17S m^r^^ns. ^Tie cell-vnnps pre rlso stiff in type ^, end elthough 



the cTosinP' boss ir. absent the cell-i'dngs are so arranged that the 

 OT^eninfr they le'^ve betv/een then at the outer side of the cell is 

 too nerroT" for the seeds to pass through v/ithout force being used, 

 as nay easily be ascertained by experinenting v.ith a needle and 

 trying to nush them through the opening either from within or 

 without the cell. 



^^ov it is certain that this complicated structure is somehow 

 arrpne"^'^ -"or the disr»ersal of the seeds. Kature does not ^"aste 

 her ener«?y for nothing. It is also certain that the seeds do get 

 out o-^ these complicated structures, as I have found old capsules 

 often to be ouite emT>ty. The capsules remain upon the plant often 

 for t'""©, three or more 7;"ears, and open and close v;ith every fall 

 of rain, and it is only at th-^t time the seeds are released, i^-ut 

 hovr'' During one of our vet summers I fixed some ripe capsules of 

 the D ty^e over a pan and placed them in the garden, where they 

 remained for many days fully exposed to rain end wind. The result 

 v.rpp thaf.T found fewer seeds ivashed out of the capsules into the 

 nan than thsre ^"ere capsules, an average of less than one seed per 

 capsule. This does no+ accord v itb v-^at I think must take place 

 when tfie cansule is on the plant in South -Africa, -'i-lso j- should 

 er-nlain th^^t I h^ve only ex-nerim.ented with capsul-^^s belonging to 

 the stemless 

 1'^;? s'^ec-'es, in v^ich the cansi\le v;ould not be raised more than from 

 half-an-inch to four "nchas above the ground, so thrt shaking by 

 vrind ■""oul^ not have much effect in dispersing them. 



I have noticed, however, that often v/hen capsules of differ- 

 ent tTH^es have been re-ex"anded by ■'■etting after a previous expan- 

 sion, a fev; (one to six ) seeds may fall out of them into the sauc- 

 er or basin in which I have nlaced them, ^o that it is possible that 

 at e-^c^, o^enlnp- O"** the cansule onlj'' a fe^' seeds ere liberated at a 

 time, the chance of their being able to germinate and obtain suffi- 

 cient moisture to grow to a size that would enable them to v/ith- 

 st'='^d the drv season, bein? a var'^'-'n-r one; for if all the seeds 

 v.'ere shed at once and the subsequent rainfall insufficient to enable 

 the seedlings to establish themselves, the efforts of the plant 

 would be vrasted for that year. In the case of the typ.e of capsule 

 A, 3, G, where the cell-wings are thin and membranous, the ~ libera- 

 te or o^ the seeds seems to be effected by the rain beating upon 

 ^■he roofin'^ cell-Kinps so as to cause them to flap iipon the seeds 

 and dis^^lrce then -from their strlks so that they can afterwards be 

 v.'ashed out O'*' thp capanle. But even in this type I do not find 

 thft all the see'^a in ^ car^sule ^re washed out at once in the ex- 

 periments T have m-f^de. It v.'ould be interesting if some ^outh 

 African botanist, having the ot^- ortunity, woul'"" observe and give 

 '^n f=ccount o^ vhat takes place under natural conditions with the 

 v^-r''ou- t""'Pes o'^ capsule. 



Tn thp extract I have made from Haworth's book he mentions the 

 c^^sule as pcsibly formlna' a character by which the eections of 

 this cenus mia'ht be se'^^rated to form nev? renera . This may be the 

 case in some instances, yet as I find the same type of cg.psule is 

 sometimes found in tv/o totally different sections, and two types of 

 car^sule sometimes found in the very same section, the character of 

 the capsule does not seer, to be .- reliable or a promising one for 

 di^tina-uishinP' genera or sections. ?or example, G in ^igs. 51, 62 

 represents the capsules o"" one of the stenless -^phaeroid group, 

 and they ■nossess distinct •'-'ell-developed cell-wings, v/hilst the 

 Qfr^'^.u'le o^ '^nother sr^ec" es- o"^ the seme group (^•^. piluliforme) is 

 nuite de?:tit.ute o-" eel 1-wlnors , en^ in this cb^.racter auite rrrees 



