This is the format of the 52 very high resolution boundary description
files. There is one file for each state, one for Washington D.C., and
one for the 48 contiguous states as a whole. The names of the files
are boundaryAK.dat, boundaryAL.dat, etc up to boundaryWY.dat for the
individual states, plus boundaryDC.dat and boundaryUS.dat.

The state files give not only the outline of the main land area, but 
also for all islands within state waters, so a large number of the
files contain multiple outlines.

The files do not contain text; all data is in a binary format.
These are the C++ or C structs needed to read the file:

struct Descriptor
{ float area;
  int number; };

struct Region
{ float min_longitude;
  float min_latitude;
  float max_longitude;
  float max_latitude; };

struct Point
{ float longitude;
  float latitude; };

All of the floating point numbers are in the four byte "float" format,
not the eight byte "double" format, and represent longitudes (x) and
latitudes (y) in degrees. For the United States (which is the only
area covered by these files), all longitudes are negative, and all
latitudes are positive.

The first 8 bytes in each file is a Descriptor. It gives the area of
the entire state in square degrees, and the number of individual outlines
that follow.

The next 16 bytes in each file describe the smallest rectangle that
contains all of the data points. This may be used to quickly determine
whether or not any of this state appears in a given area of consideration.

The remainder of each file, after those first 24 bytes, is just a
series of outlines.

Each outline begins with its own individual 8 byte Descriptor, giving the
area (in square degrees) of the outline that follows, and the number
of points in that outline. All outlines are "joined up", and make a properly
closed shape. The area may be used to quickly decide whether a particular
outline is worth considering or not. For a map covering a large area, an
island of less than 0.01 square degrees might not be visible for example.

All of the outlines in each file appear in descending order of area, so
once you find an outline that is too small to be worth displaying, you
can be certain that all subseequent outlines in that file will be just
as small.

After the descriptor for an outline, which reports the number of points
that are to follow, those points immediately follow.

After the points that make up an outline, the Descriptor for the
next outline follows immediately, unless there are no more outlines
in the file, in which case the file ends immediately.

Some state files contain only a single outline, some contain a very
large number of outlines indeed. The US file contains only one outline;
all islands must be found in their individual state files.

The outlines are very carefully coordinated to guarantee consistency:
when two states share a common boundary, exactly the same points appear
in both files, without a single bit of difference.