56 BULLETIN 82, UNITED STATES NATIONAL MUSEUM 



doninae exhibit a number of more primitive characteristics, such as a relatively small 

 centrodorsal cavity, rather inconspicuous muscular fossae, and weak synarthrial 

 backward projections. It is interesting to note that we sometimes get here radial 

 prolongations of the dorsal coelome that reach down to the ventral side of the 

 centrodorsal and there form small and shallow pits. 



It seemed very plausible to Gisle'n to assume that the second group of Macro- 

 phreata proceeded from a type nearly allied to the Thalassometridae, as well as to the 

 fossil genera Amphorometra and Conometra. If we imagine a reduction of the side 

 plates and covering plates in the thalassometrid type, we get a form extremely similar 

 to the more primitive types within group b the subfamily Zenometrinae. The 

 size of the centrodorsal cavity for which this group b among more advanced forms is 

 remarkable shows, as the synarthrial backward projection also does, a more moderate 

 development in certain Zenometrinae. If it were a question of placing the genus 

 Conometra in one of the recent groups it would be difficult to decide whether it should 

 be referred to the Thalassometridae or to the Zenometrinae. In the same way it would 

 be almost impossible, in certain cases, to distinguish between recent Thalassometrida 

 and Zenometrinae if the centrodorsal and radial pentagon only were known. On the 

 other hand, the genus Amphorometra shows, like Placometra, distinct points of con- 

 tact with thalassometrid forms with the Thalassometridae and Charitometridae. 

 It seems hardly possible, according to Gisle'n, that the genus Jaekelometra can have been 

 the ancestral form of the Atelecrinidae. There are certainly similiarites in the size 

 of the basal ring, but the relative smoothness of the cirrus facets in Jaekelometra, the 

 small size of the cavity in the centrodorsal, and the steep inward inclination of the 

 radial articular faces which are provided with relatively insignificant muscular fossae 

 differ radically from the conditions in the Atelecrinidae, in which the cirrus sockets 

 have a strong crest and the centrodorsal cavity and radial muscular fossae are extra- 

 ordinarily large. It appears, too, as if in Jaekelometra we are able to find an indication 

 of the development of a rosette through thin centripetal continuations of the basals. 

 No such formation seems to be present in Atelecrinus. Gisle'n therefore concluded 

 that Jaekelometra is a type standing nearer certain primitive Thalassometrida. 



Conclusions on the evolution of the comatulids. Gisl6n said that his investi- 

 gation of the mutual relations between the comatulids led him to assume the possi- 

 bility of three, or perhaps four, different lines of stalked criuoids of the pentacrinid 

 type having reached the eleutherozoic mode of life and become comatulids. 



The first and oldest branch of the comatulids is represented by Palaeocomaster, 

 to which the recent comasterids correspond. At the present time they still show their 

 unique nature by the strong tendency toward reduction of the cirri, by the frequent 

 occurrence of exocyclic forms, by the absence of sacculi, which are replaced by the 

 pear-shaped organs, by the combs, and by the dorsal hooks. In contrast to all the 

 other comatulids they have never acquired the capacity for swimming [not quite 

 correct; see Part 2, page 602, fourth paragraph; page 606, second and last paragraphs], 

 but still continue to employ a creeping mode of locomotion. 



The second somewhat younger branch consists of the solanocrinids. We find 

 among these in the Upper Jurassic forms having strongly biserial arms and very 

 clumsily built. Gislen assumed that these were very probably creeping forms. Pre- 

 sumably this type has died out, while less specialized solanocrinids with monoserial 



