What is the source of this very large supply? It has been suggested that the latex may be manufactured in the leaves and pass down the tree to the wounded area. Now Hevea is a deciduous tree and in Ceylon drops the whole of its leaves in February or March. There may be an interval of nearly a month between the loss of the old leaves and the time when the young leaves become fully functional. Hence, if the leaves represented the chief source of the latex supply, we should expect the yield of latex to be reduced almost to nothing shortly after the fall of the leaves, if tapping be carried on without intermission. We shall presently see that there is some falling off in yield about this period, but that the reduction is not nearly so great as would be expected on the theory of the origin of the latex in the leaves. Moreover the composition of the latex in the leaves differs considerably from that in the stem.

During the experiment the tree was steadily increasing in girth, and new latex vessels were constantly being added to the bark by the activity of the cambium. The amount of bark tissue thus added in three years over the area actually tapped was nearly equal to the amount removed, but in other parts of the tree much less. The new latex vessels contained in the renewed bark are therefore only responsible for a comparatively small proportion of the latex obtained. The greater part of the latex still unaccounted for can only have been produced in one way, and that is by the active secretion of latex in existing laticiferous tissue. In fact we arrive at the important conclusion not by any means universally accepted that the laticiferous tissue of the bark is an organ for the manufacture of latex as well as for the storage of latex.

There is evidence that this process of manufacture is specially active in the immediate neighbourhood of the wounded area. In an experiment carried out on 29 trees of an average girth of 26 inches, the whole of the outer bark on one side was removed to a height of six feet by tapping on a six-V system for a period of only six months. Tapping was immediately continued by a similar system on the other half of the tree. The cuts were twelve inches apart, and, after the first day's paring had been carried out, each further cut except the lowest was apparently draining an almost isolated patch of bark, of an average area of 216 square inches. In this case the average thickness of the laticiferous bark was not more than a quarter of an inch, and on our previous assumption the average volume of the latex vessels contained in one of the isolated patches of bark, could not have been more than 5½ cubic inches. The yield in six months from the lowest cut but one was 55 cubic inches of latex, more than 10 of which were drawn off during the first month of tapping.

There must nevertheless be a considerable movement of latex from untapped bark towards the seat of tapping. It was calculated that the largest day's yield from the old tree previously described would have drawn off the whole of the latex from a distance of at least three inches from the tapped surface. As the tubes are never completely emptied, there was probably an actual flow from at least double this distance. In the case of this particular tree the flow continued for several hours, whereas in the majority of trees the cut surface becomes blocked by coagulated rubber in less than one hour after tapping. The structure of the laticiferous system shows that movement of the latex will take place more readily in the vertical direction, but that a lateral flow is also possible. In the older networks a gradual transference of latex may take place over an extensive area.

Seasonal variation. Some idea of the variations in the flow of latex at different times of year on the Western side of Ceylon may be obtained from the following table, which embodies the results of three years' continuous tapping of the 70 trees already described as having been tapped at different intervals by groups of ten. Tapping was carried out during the last seven months of 1908, so that by the beginning of 1909 the trees may be supposed to have been responding freely to the influence of tapping. By taking the average of three years we can eliminate to some extent the effect of different stages of tapping. This is necessary because, at least when the tapping intervals are short, the yield is greatest soon after a particular area is begun, and falls off when the bark on that area is becoming exhausted. Row I is omitted because it was only tapped for a few months during 1911.

Table XIV. Average Yields Per Tapping, 1909—1911 (Grammes)

It appears that on the average the yield per tapping is lowest in April and May. This yield rises continuously until December and then steadily falls off again. The increase from April to December on the average of these three years, and taking the average result from tapping at a number of different intervals, does not fall far short of fifty per cent.