The Lake Missoula Floods and the Channeled Scabland Author(S): J

Home , Crab Creek, Dry Falls, Giant current ripples, J Harlen Bretz, Kolk (vortex), Lake Lewis

The Lake Missoula Floods and the Channeled Scabland Author(s): J. Harlen Bretz Source: The Journal of Geology, Vol. 77, No. 5 (Sep., 1969), pp. 505-543 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/30062250 . Accessed: 15/01/2015 10:22

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Geology.

http://www.jstor.org

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions THE LAKE MISSOULA FLOODS AND THE CHANNELED SCABLAND1

J HARLEN BRETZ Department of the Geophysical Sciences, University of Chicago

ABSTRACT This paper reviews the outstanding evidence for (1) repeated catastrophic outbursts of Montana's glacially dammed Lake Missoula, (2) consequent overwhelming in many places of the preglacial divide along the northern margin of the Columbia Plateau in Washington, (3) remaking of the plateau's preglacial drainage pattern into an anastomosing complex of floodwater channels (Channeled Scabland) locally eroded hundreds of feet into underlying basalt, (4) convergence of these flood-born rivers into the Columbia Valley at least as far as Portland, Oregon, and (5) deposition of a huge delta at Portland. Evidence that the major scabland rivers and the flooded Columbia were hundreds of feet deep exists in (1) gravel and boulder bars more than 100 feet high in mid-channels, (2) subfluvial cataract cliffs, alcoves, and plunge pools hundreds of feet in vertical dimension, (3) back-flooded silts high on slopes of preglacial valleys tributary to the scabland complex, and (4) the delta at Portland. Climatic oscillations of the Cordilleran ice sheet produced a succession of Lake Missoulas. Following studies by the writer, later investigators have correlated the Montana glacial record with recurrent scabland floods by soil profiles and a glacial and loessial stratigraphy, and have approximately dated some events by volcanic ash layers, peat deposits, and an archaeological site. Several unsolved problems are outlined in this paper.

GENERAL STATEMENT bounding valley bottoms. Closed basins as Although paleo-Indians probably were deep as 135 feet were bitten out of the already in North America, no human ear underlying basalt. Dozens of short-lived heard the crashing tumult when the Lake cataracts and cascades were born, the Missoula glacial dam (the front of the Pend greatest of which left a recessional gorge, Oreille lobe of the Cordilleran ice sheet) Upper Grand Coulee, 25 miles long. The burst and the nearly 2,000-foot head of im- greatest cascade was 9 miles wide. The flood pounded water was free to escape from the rolled boulders many feet in diameter for Clark Fork River valley system of western miles and, subsiding, left river bars now Montana and across northern Idaho. It standing as mid-channel hills more than 100 catastrophically invaded the loess-covered feet high. Current ripples 10 feet and more Columbia Plateau in southeastern Washing- in height diversify some bar surfaces. A ton (Bretz et al. 1956, fig. 23) and reached gravel delta 200 square miles in area was Pacific Ocean levels via the Columbia built at the junction of Willamette and River, 430 miles or more from the glacial Columbia river valleys (Bretz 1925, figs. dam. So great a flood is unknown at present 16, 17). Portland, Oregon, and Vancouver, elsewhere in the world. It has been esti- Washington, now cover some of it. Almost mated to have run for 2 weeks. It was 800 2,000 square miles of the plateau's basaltic feet deep through the Wallula Gap on the bedrock lost its preflood loessial cover which Oregon-Washington line. otherwise is on record in the 100 or more On the Columbia Plateau in Washington, "islands" 40 acres to 40 square miles in it transformed a dendritic preglacial drain- area-tracts that stood high enough above, age pattern into the amazing plexus of the or far enough from, main spillways to have Channeled Scabland (fig. 1). It flooded escaped the tremendous erosional attack across stream divides of the plateau, some meted out to the plateau's preglacial valleys. of which stood 300-400 feet above today's Prevailingly, these "islands" have steep marginal slopes and prow points at their up- 1 Manuscript received February 6, 1969. stream ends (Bretz et al. 1956, fig. 22). [JOURNALOF GEOLOGY,1969, Vol. 77, p. 505-543] Back-flooding left deposits for dozens of © 1969. The University of Chicago. All rights reserved. miles in nonglacial valleys which entered the 505

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 506 J HARLEN BRETZ plateau from Idaho on the east (Bretz 1929, another lobe (Okanogan) of the Cordilleran fig. 1) and central Washington on the west. ice sheet. Two floods found no Okanogan The above statements are generalized, lobe in the way, and most of their water es- for there probably were at least seven caped around the north and west sides of the successive burstings of the dam, each (ex- plateau via the preglacial Columbia valley. cept the last) repaired by later glacial ad- The earliest flood outlined the anastomo- vances to again impound the valley system sis of the Channeled Scabland. Later ones of the Clark Fork River in western Mon- deepened and widened its main spillways tana and make another glacial Lake Mis- and thus left earlier high-lying and less soula. Five of these floods could only cross deeply eroded channels unrefreshed, to be- the plateau, because the westward-leading come obscured by interflood weathering, capacious valley of the Columbia River rain wash, and wind-made deposits. north of the plateau was then blocked by The satellite Nimbus I has confirmed

FIG. 1.--Glaciated and flooded areas on the Columbia Plateau in southeastern Washington. Abbrevia- tions: BB, Bacon basin; B-WC, Bowers-Weber Coulee; CC, Crater cataract; CN, Chandler Narrows; DC, Devils Canyon; DHC, Drumheller Channels; EB, Evergreen-Babcock ridge; FH, Frenchman Hills; FSC, Frenchman Springs cataract; KC, Koontz Coulee; LC, Lind Coulee; OC, Othello Channels; PC, Palouse Canyon; PHC, Potholes cataract; PR, Palouse River; RC, Rocky Coulee; RP, Ringold plain; SG, Sentinel Gap; SM, Saddle Mountain; WC, Washtucna Coulee; WCR, Wilson Creek; WM, Withrow moraine; *, known occurrences of giant ripples.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 507 this judgment. One of its pictures, made sible compromise by having numerous ice- from an altitude of about 500 miles (pl. 1), berg jams along the channels-thus lo- encompasses almost the whole plateau in cal pondings and divide crossings-with Washington and, despite some atmospheric subsequent bursting of such dams to pro- obscuration, has made possible a compari- duce repeated local flood conditions. But son with the scabland map. The satellite, no one has found field evidence for such which "saw" most of the bedrock floors dams of floating ice. Flint (1938) was far of the anastomosis clearly, failed to record more conservative. He asked for invading more than 200 miles of high-lying channel- glacial meltwater streams "no larger than ways already known as such from their berg- the Snake today" and for aggradation of carried erratic boulders, a few gravel de- spillways until the violated divides could be posits, and some scabby channel-bottom crossed. He argued that the scabland was basalt outcrops. made by "leisurely" streams during removal of the fills. The terms "channels," HISTORY OF INVESTIGATIONS "bars," and "floods" were repugnant to him. The writer's first study of the Washing- Other adverse comments made largely by ton portion of the Columbia Plateau (Bretz uniformitarian-minded geologists need not 1923a, 1923b) dealt in part with Grand be noted here (see Bretz 1928c and Bretz et Coulee and its origin. But the concept of al. 1956). flood origin of the coulee or the scabland In the meantime, the writer continued his complex waited for another year and was field studies and produced nearly a dozen then termed the "Spokane Flood," with no more papers that dealt with (1) specific suggestion for the source of the water. A tracts in the scabland, (2) local scabland preliminary map of the complex (Bretz making by the Columbia River at The 1923b, pi. 3) was entitled "Channeled Dalles (Bretz 1924), Oregon-Washington, Scabland," the adjective added to the local (3) back-flooding in valleys entering the term for the bare or almost bare rock of the disputed region from nonglacial drainages, drainageways, which Nimbus I "saw" from (4) the remarkable erosional and deposi- 500 miles above them. From that time on, tional features of the Columbia River valley the writer has insisted that these spillways from the plateau westward across the Cas- were not stream valleys in the ordinary cade Range to the huge delta at Portland, sense, that they had run essentially full dur- Oregon, and (5) various alternative hy- ing their erosion and were precisely what potheses. the name implied-river channels super- Pardee had earlier announced the former posed on greatly altered preglacial stream existence of glacial Lake Missoula (fig. 2) in valleys (Bretz 1928b). Only extraordinary mountainous western Montana. He wrote: flooding could have crossed the violated "As long ago as 1885, Professor T. C. preglacial divides, and only extraordinary Chamberlin noted a curious phenomenon in velocity (born of huge volume) could the Flathead Lake region that he aptly scarify the bedrocks so tremendously. described as 'a series of parallel watermarks More than eyebrows were raised among of the nature of slight terraces sweeping geologists by this interpretation. Almost a around the sides of the valley and encircling storm of protest arose. Where did that un- the isolated hills within it, like gigantic heard-of quantity of water demanded by musical staves' " and that "Chamberlin the proposed explanation come from in so conceived the idea of a glacial dam and short a time? Why did not the great, rock- furthermore tentatively suggested that its walled Colorado Grand Canyon, with vastly location was in the Pend Oreille region with more time and only normal stream vol- outflow by way of Spokane" (Pardee 1910, ume for its development, supply the correct p. 376-386). explanation? Allison (1935) proposed a pos- By 1930 the writer had become con-

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 508 J HARLEN BRETZ vinced from a field study that the topo- interpreted as giant current ripples (fig. 3). graphic setting of the lake's ice dam and the Their heights ranged from 15 to 50 feet altitude of the lake surface had contained Clearly these discoveries were part of the the elements for a catastrophic emptying record of a collapsing dam at the north end when that dam should weaken (Bretz of the Bitterroot Range. This strongly sug- 1930a). Then in 1942 Pardee found, on three gested to the writer that other such giant low gaps in the crest of a partially sub- current ripples must exist along the scab- merged, mountainous, peninsular projec- land routes. Their discovery, however, did tion of the former lake's irregular outline, not come until 1952. that an abrupt lowering had produced great The building of the Grand Coulee Dam currents headed westward toward the dam across the Columbia near the Coulee's head 100 miles distant. He estimated the peak and the development of the Columbia Basin discharge at 10 cubic miles per hour and Irrigation Project provided a superb op- found a descending series of arcuate, asym- portunity to secure new data on the prob- metrical ridges of flood-transported debris lem. What made the final solution possible lying transverse to the slope beyond the was a cooperative field study of these new downstream end of each narrows. These he sources of information by H. T. U. Smith,

0 20 40 60 80 Scale - miles

FIG. 2.-Outline map of Lake Missoula. Adapted from Pardee (1910, fig. 4)

PLATE1.-The Channeled Scabland as "seen" by Nimbus I from an altitude of 500 miles. About forty named places can be identified in this depiction. (Nimbus AVCS photograph, Goddard Space Center, September 5, 1964.)

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions JOURNAL0! GEOLOGY~VOLUME 77 BRETZ, PLATE I.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions JOURNALo~ GEOLOGY,VoLtm~ 77 BRETZ,PLAT~ 2

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 509 then of the University of Kansas, G. E. Neff giant current ripples on gravel bars, difficult of the U.S. Bureau of Reclamation, and to identify at ground level under a cover of the writer. Their report (1956) finally es- sage brush, were indubitably present (Bretz tablished the Missoula flood theory and et al. 1956, pi. 4, fig. 1). announced that Lake Missoula floods had Since 1957, Richmond has published occurred several times during the Pleisto- nearly a dozen papers on the glacial his- cene. Aerial photographs by the U.S. De- tory which he and collaborators have department of Agriculture and the U.S. ciphered east of the Lake Missoula region Bureau of Reclamation, supplemented by but within the northern Rocky Mountain obliques taken by Smith, disclosed a dozen province. That history involves three differ- localities on the plateau scabland where ent glaciations separated by long intervals

FIG. 3.-Map of giant current ripples north of Plains, Montana. From Pardee (1942)

PLATE2.-A: Wallula Gap through the Horse Heaven Hills anticline. Seen from Pasco Basin (photograph by Bretz). B: Giant current ripples on summit of shoulder bar, junction of Palouse and Snake canyons. Aerial photo CCH-16-197 from Production and Marketing Administration.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 510 J HARLEN BRETZ of nonglacial climate when erosion of glacial glaciation and three during the Pinedale drift deposits (table 1), accumulation of in- (one of them consisting of two episodes). terglacial sediments, and weathering of soils With the three pre-Wisconsin glaciations, occurred. he thus has found six "strong" soil-making Buried soils of varying development are a intervals when glacial ice had retreated vital part of these correlations. The latest from the region and three more times of ice paper in the Richmond series (Richmond withdrawal in the later record. et al. 1965) traces these old soil markers From his data, Richmond has deduced westward into the Columbia Plateau and the occurrence of three Lake Missoula identifies the occurrence of three Missoula floods. The earliest occurred at the close of

TABLE 1 GLACIALSTRATIGRAPHY OF THE WIND RIVER MOUNTAINS,WYOMING*

Terraces at Glaciations and and Bull Lake Stades Soils Interglaciations Interstades (Feet above River)

Neoglaciation Gannett Peak Stade Interstade. Weak azonal Temple Lake Stade Altithermal interval. Immature zonal Pinedale. Late stade. Interstade Weak azonal 40 Middle stade Interstade Weak azonal Early stade. Nonglacial interval. Mature zonal Bull Lake Late stade 50 Nonglacial interval Mature zonal 100 Early stade 200 Interglaciation. Very strong zonal Sacagawea Ridge 380 500 Interglaciation Very strong zonal Cedar Ridge. 660 Interglaciation Very strong zonal Washakie Point 760

* G. M. Richmond (1965) after Wright and Frey (1965).

floods, their times separated by soil-making the pre-Wisconsin glaciations, followed by intervals and by deposition of loess. Two one at the close of the late stade of the Bull ash layers (Powers and Wilcox 1964) and Lake glaciation and another at the end of (Fryxell 1965) have been identified as from the early stade of the Pinedale glaciation. Mount Mazama (CraterLake), Oregon,and If the Pend Oreille ice front reached the from Glacier Peak, Washington, their ages site of the Lake Missoula dam in each of 6,500 years ± B. P. and 12,000 years ± B.P., these nine advances, as many burstings respectively. From these data, the earliest might be expected in the record. Bretz et al. glaciation (which Richmond has revised to (1956) believe that the scabland topography constitute three separate events) is pre- records seven such floods, and Neff (unpub- Wisconsin and the following two (Bull Lake lished) believes that at least one flood oc- and Pinedale) are of Wisconsin age. curred before a heavy caliche, whose frag- Richmond has identified two stades of the ments are so prominentin flood gravels, had Cordilleranice sheet during the Bull Lake accumulated. If correct, there have been at

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 511 least eight floods corresponding with the preglacial stream valleys they entered on nine glacial retreats in northern Montana, the plateau. Small streams with propor- Idaho, and eastern Washington (see table 2). tionately large frictional contact with bottoms and sides had found this gradient ap- ESTIMATES ON DURATION, VOLUME, VELOC- propriate, but the great rivers had a far ITIES, AND THE MECHANICS OF EROSION smaller fraction of this retarding factor to AND TRANSPORTATION OF FLOOD-DERIVED deal with, and hence a far greater rate of WASTE flow. The early estimate that a Missoula flood Estimation of velocities attained range would require 2 weeks to pass through from Flint's "leisurely" rivers to the high Wallula Gap (fig. 1 and pi. 2A) was based rates implied by the erosion transportation on Chezy's formula, where two wetted consummated. The use of Chezy's formula perimeters and a surface gradient could be by Higgins indicates a velocity through obtained (estimate by D. F. Higgins in Wallula Gap of between 17.5 and 20.6 feet Bretz 1925, p. 258). The time estimate also per second; 20 feet per second is 13.6 miles considered, incorrectly, that the full 500 per hour. The total distance involved was cubic miles would be available. The error is 15 miles, and the gorge thus was refilled al- obvious when the altitude of the Clark most hourly. Fork valley floor at the dam site, about These figures are, of course, speculations, 2,000 feet A.T., is compared with the aver- unchecked for several unknown variables age altitude of scabland channel heads (ex- such as preflood depth of the Wallula water- cept for Grand Coulee's later deepening) gap, original width, loss of energy from the of 2,500 feet A.T. Only when no Okanogan erosion that unquestionably resulted and lobe (see fig. 1) blocked the Columbia west from transportation of debris, possible later of the head of Grand Coulee, and thus no uplift in the Horse Heaven Hills anticline, detour was probable across the plateau, etc. New maps of Wallula (1964) and Uma- could the entire lake discharge have had tilla (1962) quandrangles with smaller con- free flow down the Columbia Valley. tour intervals and larger horizontal scales Pardee's estimate of discharge rates along show that the highest flood probably one of the main arms of Lake Missoula reached almost 1,200 feet A.T. much exceeds Higgin's estimate at Wallula Some critics have disliked the idea of so Gap. much floodwater erosion in so brief a time Nevertheless, the writer is convinced that as the writer has been allowing. This ob- the narrow Wallula Gap was a bottleneck at jection was answered by citing the close some maximum flood or floods, because vertical jointing of the basalt lava flows upper altitudes of back-flooding in the and the argument that not abrasion but valleys of the Snake, Walla Walla, and plucking of large fragments was inevitable Yakima rivers correspond closely with the under the high velocity of the huge rivers. Wallula's upper limit. Capacious Yakima Abrasion of large fragments to cobble and Valley in particular carries an impressive pebble sizes occurred largely during trans- record of a back-flooding of very dirty, very portation. Many boulder deposits lie but turbulent water that deposited a cover of short distances downstream from channel- silt (mostly rehandled loess) and fine angu- bottom outcrops of large-columned basalt, lar basalt granules with plenty of iceberg- and many such boulders still possess transported foreign boulders-surely not columnar outlines in part. the deposit of a static lake, surely that of a Gerard Matthes, eminent authority on strong up-valley current (Bretz 1932). river dynamics, has described (1948) the Besides great volume, the scabland rivers "kolk" in river-bed erosion and has agreed owed much of their velocities to the gradi- that its mechanics are fully applicable to ents, averaging 20-25 feet per mile, of the the scabland rivers. The kolk is a marked

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions t

RkRocky Bowers-Lind- text).

(see in fill;deep- in falls Basin summit gravelMax.ening Drumheller destroyed Quincy Channels Warden Ringold

Flat, Basin Coulee across building Eureka Bacon and Bar flood SYSTEM* Lakequadrangles Maid, Basin Lake Hud-and Old ParkCity Maximum sonTrail falls equal COULEE Hartline Initial

possibly Ce bar GRAND Crab Creek THE 2 Lower IN Beverly

ash of Creek Wil- Bowers-Lind-Rocky TABLE CraCrab Creek t EVENTS Coulee Creek peat OF in +12,000B.P. dam,mouthUpper Crab sonripples UUpper Volcanic Dry Lowest

ValleyValley bar scab- Head Ever-

CORRELATION green-Bab-cockland Coulee intervals. ColumbiaColumbia Beverly Later or

ash falls; Lake events Grand shells, of TENTATIVE and Coulee andSoap CastleGreatBlade Lower Volcanic Dry duration

Co- gla- Grand reach indicate Coulee formationfallsColumbia Valley;Lakelumbiadrained Steamboat Rock ciated;Withrow moraine; Northrup canyon not Upper NespelemSteamboat Isolated do

flood flood flood spacings flooda a no flood flood Lobe no a a Okanogan Vertical dam; dam; dam; ~ No Dam;No Dam; Dam; No

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions plateau Rockyt initiationof anasto-mosis Bowers-Lind- Functional

fill; into into Co- (early Co- (Bull Basin west-cata- Warden Quincybasin ernracts; Drumhellerinitiated fromlumbia Valley Quincybasin gravelfalls Discharge fromlumbiaValley Pinedale) Lake) Quincy Growing Three Discharge of

Basin

prefloodfill Bacon Excavation

Basin

Hartline

Crab Othel- Creek Channel, lo andKoontz Coulee Lower Initial

2-Continued Wil- Crab Creek Creek WilsonCreekripples sonripples TABLE Upper Intermediate Highest

into ValleyG Gr.HaHead scab-dis- Cra- (early Co- (Bull Ever- notchQuincy west- cata-

o below green-Bab-cockland,charge from terinto basin Pinedale) ernracts fromlumbiaValley Quincybasin Lake) Coulee Columbia Early Three Discharge of Cou- course Grand mono- Coulee monoclinalcanyon alonglee cline Lower Deepening Initial

falls over Grand retreat cas- Steam- Coulee of boat cadeCoulee monocline Upper Early Initial

flood flood a a flood flood Lobe a a Okanogan dam; dam; Dam; No Dam; No

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 514 J HARLEN BRETZ turbulence with upward lift and negative Gap to Portland, nearly 100 miles (Bretz pressures which develop at the foot of 1925), were checked by Allison (1933), who steep, downstream-facing declivities in the agreed in general with the field facts but bottoms of stream beds. In effect, it is a differed in interpretation. Allison asked for subfluvial tornado, an analogy which Mat- a dam 900-1,100 feet high to account for thes approved. He noted that the kolk is the high water downstream from Wallula. "probably the most important macro- Flint (1938) also wanted a dam to cause his turbulence phenomenon in natural streams" prescabland aggradation in the plateau and "the most powerful form of concen- valleys. A later study by Lowry and Bald- trated energy at work on stream beds." win (1952), dealing with that part of this Plucking action thus was greatly augmented stretch across the Cascade Range known as wherever submerged basalt ledges with the the Columbia Gorge, reported no evidence proper form appeared during flooded regi- that damming of any kind (volcanic, mens. Many low-cataract cliffs and their tectonic, glacial, landslide) as advocated plunge pools therefore were only "sub- by Allison and by Flint occurred at any fluvial cataracts." The comment by an time when the Columbia was carrying engineer that "basalt is a rock hard to drill glacial meltwater from Canada, Montana, but easy to blast" is apropos. Idaho, or Washington. An unusual problem of transportation A detailed study of the Portland region exists in the gritty silts with mingled (Trimble 1963) accepted the writer's in- weathered and unweathered angular basalt terpretation of the extensive gravel deposit granules and erratic boulders that were in the broad valley at the junction of Colum- carried back up nonglacial valleys which bia and Willamette rivers as flood-made. enter the plateau scabland. Only a surface The "Portland Delta" (Bretz 1925, figs. gradient could have produced the up-valley 16, 17) is composed dominantly of gravel currents. At the bottom, the flow was all with foreset bedding that dips westward in uphill! Adequate turbulence for the full most exposuresand was originally described affected length of each valley is obviously as a "subfluvial delta," a "subaqueous required, unless some means of flotation deposit" made under perhaps 100 feet of existed. For the boulders high on valley floodwater. It covers at least 200 square slopes, floating masses of ice are indicated. miles. (There are no conventional deltas in For the fines which blanket gentler slopes, a any scabland gravel deposit. All show that slurry of excessively muddy water from their surfaces were swept by strong, deep erosion of loess, perhaps charged with currents during accumulation.) Trimble minutely crushed ice, may be envisaged. has mapped even larger areas of contempo- With waning of a flood, most such fine raneously deposited sand and silt beyond debris would settle to the bottom of any the limits reached by the gravel. They lie back-flood valley, where the bulk of the southward up the Willamette and Tualatin record is found. Here, also, the excessive valleys and northward down the Columbia turbulence had diminished to a minimum, Valley. His inclusive term is "lacustrine for such valley-bottom deposits are pre- deposits." He wrote of "flood waters of al- vailingly sorted and stratified. In places most unbelievable proportions" and was there is evidence of repetition of the epi- fully aware of the source. He obtained the sodes. Yet these back-flood deposits re- ponding, up to 375 feet A.T., by means of a main, to the writer, the least understood of hydraulic damming at a constricted place all the phenomena that record the Missoula in the Columbia Valley about 30 miles floods. north of Portland-the same mechanism The writer's descriptions of massive which the writer used for the backup north erosional and depositional flood effects along of Wallula Gap. the Columbia River Valley from Wallula Earlier students of the geology of the

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 515 broad lowland between the Cascade and field evidence to support this view is at Coast ranges, now containing valleys of the "The Narrows" near the junction of John Columbia, Willamette, Tualatin, and other Day and Columbia canyons. rivers, noted evidence there of widespread Here the John Day flows parallel to the ponding with ice-rafted erratic boulders up Columbia for its last 9 miles, separated to 400 feet A.T. (Condon 1871; Diller 1896; from it by an asymmetrical, peninsular up- Washburne 1915; Allison 1935). The writer land about 2 miles wide and 1,200 feet A.T. earlier (1919) espoused Condon's idea of a Most of this upland drains to the John Day, "Willamette Sound" at sea level but later whose canyon walls are dendritically dis- abandoned it for the glacial-floodhypothe- sected and nowhere possess cliffs. sis. Lowry and Baldwin (1952) have re- In marked contrast, the Columbia can- turned to the concept of a eustatic rise in sea yon wall along the peninsular divide is a level as the cause of the ponding. continuous cliff for the entire distance and A situation closely comparableto that of ranges from 660 to 800 feet high. No ra- the Portland Delta appears to have existed vines dissect it. The Columbia's side of the at the downstream end of Wallula Gap divide has obviously been cut back marked- where the broad Umatilla structural basin ly by widening of the master valley at some was entered by floods from the scabland. geologically recent time. Here a rather featureless gravel deposit Both valleys had present depths before covers many townships thence for about 30 any glacial floods arrived. Both had been miles down valley, beyond which the river eroded well down into basalt. again encounters a rock-walled course near At the east end of the peninsular upland Arlington, Oregon. A dozen or more scab- was a low place where a flood in the Colum- land-type basalt buttes project through the bia Valley spilled a portion of its volume gravel, and several broad abandoned chan- across and made the sharply defined but nels incise it. The topography of the gravel very limited basalt scabland called The is not that of a deltaic deposit in a body of Narrows (fig. 4). standing water. It consists of broad, low Only a brief life for The Narrows spill- bars distributed through an altitude range over could have been possible before back- of 300 feet or more and declining in altitude flooding up the John Day brought it to an westward. The irregular gorge beyond the end. Yet in that time the distributary had a Umatilla basin is in places nearly 800 feet cascade down into John Day Canyon that deep, and the upland above its walls carries began as a drop of about 500 feet. This cas- flood channels with floors of 700-800 feet cade eroded the basalt floor of The Narrows A.T. Here many stranded granite boulders to depths as great as 200 feet, making cliffs lie above 1,000 feet. Only a few miles 100 feet high, a maze of rock basins, and farther downstream is a spectacular spill- vertically walled castellated buttes, and over, southward out of the Columbia gorge leaving a huge mid-channel butte whose and into the closely adjacent tributary summit, half a mile long, quarter of a mile John Canyon, its original here Day floor wide, and 956 feet A.T., recordsthe preflood 950 feet A.T. (Bretz 1928a). surface (Bretz 1928a, pi. 23, figs. 1, 2, 15). CHARACTER OF A FLOOD FRONT Debris from this erosionbuilt a bar on the By Both Snake River Valley, upstream from bottom of John Day Canyon. the time the entrance of the easternmost scabland the back-floodingin the John Day had been rivers, and Yakima Valley, southwest of consummated, this deposit had spread the plateau scabland, carry evidence of across the canyon to a depth of 360 feet high-energy back-flooding currents that over an area of nearly half a square mile. strongly suggest a steep borelike front of the It was completed as a great bar, and hence entering back-flood. But probably the best the John Day back-flood had topped it

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions the the haveof north spillover. justsouthward along scales,Some

steeplyNarrows different essentially quadranglein The on quadrangle. of Peak course

Survey, apparentlycause Satus Arlington the thethe antecedenton of Geologicalon distance, an is part U.S. heldbearing vertical hasany maps,scabland southernthat riverhas of the the Narrowson most course that quadranglefor TheHills, valley of Wasco cliffedappearsand andMost It Columbia steeply high. theis structure as Arlingtonomitted.and half upwarp,between contoursbottomabout adjoining

The valleyonlyrelations is 200-footanticlinicalthe the the sidein environs.above butmarked south and A feet theNothing on contours2,000 Oregon, all wall than quadrangle.structure. and the morevalley Narrows, of Wasco The The the flank of summit agreementa to on flows. steep lieshas basaltthe 4.-Mapbar broughtriver, of FIG. the beengravel of dipping strike

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 517 before The Narrows spillover had ceased to snake Hills anticline that crosses the supply detritus. Yakima Valley mouth to become part of The figures are simply extraordinary for the large Horse Heaven Hills anticline. such a brief episode. The Columbia flood From average widths of 15 miles, the struc- front, after initiating The Narrows spill- tural valley narrows at Chandler to 3 miles over, had only 9 miles farther to go before and the stream-made notch is less than a it must have sent a portion back up the mile wide (Bretz 1930b, figs. 8, 9). The John Day to reach the foot of the cascade. Corral Canyon quadrangle map (fig. 3) Perhaps The Narrows has carried several shows the anticlinal crest to have a topog- spillovers, but its initiation required much raphy wholly unlike anything else in the more than 700 feet of floodwater in the upfold or on any other surfaces in the dis- Columbia Valley. trict, but is duplicated in thousands of The writer is convinced that no gradual places in the Channeled Scabland complex rise in the flooded Columbia can account of the plateau. It is butte-and-basin scab- for these features at The Narrows. The land (Bretz 1930b), as isolated as that at flood arrived catastrophically. It was a The Narrows on the Columbia-John Day great wall of water, its crested front con- divide, and, like the glacial silts and stantly outrunning and overrunning its boulders for 50 miles up the Yakima, basal portion. Because of huge volume, its whose upper limit is 1,100 feet, it also has current was essentially unimpeded by an upper limit, but here only 950 feet A.T. channel bottom and side friction. There are ten rock basins strung along the Because a flood front normally becomes crest of the little anticline below 950 feet attenuated downstream and no steep front A.T., and hills of tilted basalt 100 feet high as here interpreted at The Narrows could stand among them. Total relief is more have been maintained across the broad than 200 feet in a horizontal distance of 500 Umatilla basin farther upstream, an ac- feet. cessory mechanism for regenerating one is Although the Yakima River clearly has required. Such a mechanism probably lies in never flowed on this crest, stream gravel a hydraulic damming in the encanyoned occurs in several places. A coarse, rubbly ba- Columbia (800 feet deep and 1 miles wide) salt gravel with a few granites, quartzites, a few miles upstream from The Narrows etc., lies on the western (up-valley) slopes spillover (see Arlington quadrangle map, of bedrock eminences and is foreset up the U.S. Geological Survey). This would cause Yakima, away from the Columbia. This lo- considerable backup in the Umatilla basin, cal up-valley current tore loose much boul- raising flood levels there and reconstructing dery basalt, some of it fresh, some considera- a steep front in the canyon. Evidence of bly weathered, which came to rest in the hydraulic damming with accessory effects foresets. is to be expected in all marked narrowings The writer (Bretz 1930b, fig. 6) has in- of the preglacial Columbia. dicated that the Yakima Valley upstream Another demonstration of the character from Chandler Narrows and below 1,100 of an arriving flood's front is visible close to feet A.T. has a capacity of 3.17 cubic miles the junction of the Yakima and Columbia and could be filled by the Columbia River valleys, which here lie in nearly parallel in maximum flood today up to 1,100 feet structural downwards, separated by the A.T. in about 4 days. However, only strongly expressed Rattlesnake Hills anti- tremendous velocity engendered by a steep cline. surface gradient can account for the The Yakima structural valley is notably Chandler Narrows' erosional features (fig. constricted some 10 miles upstream from 5). One thinks of repetitions of the cata- the junction of the two rivers by a minor strophic experience. The record of Missoula upfold on the southern slopes of the Rattle- floods will provide this!

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 518 J HARLEN BRETZ

The scabland torrents were excessively plateau and found en route very little loess turbid. Relict loessial hills near the north to carry off. Such floods were not only less end of the Cheney-Palouse tract are 100 turbid, but were far less impeded en route feet high, and loessial scarps overlooking to Wallula. They must have arrived in the Palouse-Snake scabland divide are 200 greater initial volume than floods forced to feet high. This denuded tract has lost nearly thread the complex of plateau channels. 1,000 square miles of its former loess cover Wallula's westernmost scabland channel to these torrents. Furthermore, the back- atop the 700-foot cliff summits (fig. 6) is floods in tributary valleys (Snake, Walla about 60 feet deep and much too wide to be Walla, Yakima, Palouse, etc.) left an almost ascribed to its preglacial predecessor. The universal deposit of the rehandled loess, gravel bar and gravel floor south of the col mixed with fresh and decayed basalt grains and the huge gravel spill thence down into and granules, up to the limits reached by ColumbiaValley indicate considerabledeep- the floodwaters. ening in the channel's headward part and In contrast, any flood that encountered an adequate spill across at the inception of no Okanogan dam followed the open, its use, perhaps of the order of 100 feet in capacious Columbia Valley around the depth. The upper flood limit may well have

FIG. 5.-Map of Chandler Narrows, Washington. From Corral Canyon quadrangle topographic map, U.S. Geological Survey; contour interval, 20 feet. Although this map of 1951 departs in no significant way from the map of 1917 (Bretz 1930b, Fig. 9), it depicts an even more rugged topography of the gutted portions of the minor Chandler anticline, shows a sharper steepening of the anticlinal nose above 1,080-1,100 feet A.T., and reveals the existence of a few undrained depressions on the lee side of the devastated portion of the anticlinal crest. The structure could hardly have been more advantageously placed for receiving the record it carries.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions FIG.6.-Scabland on summit,west side of WallulaGap. From Wallulaquadrangle topographic map, U.S. GeologicalSurvey; contour interval 20 feet. Largelymodified preflood topography. Most significant place is in sec. 31, betweenR and S in HORSE,where a gravelbar just south of the channelentering from the north has detouredthe preglacialravine entering from the southwestwhich appearsto have once continued north along the channeledroute. The bar originallydammed local drainageon the east side of the channel,making a depression25 feet deep, now drainedby postflooderosion. Basalt surfacesat this highest recordof floodwaterare of scablandcharacter. For a mile, this southward-leadingchannel descends about 175 feet on a gravelfloor and there pitches off for 500 feet in less than 2,000 feet of horizontaldistance to reach the bottom of the gorgejust south of the narrowestplace in the Gap. This pitch-offis essentiallya delta which is all frontal slope; all of its detrituscame from flood erosionin the channeland a tributary scabland-marked channel east of it.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 520 J HARLEN BRETZ been close to 1,200 feet A.T., 100 feet higher flood-made Pardee's giant current ripples in than the upper limits of back-flooded sandy the eastern part of the lake or the pro- and pebbly silt known on the slopes of nounced scarifying of lower mountain walls Yakima and Walla Walla valleys. Possibly and the building of high eddy deposits back the relatively clear floods that went around in reentrants below the lake's level. the plateau left this high record. But they The two stades, one with two episodes, of should have carried berg-borne erratic the Bull Lake glaciation are recorded on the boulders, but no one has yet specifically Columbia Plateau in Washington by three noted any lying higher than the upper definite units in the loess cover, each limits of the pebbly silt. One thinks there- carrying a caliche record of long exposure fore of the alternative, that a slight post- before burial. This compound glaciation was flood uplift along the axis of the Horse followed by another flood at, according to Heaven Hills anticline may account for Richmond, about 32,000 years B.P. these aberrantly high channels on the west Excavation for the Wanapun dam on the side of Wallula Gap, and thus abandons the Columbia (Fryxell 1962) uncovered a speculation regarding floods around, versus coarse gravel in a flood-eroded rock channel. floods across, the plateau. The gravel was almost wholly of angular, poorly sorted basalt fragments, foreset- AGE ESTIMATES AND DETERMINATIONS bedded with cut-and-fill structures and the Richmond's latest correlation chart (1965) debris open-work in part. This was clearly of Quaternary events in the northern Rocky a deposit from the operation of nearby Mountains (table 1) names three glaciations Frenchman Springs cataract and thus a as pre-Wisconsin, two as Wisconsin, and a record of Grand Coulee's functioning. In neoglaciation (Little Ice Age) of valley the gravel were fragments of wood and glacier growth. Blackwelder's pre-Wisconsin dried peat which were C14-dated as 32,700 Buffalo Gap glaciation becomes the Wa- ± 900 years B.P. The organic material was shakie Point, the Cedar Ridge, and the clearly reworked from an interstadial bog Sacajawea glaciations, verifying Black- back on the plateau. The great coulee there- welder's suspicion that more detailed studies fore operated at least once to make the would subdivide Buffalo Gap. Richmond's depression for the bog (probably in the first Missoula flood is indicated as late in Quincy basin) and once again to tear up the the Sacajawea glaciation, possibly 100,000 bog accumulation at the time the cataract years ago. operated. Only the wood and peat are 32,700 Friedman (in Richmond 1965), from hy- ± 900 years old. dration rims on obsidian pebbles in till of the The identification of two volcanic ash early stade of Blackwelder's Bull Lake members in the Palouse loesses (Powers glaciation, suggested that the event oc- and Wilcox 1964; Fryxell 1965) provided curred 80,000 years B.P. Richmond's earli- two more dates in plateau history. The est flood is indicated as preceding the early Glacier Peak eruption occurred 12,000 stade of this glaciation and was therefore years B.P., the Mazama eruption 6,500 years pre-Wisconsin. B.P. The southern part of Grand Coulee is a Alden (1953) surmised that Lake Mis- closed basin containing alkaline Soap Lake soula had been formed and emptied two or in its deepest part and diatomaceous lake three times and suggested that "lowering of silts on higher slopes (Richmond 1965). In the ice dam might result in floods of great these silts is the dated Glacier Peak ash. magnitude-even if Lake Missoula were only Grand Coulee, the greatest and functionally partially drained. Such may, perhaps, have the latest of all flood channels on the pla- been the origin of many violent floods that teau, therefore ceased to operate before the are supposed to have swept over the scab- Glacier Peak ash was distributed, 12,000 lands." However, he could not accept as years B.P. A later flood (Bretz et al. 1956,

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 521 but not noted by Richmond) found an The Telford-Crab Creek scabland tract open Columbia. Thus the Okanogan lobe's in the north-central part of the plateau has marked morainic record was made before yielded another controlling date. A bog in a 12,000 years B.P. Richmond's latest correla- rock basin near Creston (Hansen 1947) con- tion chart shows his last flood as early tains volcanic ash layers of both Glacier Pinedale, perhaps 20,000 B.P. Peak and Mazama ages. This evidence

GRANDCOULEE DAM

WATERVILLE PLATEAU

oWilson Creek se

CO* ^^ Qui'ncy QUINCY BASIN Moses Lake

Othello

*Connell RINGOLD PLAIN

FIG. 7.-Structural features involved in the history of Grand Coulee floods. Data from U.S. Bureau of Reclamation.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 522 J HARLEN BRETZ

agrees with that from Grand Coulee; all tion, it found no preglacial valley head, but the plateau floods must have occurred before crossed 20 miles or so of the northern divide 12,000 years B.P. of the plateau before encountering such a The largest scabland tract, Cheney- valley, and it trenched that divide nearly Palouse, also has a contribution to this 1,000 feet deep. effort to date the Missoula floods. In the The Columbia Plateau of eastern Wash- wild, canyoned scabland where the main ington had earlier suffered local warpings part of the Cheney-Palouse floods broke and foldings that produced originally un- across the Palouse-Snake divide (fig. 7), drained basins and narrow anticlinal uplifts (Fryxell and Daugherty (1962 and 1968 (fig. 8). In the Grand Coulee system, one press announcements) have investigated a basin (Quincy) covers more than twenty rock shelter archaeological site (Marines townships and now is deeply aggraded with Ranch). The oldest burials were beneath basaltic sand and gravel. One great mono- shell midden with a C14 age of 12,000 B.P. cline (the Coulee monocline) lifts its high ±+ 100 years, and other burials were under western side about 1,000 feet. The highest of Mazama ash (6,500 years B.P.). Thus no three marked anticlines (Saddle Mountain) Cheney-Palouse discharge has occurred rises 2,000 feet above the bottom of its since about 12,000 years B.P. closely parallel synclinal valley (Lower Crab Creek), which served as one of Grand Cou- GRAND COULEE SYSTEM lee's distributary routes. The Grand Coulee system provides the Grand Coulee is a tandem canyon that greatest display of flood-made features, heads in a notch in the south wall of the although its areal share of the plateau preglacial Columbia Valley with a floor 650 scabland is small. The great coulee is the feet above that valley bottom. Its northern deepest, longest, and steepest in gradient. 25 miles, about 900 feet in depth and Its system possesses the largest number of ranging from 1 to 5 miles wide, cross the abandoned cataracts, the widest one, a high side of the Coulee monocline. The record of the highest, the greatest cascade, glacial stream, on encountering the down- and the largest number of distributary warp slope of the monocline, descended in a canyons. It is unique in that, at its initia- cascade that very shortly became a cataract

FIG. 8.-Park Lake quadrangle topographic map, U.S. Geological Survey. The 10-foot contour intervals show that the summit of the Coulee monocline is actually a minor anticline standing about 200 feet higher than semiparallel U.S. Highway 2 less than 2 miles distant to the northwest and nearly 1,600 feet above Blue and Park lakes. The brink of the great cliff ranges between 800 and 1,100 feet above the lakes in the coulee bottom. The five named draws which head on the southeastern flank of this anticlinal divide show very little rejuvenation since the great wall was made. Knickpoints in them are very obvious. The great oversteepened wall paralleling the strike of the monocline ends north of Park Lake by turning a right angle to follow down the dip of the structure and there leads into the lower cliffs of the Dry Falls recessional gorge. North of this, the monoclinal slope into the Hartline basin is only a smooth structural slope with the preflood draws un- interrupted by knickpoints down almost to the Dry Falls platform. East of the coulee lakes and of all tilted basalt flows, this map shows, north of Dry Coulee, a terribly scarified scabland anastomosis of canyoned coulees 100-400 feet deep with four subfluvial cataract alcoves and recessional gorges 100-150 feet deep. This tract was flood river bottom throughout, eroded beforethe lake-occupied deep gash along the monocline had been gouged out and Dry Falls initiated. High Hill anticline, in the southern part of the map, stood a few hundred feet above the highest flood. A subfluvial cataract alcove and its recessional gorge at the truncated western end of High Hill are separated from the main coulee by a remarkable bladed ridge. Where narrowest (extreme southwest corner of the map) it rises more than 100 feet above the bottom of the recessional gorge on the east and more than 450 feet above Lake Lenore in the main coulee on the west. This bladelike ridge is the attenuated toe of all that survives of the monoclinal slope before it was gutted by Grand Coulee floods. The recessional gorge with its cataract east of the Blade is the record of a preflood streamway between the Coulee monocline and the High Hill anticline. Lake Lenore basin occupies most of the width of the steep toe of the moncoline as its hogback islands testify.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions JOURNALOF GEOLOGY,VOLUME 77 BRETZ, FIGURE8

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions JOURNALOF GEOLOGY,VOLUME 77 BRETZ,FIGURE 9

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 523 which receded back into horizontal basalt it encountered (Quincy basin). One of these flows of the high side of the monocline. This abandoned cataracts, 165 feet high, is today recession eventually lengthened the canyon utilized by the main canal that leads from all the way across to the Columbia Valley's the equalizingreservoir on the Upper Coulee southern wall and thus destroyed itself. floor to the same depression (Quincy) An imposing mid-coulee butte, Steamboat where the Columbia Basin Irrigation Proj- Rock (Bretz 1932, figs. 20, 21), about a ect's largest present development exists. square mile in summit area and as high as Thus the Lower Coulee is bypassed by the bounding Coulee cliffs, for a time was a man's puny river from Grand Coulee Dam "Goat Island" in the receding cataract. to the fertile but arid region of the Quincy Further evidence for this interpretation basin. The need for the bypassing is ob- is the existence under the present detrital vious, for the Lower Coulee is a closed de- floor, at least as far north as Steamboat pression dammed by its own huge bar de- Rock, of a debris-filledtrench in the basalt posits in the basin. more than 200 feet deep, a greatly elongated For about 15 miles, the Lower Coulee's plunge pool of the vanished great cataract. course is superposed directly on the Coulee Just downstream from the site of the monocline with consequent erosional fea- original cascade, the coulee floor is of un- tures unknown elsewhere in the scabland trenched basalt with a typical scabland complex. The shattered rock in the steep river-bottom topography (Bretz, 1932 fig. limb (450-60°) of the monocline yielded so 6). This is the Upper Coulee. readily to the flood from the Upper Coulee For a few miles beyond, the glacial river that a small preglacial stream valley closely spread widely in a broad synclinal valley parallel but two miles or so farther east (Hartline basin), burying most of it in was left with only relatively minor altera- basaltic waste. But a lower upwarp, the tion. The flood channel moved over into High Hill-Trail Lake anticline at the south- the shattered belt, leaving a string of deep ern margin of the basin, was shortly en- lakes on subsidence. countered and the widened flood plunged The western wall of this monoclinal across that relief feature in several cataracts course rises, 1,000 feet above the lakes, and on its way to the lowest structural feature its crest line is gabled in appearance from

FIG. 9.-Coulee City quadrangle topographic map, U.S. Geological Survey; contour interval, 10 feet. All but about a square mile of this quadrangle has been swept by Missoula floodwater. What is probably the world's largest abandoned cataract lies in its northern half, and two large, compound subfluvial cataracts with recessional alcoves lie near midlength. Structural control is evident in (1) a small portion of the Coulee monocline's flank in the northwest (largely above flood levels), (2) a part of the gravel-filled Hartline basin along the eastern side of the quadrangle, and (3) the Trail Lake and a part of the High Hill anticlines (with Dry, Hudson, and Trail Lake coulees) which trend northeast from the southwest corner of the quadrangle. The nearly 500 feet of relief in the canyoned tracts south of the great cataract is all the product of flood erosion before that cataract (Dry Falls-Castle Lake Falls) was made. The Dry Falls cliff is about 400 feet high, and the complex of recessional gorges dominated by Castle Lake Falls totals about 300 feet in height above Deep Lake. Before the Lower Coulee's great gash was made in the flank of the Coulee monocline (Park Lake quadrangle) and the great cataract was eroded, earlier floods had traversed essentially the whole of the quadrangle and had given origin to Dry, Hudson, and Trail Lake erosional gashes. The highest surviving basalt of the scabland stands just west of the Hudson Coulee cataract alcoves and suffered but little. Its altitude corresponds closely with the easternmost scabland on the southern part of the Hartline basin fill and thus indicates a relief of nearly 500 feet before the great cataract was born. The Bacon syncline once carried a prescabland gravel fill comparable to that in the Hartline basin. Stream-lined, attenuated remnants lie southeast and east of Hudson Coulee. As far as shown on the map, the only human uses of this desolate area south of Coulee City are transverses (highways, dirt roads, jeep trails, and abandoned railroad), gravel pits, springs, the main irrigation canal, and perhaps jackrabbit hunting. Recorded only by name are the tragic attempts by two brothers to farm a portion of Hudson Coulee and the location of a particular (Deadman) spring. The total vertical range of flood-eroded scabland on this quadrangle is at least 650 feet.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 524 J HARLEN BRETZ truncation of the now-hanginglittle ravines miles distant. But the volume was so great that originally descended the monoclinal that a portion rose to overflow southward slope (Bretz 1959, fig. 6). But the out- past the eastern nose of Saddle Mountain standing erosionalform of the Lower Coulee anticline by way of Othello Channels is at the head and in the broad uncanyoned (Bretz et al. 1956, fig. 12), a small edition area (part of the Hartline basin) which of Drumheller. In both the Othello and separates the two canyoned portions of Drumheller cascades, a residual hill of Grand Coulee. It is Dry Falls-Castle Lake, sedimentary rock survived the tremendous an abandoned compound cataract with erosion the basalt received. Othello has a maximum height of 400 feet and a width plucked rock basin 135 feet deep (Eagle or of 31 miles (fig. 9). It had taken origin and Scootenay Lake) on the summit of the pre- been maintained only because it was ini- glacial divide. tiated a little north of the coulee's entrance Drumheller did not carry all the flood into the shattered belt. Plunge pool lakes overlow from Quincy basin. On the west side in alcoves at the base of the cliff mark of this basin is a broad, low, north-south several more concentrated strands of the ridge (Evergreen-Babcock) separating it floodwater (Bretz 1932, fig. 50). from the closely adjacent Columbia Valley. The distributary system of cataracts In three places, saddles in the ridge took and canyons which took much of the spill some of the floodwater, and three cataracts across Hartline basin's southern rim and (Bretz 1959, figs. 13,14,15,16,17) were born out of the east side of the Lower Coulee all as their take fell over the Columbia Valley returned to the northeastern corner of cliffs. Both Frenchman Springs cataract Quincy basin a few miles east of the major and Potholes cataract were 400 feet high coulee's mouth. This large basin also re- above the bottom of their plunge pools. ceived several distributary rivers from the Both were double falls with a narrow upper stretches of the Cheney-Palouse tract dividing septum. The total width of Pot- and from all of the Telford-Crab Creek holes cataract was 1 miles. Crater cataract tract (see fig. 1). Seven of these were con- was 200 feet high. All three had the same verged into one (Upper Crab Creek) before upper limit and must have operated simul- reaching Quincy basin. taneously. The debouchureof the main Grand Cou- A startling fact that spurred critics of the lee discharge out of its rock-walled course early "Spokane Flood" to propose even into Quincy basin involved a great slacken- more startling alternatives (Bretz 1928c) ing of flood current and the consequent was that the upper limits reached initially deposition of a known maximum of 186 feet by discharge over these three cataracts of gravel and sand over the twenty town- were identical with Drumheller's upper ships involved. The topography of the de- limit, 1,300 feet A.T., although 30 miles posit is that of great, broad bars with distant. It was the relative altitudes of bars separating channels (Bretz et al. 1956, fig. and channels in Quincy basin and of the 4) which lead southward toward the cataract upper limits which led to the con- Drumheller cascade (Bretz et al. 1956, pls. cept of a succession of floods down Grand 7, 10, fig. 1). Drumhelleris 9 miles wide and Coulee. Another startling fact is that at has a maximum depth of 300 feet in basalt. some time during Lake Missoula's repeated The gradient of the water escaping from outbursts, flooding back from Columbia the basin at Drumheller was 15 feet per Valley into Quincy basin occurred through mile. That water entered the synclinal the saddles used by the westward cataract valley (Lower Crab Creek) between the discharge to the Columbia. At such times Frenchman Hills and Saddle Mountain there could have been no Okanogan dam anticlines, and much of it there turned west and no scabland rivers, not even in Grand along that valley to reach the Columbia 20 Coulee.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 525

At Beverly, where Lower Crab Creek's tributary channels (Koontz Channels) all glacial river from Drumheller Channels hang at varying altitudes above the present joined the Columbia River valley, a well- Columbia. They entered the broad lowland marked bar almost 100 feet high and 1 of the Columbia River in the Pasco basin, miles long has been built across the mouth which must have been semiponded at of the synclinal valley by a Columbia flood. different times up to the varying levels of This bar is double, its structure well shown these hanging channels. in cuts along Chicago, Milwaukee, and Pasco basin contained the major part Saint Paul Railroad (Bretz et al. 1956, of the flood backed up by the hydraulic figs. 1, 11, pi. 12). A flooded Columbia damming at Wallula Gap 30-35 miles rolled this bouldery gravel 125 feet above south of the Koontz Channels. This also river level of today and back into the trib- was the ponding which backed up Yakima utary glacial river's mouth after Grand Valley and which left the great flock of ice- Coulee had completely ceased to function. bergs stranded on the eastern lower slopes The last Missoula flood thus was later than of Rattlesnake Hills. The Koontz Channels the last Okanogan lobe's invasion of the continued to be deepened while the Pasco plateau. basin water level was being lowered as much Pardee (1918) has described an extensive as 250 feet. Their relations to a range of 250 deposit of white silt, the Nespelem forma- feet in the level of the Pasco pool are evi- tion, in the Columbia Valley upstream from dence of perhaps three different Missoula the head of Grand Coulee and on much of floods, during each of which Grand Coulee the floor of the Upper Coulee itself, but functioned. unknown farther down either drainage course. Only quiet water could have pro- EVERGREEN-BABCOCK RIDGE vided for the deposition of this silt, and The west side of this broad ridge has a only an empty coulee from that time on can pronounced structural bench which slopes explain its survival on the floor. The last southward 30 feet per mile from Crater to Missoula flood was over by this time, but Potholes and has a maximum width of 1 an Okanogan lobe dam must have held up mile (Bretz et al. 1956, pi. 11, figs. 1 and 2, this latest (and local) ponding in the Colum- pl. 13, fig. 2). For 4 miles north of The bia Valley. Potholes the bench surface is scabland and If these conclusions are correct, the flood the 100-foot cliff above it is clean-cut. But that built Beverly bar (clearly later than north of the 1,300-foot contour crossing, the last of Grand Coulee's discharges and the bench is much dissected and has no clearly a record of an open Columbia scabland or cliffs. The rise here to the Valley around the northwestern part of the summit of Babcock is over a completely plateau) preceded the last advance of the soil-covered surface much less steep than Okanogan lobe. the cliff, although held up by the same ba- The damming for the Nespelem silt salt flow. episode may well have been only a partial Floodwater in the Columbia Valley has reclaiming of that lobe's maximum stand clearly swept over that part of the bench and certainly was not the one which pushed below 1,300 feet A.T. But both cliff and over into Grand Coulee's head portion. Yet scabland record two different times of such its failure as a dam may have released the flooding. North of about 1,250 feet A.T. on very last flood down Columbia Valley, and the bench, the cliff is talus-covered almost not a maximum flood at that. to the summit and has a mantle of soil with Most of Othello's discharge went south- sage and grass cover. The terrace surface west across a plain determined by the flat- below this subdued cliff, although definitely lying Ringold superbasalt sedimentary for- scabland, has few bare rock outcrops and, mation in the Pasco basin. The several dis- like the cliff, is mantled by coarse, weath-

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 526 J HARLEN BRETZ ered debris and soil. Farther south and across the ravaged divide that swept its below that altitude, the cliff is vertical debris across and 500 vertical feet up on the bare rock with an actively growing talus far (south) side of the Snake Canyon, pick- and the structural bench carries strongly ing up well-worn cobbles from the Snake expressed scabland. Clearly, here is scab- riverbed to be mingled with the rubbly, land of two different ages, a record of two coarse gravel yielded in the divide summit's floods down the Columbia, widely separated erosion. in time, and the earlier one the greater. Fryxell reports that the enormous gravel deposits made here in Snake Canyon con- FLOOD CROSSINGS OF THE PALOUSE stituted a dam for some time afterward, SNAKE DIVIDE upstream from which the terrace record of Palouse River in preflood time crossed the postflood Snake is "completely differ- from east to west what later became the ent" from the terrace system downstream. Cheney-Palouse scabland tract. Here its A semi-isolated basalt hill here had its valley, most of its loessial uplands, and its lee slopes and most of its summit buried in lower western extension, the now-empty this flood gravel. The southbound deluge Washtucna Coulee, were severely altered but was turned westward by higher hills to go proved far from adequate to carry the dis- down the Snake. The gravel-buried lee charge coming over the plateau's northern slopes of the semi-isolated hill show strik- edge and the Cheney region. ingly long foreset stratification dipping When Washtucna Coulee could not con- back (northward) toward the Snake (Bretz tain the great surge, floodwaters rose to et al 1956, pi. 15, figs. 1, 2, 3). The square make one of the most significant records on mile of gravel flat on this hilltop has a the entire plateau. In two places, they splendid display of great current ripples topped the divide between the nearly (Bretz et al. 1956, pi. 13, fig. 1) 500 feet parallel preglacial Palouse and Snake val- above the river of today. This original in- leys, crossing it to enter the Snake Canyon. terpretation has stood the test of detailed They swept away a loessial cover more than ground-level mapping, aerial photography, 100 feet thick from 80 square miles of sum- and inspection by critics. mit area and scored the basalt thus ex- Rounded Snake River cobbles and peb- posed into a congeries of canyons, cataracts, bles in this gravel deposit and similar flood and rock basin (Bretz et al. 1956, fig. 19). deposits in the canyon show percussion One canyon was cut down so deeply that flaking and fracturing to make forms called the postflood Palouse abandoned Wash- "broken-rounds" (Bretz 1929). Sharply tucna and now reaches the Snake through a angular fractured faces and edges have 400-foot deep narrow gorge that still clearly been superposed on originally con- possesses two waterfalls in its course. A tinuous smooth surfaces, and the associated state park includes the larger fall, 196 feet fine material just as clearly contains the high. angular chips and splinters broken off in Two rock basins on the very summit of some violent experience just before deposi- the divide (Bretz et al. 1956, fig. 20) are 80 tion. The percentage of these broken- and 120 feet deep, and the H.U. Ranch rounds is far higher than in the Snake River abandoned cataract is 280 feet from its gravel now in transit. lowest lip to the bottom of its plunge pool Equally significant is the record of an basin. Striking joint control of the reces- enormous back-rush eastward up the Snake. sional gorges which this and other cataracts In addition to a pebbly, sandy silt veneer on left has been noted by Trimble (1950) and gentle slopes up to the altitude of the ripple- is obvious on the topographic maps. marked hill summit and stranded glacial The writer's original interpretation (Bretz erratics reaching the same upper limit, this 1928a) asked for a flood from the north back-rush rehandled gravel in the bottom

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 527 of Snake Canyon and redeposited it with 26), three times as deep as Niagara gorge, foresets dipping upstream all the way to and its unnamed mate are quite out of har- Lewiston, Idaho, 70 miles distant. mony with the rest of this upper scabland A smaller spill across the Palouse-Snake and could not have originated until the preglacial divide, 15 miles farther west, great cataract had retreated this far north. made a single scabland canyon (Bretz It seems very likely that they record the 1923b, fig. 13) almost as deep as Washtucna same time that the summit of Steamboat itself. If Washtucna in flood was only a Rock was glaciated and that Okanogan ice distributary of the main Cheney-Palouse was then crowding well across to the Cou- discharge, Devils Canyon, cut de novo 400 lee's eastern wall. Such a sequence would feet deep in basalt, was only a distributary mean that the Telford-Crab Creek and of a distributary! Cheney-Palouse channels, which had been abandoned because of completion of the big THE SUCCESSION OF FLOODS notch in the Columbia Valley wall, were The head region of Grand Coulee has again functioning briefly. This is the ca- been noted as involving both flooding and veat noted. glaciation. A single flood could hardly pro- The summit of Steamboat Rock has a duce the 25-mile length of the Upper Cou- scabland channel across it (Bretz 1932, fig. lee's recessional gorge. Its sill at the head is 21); therefore, the rock had become isolated about 1,000 feet lower than all other scab- before the glacial record was superposed. land channel heads, and therefore, if an But whether or not the causal cataract had Okanogan dam then existed, Grand Coulee already reached the Columbia Valley wall, must have carried virtually all floods ar- a large ice-dammed lake (Lake Columbia of riving after its great cataract had breached Richmond) existed in that valley upstream the Columbia Valley wall it so deeply from the Coulee head. Waters (1933) has notches. Therefore, there could have been interpreted a hummocky, kettly topography no plateau flood farther east after the re- in gravel deposits on the south wall of cessional gorge was completed and the great Columbia Valley, opposite the mouth of Steamboat Falls was destroyed. One ca- Okanogan Valley, as a record of a breakup veat, however, should be considered. of the Okanogan lobe's dam. If a Lake The plateau summit where crossed by Columbia then existed, that is, if Steam- the Upper Coulee has approximately the boat Falls still fell short of reaching Grand same altitude as those eastern channel Coulee's head, a flood of respectable mag- heads, and on both sides of the Coulee head, nitude down the Columbia could result. such altitudes (2,500-2,550 feet A.T.) are If the breakup was triggered by the arrival typical scabland. These rugose surfaces, of a Lake Missoula flood, a back-flooding which were made before the great cataract (eastward) through saddles in Evergreen- had migrated that far north, are contem- Babcock Ridge would be almost inevitable. poraneous with the Telford-Crab Creek When the cataract did recede through the and Cheney-Palouse channel heads. valley's wall, this lake must have emptied But after completion of the Upper Cou- catastrophically. If the emptying of Lake lee gash, the Okanogan lobe's last advance Columbia coincided with another failure of pushed glacial ice across or nearly across to the Lake Missoula dam, the stage was set the east wall. If it was complete, the Cou- for by far the greatest flood that the Coulee lee's discharge was forced again to traverse has ever known. the summit scabland on the east side. It seems probable that the flooding over On this side are two short, semiparallel into Hartline basin, across Trail Lake cataract gorges and alcoves which join the anticline and down Dry and Long Lake Coulee in the latitude of Steamboat Rock. coulee to the northeastern corner of Quincy Northrup Canyon (Bretz 1932, figs. 24, 25, basin, occurred when Grand Coulee was

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 528 J HARLEN BRETZ taking such a flood. Evidence for this is (150 feet high) left two bars at successively found in a bar at the entrance of Dry Cou- lower levels as the floodwaters escaping from lee into the basin, built back into the mouth the Bacon syncline deepened its two outlets, of the Upper Crab Creek scabland channel. Dry Coulee and Long Lake Coulee. Here, It was a complete dam originally, and fore- three discharges apparently occurred out of set bedding in it dips east up-valley. The Hartline basin across Trail Lake anticline. modern creek has cut a narrow gulch However, the high gravel flat which through the bars but, for a long period in buries a caliche-capped loess is apparently the past, it retained a lake to the east, which only an erosional remnant of a once-com- is recorded by silts on the scabland valley plete fill in an early Bacon synclinal de- floor. Therefore no flood ever came down pression. It does not have the equilibrium any of the seven distributary channels from profiles that elsewhere mark subfluvial Cheney-Palouse or the five channels which gravel accumulations. Thus, only two floods lead from the Telford region during or after can safely be read from this place. Preflood the time Dry Coulee functioned for the last gravel is also known beneath the glacial time. This seems to show that the caveat flood gravel in Hartline basin (Kirk Bryan, should be accepted, that is, that there never personal communication). There it suffered was a renewed functioning of the eastern burial instead of erosion. spillways after Grand Coulee's great gash Summarizing, the maximum discharge was completed. down Grand Coulee came after its great But there had been three different floods cataract had receded across the plateau across the Telford tract before the Dry Cou- divide and all the eastern spillways had lee bar dammed Crab Creek. The evidence is gone dry. It flooded back into Hartline in three different levels of gravel deposits in basin, thence southward over Trail Lake the scabland channel of Wilson Creek, anticline, perhaps three times. Its Dry Cou- westernmost of all Telford distributary lee discharge, late in this episode, blocked routes (Bretz et al. 1956, pl. 9, fig 1). Here Upper Crab Creek channel. But Wilson the gravel bar on the channel floor carries Creek had already left a record of three excellently shown ripple marks with wave- floods. Thus the plateau scabland complex lengths of about 300 feet, the work of the involves three floods before the Grand last flood to use this route. Two higher de- Coulee obtained complete control of all posits, ranging up to 150 feet above the plateau floodwater and left a record of at rippled floor, have been scarped by later least two later floods. floods. Each carries giant current ripples Quincy basin's great spread of gravel has above its scarped drop-off. Those on the bar profiles almost throughout. Areally, uppermost and oldest bar have an amplitude there are two groups bounded and separated of nearly 20 feet. The series can hardly be by three channels almost entirely in gravel interpreted as anything other than the (Bretz et al. 1956, fig. 4). Channel depths record of three successive floods. and bar altitudes suggest three times of Just downstream from the spillways channel erosion initiated on an unchanneled over the Trail Lake anticline and the re- earlier fill. The highest parts of this fill now sultant canyoning and cataract alcove- stand 290 feet higher than the rock bottom making in its southern limb, there are of Soap Lake at the mouth of Grand Cou- gravel deposits at three different levels that lee. The three channels, however, functioned seem to record three different discharges. late in the basin's history, and the head of The highest is an isolated hill. Later flood the lowest is hardly 100 feet above Soap erosion has left it scarped on all sides and Lake rock bottom. Nevertheless, the deep- thus exposed a preflood loess with a caliche est part of Coulee discharge flowed up this cap beneath the gravels. On the west side, 100 feet. the discharge from Hudson Coulee cataract Why the channels? Their different alti-

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 529 tudes suggest a sequence in operation. They calcareously cemented loess. From con- must be read, however, as three originally sultation with Richmond, Neff dates the contemporaneous channels, and the reason earlier flood as Bull Lake and the second for the deepening of two is the deepening of (the one noted in this account) as early Drumheller that presumably was continued Pinedale. with each flood out of the coulee. Drum- The discharge through Drumheller Chan- heller's marginal scarps, cut in a super- nels across the nose of the Frenchman Hills basalt sedimentary rock (Ringold forma- anticline could not all be accommodated by tion?), record the highest and earliest flood, the westward-leading synclinal valley (Low- 1,300 feet A.T., the same flood that initiated er Crab Creek) between the Frenchman the three cataracts on the west side of Hills and Saddle Mountain anticlines, and a Quincy basin. Drumheller outran the cata- considerable portion spilled southward a- racts and finally reached a depth in basalt cross the eastern nose of the Saddle Moun- of about 950 feet A.T. tain anticline. This produced the Othello In the deepening of Drumheller is a group of deeply bitten channels and rock record of perhaps two Grand Coulee floods. basins and a mile-wide channel in the sedi- On both eastern and western margins are mentary cover on the preglacial divide minor, high-level basalt channels in the summit there. The entire group is 3 miles sedimentary rock, each once a cataract and wide. each partially cut away later by the deep- Fully as significant as this channeled ening of Drumheller. In themselves, they divide summit is the 8-mile channel from do not establish a sequence in floods. Taken Drumheller to Othello (Bretz et al. 1956, with the Quincy basin channels in gravel fig. 12); broad, rudely terraced, and and the cataracts of Evergreen-Babcock strewn with huge boulders, some of which ridge, a sequence is clear. Add to this the had to be blasted for excavation of an ir- existence of three faintly shown scabland rigation canal. Yet there is no continuous rivers (Rocky, Bowers-Weber, and Lind) gradient for this distance. Only a steep sur- entering Quincy basin from the Cheney- face gradient of a large volume of flood- Palouse region (Bretz et al., 1956, pi. 7) but water, arriving with adequate depth, can lying south of the group that discharged to account for this channel leading to Othello. Upper Crab Creek. They were abandoned No sequence in floods through Othello so early in flood sequences that, as shown can yet be read with confidence. But the in the Nimbus I depiction (pl. 1), they be- farther course of Othello water across the came obscure before the plateau floods had Ringold plain, 10 miles or more to the south ceased. in the Pasco basin, left a record of what Uncertainties and complexities in the appears to have been three floods. The Grand Coulee system's record do not justify Koontz Coulee channel group on this plain the concept of more than two floods, but consists of two sets of channels in the soft the suspicion remains, and from Neff's rock. The heads of the highest set hang 200 later, unpublished research, the number, feet higher than, and are truncated by, the still uncertain, probably is more than two. scarfs of those of the lower set. As pre- Neff has demonstrated that two floods viously noted, all Koontz Coulee channels out of the Columbia Valley into the Quincy hang in notches in the White Bluffs sum- basin have occurred, the first one before mits along the Columbia River. the strongly marked caliche was made. His A third flood (actually the first of three) sections, near the head of Crater Notch, is postulated from the existence of a mantle contain long, poorly sorted, southeast- of glacially derived gravel with basalt and dipping beds of slightly bruised basaltic caliche fragments lying higher on the Rin- columns, boulders of soft siltstone, and gold plain than any of the channels and angular masses of an older caliche and of 500 feet above the Columbia.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 530 J HARLEN BRETZ

A fourth and last flood, limited to the intervals of exposure before another back- Columbia Valley, is unquestionably re- flooding occurred. This subject has not yet corded in the Beverly bar which blocked attracted any investigator's close attention. the mouth of the synclinal valley of Lower Crab Creek and must postdate all Grand THE LAKE BONNEVILLE FLOOD Coulee discharge. Another great Pleistocene lake's out- Successive floods down the great Cheney- burst is to be noted in this account-the Palouse spillway doubtless occurred to cor- breaking of the lowest (earthen) rim of Lake respond with those in the Grand Coulee sys- Bonneville, the ancestor of the Great Salt tem before the Upper Coulee's 25-mile- Lake in Utah, and the funneling of the long 900-foot gash had been made. But the uppermost part of its volume into the record is not as detailed or as trustworthy. Snake River at a point nearly 600 miles up- One early flood is definitely established, stream from the entrance of Lake Missoula however, from the obscuration that the water at the Palouse-Snake divide. Rocky, Bowers-Weber, and Lind coulees Gilbert (1878) was the first to discover have suffered. Neff would date this flood as the breakout at the extreme northern end Bull Lake. To serve these three distributary of that pluvial, but nonglacial, great lake's coulees, its magnitude must have been basin. The rimrock which constituted the adequate to carry it also down Washtucna divide was poorly consolidated, deeply and across the Palouse-Snake divide. Later weathered alluvial waste between the lake floods failed to reach the heads of the three basin and a north-draining valley that led obscured coulees, because clearing away of to Port Neuf River, a tributary of the the loess and enlargement of the preglacial Snake River. According to Gilbert (1890), valleys in the broad Cheney-Palouse tract the overflow from Bonneville's previously had provided adequate capacity for them. closed basin (the Bonneville level of the The number of floods on the Columbia lake) trenched into this divide for 375 feet Plateau and in its master valley is one less before encountering more resistant rock than the number of interstadials and intra- that determined a new and lower (Provo) episodes in the northern Rockies. A com- level. Gilbert estimated that 482 cubic plete correspondence is not required by miles of water (cf. Missoula's 500) escaped our main thesis that Missoula burstings down this trench, and thought that the were the source of scabland rivers. Nor is it lowering continued for about 25 years. required that Missoula filled each time to Yet he called it a "debacle." its uppermost shoreline before its dam burst. Field evidence that this discharge across Indeed, Pardee (1942) recognized that in the Snake River plain had been vastly Flathead Valley, Montana, there had been increased for a brief episode is credited to two different fillings, the later one less ex- Malde (1960, 1965, 1968). His detailed tensive than the earlier. study of that basalt plain has found a com- The writer feels that the back-flood de- plex of abandoned high-level channels and posits in preglacial valleys tributary to cataracts and a number of river bars com- what became scabland routes will yield parable in magnitude, structure, and com- more knowledge and possibly a firmer position to those along the Missoula flood chronology of successive floods. These routes. Some are more than 100 feet high, deposits appear to record a repetition some block mouths of tributary valleys, of deposition under currents of varying some are separated from adjacent canyon strength alternating with erosion into walls by "marginal troughs" 150 feet deep. deposits already laid down. Some of the Down-valley foresets are consistently shown. erosion was clearly subfluvial under up- According to Malde (1968) they are similar valley currents, but some may have been to the gravel bars in the channeled scab- subaerial and thus may record fairly long land of eastern Washington. Like the scab-

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 531 land bars, they can be accounted for by the Idaho line. There it turned northward passage of a catastrophic flood. Malde's through the Snake River's Seven Devils estimate of the time involved in this Canyon, emerged into the Lewiston down- Bonneville overflow is much less than warp, turned westward again, and entered Gilbert's that 70-mile-long portion of the Snake's He also states (1968, p. 10) that the canyon which has herein been described as catastrophic overflow of Lake Bonneville the scene of a great up-valley back-rush was "about 30,000 years ago." This would from the Cheney-Palouse flood. place it between the Bull Lake and Pine- Here is a challenging situation for the dale glaciations as dated by Richmond. historically minded geomorphologist. The He adds that "the Missoula Flood far ex- flood record in this last 70-mile stretch is, ceeded the Bonneville Flood in violence al- as far as is known, all back-rush! It sug- though it was only a little greater in gests that the last great up-valley invasion volume." was later than, and erased any record of, Malde's latest contribution on the Lake the earlier Bonneville flood. Bonneville flood revises Gilbert's opinion Altitudes of the pebbly, gritty silt and that the prism of water involved is indi- floated erratic boulders in the Snake Can- cated by the vertical distance between the yon upstream from the Palouse-Snake Bonneville (highest) shoreline and the divide transsection indicate that the back- succeeding Provo shoreline. Malde and his rush must have reached at least 1,300 feet associates, Bright and Rubin, are confident A.T. Its effects in the broad, downwarped that the lake, prior to the flood episode, was region of Clarkston, Washington, and much lower than the Bonneville shoreline Lewiston, Idaho, are complicated by other and that lava flows in the drainage of the stream-gravel deposits of various ages but Bear River, by diverting that large source cannot be expected much farther up the of water to the Bonneville basin, caused a Snake. rise in lake level perhaps 100 feet higher A gravel deposit in the valley of the than the level marked by the later Bonne- Snake a few miles south of Lewiston, built ville shoreline. It was this influx which across the mouth of Tammany Creek reached the summit of Red Rock Pass and Valley, is certainly a great river bar made initiated the debacle down Port Neuf River. under very exceptional conditions. In the However, these students hold that the light of present data, it may be assigned catastrophic erosion in the pass deepened either to a scabland flood's backwash or to a it only to the Bonneville level, where suffi- northbound Bonneville flood emerging from ciently resistant rock was encountered, and the Snake Canyon south of Lewiston. flood discharge down Port Neuf River was The deposit is a broad ridge nearly 100 succeeded by an erosional deepening rate feet higher than the creek's valley floor, 250 consonant with the stabilized outlet river. feet above the Snake and nearly 1,050 feet The flood was over. Thus Malde and the A.T. (Bretz 1929). It slopes gently toward others hold that a lowering of only about the Snake but steeply toward the blocked 100 feet yielded floodwater of about 380 valley. Its long delta-type foresets which dip cubic miles. The time estimated for this out of the Snake Valley determine this rapid deepening and lowering of outlet lake steep slope. level may have been at least 6 weeks and The bar is a plum pudding of unsorted less than 1 year. Malde's Professional Pazer boulders and cobbles in a matrix of fine 596 contains these and other quantitative gravel. From several samplings, the fines estimates made by associated students of average 89 percent angular and sub- the Bonneville field evidence. angular, 9 percent broken-rounds, and 2 All this briefly discharged volume went percent well-rounded particles. The 89 per- westward down the Snake to the Oregon- cent angular fine gravel can hardly be dis-

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 532 J HARLEN BRETZ tinguished from the product of a rock upper limits in the eastern tributary valleys crusher. make a southward-descending series in ac- Percussion scars on the edges of disk- cordance with the Cheney-Palouse spillway shaped but otherwise smoothly worn cobbles gradient (Bretz 1929). The absence of any and boulders record tremendous pounding. such deposits on spillway bars seems to Commonly, the scars are 2-3 inches across. mean that when these bars were growing, One basalt boulder carried a scar 1 inch no fines could come to rest on channel deep and 15 inches in diameter. floors. Basalt is by far the dominant constit- Surges are a necessary part of the picture uent of the deposit, but the previously of these back-floods, because only great smoothed large fragments are in large part turbulence could have maintained sus- former Snake River cobbles picked up pension for the many miles traveled up locally by the flood. Dominance of basalt some valleys. and the altitude of this debris argue for Another necessary part of the picture is derivation in a scabland back-flooding. But slower down-valley draining away that left until further field study, the suspicion re- largely intact the slope mantles and the mains that the Tammany bar may be a valley-floor deposits made at the inter- Bonneville flood product. vening pause in reversing the circulation. Only one Bonneville flood of catastrophic Flushing-in occurred, flushing-out did not. proportions is to be contemplated, and The great thickness of some valley-floor some older gravel locally exposed may also accumulations may have required recurrent be of Missoula flood origin. By Richmond's episodes of floodwater filling and emptying. chronology, Lake Missoula had a later The long distances traveled in these up- flood than the Bonneville debacle. Any valley floods is one of the most incredible Bonneville flood must have gone all the way items of the Missoula flood story. The towns down the Columbia to reach Pacific Ocean of Arrow and Juliatta in the lower Clear- levels. But we must conclude that its water Valley mark the maximum known record is probably limited to Utah and distance this material was carried back up- Idaho. No evidence for it has yet been valley, nearly 100 miles from the Palouse- found in Washington or Oregon. Snake junction. Fresh highway cuts near Arrow afford a series of sections totaling 1 THE BACK-FLOODED GLACIAL SILTS mile with depths of 30 feet maximum in The intimate mixture of tiny basalt rude, terracelike shoulders of the bottom granules in the back-flooded silts on the slopes of this valley. The sections are lower, gentler slopes of all nonglaciated dominantly of silt and fine sand, stratified valleys tributary to the Cheney-Palouse with the familiar undulatory bedding and tract of the Snake canyon above the Palouse variations in thicknesses and in contacts of River junction, of the Walla Walla River members. Multitudes of pockets, seams, basin, and of the lower valley of the Yakima and indistinct strata of coarse basaltic sand River may be ascribed, in part, to soil are well distributed along with foreign creep. Some intimate mixing, however, pebbles and cobbles. exists in flat-lying valley-bottom deposits, Freshly made highway sections in the and there it must be an original structure. bottom of the lower Tucannon river valley, Lenses and pockets, even thin strata, of a few miles up the Snake from the Palouse silt-free basaltic sand may lie in and be- River junction, were described by Bretz et tween usually thicker beds of essentially al. (1956, p. 1034) with the comment that granule-free silt, originally loess. The almost "perhaps two kinds of backwater episodes omnipresent foreign rock pebbles, cobbles, are recorded in these deposits, both with and boulders prove that these materials up-valley currents. The well-sorted, well- were introduced by glacial water. Their stratified valley bottom deposits may be

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 533 the consequence of melt-water incursions Missoula, where growth and wastage of the from normal valley train building along Cordilleran ice sheet was the climatic fac- the main scabland routes. Because, how- tor. ever, floods have all but destroyed any Missoula's numerous shorelines are so record of valley trains, survival of such faint and close set that there is little hope of valley bottom sediments seems anomalous. determining the total number or of correlat- Furthermore, their existence in Snake ing from valley to valley. No long-continued tributaries above the Palouse junction pauses in the changing lake level are on would require a Palouse canyon, an Allison record. Probably the series is the aggregate ice dam, or a Flint fill in order to cross the result of several fillings and emptyings, not divide." a strictly chronological series as Eakin and In the Yakima structural valley, silt Honkala (1952) have conceived. with foreign rock fragments extends for 50 Richmond (1965) has found glacial miles upstream from Chandler Narrows and stream gravel at 4,200 feet A.T. on the in numerous places has a definite topograph- northern nose of the Bitterroot Range, ic and stratigraphic relation to younger fea- against which the Pend Oreille lobe im- tures. Two postsilt terraces flank the present pinged. Presumably this outlet stream floodplain, each with a volcanic ash which flowed between rock on one side and glacial dates each deposit. The glacial silt higher up ice on the other. Probably Missoula dis- the slopes has a maximum thickness of 30 charged its constantly rising levels between feet near the head of the structural valley. rock and ice during whatever time was in- Its stratification shows some graded bedding volved in rising to 4,200 feet A.T. and, had -repeated couples of sand grading up into the Pend Oreille lobe crowded farther south silt. Graded bedding also occurs in a number before the dam broke, Missoula would have of Walla Walla Valley exposures. There are risen to spill across the lowest notch that altogether too many in one section, and they does exist in the Bitterroot crestline. are too thin to assign each couple to a For years after Pardee's 1910 paper, the separate flood influx. Thicknesses of the evidence for Lake Missoula was predicated total deposit seem too great, however, to on the faint shorelines etched out in the assign to one flood. Thus far, no marker for mountainside mantle waste and on certain identification and separation of successive deposits banked up in minor ravine mouths silt- and berg-laden floods, here or elsewhere, at appropriate altitudes. These were called has been recognized. eddy deposits by Pardee. Davis (1921) doubted this and hypothesized that tongues GLACIAL LAKE MISSOULA of glacial ice in such valleys left fragmentary While Lake Bonneville lay for many cen- lateral moraines. turies in a broad basin among mountainous But Pardee established his interpreta- uplifts and never had an outlet until its tion when, in 1942, he reported the spillway rising level topped the divide at Red Rock gorges and giant current ripples already Pass, Lake Missoula occupied free-draining noted (see p. 1569-1600) and the severely river valleys that only a very few times were scrubbed and largely bare rock salients briefly blocked by glacial ice. Bonneville's along that particular flooded valley (Flat- shorelines (Bonneville, Provo, and Stans- head-Clark Fork River). These strongly bury) are strongly expressed features of the eroded surfaces are below the lake's upper landscape; Missoula's are of the faintest limit, while the same slopes above have a character. Bonneville's advantages in this cover of soil and colluvium. Only Alden item were a broad expanse for wave fetch (1953), admitting the probability of cat- and a long-continued balance between pre- astrophic emptying, remained dubious cipitation and evaporation. It never had an about the "unusual currents in Lake experience with glacial ice like that of Missoula."

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 534 J HARLEN BRETZ

Bonneville's lacustrine sediments have at one time high enough to spill briefly had the attention of several investigators from its western side across unglaciated since Gilbert's classic study, but any country well south of the dam and to dis- record in Missoula's bottom deposits is only charge into the head portion of the Cheney- in the early stages of deciphering. Probably Palouse tract. Because this outlet traversed the lake arm occupying the broad Rocky loess of Bull Lake age with a strong soil Mountain trench will yield the most de- profile and because the moraine which re- tailed record. tains the modern lake is of Bull Lake age, Nobles (1952) and Bretz et al. (1956) Richmond (1965) thinks that the berg- have been concerned with the relations of floated erratic boulders reported by several Lake Missoula to the frontal moraine of the observers since Hershey (1912) and the latest advance of the Flathead lobe in the outlet channel in loess record a back- Rocky Mountain trench. At the south end of flushing-a great surge from the bursting Flathead Lake, the Poison moraine has of Lake Missoula's dam in early Pinedale Missoula shorelines inscribed on its strongly time. During Bull Lake glaciation, the lake expressed, subaerially built forms, and had been at a lower level and had discharged therefore must be older. On the west side of from its northern end, the overflow es- the lake, the Elmo moraine, a continuation caping westward between glacial ice on the of the Polson, also has the shorelines, north and mountains on the south (Bretz et though faintly developed. But west of and al. 1956, p. 1040). beyond the Elmo moraine, valley slopes This water entered another glacial lake, carry much more definite scorings of a pre- Lake Spokane, whose area, originally de- moraine stand of Lake Missoula, and out- fined by Large (1922), has been extended wash from this moraine lies in a trench cut considerably north of the city of Spokane by into lake silts. Thus an early lake level here Richmond. Lake Spokane, thus defined, was was drained away before the last ice ad- dammed on the west by the Colville lobe vance occurred, and only after that reached of Bull Lake age. Its overflow was probably its maximum did the lake level come back. southward into the Cheney-Palouse scab- Richmond, the most likely student to land tract. pursue this theme, has already established criteria for differentiating deposits of Bull SOME INTERRELATED QUESTIONS AND PROB- Lake age from those of the later Pinedale LEMS WHICH PRESENTLY LACK ANSWERS age. AND SOLUTIONS Correlations of and by back-flood silts.- GLACIAL LAKES COEUR D'ALENE The number of Missoula outbursts is still AND SPOKANE tentative, and correlations among widely The Pend Oreille lobe which dammed the separated scabland spillways remain un- Clark Fork River valley approximately settled. A promising approach appears to at the north end of Lake Pend Oreille also lie in the flood deposits on the bottoms of pushed 40 miles farther southward along back-flushed valleys. There may have been the western side of the Bitterroot Range in two kinds of experiences involved: (1) the the now-streamless valley called Rathdrum flooding-back of Missoula discharges and Prairie. Here it blocked another northwest- (2) backwater from normal glacial drainage draining river system and made another between floods. If such occurred, differentia- glacial lake, the predecessor of modern tion is yet to be found, and records of the Lake Coeur d'Alene. Its water backed up second kind of event can be expected only for miles to the southeast, drowning pre- in the lower reaches of the tributaries in- glacial river valleys which had drained the volved. Transportation of the flood silts is a western slope of the Bitterroots and the problem for which the writer has only vague Coeur d'Alene mining region. Water rose solutions. A hydraulic engineer conversant

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 535 in the field could well improve on them. Ringold-type sediments have ever been WalallaWlla basin sediments.-The basin reported beyond Wallula Gap. Is this ex- drained by the Walla Walla River is a dis- tensive deposit, believed by some to be of trict of backwater sedimentation from a Pleistocene age, due simply to a damming greatly swollen Columbia River. Except for of the Columbia by the rising Horse Heaven the Ringold formation, no tract in the en- Hills anticline as late as the Pleistocene? tire plateau carries a thicker deposit of fine- Origin of Wallula Gap.-If, following textured, well-sorted, and stratified quiet- Newcomb (1958), we date the uplift of the water sediments. Some of it is surely of Horse Heaven Hills anticline at Wallula glacial origin. Is it all from glacial water? Gap as mid-Pleistocene, we can have no Are the graded beds here and in the Yakima considerable gap during the life of his Lake Valley annual records? What chronology Ringold. How much of a gap was there when can be found in the exposures? Are the the first Missoula flood arrived? How much Glacier Peak and Mazama ash beds present? of the Gap's present depth is the erosional What relation is there between the Ringold consequence of all floods? One begins to formation, apparently from nonglacial wa- wonder if that highest scabland just west ter, and the Walla Walla deposits? of the Gap might be the oldest scabland now If successive floods are on record in this surviving. It certainly is not as old as the basin's sediments, alternating with records Old Maid record (see Prescabland Floods of interflood intervals, a sequence will be below). established when calichified or weathered An answer lies in the gravel spill from the horizons, berg-transported erratic material, highest Wallula scabland channels down up-basin erosion, transportation or deposi- more than 400 feet in a horizontal distance tion, volcanic ash layers, peat deposits, etc., of less than 2,000 feet. It could hardly lie are found. more steeply. It reaches essentially to the Lake Lewis and Touchet silts.--Dating bottom of the gorge, just downstream from from Symons's report (1882) is the concept the narrowest part. The gorge downstream employed by several later investigators from the gap itself therefore was as deep as that the finer sediments in the Pasco down- it is today when the first flood (pre-Wiscon- warped region record a glacial "Lake sin according to Richmond) arrived. Lewis." Both Allison and Flint imagined a The mid-Pleistocene uplift at Wallula dam in the Columbia Gorge across the and the draining of Lake Ringold (cutting Cascade Range as the cause of such a lake, of the Gap) could hardly have been many and Flint gave a formational name- millennia apart; yet by our argument, time Touchet silts-to such deposits. sufficient for the transection of the anti- The writer's concept of hydraulic dam- cline to the present depth must have ming at Wallula Gap provides a ponding of elapsed. the Pasco basin for each Missoula flood, The nearly flat top of the Ringold de- sees the fine sediments as deposited ir- posit is undoubtedly original with com- regularly in both time and space, and ques- pletion of this fill in the Pasco basin and is tions the Lake Lewis concept. scarcely higher than the cliff summits along The name Touchet silts is still used Wallula Gap. It seems very probable that (Richmond et al. 1965) for rhythmically the 500+ feet of this lake sediment ac- bedded silts in the lower Snake Canyon. cumulated during the uplifting of the The implication that such deposits are Horse Heaven Hills anticline and that only correlative with what are here termed back- an initial gap was eroded during the life of water silts should be investigated as should the lake. also their relations with the Walla Walla Considering the depth of the Umatilla Valley deposits. structural basin immediately downstream Lake Ringold and Wallula Gap.-No from the Gap and the short gorge of the

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 536 J HARLEN BRETZ

Gap, it may be postulated that tran- floodwaters from the Cheney-Palouse and section of the anticline was accomplished by Grand Coulee systems can hardly be ex- headward stoping of a marked series of pected to have produced in it the crested rapids, falls, and cascades. If so, the drain- flood front deduced from Chandler Narrows ing of Lake Ringold was very rapid, be- and The Narrows of Oregon while Wallula cause most of the gorge length was already Gap was open. Yet the Chandler scabland excavated by the time recession of such demands a very vigorous back-flooding up high-gradient cascades finally reached com- the Yakima structural valley. It seems pletely across the tectonic barrier. logical, therefore, to assume that enough Predeformation course of the Columbia large berg ice arrived at Wallula to make a River.-Waterworn quartzite pebbles are temporary blocking (an Allison ice jam) strewn over a northeast-by-southwest area and thus the head necessary at Chandler of the Horse Heaven Hills uplift south of the and at the less spectacular back-rush into Yakima Valley, also on gentler slopes of the the Walla Walla depression which, like western parts of the Saddle Mountain and Chandler, is at the far end of the basin. Frenchman Hills anticlines. They therefore But no field evidence for this is yet known. are older than the folding. They also occur Upper limits in Yakima Valley and on as strata in downfolded sedimentary rock Wallula summit.-Disparity on the upper in the Yakima Valley, generally considered limits reached by floods back up the Yakima to be the Miocene Ellensburg formation. Valley (1100 feet A.T.) and at Wallula Gap They are clearly from a far-traveled river 1,200 feet A.T.) may be due to inadequate gravel and appear to record a Columbia field study in the structural valley. But the course from the Beltian terrains of Mon- upper limit at Chandler Narrows is ex- tana, British Columbia, or Idaho. Are they plicable by the much greater cross section the record of a Miocene river? Do they of the valley above 960 feet A.T. record the Columbia's course just before Recalculation of Wallula's discharge rate. the foldings of the anticlines, that is, just -With remapping of the northwest corner preceding the shift of that river to the site of of the Pendleton quadrangle on a larger Wallula Gap? The undeformed Ringold oc- scale will come an opportunity to recompute cupies the broad Pasco structural basin, the discharge rate at Wallula Gap. Hy- which almost certainly dates from the time draulic damming also seems probable at the of the Horse Heaven Hills uplift. west end of the Umatilla depression, and Columbia Valley between Okanogan Valley this suggestion can be verified or disproved and Moses Coulee.-This stretch of 85 miles when the old Arlington quadrangle map is traversed by at least two Missoula floods topographically resurveyed. has never been critically examined. In the Berg deposits on Rattlesnake Mountain.- last few miles, it receives Wenatchee River, The writer noted (1930b) that the berg below which junction is an area that the mounds on the northeast slope of Rattle- writer has described as glaciated (Bretz snake Hills (Iowa Flat) are abundant 1930b). But Page (1939) has found no below an altitude of 850 feet but that only evidence in the tributary valley that glacial scattered berg-carried boulders lie between ice ever reached its mouth (see Bretz et al. that altitude and the upper limit of glacial 1956, p. 990). Much landslide topography silt, 1,100 feet A.T., on this slope. Bretz et exists here, and the suspicion is justified al. (1956) suggested that perhaps two differ- that the writer has misinterpreted the ent floods are on record here. If the berg tract. The entire stretch of 85 miles should abundance and berg load differed as much yield new data on the late Pleistocene his- as indicated, this criterion should find sup- tory of the Columbia Valley. port in further field studies in the district. Chandler Narrows flood front.-Because The larger problem is the lack of a com- of the large capacity of the Pasco basin, parable berg population elsewhere in the

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 537 semiponded tracts of the scabland or the louse Canyon. The deposit cannot be older Columbia Valley. than those bars and have survived the huge Tammany bar.-Solution of the Tam- river they record. Was it made in that great many bar problem will depend in part on stream or is it younger? a detailed field study of various gravel de- Bretz et al. (1956, p. 1005) describe a posits in the Lewiston-Clarkston basin. probable distributary discharge from Othel- The oldest gravel should date back to early lo Channels eastward up Washtucna Coulee downwarping and accompanying incision but say nothing of its farther course. The of the Snake River canyon thence down- altitudes involved permit the idea that, stream. if a Devils Canyon was already well in- Glaciation at channel heads.-The writer cised and the distributary had adequate believed that till deposits and subdued volume, it could have followed the canyon morainelike forms on that part of the to the Snake and made the deposit. This Columbia Plateau between the Spokane and would date it as after Grand Coulee had Columbia rivers on the north and the heads captured all plateau floodwater, when of Cheney-Palouse and Telford -Crab Creek Othello was operating with a maximum channels on the south recorded glaciation flood and Cheney-Palouse had ceased to almost up to the channel heads. Flint also flow. This interpretation is based almost saw this area as glaciated. Richmond has entirely on maps and aerial photographs. dismissed this view and sees the tract as A field study would verify or disprove it. flood-swept. But the total lack of isolated, Sentinel Gap.-A deep hole in the rock residual hills of loess, so prominent a fea- bottom of Sentinel Gap where the Columbia ture of flood erosion elsewhere on the pla- River crosses the Saddle Mountain anti- teau, seems to argue for glacial, rather than cline seems to indicate that kolk action has flood, removal. Perhaps both events oc- occurred in this constricted place. Was hy- curred, glaciation before flooding. draulic damming here a part of the cause? Correlations of western cataracts.-Suc- If backup of a Columbia flood did occur cessive events in the life of the three western here, did it initiate Othello Channels, flood cataracts are not yet clear. Certainly at Evergreen-Babcock bench, or discharge some time all fell from brink to plunge pool eastward through the cataract notches in bottom. But such time could not coincide the Evergreen-Babcock ridge? with the high level of floodwater in the Prescabland floods.-Alden (1953) details Columbia Valley. When the Evergreen- scattered evidences for possible Illinoian Babcock bench, the high Wallula scabland, glacial ice in Clark Fork drainage south of or the Beverly bar were under floodwater, the definitely Wisconsin moraines. The none of the three cataracts could have func- question of a pre-Wisconsin Lake Missoula tioned to the full depth of their recessional immediately rises. Richmond finds three gorges. glaciations recorded in Flathead Valley Sequential operation of Palouse and Devils and indicates the first of his three Missoula canyons.-The mouth of Devils Canyon is floods as immediately pre-Wisconsin. blocked by a gravel deposit nearly 300 feet Two features of the southwestern part of thick. On the opposite side of the Snake the plateau strongly suggest floods earlier River, this deposit extends about a mile than those in Rocky, Bowers-Weber, and downstream. Both remnants carry well- Lind coulees. One is Old Maid's Coulee, developed giant ripple marks oriented for shown on the Connell quadrangle topo- discharge from Devils Canyon (Bretz et al. graphic map. It contains remnants of a boul- 1956, pi. 9, fig. 3). The mounded surfaces are dery gravel, mostly decayed basalt with 200 feet or more higher than any terrace or pebble fragments of caliche, all under a bar down the Snake or upstream as far as marked caliche cap. The southwest-draining the great gravel bars at the mouth of Pa- coulee is described (Bretz et al. 1956) as

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 538 J HARLEN BRETZ having no scabland and no notch at the head of the Portland region. But if multiple for entrance of floodwater from Washtucna. floods (including a Bonneville flood) came Richmond suspects that it is pre-Bull Lake down the Columbia, some record of inter- in age. vals between them should be expected, The other feature is elongated Eureka at least in the silt and sand beyond the Flat, most of it on the Walla Walla quad- "Portland Delta" gravel. Trimble's investi- rangle, where it is 2 miles wide. It is gation was detailed. Exposures made since parallel to the nearby Snake for its full may, however, reveal such records. They length, 12 miles or so, and is bounded by are worth looking for. strikingly narrow, linear loessial hills in Rocky Butte.-In the eastern part of the alignment with the flat. Drainage enters Portland Delta gravel deposit, Rocky Butte it from these hills but, beyond a few miles, (fig. 10) rises to 607 ft. A.T., while the gravel leaves it to reenter the loessial hills. No flat east of it is 300 ft. A.T. and a long train bedrock and no gravel are known, but its of gravel depending westward has nearly form, proportions, and parallelism with the the same altitude. The eastern gravel flat Snake and the loessial hills surely are not fails to reach the cliffed face of the butte. simply fortuitous. Instead, there is a "hole" in the delta sur- Possibly both valleys functioned as very face, a hole in gravel on the east but with a early precaliche flood channels. If the Old craggy rock wall on the west. This fosse is a Maid gravel records an older caliche broken basin more than a mile long (north-south) up in an early flood with its fragments in- and less than a quarter of a mile wide. corporated in a later flood gravel which Along the face of the butte, it is a closed lies under a later caliche, the suspicion depression 10-25 feet deep whose bottom is arises that some caliche horizons in the approximately 100 feet lower than the delta loess of the Palouse region, thought to be surface just to the east, that is, upstream. correlative with interstades (nonflood epi- The flood which made the gravel flat had sodes), may indeed be correlative with very a huge downward-directededdy when it en- early (pre-Wisconsin) Missoula floods. Neff's countered the butte, an eddy that either earliest known flood, from the Columbia maintained a bottom hole here or eroded it. Valley back into Quincy basin, disrupted a In the scablands, this would be a rock basin, caliche that may be older than the one here it is partly in gravel. herein inferred to be immediately pre- The north end of the hole opens to the scabland. Neff has also found, on the undis- Columbia's immediate valley slopes. The sected summit plane of the Ringold forma- south end becomes two curyilinear troughs tion, a gravel containing caliche fragments. which lead westward alongside the gravel Where does this belong chronologically? flat and gradually shallow and taper out. Bacon basin.-Age and origin of the All this bespeaks, in Trimble's phrase, the gravel beneath a caliche-capped loess in the "almost incredible" discharge from the Bacon basin are uncertain. It may be a pre- Cascade Gorge into the broad Willamette- flood deposit of local origin, like the old Columbia depression, as proposed by the gravel noted by Bryan in the Hartline writer in 1925. But the unanswered ques- basin. But if found to contain erratic mation of today is the proper place for this terial and large, sporadically distributed particular flood in the presently known basalt boulders, it must be assigned to the series of floods. earliest Missoula flood ever to cross the Richmond indicates his Pinedale flood as Trail Lake flexure and thus to early stages the greatest of the three he identifies. The of cataract retreat in Upper Grand Coulee. Rocky Butte fosse could well be a part of its Repeated floods at Portland.-Trimble record. However, the flood that Neff has (1963) did not report any field evidence for identified as discharging from Columbia successive floods in his "lacustrine deposits" Valley through Crater Cataract notch into

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions feet 20

interval,

contour

Survey;

Geological

U.S.

map,

quadrangle

topographic

Oregon,

Portland,

from

Butte,

10.-Rocky

FIG.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 540 J HARLEN BRETZ

Quincy basin and the probably contempora- Richmond as late Pinedale in age. Can this neous eastward discharge through the Pot- Nespelem silt formation be dated by organic holes notch (Bretz et al. 1956) both de- remains (C14)? Neff reports that it is older mand that Drumheller and Lower Crab than the volcanic ash layers. If no cor- Creek channels then function, even without relations can be made between any part of any Grand Coulee contribution. Was this the Walla Walla sediments and the Ringold the Pinedale flood with no Okanogan dam formation, the latter will prove to be the then existing? older. But both the Walla Walla and the Then one recalls Beverly bar, built after lower Yakima structural basins logically the last operation of the Lower Crab Creek must be considered as contemporaneous channel. It would have to be post-Pinedale with the Pasco basin, and altitudes of these and would also be a post-Okanogan affair, two basin bottoms are well below the and therefore a smaller flood. Ringold's flattish summit levels. Why then It is fair to assume that the Rocky Butte should not at least the lower portions of the fosse is a maximum flood record, along with Walla Walla beds prove to date from Lake the ice-floated erratic boulders known up Ringold? A more challenging situation the Willamette Valley to altitudes of 450 exists in the Yakima Valley, where no ft. A.T. How was the fosse modified by Ringold or Walla Walla sediments have yet floods? If they were only 125 feet deep in the been reported. The Yakima's later meander- Columbia thalweg, they did not reach the ing can hardly be held responsible for clean- fosse through its open north end. Such an in- ing out that wide structure of any such sedi- terpretation would perhaps take care of the ments. Beverly bar flood, especially if it were not Sediments of Lake Missoula.-As far as is adequate for Trimble's hydraulic damming. now known, the glacial lake's sediments are But the Crater and Potholes reverse dis- varved and contain randomly distributed, charge into the Quincy basin seems to de- ice-floated rock fragments. mand a major flood and, hence, a record of Future studies will begin with sections back-flooding in the Rocky Butte fosse in postglacial stream trenches in floors of if it then existed. Dating the fosse con- various lake-occupied valleys. The shallower stitutes a neat little problem for further bays will have had complete emptying with study. each dam burst and may record flood cur- Flood sources other than Missoula.-There rents. All parts of the lake must have been is tacit acceptance throughout this paper emptied whenever retreat of the Cordilleran that collapse of the Missoula dam pro- ice sheet reestablished the preglacial Clark vided the floods on record. However, an Fork and Columbia River drainage sys- Okanogan dam burst could release a huge tem and thus provided for nonflood erosion quantity of water without any Missoula on the lake bottom. If, however, an Okano- although minor in flood, comparison. There gan lobe still blocks the Columbia, a are at present unexplored possibilities of shrunken remnant of Lake Missoula will other glacially dammed lakes in the north- linger in the deeper valleys near the site ern Cascades and the Okanogan Valley of the dam, and if Upper Grand Coulee's from which sudden release could leave some cataract has not yet breached the Columbia of the flood records known in the Columbia Valley's southern wall, a Columbia Valley from the Big Bend all the way to Lake Portland. will extend eastward, possibly past the Relations of the Nespelem formation to Colville lobe to join the shrunken Missoula. Walla Walla sediments.-Note has been Reconstruction of the Pend Oreille dam made of Lake Columbia's white silts on the will produce another Lake Missoula and Upper Grand Coulee floor, considered by another varved clay deposit. Unconform-

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 541 ities will separate every two such sedimen- sidered to be of glacial age buried by post- tary units. glacial clay. During an episode of glacial retreat and Q.E.D.? lake drainage, bogs and forests will invade the exposed floors. Return of lacustrine The International Association for Quater- conditions will bury peat beds and tree nary Research (INQUA) held its 1965 meet- stumps along the unconformities. Leaching ing in the United States. Among the many and oxidation will be added to the record of field excursions it organized was one in the erosion and plant invasion. northern Rockies and the Columbia Pla- Carbon-14 dating of woody material will teau in Washington. The party of thirty-two be the only reliable method of correlation participants representing nineteen coun- among areally separated records of drainage tries, led by Richmond, saw the Polson intervals. Exceptionally complete local rec- moraine and Lake Missoula shorelines in ords may correlate with the glacial record the Rocky Mountain trench, Pardee's in the northern Rocky Mountains and the gigantic current ripples and the strongly loessial sequence on the Columbia Plateau, scoured valley shoulders and eddy deposits but it will fall short of the scabland record in the Clark Fork Valley in Montana, and as outlined in this paper. the site of the Pend Oreille glacial dam for Astoria deep-sea fan.-Two different Lake Missoula in Idaho, and spent a day on causes for the ponding recorded by the the Columbia Plateau in Washington. The Portland Delta are hydraulic damming party crossed the head portions of the (Trimble 1963) and a higher sea level Cheney-Palouse and Telford-Crab Creek (Lowry and Baldwin 1952). Evidence bear- tracts, traversed the full length of the Grand ing on this conflict of interpretations can Coulee, part of the Quincy basin and much probably be found when the Columbia of the Palouse-Snake scabland divide, and Valley beyond Trimble's dam site is studied the great flood gravel deposits in the Snake in detail. Canyon. The writer, unable to attend, re- more data presently With than an- ceived the next day a telegram of "greetings (Nelson and Byrne 1968), the nounced and salutations" which closed with the deep-sea fan at the mouth of the Columbia sentence, "We are now all catastrophists." River may conceivably yield information on the arrival of various catastrophic floods. ACKNOWLEDGMENTS.-RoaldFryxell, Harold The summit portion of possesses the fan Malde, George Neff, Gerald Richmond, and "poorly sorted gravel, sand and silt mem- Donald Trimble have been consulted on various bers" in "coarse layers of irregular thick- items in this paper and have contributed sugges- ness with sharp upper contacts and poorly tions and criticisms. They have the writer's developed clay interbeds" which are con- lively appreciation.

REFERENCES CITED ALDEN,W. C., 1953, Physiography and glacial geol- BLACKWELDER,E., 1915, Post-Cretaceous history ogy of western Montanta and adjacent areas: of the mountains in central western Wyoming: U.S. Geol. Survey Prof. Paper 231, 193 p. Jour. Geology, v. 23, p. 97-117, 193-217, 307- ALLISON,I. S., 1932, Spokane flood south of Port- 340. land, Oregon [Abs.]: Geol. Soc. America Bull., BRETZ,J H., 1919, The Pleistocene submergence in v. 43, p. 133-134. the Columbia Valley of Oregon and Washington: Jour. 27, p. 1933, New version of the Spokane flood: Geology, v. 489-506. --- 1923a, Glacial drainage on the Columbia Geol. Soc. America Bull., v. 44, p. 675-722. Plateau: Geol. Soc. America Bull., v. 34, p. 573- --- 1935, Glacial erratics in Willamette Valley: 608. Geol. Soc. America Bull., v. 46, p. 615-29. --- 1923b, The Channeled Scablands of the

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions 542 J HARLEN BRETZ

Columbia Plateau: Jour. Geology, v. 31, p. 617- --- 1890, Lake Bonneville: U.S. Geol. Survey 649. Mon. 1, 438 p. BRETZ, J H., 1924, The Dalles type of river channel: HANSEN,H. P., 1947, Chronology of post-glacial Jour. Geology, v. 32, p. 139-149. volcanic activity in Oregon and Washington --- 1925, The Spokane flood beyond the Chan- [Abs.]: Geol. Soc. America Bull., v. 58, pt. 2, p. neled Scablands: Jour. Geology, v. 33, p. 97-115 1252. and 236-259. HERSHEY,O. H., 1912, Some Tertiary and Quater- --- 1928a, Bars of Channeled Scabland: Geol. nary geology of western Montana, northern Soc. America Bull., v. 39, p. 643-701. Idaho and eastern Washington: Geol. Soc. --- 1928b, The Channeled Scablands of eastern America Bull., v. 23, p. 517-536. Washington: Geog. Rev., v. 18, p. 446-477. LARGE,T., 1922, The glaciation of the Cordilleran --- 1928c, Alternative hypotheses for Chan- region: Science, v. 56, p. 335-336. neled Scabland: Jour. Geology, v. 36, p. 193-223 LOWRY,W. D., and BALDWIN,E. M., 1952, Late and 312-341. Cenozoic geology of the lower Columbia River Oregon and --- 1929, Valley deposits immediately east of valley, Washington: Geol. Soc. America Bull., v. 63, p. 1-24. the Channeled Scabland, Washington: Jour. MALDE, E., 1960, Snake River Geology, v. 37, p. 393-427 and 505-541. H. Evidence in the plain, Idaho, of a catastrophic flood from Pleisto- --- 1930a, Lake Missoula and the Spokane flood cene Lake Bonneville: U.S. Geol. Survey Prof. [Abs.]: Geol. Soc. America Bull., v. 41, p. 92-93. Paper 400-B, p. 295-297. --- 1930b, Valley deposits immediately west of --- 1965, Snake River Plain, in WRIGHT,H. E., the Channeled Scabland: Jour. Geology, v. 38, and FREY, D. G., eds., The Quaternary of the p. 385-422. United States: Princeton, N.J., Princeton Univ. --- 1932, The Grand Coulee: Am. Geog. Soc. Press, p. 255-263. Spec. Pub. 15, 89 p. -- 1968, The catastrophic late Pleistocene --- 1959, Washington's Channeled Scabland: Bonneville flood, Snake River Plain, Idaho: U.S. State of Washington, Division of Mines and Geol. Survey Prof. Paper 596, 52 p. Geology, Bull. 45. MATTHES,G., 1948, Macroturbulence in natural ---; Smith, H. T. U.; and Neff, G. E., 1956, stream flow: Am. Geophys. Union Trans., v. 28, Channeled Scabland of Washington: new data p. 255-262. and interpretations: Geol. Soc. America Bull., v. NELSON,C. H., and BYRNE,J. V., 1968, Astoria 67, p. 967-1049. Fan: a model for deep-sea fan deposition [Abs.]: CONDON,T., 1871, The Willamette Sound: Overland Geol. Soc. America Ann. Mtg., p. 218. p. 468-473. NEWCOMB,R. C., 1958, Ringold formation of DAVIS, W. M., 1921, Features of glacial origin in Pleistocene age in type locality: Am Jour. Sci., Montana and Idaho: Assoc. Am. Geographers v. 256, p. 328-340. Annals, v. 10, p. 75-147. NOBLES,L. H., 1952, Glacial sequence in the Mis- sion Montana [Abs.]: Geol. DILLER, J. S., 1896, A geological reconnaissance in Valley, western Soc. 1286. in northwestern Oregon: U.S. Geol. Survey Ann. America Bull., v. 63, pt. 2m, p. M., Alpine glaciation the Rept. 17, pt. 1, p. 441-520. PAGE,B. 1939, Multiple in Leavenworth area, Washington: Jour. Geology, EAKIN,G. R., and HONKALA,F. S., 1952, Cenozoic v. 47, p. 785-815. history of Missoula Valley, Missoula Co., Mon- PARDEE, J. T., 1910, The glacial Lake Missoula: tana [Abs.]: Geol. Soc. America Bull., v. 63, Jour. Geology, v. 18, p. 376-386. p. 1361. pt. 2, --- 1918, Geology and mineral deposits of the FLINT, R. F., 1938, Origin of the Cheney-Palouse Colville Indian Reservation, Washington: U.S. Scabland tract, Washington: Geol. Soc. America, Geol. Survey Bull. 677, p. 28. Bull., v. 50, p. 661-680. --- 1942, Unusual currents in Glacial Lake FRYXELL,R., 1962, A radioactive limiting data for Missoula: Geol. Soc. America Bull., v. 53, p. scabland flooding: Northwest Sci., v. 36, p. 1569-1600. 113-119. POWERS,H. A., and WILCOX,R. E., 1964, Volcanic --- 1965, Mazama and Glacier Peak volcanic ash from Mt. Mazama (Crater Lake) and from ash layers: relative ages: Science, v. 147, p. 1288- Glacier Peak: Science, v. 144, no. 3624, p. 1334- 1290. 1336. --- and DAUGHERTY,R. D., 1962, Archeological RICHMOND,G. M., 1965, Guidebook for field con- salvage in Lower Monumental Reservoir, Wash- ference E, northern and middle Rocky Moun ington: Washington State Univ., Lab. Anthro- tains, Part E: Nebraska Academy of Sciences. pology, Rept. Inv. 21, 43 p. --; FRYXELL, R.; NEFF, G. E.; and WEIS, GILBERT,G. K., 1878, The ancient outlet of Great P. L., 1965, in WRIGHT,H. E., and FREY,D. G., Salt Lake: Am. Jour. Sci., 3d ser., v. 15, p. 256- eds., The Quaternary of the United States: 259. Princeton, N.J., Princeton Univ. Press, p. 240.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions LAKE MISSOULA FLOODS AND CHANNELED SCABLAND 543

SYMONS,T. W., 1882, Report of an examination of WASHBURNE, C. W., 1915, Reconnaissance of the the upper Columbia River: U.S. 47th Congress, geology and oil prospects of northwestern Ore- 1st sess., Senate Doc. no. 186. gon: U.S. Geol. Survey Bull. 590, p. 81. TRIMBLE,D. E., 1950, Joint controlled channeling WATERS,A. C., 1933, Terraces and coulees along in the Columbia River basalt: Northwest Sci., Columbia River near Lake Chelan, Washington: v. 24, p. 84-88. Geol. Soc. America Bull., v. 44, p. 763-820. --- 1963, Geology of Portland, Oregon and WRIGHT,H. E., and FREY, D. G., eds., 1965, The adjacent areas: U.S. Geol. Survey Bull. 1119, Quaternary of The United States: Princeton, 119 p. N.J., Princeton Univ. Press.

This content downloaded from 128.193.164.203 on Thu, 15 Jan 2015 10:22:29 AM All use subject to JSTOR Terms and Conditions