ROSACEA. 3'57 



bud. Tlie androceum consists of a large number of stamens inserted 

 in verticils' round the circumference of a glandular disk which lines 

 the interior of the receptacle/ and ends in a more or less thickened 

 rim below the insertion of the perianth. Each stamen is composed 

 of a slender free filament, inflexed or crumpled in the bud, and of 

 an introrse, two-celled, more or less versatile anther, dehiscing longi- 

 tudinally/ The carpels, indefinite and free, present a sessile or 

 stipitate one-celled ovary, surmounted by a style which is con- 

 tinuous with its internal angle, and like it traversed by a longi- 

 tudinal groove, while at the apex is a more or less dilated stigma. 

 Sometimes the summits of these styles are separate ; but they may, 

 on the contrary, adhere together at a late stage, so as to simulate 

 a single column. Inside the ovary, along its internal angle, is 

 observed a parietal placenta, bearing towards its upper part a de- 

 scending anatropous ovule, whose raphe is turned towards the 

 placenta, while its micropyle looks upwards and outwards.^ Beside 

 this well-developed ovule, we sometimes find one aborted, which 



' As regards arrangement, these are an indefi- 

 nite number of " stamens of tlie Rosacece ;" we 

 always use this epithet as a phrase to denote 

 briefly the verticillate arrangement of the an- 

 droceum in this order, which we shall here 

 demonstrate compendiously. The study of or- 

 ganogeny can alone show clearly the exact 

 positions of the different stamens ; and this has 

 been done for the principal generic types by 

 Pater {Traite d' Organ. Comp. de la Fleur, 

 494-516, t. c.-ciii.). There are isostemonous 

 RoiacecB, possessing five stamens, superposed 

 either to the sepals {Slbbahlia), or to the petals 

 {Chamcerhodos). Others have a diplostemonous 

 androcenm, in which one whorl is superposed to 

 the sepals, the other to the petals {llorlelia, 

 Quillaja, &c.). But the diplostemony may re- 

 sult from another cause, and ten stamens may 

 be tound in a single whorl in pairs superposed 

 either to the sepals or to the petals {Arji-imouia) ; 

 so that this is due to a process of deduplication. 

 Again, with five petals, we may have fifteen 

 stamens, because five of the latter are superposed 

 either to the sepals or petals, while the ten 

 otliers are in pairs superposed to the petals or 

 sepals; or twenty stamens, three in front of each 

 sepal or petal, and one superposed to each petal 

 or sepal [Pninus, Ft/rus, &c.). Finally, when as 

 la the present genus, we liave a very large 

 number of stamens, it is due to one of the two 

 following causes : either there are originally few 

 staminal whorls, and each stamen superposed to 



VOL. I. 



a sepal or petal, as the case may be, is replaced 

 by a variable number of stamens ; or, as occurs 

 in the Roses, the alternating verticils (whether 

 of five or ten stamens each) are indefinite in 

 number, and extend from the orifice of the re- 

 ceptacle towards the bottom or organic apex. In 

 Sosa alpina, Payer has shown {loc. cit.) that 

 the stamens arise as in Geum ; first one verticil 

 of ten, " grouped in pairs, one stamen on the 

 right, and one on the left of each petal;" next 

 arises another verticil of ten, alternating with 

 the first, then a third, and so on. 



- On the surface of the tissue of this disk arc 

 hairs (contact with which, as we shall see, pro- 

 duces a mechanical irritation of the skin) usually 

 becoming more numerous as we ap])roach the 

 insertion of the ovaries, which also bear hairs of 

 the same structuj-e. They are simple and imi- 

 cellular, ending in a long point, and contain at tirst 

 a mixture of gas and liquid, and afterwards fine 

 granulations of a greyish or slightly orange- 

 yellow tint. At first the walls of these cells are 

 thinner than the diameter of the cavity; but in 

 the fruit they persist, and the walls become much 

 thicker in projiortioUjand tlio hairs far more rigid. 



^ The pollen-grains are marked by three longi- 

 tudinal tolds, which become narrow bauds in 

 contact with water (H. MoHL, Ann. Sc. Nat., 

 ser. 2, iii. 310). 



■• It possesses only one coat, with an unenual, 

 more or less oblique opening, sometimes irregu- 

 larly festooned or laciniatc (tig. 376). 



Z 



