248 MR. A. S. HOENE— A COiSTEIBUTIO]^ TO 



that the condition described hj Baillon may occur more frequently in places from which 

 he had not obtained material. The writer lias hitherto failed to find even traces of a 

 second ovule in H. vi7'ginica^ Zuccari7iicma, and arhorea (Text-fig. 1, A 3, A 4), but he has 

 observed biovulate loculi in Fothergilla Gardeni and major, Corj/lopsis spicata, and 

 Farrotia Jacquemontiana (Text-fig. 1, C 2). Parallel instances to these occur among the 

 Araliacete and Umbelliferae : thus two ovules suspended from terminal parietal placentae 

 develop in each loculus of Aralia and AcantJiopanax and, according to Tanfani *, of 

 Apium ; and one of each pair aborts as it does in the Hamamelidoidese cited above. 



According to Baillon'sf description, reduction in the ovary proceeds still further, 

 thus in Samamelis the ovary is sometimes completely sterile: — '*Dans un grand nombre 

 de fleurs, le gynecee demeure petit, sterile, on ne renfermant que des rudiment 

 d'ovules"; again, in Corylopsis^ **I1 y a tons les passages quant au gynecee entre des 

 ovaires completemeut steriles et vides; et ceux que contiennent des ovules bien 



■ 



constitues, dans les fleurs femelles ou hermaphrodites," Here, in two distinct genera. 



separation of the sexes is recorded as actually taking place, a feature highly significant 



when it is remembered that several genera of Hamamelidacese possess polygamous 

 flowers and in certain others the flowers are unisexual. 



The development of the ovule of Hamamelis virginkma and Liquidamhar styracifiua 

 was described by Shoemaker J in 1905, who found a nucellus of considerable bulk 

 invested by two integuments and a parietal layer of cells similar to that formed in the 

 nucellus of the Saxifragacese and Kosacese. The writer has examined the ovules of 

 Fothergilla, Cori/Iopsls, and Farrotla, and finds that they are constructed like those 

 of Hamamelis. 



Comparisons have been made from time to time by various writers between the 

 genera of the Hamamelidacege and Gornacese that are provided with conspicuous 

 involucral bracts. The writer has examined the bracts of Farrotia, and finds that they 

 differ morphologically from those of the Cornacese. The bracts of Farrotia are 

 tripartite, but this character cannot be easily observed, owing to the special development 

 of one or other of the three segments. Since the foliage leaves and the lowermost 

 bracts are stipulate, it appears that the coloured portion of the bract may correspond 

 either to a stipule or the lamina of a leaf, and so differs from the bracts of the 

 exstipulate genera, Cornus and Bavidid. 



The disposition and transformations of the stamens furnish other characters of 

 interest from an evolutionary point of view. According to Baillon §, eight stamens are 

 present in Licoryphe, but the four alternipetalous ones axe sterile, and again, in 

 Tetrathyrkim suhcordafum and Co7ylopsis, glands occur in the interval between the 

 stamens. Slioemaker [[ points out that abortive pollen is sometimes developed in the 

 nectaries present in the flowers of Liquidamlar styracifiua, and he regards these 



E. Tanfani in T^uov. Giorn. Ital. vol. sxiii. (1891) 451-469, t. 4-7. 

 t H. Baillon, Histoire des Plantes, iii. 390 {Hamamelis) ; 891 {Coryh 

 t D. :N^. Shoemaker, m Bot. Gaz. sxxix. (1905) 2Q2. 

 § H. Baillon, I.e. 392 {DicorypJie) ; 391 {Tetrathyrium). 



U D. N. Shoemaker, I c. 262. 



