Tangtsinia* S. C. Chen, a monotypic genus, is discovered in southeastern Szechuan in China. It possesses a rather ordinary monocotyledonous habit, namely with a short rhizome, non-thickend annual stem with scattered, spiral-arranged leaves and a terminal inflorescence. Its habit somewhat similar to that of some very primitive genera, viz. Apostasia, Tropidia, Cephalanthera and Selenipedilum, is of much morphological and phylogenetic interest. The primitive and significant floral features consist chiefly in the erect, hardly twisted flower with nearly regular perianth and a unilocular ovary, in the column composed of a terminal stigma and five small projections, in an erect anther with four naked pollinia, and in the absence of the rostellum. Of special interest is the occurrence of five small projections in the upper part of the column, and this is, however, a unique instance in the family, including the Apostasieae. Among the five projections three are larger and opposite to the petals respectively. Of these the two lateral ones bear a strong resemblance in texture to the two auricles in the Orchideae and some members of the Limodorinae, which P. Vermeulen considered not as staminodia, but as appendages of the single fertile stamen as usually seen in Allium. In the case of Tangtsinia, there exists the third projection which, being situated in the front of the column and thus opposite to the median petal (lip), shows no difference both in appearance and texture from the other two. These three projections are at equal distance around the terminal stigma. In view of these facts they can be no other organ than staminodia, representing the three stamens of the inner whorl. And the other two smaller projections are also staminodia which together with the single fertile stamen represent the outer whorl of three stamens. Now it is safe to say that the two auricles existing in Cephalanthera, Epipactis and the Orchideae are, in fact, also staminodia, representing the two lateral stamens of the inner whorl. In consequence, there is fairly good reason to believe that the column in Orchidaceae has developed from the union of six stamens and a central style, and this is in agreement with the conclusion drawn by Swamy from vascular anatomy of orchid flowers. Furthermore it is also an interesting fact that the pollen grains in this genus, like those of Cephalanthera, Pogonia (sensu stricto), Aphyllorchis and some species of the Vanillinae, are single, while in the vast majority of the Orchidoideae they are united into tetrads. This feature, as well as the texture of pollen grains, is of considerable significance in the classification and phylogeny of Orchidaceae. On the basis of its morphological characters mentioned above, the present genus is evidently one of the most primitive types in the subfamily Orchidoideae. It bears a strong resemblance both in habit and floral features to Cephalanthera, especially C. falcata (Thunb.) Bl. The relationship between these two genera is apparently much closer to each other than to any of other existing primitive orchids. In addition, the similarity in some of significant floral characters between Tangtsinia and the saprophytic Aphyllorchis, especially in the nearly regular perianth and a subterminal stigma of A. simplex Tang et Wang, indicates their close relationship. It is quite possible that Cephalanthera and Aphyllorchis are derived from Tangtsinia or Tangtsinia-like ancestor. Thus, Tangtsinia is here placed as the most primitive genus in the Limodorinae. Furthermore this new genus likewise shows more or less close affinity to Neottia, probably through N. gaudissarti Hand.-Mzt., in which the flowers consist of nearly regular perianth and a very primitive column with a terminal stigma and without the rostellum. On the other hand, in comparing Tangtsinia with the Apostasieae, there occurs also some similarity, but a closer investigation of their ovaries, perianthes, stigmas and some other features indicates that there is little evidence of close or direct relationship between these two taxa, although both are the ancestral types in this large family. The probable relationships between Tangtsinia and its allies may be diagrammed as follows: With regard to the pollination, Tangiorchis is found to be self-pollinated. In the great majority of cases, its flowers do not open at all, and none of which has been seen to be visited by any insects. It is interesting to note that in almost all nearly faded flowers examined by the writer the bases of the pollinia together with the base of the anther have become attached to the stigma of the same flower, and thus self-pollination has taken place. This type of pollination might be comparable with that of Cephalanthera damasonia Druce. Finally the writer should say something about its geographical distribution. This monotypic genus is confined to Gin-fu-shan (Mt. Gin-fu) and its adjacent region in Nanchuan District of southeastern Szechuan, where it occurs at scattered points within an area of no more than 250 square km. at an altitude between 700-2100 m. In view of its morphology, pollination and geographical distribution, Tangtsinia might be an ancestral relic of the family Orchidaceae and would give a possible clue as to the origin of this complicatedfamily.