I cannot close this chapter without a few words on the subject of the neat devices made by the Portuguese out of canes or bamboo, for training plants. In some instances it may be overdone, and one cannot always admire rose-trees trained on to bamboo frames in the shape of fans, crosses, or even umbrellas ; but the little arched fences as a support to lower-growing plants are used with very good effect. I have copied the idea in England with some success for training ivy-leaved geraniums in large pots or tubs, by planting four rather stout bamboos or canes, two feet or more in height, in the pots, then slipping four pieces of split cane into the hollow ends, and either forming four arches, by inserting each end of the split length into the hollow, or else a pagoda-like effect can be made by taking the split canes into the middle, and then slipping all four ends through a hollow piece of cane a couple of inches long. Side arches can be made in any number, according to the requirements of the plant or the fancy of the gardener, by making incisions in the stout bamboos at any distance from the ground, and inserting the ends of the split canes. Old carnation plants, or seedlings which bear many flower-stems, may be very successfully and neatly supported in this way.

Another contrivance  for the  increase  of their rose-trees struck me as original, and worth men­tioning, and possibly imitating, by those who garden in a  subtropical climate—this is their system of layering rose-branches.    My idea of layering carna­tions, shrubs, or any other plants, had always been to cut the plant at a joint, and peg it firmly into the ground, covering with a few inches of fine soil; but the Madeira gardeners adopt a different system, anyway, with regard to their roses.     The branch for layering is not  chosen near the ground, but often at a height of from two to four feet.    The chosen branch is passed through the hole at the bottom  of a flower-pot, or a  box with  a good-sized  hole in it answers the same purpose:   the pot or box is then supported at the necessary height on a tripod of sticks or bamboos.    The branch has an upward slit made in the ordinary way, and the pot is then filled with soil. In two or three months' time, I was assured, the branch would be well rooted and ready to be transplanted to its fresh quarters. It seemed a simple method of increasing rose-trees, which, as a rule, in climates like those of Madeira, flourish much better when grown on their own roots than grafted on to a foreign stock. The same system appears to answer admirably for the increase of shrubs and even trees, and is exten­sively adopted for creepers, especially bougainvilleas, which do not strike readily from cuttings ; so it is no uncommon sight to see pots lodging among the branches of trees, with a layered branch ready to form a new tree.