ON MERLTA XORMAXI. 669 



sponges, especially among those with a enrypylous type of 

 canal system. i 



The collar-cells vary greatly in appearance according to 

 their state of contraction. Fiofs. 8-12 on PI. 33 are drawino-s 

 of collar-cells on one and the same slide. Some are relatively 

 short and thick, with the collar contracted down and joined to 

 the collars of neighbouring cells, and with a wide opening 

 with polj'gonal outline. Others are elongated and with 

 separate cylindrical collars. The collar-cells vary in height 

 from 4"5 /_{ to 8*5 /<. The usually spheroidal but sometimes 

 hemispherical body is about 3'3^ in diameter. The spherical 

 nucleus is at or near the base of the cell. The flagelluni 

 passes down through the cell to the nucleus (PI. 33, fig. 12). 

 The name 'Miymenopylous "- is proposed for the new type of 

 canal system, which seems to be a modification of the eury- 

 pylous type. 



(2a) The Siliceous Skeletox. 



The siliceous skeleton consists mainly of upright bundles of 



' Fig. 6, PI. 33, is, I find, slightly diagrammatic in places, but it is 

 svifficieutly accurate in places to justify the statement that the collar-cells 

 are joined to each other by stellate basal processes. Further investi- 

 gations of new, well-fixed material, prepared with a good plasma stain, are 

 desirable. It is not sui-prising that we have so little knowledge concei-ning 

 the inter-relations of collar-cells in sihceous sponges. The difference 

 between the tissiies of healthy living sponges and dead ones fixed under 

 the best possible conditions must be very great, especiaUy in the matter 

 of contraction. Minchin and Reid have shown (' Proc. Zool. Soc, 

 London,' 1908, p. 674, pi. xxxvii, fig. 24) that in Clathrina contorta 

 the bodies of the collai'-cells ai'e separated by delicate extensions of the 

 gelatinous ground- substance of the sponge. Muriel Robertson and 

 Minchin (' Qiiai-t. Journ. Micr. Sci..' November, 1910, p. 621) write that 

 probably the collar-ceUs are connected across this intervening substance 

 l)y protoplasmic fibrils, though they have not yet seen such connections. 

 Certainly the manner in which a contracted Ascon with its collar-cells 

 squeezed out among the porocy tes expands again with all its collar-cells 

 " dressed " and in line would lead one to siisj)ect that here also thei*e 

 may be a kind of collar-cell membrane, as in Hexactinellids. 



* vfu'iv, ivog. membrane : tti'Xjj, gate. 



