This is what the ants do for millions of seeds, for the instances in which a few seeds appear to have sprouted within the nest in defiance of the ants, are very rare and wholly exceptional; and when after prolonged wet weather germinated seeds are seen outside the nest, it will usually be found that these have the little root cut off, and are eventually carried back into the nest and used as food. By a fortunate chance I have been able to prove that the seeds will germinate in an undisturbed granary when the ants are prevented from obtaining access to it; and this goes to show not only that the structure and nature of the granary chamber is not sufficient of itself to prevent germination, but also that the presence of the ants is essential to secure the dormant condition of the seeds. I discovered in two places portions of distinct nests of _Atta structor_ which had been isolated owing to the destruction of the terrace-wall behind which they lay, and there the granaries were filled up and literally choked with growing seeds, though the earth in which they lay completely enclosed and concealed them, until by chance I laid them bare! In one case I knew that the destruction of the wall had only taken place ten days before, so that the seeds had sprouted in this interval. My experiments also tend to confirm this, and to favour the belief that the non-germination of the seeds is due to some direct influence voluntarily exercised by the ants, and not merely to the conditions found in the nest, or to acid vapours which in certain cases are given off by the ants themselves. In order to put this latter point to the test of experiment, I confined about a hundred harvesting ants (_A. structor_), with their queen and several larvæ, in a glass test-tube eight inches long and one inch in diameter, closed with a cork and filled up to within about an inch of the cork with damp sandy soil, most of which was taken from the ants' nest. I added six peas, six cress and six millet, and then kept the tube tightly corked for nine days, only once removing the cork for a few seconds in order to sprinkle a little water on the ants, which were evidently in need of it. On the ninth day I turned out the contents of the tube and found that all the peas, millet and cress, had germinated and were growing strongly. One of the cress, however, had had its root, which lay across the gallery constructed by the ants, gnawed off; four clover seeds, which had come with the soil taken from the nest, and which had formed part of the ants' stores, had germinated also. Here the small quantity of air contained in the test-tube must certainly have become saturated with any vapour which the ants may be supposed to give off, and we cannot therefore accept this as the cause of the dormant condition of the granary seeds.