THE STUDY OF THE EVOLUTION OP THE ELOWEE. 253 



suspended axially (D, C 1-C 3 ; loci). The structural agreement in this respect between 

 Ahelia and lAnncea is so close that it is impossible to distinguish between them. The 

 genus Dipelta, established by Maximowicz *, possesses an ovary with two pluriovulate 

 and two uniovulate compartments. Seeds are not formed in the pluriovulate loculi. 



Symi^horicarpus forms the next member of a remarkable series in which that portion 

 of the ovary corresponding to the superior half in Leycesteria fails to develop. The 

 adult flowers of S. racemosus are four-chambered below (E 1-E .3), but two-chambered in 

 the very small upper portion, owing to the fact that the upper portion of the lateral 

 carpel is incompletely developed (E4; loc2, loc4^). Each lateral chamber, locly loc3, 

 contains a single axially suspended ovule. The loculi at right angles to these are pluri- 

 ovulate, and the ovules which they contain are arranged in biseriate rows, suspended 

 from the small adpressed parietal placentae (E 3, pi) present in the upper portion of thft 

 ovary, with the exception of the lowermost ovule, which is axially borne. Only a 

 single specimen of an entirely pentamerous flower w^as found ; in this case the inferior 

 portion of the ovary was five-chambered and the additional loculus uniovulate. 



Abortion in the parietal portion of the ovary is carried still further in Vihirmim, 

 where, if the very young flowers of V. Lantana be examined, three united j)airs of 

 parietal placentge may be observed (Text-fig. 3, E 2, E3; pll, 2, 3). These are almost 

 in contact with one another, so that they separate three compartments {loc 1, 2, 3). 

 During subsequent growth only a single loculus (E1-E3; loc 2) develops as indicated 

 in Text-fig. 4, leaving two upper arrested compartments {loci, 3). The latter are pluri- 

 ovulate as previously recorded by Vidal f, and the ovules in each are arranged in two or 

 more series (Text-fig. 3, E4i ; ser 1, ser 2). At the same time Vidal, who made a special 

 study of the development of the ovary in the Caprifoliacese, entirely overlooked the 

 fact that the fertile ovule (E 1, ov) does not extend as it normally should do into either 

 of the adjacent compartments (E 1-E 3 ; loc 1, 3), but into the loculus which has 

 developed from the originally opposite compartment (E1-F3; loc 2). Again, the 

 placentae {pi 2,pl^) do not bear ovules as they normally should do on the sides adjacent 



1 



to loc 2. I'^inally, the fertile ovule does not actually form a member of the series of 

 sterile ovules that arise from pll and extend in loci and 3. These features furnish 

 an important clue to the interpretation of the peculiar morphology of Viburnum, which 

 will be discussed after the vascular system has been described. 



Samhiwus forms the extreme member of the series under discussion. Tlie nuhiber of 

 loculi in S. nigra varies from two to five, but there arc usually three. A single tcrmmal 

 axially situated ovule occurs in each loculus (G2). The area above the ovules, which 

 appears to form a part of the lower tissue region of the style, is in reality an exceed- 

 ingly rudimentary parietal portion of the ovary, wherein the ovules are reduced to tiny 

 swellini?s, arranged in short biseriate rows above the normal ovules. Eigs. G4-G8 

 show the ovular rudiments arranged in the series a, a \ h, h' \ and c^c above the ovules 

 A, B, and 0, respectively (G2). Hence it* appears that the terminal ovules of Sambiicus 



* Maximowicz in Bull, Acad. Imp. Sci. Saiut Potersbourg, xxiv. (1878) 50. 

 t Vidal in Bull. Soc. de Statistl<iuc, Grenoble, 4 ser. t. iv. (1899; 157. 



