So soon as a gully is started, the water flowing into it from either side wears back the slopes. The least inequality of slope, or the least variation in the character of the material, is sufficient to make the lateral erosion unequal at different points, and unequal erosion in the slopes results in the development of tributary gullies. The oldest tributaries may be nearly as old as the main which they join, and from which they developed, for the possibilities of unequal side erosion exist as soon as a gully is opened. While the main gully is developing into a ravine, and the ravine into a valley, the tributary gullies are likewise developing into maturer stages. Tributary to a young valley, therefore, there may be gullies near its head, ravines in its middle course, and small valleys along its oldest portion. It is not to be understood, however, that the oldest tributaries are necessarily the largest, for because of more favorable conditions for growth the younger tributaries often outstrip the older. [Illustration: +Fig.+ 62.—Diagram to illustrate the oblique position of a tributary gully at its inception, and its later normal change of direction.] The position of tributaries with reference to their mains is worthy of note. The water flowing down a slope follows the line of steepest descent. A gully is usually wider at its lower end, and narrower at its upper. Wherever this is true the line of steepest descent down its side is not a line perpendicular to its axis, but a line slightly oblique to it (_ef_, Fig. 62), and oblique in such a direction that it meets the axis with an obtuse angle below and an acute angle above. It is in the direction corresponding to this line that tributary gullies tend to develop. Thus at the inception of its history a tributary gully is likely to join its main with an angle slightly acute on the up-stream side. If the tributary did not begin until after its main was farther advanced this tendency would be less and less pronounced. Inequalities of material or slope would often counteract this tendency, which, at best, would cause the courses of tributaries to depart but little from perpendicularity to their mains. After the head of a tributary has worked back from the immediate slope of its main every condition which determines the course of a gully is likely to affect it, and it is by no means certain that it will continue to lengthen in the direction in which it started. Since the general slope of the surface into which the tributary works is likely to be seaward, more water is likely to enter from the landward than from the seaward side of its head, so that, except where there are notable irregularities of slope, its tendency will be to turn more and more toward the direction of its main (_efg_, Fig. 62). In depth the tributary is always limited by its main. The principles which determine the length and width of a main valley determine also the length and width of a tributary (see p. 67 et seq.).