ABSTRACT. Migration maps represent patterns of geographical movement by arrows or bands between places, using information arriving in "from-to" tables. In the most interesting cases the tables are of large size, suggesting that computer assistance would be useful in the preparation of the maps. A computer program prepared for this purpose shows that graphical representation is feasible for tables as large as fifty by fifty, and possibly larger. The program contains options for alternate forms of movement depiction, and rules are suggested for the parsing of migration tables prior to the cartographic display, without loss of spatial resolution. KEY WORDS: Computer cartography, geographic movement, thematic maps.Maps that show patterns of geographical movement function as particularly effective illustrative and research tools. Like most graphical aids their value increases in direct proportion to the complexity of the data. The general style of these maps has not changed much since the time of Harness, Belpaire, or Minard in the last century (Brinton 1914;Robinson 1955Robinson , 1967Robinson , 1982Tufte 1983). The areas between which a migration, or other movement, occurs are connected by a "band" whose width represents the quantity moved. What is new today is that computers can aid in their construction.As a generic case for all movement mapping problems it is assumed here that the data take the form of an N-squared table of geographical interactions. These "from-to" tables indicate all of the N by N possible movements between N geographical areas during some interval of time. When the movements are between only a few of these possible pairs of places, a small subset of the N by N pairs for example, then complete from-to tables do not provide a particularly convenient formulation of the problem. These special cases might include movement of people on a highway network, where only the link volumes are known. Here the detailed path of the movement may be shown but the individual nodes of such a network are directly connected to only a few, adjacent, nodes. Most of the entries in a complete node-to-node table for such a network would be empty. Similarly, travel from many agricultural zones to a few urban centers would result in a from-to table with a very special, sparse, structure. Conceptually it makes sense to think of these cases as large but incomplete N by N from-to tables but from a data storage or data entry point of view this is not very practical, even though the types of displays described here are still relevant.For the general N by N situation the cartographic challenge is when N is greater than ten, with over one hundred possible flows. Now it begins to pay to use a computer. Tables of smaller size usually have such a coarse spatial resolution as to be geographically uninteresting and the depiction of the few flows not a difficult problem. Thus the U.S. Bureau of the Census often shows migration between only four regions -Northeast, North Central, South, and West, but actually collects data by counties...
