THE ANGIOSPERMAE 1183 



Xanthorrhoea, we cannot consider the sporangia to be marginal. The 

 separation of the sporangial rudiments on the primordium is also evidence 

 against the view that there are only two sporangia which are subdivided 

 into four by trabeculae of sterile tissue. In our present ignorance of the 

 evolution of the stamen we cannot even be sure whether the position of 

 the sporangia is a real problem or not, but it is worth noting, in passing, 

 that ovules may be either marginal or superficial. 



The attachment of the sporangia to the connective is usually longitudinal, 

 though transverse sporangia are also known {Pinguicida) as well as anthers 

 in which the sporangia are spherical and form a rectangular group, either 

 horizontally or vertically. Sporangia which are not only transverse but 

 terminal, lying apparently across the top of the filament, are known in 

 Verbascutn, and this and some other abnormalities of position are usually 

 ascribed to displacements during development, though little is known of the 



facts. 



Two members of the parasitic family of Loranthaceae possess ring- 

 like sporangia. In Korthahella there are three stamens with coherent 

 anthers. At maturity all the pollen loculi fuse into a ring. In Arceutlwbiinn 

 the one pollen loculus is annular from the beginning, surrounding a central 

 columella. 



Several anomalies of anther form are found in the Cucurbitaceae. The 

 stamens are concrescent in most of the genera and the sporangia, which are 

 attached to the outer surface of the connective, are very long and sinuous, 

 forming a compressed S-shape. There is also the remarkable case of 

 Cydanthera, mentioned on p. 1173, in which the male flowers contain but 

 one, apparently terminal, stamen, around the top of which run two trans- 

 verse, ring-shaped pollen-sacs. This may be either the result of concres- 

 cence or it may be a single stamen, the other four stamens of the normal 

 whorl of five having aborted, and the two pollen-sacs having resulted from 

 the union in pairs of the original four. The unification may have resulted 

 in this case from the transformation of intermediate sterile tissue into 

 archesporium. The monosporangiate anther of Alchemilla has probably 

 arisen from this latter event. It also seems to have occurred in some 

 species oiNaias {e.g., N.flexilis) where there is a single terminal stamen, with 

 a central core of archesporium, which appears to be formed by the union of 

 four sacs, accompanied by the transformation of the central connective 

 tissue into additional archesporium. Naias major has normal anthers and 

 there are no transitional stages between the two conditions. 



The number of sporangia can also be reduced by abortion. This is 

 shown by Piper betel, where the number in each anther may be reduced 

 from four to one or even none. A similar variation occurs in the anthers of 

 the cleistogamic flowers (see p. 135 1) of several genera, notably in Viola. 

 The occurrence seems to be connected with general floral reduction in both 

 the above cases. Bisporangiate anthers are not uncommon (Amarantaceae, 

 Scrophulariaceae, Epacridaceae) and are not necessarily linked to general 

 floral reduction. An apparently bisporangiate condition may arise through 



