Sorry, I don't have the images in yet.


Integrating 19th-century Documents and a Geographic
Information System


Erich K. Schroeder
Illinois State Museum
Springfield, Illinois 

Paper presented at the Mid-America GIS Symposium Overland Park,
Kansas, May 4-7, 1992

ABSTRACT: The Illinois State Museum has made several efforts in
integrating 19th-century documents and a GIS. These efforts have focused
on two groups of documents; land entry records produced at federal land
offices, and commercial county atlases. By relating the legal location
recorded on the land office documents with information digitized from
modern topographic maps it is possible to produce various maps reflecting,
for example, earliest or last date of entry per section, an average date of
entry per section, or the distribution of a particular individual's purchases.
This process has been completed for the entire state of Illinois. Most 19th-
century commercial county land ownership atlases show the locations of
farmsteads and other cultural features in rural areas. Several methods for
transferring information to the GIS have been tried, including: recording of
structure counts per section, and digitizing from photocopies. The current
method is to hand-transfer the feature locations to township-and-section
outline plots produced at the scale of the original documents. This method
minimizes distortion due to inaccuracies in the atlases. Transfer errors are
controlled by producing plots in a format that can be overlaid on the
atlases. This process is presently ongoing at the Illinois State Museum. 


Introduction

    There are always technical difficulties in transferring
information from paper records to an electronic medium. These difficulties
are often increased when dealing with older records. Problems differ
depending on whether the documents are texts or maps (Figure 1). In the
case of texts, the general problem stems from the difficulty in translating
text to spatial information. In particular, the information may be expressed
in unknown or unusual coordinate systems, the writing may not be legible,
and there may be a low degree of standardization in how the information
was originally coded. On the other hand, maps may have been produced in
unknown projections, the printing may be muddy or illegible (particularly
when dealing with a copy of the original), the map may have been produced
with an unknown level of accuracy and precision, and there is often a
difficulty in identifying needed registration points needed by a GIS. In both
the cases of texts and maps, the original documents may be in fragile
condition, forcing the use of reproductions.

My goal in this paper is to present methods of integrating two forms of
documents and a Geographic Information System (GIS). One set of
documents consists of tabular data recorded from 1814 to the 1890s at
federal land offices located in Illinois. The other set consists of 19-century
commercial county atlases. 

Land Office Records Mapping Project

The land office records mapping project is an example of going from
text-based information to a GIS. I will begin with a brief history of the
public domain documents.

When the Treaty of Paris ended the War of Independence in 1783, the
western boundary of the United States was suddenly extended to the
Mississippi River. The new government was faced with the problem of
how to dispense and settle the new lands, known as the public domain. In
general, the government decided to make land available to private
individuals in the hope of settling the new land with privately-owned
farms, while at the same time providing funds to help support the new
government. To facilitate this, the familiar rectangular survey system was
developed based on townships 6 miles on a side. The government provided
sets of instructions to surveyors, and most of the United States has been
mapped using these methods in some form.

The laws that set up surveying methods also stipulated rules for the
transference of the public domain lands into private ownership. Land could
not be sold until it had been surveyed. It could only be sold in parcels that
could be referred to in terms of the township, range, section, and fraction of
section (McEntyre 1978:38-40). The minimum allowed size of land parcel
changed through time, and was also set by law (Hart 1974:77). When a land
transaction took place it was recorded as an entry into a log book at the land
office. Since this transaction might have been a purchase, or a grant in
which no money was involved, it is best to use the generic term "entries" to
refer to these transactions. 

Surveys began in Illinois in 1804, and the first property was sold at the
Shawneetown land office in southern Illinois in 1814 (Matousek 1971;
Gaynon 1975; Schroeder 1991). By 1834 there were 10 active land offices
in Illinois. As the amount of land remaining in the public domain dwindled,
the land offices were closed and consolidated at the Springfield office.
When the Springfield office closed in 1876, the records were transferred to
Washington, D.C. Later, the federal government returned many of the
records to some state agency, in Illinois' case the Illinois State Archives
(Gaynon 1975; Smith 1975). Notable to states settled after 1863,
Homestead Act records are retained by the Bureau of Land Management.

In 1975 the Illinois State Archives began a 9-year project to computerize
the land records in their possession (Kremm 1975). The goal of this project
was to make the information more accessible to the public while protecting
the original documents from overuse. The project was funded through a
combination of a grant from the National Endowment for the Humanities
and the Illinois State Archives' summer internship program, and a
dedicated group of volunteers. Early in the project some nonvoluntary labor
was also used in the form of prisoners under the direction of the Illinois
Department of Corrections. This was discontinued due to supervisory
problems (Robert Bailey, personal communication).

As a minimum, each coded record consists of: the date of the transaction,
name of the grantee or purchaser, number of acres, price per acre, total
price, legal description of the land parcel (principle meridian, township,
range, section, and quarter-quarter description or lot number), the type of
transaction, and the volume and page of the entry. The county or state of the
grantee was also coded, although this was not consistently recorded by the
land office personnel. Three additional variables were added: a one-
character variable was coded to indicate the sex of the grantee, based on the
coder's interpretation of the name; an unique record number was assigned
to each record; and a numeric code indicating the present-day county
within which the land parcel is located. These last two were added
automatically after the records had been read into the computer.

Since the project was originally designed to be punched on to cards, the
records were limited to 80 columns. As a result of this limitation, over 200
codes and abbreviations were devised during the project. Duplication of
codes was controlled by keeping a master set of definitions on index cards.
Although a few duplications did make it through the process, there is a
record.

     The project was finished in 1984, and the resulting file contains
more than 500,000 records of land entries. The Archives has produced three
microfiche hard copies of the file (Figure 2). One is sorted by legal location
and the others are sorted alphabetically by the last name of the grantee, one
by county and the other for the entire state. The file has been backed up on
tape, and is not maintained on line by the archives. In the early 1980s the
Illinois department of Energy and Natural Resources contracted with
Environmental Systems and Research Inc. to produce the basic state-wide
coverages for the Illinois Geographic Information System (ESRI 1984).
One of these ARC/INFO coverages contains the township, range, and
section boundaries, digitized from the 1:24,000 7.5 minute USGS
topographic maps. Each section is a separate polygon, and has the legal
location associated as variables for: principle meridian, township and range,
and section.

In 1988 the Illinois State Museum obtained a tape of the public domain land
records from the State Archives. The tape was produced on the Illinois
Secretary of State's Honeywell computer to the requirements of the Illinois
Department of Energy and Natural Resources' Prime 750, where the GIS
resides.

     Combining the tabular and geographic information required
several steps (Figure 3). The first, reading the tape, was done by GIS system
programmers. After the position of the numeric variable signifying the
county was determined, a FORTRAN program was written to separate the
file into county subsets. The steps of translating the file into INFO and
rearranging the legal location variables were done through the INFO
programming language. 

Once the structure of the original text file is known it can be easily
imported into an existing INFO file structure. However, after attempting it
the first time we realized that the variables had to be rearranged before the
file could be used. Rearrangement was necessary both in the placement of
the variables in the file structure, but also in how some of the variables had
been coded. In particular, the original numeric data had been coded with
leading zeros as place- holders for the townships, ranges, and sections. This
was accounted for by importing the data in a two- stage procedure, using a
temporary file that was then related to an empty file with the correct
structure. 

After rearranging the legal location variables, the final steps of relating
tabular and geographic information and producing maps of derived values
are relatively straightforward. Examples of derived variables include, the
average date per section, the percent of the section purchased by a certain
date, first or last date of entry per section, or the distribution of a particular
individual's purchases.

The resulting maps give a view of the settlement of Illinois that has never
been seen before. The broad trends of land speculation, early state and
national road systems, railroad and canal construction, and environmental
preferences are easily visible. Also, the land purchase data can be overlain
on other geographical data existing within the GIS, leading to more
in-depth modeling. The spatial patterns of land acquisition in an area are
the results of an interplay of social, ecological, economic, and legal
dimensions. Selective overlays of the land records on other variables will
result in an analysis of the relative importance of each dimension, and the
change in importance through time.

There are, however, some weaknesses in this data set as it has been
constructed. Chief among these is the limitation to the scale of the section.
The original data, and the INFO file that has been produced from it,
contains a legal description down to the quarter- quarter, or about 40 acres.
No way has yet been identified to automatically create polygons matching
these sub-section legal descriptions on a large scale. More detailed
information must be hand-digitized. This limitation sets the scale of detail
for models that can be produced with this information.

Commercial County Atlas Mapping Project

Commercial county atlases are one of the most-used documents for
cultural resource studies concerned with 19th-century Euroamerican
archaeological sites. The heyday for these atlases in the Midwest was
approximately from the late 1860s to the second decade of the 20th century
(Conzen 1984a). At their best, these county atlases were produced through
careful field observation and are accurate representations of the rural
landscape (Conzen 1984b).

     The problem of moving information from atlases into the GIS
has been approached in 3 different ways, by three different projects (Figure
4). In one of these projects (SSC) the goal was to produce a predictive
model for 19th- century rural sites. In the other two projects, the goal was
primarily to serve as a document of potential site locations. However, the
ongoing atlas conversion project will yield data that could be used in a
modeling effort.

Superconducting Super Collider

     The first of the projects was part of Illinois' bid for constr uction
of the Superconducting Super Collider, or SSC. In this study predictive
models were developed for prehistoric and historic period archaeological
sites, and for paleobiological sites (McGimsey et al. 1986). The historic site
predictive model used both environmental and cultural variables (Figure 5).
The dependant cultural variable to be described by the other variables was
expressed as structure density per section (Schroeder 1986).

Structure density per section was produced simply by counting the number
of structures shown in a section on the county atlas. This value was then
introduced as an attribute in the modeling process by overlaying on the
other variables. The modeling process consisted of calculating the average
section-structure density for polygons representing all occurring
combinations of the other variables. These values were assigned low,
medium, or high site probability status, and mapped (Figure 6). Although
some interesting results were derived by this process, discussion of the
model is beyond the scope of this paper. 

     The SSC model suffered from several limitations. Chief among
these was the observation that point digitization of the structure locations
would have produced a more fine-tuned model. The generalization of the
data due to the expression as counts per section probably masked a great
deal of information, particularly information on where sites are not
located. This tendency was further aggravated by the averaging necessary
for the manipulation of the density data. (Schroeder 1986:65). 

In spite of the limitations, this method for moving information from atlas
to GIS had at least two strengths. First, coding the density of structures per
section was very quick-- perhaps taking 10 percent of the time it would
have taken for point digitization. This made it possible to gather
information on a 636 mile2 area in a matter of a couple of days. Second,
fewer assumptions of the accuracy of atlases were made than if structures
had been digitized as points. For example, we can probably assume that
most structures are mapped in the correct section, but not necessarily that
they are placed correctly within the section. For some purposes, simple
density coding may give adequate information.

The Knox County Project

In another project, the the problem of gathering information on structure
location was addressed by directly digitizing the structures from
photocopies (Wiant 1990). The structure locations were to be treated as
potential archaeological sites in the report, so point locations were
required. The sequence of steps was as follows:

1. Photocopies were produced for all the atlas maps of interest. Some were
from the original documents, others from microfilm. In every case, the
photocopy page was not large enough to contain the entire township.
Therefore, multiple copies were made and then pieced together.

2. New tic marks were created in the lambert coordinate system at the
corners of the townships. These tic marks are necessary to register the
digitized locations with the GIS.

3. The photocopies were oriented into the GIS coordinate system, and all
structures shown on map were digitized as points. 

4. The resulting coverages were plotted, and the plots were examined for
accuracy.

Again, this method has both strengths and weaknesses. Perhaps the greatest
strength is the absence of a hand-transfer step. This should limit the
number of human errors that are introduced to the process. The greatest
weakness was a lack of concordance with the digitized locations and
features already in the GIS coordinate system. In some cases this
registration problem resulted in structures digitized into the wrong section
or on the wrong side of a road, requiring the location to be changed (Nick
Klobuchar, personal communication 1992). Distortion could have been
introduced at several steps: the original production of the atlas, the
microfilming, the photocopying, the piecing of the photocopies, the
addition of registration points, or the digitization. 

State-Wide Atlas Project

Finally, a project is presently ongoing at the Illinois State Museum to
digitize the locations of cultural features from all 19th-century county
atlases for Illinois. This project is just beginning--the 1870s atlases from
15 counties have been completed since January of this year. If it is
continued to completion, a state-wide coverage containing locations of
potential Euroamerican historical archaeological sites will provide planners
with information useful on the same scale as the state-wide coverage of
prehistoric archaeological sites. The information can also be used to
upgrade the modeling methods used in the SSC project.

The present method has been developed in response to the previous studies.
The current method is to hand-transfer the feature locations to
township-and-section outline plots produced at the scale of the original
documents (usually 1:31,680, or 2 inches = 1 mile). At this scale, on the
plotter that we use, at the most we can produce a plot of 6 townships (2
high by 3 wide), or an area of approximately 216 square miles. Each county
requires 4 or more of these plots to be produced, depending on its size and
shape. The plots are simple in appearance, consisting of the township
outlines in heavy red lines, section outlines in thin black lines, and tic
marks corresponding with the corners 7.5 minute USGS topographic maps,
the standard registration points used in the Illinois GIS.

Since the plotter medium is translucent the plots can be overlain on the
original documents, and the feature locations can be traced. We are only
recording features outside of urban areas. Most of the features are recorded
as point locations, and most of these are farmhouses, however there have
also been schools, churches, mills, blacksmith shops, cheese factories, and
others. Also, municipal boundaries and the boundaries of cemeteries are
mapped and digitized as polygons.

The strengths of the hand-transfer method include:

1. All of the map interpretations are done before digitizing. When features
are digitized directly from photocopies considerable time can be spent at
the digitizer while making decisions about points. In an institution that
makes extensive use of GIS this can cause bottlenecks at the digitizer.

2. This method minimizes distortion due to inaccuracies in the atlases. This
is because the overlay plot can be centered on each section, minimizing
scale or projection discrepancies in the atlas.

3. If scale or other discrepancies exist, they are seen early in the process.

Weaknesses in this method include:

1. The method is only feasible for atlases that can be obtained as the
original document. A large percentage of the 19th-century atlases for
Illinois are easily obtainable only on microfilm. Another method will have
to be used for these.

2. Although transfer errors are controlled by producing plots in a format
that can be overlain on the atlases, such errors are still possible.

3. Occasionally, a plot is produced that does not contain the required 4
registration points. This occurs when the location of townships and USGS
maps do not coincide. In these cases tic marks must be added to the
coverage, possibly introducing distortion.

4. There are areas of Illinois that were not part of the rectangular survey.
Primarily these are French land grants. These areas are difficult to register
to today's maps.

Summary

In this paper I have covered the methods used in four projects at the Illinois
State Museum to integrate 19th-century documents and a GIS. Problems
involved in these or similar projects include the general problems involved
in transfer of information from paper to electronic, and text to graphic
formats. Additional problems may arise because of the age of the
documents. In particular, the documents may not be readable, and maps
may be drawn in unusual projections or contain elements that no longer
exist.

The land records project is an example of translating text information to a
GIS. The most time-consuming part of the project, that of moving
information from paper to electronic format, was done by the Illinois State
Archives. This was a long-term project, lasting 9 years. It required people
who were familiar with the documents, and who had practice in reading the
handwriting of the 19th century. Integrating the text information and the
GIS was made possible by the rational basis of the legal location method of
land description. Areas of the United States settled before the development
of the rectangular survey would be difficult to work with. A negative point
is the present scale limitation to the section level.

Three different ways of moving information from 19th-century county
atlases to the GIS have been tried at the Illinois State Museum. These are:
simple coding of counts of structures per section; directly digitizing from
photocopies of the atlases; and hand-transfer of feature locations to outline
plots. Although there are both strengths and weaknesses to each of these
methods, the third method appears to be the most satisfactory at this time.

The choice of method to move information from documents to GIS is
partially dependent on the nature of the documents. But it is also dependent
on the goal of the particular project. If the goal is to produce a graphic
inventory of cultural resources, then perhaps little information beyond
location need be coded. If the goal is to produce a detailed model of some
cultural process, then perhaps a great deal of information beyond location
will be coded. When working on a large scale, the data-gathering step will
probably be the longest part of the project. Therefore, a great deal of
thought should be put into how the information transfer is accomplished.

References Cited

Conzen, Michael
1984a The County Landownership Map in America, Its Commercial
Development and Social Transformation 1814-1939. Imago Mundi
36:9-31.

1984b Maps for the Masses: Alfred T. Andreas and the Midwestern County
Atlas Map Trade. Chicago History 13(1):47- 63.

Gaynon, David
1975 Evolution and History of the Illinois Land Records. Newsletter of the
Illinois State Archives, 1(3):4-8.

Hart, John F.
1974 The Spread of the Frontier and the Growth of Population. Geoscience
and Man, vol. V, pp. 73-81.

Illinois State Archives
1984 Public Domain Computer Conversion Project. From data in record
group 952. Illinois State Archives, Springfield.

Kremm, Thomas
1975 Land Records Data Project. Newsletter of the Illinois State Archives,
1(3):8-9.

McEntyre, John G.
1978 Land Survey Systems. John Wiley & Sons, New York.

Matousek, Ladislav
1971 The Beginning of Illinois Surveys. Illinois Libraries 53(1):23-44.

Schroeder, Erich K.
1991 The Illinois Public Domain Land Sales and the Settlement of Central
Illinois. In Landscape, Architecture, and Artifacts: Historical Archaeology
of Nineteenth-Century Illinois, edited by Erich Schroeder, pp. 22-38.

1986 A Model of Historic (ca. 1860) Site Location Probability. In Siting
the Superconducting Super Collider in Illinois, by C.R. McGimsey, M.A.
Graham, E.K. Schroeder, R.W. Graham, M.D. Wiant, and R. Druhot, pp.
49-65. Illinois State Museum, Springfield, Illinois.

Smith, Jane F.
1975 Settlement on the Public Domain as Reflected in Federal Records:
Suggested Research Approaches. In Pattern and Process: Research in
Historical Geography, edited by Ralph E. Ephramberg, pp. 290-304.
Howard University Press, Washington D.C.

Wagner, Mark J.
1991 Squatters, Speculators, Improvers, and Settlers: An Examination of
Frontier Settlement (1811-1830) in Marion County, Illinois. In Landscape,
Architecture, and Artifacts: Historical Archaeology of Nineteenth-Century
Illinois, edited by Erich Schroeder, pp. 39-74. 

Wiant, Michael D.
1990 Cultural Resources. In Illinois Land Report: Salem Township of Knox
County, Draft. Illinois Department of Energy and Natural Resources,
Springfield.



Erich Schroeder (erich@museum.state.il.us)