A new theory of migration in butterflies is outlined and present concepts are examined. During the course of evolution many butterflies have become adapted in the larval stage to foodplants that occur in small and scattered localities, the distribution of which changes constantly. It is argued that whenever this happens selection might be expected to produce a butterfly which flies from one foodplant site to another. Further it is argued that while they were crossing areas devoid of foodplants selection would have favoured those individuals that flew at a constant angle to the sun. At first all angles to the sun would be represented equally in the population but each individual would pass on to its offspring a bias towards its own particular angle. It is suggested that the temperature gradient experienced by a butterfly dispersing in this way would constitute a marked selective pressure. This selective pressure would cause an increase in the number of individuals flying at certain angles and a decrease in the number flying at others. The effects of temperature on rate of development and fecundity were demonstrated for Pieris rapae and P. brassicae in the laboratory. The effects of seasonal and geographical temperature variations on these two species in the field were also demonstrated. Based on these results the relative selective advantage of each flight direction was calculated for different times of the year. As a result of these calculations it was predicted that P. rapae should fly at 159<latex>$^\circ$</latex> to the sun until 27 August, when it should fly at 0<latex>$^\circ$</latex>. For P. brassicae it was predicted that the first brood should fly at 159<latex>$^\circ$</latex> and the second brood at 339<latex>$^\circ$</latex>. Observations of flight direction of these two species from August of one year to October of the following year agreed well with these predictions. The observations of flight direction also showed that P. rapae, and probably also P. brassicae and Vanessa atalanta, were using the sun as the environmental clue by which they were orientating themselves. There was no compensation for the sun's movement during the day or season. Experiments showed that P. rapae is sensitive to photoperiod during the adult stage. It is by this means that the same individuals can change their flight direction from 159<latex>$^\circ$</latex> to 0<latex>$^\circ$</latex> at the most selectively advantageous time. A calculation based on the results of this investigation suggested that a return flight would be a selective disadvantage to both P. rapae and P. brassicae. Observation of these two species suggested that in neither does the southward movement function as a return flight, i.e. is equal in distance to the northward movement.