were
requestedby
the NationalPark
Service to lend assistance.The
State of
South
Dakota, through theW. H. Over Museum and
theSouth Dakota
ArchaeologicalCommission,
put field parties into this area in both 1950and
1951,and
the University ofKansas and
theNebraska
State Historical Society agreedtointerrupt pursuit of their research interests in theirown
States to contribute tothe effort.The Nebraska
institution, having undertaken the salvaging of sites in the criticalTrenton
Reservoir area in that State in 1950,was
unable to beginwork
in Fort Randalluntil 1951, but the University ofKansas
sentexcavation parties intothe area both years.The
operations of theSouth Dakota
ArchaeologicalCommission-
W. H. Over Museum
project during both seasonswere under
the supervision ofWesley
R. Hurt, Jr.,who
has published reports cover- ing the complete investigations (Hurt, 1951, 1952). In 1950, agroup
of amaximum
of 16workers
excavated in theSwanson
site(39BR16),
acompact
villageon
alow
terraceborderingthe Missouri River bottoms approximately 6miles above Chamberlain.When
dis-covered, the site
was
apparenton
the surface as anumber
of large, conspicuous,more
or less circular depressionson
a small pointwhich was
isolatedfrom
thebody
of the terraceby
a shallow linear depres- sion. Excavation revealed that a ditch approximately 3 feet deep lay beneaththe latter featureand
that the other depressionswere
under- lainby
deep house floors.No
evidence that a stockadewas
part of the defensive systemwas found
in the intensive tests.Four
houseswere
completelyuncoveredand
six miscellaneous trencheswere
exca- vated.The
latter revealed that therewere
in the sitean unknown number
of houseswhose
locationswere
not evidenton
the surface.Although
therewas
evidence, in the intrusion of houses into cache pitsand
vice versa, that the villagehad
existed forsome
time, the relativehomogeneity
of the cultural materialsand
the lack of evidence of superposition of dwelling structures suggests that the length of occupationwas
not of great magnitude.The
typical housewas an
oblong rectangular structure,4
to 5 feet deep, with posts rather closely spaced along the side wallsand
rarely along the ends.A
singlerow
of large postson
the long axis or a doublerow
straddling the midline constituted the other verticalmembers
of the superstructure.The
entrancewas
characterizedby
avestibuleand
aramp
leadingfrom
it into the interior of the house.With
one exception,where
therewas
a single central fireplace, therewere two
hearths,bothon
the midline but situatedbetween
the centerand
the end walls. All in all, the structures here resemble rather1
NO. 2
SALVAGE PROGRAM,
I95O-I95I— COOPER
85 closely those of the secondcomponent
attheDodd
site, near Pierre,Specimens found
in relativeabundance
in the fill of houses, in the cache pits,and
on the generalvillage level indicateheavy
reliance for subsistenceon
agricultureand
huntingand
rather little emphasison
fishing or the collection of shellfish.
The
artifact inventory, notably the pottery, together with the architecture, suggests a close relation- ship to villageson
theJames and Big Sioux
Rivers to the east—
among them
theMitchelland Brandon
sites, previouslyexcavatedand
reportedby
theW. H. Over Museum — and Hurt
has assigned theSwanson
site, together with these others, totheOver
focus.In 1951, with a
maximum
party of 13 individuals.Hurt
reinvesti- gatedtwo
sites,39GR1
(ScalpCreek
site)and 39GR2
(EUisCreek
site),
which had
been partially investigated in 1941by
theW. H.
Over Museum,
withWPA
assistance, butwhich
required additional field study tomake
laboratory analysisand
reporting possible.The
ScalpCreek
site appeared superficially to be simply a small fortified earth-lodgevillage, withperhaps 15 houses,on
the point of a terrace cutoffby
theremainsofaditch,but previous excavationhad
revealed thatmaterials relatingto at leasttwo
culturalcomplexeswere
present.The work
during thetwo
seasons, including the complete uncovering of nine earth lodgesand
the excavation of anumber
of trenches, demonstrated that a village of earth lodges is underlainby
materials ofWoodland
affiliations.The upper
village,surroundedby
a stockade ofpostsspaced i to3 feetapartand
defended with aditch 3 feetdeepon
the side not protectedby
a slope, contains circular houses with central fire basin, four center posts, varyingnumbers
of outer roof supports, arow
of leaners,and
a covered entrance passage. Unlike the situation inmany
earth-lodge villages of the region, cache pitswere
rather scarceand were
usually small.A
subsistence pattern basedon
agricultureand
hunting, especially of bison, is indicatedby
the specimens recovered.The
pottery is simple-stampedand
deco- rated primarilyby
incising, bothon
the rimand
body,and
appearsto resemble that previously recoveredand
reportedfrom
theLa Roche
site,
some
distanceup
the Missouri River (Meleen, 1948).Although
thereseem
to besome
differences, at least in proportions of various pottery typesand
perhaps in architecturaland
other traits.Hurt
has assigned the upper(Wheeler) component
of the ScalpCreek
site to theLa Roche
focus.In the lowest cultural deposits at the Scalp
Creek
siteand
at the EllisCreek
site, situatedon
a terracesome
2 miles downriver, arti- factswere
recoveredwhich
included pottery very similar to the cer- amics at siteson Loseke Creek and
EagleCreek
in Nebraska, the86 SMITHSONIAN MISCELLANEOUS COLLECTIONS
VOL, I26former
ofwhich
have been describedby
Kivett (1952). This pot- tery is characterizedby
decoration of the rim area, predominantly withsingle-cord-impressedlines or nodes.Hurt
hasalsodistinguished anothercomponent
in the upper part of theWoodland
deposits at the ScalpCreek
site in w4iichthere arecord-marked
vessels withflar- ing, undecorated rims (or with only the lip decorated)and
vesselswhose
outerrimsaredecorated withhorizontal incisedortrailedlines, usually below aband
of short impressed lines. Shallow fireplaces occurred throughout theWoodland
level, but evidences of structureswere
absent except fortwo
patternless clusters of small postmolds and
areaswhich
suggested floors.Rather
small shallow pitswere numerous, and
large bell-shaped cache pitswere
entirely absent.No
evidence of cornor other cultivated plants
was
found, although afew
scapula hoeswere
present.The
suspicion that corn cultivationwas
practiced
seems
warranted, inview
of the probable relative recency of thecomplex
and, especially, inview
of Kivett's report of corn in one of the relatedLoseke Creek
sites (Kivett, 1952, p. 57).Site
39BF3
(the TalkingCrow
site),an
earth-lodge villageon
alow
terrace in theupper
part of the Fort Randall Reservoirarea, has intrigued archeologists forsome
time, primarilybecauseof the variety of the ceramic remains collectedfrom
its surface.The
University of Kansas, through Carlyle S. Smith,assumed
the responsibility for the intensive investigation of this site,which proved upon
excavation tobe ascomplex
aswas
suspected.A
partyof a dozen to 15workers
spent approximatelytwo months
each of theyears 1950and
1951 in the TalkingCrow
siteand
an additional season'swork was
planned for 1952.By
the end of the 1951 field season, seven houses, not all ofwhich were marked by
surface depressions; a refusemound
;and many
cache pitsand
other featureshad
been excavatedand
test trenchinghad
uncovered parts of such defensive features as the bastioned ditch, stockade,and
amound
thatmay
have served in place of the stockade at the eastern edge of the site.A few
burialshad
been encountered in various parts of the village.Houses were
basically circular,with centralfireplace, four center posts, outer posts generally closely spaced
around
the perimeter,and
vestibule entrance.A number
of featureswere found
in stratigraphic relationship,and
the contents of the refusemound,
excavatedin6-inchlevels, appeared toshow
ceramic changesfrom bottom
to top.Two
pottery wares, not previously isolated, have been describedand named by
Smith.They
are the TalkingCrow and Campbell
Creek
wares.The former
is characterizedby
simple-stamped globu- larbodies, frequentlysomewhat
flattenedshoulders,and
straight highNO. 2
SALVAGE PROGRAM,
I95O-I95I— COOPER
87 rimson which
the lipsare often decoratedby
impressing or notching.Shoulders are frequently incised.
Campbell Creek ware
embraces globular vesselswithrounded
shouldersand
simple flaringorcollared rims. Surfaces are predominantlycord-marked and
decoration, con- fined mainly to the rim, consists of notching or pinching of the lipon
flaring rimsand
incisingon
collared rims. TalkingCrow ware
resembles inmany
respects the ceramics of theLower Loup complex
in Nebraska, while the
Campbell Creek
pottery suggests rather close affinities to that of theNebraska and Upper
Republican aspects.These two
categories include thebulkof the potteryfound
inthe site,butotherkinds occurinsmallernumbers.
These
includeafew
sherds of Stanley ware, predominant in sites fartherup
the river,and some- what more abundant
fragments of vessels with horizontally incised rimsand
shoulderswhich
are usually decoratedby
incisingand
punctating.At
the end of the 1951work Smith
felt that, in addition to a late nineteenth-centuryDakota
occupation, three periods—
definedby
pot-tery, since other artifacts
seemed
to bemuch
thesame
throughoutthe occupation—
could bedistinguishedinaculturalcontinuum.Campbell Creek ware
predominated in the earliest period, TalkingCrow ware
in the latest, with pottery of the
two
represented in approximately equal proportions in the middle period. Small quantities ofWhite
trade materials
were
present in proveniences attributable to all periods. If further excavationand
final analysis sustain Smith's im- pression, basedon
incomplete results, of continuity rather than a series of discrete occupations of the site it will be of considerable interest, since continuous occupationwould seem
to imply an unex- pected compression of the history of the ceramic types present.Among
other things, pottery apparently closely related to theCamp-
bell
Creek
types,and
perhaps evenassignabletothem, occurs in other contexts without evidence ofWhite
contactand
even the later pot- tery with horizontally incised rimsseems
elsewhere to beprehistoric.Evidences ofintensive aboriginaloccupation
on
alevelterraceadja- cent to thetown
ofOacoma,
across the Missourifrom
Chamberlain,were
assigned thenumbers 39LM26 and 39LM27 by
the River Basin Surveys in 1947 and,more
recently, thenames Sharpe
siteand Dona- hue
site byMarvin
F. Kivett,who
began their excavation for theNebraska
State Historical Society in 1951,under an
agreement with the NationalPark
Service.The
sites, if not actually parts of a single village, appear at least to represent a single cultural complex.A
party of five
worked
there for approximatelytwo months
that year.Two
houses (onein eachof the sites)and
apart of athird, a refuse88 SMITHSONIAN MISCELLANEOUS COLLECTIONS
VOL. I26mound,
asegment
of aditchand
stockade at39LM26, and
anumber
of storagepits
were
excavated. Additional investigationwas
planned for 1952.The
house remains thatwere
completely uncoveredwere
circular, with central fireplace, square central roof support arrange- ment, closely spaced outer posts,
and an
intermediate set of posts that did notconform
to a well-defined pattern.The
vestibule en- trances faced to the southand
southeast.The house
in39LM26
appeared to
have
been extensively remodeled, as indicatedby
the presence oftwo
fireplaces, one intruding into the other,and by
evi- dence ofabandonment and
filling of old post holesand
the setting ofnew
posts.There was
evidence, too, ofan
attempt to seal off old refuse depositsby
blanketingthem
with clay at pointswhere
the house wallswere
expanded. Storage pitswere
oftwo
kinds, bell-shaped
and
cylindrical.The
walls of theformer were
often plastered with red clay, but the cylindrical pits lacked such treatment.The
nature of the artifact inventory is consistent withan
early contact date for the sitesand
this is borne outby
the consistent presence of limitedamounts
of materials derivedfrom White
culturein the vari- ous features.A
large proportion of the pottery is of the TalkingCrow
StraightRim
type definedby
Smith, but other TalkingCrow
types are presentalso,as are cloistered rimslikethose