TlIK CkIXoIDS I-K(pM |)1!. S. I!(I( k"s IvM'KDI'I'IhX 1(1 .IaI'AX I'.MI. .").") 



l'"iirllH'i' it may lie tliat (Ik'sc possihly cxisling small s|iefies 

 partl\' make li.\ luiils lictwccii tiicmsolvos and |iorha|)s parlly also rurm 

 crossings with |M(iximal specie's in tlio ,Q;eiiei'a Cnmaxler. OovHuiilieria 

 and Comantliiis. If tla-sc iiylirids iwo fcrtilo they will al'tci- a conplc 

 of generations he split into countless different types. Sujiposing. f(ir 

 instance, that a form without cirri, w ith ill and IV Br-s : 4. and with 

 a ComanlliHs-comh (restricted to prctximal pinnules) were crossed witii 

 a form with cirri, with ill and I\' Hr-s : 2, and with C'oinaslcr-c-omh 

 ((hat is short, iiigh comhs (hsjiersed to distal pinnules|. h^ui-ther. sup- 

 posing that such a hxhrid were fertile and therefore in the 2'' gene- 

 ration was normallx split in Ihe different jiossibilities of coml)ination. 

 Then one would get forms (hat might be referred to all the genera 

 known within (he sub-fam. Coinastei-iiKr. (If, e. g., the first 3 gens 

 as heterozygotic might be I'cpresented with (lifferen( forms: i-udimen- 

 tary cirri, mixture of '2 and 4 components in (he III and 1\' division- 

 series, one would get lOS different possil)ilities). The great \ai'ialiili(y 

 in the division-series even in the same specimen seems to be a pi-oof 

 of the probability of the heterozygotical natui-e of eerlain specimens of 

 (-'. /ini'ricin'ii. 



In a special papei' 1 will trea( a 8'' j)ossibIe influencing cause, 

 viz. alteration in the mode of ca(ching food. 



Though ill man.\- cases (he Comastcrids show primidve charac- 

 eristics such as clumsy formadnn. frequently (in (he genus Coinaster) 

 a primiti^"e way of arm-di\ ision. a number of components in the divi- 

 sion-series (hat is \ cry indelini(e. it is eviden( (hai in other charac- 

 tei'istics (he family is in fas( development and has enlered qui(e a 

 specific direction of evohuion. A lot of (he charac(eristics which are 

 engaged in such a de\elopment have this apparently cansed by the 

 changed maiinci- (if c.ifchiiig food and are (I'eated in coimeetion \\i(h 

 this ques(iiin. 



The \ariabili(y of the cii'ii is pi'obaidy caused liy another 

 factor and tiieir regression, which is observable in different species of 

 the Comastpri(J family, I suppose lo be caused by (he dorsal hooks on 

 the distal piiimdes ha\iiig taken over (he anch(»ring func-tion that 

 otherwise falls updu the cirri. Those dorsal luMdvs arc to lie found in 

 most of (he ('niDiiliilids^ though usually visible only liy sli'ong magiii- 



