496 
TIGER Polygons 
Although the TIGER data base is not a GIS in the 
traditional context of such systems, the way in which the 
Census Bureau collects and tabulates data also is by 
“polygon.” The geographic structure of the 1990 census, 
as documented in the TIGER data base, provides 
machine-usable geographic units (polygons) similar to 
the ones used by the people collecting and studying soils 
or land use/land cover information. More importantly, 
in creating its “people polygons,” the Census Bureau 
routinely uses as the boundaries for its geographic 
entities the same earth surface features most people 
appear to want — and that many already include — in 
their GIS: streets, roads, streams, railroads, govern 
mental unit boundaries, and so forth. Thus, the TIGER 
data base is a very valuable source of information that 
can form a critical component of a GIS (Marx, 1988). 
To extend the notion of people polygons to the context of 
putting people in a GIS, I have coined a new term - 
censels. Each “censel” (people polygon) has a string of 
data associated with it in the Census Bureau’s sum 
mary tape files — just as a pixel does in a satellite 
transmission (Marx, 1990a; U.S. Bureau of the Census, 
1989a). Further, each censel provides a different level of 
resolution depending on the nature of the geographic 
entity that it documents. (See Figure 1) The following 
paragraphs highlight the entities used most frequently 
for general purpose data analysis. 
• At the coarsest level, the Census Bureau provides 
data for one large censel — the United States; censels 
comprising the 4 regions and 9 divisions of the United 
States, each composed of groups of states; and up to 57 
component censels comprising the 50 states and statis 
tically equivalent entities in which the Census Bureau 
conducts or assists with the several censuses and statis 
tically based sample surveys that are the primary 
mission of this Federal agency. This level of resolution 
is useful primarily for studies on a global scale involving 
nations and their first-order subdivisions. 
• There are more than 3,200 censels comprising the 
first-order divisions of the states; these entities generally 
are called counties, but also include a number of statis 
tically equivalent areas. They provide fairly coarse 
resolution of the Nation’s people and housing unit 
characteristics, usually for relatively small portions of 
the earth’s surface. 
• There are more than 60,000 censels comprising the 
units of local government and statistically equivalent 
entities subdividing those 3,200-plus counties - town 
ships, cities, villages, census designated places, and so 
forth. They provide nearly a twentyfold increase in the 
number of polygons available for use in a GIS, but gener 
ally still offer fairly coarse resolution. On the plus side, 
demographic data are available for all these govern 
mental entities and economic data are available for' 
about 20 percent of the most populous entities. • 
• There are more than 60,000 censels called census 
tracts and block numbering areas (BNAs) that subdivide 
the United States on a fairly uniform population basis for 
the 1990 census -- a significant increase over the number 
of these entities delineated for the 1980 census because 
now these entities cover the entire United States and its 
possessions. Whole census tracts and BNAs contain, on 
the average, about 4,000 people, with a range from about 
2,500 to more than 8,000. In the more populous govern 
mental entities, the census tracts and BNAs provide a 
finer-grained resolution of the reported statistical data, 
letting data analysts see variations in characteristics 
that might occur in different parts of the entity. In the 
less populous governmental entities, and in cases where 
entity boundaries do not follow easily recognized map 
features, the census tracts and BNAs tend to include all 
or parts of several governmental entities; in these cases, 
the Census Bureau provides separately tabulated data 
for each governmental entity/census tract-BNA portion, 
creating even more (141,100 E ) censels for data analysis. 
• The census tracts and BNAs are further subdivided 
for purposes of data tabulation into nearly 229,000 
censels called block groups. (For the 1980 census, some 
of these tabulation entities were called enumeration 
districts, or EDs; in combination, the 1980 census EDs 
and block groups covered the entire United States.) The 
block groups further segment the governmental entities 
for purposes of data presentation, providing a relatively 
uniformly populated “grid” across the United States in 
which the censels (353,000 E ) average just over 700 people. 
At this level, a GIS user has access to the full range of 
decennial census data tabulations - those data items 
collected about all people and housing units in the 
United States and those data items collected about only a 
sample of these people and housing units. 
• Finally, for the 1990 census, the Census Bureau 
assigned block numbers nationwide to the map feature- 
bounded polygons nesting within the block groups. The 
result is slightly more than 7 million censels called 
census blocks - people polygons - for which the Census 
Bureau will tabulate data following the 1990 census. As 
with the census tracts and BNAs, this represents a huge 
increase in the number of fine-grained entities over the 
1980 census when block-level data were available pri 
marily for the urban cores of metropolitan areas. 
Although census blocks vary widely in area and 
population, these “millions and millions” of blocks 
provide a very fine-grained censel resolution to the 
demographic data sets available from the 1990 census for 
the items asked about all people and housing units. 
The TIGER data base links the codes identifying the 
Census Bureau’s geographic entities (censels) directly 
with the underlying network of road, railroad, and 
hydrographic features, as well as the governmental unit 
and other tabulation entity boundaries. In a GIS envi 
ronment, a data analyst using the TIGER data base can 
examine the data sets that flow from the decennial, 
economic, and agriculture censuses of the United States 
along with many other locally available, geographically 
distributed data sets -- soil categories, hazardous waste 
sites, water quality, land use/land cover, and so forth. 
At the same time, the data analyst can examine these 
data sets in the context of the governments responsible 
for managing an area, in the context of the char 
acteristics of the people who occupy the land, in the 
context of those people’s homes and farms (U.S. Bureau 
of the Census, 1990a), and in the context of the 
businesses and industrial activities that support the 
economy (Carbaugh and Marx, 1990). 
AVAILABILITY 
The Census Bureau completed release of the first public 
product from the TIGER data base - a series of extracts 
called the Prototype TIGER/Line files - in April 1989. 
These files gave anxious data users an opportunity to 
examine the content of this new product and provide the 
Census Bureau with suggestions for improving it. The 
process worked well, and the Census Bureau completed 
release of a revised and updated version — the Precensus 
TIGER/Line files - in February 1990; they contain more 
than 19 gigabytes of geographic information that can be 
used in a GIS. (See Figure 2) 
Current Files 
The Precensus TIGER/Line files (U.S. Bureau of the 
Census, 1989d) generally correspond with the 1990 
Precensus Maps that the Census Bureau sent to all