A well-corroborated morphological scheme of interrelationships for centipedes, once broadly accepted, has been in conflict with molecular data with respect to deep branching events. Expanded taxonomic coverage compared to previous analyses adds longer fragments for 28S rRNA and a structural alignment as part of a sample of four genes (two nuclear ribosomal and two mitochondrial) for 111 extant species; these sequence data are combined with morphology under parsimony and maximum likelihood, explor- ing both traditional multiple sequence alignment and direct optimization approaches. Novel automated procedures to incorporate secondary structure information are also explored. The molecular data in com- bination yield trees that are highly congruent with morphology as regards the monophyly of all centipede orders as well as the major groups within each of the large orders. Regardless of the optimality criterion or alignment strategy, the Tasmanian/New Zealand Craterostigmomorpha is resolved in a different posi- tion by the molecular data than by morphology. Addition of morphology overturns the placement of Crat- erostigmomorpha in favour of the traditional morphological resolution and eliminates the need to posit major character reversals with respect to developmental mode and maternal care. Calibration of the tree with Palaeozoic and Mesozoic fossils for a relaxed clock analysis corroborates the palaeontological signal that divergences between centipede orders date to the Silurian and earliest Devonian, and familial diver- gences are likewise almost wholly Palaeozoic.

The only well-documented data on female brooding posture in the geophilomorph family Mecistocephali- dae come from Dicellophilus carniolensis (C.L. Koch, 1847), in which the mother coils around the eggs and hatchlings with the dorsal surface outwards. This posture is shared by Craterostigmomorpha and Scolopen- dromorpha but not by other Geophilomorpha (united as Adesmata), which coil with the ventral surface out- wards. The change in brooding behaviour has been thought to coincide with the evolution of ventral glandu- lar pores in Adesmata and defends the basal split of Geophilomorpha into Placodesmata (Mecistocephalidae alone) and Adesmata. However, a brood of another mecistocephalid, Mecistocephalus togensis (Cook, 1896), documented in situ in Cameroon, shows the mother to guard the hatchlings with the ventral surface out- wards, in the manner of Adesmata rather than that seen in Dicellophilus. This observation suggests that the brooding posture may be more subject to convergence or evolutionary reversal than previously expected.

The phylogenetic relationships among the ‘‘archaeogastropod’’ clades Patellogastro- poda, Vetigastropoda, Neritimorpha, and Neomphalina are uncertain; the phylogenetic place- ment of these clades varies across different analyses, and particularly among those using morphological characteristics and those relying on molecular data. This study explores the re- lationships among these groups using a combined analysis with seven molecular loci (18S rRNA, 28S rRNA, histone H3, 16S rRNA, cytochrome c oxidase subunit I [COI], myosin heavy-chain type II, and elongation factor-1a [EF-1a]) sequenced for 31 ingroup taxa and eight outgroup taxa. The deep evolutionary splits among these groups have made resolution of stable relationships difficult, and so EF-1a and myosin are used in an attempt to re-examine these ancient radiation events. Three phylogenetic analyses were performed utilizing all seven genes: a single-step direct optimization analysis using parsimony, and two-step approaches using par- simony and maximum likelihood. A single-step direct optimization parsimony analysis was also performed using only five molecular loci (18S rRNA, 28S rRNA, histone H3, 16S rRNA, and COI) in order to determine the utility of EF-1a and myosin in resolving deep relationships. In the likelihood and POY optimal phylogenetic analyses, Gastropoda, Caenogastropoda, Neritimorpha, Neomphalina, and Patellogastropoda were monophyletic. Additionally, Neomphalina and Pleurotomariidae fell outside the remaining vetigastropods, indicating the need for further investigation into the relationship of these groups with other gastropods.

Gelatinous egg masses are common in a number of animal phyla. However, they are virtually unknown in marine bivalves, with structures that could be thought of as gelat- inous egg masses being reported for only five species. We describe the gelatinous egg mass and intracapsular development in the tropical lucinid Phacoides pectinatus. The embryos devel- oped within individual capsules embedded in a large flimsy, spherical mass. Swimming veligers hatch at 198 mm shell length. They did not feed, settled within several days of hatch- ing, and metamorphosis was completed within 2 weeks of hatching. Gelatinous egg masses might be detected in members of more lucinid species if studies of development included field or in vivo observations of reproduction in addition to producing embryos by stripping the gonads.

The cyphophthalmid family Pettalidae in South Africa is revised and seven new species are described from museum material collected between 1939 and 1985. Two of these are placed in the genus Purcellia and five in Parapurcellia, bringing the total number of described South African cyphophthalmids to 15. In addition, Purcellia peregrinator is transferred to the genus Parapurcellia. Phylogenetic analyses of discrete morphological and continuous morphometric characters, both separately and in combination, support the generic assignments and contribute towards a more detailed understanding of the systematics of the group in South Africa. In order to assess the stability of our phylogenetic results, the different morphological datasets were analysed under equal and implied weighting, as well as under several weighting schemes that varied the respective contribution to tree length of the discrete and continuous data partitions. These variations generated two phylogenetic hypotheses: (1) monophyly of the South African pettalids + Austropurcellia from north-eastern Australia as a derived clade within Pettalidae; and (2) polyphyly of the South African pettalids with Parapurcellia basal within Pettalidae. The latter hypothesis is congruent with previous molecular phylogenies of Cyphophthalmi, and has moderate bootstrap support. The sisterhood of Purcellia griswoldi, sp. nov. and P. lawrencei, sp. nov. receives high nodal support across analytic methods. New combination: Parapurcellia peregrinator (Lawrence, 1963).