margin is the rostellum. This is formed of large cells, covered with a very delicate membrane. If this be touched with a bristle, it is almost instantly ruptured, and a small, very viscid drop of matter exudes. In withdrawing the bristle the pollinia are brought away with it. The anther is terminal (posterior), and has broad lateral projections. The pollinia are four in number, in two pairs, and in the form of plates. The flowers do not appear to secrete any nectar, but when the surface of the labellum is slightly punctured, a considerable amount of sweetish purple juice exudes, which is probably grateful to insects. From the shape of the flowers, it is necessary to cut them longitudinally to see the parts. Looking at the position of the anther and stigma, it appears to me almost impossible that self-fertilization can take place; at the same time it is somewhat difficult to suggest any satisfactory way in which an insect could accomplish either this or cross-fertilization. I presume that any insect entering the flower would have to back out again by the same way as it entered, and in doing so it would come in contact with the rostellum, and would remove the pollinia on its head. It is also probable that, in endeavouring to obtain from a second flower any of the sweet juices from the tissue at the base of the labellum, it would slightly advance its head, so as to bring the pollinia attached to it on to the stigma. Again, it is possible that self-fertilization might be secured by an insect thus getting the pollinia on its head, and then endeavouring to push its way down through the small lateral apertures. In doing so, it would almost certainly smear the stigma with pollen from the same flower, and I have sometimes been inclined to think that such did take place. At the same time, this would seem like putting an unnecessary difficulty in the way of what is usually a very simple process, and therefore no great value is to be attached to this idea.

For a time I could not understand why spiders frequented these flowers so much, but I soon found a sufficient cause. The only insects capable of removing pollen which were found about the flowers were small Diptera—probably a species of Culex. In several cases these small flies had penetrated into the tube of the flower, and, in their eagerness after the sweet juices found there, brought their heads in contact with both rostellum and stigma, and partly owing to the viscidity of these parts, and partly to the narrowness and bending of the tube, were unable to withdraw backwards. In some flowers insects were thus found still alive, in others they were dead, while in many others only portions of them, such as legs, wings, etc., were left, the spiders having devoured the rest. In every case in which a captured insect was withdrawn from its trap, the pollinia were removed also, securely attached to the front of the head.

I closely examined 143 flowers, and found that in 47 the pollinia were

still in the anther cells; from 90 they had been removed, while in 6, dead or living insects were found glued to the stigma. Of the whole number examined, only a small proportion ultimately produced capsules.

The flowers of this genus will well repay examination.

(4.) Microtis porrifolia.

In the flowers of this species the column is protected by a broad, flat hood, formed by the posterior sepal and the two lateral petals. The lateral sepals are completely reflexed, and lie back against the ovary. The labellum is large and pendulous, hanging out from the front of the flower like a tongue. It is rectangular in shape, rather longer than broad, with the margin crimped and curled, and bearing three glandular projections on its surface. Two of these are situated together near the base, and enclose a small depression or pit. This, from its position and appearance, I take to be a nectary, but I was unable to detect any liquid in it. The third gland is formed by an irregular wart-like mass of cells, and is situated near the apex of the labellum. I have not investigated its functions, nor do I know how its presence can be accounted for. The column is very short, and stands almost square, this appearance being caused by the wings or auricles which stand up on each side. Beneath these is the hooded anther, enclosing four pollinia, which lie very loosely in their cells. They present the appearance of two masses, but each is composed of a large outer and a smaller inner sheet, of a reniform shape, united by their threads to a short caudicle. In front of and somewhat below them is the viscid rostellum, towards the apex of which a minute white point is visible, which marks their point of attachment. The rostellum projects considerably outwards, so that the stigmatic surface is placed in a recess. The slightest touch on the viscid disc suffices to bring away one or both pollinia, the matter being excessively viscid. An insect alighting on the rostellum, and advancing its head to examine the glands at its base, would be certain to touch the rostellum and bring away the pollinia. These fall slightly by their own weight, so that on entering a second flower, they would be in such a position on the front of the insect's head as to touch the stigma immediately under the rostellum. In the first spike examined by me, 83 flowers were fully opened, and all but the top one had their pollinia removed.

Even when not fertilized by insects however, these flowers are readily self-fertilized, and during the past season this appears to have been the case with the great majority. After a time, the pollinia appear withered and brown, and somewhat dragged forward from their anther cells, while the ovary begins to enlarge, showing that pollination has taken place. If such flowers are examined carefully, it will be found that the pollen grains have emitted a great mass of tubes, which penetrate the upper margin of

the stigma, thus ensuring fertilization. I found this to be the case in several hundred flowers which I examined. The position of the labellum on the underside of this flower is caused by the usual twisting of the pedicel or ovary, which is so common in many orchids. But in young buds the posterior sepal is lowest and placed on the side farthest from the axis of the spike; and it is during the gradual maturing of the flower that the twisting takes place, so that, by the time it opens, the labellum and posterior sepal have changed places.

This species, as might be expected from its facilities for reproduction, is one of the commonest plants of the class.

(5.) Caladenia bifolia.

Chiloglottis traversii, Müeller.

This is a most abundant orchid in upland districts as an elevation of 1500 to 3000 feet. The flower is solitary on an erect scape, three to four inches in height. The upper sepal is obtuse, somewhat arched forward, and slightly keeled. The lateral sepals are placed under the labellum, and extend forward almost horizontally. The labellum is broad; on each side of the expanded portion is a yellow-coloured patch bearing two or three brownish spots, while extending from the middle to the base are two rows of yellow glands. The column is long and erect, slightly winged above, and bearing a terminal anther which encloses four pollinia. The stigma is rounded and slightly hollowed out, and is placed in close contiguity to the anther. The arrangement of the parts is so simple that an insect alighting on the labellum and advancing its head into the base of the flower could hardly fail to remove the pollinia; nor could one entering with pollen on its head fail to leave these on the stigma, for in withdrawing pollinia from a flower they are always slightly depressed by the cap of the anther. The pollen of this plant is very incoherent, and the lower surface of the stigma projects a little, so that I am inclined to think self-fertilization takes place in flowers which have not been visited by insects. The majority of the flowers appear to set good capsules, and flowers which I fertilized artificially, produced good full seed-vessels. I examined one sunny day twenty-two flowers growing in the open; of these only three had both pollinia removed; in one the pollinia were removed from one anther lobe; in five others the pollen masses appeared more or less disturbed, while in the remaining thirteen the anthers were untouched.

(6.) Pierostylis banksii.

The fertilization of the flowers of this genus has been so well described by Mr. Cheeseman, in the Trans. N.Z. Inst., Vol. V., p.352, that I cannot well add to it, but my observations on them more than over induce me to cousider that thero has been an unusual scarcity of insect life during the

past season. Out of all the flowers of the above species, and of P. graminea, examined, not one had the pollinia removed. The flowers are incapable of self-fertilization. Certain experiments made by me to test whether they were fertile with their own pollen were rendered useless by being conducted in the open, where the flowers were liable to be destroyed.

The rostellum of this orchid, when examined in bud, lies in front of and between the bases of the pollinia, but quite separate from them. At this early stage it consists of an oblong, pearly-white body, composed of large rounded cells, filled with granular fluid. The pollinia stand in a small hollow on the top of the column, and at this stage are attached only by a small posterior ligament at their base.

(7.) Chiloglottis cornuta.

In this species the flower is solitary, on a short scape, which lengthens after flowering, and is partly covered by an acute, sheathing bract. When fully developed, all the parts stand nearly erect, and thus leave no landing place for insects. The labellum is acutely trowel-shaped, with one broad central, and several narrow, lateral, longitudinal, purple glands. The column is curved back at the base, and then ascends in front of the upper sepal. The stigmatic surface is large, almost circular, quite flat and excessively viscid, there being no distinct rostellum. The anther is terminal, and encloses four plate-like pollinia, which are coherent, and are attached by their bases to the upper margin of the stigma (rostellum). Before the flower is open, and while yet almost sessile, and sheathed by the bract, the stigmatic surface becomes excessively viscid, and smears all the portion of the labellum immediately opposite to it. I could not ascertain how the pollen got on to the stigma, but in the few flowers I was enabled to examinc, all four pollinia were on the stigma, and the anther cells were empty.

From the position of the flower when the parts are ripe for pollination, viz., low down between the two leaves, from its inconspicuous greenish colour, and the fact that viscidity is strongest in the unopened flowers, I am of opinion that this species is exclusively adapted for self-fertilization. The subsequent lengthening of the scape is probably only to aid in the dispersion of the seed.

(8.) Lyperanthus antarciticus.

In this orchid the flowers are solitary, or two on a scape, partially covered by a relatively large concave bract, and of a green colour through-out. The posterior sepal is large and broad, arched forward, and covering the column like a hood. The labellum is flat, broadly ovate and acute, quite glabrous, with two lateral and four median ridges. The column is broad, somewhat arched forward, and terminated by the acute anther. The rostellum placed directly above the stigmatic chamber, impinges on the base