Gregory has made a careful study of the glands on peach-leaves.[164] We publish here the most important facts he brings out. "In a large number of cases the glands are stable and can be safely used to aid in the identification of certain varieties. There are also varieties in which the glands are exceptionally unstable, being on the border line between the two types--reniform and globose--and having what might be termed mixed glands. These mixed glands are of two kinds: one in which the majority of the glands are reniform, with some globose intermingled; the other in which the globose form predominates. It would be quite possible, as Carriere (1867) suggests, to distinguish a third type of glands--the mixed type. It is important that leaves should be chosen from healthy branches on bearing trees. It is also best to obtain a large number of leaves or to examine the tree carefully before making the final selection of leaves. Mature leaves are best because their glands are full-sized and correctly shaped, while on young leaves the form of the glands is usually obscure. This is particularly true of the reniform glands. On the other hand, old, partly decayed, globose glands frequently have much the appearance of reniform glands. The structure of the glands shows that they are true glands, having an upper layer of long, rectangular, secretory cells that produce a sweet substance, the function of which is not apparent. After the glands have ceased secreting they begin to decay, becoming brown on the upper surface and slowly disappearing until almost nothing is left. This decaying is a very complicated process, being preceded in every case by a suberization and thickening of the cell walls. The spines of the leaf are very similar to the glands in structure, having the same upper layer of long cells, but with much more heavily cutinized walls. A study of the transitional forms indicates that the glands are merely modified leaf spines. The leaves with reniform glands are apparently the highest type and the glandless leaves the lowest, with the transition through the globose type. In support of this view is the fact that whenever typically glandless leaves become possessed of glands they are always of the globose type. The serrations of the glandless leaves are very strikingly different from those on a leaf with glands. The former leaves are deeply and doubly serrate, while the margins of the latter are always single and crenate. Almost invariably, when glands develop on a normally glandless leaf, the serrations are transformed to crenations, indicating that there is a very close correlation between the glands and the crenations on the edges of the leaves."