THE BASES OF ANGIOSPERM PHYLOGENY: FLORAL ANATOMY RICHARD H. EYDE1 ABSTRACT An eclectic ramble through phylogenetic aspects of floral structure includes the following: (1) Sterling's view that the ancestral flowers of Rosaceae had only two ovules per carpel is examined and rejected. (2) Recent observations on the direction of androecial development in various taxa are reviewed, and it is concluded that centrifugality is not as valuable a phylogenetic indicator as some systematists had hoped it would be. (3) An attempt is made to reinterpret the inverted placental bundles of Capparales and the inverted "recurrent" bundles of Nestronia along morphogenetic lines. It is suggested that the inverted orientation is causally related to the initiation and differentiation of these bundles in isolation from previously formed vascular tissue. Floral anatomy turns some botanists into fantasts, others into iconoclasts. Butdespite the frequent speculative excesses, the occasional overreaction, and therecurring disagreements that are a part of the field, serially sectioned and clearedflowers continue to provide essential phylogenetic information. To begin witha straightforward example, consider Cronquist's (1968) suggestion concerningthe origin of the Proteales, which he defines as Proteaceae plus Elaeagnaceae.Stressing similarities between the Proteales and the Thymelaeaceae (a point ofdifference with Takhtajan, 19702), Cronquist postulates that the origin of theorder was in the Myrtales. For this to be true, the gynoecium in Proteaceae andElaeagnaceae must be pseudomonomerous; in other words, it must be a syn-carpous gynoecium that has acquired through evolutionary processes the super-ficial appearance of a single carpel. Noting that the Myrtales, which aresyncarpous, must be excluded as possible ancestors if the gynoecium of theProteales should turn out to be a solitary carpel, Cronquist adds: "The mostlikely origin of the Proteales would then be in the Rosales." Serial cross sectionsthrough the gynoecia of various Proteaceae make Cronquist's favored positionfor the Proteales untenable, for there is no sign of pseudomonomery. Instead,each gynoecium has the three major vascular bundles and the ventral suture ofa single carpel (Fig. 1). The same is true of Elaeagnaceae (Fig. 2; see alsoEckardt, 1937: 47). The ancestry of the Proteales must therefore be sought ina group with apocarpous members such as Cronquist's Rosales or Takhtajan'sSaxifragales. The conviction that an apocarpous gynoecium did not originate from asyncarpous gynoecium will not be challenged in this forum because evidence isoverwhelming that apocarpy preceded syncarpy in many groups of flowering 1Department of Botany, Smithsonian Institution, Washington, D.C. 20560. 2I cite the Russian version of Takhtajan's Flowering Plants: Origin and Dispersai (1970)rather than the English version (1969). Although the English version was translated from aRussian manuscript, the printed Russian version appeared later and differs in a number ofways (see, for instance, the newly segregated families in Comales). While my symposiumcontribution awaited publication, Fischer Verlag published a German version: Evolution undAusbreitung der Bliitenpflanzen (1973).ANN. MIssoUrm BOT. GARD. 62: 521-537. 1975.