§ 17. Let X, Fig. 54, represent a shoot of any opposite-leaved tree. The mode in which it will grow into a tree depends, mainly, on its disposition to lose the leader or a lateral shoot. If it keeps the leader, but drops the lateral, it takes the form A, and next year by a repetition of the process, B. But if it keeps the laterals, and drops the leader, it becomes first, C and next year, D. The form A is almost universal in spiral or alternate trees; and it is especially to be noted as bringing about this result, that in any given forking, one bough always goes on in its own direct course, and the other leaves it softly; they do not separate as if one was repelled from the other. Thus in Fig. 55, a perfect and nearly symmetrical piece of ramification, by Turner (lowest bough but one in the tree on the left in the "Château of La belle Gabrielle"), the leading bough, going on in its own curve, throws off, first, a bough to the right, then one to the left, then two small ones to the right, and proceeds itself, hidden by leaves, to form the farthest upper point of the branch. The lower secondary bough--the first thrown off--proceeds in its own curve, branching first to the left, then to the right. The upper bough proceeds in the same way, throwing off first to left, then to right. And this is the commonest and most graceful structure. But if the tree loses the leader, as at C, Fig. 54 (and many opposite trees have a trick of doing so), a very curious result is arrived at, which I will give in a geometrical form. § 18. The number of branches which die, so as to leave the main stem bare, is always greatest low down, or near the interior of the tree. It follows that the lengths of stem which do not fork diminish gradually to the extremities, in a fixed proportion. This is a general law. Assume, for example's sake, the stem to separate always into two branches, at an equal angle, and that each branch is three quarters of the length of the preceding one. Diminish their thickness in proportion, and carry out the figure any extent you like. In Plate 56, opposite, Fig. 1, you have it at its ninth branch; in which I wish you to notice, first, the delicate curve formed by every complete line of the branches (compare Vol. IV. Fig. 91); and, secondly, the very curious result of the top of the tree being a broad flat line, which passes at an angle into lateral shorter lines, and so down to the extremities. It is this property which renders the contours of tops of trees so intensely difficult to draw rightly, without making their curves too smooth and insipid. Observe, also, that the great weight of the foliage being thrown on the outside of each main fork, the tendency of forked trees is very often to droop and diminish the bough on one side, and erect the other into a principal mass.[1]