466 MEMOIRS OF THE CARNEGIE MUSEUM. 



these connections become more frequent, are heavier, and stand closer together. 

 Very often we see short connections grouped together either in a reticulate form, 

 or in rows parallel to the gill-filaments, ie., vertical to the edge of the gill (See text- 

 figs. 1 and 2 i, on pp. 455-6), and it should be noticed that precisely the same ar- 

 rangement is also seen in the Australian Hyridella (Ortmann, 1912, pp. 100-103, 

 fig. 1). The connections of adjoining vertical rows may alternate in an irregidar 

 way, and may thus form irregular transverse or oblique rows. Towards the edge of 

 the gill, and near the base, the connections are often somewhat elongated, stand 

 closer together in the same row, and thus a more distinct arrangement into vertical 

 septa is brought about, separating more distinct water-tubes (ovisacs), which, how- 

 ever, are laterally connected with each other by the interruptions of the septa 

 (See PI. XLV, figs. 16, 2b, 3; PI. XLVI, figs. 1, 2, 3, 4, 6, 7a, 7b; PI. XLVII, fig. 1). 



There is no doubt that the arrangement described above is primitive, for it 

 most closely approaches the condition seen in the male gill. The marsujiial struc- 

 ture is simply brought about by a more frequent and heavier development of the 

 intei'laminar connections of the male, which approach each other, and partly ar- 

 range themselves in vertical rows. 



A step in advance is observed when the vertical rows prevail over the reticulate 

 arrangement, and extend over the whole marsupial portion of the gill. Such cases 

 are found, and in them the interlaminar connections may be shorter or longer, 

 but they always stand close together in each row, so that we have all through the 

 marsupium the appearance of distinct septa, which, however, are perforated by 

 holes, so that the water-tubes (ovisacs) communicate with each other; this is most 

 evident in Diplodon piceus (PI. XLVI, fig. 2). 



Furthermore in a few species {D. decipiens, PI. XLV, fig. 46, and D. mogymirim, 

 PI. fig. 5c; see also PL XLVII, fig. 7, and PI. XLVIII, figs. 2a, 2b), I have observed 

 a further advance in this structure. Here the septa become solid, the interruptions 

 are missing, and each septum runs from the base of the gill vertically towards the 

 edge, exactly in the manner which we know to be the characteristic condition in the 

 Unionidw. Well-defined water-tubes, which do not communicate, are thus formed. 

 The species, which show this structure, have in the marsupial part of the gill 

 hardly any indications of interruptions of the septa. In the non-marsupial por- 

 tions, as well as in the outer gill of the female, and both gills of the male, the usual 

 structure of Diplodon is present, showing few and scattered connections. This 

 reveals that the solid septa of these species of Diplodon represent the highest stage 

 in the marsupial development in this genus, and that this feature should not be 

 considered as homologous to the septa of the Unionidce. It is analogous, and has 

 been independently^ acquired. 



