Abstract: Clematoclethra (Franch.) Maxim. is a genus endemic to China, and has its distribution center in the area from W. Henan eastwards and to N. Guangxi southwards. For the whole genus, only 1 species comprising 4 subspecies is recognized in this treatment (see Fig. 3). Its members are woody vines and grow in thickets. They are very attractive for their medium-sized white to rosy flowers in spring and scarlet or purplish black to black berry-like fruits in autumn. It is due to the inadequate knowledge of variation within the genus that Liang and Chen (1984) have recognized 20 species and 4 varieties in the Flora Reipublicae Popularis Sinicae, vol. 49 part 2, including 10 new species and 2 new varieties, just half of the total number of species in the genus. Nearly simultaneously, C. Y. Chang (1983) described 2 new species from Sichuan in addition. The binominal inflation has reduced the taxonomy of the genus to a state of confusion and made floristic analysis difficult. Based on the principal key characters taken from the descriptions in Liang and Chen (1984) and Chang (1983), 22 polygonal graphs are presented for the 22 species in this article (Fig. 1: 1-22). Although the extreme forms, such as those of C. argentifolia and C. actinidioides and that of C. scandens, are very different, there are a series of intermediate polygons, which fall within the range bounded by the extremes and bridge these extremes. We are disposed to consider the genus monotypic in view of this fact. Actuall. Airy Srow (1936) already pointed out “Certain forms appear more distinct from others, but on general it may be said that the ‘species’ o this genus are exceptionally arbitrary”. But unfortunately, Airy Shaw’s warning was neglecled. In the present article, the outer morphology of the plants was extensively studied, which discloses that leaf-form, and all parts of flowers are of little taxonomic significance. The statistical method and histograms were used for evaluating the infraspecific variation. 473 herbarium sheets were treated. Based on the vestiture of annual young branch, leaf, pedicel and calyx, flower numbers per inflorescence, and fruit colour, the species, C. scandens, may be divided into 4 forms (Tab. 1). They each have a particular range but overlap somewhere (Fig. 3). Moreover, three forms occur at different elevations in Mt. Emei, form 1 occurring at 1600-2000 m, form 3 at 1100-1800 m and form 4 at 1850-3100 m (Fig. 4). Inasmuch as the 4 forms are vicarious geographically and ecologically and have differentiated relatively distinctly in 1-2 characters, they are here treated tentatively as 4 subspecies, on the basis of herbarium study alone. A key to 4 subspecies is given as follows. 1. Annual young branches setose, fruits always scarlet at maturity ............ subsp. scandens (form 1) 1. Annual young branches without seta, if present, very rare and sparse, fruit always purplish-black to black at maturity 2. Pedicel and calyx usually woolly, inflorescence usually with 3-7 (-12) flowers ........................ ......................................................................................................... subsp. hemsleyi (form 2) 2. Pedicel and calyx usually not woolly 3. Leaf blades usually tomentose below, inflorescence usually with 3-6 flowers ........................ ................................................................................................ subsp. tomentella (form 3) 3. Leaf blades usually glabrous or sparsely pilose below, inflorescence usually with 1-3 flowers .................................................................................... subsp, actinidioides (form 4) As a result of the study, more natural and reasonable boundaries between the taxa may have been found. Someone may ask why the conclusion here made is so distinctly different from those of the previous authors. We believe that the major concern in taxonomy, and also the first step of classification, should be the assessment of characters and the understanding of their variation, and then the choice of correct names of the taxa worthy of naming. The delimitation of species should be based on correlation of characters and discontinuity of variation but not on the distance (or difference) from type-center. Here we by no means diminish the significance of type specimens in taxonomy. They are indeed very important, but only used in nomenclature for correct application of names. Davis and Heywood (1963: 10) are right when they said “only names have types, not species”. It must be reminded that attention should be paid to the noticeable inconsistency in delimitation of species in various volumes, even within a single family written by different authors in Flora Rep. Pop. Sin., which makes comparison difficult. We eagerly hope that the authors of Flora Rep. Pop. Sin. do more to alleviate this shortcoming in next edition. Judging from the distribution pattern of its close relatives, the genera Actinidia and Saurauia, We suggest that Clematoclethra is a new endemism. Actinidia is mainly distributed in East Asia, with only 2 species extending fron temparate Asia to trapical regions of Malaysia. Saurauia mainly occurs in tropical America and Asia, with a few species extending to south China through Southeast Asia and one species reaching Sichuan. This reassessment is just the beginning of study on Clematoclethra, but it is hoped that this may help to stimulate further synthetic research on the group. We also confess to having not seen many type specimens, and thus some synonyms may have been misplaced here. The decisions have been made from the original descriptions only.

Key words: Actinidiaceae, Clematoclethra, Taxonomy, Endemic genus