 3 ,-. j o H H/ M y p o co fj CD p H- H. O CO P' p, cf c?- iJ- o ,-3 >-j o o O o ra t" c o 1 a o P, r: : o ct- o l- 1 MO M M r*- o i- p H- ft , -

f5 " (

H' 0) M O C1" O ° 5; CT ct- ;J P P M '" 0 ® tL O P op. P- M O ;3 OH- H- C fl ' c*- ct- '1 O O £ O p P H 3 U'. 0S'£ 0 w cj ° p.o

' D ' ,

-j ',

' 'i / J 1

) f ( ( M M 00 J- 0 0 ..- -p 0 O X O O O O O '--I O P"

rf rr i cr rf cr ; i .-/ c*- ct- oj h KJ. r*" O D o cr co o o

" P c- b h rJ P g '-s :: H- o

P CD CD t' 3 C) .J' !- J P o n

CJ M ' CD O CD o i w

-'-I zr o :J <-fl a M o J :o c- P M U CD co M ct- p a P M M o .a CD to I ' p ! ' L~' t- 1 JlJ

"' y i"t" i"j UJ> i.-j M. p. ,a. o o 6 6 -J

h GO '.) ti :J M CD

CD m M ct- ct- CD CD

Ct"OOO ;Ct" o J rs P. o j P- :3 P- o CD ct- r-'

O Cd C+1 O 3 O O O O "J 07 M

CD O n P -- j M P' 'P -3 P- H' :-j o crop M H- CD O M M WO ;j £ Cf Ct- HJ

ct- f-i. TJ >j- a O .' O DO *-x r j

CD H3 ' O % vO M O ro o

Cz)

-<3

O Oi ' tl AJ Oi ' N

p /j .P H :i :' ; o "-J -f a c o ct- c- ' ; o c*- H- M f-1 CD b b H P i. C-i 0 O ct- CD M P C O M M 2 H' P T

"' i , ' CD h t CD O '- C iS f £ !f J CD i- 1 - w ,

o" CD' c 'J - -d c; -' co M'; o p' £ £ M CD ; o J C1 w. b. p - cf b w r CD O f Of" to f o :-i be : "5 MO [i § U) 'CJ1

vO O "' "

C O C C; 7 Vi vi O O O O O

hi O ! ? O fj P "j C J CJ O - C) P-> IJ P 't -f ' '/ i n ' CD" r- H- rv. P r° o c i TJ tj< i 1 fj* " 0 rf 0 M M n CO 0 -J M ctD ?J CD r; o Ci o o o cj o c: n o a P o U) H a rf f a ;L

Mi- o ci ? p P P M D H' C' M 0 P CD M O p 0 CD 5" 3 CD CD CD -3 p P, c- OMWCO Q, rf V d- CT CO CD

CD u O O O C+- O (0 CD P- O .T* ctD CD CD o H- 'jt CD M P. ': ) ? H' O D CO P.

rt f,o j U o' -t D J

-j. Vn O

.'i vi i -J " - -o vi vn ( ' i " i f O " i r f i o v>i T:T b o 006 C i ci c ' n o' P n P f j :i f-1 r - P A p- . o

;r co ." r.13 do : :r -i CJ (-' CO C J H- P '- cj j :/ M ir *-j c- M cf (J CD CD O M cs i cj f? o :? M c>j

'

p Ji n p M & P en CD, S1 u. OOP. H- & Q H0 P rr -o

i-J. r-J c+ o o ";5 ?- o g n 1? f? h'- J CIN C) vO C) V.O V~" Is " O M si"

Co Oj

( :

1 -J

VJt O

P-r

K" r. K

&

C" (1 M r- /

f;

Jjl ,-, c'

p s/. H1 -' l-j v*t

3 V - 7 r-| ', ' l , t :

K '- C' r. a .;

v. ff- r f-' h- C3

r- B 1? I':' CT* i ' ( - - i C'

; c

- ' ; ("

J V..J

Q p c' I J f: r; r- "- c' p - ? O 1

T

1 - J

;T

f o °: r 7! 1°

' '? C*"

i'J1 " i i) s, , Jo (D i

' If.1 o I -: o o Cj

PecV 38] -A -I.icCone County VD5 84"-C Vi'llliston rol-a l,Teri3cn Anticllno L- (J1-J f Fort Brthold T*'-'J ?ort Berthcld : 1-A, -Pl-C Linot OOf-B

'ibhburn ivf. ol era 7.'!C-A Crnr.cnbe.il River 541-u ' Standing Rock and Cheyenne River Northwestern South DeUota V,-.nunrth otjr.tinel Dutte ?'41-A

Sidnoy 471-D Glendivo ( 471-D Terry ' 47i-D Bakor -*71-D Zkalvka 7'U-F JJirpah Miles- City Ml-A Little Gheop ?7 Rosebud 847-DHP Forcyth R1.--A r>9 ' Tullock CreeK Pine Ridc;o 5O.-H Dull I'.ountnin Elf, Horn County yorthw*rd eztonsion of ohoridnn 60G-B Sheridan 3-11-B Buffalo 3C1-B Berber Pumpkin Buttas B06-A Sussex 471-jj1 Glenrock Lost opring Gillstte 796-A

Powder River 381 -B Littlo Powder Hiver 471-A Coalwood

973-B This report

'rPFT/TFOIK CRETACEOUS? Lr?fc7AC.LUU2> TERTIARY? TFFTIARY n

o "2 o rr; o n o3n -n o Fort Knion fom.ction c£" 2 a -v. P : o- sr! V, O-o ! -i-L-J

Lance formation Fort 11 nion Fcsatch formation formation 5? Gcr rr o D V) O. O Cretaceous Montana Eocene (?) Lance ' formation Locene

' \ii

CRETACEOUS 3 c/ Fort .'ru'on formation a ; Fort Union(?) formation ! " 3 ; %

CD f. o 5-, n TERTIARY KJ- O9 !;rr rrrrncrou*; ory O O ., o

§ i , CRETACEOU Upper Cretaceous ] Paleoccoe c-Dccne j O O Montana group ! ! H W "J ' nTP : o

' &

o 2. S .' n -u J

r; ° ?T !

Fort Union ' 3 '!r i '

! a" a 5" ! S 2 ?

. ; Cs ' fe fcrmatzoo s a 2 I

m

tfcsatcb fro.

: n <y, t £

3 o

O

Cr n.

J5

B t

3ai

CO o

en o CRETACEOUS TERTIARY

Upper Cretaceous Paleocene-

Montana

! J?. t tA ! 1n group gsr 3ft. If K

5 r

Jva G J; 2 o n Fort Union for cation

j o

' 1 T n 7) H f w

o S 2.

,

y n " '

a

n °

cr

& p Q

s: :

n O 2

yJSJXJvTvK . /"" fXi 1

S'

f9 S ry Ot %

p. o n 'j T-

! O n Q3 O J

r3 i"

O3 S

o o i

S

". 1 IS "I n ;