The Arhopala butterflies described by Fabricius: A. centaurus is from Java, A. democritus from Phuket (Lepidoptera: Lycaenidae)

Taxonomic history

Papilio centaurus Fabricius, 1775, has become one of the most thoroughly misinterpreted of Fabrician butterfly names. As related below, Steven Corbet (1941a) must bear some responsibility for initiating the recent history of errors. Usage for centaurus was previously well established, although not entirely soundly, by Fabricius (1793), Godart (1824), Horsfield (1829), Hewitson (1862), Butler (1870), Distant (1885), and de Nicéville (1890). Among the 19th century authors, only Doubleday (1847) made mistakes similar to those introduced by Corbet. Despite Corbet’s unfortunate intervention, the correct application of the name centaurus to an Oriental insect was otherwise universally maintained right up to Evans (1957), by most authors until the 1990s, and by some to this day. So far as we have been able to ascertain, after Doubleday (1847) and Corbet (1941a), the first recent author to apply centaurus to an Australian butterfly appears to have been D’Abrera (1977), acting on well-meant but incorrect advice received from the late John Eliot and Charles Cowan. Since then, more and more have followed this incorrect path. It is this confusion that necessitates the historical review presented here.

Fabricius (1775), who worked in London during the summer months of 1772–1775 (Hope 1845: viii), originally described Papilio centaurus from an unspecified number of specimens from the Banks Collection, and gave its origin as “Habitat in nova Hollandia” (=Australia). His entire Latin description, in its original layout, is reproduced here:

The Banks Collection (BMNH) now includes two very old specimens over the name centaurus. In addition to this obvious source for original material, we have searched the Sehested and Tønder Lund collection (ZMUC) and the HM, where we also found very old Arhopala specimens associated with the name centaurus. Banks allowed Fabricius to take certain duplicates, and on occasion Fabricius passed some of these to other colleagues (Fabricius 1784: 123). As noted in his autobiography, Fabricius had close links with Banks and the Hunter brothers while in London (Fabricius 1805; see Hope 1845: vi, viii), and with Niels Tønder Lund in Copenhagen (Hope 1845: xv). However, as detailed below, the Glasgow and Copenhagen specimens are not conspecific with authentic centaurus, nor is there any proof that these specimens originated from the Banks collection, and so we exclude them from the type series.

Fabricius (1775, 1781, 1787) described the butterfly as “blue with fuscous edge” (“. . . coerulescentibus; limbo fusco . . .”). Later, Fabricius (1793) moved centaurus from Papilio to Hesperia, and gave a slightly extended description. Although the color expression “shining dark purple-blue” or “violaceous-blue” used by modern authors to describe the male of P. centaurus has no parallel in any of Fabricius’s descriptions, it was primarily his citation of Australia as the provenance for centaurus that ultimately led to erroneous interpretations.

Following Fabricius (1775), the next person to study the Banks Collection was William Jones, who illustrated both the upper and under-surfaces of a specimen of Papilio centaurus as a Fabrician species in his celebrated, but unpublished Icones (volume 6: pl. 22, fig. 6). Although the Icones are undated, all six of the Jones volumes seen by Fabricius during 1787 appear to have been completed well before his return to London in the summer of that year (Hope 1845: xii). According to Salmon (2000: 120), the illustrations were all made by Jones in the period 1783–1785. Fabricius had access to Jones’s paintings throughout his 1787 visit (E.B. Poulton 1938 in Waterhouse 1938: 149). However, his first references to the Jones illustrations and new descriptions based on them did not appear until 6 years later, when he next published on butterfly taxonomy (Fabricius 1793).

Banks’s collection was bequeathed to the Linnean Society, arriving there in 1823 from Banks’s London residence in Soho Square. Horsfield (1829: 102, 103), while identifying the EIC collections that he and others had made in Java (for an account of the history of the EIC, see Cowan 1975), compared their specimens to the Banks material, commenting at the end of his Latin diagnosis of Amblypodia centaurus: “Hospitatur in Musæo Domini Banks, Societati Linneanæ Londinesi munificè donato, nomine à Fabricio ipso inscripto” (our italics). On the next page he states in English, “I have been able to identify this species with the Hesperia centaurus of Fabricius (1793: 275), by the examination of a specimen contained in the Banksian Museum, bearing a ticket in his own (Fabricius’s) hand-writing,” going on to note that the EIC collection contained “three male and six female specimens of this very beautiful species.”

Thus the identity of Papilio centaurus, at least to Horsfield, was certain. For him the “type” specimen that he examined in the Banks Collection (LS), the EIC material from Java, and Fabricius’s description were all positively established as referring to one and the same species from the Oriental region. However, Horsfield did not specifically state that, in his opinion, the Banks material came from Java.

No further confusion would have arisen if Doubleday, while compiling his List of . . . Lepidopterous Insects in . . . the British Museum, had followed the positive conclusion of Horsfield. However, on the testimony of Fabricius’s published “habitat” of Nova Hollandia, Doubleday (1847) displaced centaurus (together with Godart’s 1824 citation) to Australia. He then named Horsfield’s nine specimens (by reference to Horsfield’s description of centaurus), together with other material from Java, India and Sri Lanka listed as in the BMNH, as a new species, Amblypodia pseudocentaurus. Doubleday did not mention (or apparently study) the putative type specimen in the Banks Collection. When Doubleday made his list, the Banks Collection was still in the LS and Horsfield’s collection was still at the EIC. However, it would appear from Doubleday’s text that at least one Javanese specimen from the EIC had by then been presented to the BMNH, most probably one of the nine specimens recorded by Horsfield (1829). Doubleday was also able to examine a specimen of what he believed represented the true Fabrician Hesperia centaurus, given to the BMNH by the Earl of Derby and originating from Repulse Bay, Queensland. Unfortunately, we have been unable to locate this specimen and so ascertain its identity according to current taxonomy.

Horsfield and Moore (1858) did not accept Doubleday’s notion that Fabricius’s taxon was from Australia. However, by recording two pairs of Amblypodia centaurus, one from “Cherra Poonje” (ex Buckley) and the other from “Darjeeling” (ex Schlagintweit), and at the same time recognizing Doubleday’s pseudocentaurus as a separate species from Java, they effectively introduced a third concept for the species. For pseudocentaurus they listed three Javanese specimens, representing both sexes, “from Dr Horsfield’s Collection” in the EIC. These three are now in the BMNH: two females and one male. Presumably they are a subset of the three males and six females originally recorded by Horsfield (1829). During the ensuing 30 years, four females and two males must have been lost, destroyed or given away, one probably being the Javanese specimen listed in Doubleday (1847), of which we can find no trace. We believe, however, that we have located three of these other Horsfield specimens: two in London, which reached the BMNH by way of the Felder and then the Rothschild collections, and one in Oxford (ex Westwood Collection, OUM).

Felder and Felder (1860) described a new nominal species from peninsular Malaya as Amblypodia nakula. Five years later the Felders transferred nakula to Arhopala, stating the provenance to be “Malacca interior and Sumatra,” and comparing it to Fabricius’s description of centaurus (Felder & Felder 1865). Hewitson (1862) synonymized both Amblypodia pseudocentaurus Doubleday and A. nakula Felder and Felder (and by implication, Horsfield and Moore’s interpretation of centaurus) with Fabricius’s centaurus, a decision followed by Kirby (1871) in his influential world catalog.

The next author to study the material in the Banks and BMNH collections, Jones’s Icones, and all post-Fabrician taxonomic interpretations, was Butler (1870). The Banks Collection passed from the LS to the BMNH in 1863, and Butler would have been able to make direct comparisons with the rapidly growing BMNH collection, and consult all the literature. He stated that the type of centaurus agreed with the Felders’nakula. At first sight it seemed surprising to us that he did not mention any of the BMNH specimens listed by Doubleday (1847), or the three Horsfield specimens from Java that had arrived only 10 years before (registered as “1860–15” in the BMNH annual registers). Instead, Butler notes just one example of centaurus in the BMNH, a Sumatran specimen obtained in 1854 that formerly belonged to Sir Stamford Raffles. Again, regrettably, we have been unable to trace this specimen for comparison (Sumatran centaurus is considered to belong to the same race as material from the Malay Peninsula, the type-locality of nakula; see below). However, examination of other entries in Butler’s catalog shows in almost all cases the same pattern: citation of just a single BMNH specimen. The implication is that Butler simply considered the particular specimen cited to be “typical,” in terms of phenotype, of the Fabrician taxon in question. Butler is thus (unfortunately) silent regarding the full range of BMNH material that was then available to him for comparison.

Distant (1885) summarized the discussions to date and stated that the true Fabrician centaurus, of which he had examined the type “in the Banksian collection in the British Museum,” was an Oriental species. De Nicéville (1890) reproduced all relevant previous descriptions and stabilized the nomenclature of the Fabrician taxon as Arhopala centaurus. This was followed by virtually all taxonomists and faunistic workers on the Oriental region, including Bell (1919: 447), who published the first detailed life history of the species, from southern India. Thus Moulds (1977), in his bibliography of Australian butterflies, noted centaurus, stated that it was not Australian, and claimed that the published type locality was an error on Fabricius’s part (cf. Watkins 1923).

Thus from 1870 until Corbet’s intervention in 1941 (and even after that), all workers accepted that centaurus s.s. applied to an Oriental rather than Australian taxon. However, there was still uncertainty regarding its origin. This is particularly clear in the work of those who did not switch to the polytypic species concept, as introduced by Karl Jordan and others in the mid-1890s. Amongst butterfly specialists, Frederic Moore was almost at the opposite extreme to Jordan: Moore’s genera equated to species groups or even single collective species in the new taxonomy, whereas his species were mostly equivalent to what were increasingly regarded as subspecies.

Moore stayed with this approach throughout his magnum opus, the remarkable Lepidoptera Indica. This great series was continued after his death in 1907 by Charles Swinhoe, who evidently did not see fit to alter Moore’s taxonomic “style.” Thus, in dealing with what we would now consider to be a single polytypic species, Swinhoe (1911) recognized four separate species-level taxa: A. pirithous (“Sikkim, Assam”), A. centaurus (“Burma, Singapore, Borneo, Sumatra, Nias,” with nakula as a synonym), A. coruscans (“Andamans, South India, Ceylon”), and pseudocentaurus (“Java”). In this way the idea that the type material of centaurus could have come from the Asian mainland, as originally but incorrectly proposed by Horsfield and Moore (1858), was maintained. This view was reinforced a few years later by Watkins (1923), who stated that the type of centaurus in the Banks Collection, “though described from ‘New Holland,’ is certainly non-Australian.” Watkins then continued: “As pointed out by Butler . . . it belongs to the race nakula[sic!], Felder, and is probably from Malacca.”

As already indicated, Corbet (1941a) was responsible for introducing several new errors. First, he claimed that the male Arhopala specimen in the Banks Collection, which was long regarded as the type of centaurus, was not authentic, partly on the grounds that it (and various other specimens in the Banks Collection) did not have any original labels. Secondly, that the original description fits a female rather than a male. Corbet further suggests that neither of the two specimens of centaurus in the Banks Collection could have been present at the time when Fabricius made his original description (prior to 1775), and that the putative type “was probably obtained by Koenig in the Malay Peninsula or Peninsular Siam.” As Corbet correctly points out in the introduction to his paper, Johann Gerhard Koenig collected in those areas during 1778 and 1779 (see also below), thus implying that the supposed type of centaurus was a later addition to the Banks Collection, and that the original material that he presumed Banks “obtained in North Queensland” must have been lost or destroyed, either before or subsequently. A reprint of Watkins (1923), annotated by Watkins and held in the BMNH, may have influenced Corbet’s thinking on possible type localities.

Despite Corbet’s views, those writing taxonomic or faunistic accounts of the Oriental region and of Australia initially took no notice of these suggestions, and in this they were supported by Evans (1957), who explicitly stated that he saw no reason to upset stability. A further suggestion of Corbet (1941a) was that Fabricius’s original description of centaurus best fits an Australian taxon, Arhopala eupolis (Miskin 1890). Evans (1957: 136), who revised the whole of this large and difficult group, dismissed this idea as well, and acted to maintain the Fabrician name for an Oriental species. The increasingly influential Eliot (1972) at first followed Evans but, in the revised third edition of The Butterflies of Malaya Peninsula (Eliot, in Corbet & Pendlebury 1978: 468), he unfortunately rejected Evans’s plea for stability and upheld Corbet’s interpretation. This had been foreshadowed in print a year earlier by D’Abrera (1977), who noted that “Eliot has drawn my attention to this fact [that the type locality for centaurus was originally given as ‘Nova Hollandia’] and to a rather extensive correspondence between himself and Col. C.F. Cowan on the subject, confirming Corbet’s arguments.” Seemingly Eliot did not do any research on this issue himself but, in his revision of Corbet and Pendlebury (1978), added the information supplied by Cowan (in litt.) on Koenig’s collecting activity in Malaya and Thailand.

Eliot (in Corbet & Pendlebury 1978: 468), closely echoing the words of Corbet (1941a: 100), confidently stated “there is every reason for supposing that it [the centaurus specimen] was not in coll. Banks when Fabricius described centaurus.” However, we have been unable to discover what this “every reason” might be! Neither Corbet nor Eliot gave explicit grounds for rejecting the Banks “type” specimen as part of the type series. As we demonstrate below by a re-examination of various specimens and an analysis of Cook’s itinerary, we believe the conclusions of Corbet, as compounded by Eliot, are erroneous. The changes they proposed have now been followed uncritically by many taxonomists in the Indo-Australian region, but not by all, and some authors have inadvertently used both centaurus and pseudocentaurus as if they applied to two different species, when in reality they were dealing with one and the same insect (e.g. Robinson et al. 2001).

Taxonomic conclusions

To arrive at defensible conclusions concerning application of the name Papilio centaurus Fabricius, two issues must be resolved. First, what is the identity and status of the “centaurus type specimen” that has been in the Banks Collection for so long? Apparently it was there before 1791, the last time Fabricius studied in England, and at least one specimen seems to have been depicted by Jones as a Fabrician species during 1783–1785 and thus, critically, before Koenig’s material reached London. Secondly, we must establish the true type locality.

While traveling between Queensland in Australia and the Cape of Good Hope, HMS Endeavour touched land only in New Guinea (very briefly), the small island of Savu (or Sawu) south-west of Timor, and for about 2 months at Batavia (near modern day Jakarta), in Indonesia. Although there is no direct evidence in the Banks, Solander and Cook journals to show what kinds of butterflies (for that matter, insects in general) were collected where, there is evidence that they did collect in Queensland, near Jakarta (Java), and finally at the Cape of Good Hope in South Africa. There is no evidence that they collected any material on Savu (H. Gaonkar, unpubl. data, 2003).

Fabricius generally knew with confidence which insects Banks collected in Madeira, Rio de Janeiro, Patagonia, Tahiti and New Zealand, and most of those obtained from Australia. He did describe a “Zygaena” from Madeira, a type of moth that could not possibly have come from there (O. Karsholt, pers. comm., 2004), and there may be other individual mistakes. However, Fabricius was frequently confused about the provenance of specimens that were collected when Endeavour made landfall in Java and the Cape. Some of these cases were correctly interpreted (and some also misinterpreted) in the late 19th century. The reason for these errors regarding butterflies from Java, the Cape and, in a few cases, Australia, was simply that neither Banks nor Solander kept (or could keep) any records of their natural history activities in Java or the Cape. Their serious illnesses, and how numerous crew members of Cook’s HMS Endeavour died in Java and on the way to the Cape, are now so well documented that we need not go into detail here (Banks in Beaglehole 1962; O’Brian 1987).

For example, Papilio astenousFabricius, 1775, a taxon that Walter Rothschild interpreted as belonging to Troides helena (Linnaeus, 1758), was undoubtedly collected in Java, and could only have been collected there. But Fabricius gave the habitat as “in Capite Bonae Spei Mus. Banks . . .”, that is, the Cape of Good Hope (Rothschild 1895). In this way, Fabricius gave the habitat for many insects collected in Java either as “Nova Hollandia” (Banks’s collecting stop before Java) or “Capite Bonae Spei” (the Cape of Good Hope, which was Banks’s stop after Java).

The original material of Papilio centaurus must have reached the Banks Collection before 1772 (Cook’s Endeavour, with Banks and his collections on board, returned to England in 1771; Beaglehole 1962). The Banks Collection (BMNH) now contains two specimens over the name centaurus. We consider that the assumption of Corbet (1941a), that both of these specimens came from Malacca, must be wrong. Koenig certainly collected a number of butterflies in Malacca during 1778 and 1779. Most of these specimens, now in ZMUC and BMNH, were described by Fabricius in 1781 and 1787, and a few (smaller) species in 1793. Any Koenig material from Malacca would have come to Banks in approximately 1786, because all the specimens that Koenig bequeathed to him came to London after Koenig’s death in India in 1785 (Dryander 1800: vol. 5, 309). Fabricius would then have most certainly described any new species in these accessions, either in his Mantissa Insectorum in 1787 or in his Entomologia Systematica in 1793 (e.g. see the account of democritus below). But centaurus was described in 1775, long before any of Koenig’s material had reached Banks. It seems that because Corbet appreciated this fact, and accepted the “Nova Hollandia” source for centaurus without question, he convinced himself that the material in the Banks Collection could not have been collected during the Endeavour voyage, and most likely had been substituted by material from the later Koenig accession. That one or both of the specimens might have been collected by Banks or other members of the Endeavour crew in Java does not seem to have occurred to him. As related further below, it does indeed seem plausible that the second centaurus specimen in the Banks Collection is just such a later addition, and could well have been collected in the Malay/Thai peninsula by Koenig. But we are confident that this is not the case for the other Banks specimen that, for so long, was regarded as the type. We now believe that this example was collected during the Cook voyage, and it came from Java (see further below, under “type material”).

Consequently, the conclusions of the majority of classical authors are here reinstated. In particular, Doubleday’s pseudocentaurus from Java must be regarded as a junior subjective synonym of Arhopala centaurus (Fabricius, 1775), sensu stricto. The name to be applied to the race of centaurus that flies in the Malay Peninsula is not subspecies centaurus, but centaurus nakula (Felder & Felder, 1860). The name of the separate Australian species that has been confused with centaurus (and A. araxes) is Arhopala eupolis. The results of the research outlined above are in harmony with the Code (see ICZN 1999: Articles 23.9 and 23.9.6, but see also 3.2). The overwhelmingly continuous usage of this name for the Oriental taxon for more than 200 years (and its continued use in many local checklists) is enough reason to maintain stability in this case. Finally, the type locality of the nominotypical subspecies of Papilio centaurus should be regarded as Java (near Jakarta), not “Malaya” as proposed by Evans, “Nova Hollandia” as given by Fabricius, or “North Queensland” as suggested by Corbet.

Synonymy of Javanese butterfly to be known as Arhopala centaurus centaurus (currently known as A. pseudocentaurus pseudocentaurus)

Papilio (Plebeji Rurales) CentaurusFabricius, 1775: 520, no. 329. Lectotype male, “Nova Hollandia” (recte Indonesia, Java, near Jakarta), in Banks Collection, BMNH, here designated (examined). BMNHE♯ 668233. (Fig. 1.)

Papilio centaurus Fabricius, 1775: Fabricius (1781): 117, no. 523; Jones (MS 1783–1785, volume 6: pl. 22, fig. 1, OUM); Fabricius (1787): 68, no. 646; Butler (1870): 179; Evans (1957): 135; Zimsen (1964): no. 118.

Hesperia (Rurales) Centaurus (Fabricius, 1775): Fabricius (1793): 275, no. 63.

Polyommatus centaurus (Fabricius, 1775): Godart (1824): 658.

Amblypodia centaurus (Fabricius, 1775): Horsfield (1829): 102; Hewitson (1862): 3,4 (may be a misidentification according to Distant 1885); Butler (1870): 179; Kirby (1871): 419; Norman (1950): 813.

Amblypodia (Arhopala) centaurus (Fabricius, 1775): Watkins (1923): 205.

Arhopala centaurus (Fabricius, 1775): Felder and Felder (1865): 222; de Nicéville (1890): 234; de Nicéville and Martin (1896): 464; Swinhoe (1911): 147; Bell (1919): 448–450 (biology); Evans (1957): 113; Eliot (1963): 206; Fleming (1975): 39; Eliot (in Corbet & Pendlebury 1978): 468; Robinson et al. (2001): 63; Vane-Wright and de Jong (2003): 124.

Amblypodia pseudocentaurus Doubleday, 1847: 24. Lectotype female, Indonesia, Java, BMNH, here designated (examined). BMNHE♯ 147889. Formally synonymized with Arhopala centaurus (Fabricius) by Bethune-Baker (1903): 40. Syn. rev. (Fig. 5)

Amblypodia pseudo-centaurus Doubleday, 1847: Westwood (in Westwood 1852): 478.

Amblypodia pseudocentaurus Doubleday, 1847: Hewitson (1862): 4, as synonym of centaurus; Corbet (1941a): 100.

Narathura centaurus (Fabricius): Distant (1885): 261.

Nilasera centaurus (Fabricius): Moore (1881): 115.

Amblypodia amazona Pagenstecher, 1890: 107 (Java), nomen nudum: Bethune-Baker (1903): 40.

Arhopala centaurus pseudocentaurus (Doubleday, 1847): Fruhstorfer (1914): 158; Corbet (1941b): 163.

Arhopala pseudocentaurus (Doubleday, 1847): Corbet and Pendlebury (1978): 285; (1992): 275; Fleming (1983): 85; D’Abrera (1986): 576; Seki et al. (1991): 53; D’Abrera (1998): 162; Bascombe et al. (1999): 209; Osada et al. (1999): 170; Robinson et al. (2001): 63; Megens (2002): 35; Monastyrskii and Devyatkin (2003): 37; Huang and Xue (2004): 195; Megens et al. (2004a,b).

Type material of Papilio centaurus Fabricius

Papilio centaurus was described from an unstated number of specimens in “Mus. Banks,” purportedly from “Nova Hollandia” (Australia). As noted above, there are two male centaurus in the Banks Collection. On the testimony of Horsfield (1829), at least one should have had “a ticket in his own (Fabricius’s) hand-writing” but, as Corbet (1941a) pointed out, neither of these specimens has a locality label, nor does either of them have an original Fabrician “ticket.” This helped convince Corbet that neither of them could represent original material.

One of the specimens (BMNHE♯ 668232) has no label at all. It is exceedingly low-set, on a coarse pin that is different from the pin of the second specimen. The underside of this unlabeled, low-set specimen is dull, like that of A. centaurus nakula from the Malay Peninsula and elsewhere on the South-East Asian mainland (see the underside of a specimen from Laos, beautifully illustrated by Osada et al. 1999: 170). In our view it is very plausible that this is a Koenig specimen from Peninsular Thailand, collected during 1778–1779, and added by Banks to his collection sometime around 1786. However, there is no independent evidence that would back such an assertion. The real importance of this specimen is that it is typical of nakula, and surely helped convince Corbet (who in his own mind was only trying to decide between Malay Peninsula versus Queensland as the likely origin) that the Banks specimens must have been from the former, and therefore could not be original material for Fabricius’s 1775 description.

The second specimen (BMNHE♯ 668233), higher set and on a different type of pin, is clearly the one referred to by many authors as “the Fabrician type.” It has an old label (“Papilio P.R. Centaurus, Fab. Entom. p. 520, n. 329”), in addition to a more recent circular label relating to its accession to the BMNH, a circular “type” label, and a drawer label added by Watkins (Fig. 1). However, the old label offers no proof of direct connection with Fabricius, and is probably not the “ticket” that Horsfield referred to. Lengthy but superficial (in the sense that we have not undertaken any biometric or colorimetric analyses) comparison with a long series of both Malayan and Javanese centaurus leaves us a little uncertain regarding its origin, even if we restrict ourselves to a choice between just Malaya and Java. This insect is more variegated beneath than the other Banks Collection specimen, being on the “bright” end of the range for the Malay Peninsula population, and at the dull end of the range for Java. As pointed out by Evans (1957), the mean difference between the subspecies is clear, but there is a lot of individual variation. The illustration in Jones’s unpublished Icones (MS, volume 6, plate 22, OUM; reproduced here as Fig. 4) is, to our eye, well fitted to the Javanese phenotype.

However, we also consider that Jones’s illustration fits Banks BMNHE♯ 668233 very well, including wing shape, setting angle, shape and disposition of underside spots, and even color (Figs 1,4). A further insight into the origin of this specimen is afforded by Bethune-Baker (1903: 40), who evidently considered the Malay Peninsula form to be “typical”centaurus, and usually a duller blue than the Javanese population. We quote (our emphasis added):

We imagine, had Bethune-Baker been apprised of what we consider to be the true origin of the Banks material, his observation would not have seemed curious to him at all.

The loss of the original Fabrician label from the Banks Collection “type” specimen, after so many years and the vicissitudes that befall collections, is not convincing evidence of replacement. Taking all the information together, we are thus quite satisfied that this specimen (Fig. 1) was originally collected by Banks in Java during the very difficult phase of the Endeavour voyage, and is not a later Koenig specimen originating from the Malay Peninsula, or anywhere else. Accordingly, we designate the Banks Collection “type” (BMNHE♯ 668233; Fig. 1) as the lectotype of Papilio centaurusFabricius, 1775. We exclude the second Banks Collection specimen (BMNHE♯ 668232) from the type series, as well as material labeled “centaurus” that we found in the HM and ZMUC collections:

As already indicated, based on Fabricius’s known links with William Hunter and Tønder Lund, we searched for possible Banks Arhopala material in the HM (Glasgow) and the Sehested and Tønder Lund and Fabrician collections in the ZMUC (Copenhagen). In the HM we found a single Arhopala male, labeled “Pap. Centaurus Fabr p. 117 No. 523” (Fig. 2). It is quite similar to A. centaurus, and initially we thought it might be a syntype, not least because it is set in such a similar fashion to the lectotype (cf. Figs 1,2). However, on the evidence of the wide black margin to the upperside wings, and details of the underside pattern, we believe that this is an example of the related species A. amantes (Hewitson, 1862), a butterfly found from Sri Lanka and southern India to Myanmar and Laos. It corresponds well to the image given in D’Abrera (1998: 161). Cook’s Endeavour did not make landfall anywhere in this region, and thus this specimen could not have been collected by Banks (prior to 1775). Most likely it is a contemporary specimen from India, an area from which, by the mid-late 18th century, English collectors were receiving much Lepidoptera material (Vane-Wright & Hughes 2005: 254).

In the Sehested and Tønder Lund Collection (ZMUC), there are male and female specimens of Arhopala, both of which carry the name centaurus on old labels. The female, which we identify as A. anthelus (Westwood, 1851), is labeled “H. Centaurus ex Ind. Or.” More challenging is the male. It bears a label that appears to have been written at two different times, in two different hands (Fig. 3). At the bottom, in faded brown ink, is the single word “Centaurus.” At the top, in black ink, it appears to say “H. n. sp. ex Ind. orient, e Sumatra”. As pointed out to us by Ole Karsholt (pers. comm., 2004), the work of Zimsen (1964: 12) suggests that most specimens in the Sehested and Tønder Lund Collection had their original labels replaced by new ones, written by some unknown person working in the first half of the 19th century. In this context, inclusion of “Sumatra” may be significant. So far as we are aware, the first people to note in the literature that A. centaurus occurs on Sumatra were Felder and Felder (1865) and Butler (1870), the latter documenting a specimen then in BMNH dating from 1854. Possibly this replacement Sehested and Tønder Lund label was made rather later than Zimsen suggests, perhaps even after the appearance of the Butler catalog. However, as Ole Karsholt (pers. comm., 2006) has further pointed out to us, it is difficult to imagine anyone working in Copenhagen in the 1800s having such detailed knowledge of tropical butterflies.

With respect to the specimen itself (Fig. 3), initially we thought that this might also be a genuine A. centaurus, and so conceivably ex Banks. However, it has wider black borders to the wing uppersides than true centaurus, and the (damaged) hindwing underside cannot be matched to any centaurus material that we have seen. Notably, it has a well-developed tornal lobe. Frustratingly, the critical forewing underside pattern is not visible. It has been suggested to us that it could be an example of the Australian species A. madytusFruhstorfer, 1914, but having compared it with A. madytus material in the BMNH, we do not think it fits this species either (cf. illustrations in D’Abrera 1977: 312; Braby 2000: 689). It seems to us more comparable to the Oriental species A. bazalus (Hewitson, 1862). While we remain uncertain as to its identity, it clearly does not correspond to the lectotype of centaurus (Fig. 1), or to Jones’s painting (Fig. 4). The pin on which it is fixed is old, short, and of a different gauge from that of the lectotype. Ole Karsholt (pers. comm., 2006) has pointed out that Fabricius was not in the habit of giving away specimens, unless duplicates, from his own collection, so it is unlikely that he would have given such a specimen to Tønder Lund when no such specimen can be found in his own collection in ZMUC. We conclude that this specimen cannot have been part of the original type series, and there is no evidence to suggest that it was collected by Banks.

Type material of Amblypodia pseudocentaurus Doubleday

Amblypodia pseudocentaurus Doubleday, 1847, was described from eight syntypes in the collection of the British Museum: one from “Java. Presented by the Hon. East India Company,” one from “Moulmein,” four from “Silhet,” and two from “Ceylon,” together with eight or nine additional specimens originally in the EIC that should be included as part of the series by virtue of the clear reference to Horsfield’s (1829) account of centaurus. In the BMNH collections we have found one male and two female specimens of A. centaurus received from the EIC in 1860, registered as BMNH 1860–15 (according to Cowan 1975; the EIC collections were “dispersed to Kew and South Kensington during 1879,” but this is clearly not the case, at least for much of the insect material). These must be the three specimens listed by Horsfield and Moore (1858) as remaining in the EIC collection at that time. In addition, we have also located two specimens apparently presented by Horsfield to the Felders that probably originated from the EIC, and one apparently presented to Westwood. The other three of nine EIC specimens noted by Horsfield (1829) remain unaccounted for, including the one that evidently passed to the BMNH sometime between 1829 and the mid-1840s (Doubleday 1847). Of the eight specimens listed by Doubleday as being in the BMNH collection in 1847, we have located only the two from Sri Lanka. From this total of eight putative syntypes that we have located, we designate the ex-EIC female long ago placed as the “type” in the BMNH as the lectotype of Amblypodia pseudocentaurusDoubleday, 1847. This specimen (BMNHE♯ 147889) has been labeled accordingly (Fig. 5), and the other seven specimens have been labeled as paralectotypes (six in BMNH: BMNHE♯ 147890–147895; one male in OUM labeled “W Horsfield Java/Westw/Arhopala centaurus f. pseudo-centaurus Doubld. Named by H.H. Druce 1900–1904.”). Horsfield’s material originated largely or entirely from East Java, if it is necessary to delimit the type locality more precisely.

Identity of Arhopala centaurus

Having located the authentic specimen on which Fabricius’s Papilio centaurus is based, and established its origin, it remains to confirm the identity of centaurus within the current framework of Arhopala systematics. In a key with 244 entries, Eliot (1963) divided the species found in Malaysia into 27 species groups, of which the centaurus-group was one, noting that the group included “about 15 species – mostly from Papua.”Parsons (1998) now recognizes 18 species in the group. We confirm that the lectotype of A. centaurus (Fabricius, 1775), as designated here, runs to Eliot’s couplet 155 (Eliot 1963: 206) where, based on the false type locality of Malaysia due to Evans (1957), Eliot recognized the Malaysian race of this butterfly as centaurus centaurus. In the later key to Malaysian Arhopala revised by Eliot for Corbet and Pendlebury (1978, 1992), the lectotype of centaurus runs out very early, at entry 4, to A. pseudocentaurus. Following Corbet and Pendlebury (1978, 1992), amongst the Malayan fauna, A. pseudocentaurus is distinguishable from all other Malaysian species “by the silvery green lines edging the markings in the cell on the forewing beneath.” Again, we confirm that this characteristic is evident in the lectotype of centaurus. Corbet and Pendlebury note two other species in the Indo-Malayan region with “similar venation and silvery markings in the forewing cell beneath,”A. araxes Felder and Felder from Sumatra, Sulawesi, etc., and A. amantes from southern Myanmar.

Distribution of Arhopala centaurus

Because of the confusion over application of the name centaurus, as well stated by Distant (1885) more than 100 years ago, accounts of the distribution of this species are also confused in the literature. Following Seki et al. (1991), we consider A. centaurus to be a polytypic Oriental species, the collective taxon extending from Sri Lanka and Nepal in the west to southern China, and the western Lesser Sunda Islands and Philippines in the east. Despite the claim by Bascombe et al. (1999), according to Vane-Wright and de Jong (2003) there is no reliable record of A. centaurus from Sulawesi. Thus although Detani (1983) recorded A. pseudocentaurus (here regarded as a strict synonym of A. centaurus) from the Banggai Islands, immediately to the east of central Sulawesi, the identity of the species recorded by Detani must be regarded as very uncertain (Vane-Wright & de Jong 2003). Based on Evans (1957), Eliot (in Corbet & Pendlebury 1978) and Schroeder and Treadaway (1999), 12 subspecies of A. centaurus are currently recognized:

Synonymy of Malayan taxon to be known as Arhopala centaurus nakula (currently known as A. pseudocentaurus nakula)

Type material of Amblypodia nakula Felder and Felder

Felder and Felder (1860) originally described Amblypodia nakula in a paper listing new butterflies from the Malay Peninsula. No specific distributional data were given for nakula, or any of the other species described at the time. Five years later the Felders moved nakula to the genus Arhopala, and gave two localities for it based on material in their own collection: Malacca Interior (Com. de Castelnau) and Sumatra (Wallace). Ex Felder material from both sources is now present in the BMNH (received as part of the Rothschild Bequest). In our view only Malayan ex Felder specimens can qualify as potential syntypic material, as Sumatra was not mentioned in the original description, whereas the Malay Peninsula was. Accordingly, we have selected a specimen ex Felder Collection (Fig. 6) evidently collected by de Castelnau as the lectotype of Amblypodia nakulaFelder and Felder, 1860. This specimen (BMNHE♯ 147896) bears a label in W.H. Evans’s handwriting dated 1955 (Fig. 6), stating that the specimen was “Incorrectly placed over type of vihara,” and “Is identical with type of nakula” (i.e. the Sumatra specimen now excluded as a valid type).

Identity of Arhopala centaurus nakula

If it is accepted, as established above, that Fabricius’s centaurus is based on material from Java, then the correct name for the South-East Asian race that includes Malay Peninsula populations is A. centaurus nakula.

Distribution of Arhopala centaurus nakula

Arhopala centaurus nakula occurs from Indo-China throughout most of Neomalaya (Moulton 1915), from Myanmar, Indo China, Hainan, Thailand and the Malay Peninsula to Singkep, Sumatra, Nias, Belitung, Bangka, the Natuna Islands, Borneo, and Dumaran and the Calamian group in western Philippines (Evans 1957; Seki et al. 1991; Schroeder & Treadaway 1999). According to Chou (1994), this taxon also occurs in China but, as noted above, this requires confirmation. Bascombe et al. (1999) considered the Hong Kong population to belong to subspecies pirithous.

Synonymy of Australian taxon to be known as Arhopala eupolis Miskin, 1890, sp. rev. (currently known as A. centaurus)

Type material of Amblypodia eupolis

Originally described from a syntypic series of both sexes, collected at Cooktown & Cardwell, Queensland. See Hancock (1995) for details.

Discussion of Arhopala eupolis

Following Corbet (1941a), D’Abrera (1977) and Eliot (in Corbet & Pendlebury 1978), the name centaurus Fabricius has in recent years been applied erroneously by various authors to an Australian member of the centaurus group. The species concerned should now be known as A. eupolis, returned here to full and valid species status. Corbet (1941a: 100), because of his mistaken belief that centaurus had an Australian provenance, considered the Australian eupolis (type localities Cooktown and Cardwell) to be a strict synonym of centaurus (the most likely type locality, if based on Australian material, would have been Cooktown: Watkins 1923). Evans (1957), followed by D’Abrera (1971) and Quick (1974), regarded eupolis as a subspecies of a somewhat more restricted polytypic species, A. araxesFelder and Felder (1865). The latter (type locality Sulawesi) is now regarded as a separate species that occurs from Sumatra and Java to the Lesser Sunda Islands and the Sulawesi Region (see Vane-Wright & de Jong 2003).

Distribution of Arhopala eupolis

Taking the works of Evans (1957), Common and Waterhouse (1981), D’Abrera (1990), Parsons (1998), Braby (2000), Edwards et al. (2001), Tennent (2002) and Vane-Wright and de Jong (2003) together, this species apparently occurs as just two weakly separated races. Since Parsons (1998) proposed that the New Guinea region populations formerly regarded as subspecies philtronFruhstorfer, 1914, cannot be separated from Queensland butterflies, nominate A. eupolis eupolis is seen to extend along the Queensland coastal belt, through the islands of the Torres Strait and the whole of New Guinea, westwards to Aru and Kep. Kai, and eastwards along the Louisiade Archipelago to Tagula. Arhopala eupolis asopusWaterhouse and Lyell, 1914, is found in north-western Australia and Northern Territory, including Groote Eylandt. Both Parsons (1998) and Tennent (2002) regard the Bismarck–Solomons taxa previously included within this collective taxon to comprise a separate species, A. eurisus Druce, 1891.

Synonymy of Arhopala democritus

Type material of Hesperia democritus

Hesperia democritus was described by Fabricius from an unspecified number of specimens (but probably just one) in “Mus. Dom. Lund . . .”, collected by Koenig (Corbet 1941a). A single male in the Sehested and Tønder Lund Collection, ZMUC, illustrated together with its labels in Fig. 7, is hereby designated the lectotype of Hesperia democritusFabricius, 1793. This selection reflects Corbet’s (1934) earlier valid restriction, in which he stated that “there appears to be no valid reason for regarding this specimen as other than the type of democritus.”

Taking into account the known distribution of butterflies with a typical democritus phenotype (see below), the original material could only have been collected either in western Malaysia (Kedah), or on the island of Pulau Salang or “Junk Ceylon,” now known as Phuket, the familiar holiday destination in Peninsular Thailand. Corbet (1941a) concluded that it must have been the latter, and since there is no evidence confirming or contradicting this, we accept his inference. Koenig spent several weeks on Phuket in 1778 and 1779.

Current status and distribution of Arhopala democritus

Arhopala democritus is not found in southern India, but is distributed from Orissa (H. Gaonkar, unpubl. data, 2004) to north-eastern India, Myanmar, Thailand, Malaysia and Neomalaya (Indonesia west of Sulawesi and Java, including northern Borneo). The following three subspecies are rather dully marked beneath: A. d. olinda (H. Druce, 1873), endemic to Borneo; A. d. buxtoni (Hewitson, 1878) restricted to Sumatra; and A. d. lycaenaria (Felder & Felder, 1860), the population that occurs south of Kedah to Singapore. These three could collectively constitute a separate species (to which the name Arhopala lycaenaria would apply) from democritus sensu stricto, which flies from Kedawi north to Thailand and Myanmar, and is brightly marked with silvery-white beneath (as suggested by the synonym albopunctata). Eliot (in Corbet & Pendlebury 1992) notes a specimen of the bright phenotype collected as far south as Fraser’s Hill, suggesting that democritus democritus and democritus lycaenaria overlap without intergrading (see also Fleming 1975, who implies overlap in Kedah). Material from Orissa and NE India has yet to be determined to subspecies.