The Marginellas, and the Lip They Build Last

What a marginella is, and where the name comes from

A marginella is a small sea snail with a shell that looks like it was dipped in glaze. Most on this coast would sit under a fingernail. The shell is short and oval or pear shaped, with almost no spire showing, a long narrow opening down one side, and a surface smooth enough to reflect a window.

Marginellidae J. Fleming, 1828 is an accepted family name, AphiaID 23025 in the World Register of Marine Species, which is also MolluscaBase where mollusks are concerned. It sits in the superfamily Volutoidea among the neogastropods, and it is marine only. The register carries three vernacular names for it: marginellas in English, a German one that translates literally as margin snails, and a Japanese one.

That is the point of the name. Both the scientific name and the English one track the Latin margo, margin or edge, and the edge in question is the thickened outer lip of the aperture. The family is named after one wall of its own doorway.

Seventeen characters make this family, and only two are the shell

The most useful published work on this family is also the most counterintuitive. Paulino Jose Soares de Souza Junior's 2007 doctoral thesis at the Museu de Zoologia of the Universidade de Sao Paulo scored 93 characters across the marginelliform snails, 12 from the shell and 81 from the soft body, against volute, cystiscid, nassariid and murex outgroups.

The result: Marginellidae holds together on 17 shared derived characters, of which only two are shell characters. Seven are modifications of the odontophore, the cartilage and muscle cradle that drives the rasping tongue. Six are modifications of the reproductive system.

Read plainly, that says what makes a marginella a marginella is a rebuilt feeding apparatus and a rebuilt reproductive tract, neither of which anyone holding a beach shell will ever see. The famous glassy shell contributes two characters out of seventeen. Most of this page is about those two, because they are the part you can hold. It is worth knowing they are the minority report.

Why the gloss survives the surf

Here is the explanation in wide circulation: the snail keeps its mantle wrapped over its own shell for life, continuously buffing it, continuously laying down a thin glassy layer. It is a satisfying story. The anatomical literature does not support it, and the last clause is contradicted outright.

In Souza's matrix, shell surface is character 1, scored only as opaque or shiny, and the explanation offered is developmental rather than behavioral. The ancestral condition for a sea snail is a shell covered by periostracum, a protein skin, over an opaque mineral surface. The derived condition, the marginellid one, is a surface laid down smooth and bright, reaching what he calls an almost vitreous extreme. The same thing has arisen independently many times elsewhere among gastropods, and its consistency index in his own dataset is only 0.5, so the gloss is not even a reliable marker inside the group he analyzed.

Of the 81 soft-part characters scored, not one codes for a mantle covering the shell. Boyer's 2019 review of the same group does not use the word mantle anywhere. And "continuously" cannot be right, for the reason the next section covers. These animals stop growing.

The test that settles it is one anybody can run on a wrack line. Empty marginellas, long dead, with no tissue near them for months, still shine. A finish that needed maintenance would not outlive the animal.

So the gloss is structural, not maintained. There is no dull organic overcoat to chalk and abrade off, and the calcium carbonate underneath went down glassy to begin with. Marginellids do extend mantle lobes over the shell in life and that shields it, the way olive snails do. Shielding is not polishing. The olive page handles coverage, a real and separate mechanism.

The lip is the last thing the animal builds

Character 11 is the thickening of the outer lip, the character the family is named for, and it scores a perfect consistency index of 1.0.

The mechanism is the part to carry home. Marginellids have determinate growth. Souza states it directly: the thick lip forms at the end of the animal's growth. The shell reaches a final size, stops, and lays the heavy lip down as a single terminal event.

Three things follow, all checkable in a tray of shells.

• Adults of one species arrive at about the same size, because growth ends rather than slows.
• There is no half-grown adult with a thin lip. A thin, sharp-lipped marginella is a juvenile. Not a variety, not a runt, a juvenile.
• The lip is disproportionately heavy for a shell that small because it was never incremental. It is a one-time build.

The general rule, thick lip means finished growing, is covered on the shell identification guide; the marginellid version is stricter, and determinate growth is why. In some Prunum the thickening also shows externally, reflected over the dorsal face of the lip and leaving a groove.

The lip teeth run on the same clock. Character 10 covers denticles set inside the thickened edge, present in Dentimargo, Eratoidea and Marginella. Souza and Coovert reported in The Nautilus in 2001 that in this family they are among the last structures to form, and Nehm, working on Prunum in the Journal of Paleontology the same year, read them as anti-predator armament that narrows the opening. Neither paper was read for this page; both are reported as Souza cites them.

The plaits, and the one that fuses

Look into the aperture and there are folds running across the columella, the pillar at the shell's core. Those are the columellar plaits, and they are the first thing anyone points at.

They are not diagnostic on their own. Souza says so plainly: columellar plaits are not exclusive to Marginellidae, and various gastropods carry them, among them the Turbinellidae, Mitridae, Costellariidae and Volutidae. Do not identify a marginella by the presence of folds.

Two refinements do the work, and both score a consistency index of 1.0. The plait count is fixed from the first adult whorl, while the cystiscid used for comparison does the opposite and adds plaits as it grows. And the first plait fuses with the front of the outer lip, present in every marginellid tested and absent in both the volute and the cystiscid. The volute shares the fixed count. It does not share the fusion. Four is the working number, constant within a species and variable between species.

The folds are not decoration either. Character 18, another at 1.0, is a lobed columellar muscle in the subfamily Marginellinae. That muscle is the retractor that hauls the animal back inside. Volutes, cystiscids and the deepwater Marginellona all have plaits, but only in the Marginellinae is the retractor's shape significantly altered by them. Counting folds in an aperture is counting the channels of a muscle.

One trap comes with that. Character 9 covers parietal teeth on the inner wall opposite the columella. They appear only in adults that already have a thickened lip, run less than half a whorl back, and pass for plaits. Coovert and Coovert named the effect false plication. A fifth fold near the parietal wall may not be a fifth plait.

The spire the last whorl swallowed, and the door that went with it

A marginella reads as one smooth bean, and that is three characters moving together. Character 4 scores the spire from high through low through almost involute to totally involute. Character 3 scores the aperture, which in many marginellids runs past four fifths of the total shell length. Character 2 records that the suture between whorls becomes indistinct in nearly every member of the family.

Souza reads those as one transformation rather than three: the aperture migrates from the front of the shell around to the side, parallel to the coiling axis, narrowing as it goes, which makes the shell more cylindrical. The spire does not shrink so much as get swallowed by the last whorl. The earlier whorls are still in there, overgrown, their seams smoothed over, which is why the outside carries almost no record of its own coiling.

The operculum went with it. Most sea snails carry one, the door on the back of the foot that seals the aperture. Marginellids do not. Character 14 ties the loss to the aperture change: the standard gastropod arrangement folds the foot across its width, which positions a door over a roughly round hole, while marginellids fold the foot lengthwise, which fits a long narrow slot. Character 13 adds that the foot retracts about three quarters of a whorl, not far, because there is not much shell behind the aperture. An empty marginella never had a door in it.

One tooth, or none at all

Marginellids are predators and scavengers on small animals. That is the safe description and where it should stop.

The rasping tongue, the radula, is reduced about as far as it goes while still existing: uniserial and monodont, one central tooth per row, side teeth gone. Coovert and Coovert named that the "type 6" radula in 1990, against the broader "type 7" of the deepwater giant Marginellona gigas. The two main sources on this family disagree about the average cusp count on that single tooth, so no number appears here.

The stronger finding is that some marginellids have given up the radula entirely. Souza's matrix records complete loss in three species, and Boyer refers to full loss in some marginellid lineages. Two of the three are recorded from the Gulf of Mexico. One honest wrinkle: the anatomical literature files those two under Eratoidea while the World Register places them in Dentimargo, and the equivalence rests on matching epithets rather than a published synonymy. Very likely, not settled.

What marginellids do not do is inject venom. A claim circulates online about a paralytic venom delivered through the proboscis, with no literature behind it. They are volutoideans, not cone snails, with no harpoon and no venom gland of that kind. The cone snail page covers the group that genuinely carries that equipment.

The marginellas on record in the Gulf

Every name below is accepted in the World Register of Marine Species, with its AphiaID, and each carries a valid Gulf of Mexico distribution record except where the note says otherwise.

The one most people meet is Prunum apicinum. The Bailey-Matthews National Shell Museum describes it as reaching about 10 mm, pear shaped, highly polished, with four folds on the columella and a thickened outer lip usually carrying two small brown spots, and very common in seagrass beds. Everything above applies to it: fixed adult size, terminal lip, structural gloss.

One correction, because regional lists get it wrong constantly. Prunum and Volvarina are not close relatives inside the family. Prunum sits in the subfamily Pruninae, while Volvarina, Dentimargo and Eratoidea sit in Marginellinae. The family currently holds four accepted subfamilies: Marginellinae, Pruninae, Granulininae and Austroginellinae.

Names in this family that will trip you

• Marginella is essentially not a Florida genus. True Marginella Lamarck, 1799 is centered on West and South Africa. Every Florida species once filed there has since moved to Prunum, Volvarina, Dentimargo, Hyalina or Eratoidea. "Marginella apicina" is a superseded combination, not a name in current use.
• Gender agreement. The Bailey-Matthews National Shell Museum, this coast's reference institution, lists Dentimargo eburneolus and Dentimargo aureocinctus. The register rejects both masculine forms. Aureocinctus is flagged unaccepted for wrong grammatical agreement of the species epithet, and eburneolus returns no record at all. The accepted spellings are eburneola and aureocincta. That is a Latin agreement update, not an error of expertise.
• The avena homonym. Two authors used the name Marginella avena. Kiener's, from 1834, is the Gulf animal and resolves to Volvarina avena. Sowerby's, from 1846, is an invalid homonym that resolves to a different species entirely. Written without its authority, "Marginella avena" is genuinely ambiguous and should never be used bare.

Which family is this, exactly, and what is still argued

The marginelliform snails, the glossy small ones as a visual group, have been split, merged and split again, and plenty of regional lists still run on the old arrangement.

Stimpson erected Cystiscidae in 1865. Coan demoted it in 1965, folding it in as a subfamily of Marginellidae. Coovert and Coovert re-split it in 1995 in The Nautilus, arguing that the cystiscids resemble marginellids only superficially and sit closer to the olives than to the volutes. That split is what the World Register follows now: Cystiscidae is an accepted family, separate from Marginellidae. Which is why Persicula, Gibberula and Cystiscus are cystiscids, not marginellids. Marginellonidae and Plesiocystiscidae have also been split off, and both are accepted.

Three questions are genuinely open, and this page will not pick winners.

• Granulina. The World Register places it inside Marginellidae, as the subfamily Granulininae, and carries the family-rank name Granulinidae as a superseded rank. Boyer's 2019 review ranks it as a full family outside Marginellidae instead. Both agree it is closer to the marginellids and volutes than to the cystiscids, so the disagreement is purely about rank.
• How many families. Boyer ends at five; the register accepts four of those five. He also concludes that the marginelliform group as a whole is polyphyletic, which is to say the family resemblance is not by itself evidence of relatedness.
• The genera themselves. The abstract of Souza and Simone's 2019 revision states that Prunum, Marginella and Dentimargo are non-monophyletic, and raises the possibility that Marginellidae is only a branch of Volutidae. That paper is paywalled and only its abstract was available here, so it is reported exactly as its authors summarized it. They are the accepted names. They may not be the final ones.

Beads, and an object in Oxford

The architecture described above turns out to make an almost ideal bead.

The Maryland Archaeological Conservation Laboratory, a state archaeology lab, identifies the bead species as the common Atlantic marginella, Prunum apicinum. The manufacture is the detail that matters here: the holes were usually made by grinding the shells on rocks or gouging them with a tool, not by drilling. The wall is thin enough that abrasion opens it. The beads were worn as ornaments, woven into belts, sewn onto garments, and used as currency in the seventeenth century.

The most famous object made this way is Powhatan's Mantle, in the Ashmolean Museum in Oxford. The Ashmolean describes four white-tailed deer hides sewn together with sinew, the shell beadwork forming a central standing human figure flanked by two opposed four-legged animals and surrounded by thirty-four circles. The museum dates it to about 1600 to 1638, places its likely making in the southern Chesapeake, and records it in the Tradescant collection in London by July 1638. The leading explanation for how it reached England, offered as an explanation rather than a fact, is that it was among gifts presented by Powhatan to Captain Newport for King James I in 1608. It is named for Wahunsenacawh, and the museum does not assert it was personally his.

Two figures repeated everywhere about this object are not in the museum's own account, which names no shell species and gives no shell count. The quoted "Marginella roscida" and "nearly 20,000 shells" come from secondary sources, none citing a published identification. That name resolves to Prunum roscidum, the seaboard marginella, which carries no Gulf of Mexico distribution record and is the more northerly animal. A Chesapeake object made from a Chesapeake-range species is entirely coherent, so the sources genuinely differ here.

The same idea, on the other side of the Atlantic

West Africa has a parallel, handled more carefully here because the sourcing is thinner. The scholarship reporting it, in the archaeology of cowrie currency, could not be read in full for this page, so what follows is secondhand.

What is reported is that marginella shells turn up at inland West African sites, mostly trading settlements, and that historical accounts describe them used as currency at Timbuktu in the eighteenth century, when war or insecurity cut the cowrie routes. A fallback currency, in other words, not a first choice. The same literature reports that Theodore Monod collected ten thousand of them at Timbuktu in May 1935, identified then as "Marginella amygdala." That figure is secondhand too, repeated here as a report rather than a verified count.

The taxonomy is checkable even where the history is not. "Marginella amygdala" is unaccepted; the accepted name is Prunum amygdalum (Kiener, 1841), AphiaID 473945. So the West African currency shell and the North American bead shell are both Prunum, the same subfamily, on opposite sides of an ocean, put to the same use by people with no contact.

Why this family and not another is what the rest of this page answers. Determinate growth delivers shells that all arrive at the same finished size, so beads match without sorting. The wall is thin enough to open by abrasion, so no drill is needed. And the gloss is built into the surface rather than kept up by a living animal, so it does not go dull in a belt.

To see the water these animals live in, reserve a seat. The Shell Monographs hub collects the other families, and how a shell is actually built covers what a mantle normally does.