Number 29, Winter 1998 21

points in their symbol designs, which assists accu rate percentage esti-
mates. The symbol form in Figure 8b is known to statisticians as a framed
rectangle (Cleveland 1985, p. 208; Dunn 1987, 1988), and Monmonier
(1993, p. 65) describes its usefulness in representing both absolute magni-
tudes and intensities. Dunn (1987) also suggests scaling the widths of the
rectangles to total data values (to total disbursements for Figure 8b). In a
conversation about Figure 8, MacEachren mentioned a successful student
project from an ACSM design competition th at divided graduated squares
into ten-by-ten grids such that each cell represented one percent, and each
row ten percent, of the whole for all symbol sizes (rather than using grid
cells of a fixed size; a hybrid of Figure 8c and 2e).

The work of students new to cartography has been an inspiration. With CONCLUSION
modern computing at their disposal and freedom from the conventions of
traditional thematic mapping, students are producing a wide variety of
creative symbols for mapping data usually symboli zed by graduated
circles or squares. For example, a recent student assignment with eco-
nomic data produced creative ideas ranging from a propo rtionall y scaled
\ilonopoly game character (with pockets turned out or clutching money
bags) to weight scales drawn tipped to illustrate balances be tween dollar
amounts (by Elliott Westerman and Erika Bozza respectively). Are these
effective symbols? Wi th so little resea rch into multivariate representations,
one can not say, but this type of experime ntation can be encouraged until
there is e\·idence to the contrarv.
"... there are wide-ranging
The intent of this paper has been to review some of the major topics on
w hi ch grad uated symbol research has been conducted and to demonstrate possibilities for applyi11g
the wide variety of ways to map univariate data as well as biva riate creative, eye-catching symbol
comparison and proportional relationships. The survey \·v as not exhaus- designs to summarize mid
tive, but if it h as inspired recall of other symbol forms that have been synthesize quantitative spatial
missed, then it has been successful in provoki ng consideration o f the wide distributions."
range of possibilities available •vith point symbo ls. Modern computer
mapping allows mapmakers greater flexibility in desig ning creative
graduated point symbols. This increase in fl exibility increases the impor-
tance of research examining symbol design issues. As this article has
illus tra ted, there are wide-ranging possibilities for applying creative, eye-
ca tching symbol designs to summarize and synthesize quantitative spatial
dis tributions.

Comments on the manuscript by Judy Olson and by an anony mous ACKl\'OWLEDGMEl\'TS

reviewer are acknowledged with appreciation. This research was funded
by a George H. Deike Jr. Research Grant from the College of Earth and
Mineral Sciences at Penn State.

Abler, R.F. 1987. The >Jational Science Foundation National Center for REFEREl\'CES
Geographic Information and Analysis. 1Htcmatio11a/ /011mal of Geograpl1ical
/11for111atiou Syste111s 1(4):303-326.

Balogun, 0.\/. 1978. The decagraph: A substitute for the pie graph? T/ie
Cartograplric journal 15:48-85.

Bernes, C., and C. Grundsten (eds.). 1992. Tire Natio11a/ A tins uf Swede11: The
E1wiro11111e11t, vol. 4. Stockholm: SNA.