TRANSACTIONS OF SECTION D. 435 



bengalensis, are known from Celebes. These are brackish-water fishes, and the 

 distribution of the species of Symbranchus leaves little doubt that they some- 

 times descend to the sea ; indeed S. bengalensis has been recorded from Western 

 Australia. Monopterus javanensis parallels Anabas scandens in its wide distri- 

 bution, its importance as a food fish, and its vitality. 



Two endemic Cyprinodonts are found in Celebes, one belonging to the 

 Asiatic genus Haplochilus, the other to a peculiar but related genus. There 

 can be little doubt that these originally reached Celebes by sea. Many fishes 

 of this family are marine, and although the Indian species are usually reckoned 

 as fresh-water, Day tells us that they are nearly always found in estuaries 

 or not far from the sea. 



In my opinion, the Indian element in the fresh-water fish fauna of Celebes 

 consists of (1) introduced species and (2) species which have journeyed by sea; 

 the same is true of the islands from Lombok to Timor, which have only a 

 Haplochilus and the two Labyrinthic fishes, Anabas scandens and Ophiocephalus 

 striatus, already disposed of. Bali is a small island, and its fresh-water fish fauna 

 is not too well known, but is doubtless poor in proportion to its size ; neverthe- 

 less, we know that it is inhabited by a siluroid of the genus Clarias, two species of 

 Cyprinids, and an Anabantid in addition to Anabas scandens. Java has a fresh- 

 water fish fauna extremely similar to that of Sumatra and, considering its 

 smaller size, is probably as rich in species; it is rather straining a point to call 

 the absence of Luciocephalvs pulcher, the only species of its family, an im- 

 poverishment of the Indian fauna, and still more so to apply this term to the 

 absence from Java of the Sumatran and Bornean Scleropages formosus, since 

 the only other species of Scleropages inhabits Queensland. 



For fresh-water fishes Wallace's Line is neither ' exploded ' nor ' unfor- 

 tunate,' but of fundamental importance. The rich and varied fish fauna of 

 Java, Sumatra, and Borneo, with most of the genera and many of the species 

 identical with those of the mainland of Asia, indicates clearly that these islands 

 formed part of the continent quite recently. The absence from Celebes of true 

 fresh-water fishes shows that this island has not been connected with Asia 

 during the Tertiary Period. The Ostariophysi are an ancient group, and prob- 

 ably most of the families were differentiated in Cretaceous times ; Siluroids are 

 known from Lower Eocene deposits in various parts of the world; the North 

 American Catostomiche are also known from the Lower Eocene; the Cyprinids. 

 and Loaches of the Oligocene and Miocene of Europe differ only specifically from 

 their modern representatives, and this is true of the Cyprinoids and Siluroids 

 of the shales and lignites of Sumatra, usually regarded as of Eocene age, but pos- 

 sibly Miocene. Professor Max Weber thinks that Celebes did not separate from 

 Asia until it had received its mammals. It seems to me significant that the 

 mammalian fauna of Celebes consists so largely of arboreal types, specially 

 liable to accidental transmission, or of strong swimmers. 



Celebes is an anomalous island, as Wallace well called it ; belonging to the 

 Australian Region, yet from its remoteness and its long isolation deficient in 

 Australian types. During its long proximity to the Oriental Region it has ac- 

 quired in one way or another a certain number of Indian forms which obscure 

 its real affinity. The Timor Group seems to have had a somewhat similar history ; 

 its zoological distinctness from the Indo-Malay Islands is confirmed by the 

 fresh-water fishes. It may therefore well be the case that Wallace's Line marks 

 the severance of Australia from Asia at or before the commencement of the 

 Tertiary Period. 



F f 2 



