Front Cover
 Title Page
 Table of Contents
 Geographical Background
 Field Work
 Discussion and Dating
 Back Cover

Group Title: American studies report - William L. Bryant Foundation ; no. 6
Title: Three archaic sites in the Ocala National Forest, Florida
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00053719/00001
 Material Information
Title: Three archaic sites in the Ocala National Forest, Florida
Series Title: American studies report - William L. Bryant Foundation ; no. 6
Physical Description: Book
Language: English
Creator: Bullen, Ripley P.
Bryant, William J.
Publisher: William L. Bryant Foundation
Publication Date: 1965
Spatial Coverage: North America -- United States of America -- Florida -- Ocala National Forest
 Record Information
Bibliographic ID: UF00053719
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 17166708

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Page i
        Page ii
        Page iii
        Page iv
    Table of Contents
        Page v
        Page vi
        Page 1
    Geographical Background
        Page 2
        Page 3
        Page 4
    Field Work
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
    Discussion and Dating
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
    Back Cover
        Page 32
Full Text


American Studies

Report Number Six

Ripley P. Bullen

William J. Bryant

Plate I. Ocala National Forest



American Studies

Report Number Six


Ripley P. Bullen
William J. Bryant



Another gap in the archaeology of Florida's inland shell heaps
has been successfully bridged by this, the sixth in the American
Studies series by the William L. Bryant Foundation. This work
has definitely eliminated many earlier suppositions as to the age
of the shell middens and length of time required for their forma-
tion. The three shell middens discussed in this report are located
in the Ocala National Forest and were among the earliest to have
been inhabited in Florida.
Examination of these sites has also added to our knowledge of
the culture of past inhabitants. It is significant in itself that defi-
nite radiocarbon analyses have established human inhabitation
dating to about 3000 B.C., and that because of changes in land
forms or ecological conditions, the middens were all but aban-
doned one thousand years later.
These three sites are protected under the provisions of the
American Antiquities Act of June 8, 1906 (34 Stat. 225, 16 U.S.C.
433). The land on which they are located is administered by the
Forest Service, United States Department of Agriculture, through
the District Ranger at Eustis, Florida. After obtaining clearance
from the Smithsonian Institution, a Special Use Permit for the
archaeological investigation was issued by the Forest Service, who
will continue to protect these middens but will also make them
available for public use.
These and other sites of archaeological or historical significance
on National Forest lands will be developed in whole or part as a
segment of the Forest's recreation resource. The management

objective on sites such as these is to protect the sites from depre-
dation while providing ample opportunities for use for the public.
To attain this objective, these sites have been set up as archaeologi-
cal areas in Multiple Use and Recreation Management Plans.
Middens No. 1 and No. 2 are the two largest mounds within
the Bowers Bluff Archaeological Area. To facilitate public use,
the Forest Service plans to develop a good access road and parking
area, with water and sanitary facilities. A foot trail through the
cabbage palm hammock will lead from the parking lot to both
Midden No. 3 is the largest of 3 mounds within the 1100 acre
Kimball Island Scenic Area, which is accessible only by boat via
the St. Johns River or Alexander Spring Creek. Current develop-
ment plans for this area stipulate construction of a boat dock and
shelter at Kimball Mud Lake, with foot trails leading to the
The foot trails will serve two purposes. They will enable the
visitor to commune with nature through the observation of the
area's flora and fauna, and they will also be the medium for be-
ginning the mound interpretation program. The Forest Service
plans to interpret the mounds initially through the use of signs
and guided tours. Any display of artifacts and accompanying art-
work telling the prehistoric story, will be located in manned visitor
information centers.
Much of the information to be presented in the interpretive
program will come from this report. It is reports such as this,
which analyze technical data and present it in layman's terms, that
make the study of prehistory interesting as well as educational.
R. J. Riebold, Forest Supervisor
Forest Service




Geographical Background .

Field Work .

Midden No. 1 .

Midden No. 2 .

Midden No. 3 (Kimball Island)

Discussion and Dating .

References Cited .


Plate I. Ocala National Forest Midden No. 2

Plate II. Excavating at Ocala National Forest
Midden No. 1 ..

Plate III. Specimens from Ocala National Forest
Midden No. 1 ..

Plate IV. Ocala National Forest Midden No. 2
during excavation .

Inside front cover



. 14









Plate V. Specimens from Ocala National Forest
Midden No. 2 16

Plate VI. Excavating in lower part of Ocala National
Forest Midden No. 2 Inside back cover

Figure 1. Map of part of the St. Johns River Valley 3

Figure 2. Section of west profile, Ocala National Forest
Midden No. 2 13

Figure 3. Contour map of Kimball Island Midden 18

Table 1. Vertical Distribution of Specimens at
Midden No. 1 7
Table 2. Vertical Distribution of Specimens at
Midden No. 2 12
Table 3. Identified Animal Bones from Midden No. 2 15
Table 4. List of Radiocarbon Dates 23
Table 5. Some Bicarbonate Values for St. Johns River 24
Table 6. Bicarbonate and Calcium Carbonate Values for
St. Johns River. 26
Table 7. Bicarbonate and Calcium Carbonate Values for
some Florida Springs 26


During the winter of 1955-56 a snail shell midden, located in
the Ocala National Forest and situated some distance from run-
ning water, was brought to the attention of the William L. Bryant
Foundation by Frederick W. Sleight, then Director of the Central
Florida Museum. The site appeared to be preceramic and hence to
have respectable antiquity while presenting problems in dating and
An archaeological investigation for the spring of 1957 was
planned in which Sleight, Ripley P. Bullen, Curator of Social
Sciences at the Florida State Museum, and William J. Bryant of
the Bryant Foundation were to participate. Heavy rains over
several years, which made the site practically inaccessible except
on horseback, caused postponement until 1962 at which time
Sleight was unable to participate in the field work. However, he
arranged with the United States Department of Agriculture for
the necessary permission to conduct archaeological explorations in
the Ocala National Forest and for renewals of this permit until the
completion of field investigation in 1964.
We appreciate the assistance given us by Messrs. L. S. Newcomb
and R. J. Riebold, Forest Supervisors, and John Sieker, Director
of the Division of Recreation and Land Uses, Forest Service, United
States Department of Agriculture, for assistance in securing and
renewing our work permit. We are also greatly indebted to Mr.
Robert K. Dodson, District Ranger, Ocala National Forest, and his
staff for location of sites, marking of trails, and transportation by
water to Kimball Island.


Three similar sites were investigated. Bullen and Bryant col-
laborated in the original field work at each of the three sites but
Bullen subsequently returned to the first two to secure material for
additional radiocarbon tests. We are happy to include in this re-
port a contour map of the Kimball Island site. It was compiled by
H. Bruce Greene II, then of Winter Park, Florida, and is based on
independent field work done by him and Carl A. Benson of Orlan-
do, Florida.
This report is concerned with attempts to date in terms of our
calendar three fresh water snail shell middens located near the
St. Johns River in the eastern part of the Ocala National Forest
about 35 miles west of Daytona Beach, Florida. These sites are
known as the Ocala National Forest Midden No. 1, Midden No. 2,
and Midden No. 3 or Kimball Island mound (Fig. 1).
All three middens are large, consist almost exclusively of Vivi-
parus georgianus Lea shells, and are situated at a substantial dis-
tance-about half a mile-from a river or lake where these shellfish
might have been collected. Before describing these middens and
our work at them, it seems desirable to briefly outline the environ-
mental and archaeological situation in this part of the St. Johns
River valley.
The St. Johns River flows northward some 160 miles until it
turns towards the east at Jacksonville to leave Florida and join the
Atlantic Ocean. Throughout this distance there are no falls or
rapids and the total difference in elevation is only 20 feet (U. S.
G. S. Deer Park S E quadrangle; Brown, Kenner, and Brown
1957: 32). Today the channel of the St. Johns appears to us to be
stable but such has not been the case throughout recent geologic
time. In places it is now maintained by dredging.
The valley of the St. Johns may be divided into three parts:
an upper or southern third, a middle portion, and a northern or
lower third. In the upper reaches, the St. Johns consists of a small,
shallow, meandering stream draining a considerable swampy area.
In this region there are no large springs and the size of the river
is determined by rainfall. The northern end of this portion is
at Lake Harney where the St. Johns River enters the eastern part
of the lake region of Florida.



V) )



- a)


0 C


a) d d
e z Z
u -

o- l


- LL o
. a. 0 o
I- 01 u o o
iarf; j" nu>i


In its mid portion the St. Johns River flows from Lake Harney
via Lake Monroe, Lake Dexter, and Lake George to the northern
end of the lake district a little south of Palatka. In this region it
absorbs water from Lake Jessup, the Wekiva River, Lake Wood-
ruff and DeLeon Springs, Alexander Springs, Salt Springs, the
Ochlawaha River and Silver Springs, and Dunns Creek and
Crescent Lake as well as from other minor streams and springs.
Here the flow of the river is dominantly determined by water
flowing out from a great many springs each with a different
mineral content. The chemical composition of the water of the
St. Johns is not only greatly modified as it passes through the lake
region, it is varied with each new intake.
In this region, particularly south of Lakes Dexter and Wood-
ruff, are to be found evidence of channel migration in the form
of old "dead rivers," cutoff oxbows, and isolated lakes (Fig. 1).
The western part of this extremely riverine region is included in
the eastern part of the Ocala National Forest. It is in this area
that our three midden sites are located (Fig. 1).
From Palatka northward the St. Johns, swollen by lake and
spring water, flows slowly but majestically to the sea. Here the
river is very wide and partakes of the nature of an estuary. Be-
tween Jacksonville and the ocean meanders and old channels are
again in evidence.
A great many archaeological sites are to be found throughout
the St. Johns River valley. Large sites pertinent to this report
have been located in Figure 1. Four of them-Bluffton, Dexter
Point, the Harris Creek midden on Tick Island, and the St. Francis
midden-are extremely large and culturally similar. Typically
such sites have lower extensive preceramic Archaic zones of the
Mt. Taylor period (Goggin 1952: 40-43) producing stemmed points
in reasonable quantities, sizeable intermediate fiber-tempered
ceramic zones of the Orange period (Goggin 1952: 43-47, Bullen
1955), and higher but smaller zones of the St. Johns II period
(Goggin 1952: 53-58, Bullen and Sleight 1960) producing St.
Johns Check Stamped pottery. The post-Orange and pre-St. Johns
II, Florida Transitional (Bullen 1959) and St. Johns I (Goggin
1952: 47-53, Bullen and Sleight 1960) periods are poorly repre-
sented. The Spring Garden Creek site is smaller than the others


and may not have a lower preceramic component but is otherwise
These sites, as well as similar ones farther up and down the
river, are composed of a mixture of small snail (Viviparus georgi-
anus Lea), large snail (Pomacea paludosa Say), and mussel (Unio
or Elliptio sp.) shells. There is a tendency for these shells to occur
in alternating layers, but there is no "zone" exclusively composed of
any one of these shells, nor can any one of them be said to be
typical of the lower or of the upper levels. Radiocarbon dates in-
dicate occupation at these sites before 2000 B. C., that the Orange
period lasted from about 2000 B. C. to around 1000 B. C. (Bullen
1961) and that the beginning of the St. Johns IL period occurred
about A.D. 800 (Bullen and Sleight 1960).
Field work was started at Midden No. 1 on January 25, 1962,
and at Midden No. 2 on February 27, 1962. About three days were
spent at each site with a crew of four workmen. Transportation was
by means of the Florida State Museum's 4-wheel drive Jeep which
overcame deep sand, mud, and dense underbrush in getting the
crew to the sites. The boat trip to Midden No. 3 occurred Feb. 19,
Midden No. 1
The location of Midden No. 1 is indicated on Figure 1. This
midden is roughly oval in shape, some 80 feet wide, and about
150 feet long, with its major axis oriented in the north-south
direction. It rises to a height of 13 feet above the surrounding land.
The long western side is curved and gradual in slope compared
with the eastern side which is noticeably straighter and steeper.
Midden No. 1 is situated at the western edge of the present
flood plain of the St. Johns River. Here there is a slight rise in
the elevation of the ground and a significant change in the flora.
To the east the surface is level, damp, and sometimes eroded by
shallow drainage sloughs. Overhead is a canopy of leaves which
keeps out sunlight and, hence, prevents the growth of underbrush.
Travel by foot is easy as vines and bushes are at a minimum. This
flood plain is periodically inundated at times of high water which
leaves behind a small amount of clay and sand.


Plate II. Excavating at Ocala National Forest Midden No. 1.

To the west of the edge of the flood plain, the land rises ir-
regularly and is generally drier. Tree growth does not lorm a com-
plete overhead cover and bushes and underbrush are a problem to
the traveller. Gradually other plant associations become dominant.
Vegetation on the midden itself varies considerably from that
of the surrounding flood plain. Not only does some sun penetrate
to give encouragement to bushy growth but the alkalinity resulting
from the shell composition of the midden is a determining factor
in plant growth. Apparently, poison ivy does not suffer from over-
A 10- by 10- foot test, located near the highest point, was sunk
in Midden No. 1. Results in terms of specimens are presented in
Table 1. The top 30 inches of the midden was composed of a
mixture of small snail shells (Viviparus georgianus Lea) and black
dirt with some admixture of ash and heat-cemented shells. Mixed
with the small snail shells were appreciable quantities of even
smaller snails (Goniobasis Floridanus Reeve and Helisoma scalare
Say). Samples of these shells were kindly identified by William
J. Clench, Curator of Mollusks, Museum of Comparative Zoology,

Table 1


Depths Below the Surface
Specimens Inches Feet
0-6 6-12 12-18 18-24 2-3 3-4 4-5 5-6 6-7 7-8 8-10

St. Johns Plain 29 4 2
Sand-tempered plain 6
Orange Plain 1 1
Pecked sandstone disc 1
Strumbus celt
Fragments, conch shell 1 1
Fragments, bone pins or awls 1 1 6 3 1
Concentration of animal bones x x x x x
Basally-notched point 1
Stemmed point 1
Tip of a point 1
Flake-scraper 4 4 1
Unused chert chips 1 1 2 3 1


Harvard College, Cambridge, Massachusetts (letter of Feb. 7, 1962).
At a depth of about 30 inches below the surface we encountered
a compact zone of heat-cemented shells, approximately 3 inches
thick, over all the excavated area. Associated with these cemented
shells was a considerable quantity of ashes. Below this level down
to a depth of 50 inches the deposit, as before, consisted dominantly
of Viviparus shells with the usual admixture of some Goniobasis
and Helisoma shells. Large snail (Pomacea sp.) and mussel (El-
liptio sp.) shells represented much less than .1 per cent of the total
volume. The presence of numerous small fish bones was noted.
At about 50 inches below the surface another zone of cemented
shells, 3 to 5 inches thick, was met.
Our test continued downward to a depth of 10 feet but, of
course, the excavated areas became smaller as it was impossible to
maintain vertical walls in the loose snail shell deposit. Lower
levels were much the same as higher levels except that we en-
countered a little more ash and a relatively greater quantity of
fish bones. At a depth of 7.5 feet a small, charcoal-impregnated
zone divided the lower deposit into two parts. Just below this
zone, in the 8- to 9-foot level, we noted a concentration of deer and
turtle bones. Throughout this test Viviparus shells consisted of
99 per cent of the volume of the midden.
Ten feet west of the apparent edge of the midden, we made a
small test which disclosed a profile from the surface downward of
6 inches black dirt, 17 inches shell and grey sand, and then sterile
grey sand. Fifty feet further to the west, we made another small
test. There the black dirt was 10 inches thick and rested directly
on the grey sand. Apparently, this is the normal profile. The
underlying gray sand is medium fine and seemed to contain a
little clay.
Subsequently, on April 7, 1964, the senior author accompanied
by C. J. Clausen, then a graduate student at the University of
Florida, returned to Midden No. 1 to secure a sample of Viviparus
shells for radiocarbon dating. Removing the top 9 inches, the
sample was taken from between depths of 9 to 16 inches in a loca-
tion 3 feet west of the test described above. Results of radiocarbon
analysis of this sample known as Sample M (Isotopes, Inc. No. I-
1345) will be discussed later. In collecting this sample a broken


rz~..-- --.- ;<
o-j .. ...- if

l l.

C entlmeter

Plate III. Specimens from Ocala National Forest Midden No. 1.
a, basally notched point; b-c, flake-scrapers; d-e, St. Johns Plain sherds;
f, long bone point.


stemmed point and a sherd of St. Johns Plain pottery were found
The latter was about 5 inches below the surface.
Specimens from the main test at Midden No. 1 are listed by
arbitrary levels in Table I. Three quarters of the St. Johns Plain
sherds (P1. III. d-e) are soft and exhibit occasional red pieces of
sherd temper. The balance are similar but contain both sherd tem-
per and fine sand in small quantities. The sand-tempered sherds
contain a large amount of white quartz sand as temper.
The basally-notched point is intermediate between a Hernando
and a Citrus point (Bullen and Bullen 1963: 89-90). Other speci-
mens merit no special comment except for what we have designated
"flake scrapers" (P1. III, b-c). These tools are made intentionally
from small, thin flakes by steeply chipping one edge by the removal
of minute bits of chert.
The vertical arrangement in Table 1 suggests the anticipated
culture sequence for the region: a preceramic Mt. Taylor period
followed by the Orange period with fiber-tempered pottery and
then by a period when St. Johns ceramics became dominant.
Whether the highest levels represent a St. Johns I period or one
transitional between the fiber-tempered and chalky ceramic epochs
is debatable. The lack of any substantial Orange period deposit
separates this site from the large shell heaps typical of the St.
Johns River. These matters will be discussed further in the con-
cluding phase of this paper.
Examination of Table 1 indicates three zones in which speci-
mens and food bones concentrate: near the surface, around a
depth of 3 feet, and between depths of 8 and 9 feet. Unfortunately
we have no specimens to define the material culture of the Indians
who ate the contents of the shells of the lowest zone.
Animal food remains included bones of deer, turtle, fish, and
racoon and appeared to be the same as those listed later for Mid-
den No. 2. While deer bones were found in fairly equal numbers
down to a depth of 4 feet and again between depths of 8 and 9
feet, turtle bones were limited to the three zones of concentration
mentioned above.
As Midden No. 1 is today remote and difficult of access, it is
of interest to note the fragments of marine shells found at depths
of 6 to 12 and 24 to 36 inches. Communication may have been
easier and the site less isolated many years ago.


Midden No. 2

Midden No. 2 is located about a half mile south of No. 1 (Fig.
1). The environmental situation is exactly the same for both mid-
dens. A small brook or slough flows towards the east about midway
between the two middens. Much the larger, Midden No. 2 is ap-
proximately 380 feet long, about 100 feet wide, and 15 feet high.
Oriented with its major axis extending in a north-south direction,
the east side has a gradual slope with irregular shape and surface
contours while the west side is rather straight and steep.
Our test, 10 by 20 feet in size, was located a little south of the
center of the midden as roots of large trees made excavation diffi-
cult farther north. The vertical position of specimens is given in
Table 2 and the locations of samples taken for radiocarbon dating
are indicated on the west profile (Fig. 2).
As shown by our test, 98 per cent of Midden No. 2 consisted of
small snail (Viviparus) shells. Compared with Midden No. 1,
large snail and mussel shells, while not common, were more fre-
quently encountered, especially at lower depths. The same was
also true of ashes and heat-cemented shells. As before, some very
small snail shells (Goniobasis and Helisoma) were present in small
but noticeable quantities.
By the time the excavation had reached a depth of 8 feet, our
trench was reduced to an area 6 by 4 feet in size. We were able
to get down to 10 feet below the surface without difficulty although
our hole became of necessity still smaller. Below that depth we
were unable to dig and maintain levels because of the looseness of
the shells. Every shovel full was immediately replaced as soon as
it was removed. Our maximum penetration was to a depth of 11.5
feet. We regret that we were unable to reach the base of Midden
No. 2. It is doubtful if the knowledge to be gained would have
merited the substantial enlargement of our test.
This fairly homogeneous deposit was divided vertically into
several zones as shown in Figure 2. Between depths of 30 to 36
inches was a thick layer of cemented shells, charcoal, and black
sand. Charcoal from this level formed our radiocarbon sample
"F." At a depth of 5 feet we uncovered what appeared to be a
hearth, 6 by 2 feet in size, composed of cemented shells. Charcoal
in large quantities was not present. At 8 feet was another zone of

Table 2

Depths Below the Surface
Specimens Inches Feet
0-6 6-12 12-18 18-24 2-3 3-4 4-5 5-6 6-7 7-8 8-10

St. Johns Plain 1
Fragment, conch shell 1
Peg-topped bone pin 1
Bone points 2 4
Bone awls 2 3
Worked bone 2 3 1 7 3 3 1 1
Concentration of animal bones x x x
Stemmed point 1
Flake-scraper 1
Drills I 1




E F Black san char-
coal 8 cemented sh

ert of ceme hes


Bloc line

Figure 2. Section of west profile, Ocala National Forest Midden No. 2.











heat-cemented shells on top of which Indians had deposited a large
amount of clean sand. Radiocarbon sample "B" was collected
from just below this layer of cemented shells. Sloping downwards
between depths of 9.5 and 11 feet (Fig. 2) was a narrow dark line of
burnt mussel shells. Our lowest radiocarbon sample, Sample C,
came from below this line. Apparently, shells in this part of our
trench collected rapidly as radiocarbon tests of Samples B and C did
not indicate a different age.
As shown in Table 2 the only sherd of pottery we discovered
in our trench came from the first 6-inch level. To check further
the ceramic picture of Midden No. 2, we dug a series of nine small
tests. Three of these, each 2 feet in diameter and 18 inches deep,
were located in the central and northern part of the top of the
midden. They were sterile. The fourth, 2 by 2 feet in area, was
located a little further to the northeast in a small level area about
2 feet lower than the top of the midden. This test produced two
St. Johns Plain sherds. The other five tests, all 2 by 2 feet and
located at presumed to be strategic spots at lower elevations around
the midden, were sterile. We decided that any use of this midden

Plate IV. Ocala National Forest Midden No. 2 during excavation.


during ceramic periods must have been very superficial. Sherds
at Midden No. 2 were both less common and shallower in pro-
veniences than at Midden No. 1.
The one St. Johns Plain sherd from our main test at Midden
No 2 was, like the two mentioned above, similar to those from
Midden No. 1 (P1. III, e). Other specimens listed is Table 2 include
bone and stone tools and animal food bones.
A peg-topped bone pin (P1. V, i) was found in the 12- to 18-
inch level. Six fragments of bone points (PI. V, f-g) were located
between depths of 18 and 36 inches. Five bone awls (P1. V, d-e, h)
were concentrated between depths of 2 and 4 feet, Pieces of worked
bone had general distribution above a depth of 5 feet. The number
of bone specimens greatly exceeded those found in Midden No. 1.
Only one flake-scraper, like those found in Midden No. 1 (PI.
III, b-c), came from Midden No. 2. Its relatively higher proven-
ience, in the 6- to 12- inch zone are opposed to the 18- to 48- inch
zones, agreed with the scarcity and shallowness of pottery in sug-
gesting the highest levels of Midden No. 2 were older than the cor-
responding levels of Midden No. 1.
Other stone specimens from Midden No. 2 included two drills
or perforators (P1. V, a-b) and a stemmed point (P1. V, c), all
from the highest foot of the midden. A piece of sandstone from
the 18- to 24-inch level completed the stone inventory.

Table 3

Mammal rabbit Sylvilagus sp.
raccoon Procyon lotor
white-tailed deer Odocoileus virginianus
Amphibian siren Siren lacertina
Reptile musk turtle Kinosternon sp.
box turtle Terrapene carolina
terrapin Pseudemys sp.
soft-shelled turtle Trionyx ferox
Fish gar Lepisosteus sp.
bowfin Amia calva
redear sunfish Lepomis microlophus


a b c d

e f g hL j


Plate V. Specimens from Ocala National Forest Midden No. 2.
a-b, stone drills or perforators; c, stemmed point; d-e, bone awls; f-g, long bone
points; h, long bone awl; i, peg-topped bone pin.


Food bones were evenly distributed throughout our test ex-
cept for the heavy concentration around a depth of 30 inches,
where they were associated with the thick layer of cemented shells
and charcoal found at that depth. The above list of the ani-
mals represented by those bones has been prepared by Elizabeth S.
Wing, Assistant Curator of Zooarchaeology, Florida State Museum.
After the results of the first radiocarbon tests were received
and it was decided to run shell samples from Midden No. 2, we
realized we had neglected to get a sample from the top of our
section. The senior author consequently returned, accompanied
by Howard A. Chamberlen, Museum Technician at the Florida
State Museum, and Morgan MacLachlan, then a graduate student
working at the Museum, and collected a sample at a depth of 6
inches from undisturbed shell adjacent to our trench. It became
known as Sample A.

The Kimball Island Midden

Usually referred to as the Kimball Island Midden or Mound,
Midden No. 3 (Fig. 1, 8) is more remote than either of the others.
Under the right water conditions and with a good guide and boat,
the trip is very pleasant. We were taken up Alexander Spring
Creek and across 1-foot deep Kimball Mud Lake to the western side
of Kimball Island (Fig. 1). From there it was only a short walk
to the midden. This was the same route as that used by Greene
and Benson when they surveyed the midden and collected the data
for their contour map of the mound (Benson and Greene 1962:
113). This map, a copy of which was kindly given by them to us,
has been reduced and presented here as Figure 3.
The surface of Kimball Island is only a foot or two above that
of the surrounding water. The midden is divided into two ovate
areas. The smaller, to the northwest, is also the lower in altitude
as it rises to a height of only 13 feet while the two high points in
the larger area, towards the southeast, both reach 17 feet above
the surrounding land. The Kimball Island midden is considerably
larger in area than either of the other two sites previously discussed.
The contour lines may be interpreted as suggesting that a group of
four huts were located at the site.

0 80 160 Feet '/


Figure 3. Contour map of Kimball Island Midden.


Our examination revealed that while dominantly formed of
small snail (Viviparus) shells, more large snail and mussel shells
were present than at the other sites. The presence of several "pot-
hunters" holes gave us an idea of the internal structure of the
shell deposits. While similar to those at the other middens, more
dirt seemed to be present and in one place a sand zone was noted.
Examination of the surface, of pot-hunter's spoil piles, and of
the edges of their holes failed to disclose any pottery. Benson and
Greene during the three days they worked at the site-two days
surveying, one day testing the strata-found only three sherds.
These, two St. Johns Plain and one St. Johns Check Stamped,
were in the first 6-inch levels of several pits (Benson and Greene
1962: 114). Pottery was as scarce at Kimball Island as at Midden
No. 2.
Along the steep western side of the midden and in the curved
area between the two lobes of the site, the surface seemed sunken
as if a ditch had been present. One is tempted to see here the
remains of a cutoff oxbow, now nearly filled up, in which shells
may at one time have grown. Alternatively, this depression may be
the remains of a creek.
Examination of the air photo of the Kimball Island region
at the United States Agricultural Stabilization and Conservation
Service office in Gainesville, Florida, indicates a "slough" extends
from the southern end of the lake, located to the northeast of the
site, southwesterly to the "bay" located below the "n" of the word
"Island" (Fig. 1). The Kimball Island mound is not on this
slough but the depression mentioned above may have been a con-
necting creek. Years ago the site on Kimball Island may have been
more accessible to shellfish producing waters than is the case today.
The Kimball Island mound is so similar to the other middens
covered by this report that it seems likely it was occupied con-
temporaneously by the same or culturally similar people.


Our investigations indicate there is a series of sites in and
along the edge of the St. Johns River flood plain which have cer-
tain consistent characteristics. While these investigations were


limited to three sites, it is believed other with the same charac-
teristics could be located with additional field work. At least we
have heard comments of hunters and others which suggest the
existence of similar sites.
Midden No. 1, Midden No. 2, and the Kimball Island midden
are all located at a substantial distance from water, are dominantly
composed of small snail shells with only a trace of large snail or
mussel shells, contain sand zones, and do not produce pottery in
large amounts. At Midden No. 2 and Kimball Island the sherds
found were both few in number (a total of six St. Johns Series
sherds) and limited to the top 6 inches. We feel they can easily be
accounted for by casual use of the sites by campers during the St.
Johns period. We conclude that these two middens may be con-
sidered as having been entirely built during the Mt. Taylor or
preceramic part of the Archaic period. This would indicate
abandonment before the introduction of fiber-tempered ceramics
or before 2000 B.C.
At Midden No. 1 the situation was different. There St. Johns
Plain sherds were found in fair quantities (Table 1). While con-
centrating in the highest zone a few were found as deep as 12 to
18 inches. Use of the site during the Orange period (2000-1000
B.C.) is suggested by two Orange Plain sherds (Table 1). Whe-
ther the Orange period is represented by deposits between depths
of 6 and 18 inches or whether deposits of that period extend
deeper is not certain from our limited test. Certainly the lower
levels of this midden must be considered preceramic in date.
Midden No. 1, then, was started during the Mt. Taylor period,
occupied during at least part of the Orange (or Transitional) per-
iod, and used for habitation in the early part of the St. Johns I
period. This sequence is similar to that at the large middens along
the present or recent channel and lakes of the St. Johns River such
as Bluffton, Dexter Point, Tick Island, and St. Francis (Fig. 1). It
seems reasonable to equate Midden No. 1 temporally with the early
and middle parts of these St. Johns River middens. The latter
sites contain St. Johns Check Stamped pottery in the highest
zones indicating use in the St. Johns II period. Occupation at
Midden No. 1 did not last that late.
We believe that all of Midden No. 2, all of the Kimball Island
midden, and the lower part of Midden No. 1 are culturally and


chronologically the equivalent of the lower, preceramic zones of
the St. Johns River sites while the upper part of Midden No. 1
equates with the middle period of the St. Johns River middens.
These correlations should be reflected in radiocarbon dates.
It should be emphasized that the only difference between the
composition of the Ocala National Forest sites and that of the
preceramic layers of the St. Johns River sites lies in the fact that
the former are composed almost exclusively of small snail shells
while the latter also contain large quantities of large snail and
mussel shells.
While excavating at Midden No. 2, the point was forcibly
brought to our attention that we would gain very little informa-
tion from the artifacts we were (not) finding that would be useful
in reconstructing life at the site or in dating the deposits other
than as preceramic. The lack of other than small snail shells sug-
gested a local environment different from that of the St. Johns
River proper. Perhaps, if we could gain some idea of the length
of time it took for the midden to accumulate, it would help to
eliminate one of the unknowns in population estimates. Radio-
carbon dating of the shells should give us temporal differences
between the top and the bottom of the midden and tell us when
occupation had occurred in terms of our calendar. At the time of
excavation no radiocarbon dates were available for the preceramic
part of the Florida Archaic period. To this end we collected
samples of Viviparus shells.
To secure a sample of recent Viviparus shells for a test to find
out if such shells when growing in limestone waters were con-
taminated by antique carbon, Bullen on March 2, 1962 went down
Spring Garden Creek (Fig. 1) from DeLeon Springs to a place
where he remembered seeing Viviparus shells. Specimens were
collected from around and near the roots of a tree uprooted by
Hurricane Donna in 1960. These specimens were dead but some
of the skin was present and it was presumed they had not been
dead very long.
When this matter was discussed with Isotopes, Inc., Westwood,
New Jersey, they felt that the "recent" shells from Spring Garden
Creek might have been contaminated by bomb fallout C-14 and
asked if at all possible, that we secure shells collected at least 15
years ago. A check of the research collections of the Florida State


Museum disclosed a few Viviparus shells (Lot No. 4492) collected
in 1938 from the north shore of Lake Monroe, 2 miles west of
Sanford, Florida. Not realizing at the time the extreme variation
in the chemical content of the water in the St. Johns River, some
of the Lake Monroe shells were sent to Isotopes, Inc., to serve as
a modern standard. In the same shipment went charcoal we had
collected from a depth of 33 inches in Midden No. 2.
These samples are known as Samples G and F respectively and
the results of their radiocarbon runs are given in Table 4 which
also includes the results of tests on our other samples. The
charcoal, Sample F, gave a date of about 2775 B.C. which seemed
reasonable as it was comfortably earlier than the date for the intro-
duction of pottery (2000 B.C.). When secured in 1962, this was
the earliest Florida Archaic radiocarbon date. As the Lake Monroe
Viviparus shells proved to contain 98.7 per cent of the C-14 activity
of the National Bureau of Standards oxalic acid standard and to
yield a date equal to or less than 85 years B.P., we decided to con-
tinue our efforts to date Midden No. 2 by dating the shells which
formed it.
Samples A, B, and C were sent to Isotopes, Inc., and dated. The
results-4500 B.C., 5450 B.C., and 5300 B.C. (Table 4)-had a cer-
tain amount of internal consistency but differed considerably from
the date (2775 B.C.) previously secured from the charcoal sample,
Sample F, which was stratigraphically located between Samples A
and B. Assuming the charcoal date to be approximately correct,
our modern sample from Lake Monroe was not a proper standard
for the shells from Midden No. 2. In other words the environment
in which they grew differed considerably from that of Lake Monroe.
This led us to a consideration of the chemical content of the
waters feeding the St. Johns River. Some of the available analyses
are presented in Tables 5-7. In order to correlate the dates from
Midden No. 2 with our calendar we had a test run on Viviparus
shells (Sample B) from the same zone as the previously tested char-
coal sample (Sample F). To test the effects of an extreme limey
environment on Viviparus shells we also had the sample from
Spring Garden Creek (Sample D) tested.
These runs gave the expected result. Sample E from the 33 inch
depth of Midden No. 2 produced a date nicely intermediate between

Table 4

Bryant- Isotopes, Before Approximate Adjusted
F.S.M. Inc. Material Provenience Present calendar date
number date date

G 1-564 Viviparus Lake < 85 present
shells Monroe
D 1-684 Viviparus Spring Garden 4300 175 2350 B.C. ??
shells Creek
A 1-634 Viviparus Midden No. 2 6450 250 4500 B.C. 2225 B.C.
shells 6 inch depth
E 1-683 Viviparus Midden No. 2 7000 250 5050 B.C. 2775 B.C.
shells 33 inch depth
F 1-563 charcoal Midden No. 2 4725 180 2775 B.C. 2775 B.C.
33 inch depth
B 1-635 Viviparus Midden No. 2 7400 250 5450 B.C. 3175 B.C.
shells 8.5 foot depth
C 1-636 Viviparus Midden No. 2 7250 250 5300 B.C. 3025 B.C.
shells 10 foot depth
H 1-1345 Viviparus Midden No. 1 3640 110 1690 B.C.
shells 12 inch depth


Table 5
(parts per million)

Dates of samples
Location of samples June 7, 1962 July 7, 1962 Aug. 15, 1962

At entrance to Lake Monroe
(Highway 415) 66 66 46
At exit from Lake Monroe
(Highway 17) .96 68
At Crows Bluff (Highway 44) 152 120 80
At Astor Park (Highway 40) 116 102

those from Sample A (6 inch depth) and Sample B (8.5 foot
depth). The date from Sample D from Spring Garden Creek,
2350 B.C., indicated contamination by antique carbon but was
nowhere near as old as those from the presumed substantially older
shells of Midden No. 2 (Table 4).
We believe that the difficulties encountered in dating these
Viviparus shells are caused by variations in the environments in
which they grew, which we interpret to be differences in the per-
centages of carbonates containing antique carbon present in the
water. Springs in the lake region of Florida emit water which
comes from an aquafer composed of Eocene limestones. Carbonates
in this water (Brown, Kenner, and Brown 1957) must contain
antique carbon which would dilute the radiocarbon activity of
carbonates (HCO, or CO2) of any body of water into which
such springs flow. The resultant effect, of course, would be
a question of relative volumes and percentages. Radiocarbon an-
alyses of organisms living in such water and securing their carbon
from the carbonates in these waters would be expected to produce
abnormally old dates.
Recent bicarbonate values (HCO,) for four locations on the St.
Johns River are given in Table 5. Month to month variation is
evident. Data taken at one location near DeLand on the St. Johns
River over 1948 and 1949 show that these carbonates are two to
three times as plentiful in the summer, when the river temperature
is high, as in the winter, when the temperature of the water is low.
These data were kindly given us by Kenneth A. MacKichan (letter


of Sept. 19, 1962), District Engineer, Quality of Water Branch,
United States Geological Survey, Ocala, Florida. If shellfish should
grow more rapidly in the summer than in the winter the effect
would be cumulative.
We do not understand why the HCO, values increased between
the entrance and exit of the St. Johns River at Lake Monroe, but
clearly the lowest value is that given for Lake Monroe (Table 5).
The increase between Lake Monroe and Crows Bluff reflects the
entrance of water from Wekiva and Rock Springs, sources high
in HCO,. Our Viviparus sample from this lake, as was the case,
would be expected to give a radiocarbon date more nearly correct
in terms of our calendar than would samples from farther down the
river where the content of carbonate containing antique carbon is
greater. The north shore of Lake Monroe is not on the direct
line of river flow. Possibly our sample, which came from that
shore, may have been growing in a location where the concentra-
tion of antique carbon was relatively low.
Available bicarbonate and calcium carbonate values at various
points on the St. Johns River for a three day period in 1954
(Brown, Kenner, and Brown 1957: 94-110) are given in Table 6.
The table is arranged downstream from south to north. The in-
crease in these values below Christmas is very evident as is the
large variation from collecting station to collecting station. Our
sites are located off the river but between Crows Bluff and Astor
where the highest HCO, values are indicated. These values reflect
the addition to the St. Johns River of water from different tribu-
taries many of which originate in springs. In order to show the
variation in the waters from these springs, partial analyses of the
principal springs feeding the St. Johns River have been given in
Table 7 (Ferguson, et al. 1947: 119-67). It is evident that the mico-
environment of growing shells in this area is very critical in regard
to the availability of antique carbon.
We do not know the source of shells forming Midden No. 2,
nor is there any way of ascertaining it. Because of the internal con-
sistency in dates, it seems likely all the shells from this midden came
from the same source. Sample H from Midden No. 1 differs con-
siderably in its date (Table 4) from any of the Viviparus dates
from Midden No. 2. It seems likely shells of Midden No. 1, at


Table 6
(parts per million)
May 17-19, 1954





Near Malabar 33 34 Near Mins 76 1240
Near Melbourne 18 29 Near Geneva 48 320
Near Bonaventure 28 36 At Osceola 38 265
At Lake Winder 26 41 At Lake Jessup 43 246
At Lake Poinsett 42 81 At Lake Monroe 64 220
Near Delespine 22 88 Near Sanford 58 202
Near Christmas 35 130 At Crows Bluff 94 204
At Kyser Ranch 66 573 At Astor 93 212

least those from the top of that midden, came from a different
environment than did the shells of Midden No. 2.
Sample D from Spring Garden Creek grew in water from
Ponce de Leon Springs which is one of the higher contributors of
antique carbon into the St. Johns River system (Table 7). When
we collected this sample, we thought the shells composing it had
only very recently died. Greater knowledge of the habits of Vivi-

Table 7
(parts per million)
All flow into the St. Johns River

Springs HCO3 CaCO3 CO, HCO3 CaCO, CO

Seminole County Lake County
Sanlando 125 105 13 Alexander 98 176 20
Sheppard 141 126
Marion County
Orange County Salt 87 1290
Rock 105 107 -.. Silver Glen 85 406
Kekiva 117 94 ..- Silver (near
Volusia County Ocala) 201 209
Blue 148 399 6
Ponce de Leon 130 340


parus shells sheds some doubt on that assumption. These shells
were exposed by the roots of a tree blown down by hurricane
Donna, but if alive at the time the tree blew down they must have
been living "in" instead of "on" mud. It seems likely they were
washed in a dead state onto a mud flat-or died on the flat- to be
then covered with more mud and to be later exposed when the
tree blew down. They thus would predate the growth of the tree.
While we cannot tell the exact origin of these shells, their ex-
tremely old date (2350 B.C.) certainly reflects their growth in
Spring Garden Creek, an environment abounding in antique car-
bon. Parenthetically, there is no preceramic midden on Spring
Garden Creek from which the shells of Sample D could have
There are some references in the radiocarbon literature to dates
on contemporary shells which were 1350 to 1800 years older than
present (Crane and Griffin 1959: 176-77). There is also a note in
Radiocarbon Supplement (Vol. 4: 46-48) under the Kincaid
Shelter series that "all dates from the same excavation units are in
proper sequence. Dates on charcoal are for the most part reason-
able, while dates on snail shells are older than expected." This
seems to match our experience.
In 1964 in an article in Science, Broecker discussed the problem
of C-14 and C-13 deficiencies in lake and river mollusks and decided
that "solution of carbonate rocks can be conclusively demonstrated
to be a major source of dissolved carbon and an entirely adequate
source of the observed C-14 deficiency." Our findings would sup-
port Broecker's conclusions.
The investigation reported here is not a quantitative study of
the effects of antique carbon on Viviparus shell radiocarbon dates.
We hope others will collect live samples from spring waters of
different composition and from river and lake waters of known low
concentrations of carbonates. Radiocarbon runs on such samples
should indicate quantitatively the relationships between B.P. dates
and the percentages of carbonates and, hence, antique carbon
This brings us to the question of the dates for Middens No. 1
and 2. As shown in Table 4, we have assumed that the radiocarbon
date from the charcoal (Sample F) from the 33 inch zone at Mid-


den No. 2 is approximately correct. The date for Viviparus shells
associated with this charcoal was greater by 2275 years. Assuming
this difference to be the correct adjustment, we have subtracted
2275 years from the other Viviparus dates for Midden 2 to get the
adjusted dates shown in the last column of Table 4. Archaeo-
logically, these dates are, if anything, conservative. They indicate
the abandonment of Midden No. 2 around 2225 B.C. or over 200
years before the introduction of fiber-tempered pottery. This is
a reasonable correlation as no pottery was found in Midden No. 2.
Dates for the basal part of the midden, but not the bottom, are
around 3100 B.C. This suggests Indians started living at the site
around 3200 B.C. and continued-possibly intermittently-living
there until about 2200 B.C. or that it took about 1000 years for
Midden No. 2 to accumulate. Again this seems reasonable. This
implies that a small group, not more than one extended family, lived
at the site for a long time. Such a small group living in balance
with the local ecology seems more likely than that the midden
represents the accumulation of a large group who, in a short period
of time, might be expected to kill off shellfish by over collecting.
We have only one date for Midden No. 1. This date, around
1690 B.C., came from a depth of about 12 inches. As shown in
Figure 1, this depth could represent an early St. Johns I (circe
A.D. 0-500), a Transitional (circa 750 B.C.), or an Orange (2000-
1000 B.C.) period. If the date were to be taken as indicated, it
would fit nicely at about the middle of the Orange period. If the
date reflects the presence of some antique carbon, as is likely the
case, it might fit either of the other two periods mentioned above.
We feel there is plenty of evidence, as mentioned earlier, to
believe that the channel of the St. Johns River has not been con-
stant over thousands of years (Fig. 1). It seems to us likely that
both Midden No. 1 and No. 2 were located on one or more oxbows
of the river when first settled by Indians. If the oxbow became
cut off from the main channel, Viviparus shells would continue to
grow and, hence, the Indians would not have to move as their
dominant food supply would still be present. Over a long period
of time, flood waters would leave enough sand and clay behind
to eventually fill the oxbow lake with dirt and other debris and
the Indians would be forced to move. Judging from our radiocar-
bon dates, this process took a long time.


It has been our experience that middens bordering lakes,
streams, or bays are steep on the water side and slope more gradu-
ally and more irregularly on the landward side. In other words,
middens grow upward and away from the water. In our description
of both Middens No. 1 and No. 2, we mentioned that one side was
steep and the other more gradual. We would postulate that the
steep side was towards the oxbow lake. A similar situation occurred
at Kimball Island. Of course, an alternative hypothesis might
be that now-completely-hidden springs once existed at these sites
and that Indians exploited shellfish growing at these springs.
We conclude that around 3000 B.C. the channel of the St.
Johns river was in a different location than today. Indians lived
not only on the large bodies of water such as Lake Dexter but also
on small oxbows or oxbow lakes. Middens No. 1, No. 2, and
probably No. 3, were so situated. There the women and children
collected Viviparus shells, turtles, and probably fish while the
men and youths hunted deer and trapped rabbits and raccoons in
the neighboring forests and swamps. Some vegetable products such
as nuts, seeds, and the hearts of cabbage palms undoubtedly
broadened this high protein diet.
Occasionally they met Indians from other small settlements.
Villages on the large lakes prospered and became towns but those
on the cut-off oxbows remained small as their supply of the stable
shellfish was limited. Finally, shortly before 2000 B.C., the old ox-
bow lakes became filled and ceased to supply Viviparus shells in
any quantity. When this event happened, it forced the abandon-
ment of these small settlements and their inhabitants moved else-



1962. "The Kimball Midden, Lake County." Florida Anthropologist, Vol.
15, No. 4, pp. 113-14.
1964. "Radiocarbon Dating: A Case against the Proposed Link between
River Mollusks and Soil Humus." Science, Vol. 143, No. 3606, pp.
1957. "Interim Report on the Water Resources of Brevard County,
Florida." Florida Geological Survey, Information Circular No. 11.

1963. "The Wash Island Site, Crystal River, Florida." Florida Anthro-
pologist, Vol. 16, No. 3, pp. 81-92.

1955. "Stratigraphic Tests at Bluffton, Volusia County, Florida." Florida
Anthropologist, Vol. 8, No. 1, pp. 1-16.
1959. "The Transitional Period of Florida." (15th) Newsletter of the
Southeastern Archaeological Conference, Vol. 6, pp. 45-53.
1961. "Radiocarbon Dates for Southeastern Fiber-tempered Pottery."
American Antiquity, Vol. 27, No. 1, pp. 104-6.
1960. "Archaeological Investigations of Green Mound, Florida." The
William L. Bryant Foundation, American Series, No. 2.
1959. "University of Michigan Radiocarbon Dates IV." American Journal
of Sciences, Radiocarbon Supplement, Vol. 1, pp. 173-98.

1947. "Springs of Florida." Florida Geological Survey, Geological Bulle-
tin, No. 31.
1962. Radiocarbon Supplement, Vol. 4. American Journal of Sciences.
1952. "Space and Time Perspective in Northern St. Johns Archeology,
Florida." Yale University Publications in Anthropology, No. 47.

I'late \I. X Excaaating iII In Iower part of Ocala Nationial IFrcst Mliddti 'No. 2.


Report Number 1. Archaeological Investigations of the Castle
Windy Midden, Florida, by Ripley P. Bullen
and Frederick W. Sleight. Price $1.00.
Report Number 2. Archaeological Investigations of Green Mound,
Florida, by Ripley P. Bullen and Frederick W.
Sleight. Price $1.00.
Report Number 3. Archaeological Reconnaissance of the Island of
St. John, United States Virgin Islands, by Fred-
erick W. Sleight. Price $1.00.
Report Number 4. Ceramic Periods of St. Thomas and St. John
Islands, Virgin Islands, by Ripley P. Bullen.
Price $1.00.
Report Number 5. The Krum Bay Site: A Preceramic Site On St.
Thomas, United States Virgin Islands, by Rip-
ley P. Bullen and Frederick W. Sleight. Price

Report Number 6. Three Archaic Sites in the Ocala National For-
est, Florida, by Ripley P. Bullen and William
J. Bryant. Price $1.00.

rTl.r 2ro in 'I1?'Ste Horida Nfs.ei
^gEE~~nS&-[ye^Orlandor FktkIte.

L jary o Gn~alltuaLj:


University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs