Figure 5.      The need for geographical modelling is illustrated in these hypothetical examples: (a) malaria surveys are unevenly distributed within and between two bordering countries for various reasons; the point size reflects prevalence; (b) the actual malaria endemicity can not be derived for the whole area from the available parasite ratios alone.  The effect of environmental gradients (elevation contours) and features (large rivers), as well as the effect of epidemics and low-level variability on parasite ratios is illustrated.