In the wider tubes, the work proceeds quite differently. At the moment when the Osmia disgorges her honey and especially at the moment when, with her hind-tarsi, she rubs the pollen-dust from her ventral brush, she needs a narrow aperture, just big enough to allow of her passage. I imagine that in a straitened gallery the rubbing of her whole body against the sides gives the harvester a support for her brushing-work. In a spacious cylinder this support fails her; and the Osmia starts with creating one for herself, which she does by narrowing the channel. Whether it be to facilitate the storing of the victuals or for any other reason, the fact remains that the Osmia housed in a wide tube begins with the partitioning. Her division is made by a dab of clay placed at right angles to the axis of the cylinder, at a distance from the bottom determined by the ordinary length of a cell. The wad is not a complete round; it is more crescent-shaped, leaving a circular space between it and one side of the tube. Fresh layers are swiftly added to the dab of clay; and soon the tube is divided by a partition which has a circular opening at the side of it, a sort of dog-hole through which the Osmia will proceed to knead the Bee-bread. When the victualling is finished and the egg laid upon the heap, the whole is closed and the filled-up partition becomes the bottom of the next cell. Then the same method is repeated, that is to say, in front of the just completed ceiling a second partition is built, again with a side-passage, which is stouter, owing to its distance from the centre, and better able to withstand the numerous comings and goings of the housewife than a central orifice, deprived of the direct support of the wall, could hope to be. When this partition is ready, the provisioning of the second cell is effected; and so on until the wide cylinder is completely stocked. The building of this preliminary party-wall, with a narrow, round dog-hole, for a chamber to which the victuals will not be brought until later is not restricted to the Three-horned Osmia; it is also frequently found in the case of the Horned Osmia and of Latreille's Osmia. Nothing could be prettier than the work of the last-named, who goes to the plants for her material and fashions a delicate sheet in which she cuts a graceful arch. The Chinaman partitions his house with paper screens; Latreille's Osmia divides hers with disks of thin green cardboard perforated with a serving-hatch which remains until the room is completely furnished. When we have no glass houses at our disposal, we can see these little architectural refinements in the reeds of the hurdles, if we open them at the right season.