Revision of Agathodesmus Silvestri , 1910 ( Diplopoda , Polydesmida , Haplodesmidae )

Agathodesmus Silvestri, 1910 includes A. baccatus (Carl, 1926) comb. n. from New Caledonia, A. bucculentus (Jeekel, 1986) comb. n. from Queensland, Australia, and A. johnsi sp. n. and A. steeli Silvestri, 1910 (type species) from New South Wales, Australia. A. baccatus and A. bucculentus were formerly placed in Atopogonus Carl, 1926 syn. n. Th e identity of the apparently congeneric Inodesmus jamaicensis Cook, 1896 sensu Loomis, 1969 from Jamaica is still uncertain, and Inodesmus Cook, 1896 remains a nomen inquirendum.


Revision of Agathodesmus
Introduction Filippo Silvestri (1910) erected the polydesmidan genus Agathodesmus Silvestri, 1910 for a single female specimen from Australia.Because millipede taxonomy is based largely on male genitalia, the classifi cation of Agathodesmus and its type species A. steeli Silvestri, 1910 has been uncertain for almost 100 years.
In this paper I redescribe A. steeli from the holotype and from recently collected, nearly topotypical males and females.A second Agathodesmus species is described from a site ca.180 km from the A. steeli type locality.Th e two species are shown to be congeneric with species of Atopogonus Carl, 1926 in Haplodesmidae.

Materials and methods
Th e type locality of A. steeli Silvestri (1910) reported that the A. steeli holotype was collected by Th omas Steel at Avoca, New South Wales (see Appendix).No additional information is on the holotype slide label (Fig. 1).At the beginning of the 20th century there were three New South Wales places called "Avoca": one in the dry southwestern corner of the State, one in the highlands west of Wollongong and one on the coast near Gosford.Th e latter "Avoca" is today known as Avoca Beach.
Independent evidence that the type locality is the highlands Avoca comes from records concerning Th omas Steel (1858Steel ( -1925)).Steel was an industrial chemist and was based in Sydney from 1893 to 1918 (Carter 1926).In his spare time Steel was a fi eld naturalist with a taxonomic interest in Onychophora and terrestrial fl atworms; he donated many specimens to the Australian Museum in Sydney (G. Milledge, pers. comm.).Th e Museum's collection database lists earthworms, onychophorans, terrestrial fl atworms, frogs, lizards and snakes collected by Steel between 1895 and 1908 from Bundanoon, Moss Vale, "Avoca, near Moss Vale", Wildes Meadow and Robertson.Th ese fi ve localities are clustered in the southern highlands of New South Wales (Fig. 2C).In Steel's time they were easily accessed by train from Sydney, and the area had been popular with rail tourists since the 1880s (Garran 1886).Th ere are no Steel collections in the Australian Museum from either of the other two "Avoca" localities in New South Wales (G. Milledge, pers. comm.)Th e date of the A. steeli collection is unknown, but the Australian Museum collection database lists a terrestrial fl atworm collected by Steel at Avoca in February 1905, and a leech, an onychophoran and fl atworms collected by Steel at Avoca in January 1907.Both dates are consistent with the 1910 date for Silvestri's description.I have been unable to locate any correspondence between Steel and Silvestri, nor have I found any other information indicating how Silvestri acquired Steel's specimen.
Fig. 2C shows that Avoca (a named locality, not a town) is only ca. 4 km and 14 km, respectively, from the Steel localities Wildes Meadow and Moss Vale, and only 4 km from Fitzroy Falls, a tourist destination that was as well known by that name in Steel's time (Garran 1886) as it is today.It is likely, then, that the place name "Avoca" as used by Steel refers to a small area, perhaps only a few square kilometres.

search for A. steeli
Th e Avoca locality today is cleared farmland on a small patch of Tertiary basalt (Fig. 9 in Bowie 2006).Although closed rainforest grew on basalt further to the east, the isolated Avoca basalt patch is thought to have carried open eucalypt forest, not rainforest, prior to clearing in the mid-to late 19th century (K.Mills, pers. comm.;Fig. 3.1 in Mills and Jakeman 1995).Th omas Steel may have collected A. steeli in an intact forest remnant at Avoca, or among logs and other woody residues of clearing operations.
On 6 and 7 April and 14 May 2009, I searched for A. steeli at the forested places near Avoca marked in Fig. 2C.I looked under stones, under loose bark on fallen trees, in and under rotting wood on the ground, and in fallen leaf and bark litter.
Th e fi ve April sites carried open eucalypt forest on soils derived from shale and sandstone.Th ese soils are relatively infertile and there is a large area of never-cleared forest on shale and sandstone south and west of Avoca.Spirobolida, Spirostreptida and paradoxosomatid Polydesmida were abundant at the fi ve sites, with scattered occurrences of dalodesmid Polydesmida, Siphonophorida and Polyzoniida.I saw no Polydesmida in the A. steeli size range (ca.6 mm long) other than early-stadium juvenile paradoxosomatids and dalodesmids.
Th e four May sites were all on basalt and carried either closed rainforest or tall eucalypt forest with a variably dense understorey of small trees.Uncleared forest on fertile basalt soil is rare in the area, and the remnants I searched have survived largely because the ground under the trees is too stony for farming.Th e millipede fauna appeared to be the same at the basalt sites as at the sandstone and shale sites, but the dalodesmid Orthorhachis christinae Mesibov, 2008 was noticeably more abundant.I found 11 adults and four juveniles of A. steeli in a single rotting log in eucalypt forest at Knights Hill, ca.20 km east of Avoca.No other small, adult Polydesmida were seen at the basalt sites.

Other specimens
Th e Australian Museum in Sydney holds many millipede samples from New South Wales, mostly sorted to order or below.As part of a paradoxosomatid mapping project in 2006-07 (Mesibov 2008), all samples sorted to Polydesmida were examined by Cathy Car (Charles Sturt University) or myself.We sorted these further to Dalodesmidae, Paradoxosomatidae and "other".Among the latter samples I found specimens of a second Agathodesmus species collected in 1966 from the Brindabella Ranges west of Canberra, at a site ca.180 km west and south of Avoca (Fig. 2B).
Th e Australian National Insect Collection in Canberra holds a small number of unsorted samples obtained from Berlese extraction of forest litter from the area I searched (Fig. 2C).I examined these samples but found no Agathodesmus.

Note on geography
Th e new Agathodesmus species described below was found on the summit of Mt Aggie.According to the Australian Museum database, Mt Aggie is in the Australian Capitol Territory, which is embedded within the much larger Australian state of New South Wales (Fig. 2B).Th e Commonwealth of Australia gazetteer (searchable online at http:// www.ga.gov.au/place-name/)locates Mt Aggie in both the Australian Capitol Territory and New South Wales, because the mountain straddles the border between the two, and the border crosses the summit.Th e new species undoubtedly occurs on both sides of the border on Mt Aggie, and for practical reasons is here regarded as a New South Wales species.

Specimen treatment
"Male" and "female" in the text refer to stadium VII individuals unless otherwise indicated.Specimens are stored in 80% ethanol in the Australian Museum.Some specimens were cleared in 80% lactic acid, then temporarily mounted in 60% lactic acid for optical microscopy.SEM images were acquired digitally using an FEI Quanta 600 operated in high-vacuum mode; alcohol-preserved individuals and body parts were air-dried before sputter-coating with gold.
Diagnosis.Small Polydesmida with head and 19 or 20 rings; body not curling in spiral; head and telson facing downwards; metatergites with numerous tubercles of different sizes, the largest sometimes bearing a single seta; ring 2 tergite extended laterally, basally and anteriorly, and edged with large tubercles; no paranota on posterior rings, sometimes replaced by short row of large tubercles just above leg bases; gonopod with neither cannula nor prostatic groove, telopodite consisting of more or less cylindrical basal portion with broad, fl attened structure arising posterodistally on basal portion of telopodite and bent basally or basolaterally.
Diagnosis.Head + 19 rings; gonopod telopodite with distal portion directed basally and slightly laterally near origin and with broad lateral branch apically expanded and divided into three anterobasally curving lobes.
Description.Th e original description (Silvestri 1910) is quoted in the Appendix.What follows is based on my examination of both the holotype and the Knights Hill material.
Adult with head + 19 rings (Fig. 4A).Live and freshly preserved adults pale with faint reddish pigmentation dorsally; in some individuals, pigment concentrated in transverse band at rear of metazonite.Male/female approximate dimensions: length  3D) about as wide as collum, overall body width almost uniform, tapering only slightly posteriorly from collum.Head facing downwards (Fig. 3C), with clypeus, frons and ventral part of vertex almost parallel to substrate and only slightly convex.Antenna (Fig. 3D) short, stout, clavate, held close to head, antennomeres 2 and 3 lying in broad, shallow excavation on head; antennomere 6 widest and longest; antennomeres 2-5 about equal in length, decreasing slightly in diameter from 5 to 2. Collum with slightly convex anterior margin and broadly convex posterior margin; corners rounded (Fig. 3C).Ring 2 tergite largest, extending basally, later-
Integument richly and densely sculptured (Figs 3D, 4C, 5A, 5C, 6A-C, 7A, 7B).Most of body covered with cuticle raised in cellular mesh of narrow folds, often with minute bumps (adorned with even smaller bumps) at or near fold junctions.Integument raised further as tubercles (Fig. 6A) of varying sizes on head, collum, tergites, metatergites and telson, the largest tubercles forming paranotum-like extensions on posterior rings; tubercles and some other parts of integument with minute, fi nger-like projections (Figs 4D, 6A, 6B), often arising along 'mesh-cell' boundaries.Cell boundaries at rear of metazonite extended as lappets, forming secondary limbus above primary limbus of uniform, triangular elements (Fig. 6C).Setae of normal type on legs and some other surfaces; a bisegmented seta with fl attened, expanded tip (Fig. 6B) on each 'paranotum' tubercle and in association with some dorsal tubercles.
Male with gonopore opening at tip of cylindrical projection about 1/3 the length of leg 2 coxa, arising distomedially on the coxa.First legs somewhat swollen (Fig. 3C), no other anterior legs enlarged; neither sphaerotrichomes nor brush setae on any legs.Leg 7 bases well separated; leg 6 bases slightly separated, with a pair of short, rounded projections between coxae.Gonopod aperture oval (Fig. 7A), rim a little raised laterally.Gonocoxae (Fig. 7A) occupying full width of aperture; tapering a little distally; with mesh-like integumental sculpture and without setae; fi rmly joined medially near distal end.Telopodite (Fig. 7A) short, compact, when retracted reaching leg 7 base; broadly joined to gonocoxa (Figs 5B, 7A); no trace of cannula or prostatic groove; no integumental sculpturing; divided into more or less cylindrical basal portion and fl attened distal portion.Basal portion of telopodite with blunt, basally directed projection arising posteromedial to junction with gonocoxa; portion terminating in fl at, rounded tab bending posteriorly; with a few short setae on basal half of posterior surface of portion and three large setae in a row on lateral edge of posterior surface of terminal tab.Distal portion of telopodite a large, fl attened structure arising on posterior surface of basal portion of telopodite just below terminal tab; curving basally and slightly laterally; divided near base into narrow medial branch, fl attened apically with minute, spine-like protrusions on posterior and medial surfaces, and much larger lateral branch, the latter  7A).Female longer and more robust than male (Fig. 4A); epigynum inconspicuous, posterior margin barely raised; cyphopods not examined.
Distribution and habitat.Known so far from eucalypt forest (historically in the case of Avoca) at two localities ca.20 km apart in southeastern New South Wales  2C).Both sites are above 700 m with annual rainfall probably >900 mm, in a temperate climate with cool winters.At the Knights Hill site, the 15 A. steeli specimens were found in narrow spaces in part of a large, moist, well-rotted log, either a Eucalyptus species or Acacia melanoxylon.Also in that part of the log were Siphonophorida, Symphyla, Cryptops sp.centipedes, fl y and beetle larvae and terrestrial isopod crustaceans.
Remarks.Live A. steeli are very slow-moving and do not curl up, even when disturbed.Unlike adults of the morphologically, ecologically and behaviourally similar species of Asphalidesmus Silvestri, 1910(Mesibov 2002, 2009), A. steeli adults are not heavily encrusted with soil particles.
Th e apparent absence of well-defi ned spiracles in A. steeli is remarkable.I have so far been unable to detect spiracles either with light microscopy (cleared specimens) or scanning electron microscopy (see also Fig. 5C).A histological study is needed to determine whether the tracheal system is also modifi ed from the norm in Polydesmida.Paratypes.12 males, 12 females, details as for holotype, AM KS94156 (two males dissected).

Agathodesmus johnsi
Diagnosis.Head + 19 rings; gonopod telopodite with distal portion directed laterally near origin and with broad lateral branch without notches, apically tapering and curving basally rather than anterobasally.
Description.Colour uniformly light yellow-brown after long preservation.Males and females as for A. steeli in all details so far noted, including apparent absence of spiracles (Fig. 5C), but male gonopod telopodite (Fig. 7B) with more short setae on posterior surface of basal portion, terminal tab narrower and less bent posteriorly, and distal portion directed laterally near base rather than posterolaterally, with narrow medial branch curving to lie against broad lateral branch, the latter undivided, curving basally and tapering to bluntly rounded tip.
Distribution and habitat.Th e only known locality is the summit of Mt Aggie at ca. 1500 m, where according to the collector, "Th e site was in scrubby high altitude Eucalyptus, a few bits of snow were around and the ground was quite damp" (P.M. Johns, in litt.).
Etymology.Adjective, genitive singular, for the collector Peter M. Johns.

Identifi cation of A. steeli
Th e A. steeli type is mounted on its side and has evidently degraded over 100 years; not all of its characters are clearly visible.However, the type agrees with adult Knights Hill females in size, number and shape of rings, orientation of head and telson, and sizes and shapes of antennomeres and podomeres.More importantly, both the type and the Knights Hill specimens have variably sized tubercles with microsculpture, bisegmented setae, 5-6 large tubercles edging the ring 2 tergite (Figs 3A-C) and paranotum-like, four-tubercle lateral extensions on rings 3-18.Th e Knights Hill specimens are clearly in the same genus as the type.It is still possible that they are not conspecifi c with A. steeli, which was collected ca.20 km to the west in forest habitat lost many years ago.If the Knights Hill form also occurs in the remnant forests south and west of Avoca, where so far no Agathodesmus specimens have been found (Fig. 2C), I would be more confi dent in its identifi cation with A. steeli.
It is curious that Silvestri (1910; see Appendix) noted setae arising laterally in the type, but did not record the more obvious fact that the lateral extensions and the ring 2 tergite have lobed margins.

Agathodesmus and Atopogonus as synonyms
As indicated in the Agathodesmus diagnosis (above), A. johnsi and A. steeli share several non-genitalic apomorphies with the two species described under Atopogonus.All four species also have inconspicuous ozopores located in non-tuberculated 'clear' zones low on the metatergites (Fig. 4B; Fig. 24 in Carl 1926;Jeekel 1986, p. 47).Th e most striking similarities, however, are in details of gonopod telopodite structure, as seen in Figs 8A -8E.Th ese include a rounded terminal tab on the basal portion with three long setae on the posterolateral surface, and the division of the distal portion by notches into lobes in A. baccatus, A. bucculentus and A. steeli.Th e similarity in telopodite structure of the last two species is particularly striking (compare Fig. 7A with Figs 8D and 8E), and justifi es placing them in the same genus.
Like Agathodesmus, the pyrgodesmid genus Poratia Cook and Cook, 1894 contains species with either 19 or 20 body rings (Golovatch and Sierwald 2001).Adis et al. (2001) suggested that the 19-ringed Poratia digitata (Porat, 1889) form could have evolved by neoteny from a larger, 20-ringed ancestor.Similarly, the 19-ringed ancestor of the miniscule A. johnsi and A. steeli might have evolved from a 20-ringed lineage represented today by the larger A. baccatus and A. bucculentus.Carl (1926) assigned his A. baccatus to Rhachidesmidae Carl, 1903.Th e remarkable lack of both a cannula and a prostatic groove inspired Verhoeff (1941) to establish Atopogonidae Verhoeff , 1941 for this species within the superfamily Rhachidesmidea Verhoeff , 1941, which also included Rhachidesmidae. Hoff man (1980) regarded Atopogonus as a genus of uncertain family position within the suborder Polydesmidea, but Jeekel (1986) placed it in Haplodesmidae.Hoff man (1999) agreed, but nevertheless felt that "the gonopods present a singular and highly disjunct pattern that invites taxonomic recognition at a level no less than subfamily (or family)" (p.483).Golovatch et al. (2009) were unwilling to distinguish Atopogonus in this way, and listed it as one of six genera in Haplodesmidae after a careful revision of the family.

Family placement of Agathodesmus
Here I follow Golovatch et al. (2009) in leaving Atopogonus, now synonymised with Agathodesmus, in Haplodesmidae.A haplodesmid character state demonstrated here for A. johnsi and A. steeli is bisegmentation of setae on the tergites (Golovatch et al. 2009); whether the other two described species in the genus have bisegmented setae is yet to be determined.

Th e Inodesmus problem
Th e name Agathodesmus may fall into synonymy with Inodesmus Cook, 1896 if "an intriguing and very diffi cult problem" (Hoff man 1999, p. 483) can be resolved.Th e problem has been briefl y discussed by Jeekel (1986) and Hoff man (1999).Here I explore the problem in more detail.
In his self-published journal Brandtia, O.F. Cook added the new genus Inodesmus to Comodesmidae Cook, 1896 with the following words (Cook 1896b, p. 25): From a cave in Jamaica I have specimens of a genus related to Comodesmus.
Diff ering from Comodesmus in the somewhat more slender, moniliform body, obsolete carinae, more projecting last segment, and normal pore-formula, the pores located in shallow depressions in the lateral middle of the segments, not in front of the middle as in Comodesmus.Th e only known species, I. jamaicensis, is about equal in size to Comodesmus lanatus, and is lighter brown in color, but may be faded.Th e type of this family is a small, reddish-brown, subcylindrical form, very rare, and also inhabiting the denser parts of the forest [in Liberia].Th e pore formula is unique: 5,7,9,12,15,17,18.Th e pores are located in the front part of the posterior subsegments.Th e dorsal surface is beset with conic piliferous granules, giving a woolly appearance.Th e last segment is scarcely produced beyond the anal valves, but is rounded off at apex as in many Iulidae.Th e head is not concealed by the fi rst segment, which is narrower than the second and somewhat included between the carinae of the latter, much as in Scytonotus granulatus (Say).(Cook 1896a, p. 415) Comodesmus lanatus.Antennae distinctly clavate; last segment decurved, the immediate apex small, projecting, truncate; lateral carinae present only as a longitudinal row of large tubercles, above which the tubercles are gradually smaller; length 8 mm., width 1 mm.(Cook 1896c, p. 258) Th e location of the types of I. jamaicensis Cook, 1896, if they still exist, is not known.It is particularly ironic that the types may have been lost, considering that Cook understood the value of type specimens: Th e importance of preserving type specimens with special care is now recognized throughout the scientifi c world, and where specifi c types are lacking, naturalists are endeavoring to supply their place by specimens collected in the original localities.Th is may be taken as a general admission of the obvious fact that purely descriptive methods are generally insuffi cient for scientifi c accuracy and need to be supplemented by actual specimens if correct identifi cations are to be permanently assured.(Cook 1900, p. 481) Nearly 40 years after the establishment of Inodesmus, H.F. Loomis (1934) described another species in the genus from a set of females collected in Dutch Guiana (now Suriname).At the same time Loomis synonymised the monotypic Lasiodesmus Silvestri, 1908 with Inodesmus.Th ere were now three species in Inodesmus: I. jamaicensis (Jamaica), I. peduncularis Loomis, 1934 (Suriname) and I. caraibicus Silvestri, 1908 (Puerto Rico).Discussing the new synonymy, Loomis wrote After comparing the present species, from a generic standpoint, with Silvestri's description of Lasiodesmus and with the brief characterization of Inodesmus Cook, there appears to be no reason for maintaining Silvestri's genus... Th e question of the distinctness of Inodesmus jamaicensis Cook and I. caraibicus (Silvestri) cannot be decided until comparison is made of the types or of specimens undoubtedly similar to the types.(Loomis 1934, p. 65) It seems clear from these statements that Loomis had not examined the types of I. jamaicensis.He was confi dent that his I. peduncularis and L. caraibicus were conge-neric, because he had specimens of the former and a clear, beautifully illustrated description of the latter (Silvestri 1908).However, he was also confi dent that both were congeneric with I. jamaicensis, for which he had neither specimens, nor illustrations, nor an adequate description.
Th irty years later Loomis (1964) added another species to the genus, I. globulosus Loomis, 1964 from Panama, with only minor diff erences distinguishing it from I. peduncularis.
It is not clear whether Cook (1896b) had examined males of I. jamaicensis, but we know from the original descriptions that the three species assigned by Loomis to Inodesmus were based on all-female samples.In the late 1960s, samples of Inodesmus spp.including males were collected by Stewart Peck in Jamaica and Panama.Th e Jamaican millipedes, from two caves, were identifi ed by Loomis (1969) as I. jamaicensis and redescribed.Th e gonopods of the Jamaican species were very diff erent from those of the males collected by Peck in Panama and identifi ed by Loomis as I. globulosus, so Loomis (1969) transferred the latter species to a new genus, Hypsoporus Loomis, 1969.Th ere is no evidence that Loomis compared the new Jamaican specimens with types of I. jamaicensis, but he was confi dent that they were conspecifi c: Also present [in the Peck collection] were both sexes of the genotype species, Inodesmus jamaicensis, which O.F. Cook (1896) diagnosed very briefl y in erecting the genus and which species has not been reported since.(Loomis 1969, p. 141) Th e exceedingly brief description of this species, its type locality given merely as "a cave in Jamaica," failed to mention many of the following characters which are of importance.(Loomis 1969, p. 144) Th e other two Inodesmus species, I. caraibicus and I. peduncularis, have since been shown to be synonyms, like Hypsoporus globulosus, of the pantropical "tramp" species Cylindrodesmus hirsutus Pocock, 1889 (Golovatch et al. 2001).Th e question remains: did Loomis (1969) describe Cook's I. jamaicensis, or a species in a diff erent genus?Th e descriptions of I. jamaicensis by Cook and Loomis agree only in body length (8-9 mm) and in the position of the ozopore (near the transverse midline of the metazonite).If we assume that Cook's I. jamaicensis shares with his Comodesmus lanatus those features not said to be diff erent, then Cook's I. jamaicensis appears to agree with Loomis' I. jamaicensis in having the dorsal surface covered with more or less conical tubercles bearing long setae, and in having the paranota of ring 2 extending slightly forward.It does not seem enough on which to base an identifi cation, and the fact that both came from Jamaican caves is hardly relevant, since Loomis (1969) says that his specimens have no "modifi cations indicating restriction to cave life" (p.141), and Cook's specimens were pigmented.
Th e identity question became important when it was recognised by C.A.W. Jeekel that the gonopod of Loomis' I. jamaicensis (Fig. 8F) closely resembled that of the two Atopogonus species: "it is quite obvious that Inodesmus jamaicensis sensu Loomis, 1969, is a species congeneric with Atopogonus baccatus and the presently described species [A.bucculentus]" (Jeekel 1986, p. 46).However, Jeekel (1986) was not convinced that Loomis had redescribed I. jamaicensis, and he refrained from synonymising Atopogonus with Inodesmus.R.L. Hoff man (1999) was also unconvinced by Loomis' identifi cation.Further, he was puzzled by the geographic disjunction between Australian and New Caledonian Atopogonus and Jamaican Inodesmus sensu Loomis, 1969. Jeekel (1986) had suggested that the genus was "in essence...a continental Australian taxon" (p.46) which might have been carried by humans to New Caledonia and Jamaica.Hoff man (1999) reported that there was a second, undescribed species of Inodesmus sensu Loomis, 1969 in Jamaica's Blue Mountains rainforest, and argued: "While a multiple transport of rare and localized species from the Antipodes to a single West Indian island is not impossible, it does appear improbable.How else, then, can this distribution be accounted?If natural, it can only represent an astonishing case of reliction of a formerly widespread parental lineage."(p.483).
If Cook's I. jamaicensis types are indeed permanently lost, then I. jamaicensis sensu Loomis, 1969 could be renamed as a species of Agathodesmus, leaving Inodesmus as a nomen inquirendum.Alternatively, it could be assumed that Cook's types are lost and that Loomis' identifi cation is correct.A neotype of I. jamaicensis Cook, 1896 could then be selected from among the specimens examined by Loomis (1969), as suggested by Golovatch et al. (2009), and Agathodesmus would become a junior subjective synonym of Inodesmus.
A third and more satisfying possibility would be to fi rst make a thorough inventory of Jamaican Polydesmida, both in and out of caves.If it could be shown that the only Jamaican genus fi tting Cook's description is the one represented by the species described by Loomis (1969), then the three known Agathodesmus species should be moved into the older genus Inodesmus.I suspect, however, that there are other genera of small Jamaican Polydesmida with dorsal tuberculation and an anteriorly extended ring 2 tergite, and it is possible that Cook (1896) may in fact have redescribed Cylindrodesmus hirsutus as I. jamaicensis.
In this paper I leave the Inodesmus problem unresolved.

Since
Cook assigned Inodesmus to Comodesmidae Cook, 1896 and compared it with the the type species of Comodesmus Cook, 1896, it is worth examining what Cook wrote about these taxa.I have corrected minor spelling and typographical errors in the following extracts: Family Comodesmidae, new.