257
teractive system support (quick response, simple command language, feed
back) so that he can try out various solutions and compare them easily.
The different methods of defining class intervals are well documented in
the literature (e.g., JENKS and COULSON 1963, KREBS et al. 1977) and
will not be discussed any further here. The obvious task when designing
an interactive mapping system is to incorporate the various standard so
lutions as options.
The symbol code assignment operation, on the other hand, needs to be
looked at in some detail. In particular, we are interested in the pro
duction of T6(c,g) when the output device is a color CRT as discussed
in the previous section. The operation then consists of selecting a par
ticular color tone for each class. A color CRT functions according to
the RGB (red-green-blue) principle. Thereby each of the three primary
colors can be generated in a number of discrete intensity levels from
zero (black) to maximum. The additive mixture of these primaries in va
rious proportions produces intermediate color tones. Mixing them with
identical intensities results in neutral gray tones. Geometrically, the
RGB system can be visualized as a 3“dimensional discrete coordinate sy
stem, whereby the space of realizable colors forms a cube (LANG 1978,
GRAF 1980) (see Fig.2).
Fig.2: RGB color cube with intensity axis (i) from black (0) to white
(W), spectral periphery with changing hues (H) from red (R) over
yellow (Y), green (G), cyan (C), blue (B) and magenta (M) back
to red, and three different intensity planes: il = P, i2 = 3/2P,
i3 = 2P, where P is the maximum intensity level of a primary.
For univariate mappings it would seem to make sense at first sight to
choose systematic color scales by algorithmic means (SIBERT 1980, STEI
NER 198lb). The simplest procedure would be to intersect the color cube
with a straight line and to select from the color tones encountered
along such a line a systematic sample depending on the required number
of classes. It is found, however, that nominally equal level differen
ces of the primaries may not be perceived as equal contrasts by the hu
man eye. Consequently, the user must be provided with an interactive
