UMBELLIFERAE 461 



umbellules of the primary umbel. The umbellules of tertiary umbels display a still 

 further diminution of hermaphrodite flowers, and an increase of male ones : either 

 a few marginal hermaphrodite flowers still remain, or these have entirely disappeared, 

 so that the entire umbel is often purely male. If the well-marked protandry of 

 Umbelliferae promotes cross-pollination in a high degree, this is further accentuated 

 by the peculiar distribution of the sexes. Warnstorf found that the anthers of our 

 native species are neither introrse nor e.xtrorse, but dehisce laterally. As, however, 

 the two outer anther-valves curve back towards one another, while the two inner ones 

 retain their original position, the shed pollen faces outwards. 



The protandry varies greatly in degree. Beketow (' U. d. Proterand. d. 

 Umbellif.,' 1890) found it to be most strongly marked in Anthriscus sylvestris 

 Hoffm. and Carum Carvi I.., the flowers appearing at first purely male, and 

 afterwards purely female. The first (terminal) umbel, borne by the primary 

 axis, is here very feebly developed; the lateral umbels are much larger, and by 

 elongation of the secondary axis attain a higher level. The terminal umbel is 

 purely female at the time when the lateral ones are still male ; the female flowers 

 being at a lower level than the male ones, their geitonogamous pollination is 

 ensured. 



Beketow states that protandry is much feebler in Heracleum Sphondylium Z., 

 Aegopodium Podagraria Z., and Angelica sylvestris Z. In these species the primary 

 umbel is larger and at a higher level than the lateral ones. It is possible that this 

 relation between the degree of protandry and the development and position of the 

 various umbels is of more general application. (Rothert, Bot. Centralbl., Cassel, xlv, 

 1891, p. 381.) 



Schulz summarizes his observations somewhat as follows (' Beitrage,' II, 

 pp. 90-1).— 



Male flowers are generally associated with hermaphrodite ones ; they are 

 either borne on the same or on different stocks. In the former case, they are 

 distributed in one or other of the two following ways. — 



1. Hermaphrodite and male flowers are associated either in all the umbels 

 (mixed umbels), or only in some, usually those of a higher order. In the latter 

 case, umbels of a lower order are as a rule entirely hermaphrodite. In mixed 

 umbels both kinds of flower are usually present in all the umbellules ; purely male 

 umbellules are not common in most species, and as a rule only occur in the centre of 

 umbels of a higher order. Now and then, however, individual umbels are purely 

 male ; in some species, such as Oenanthe fistulosa Z., those of the highest order are 

 always so. Most of the species investigated by Schulz belong to this group. The 

 umbellules either bear,— 



(a) marginal hermaphrodite and central male flowers, except in certain genera 

 where the terminal flower, when present, is usually hermaphrodite ; or, 



(b) the hermaphrodite flowers at the periphery, in the centre (Sanicula europaea 

 Z.), or an intermediate zone between marginal and central male flowers (Astrantia 

 major Z.). 



2. Hermaphrodite and male flowers are but very seldom associated in the same 

 umbel ; only umbels of a higher (or even the highest) order are male. Sometimes, 

 however, all the umbels of a plant are completely hermaphrodite ; more rarely flowers 



