JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA VOLUME 58 PART 1 MAY, 1975 PRICE TWO DOLLARS REGISTERED FOR POSTING AS A PERIODICAL-CATEGORY B THE President Vice Presidents Past President Joint Hon. Secretaries Hon. Treasurer Hon. Librarian Hon. Editor ROYAL SOCIETY OF WESTERN AUSTRALIA PATRON Her Majesty the Queen COUNCIL 1974-1975 .... G. A. Bottomley, B.Sc., Ph.D. .... B. E. Balme, D.Sc. P. R. Wycherley, O.B.E., B.Sc., Ph.D., F.L.S. .... A. F. Trendall, B.Sc., Ph.D., A.R.C.S., F.G.S. .... M. Perry, B.Sc. (Agric.) (Hons.) G. Perry, B.Sc. (Hons.) .... S. J. Curry, M.A. .... A. Neumann, B.A. .... A. J. McComb, M.Sc., Ph.D. R. M. Berndt, M.A., Dip.Anth., Ph.D., F.R.A.I., F.F.A.A.A. C. E. Dortch, B.S., M.Phil. L. J. Peet, B.Sc., F.G.S. P. E. Playford, B.Sc., Ph.D. P. G. Quilty, B.Sc. (Hons.), Ph.D. J. C. Taylor, B.Sc,, Ph.D., A.R.C.S. P. G. Wilson, M.Sc. 1. — Preliminary investigation of the palynology of the Upper Eocene Werillup Formation, Western Australia by D. Hos^ Manuscript received 19 March 1974; accepted 19 November 1974 ABSTRACT A diverse Late Eocene plant microfossil assemblage was recovered from the Werillup Formation (Plantagenet Group) in Werillup No. 17 bore near Albany. The assemblage is composed predominantly of angiosperm pollen, with minor cryptogam and gymnosperm ele- ments. Common species are Nothofagidites spp. (mainly brassi group), Proteacidites spp., Haloragacidites harrisii, MalvacipoUis diversus, Myrtaceidites spp., Tricoloporites prolata, Cya- thidites minor, and podocarpaceous species. Most species are long ranging, however, Proteacidites concretus, Cupanieidites reticu- laris and Triporopollenites gemmatus restrict the assemblage to the Eocene and four other species are restricted lo the Early Tertiary. Dinoflagellates in the borehole confirm the Late Eocene age determined by invertebrate fossils elsewhere in the Formation. The following species are recorded for the first time in Western Australia; Proteacidites concretus Harris, P. granulatus Cookson, P. reticulatus Cookson, P. subscabratus Couper, Simplicepollis scabratus McIntyre, Tricolporites microreticulatus Harris, and Triporopollenites gemmatus Harris. The assemblage from the borehole, especially Beaupreaidites spp., Cupanieidites spp., and Myrtaceidites mesonesus, suggests a warm and humid climate which is consistent with other palaeoclimatic and palaeomagnetic evidence. Introduction A transgressive sequence of fine-grained marine and paralic sediments was deposited on the south coast of Western Australia during the Late Eocene. The sediments compose the Plantagenet Group and contain a diverse fauna and flora that has been confidently dated. The Plantagenet Group was intersected by a bore Werillup No. 17, drilled by the Geologi- cal Survey of Western Australia near Albany. Western Australia, and the samples collected from the sequence were examined palynologic- ally. The aim of this study was to describe and identify the species of spores and pollen in the samples and to determine the distribu- tions and frequencies of each species. The samples are from the lower part of the basal formation of the Plantagenet Group. They consist of silt, sand and mud* and most have an abundant microflora. The micro- floral assemblage from the bore is here related to other microfloras from the Plantagenet Group and Lower Tertiary sediments elsewhere in Australia. The assemblage is also discussed in relation to the Late Eocene climate of the I Geology Department. University of Western Australia, Nedlands, Western Australia 6009. south coast of Western Australia, the palaeo- geography of the region, and the environment of deposition of the sediments. Stratigraphy of the Plantagenet Group The stratigraphy of the Plantagenet Group was revised by Cockbain (1968) and a review of the literature is given in his paper. The Group consists of two formations: the Pal- linup Siltstone and the underlying Werillup Formation, from which the samples for this study were collected. The distribution of the Group is shown in Fig. 1. The Werillup Formation is a sequence of dark-coloured paralic siltstone, sandstone, car- bonaceous claystone and lignite that has a sporadic distribution in Precambrian basement lows. The transgressive sequence includes basal conglomerates and lignites in places, deltaic and lagoonal sediments and deeper- water siltstones and claystones. Similar lig- nites are also found at Nornalup, Denmark and Fitzgerald River. The sequence passes up into the Pallinup Siltstone or is unconform- ably overlain by Quaternary sands. The bryozoal Nanarup Limestone Member of the Werillup Formation is of restricted lateral extent and highly fossiliferous. It only occurs at Nanarup (about 18 km ENE of Albany) and has been dated by Quilty (1969) as uppermost Eocene. The Pallinup Siltstone consists of a light- coloured siltstone and spongolite and either conformably overlies the Werillup Formation or onlaps the Precambrian basement. It is up to 60 m thick and outcrops from Walpole to 160 km east of Esperance where it passes laterally into the Toolinna Limestone of the Eucla Basin. Where terrigenous material was negligible and conditions favoured sponges, the sediment became extremely rich in sponge spicules. Lithistid sponges in the sediments indicated the depth of deposition may have been from 20 m to 200 m (de Laubenfels 1953). The same depths when applied to spongolites of similar age and type now occurring 273 m above sea level at Norseman, indicate that the Pallinup Siltstone may have been deposited in depths up to 474 m (Clarke et al. 1948). Palaeontology of the Plantagenet Group The Werillup Formation has abundant in- vertebrate remains including gastropods, ce- phalopods, bivalves, echinoids, sponges, fora- minifers and bryozoans. The foraminifers and echinoids have been used to date the sediments Journal of the Royal Society of Western Australia, Vol. 58, Part 1, May, 1975. 1 Figure 1. — Map showing location of Werillup No. 17 borehole and the distribution of the Plantagenet Group. as Late Eocene (Quilty 1969); and Asterocyclina from the formation has been used as a warm- water indicator (Cockbain 1967). Cockbain (1969) recorded remains of dasycladacean algae in the Werillup Formation near Esperance which also indicated warm tropical waters. The Kojonup Sandstone (Churchill in McWhae et al. 1958) was correlated with the Plantagenet Group and contains leaves and wood of Nothofagus, Banksia and Araucaria, a frond of Gleichenia, leaves of Moraceae. Pro- teaceae, a palm, and an unidentified monoco- tyledon. The Pallinup Siltstone has a poorly pre- served fauna of molluscs, sponges and bryo- zoans. Nautiloids and foraminifers are rare but both suggest a Late Eocene age (Cockbain 1968). Chapman & Crespin (1934) described leaf and wood impressions of Agathis. Notho- fagus and Bombax from sediments considered to be part of the Pallinup Silt tone at Cape Riche (Cockbain 1968). Silicified coniferalean and proteaceous wood that is probably weather- ed from the Plantagenet Group is common along the south coast (Balme in de Jersey 1968). Palynological work from the Lower Tertiary of Western Australia was reviewed by Balme (in de Jersey 1968) and no further contribu- tions have been published since then. The earliest work on material from the Plan- tagenet Group is by Cookson (1953), who re- corded Phyllocladidites ( Dacrydiumites) maw- sonii from the Group. Cookson & Pike (1953a, b, 1954a, b) recorded and described several new species from the same material. Cookson (1954b) listed all the species that had been recorded from the samples and she was able to relate the assemblage to ‘Microflora C’ from No. 1 Bore, Birregurra in Victoria (Cookson 1954a). The assemblage was also similar to Late Eocene microfloras from New Zealand (Couper 1953). Further records of species of Nothofagi- dites in the Plantagenet Group were given by Cookson (1954b). The assemblage reinforced plant macrofossil evidence of a pan-Australian Early Tertiary flora that had a tropical aspect (Burbidge 1960). A Late Eocene microflora from sediments in bores and deep leads at Coolgardie was de- scribed by Balme & Churchill (1959), and on the baus of this microflora the sediments were correlated with the Plantagenet Group. The Coolgardie sediments were further examined by Churchill (1962, unpublished), who also de- Journal of the Royal Society of Western Australia, Vol. 58. Part 1. May, 1975. 2 scribed the microfloras from lignites at Esper- ance and Norseman, and samples from Albany Bore No. 6 (near Lake Munrillup, north of the Stirling Ranges). Upper Eocene plant microfos il assemblages from the Plantagenet Group in the Albany area were examined by Ingram (in Cockbain 1968) and similar microfloras have been found in other bores in the Albany area by the Geological Survey of Western Australia (unpublished reports) . Material The material on which this study is based consists of 17 sludge samples from Werillup No. 17, a water bore which is located about 9.5 km west of Albany townsite at 35‘’02'05"S latitude and 117°48'20"E longitude (Fig. 1). The bore was drilled by the Geological Survey of Western Australia to a depth of 61.9 m. from an eleva- tion of about 12 m above sea level. Werillup No. 17 is one of a series of bores that are named after a nearby trigonometrical station. Werillup No 17 was drilled by a percussion rig in 1968 and sludge samples were generally col- lected at 3 m intervals. The lithology of the bore and the location of the samples down the hole is shown in Fig. 2. The top 15 m of the hole consists of Quaternary quartz aeolianite and this has not been examined palynologically. From 16.8 m to 49.4 m the strata consist of dark coloured silts and silty sands, and from 49.4 m to 61m of kaolinitic clay. The bottom 0.9 m of the hole was logged as weathered granite and the bore ended in granite. Ingram (1969, unpublished report) reported Gramineae and Compositae pollen in samples from 16.8 m to 18.9 m, but these may be modern contaminants. The white clay from 49.4 m to the bottom of the hole can be interpreted as weathered granite; however, as it contains a microflora that does not differ greatly from samples higher up the bore, it is considered to be a sediment. Thus, from 16.8 m to 61 m the sediments are interpreted as belonging to the Werillup Formation. Palynological techniques The technique used to macerate and concen- trate the acid-insoluble microfossils was a modified hydrofluoric acid — Schultze’s solution — alkali technique similar to that outlined in Figure 2. — Diagram showing sample depth and lithology in Werillup No. 17 borehole and the distribution of the pollen and spores of the major plant groups. The dinoflagellates are shown as a percentage of the total number of spores and pollen. Journal of the Royal Society of Western Australia, Vol. 58, Part 1, May. 1975. 3 Balnie & Hassell (1962). Further concentration was effected by heavy liquid separation in a zinc chloride solution of specific gravity 2.0. Permanent strew mounts were prepared from each residue by smearing a drop of the residue on a cover slip with Clearcol (a mounting medium, H. W. Clark, Melrose, Mass., U.S.A.). The spores and pollen are able to settle in favourable orientations close to the coverslip while it is left to dry. The cover slip was then fixed to a slide with a drop of Xam (G. T. Gun* Ltd, London). Single grain mounts of some of the species were prepared by placing the specimens in glycerine jelly under a cover slip sealed with beeswax. The residues contained, with few exceptions, large numbers of spores, pollen and other acid- insoluble plant I'emains. Algal (dinoflagellate) cysts and reworked spores occurred in some samples. Species counts were made for all except three samples which were practically barren of spores and pollen. Most samples required three or more slides to be counted to give a representa- tive number of grains (usually about 100). The total pollen content was not determined as the samples are a sludge that may have resulted from the mixing of highly contrasting litho- logies. Also, the samples may have been con- taminated from higher up in the borehole (discussed below). All samples, residues and slides used in this study are stored in collections of the Depart- ment of Geology, University of Western Aus- tralia, and all numbers given to slides and samples are from the general catalogue of the Department of Geology. The location co- ordinates cited in the text after the slide numbers for each specimen refer to the stage of Reichert microscope. No 256,251. A reference slide with a located point is included in the slide collection. The specimens were photo- graphed using a Leitz Orthoplan microscope and a Leitz Orthophot camera body. Composition and age of the microflora Most of the spores and pollen found in the Werillup samples can be referred to previously described species and these are listed in Table 1. However, there are some species that do not appear to have been previously described or are slightly different from existing species, and these are listed in Table 2. Unfortunately, these species are too rare to determine the signifi- cance of the differences: or to warrant detailed description at present. The microflora in Werillup No 17 was ex- tracted from sediments that have been pre- viously placed in the Upper Eocene Werillup Formation (Cockbain 1968). The sediments appear to be no younger than this, as the following species (see Table 1) are restricted Table 1 List of 'previously described species found in Werillup No. 17 borehole. Ranges shown are those given in the literature for the species in southern Australia and New Zealand. 26. 36. 12 . 1. 27. 40. 9. 2 . 22 . 21 . 14 24. 19. 3. 10 . 4. 23. 5 . 20 . 32. 33. 13. 34. 8 . 6 . 7. J.5. 17. 16. 38. 15. 18. 28. 11 . 39. 30. 37. 29. 31. 25. 41. SPORE-POLLEN SPECIES (ALPHABETICAL LISTING) Bfinkson id i tcs minimus Bcnuprcfi id i tcs clcgansi f ormis C iciit r icos i spar i tcs pscudot r ipart i tiis Cingut r i ictcs cifivus Cupnn i c id i t cs ortbotci elms C. reticularis Cyathid i tcs minor Cycadopites sp. Daerycarpi tes austral iensis Dacryd iumi tcs f I or ini i Di Iwini tcs granulatus D. tubcrculatus Haloragac idi tes hair is i i Laevigatosporites ovatus Li I i acid i tcs variegatus Lycopod i umspor ites oust roc I ava t id i t cs .Valvac ipol I is diversus .Vicrocachryiditcs antarct icus .Vyrtaceidi tcs cucalyptoidcs il. mesonesus ,V. parvus Nothofagidites sp.fhrassi group) N. sp. (fusca group) Phyl loclodidi tcs mawsoni i Podocarpidi tcs cl 1 ipt icus P. microrct iculoidatus Polycolpitcs csobal tcus Protcacidites ixlcnanthoidcs P. annularis P. concrctus P. granulatus P. incurvatus P. pachypolus P. parvus P. rcticulatus P. subscabratus SimpI icepol I is scabratus Tricolporitcs microrct iculatus T. prolata 'Triorites' psilatus Tr iporopollcnitcs gemma t us 27 3 12 4 28 17 13 5 29 30 31 32 .33S04 6 35 14 36 37 38 9 18 10 15 16 39 40 41 19 20 42 21 11 43 22 23 24 25 44 26 LATE CRETACEOUS PALEOCENE EOCENE OLIGOCENE MIOCENE PLIOCENE QUATERNARY SPORE-POLLEN SPECIES (STRATIGRAPHICAL LISTING) C. clavus Cycadopites sp. L. ovatus L. aus t roc I ava t id i t cs il. antarct icus P. clipticus P. microrct iculoidatus P. mawsoni i C. minor L. variegatus P. parvus C. pscixiotr ipart itus N. (hrassi group) D. granulatus P. granulatus P, annularis P. adenanthoides P. incurvatus H. harrisii ,1/ . e LK'.-j / yp f o ides D. florinii D. austral iensis .1/, d iversus D. tubcrculatus T. psilatus B. minimus C. orthoteichus P. pachypolus T. mic rorc t icul at us P. subscabratus T. prolata .!/. mesonesus il. parvus N. ( fusca group) P. csobal tcus B. elegansiformis S. scabratus P. concrctus P. ret iculatus C. reticularis T. gemma t us Journal of tUe Royal Society of Western Australia, Vol. 58, Part 1, May. 1975. 4 Table 2 List of species found in Werillup No. 17 'borehole that do not appear to have been previously described, or which could not be assigned with confidence to existing species due to iheir very rare occurrence. Bombacacidites sp., fig. 45, rare Ceratosporites sp. cf. C. equalis Cookson & Dettmann, 46, rare Clavatipollenites sp. cf. C. ascarinoides McIntyre, fig. 47, rare Liliacidites sp. cf. L. aviemorensis McIntyre, fig. 48, rare Liliacidites sp., fig. 49, common Monosulcites spp., figs 50 and 51, rare Polypodiidites sp., fig. 52, rare Polyporina sp., fig. 53, common Proteacidites sp. cf. P. annularis Cookson, fig. 54, common Proteacidites sp. cf. P. crassus Cookson, fig. 55, rare Proteacidites sp. cf. P. minimus Couper, fig. 56, rare Proteacidites sp. cf. P. parvus Cookson, fig. 57, rare Proteacidites sp. 1, fig. 58, common Proteacidites sp. 2, fig. 59, rare Proteacidites sp. 3, fig. 60, rare Proteacidites spp., figs 61 to 68, abundant Retitricolporites sp., fig. 69, rare Tricolpites sp. cf. T. aspermarginis McIntyre, fig. 70, rare Tricolpites sp. cf. T. lilliei Couper, fig. 71, rare Tricolpites sp. cf. T. matauraensis Couper, fig. 72, rare Tricolpites sp. cf. T. pachyexinous Couper, fig. 73, rare Tricolpites sp., fig. 74, rare Tricolporites spp., figs 75 and 76, rare ‘Triorites' sp. cf. T. minisculis McIntyre, fig. 77, common ‘Triorites’ sp. cf. T. minor Couper, fig. 78, rare ‘Triorites’ sp. cf. T. orbiculatus McIntyre, fig. 79, rare ‘Triorites’ spp., figs 80 and 81, rare to the Eocene; Proteacidites concretus, P. reticulatus, Cuvanieidites reticularis (described from two localities in Victoria) and Triporopol- lenites gemmatus. Also, the Nanarup Limestone Member, which is stratigraphically higher than the Wenllup No 17 sequence is no younger than uppermost Eocene (Quilty 1969). The age of the base of the Werillup Formation has not been determined precisely and whether it extends into Middle Eocene cannot be resolved satisfactorily with the present data. The species identified as Proteacidites incurvatus and '‘Triorites” psilatus require further study before their chronostratigraphic significance can be fully assessed. Triporopollenites gemmatus is the only species useful in placing a lower limit on the age of the microfiora as it is known only from Middle and middle-Upper Eocene strata in South Australia and the Great Artesian Basin (Harris 1972). Some of the dinoflagel- lates in the samples, e.g. species of Wetzeliella (figure 82) and Cordosphaeridium (figure 83), support the assignment of the sediments to the Upper Eocene (Dr. B. E. Balme 1972, personal communication) . Several specimens of reworked spores were encountered in the uppermost samples (figures 84, 85 and 86) and they have been identified by Dr. B. E. Balme as characteristic Lower Cretaceous forms. Lower Cretaceous sediments are found in the Eucla Basin (Ingram 1968) and in the southern Perth Basin (Lowry 1965). However these are too far away to be considered as possible sources. There may have been pockets of Lower Creta- ceous sediments nearby that were eroded during the transgression and have since been covered up or completely removed. Comparison of the microflora with other assemblages The microfloral assemblage from Werillup No 17 is similar to Late Eocene microfioras previously described from southern Australia and New Zealand. Differences between the Werillup microflora and other assemblages from sediments correlated with the Plantagenet Group by Cookson (1954b), Balme & Churchill (1959) and Churchill (1962, unpublished) have no stratigraphical significance as the species are long ranging. They may have a phytogeo- graphicai significance, but there are insufficient data on which to base evaluations. Cookson (1954b) correlated the microflora from the Plantagenet Group with Microflora C from Victoria, mainly on the basis of Proteaci- dites pachypolus, which is now known to be long-ranging. Harris (1971) related Microflora C to the Triorites magnificus Zonule of the Otway Basin (Middle to Late Eocene). There are only 5 species from the Werillup micro- flora that are common with the T. magnificus Zonule, but they are of limited stratigraphical value as they also occur in the uppermost Palaeocene Cupanieidites orthoteichus Zonule. Correlations with the other palynological zonules of Harris (1971) are also not possible. The Upper Eocene "Aglaoreida barungensis” (Harris, unpublished) Zonule could not be recognized as it does not have any characteristic species that are present in the Werillup bore- hole. The Werillup microfiora contains at least 14 species that are present in the informal Oligocene V errucatosporites Zonule in the Otway Basin. Most of the species in the microfiora are found in the Early Tertiary Unit 1 set up by Hekel (1972) and in the Late Eocene micro- floras of New Zealand (Couper 1960, Wilson 1968). However, the species only allow a broad correlation. McQueen et al. (1968) note that from the Kaiatan to Runangan there is an abrupt change from the Nothofagus fusca group to a dominance of the Nothofagus brassi group. The latter group is similarly dominant in the Werillup microflora and it is the same age as part of the Runangan Stage of New Zealand. The apparent uniformity of the microfloras throughout southern Australia and New Zealand during the Late Eocene suggests that the cli- mate over the region was very uniform. Palaeoecology of the microfiora The palaeoecological interpretation of a micro- flora depends on the reliabilty with which the spores and pollen can be related to modern plants. Such determinations become less reli- able as older spores and pollen are examined and are usually not attempted with pre-Tertiary fossils. However, many of the species in the Werillup microfiora have been related fairly Journal of the Royal Society of Western Australia, Vol. 58, Part 1, May, 1975. 5 Journal of the Royal Society of Western Australia, Vol. 58, Part 1. May, 1975. 6 confidently to living genera and families. Most workers accept that the ecological tolerances of the plants that are represented in Early Terti- ary microfloras have not altered dui’ing the Cainozoic (Burbidge 1960). The microfiora as a whole will be discussed first, and then the variations within the microflora. The palaeoecology of the asserriblage The general microfloral assemblage in the samples does not differ greatly from other as- semblages described in the Plantagenet Group, and the additional species encountered do not alter the palaeoecological conclusions of pre- vious authors. Three floristic elements have been recognized in the Lower Tertiary floras of southern Australia (Burbidge 1960). They have been called the ‘Antarctic’, ‘Tropical’ and ‘Aus- tralian’ elements, and the spores and pollen that belong to these groups will now be dis- cussed. The ‘Antarctic’ elements in the microflora are species of Nothofagidites and Podocarpaceae whose major development has been in the south- ern hemisphere and appear to have migrated to Australia and New Zealand from Antarctica. Nothofagidites is the most commonly repre- sented pollen in the assemblage ( though pollen from other plant groups is more abundant in several samples) and it is predomi- nantly of the type that has been closely related to the pollen of species of the Nothofagus hrassi group (Cookson & Pike 1955). The Nothofagus hrassi group at present grows in areas of moderate to high rainfall in New Caledonia (above 600-900 m) and in New Guinea (above 2 400 m). It indicates a climate of constant humidity and one warmer than that occupied by the N. fusca . group (McQueen et al. 1968). Pollen from the hrassi-. group is recorded from Eocene to Pliocene in New Zealand and is abun- dant in Queensland from Eocene to Miocene. The N. hrassi group is an evergreen forest domi- nant that requires dense forest for regeneration. It is a very heavy pollen producer and is prob- ably over-represented in the pollen spectrum (Cranwell 1964). Gymnosperms were probably not an important part of the vegetation as they make up only 9% of the microflora. They are mainly podocarps with extant genera now living in a wide range of climates. High proportions of Pcdocarpus pollen in a Quaternary core off Argentina was used as an indicator of cooler climates when compared with the proportions of pollen from Nothofagus, Cupresssaceae and Weinmania (Groot & Groot 1966). Dilwynites has a down-hole distribution similar to the conifers and has been com- pared to several living conifers (for example Callitris) , however, Harris (1965) considers an angiospermous affinity for the pollen more likely. Dacrycarpites autraliensis was compared to the pollen of Podocarpus section Dacrycarpus which has its major development in New Guinea (Cookson and Pike 1953a). Micro- cachryidites antarcticus and Phyllocladidites mawsonii are common compared with the other podocarps in the assemblage, and have been related to two living podocarps, Microcachrys tetragona and Dacrydium franklinii respect- ively, both of which are restricted to Tasmania (Cookson 1947). These podocarps flourish under cool temperate conditions with a mod- erate rainfall. Podocarpidites represents pollen with gen- eral podocarpaceous affinities, and which may now be found in plants living in both tropical and temperate conditions. A similar distribu- tion is found for a number of species of Dacry- dium, which have the same pollen as Dacry- diumites fiorinii (Cookson and Pike 1953b). Thus the mesophytic ‘Antarctic’ species of Nothofagus and the podocarps, which are now restricted to cool temperate and tropical sub- montaine regions, indicate a rainforest vege- tation. The moss Sphagnum is repi’esented in the assemblage by Cingutriletes clavus (Dett- mann 1963) and indicates swampy conditions. Lycopods are represented by Lycopodium- sporites, and Selaginella by Ceratosporites (Dettmann 1963). Pteridophytes make up a small proportion of the assemblage and several families are present, including tree ferns. Cyathidites minor has been related to ferns such as Cyatheaceae and Dicksoniaceae, and Cicatricosisporites resembles some modern spores of Anemia in the family Schizeaceae (Dettmann 1963). Laevigatos- porites and Polypodiidites have been compared to spores from Polypodiaceae. The pterido- phytes have a broad distribution in high rain- fall areas of the tropical and temperate regions and their presence in the microflora indicates conditions wetter than at present. The “Tropical” elements in the microflora from Werillup No. 17 are species that have been related to families now distributed mainly in the tropical and subtropical regions of Aus- tralia and the Indo-Pacific. The species are Beaupreaidites elegansiformis, Bomhacacidites sp., Cupanieidites sp., Malvacipollis diversus and Figures 3 to 22. — All figures at xlOOO unless otherwise stated. 3. — Beaupreaidites elegansiforviis Cookson. Slide 69428b. 132.9 X 42.5. 4.— Cingutriletes clavus (Balme) Dettmann. Slide 69445. Cycadopites sp. Slide 63428a, 129 0 X 47.5. Q.—Laevigatosporites ovatus Wilson and Webster. Slide 69444a, 115.7 x 24.9. 1 —Liliacidites variegatus Couper. Slide 69451. 8.—Lycopodiumsporites au.troclavatidites (Cookson) Potonie. Slide 69428b, 120.2 x 32.2. Q. —Nothofagidites sp. (Nothofagus hrassi group). Slide 69486. 10.— Phyllocladidites mawsonii Cookson. Slide 69450. ll.—Proteacidites parvus Cookson. Slide 69431a. 125 9 x 43.7. 12.— Cicatricosisporites pseudotripartitus (Bolkhovitina) Dettmann. Slide 69428b, 123.2 x 32.2. IZ.— Cyathidites minor Couper. Slide 69428b, 123.9 x 35.0. \A.—Microcachryidites antarcticus Cookson. Slide 69428b, 128.9 x 28.8. IZ.— Podocarpidites tllipticus Cookson. Slide 69428b, 133.8 x 45.6. 16.— Podocarpidites microreticuloidatus Cookson. Slide 69428a. 113.1 x 37.4. 17.— Cupanieidites reticularis Cookson and Pike. Slide 69428b. 125.2 x 35.7. IS.— Nothofagidites fp. (Nothofagus fusca group). Slide 69438b. 131.4 x 39.5. 19.— Proteacidites concretus Harris. Slide 69428b, 136.3 x 25.8. 20.™ Proteacidites granulatus Cookson. Slide 69428b, 132.7 x 25.8. 21.— Proteacidites pachypolus Cookson and Pike. Slide 69428b, 134.2 x 35.4. 22.— Proteacidites subscahratus Couper, Slide 69428b. 135.6 x 38.5. Journal of the Royal Society of Western Australia, Vol. 58, Part 1, May, 1975. 7 23 Journal of the Royal Society of Western Australia, Vol. 58. Part 1, May, 1975. 8 Myrtaceidites mesonesus. Additional “Tropi- cal” species previously described from the Plan- tagenet Group but not found in the present study are Anacolosidites sp., ‘"Palmidites” sp. and ‘‘Santaluviidites cainozoicus” . Beaupreaidites has been related to the pollen of Beauprea (Proteaceae) which occurs in New Caledonia and New Guinea (Cookson 1950). The determination of Bombacacidites sp. is not reliable, however, the pollen is common in the Eocene of New Zealand (Couper 1960). Leaves of Bombax (kapok tree) are recorded from the Plantagenet Group at Cape Riche (Chapman and Crespin 1934), thus there is evidence that the tropical species was present on the south coast of Western Australia during the Eocene. The two species of Cupanieidites in the Weril- lup microflora have been compared to the pollen from Sapindaceae tribe Cupanieae, which is a component of tropical and south- eastern Australian rainforests (Cookson and Pike 1954b). The family Malvaceae has its major distribution in the tropics at present, and is represented in the microflora by Malva- cipolUs diversus. Myrtaceidites mesonesus has been related to a species of V/hiteodendron now living in Indonesia (McWhae 1957). Anacolosidites is from the family Olacaceae in which the genera are now restricted to the tropics. It was pan-tropical from Palaeocene to Eocene, and was common in Borneo and Queensland throughout the Tertiary (Germc^raad et al. 1968, Hekel 1972). “ Santalumidites caino- zoicus” is considered (in part) a synonym of Florschuetzia levipoli (Germeraad et al. 1968). This species is similar to the pollen of Sonneratia caseolaris, a mangrove at present growing in estuaries along the Indo-Malesian coasts. The form genus 'Palmidites' is of un- certain reliability, however, palm pollen is re- corded from the Upper Eocene of New Zealand, and supports the evidence for warmth (McQueen et al. 1968). The above ‘Tropical’ species indicate that the Late Eocene climate on the south coast was sub-tropical or warmer. These ‘Tropical’ species are also represented elsewhere in the Late Eocene of southern Australia, and in New Zealand (Harris 1971, Couper 1960), and suggest that the Australasian region was characterised by a warm and humid climate during this period. The ‘Australian’ elements in the microflora are species such as Myrtaceidites eucalyptoides, Haloragacidites harrisii, Banksieaeidites and Proteaceidites. Banksieaeidites is similar to the pollen found in Banksia and Dryandra (Cookson 1950). Proteacidites adenanthoides has been compared to the pollen of Adenanthos and P. annularis has been related to the pollen of Xylomelum occidentale, a tree occurring in the Jarrah forest of the south-west of Western Australia (Cookson 1950). Other species of Proteacidites found in the Plantagenet Group by previous workers include P. rectomarginis, which has a possible affinity to Petrophile, and Proteacidites symphyone- moides, related to Symphyonema (Cookson 1950). The family Proteaceae diversified in the Late Cretaceous and possibly originated in the rain- forest environment of south-eastern Australia and South-West Pacific (Muller 1970). The Werillup microflora includes over 30 proteaceous species and these compose up to 24% of the total microflora. Proteaceous pollen is the most abundant pollen type in several samples, and it is more abundant towards the bottom of the borehole. The wide diversity and abundance of proteaceous pollen suggest that Proteaceae were a major constituent of the vegetation and that the climate was warmer than cool temperate. The form species Haloragacidites harrisii principally represents pollen from Casuarina, though other plants with similar pollen were probably present in the vegetation. The form species composes up to 16% of the total micro- flora and it is considerably more abundant in some samples. The less markedly aspidate pollen grains placed in the form species have been related to pollen from Haloragaceae and genera such as Geniostoma and Canacomyrica. These are now living mainly in the tropics (Mildenhall & Harris 1971). The broad palaeovegetative pattern gained from the spores and pollen is that of a tropical to subtropical rainforest (closed-forest). The predominance of tree pollen relative to the proportion of non-tree pollen is indicative of a rainforest (Churchill 1962, unpublished). The subtropical vegetation suggests a climate con- siderably warmer and more humid than the present climate along the south coast of West- ern Australia, which is mild, cool temperate, with a warm to hot dry summer. The southern beeches (Nothofagus) appear to have been forest dominants, with Proteaceae and Casuarina as subdominants. Other ele- ments of the forest were podocarpaceous coni- fers, pteridophytes and the ‘Tropical’ species. Figures 23 to 44.— All figures at xlOOO unless otherwise stated. 23. — Simplicepollis scahratus McIntyre. Slide 69473. 24. — Tricolporites microreticulatus Harris. Slide 69439a, 123.1 x 31.2. 25. — Tricolporites prolata Cookson. Slide 69428b, 132 x 40.7. 26. — Triporopollenites gemmatus Harris. Slide 69428d. 119.0x44.1 (single grain). 27. Bank- sieaeidites minimus Cookson. Slide 69439a, 127.0 x 26.4. 28. — Cupanieidites orthoteichus Cookson and Pike Slide 69428a, 121.7 x 50.7. 29. — Dacrycarpites australiensis Cookson and Pike. Slide 69428b, 125.8 x 30.2 30.~ Dacrydiumites florinii Cookson and Pike. Slide 6944 9. 31. — Dilwynites granulatus Harris. Slide 69447 S2.—Dilwynites tuberculatus Harris. Slide 69448. 33, 34. — Haloragacidites harrisii (Couper) Harris. Fig. 33’ slide 69453. Fig. 34, slide 69428a, 129.8 x 27.2. 35. — Malvacipollis diversus Harris. Slide 69439a, 123 7 x 48 l’ Myrtaceidites eucalyptoides Cookson and Pike. Slide 69471. 21 —Myrtaceidites mesonesus Cookson aiid Pike. Slide 69428b, 121.5 x 49.3. 22.— Myrtaceidites parvus Cookson and Pike. Slide 69472. 29 —Polycolpites esobalteus McIntyre. Slide 69428b, 132.7 x 32.8. 40. — Proteacidites adenanthoides Cookson. Slide 69444a 115 7 x 24.9. 41. —Proteacidites annularis Cookson. Slide 69439a, 133.5 x 39.3. 42.— Proteacidites incurvatus Cookson Slide 69431b, 121.3 x 29.5. 42.— Proteacidites reticulatus Cookson. Slide 69457. 44.—“Triorites’' psilatus Harris Slide 69428a, 121.8 x 45.5. Journal of the Royal Society of Western Australia, Vol. 58, Part 1, May, 1975. Journal of the Royal Society of Wesle.*n Australia, Vol. 58, Part 1, May, 1975. 10 The diverse ‘Australian’ element, especially the Proteaceae, is now characteristic of a climate with a marked dry season. It is difficult to explain the presence of this element in a vegetation growing in humid conditions and that includes Nothofagus, which cannot with- stand long periods of dry conditions. A Middle Eocene flora from Maslin Bay, South Australia includes Araucaria, Casuarina and Proteaceae, and a leaf macroflora which has a broad similarity to some leaf litters from present day Queensland wet forests (Lange 1970) . Thus, it is possible that the Early Tertiary rain- forest vegetation on the south coast included a high proportion of Proteaceae. An alternative interpretation is that the “Australian” element represents species from a flora further inland in a drier and more seasonal climate. Analysis of down-hole variations in the micro- flora Palaeoecological interpretations from palyno- logical data are subject to many reservations, and when based on the results from one bore- hole, must be regarded with even more caution. The importance of the relative pollen produc- tion of the plants and the hydrodynamic properties of the grains have been stressed: yet there is very little quantitative information on these parameters (Davis 1963, Brush yi and its hemiparasite, Amyema nestor under optimal and stress conditions. J. Ecol. 59: 351-363. (1971). — Ibid. V. Heat resistance limits of photosynthetic organs of different seasons, their relation to water deficits and cell sap properties and the regeneration ability. J. Ecol. 59: 365-374. (1971). — The inverted microscope for direct observation of stomata. Search 2: 142-143. Hellmuth, E. O.. and Grieve. B. J. (1969).— Measure- ment of water potential of leaves with par- ticular reference to the Schardakow method. Flora. Jena 159: 147-167. Holland. A. A. (1953). — Ecology of the south western and southern eremean Eucalvpts. M.Sc. Thesis. University of Western Australia. Pearman, G. I. (1965). — Preliminary studies of the loss of heat from leaves under conditions of free and forced convection. Aust. J. Bot. 13- 153-160. • (1966). — The refiection of visible radiation from leaves of some Western Australian species. Aust. J. biol. Sci. 19: 97-103. Shelton, W. E. (1921).— Xerophytism in the Swan River District. J. and Proc. Roy. Soc. W A 7 • 95-107. (1934). — Plant response to the dry phases of the climate of south Western Australia. Presidential address. J. Roy. Soc. WA 19- 13-36. Plant Physiology Broughton, W. J. (1968). — Infiuence of gibberellic acid on nucleic acid synthesis in dwarf pea inter- nodes. Biochim bicphys. Acta 155: 308-310. (1969). — Relations between DNA. RNA and protein synthesis and the cellular basis of the growth response in gibberellic acid- treated Pea internodes. Ann. Bot. 33: 130 227-243. Broughton. W. J., and McComb. A. J. (1967). — The relation between cell wall and protein syn- thesis in dwarf Pea plants treated with gibberellic acid. Ann. Bot. 31: 359-366. (1971). — Changes in the pattern of enzyme development in gibberellin-treated Pea inter- nodes. Ann. Bot. 35: 213-228. Broughton, W. J., Hellmuth, E. O., and Yeung. D. (1970). — Role of glucose in development of the gibberellin response in Peas. Biochim. biophys. Acta. 222: 491-500. Dakin, F. J. (1917/18). — The Western Australian Pitcher Plant (Cephalotus follicularis) and its phy- siology. J. and Proc. Roy. Soc. W.A. 4: 37-53. Edwards, E. T. (1935). — SUidies on Gibberella fujikuroi var. subglutinans, the hitherto undescribed ascigerous stage of Fusarium moniliforme var. subglutinans and on its pathogenicity on maize in New South Wales. Dept, of Agric. N.S.Wales. Sci. Bull. 49. Grieve, B. J. — The biology of Nuytsia floribunda. a root parasite. (Unpublished.) See a short note in "The West Australian”, 1963. Grieve, B. J.. and Doley, D. — Parasitic attack of Nuytsia floribunda (Western Australian Christmas Tree) on \mdergro\md electric cables. (Un- published.) Herbert, D. A. (1919).— See under Plant Anatomy. Jennings, R. C. (1968). — See under Algology. (1969). — Cytokinins as endogenous growth regulators in the algae Ecklonia (Phaeo phyta) and Hypnea (Rhodophyta) . Aust. J. biol. Sci. 22: 621-627. (1969). — Gibberellin antagonism by material from a brown alga. New Phytol. 68: 683-688. Jennings, R. C., and McComb, A, J. (1967).— See under Algology. Lamont, B. B. (19714-). — The effect of soil nutrients on the production of proteoid roots by Hakea species. Aust, J. Bot. (submitted for publication). (1971-h-). — The morphology and anatomy of proteoid roots in the genus Hakea. Aust. J. Bot. (submitted for publication). Lloyd, F. E. (1936). — Notes on Utricularia, with special reference to Australia, with descriptions of fo\ir new species. Vic. Nat. 53: 91-112. Main, A. (1947). — Artificial propagation of Nuytsia flori- bunda. W.A. Nat. 1: 1-7. McComb, A. J. (1965). — The control of elongation in Callitriche shoots by environment and gib- berellic acid. Ann. Bot. 29: 445-458. (1966). — The stimulation by gibberellic acid of cell-wall synthesis in the dwarf Pea plant. Ann. Bot. 30: 155-163. (1967). — The control by gibberellic acid of stem elongation and flowering in biennial plants of Centaurium minus Muench. Planta IBerl.) 76: 242-251. McComb. A. J., and McComb, J. A. (1970). — Growth substances and the relation between pheno- type and genotype in Pisum sativum. Planta (Berl.) 91: 235-245. McComb, A. J., and Rizvi, S. A. (1969).— The production of gibberellin by strains of Gibberella fuji- kuroi var. s^ibglutinans isolated in Aus- tralia : Confirmation of a suggestion made in 1935. Aust. J. Sci. 32: 161-2. Mott, J. J. (197H-). — Germination studies on some annual species from an arid region of Wes- tern Australia (in preparation). Wong, C. H.. and McComb, A. J. (1967). — An anatomical investigation into the effects of gibberellic acid on the expansion of Callitriche shoots. Aust. J. biol. Sci. 20: 1053-62. Appendix History of the development of botanical institutions in Western Australia In the course of the preceding survey emphasis naturally has been upon developments in botani- cal research in the different fields. But it may perhaps be helpful to illustrate how the advance- ment of botany has been related to personnel and establishments. Dealing first with the official Government appointments it may be noted that the positions of Economic Botanist (Stoward, 1911-17; Her- bert, 1918-21; Praaf, 1921-22; Campbell, 1922-23; Came, 1923-28), Government Botanist (Gardner, 1929-60) and Ofticer-in-Charge, Botany Section, and Curator of the Herbarium (Royce, 1960+) were held within the State Department of Agri- culture. The State Herbarium has been succes- sively housed in part of the Observatory Building (1933-58), in the Department of Agriculture Laboratories, Jarrah Road, South Perth (1958- 69), and finally from March, 1970, in its own new building. Currently a staff of nine profes- sional botanists is engaged on taxonomic and e+logical studies. A new journal “Nuytsia” pro- vides for the publication of the results of their researches. Botany in the University of Western Australia developed in the Biology Department, commenc- ing in 1914. Professor W. Dakin (primarily a zoologist but with an interest in the physiology Journal of the Royal Society of Western Australia, Vol. 58. Part 2. July, 1975. 52 of plants as shown by his study of Cephalotus follicularis, the Albany pitcher plant, and his translation of a large part of Diels’ “Die Pfian- zenwelt von West Australien” ) and Dr. Cayzer (taxonomic botanist) comprised the staff. Pro- fessor Dakin was succeeded by Profes- sor Nicholls (zoologist) in 1921 and Dr. Cayzer by Miss E. Reed about the same time. Towards the end of the 1920s Botany achieved autonomy and moved from the Biology build- ing north of Stirling Highway to a separate building on the main campus at Crawley. Dr. Armstrong succeeded Miss E. Reed as Head of Department in 1930-31 and with Miss A. M. Baird, (who was appointed to the staff in 1934), continued until world war II commenced in 1939. On the outbreak of war Dr. Armstrong joined up and Miss Baird acted as Head of Department from then until 1947 when the influx of ex- servicemen necessitated expansion of staff. Dr. B. J. Grieve was appointed as Head in 1947 and foundation Professor in 1956. In 1970 a new Botany building well equipped for studies in descriptive and experimental botany was completed and occupied on Hackett Drive at the southern end of the campus. A small teaching botany garden was developed being laid out to show a possible evolutionary sequence of plant development and incorpora- ting a variety of ecological niches. The profes- sional staff at that time numbered nine and the major facets in Botany were represented by specialists. In 1959-60, following a report by a Govern- ment appointed committee and a further recom- mendation by Dr. W. Stewart, Director of the Los Angeles State and County Arboretum (who at the time was a Fulbright Scholar in the University Botany Department) , a Botanic Gar- dens essentially for native plants was approved for establishment in King’s Park. In 1961 Dr. J. S. Beard was appointed as the first Director. Following four years of development the Botanic Gardens was officially opened in 1965. Botanists both professional and amateur, through publications in the Royal Society of Western Australia, the Western Australian Naturalists’ Club and a variety of specialist Aus- tralian and overseas scientific journals, have contributed considerably to the development of botany in this State as outlined in the course of the preceding presidential address. Journal of the Royal Society of Western Australia, Vol. 58, Part 2, July, 1975. 53 4. — New finds of sand fulgurites from the Perth Basin, Western Australia by J. E. Glover^ Manuscript received 17 September 1974; accepted 19 November 1974 Abstract The first fulgurite recorded in Western Aus- tralia was recovered from West Popanyinning in 1931. In 1974 fulgurites were recorded from Willetton, East Victoria Park, Canning Vale and near Broome, and they are now reported from Jandakot, Lynwood, Welshpool, Malaga, Beech- boro. Upper Swan, Wanneroo, Guilderton and near Northampton. Eleven of the fourteen localities are in the central Perth Basin, and ten are near Perth. The spate of recent finds near Perth is due to close local investigation. Fulgurites are also probably fairly common in the rest of the central Perth Basin, which is similar in its topography and sandy soil, and has about the same annual lightning frequency. The fairly common association of Aboriginal artifacts with fulgurite fragments in sand blow- outs is probably generally fortuitous, because both remain as lag when sand is removed. However, isolated fulgurite fragments do not necessarily post-date associated artifacts. Introduction Sand fulgurites are glassy, tube-like bodies fused from sand by lightning on or near the Earth’s surface. Reports of fulgurites seen to form from lightning strikes have been presented by Pfaff (1822), Wicke (1859), Van Bastelaer (1883), Wood (1910), Simpson (1931), and Fen- ner (1949). Beadle (1940) has shown fairly conclusively that bushfires do not attain the temperatures necessary to fuse quartz (Rogers 1946) and thus simulate fulgurites. Fulgurites generally consist of vesicular lecha- telierite with a mean refractive index close to 1.461 ± 0.002, and have a smooth, translucent, vitreous lumen with a rough, opaque, commonly flanged exterior containing embedded quartz grains, or partly fused quartz grains. Most fulgurites recovered recently from Western Aus- tralia are fragmentary, probably because of sand movement after the fulgurites formed, and represent parts of tube walls and flanges. The original appearance of the fulgurites can be reconstructed from two that were found in situ and unbroken, namely the West Popanyinning fulgurite (Simpson 1931) which was tubular, highly flanged, vertical and about a metre long, and the East Victoria Park fulgurite (Glover 1974) which was tubular, vertical and may have been over two metres long. Tube walls are commonly one or two millimetres thick, and generally contain vesicles with their long axes normal to the surfaces. It has been shown on chemical grounds that fulgurite fragments from » Geology Department. University of Western Australia, W.A. 6009. Willetton were formed from the fusion of the white sand in which they were found (Glover 1974). The newly recorded material, which is all fragmentary, is described briefly, and its signi- ficance is discussed. The colours and corres- ponding numerical designations refer to the Rock-color Chart distributed by the Geological Society of America (Rock-color Chart Com- mittee 1963). Each fulgurite fragment, or group of fragments from one locality, has been allotted a number by the Geology Department of the University of Western Australia, with the excep- tion of the Wanneroo object, on loan from the Commonwealth Scientific and Industrial Re- search Organisation, Floreat Park. The distri- bution of the fulgurite localities is shown in Figures 1 and 2. Petrography General Most of the fulgurite fragments have been recovered from sand blow-outs, or from com- mercial sandpits, and because of probable sand movement, their precise stratigraphic position is uncertain. It is clear in some places, however, that they come from about a metre or less below the original surface. Aboriginal artifacts, dominantly quartzite and chert flakes, have been found at ten of the fourteen localities. The newly recorded material consists mainly of irregularly shaped, crinkled to roughly flat fragments up to 4 cm x 2 cm x 1 mm in size, but there are some tubular fragments up to 4 cm long. One side of each fragment is almost invariably vitreous, translucent, smooth and somewhat mammilated: this side is always the lumen in tubular fragments. The other side is rough, opaque and contains embedded sand grains. The glass is not uniformly coloured even within the one fragment, and the smooth side of the fragments commonly ranges from very light grey (N8) to light grey (N7), but other colours including white (N9), pinkish grey (5YR8/1), greyish orange (10YR7/4), dark yel- lowish brown (10YR4/2), medium dark grey (N4), and greyish black (N2) have been ob- served. Dark grey (N3) to black (Nl) schlieren about one mm long, or smaller spots, are com- monly found in the light-coloured glass. The overall colour of the rough outer surfaces is generally a little different because of embedded sand grains and superficial iron staining. Under the microscope, some quartz sand grains show cracks filled with glass, or grade Joiirnal of the Royal Society of Western Australia, Vol. 58, Part 2, July, 1975. 54 Fig. 1.— Locality map of Western Australia showing Perth Basin. Journal of the Royal Society of Western Australia, Vol. 58. Part 2. July, 1975. 55 INDIAN OCEAN Fig. 2. — Map of Perth area showing fulgurite localities (small solid circles). Journal of the Royal Society of Western Australia, Vol. 58, Part 2, July, 1975. 56 Pig. 3. — Wanneroo fulgurite (left) and two fragments from the Widgee Road Sandpit, Beechboro (centre and right). Note the holes in the Wanneroo material. The Widgee Road material has a smooth, shiny inner surface, and a rough outer surface with embedded sand grains. Length of Wanneroo tube 2.8 cm. into glass, and the glass of the walls is finely vesicular. The mean refractive index of the glass ranges from 1.460 to 1.467 ± 0.002, but is most commonly within the range 1.461 to 1.463 ± 0.002. Dark glass from a parti-coloured frag- ment commonly though not invariably has a slightly higher refractive index than colourless glass from the same fragment. The Wanneroo object, listed with the others below, is unique. It is an unfianged tube in which the lumen is smoother than the outer surface, but the clearly discernible sand grains characteristically embedded in the outer surface of other Western Australian fulgurites are not evident. The tube walls are unusual in contain- ing a network of irregularly shaped holes up to 2 mm in diameter (Fig. 3). The localities and main features of the newly recorded material are listed below. Jandakot Lechatelierite, Geology Dept. No. 73365 Locality: Jandakot Readymix Sandpit, Forrest Road (see Fig. 2). No. of fragments: Two (includes one tube). Associ- ated artifacts. Colour: Very light grey (N8) to light grey (N7) with a few black (Nl) spots. Mean refractive index: 1.460 ± 0.002. Lynwood Lechatelierite, Geology Dept. No. 73366 Locality: Sand blow-out north of Bannister Lagoon on Riley Road (see Fig. 2). No. of fragments: One. Associated artifacts. Colour: White (N9) to very light grey (N8) with a few black (Nl) spots. Mean refractive index: 1.461 ± 0.002. Welshpool Lechatelierite, Geology Dept No 73367 Locality: Sand blow-out north side of Dowd Street near its eastern end. north of Welshpool Road (see Fig. 2). No. of fragments: Two (includes one tube). Associ- ated artifacts. Colour: Light grey (N7) with a few black (Nl) spots. Mean refractive index: 1.461 ± 0.002. Malaga Lechatelierite, Geology Dept. No. 73368 Locality: Road cutting in sand, east side of Beech- boro Road, 1.6 km north of King Road (see Fig. 2). No. of fragments: One. Associated artifacts. Colour: Pinkish grey (5YR8/1) with medium dark grey (N4) to black (Nl) spots. Mean refractive index: 1.461 rt: 0.002. Beechboro Lechatelierite, Geology Dept. No. 73369 Locality: Disused sandpit, Widgee Road (see Fig. 2), No. of fragments: 250, with numerous tubes (see Pig. 3). Associated artifacts. Colour: Light grey (N7) to greyish black (N2). Mean refractive index: 1.463 ± 0.002. Upper Swan Lechatelierite, Geology Dept. No. 73545 Locality: 20 km N.N.E. of Perth, Bell Bros. Sandpit, northern side of Gnangara Road, 2.1 km west of West Swan Road turn-off. North of area covered by Figure 2, and not shown. No. of fragments: Three. Associated artifacts. Colour: Very light grey (N8) to light grey (N7) with dark grey (N3) to black (Nl) spots and streaks. Mean refractive index: 1.462 rt 0.002 with some darker glass close to 1.464 ± 0.002. Wanneroo Lechatelierite, C.S.LR.O. No. 9073 Locality: Sand at Wanneroo, about 25 km north of Perth, precise locality not recorded. No. of fragments: One tube. Colour: Very light grey (N8) with rare small black (Nl) spots. Mean refractive index: 1.463 ± 0.002. Guilderton Lechatelierite, Geology Dept. No. 73370 Locality: Sand blow-out near mouth of Moore River, south bank, opposite Guilderton (see Fig. 1). No. of fragments: Five. Associated artifacts. Colour: Greyish black (N2) to dark yellowish brown (10YR4/2). Mean refractive index: 1.461 ± 0.002. Northampton Lechatelierite, Geology Dept. No. 73371 Locality: Sand blow-out 3 km west of Howatharra Homestead, which is 22 km south of Northamp- ton on Highway No. 1 (see Fig. 1). No. of fragments: One. Associated artifacts. Colour: Dark grey (N3) to greyish orange (10YR7/4). Mean refractive index: Mainly close to 1.463 ± 0.002, but ranging from 1.461 to 1.467 0.002. Discussion Some sandy areas of the Earth, such as the south-eastern portion of the Kalahari Desert in southern Africa, are notable for their concen- tration of fulgurites, and Lewis (1936) in his discussion of the region guessed that there were not less than 2,000 fulgurites within an area of about 20 sq km. He thought it reasonable, after questioning local inhabitants, to assume that the area had been struck recently only about once every fifty years by lightning. He there- fore proposed that the sands might be about 100,000 years old, if there had been unchanged climate throughout. Journal of the Royal Society of Western Australia, Vol. 58. Part 2, July, 1975. 57 Meteorologists chart thunderstorm activity by isobronts, lines that join places of equal annual thunderstorm activity. The central Perth Basin occupies a 20 to 30 isobront area, a thunder- storm frequency described by the Bureau of Meteorology (1967) as “relatively high”, and exceeded in Western Australia only in the tropical northern part of the State. The West- ern Australian statistics suggest a far higher annual frequency of lightning in the central Perth Basin than that now experienced in the south-eastern Kalahari area, if the assumptions of Lewis are accepted. It should be added, how- ever, that the isobrontic map of Ramakrishnan and Rao (1955) shows a higher thunder fre- quency in the Kalahari than around Perth. The Western Australian statistics suggest a far higher annual frequency of lightning in the central Perth Basin than that now experienced in the south-eastern Kalahari area, if the assumptions of Lewis are accepted. It should be added, how- ever, that the isobrontic map of Ramakrishnan and Rao (1955) shows a higher thunder fre- quency in the Kalahari than around Perth. Ten of the fourteen known Western Australian fulgurite localities are also rich in Aboriginal artifacts. There is not necessarily any signi- ficant connection between the association of artifacts and fulgurites: the former are normally sought in blown-out sandy areas because they are concentrated as wind removes the sand, and the fulgurite fragments, which are similarly concentrated, were found when the localities were examined for artifacts. The time of arti- fact manufacture in Western Australia ranges from the ethnographic present back 25,000 years and probably longer (see Dortch and Merrilees 1973). It is known, in the light of the West Popanyinning event, that fulgurites are forming at present, but it cannot be assumed that isolated fulgurite fragments are necessarily younger than the artifacts with which they were found. It is not possible to say what proportion of the recovered fulgurite material formed under the current climatic regime. The recent spate of fulgurite finds in the Perth area, which is similar in its topography, sandy soil and meteorology to the surrounding region of the central Perth Basin, is probably a reflection of the high number of studied sites of former aboriginal occupation, rather than any special abundance of fulguritic material. Fulgurites are therefore probably fairly common throughout the central Perth Basin. Other sandy areas in Western Australia, such as the Gibson Desert and part of the Great Victoria Desert, are of equal isobrontic frequency, and the northern part of the Great Sandy Desert has a higher frequency. These areas will almost certainly contain numerous fulgurites. Acknowledgements. — Fulgurite fragments from Guil- derton, Upper Swan and Jandakot were supplied by Mrs. S. J. Hallam, Department of Anthropology, Univer- sity of Western Australia, and the Welshpool material was supplied by Miss A. D. McConnell, Department of Geology, University of Western Australia. The Wan- neroo material was loaned by Dr. D. R. Hudson. Com- monwealth Scientific and Industrial Research Organisa- tion, Ploreat Park. References Beadle, N. C. W. (1940). — Soil temperatures during forest fires and their effect on the survival of vegetation. J. Ecol. 23: 180-192. Bureau of Meteorology (1961) .—Average Annual Thun- der Day Map of Australia (1954-1963) with Explanatory Notes. (Bureau of Meteorology, Department of the Interior.) Dortch, C. E., and Merrilees, D. (1973).— Human occupa- tion of Devil’s Lair, Western Australia, dur- ing the Pleistocene. Arch. Phys. Anthrcp. in Oceania 8: 89-115. Fenner, C. (1949). — Sandtube fulgurites and their bear- ing on the tektite problem. Rec. S. Aust. Mus. 9: 127-142. Glover, J. E. (1974).— Sand fulgurites from Western Australia. J. Roy. Soc. West. Aust. 57: 97- 104. Lewis, A. D. (1936).— Fulgurites from Witsands on the south-eastern borders of the Kalahari. S.A. Geogr. Jl. 19: 50-57. Pfaff, C. H. (1822).— Beobachtete Entstehung einer Blitz- rohre durch den Blitz. Ann. Phys. (Gilbert’s Ann.) 72: 111. Ramakrishnan, K. P., and Rao. D. S. V. (1955). — D.stri- bution of thunderstorms over the world. Ind. J. Meteorol. Geophys. 6 (2): 1-5. Rock-color Chart Committee (1963). — “Rock-color Chart’' (Geol. Soc. Am., New York). Rogers A F. (1946).— Sand fulgurites with enclosed lechatelierite from Riverside County. Cali- fornia. J. Geol. 54: 117-122. Simpson, E. S. ( 1931) .—Contributions to the mineralogy of Western Australia — Series VI. J. Roy. Soc. West. Aust. 17: 137-149. Van Bastelaer, D.-A. (1883).— Sur un fulgurite forme en presence de pluslers t^moins a Gougnies pres de Charleroi. Bull. Acad. Roy. Sci. Brux. 6 (3rd Ser.): 144-152. Wicke, W. (1859).— Directe Beobachtungen fiber Entste- hung von Blitzrohren. Ann. Phys. 106: 158- 159. Wood. R. W. (1910). — Experimental study of fulgurites. Nature, Lond. 84: 70. Journal of the Royal Society of Western Australia, Vol. 58, Part 2. July. 1975. 58 5. — Geometric microliths from a dated archaeological deposit near Northcliffe, Western Australia. by C. E. Dortch’ Manuscript received 11 March, 1375; accepted 15 April, 1975 Ab.stract Preliminary examination and a test excavation made in a sandy podzol overlying a silcrete formation near Northcliffe. Western Australia, established that prehistoric man quarried the silcrete and used the immediate locality as a factory for manufacturing stone tools. Two radiocarbon dates based on charcoal samples collected at the excavation site indicate that geometric microliths were manufactured there from about 6000 to about 3000 years BP, and that the site had been a silcrete quarry-factory for some time previous to this. Analysis of pollen samples taken from the dated deposit show that Eucalyptus diversicolor, E. calcphylla, and E. marginata existed in the locality prior to about 6780 years BP and that the two former species and possibly the latter were present at times since. Introduction Geometric microliths are very small, abruptly retouched stone tools known in many Old World stone industries dating to the terminal Pleisto- cene and Recent periods, and they are a characteristic feature of stone artifact assem- blages from many districts in the southern two- thirds of Australia (Mulvaney 1969, Fig. 28). Radiocarbon assay has shown that geometric microliths were used in Australia from about 5000 to 6000 years BP until Modern times (Pearce 1974, Table 2). Various prehistorians (e.g. Mulvaney 1969) have interpreted the in- troduction of geometric microliths and a range of other distinctive forms of flaked stone tools (e.g. tula adze flakes and invasively flaked points) into the archaeological sequences of a number of Australian sites as a definitive event marking the beginning of a phase or period in continental prehistory. The Northcliffe quarry-factory site Several years ago Mr. G. Gardner, a naturalist from Northcliffe, W.A., identified an outcropping formation of sedimentary rock located about 10 km west of Northcliffe (Pig. 1) as an Aboriginal stone quarry. The rock has since been identified as silcrete (J. E. Glover, pers. comm.). On different occasions Mr. Gardner collected quarrying debris and stone tools from an area of several hectares around the silcrete forma- tion, and he found silcrete artifacts in the section of a bulldozed cutting in a sandy soil overlying the formation. In 1973 I took a char- 1 Western Australian Museum, Francis Street Perth W.A. 6000 Figure 1.— Map of the south west, Western Australia showing locations of archaeological sites mentioned in the text. 1, Northcliffe quarry-factory; 2, Frieze Cave; 3, Devil’s Lair. coal sample from the upper part of the cutting face at a position (50 cm below a datum level later established at the site) where several artifacts (B1716) were visible. The sample, KS 1 (Table 1), has since been radiocarbon assayed at 3080 ±: 75 years BP (ANU 1131; H. Polach, pers. comm.). Mr. W. M. McArthur of the Division of Land Resources Management, CSIRO, Perth, who was present at the site, identified the sandy deposit Journal of the Royal Society of Western Australia, Vol. 58, Part 2, July, 1975. 59 as an iron humus podzol (pers. comm.). He also found several flakes (B1715) 30-40 cm deep (i.e. at about the same depth as the artifacts visible in the section) in a 15 x 20 cm test pit dug into undisturbed soil about 2 m away from the cutting. He and I regard these flakes as evidence that occupation took place on the surface of the sand as it accumulated, and that the artifacts we found in the face of the cutting are in situ. In February 1974, Mr. Gardner, Miss A. Mc- Connell and I excavated a lx 1.5 m test trench (Trench 1) in the bulldozed cutting at the position where a year previously I had collected charcoal for sample KS 1. We excavated Trench 1 in two, three and four cm arbitrary levels and screened all excavated material through three or five mm sieves. The deposit as shown in the north section of Trench 1 (Fig. 2) shows a surface plant fibre and leaf zone overlying a dark humic zone about 20 cm thick. This merges into a leached zone about 50 cm thick. At the base of the leached zone is a five cm thick layer of dark sand which in turn rests on a hard pan of iron stained, cemented sands. The first stone artifacts we excavated in Trench 1 were in the upper part of the leached zone at a depth of 46 to 51 cm below datum. Below this, the number of artifacts increased to a depth of about 75 cm and continued in smaller numbers to the surface of the basal cemented sands (Fig. 2, Table 2). This surface contained several pits or channels 10 to 30 cm deep, and a few stone artifacts were excavated from these. We did not continue systematic excavation into the cemented sands; we screened about 20 kg of these sediments through a five mm sieve but found no archaeological material. Table 1 Kadiof'.arbon dates^ hafted on charcoal samples collected from Trench 1, yorthcliffe quarry-factory site. Western Australia Sample Date Kadiocarboii age in years BP Provenance, with depth below datum KS 1 AXU 1131 3 080 i 75 upper part of leaclied zone, 50 cm KS 2 SUA 379 6 780 ± 120 lower part of leached zone, 85-95 cm Table 2 Distribution of stone artifacts in Trench 1, Northcliffe quarry- factory site. Western Australia Silcrete artifacts Artifacts of other stone Radiocarbon Depth Geometric Notched Blades Blade- Flakes Chips* Cores Debris dates in cm microliths ^ flakes lets below and datum other re- touched tools AXU-1131 3080 ± 75 46-51 2 51-53 2 8 6 + 53-57 1 atypical 1 3 39 47 s- 57-59 2 atypical 2 2 48 57 -r 1 quartz chip 59-60 2 6 57 110-S 60-62 3 2 2 76 SOS- 62-64 2 1 1 54 50 -r 2 quartz chips 1 quartz flake 64-66 2 6 5 81 50 S- 3 1 quartz bladelet 66-70 6.2 atypical 2 3 11 149 250 S- 6 quartz flakes and fragments- 70-72 4,1 atypical 2 4 2 106 250 2 1 pebble 72-74 2 4 5 1 97 250 S- 5 3 quartz fragments 74-78 1 atypical 3 1 11 112 250 r 4 1 2 quartz fragments 1 fragment of gneiss 78-82 3 1 38 100 s- 1 82-85 1 1 14 20 S- 8 quartz chips SUA 379 85-90 2 20 + 1 1 1 quartz fragment 6780 ± 120 90-96 1 5 2 + 1 1 (|uartz flake 96-98 5 9 }- 1 quartz core 1 quartz flake 98-103 7 1 + 1 (juartz bladelet 2 quartz flakes 2 quartz fragments 103-c,120 5 4 + 2 1 quartz fragment 2 quartz flakes 2 chert flakes 18 typical 25 22 46 903 1526-{- 12 11 40 7 atypical 1 atypical indicates irregular or partly retouched specimens. ^ -i- indicates that total chips were not recovered. Journal of the Royal Society of Western Australia, Vol. 58, Part 2. July, 1975. 60 DATUM 0 depfhs in cm ]Q 20 30 40 50 60 70 80 90 100 no 120 ANU 1131 3080 - 75 BP uppermost geometric microlith [cv^errr.cst gecmetric microlith lUA :79 ^7CC i I20EP Figure 2. — North section of Trench 1, Northcliffe quarry- factory. Trench 1 contained no vertebrate remains or mollusc shells. Charcoal occurred in varying quantities throughout and we collected several samples for radiocarbon assay. One of these samples, KS 2, taken from a dept of 85 to 95 cm in the lower part of the leached zone (Fig. 2), was submitted for dating and yielded an age of 6780 ± 120 years BP (SUA 379: R. Gil- lespie, pers. comm.) (Table 1). Stone artifacts The largest quantities of stone artifacts were in the leached zone between 65 and 80 cm (Table 2). The assemblages from arbitrary levels at this depth comprise retouched tools, large irregular flakes or fragments, numerous small flakes and very many tiny chips. These last are flakes which have a maximum dimen- sion of less than one cm; many of these (measuring < .5 cm) were not retained. There are also many blades and bladelets which, fol- lowing Tixier (1963, p. 38), can be distinguished on the basis of width, bladelets having a maxi- mum width of 1.2cm (e.g. Fig. 3d). No blade cores were recovered from Trench 1 though these have been collected from other sites in the district. Some of the blades and bladelets are snapped short, perhaps deliberately. None is retouched. All of the 13 cores from Trench 1 were used in flake production. One of these, from 74 to 78 cm depth, is a typical discoidal core. There are also a number of short (2-4 cm long) flakes with broad, thick faceted butts and oval or triangular forms which were probably produced on discoidal cores. The single quartz core (depth 96-98 cm) is of bipolar or scalar type (White 1968). Half of the retouched tools are geometric microliths of trapeze (Fig. 3a), crescentic (Pig. 3b) or triangular form (Fig. 3c). We recovered 18 typical geometric microliths from Trench 1. Almost all of the other retouched tools are flakes with one or two notches, the exceptions being retouched flakes. The very high propor- tion of dehitage ranging from flakes or frag- ments to hundreds of tiny chips probably result- ing from the manufacture of finely retouched tools such as geometric microliths, as well as the retouched tools themselves, indicate that the site is not only a quarry but also a factory where geometric microliths were one of the main products. Only two flakes, both made of silcrete, oc- curred above the position of ANU 1131. The uppermost geometric microlith came from a depth of 53 to 57 cm or about three to seven cm below the sample position of ANU 1131 (Fig. 2, Tables 1, 2). The lowermost geometric micro- lith (this piece is illustrated in Fig. 3b) was recovered one to three cm above the upper limit of the level from which we collected char- coal for SUA 379 (Pig. 2, Tables 1, 2). These radiocarbon dates thus show that geometric microliths were being manufactured at the fac- tory from about six thousand to about three thousand years ago. Relatively few artifacts were found below the sample position of SUA 379. However the sil- crete artifacts from this depth and below are similar (i.e. quarried fragments, tiny chips and flakes of small size, some with small butts and parallel dorsal flake scar ridges) to the silcrete dehitage in the assemblages containing geometric microliths higher up in the sequence, thus sug- Journal of the Royal Society of Western Australia. Vol. 58. Part 2. July, 1975. 61 Figure 3. — Geometric microliths and a bladelet excavated from Trench 1. Northcliffe quarry-factory. gesting that essentially the same industrial activity took place. It is possible then that a larger artifact sample from the lower part of the deposit would contain geometric microliths. Most of the artifacts from Trench 1 and all of the retouched tools are made of silcrete, the few exceptions being quartz pieces (e.g. B2418), a fragment of gneissic rock (B2412) and two flakes (B2418, B2420) made of a distinctive form of fossiliferous chert (J. E. Glover, vers. comm.). This chert is similar to that used for artifacts from many sites in the Perth area (Pig. 1: Glover and Cockbain 1971; Hallam 1972) and in the late Pleistocene layers of Devil’s Lair (Fig. 1: Dortch 1974; Dortch and Merrilees 1973: Glover 1974). One of the two chert flakes comes from a deep depression in the surface of the basal cemented sands (Pig. 2) and the other comes from dark sand just above the basal sands. These flakes may be considerably older than the artifacts in the leached zone, and they may have dropped down to the hardpan during the deflation of an earlier soil. Hallam (1972, pp. 14-15) has postulated that fossiliferous chert of this kind predominated in the early assemblages from sites on the coastal plain to the west. However on the south coast there is evidence to suggest that here fossiliferous chert continued to be used in quantity during the later prehistoric period. Gardner has collected numerous fossiliferous chert artifacts from sur- face sites in the Northcliffe area which also contain quantities of geometric microliths and bladelets of silcrete and quartz typical of later Australian stone industries. He has also located an outcrop of fossiliferous chert about ten km east of Northcliffe which would have been accessible during the later prehistoric period. It seems likely then that fossiliferous chert was used until relatively recent times in the North- cliffe area. Pollen Analysis Pollen samples were taken at different depths in Trench 1. These were submitted to Dr. B. E. Balme of the Geology Department, University of Western Australia for analysis. His report (B. E. Balme, pers. comm.) stated that a sample from a depth of 47 cm contained pollen grains, the dominant species being Eucalyptus caZo- phylla and E. diversicolor though some grains “of E. marginata-type were also fairly common”. A sample from 68 cm contained a similar plant microfossil assemblage to that of the previous sample. The next sample from 87 cm contained no pollen. The lowermost sample from the dark sand resting on the basal cemented sands (depth 121cm) contained pollen grains of the three above Eucalyptus species with E. diversi- color appearing to be “relatively more abun- dant” than in higher samples. These three Eucalyptus species are at present dominant in the Northcliffe area. The pollen analysis suggests then that climate in this locality was at various times during the Recent period much the same as it is at present. In his recent study Churchill (1968, p. 146) con- cludes that “the climate [of the extreme south west] from 4000 to 3000 B.C. was favourable for E. diversicolor” . The oldest Northcliffe pollen sample above suggests that conditions were favourable for E. diversicolor even earlier during the Recent period. Western Australian Museum catalogue numbers of silcrete artifacts illustrated in Fig. 3 a. B2410 b. B2414 c. B2413 d. B2410 Stone artifacts excavated from Trench 1 at the North- cliffe silcrete quarry-factory are listed in the Western Australian Museum register under catalogue numbers B2401-E2421. All catalogue numbers mentioned in the text pertain to the Western Australian Museum register. Discussion Hallam (1972, pp. 16-17) has published four self consistent radiocarbon dates ranging from 3090 ± 240 years BP (ANU 830) to 110 ± 70 years BP (ANU 827) which are associated with tool assemblages containing geometric micro- liths excavated at Frieze Cave near York, W.A. (Fig. 1) some 300 km north of Northcliffe. The Frieze Cave dates and those from the Northcliffe quarry-factory (Table 1) indicate that the use of geometric microliths in south western Aus- tralia persisted from about 6 000 years ago until the Modern period, this being the longest dura- tion for microlithic industries yet recorded for Australia (see Pearce 1974, Table 2). The Northcliffe dates do not support Pearce's recent hypothesis (1974, p. 307), based on pre- viously available radiocarbon dates (e.g. ANU 830), that “the introduction of backed blades [geometric microliths] was earliest in New South Wales and progressively later away from that area”. The earlier date from Northcliffe shows that geometric microliths are not a relatively late innovation in the south west, and that it is possible that they occur here earlier than in south eastern Australia. Until more data be- Journal of the Royal Society of Western Australia. Vol. 58, Part 2, July, 1975. 62 come available I see no reason to assume that the beginning date of the microlithic industries in Western Australia differs greatly from that of the microlithic industries in eastern regions. The dates reported here from Northcliffe, those from Frieze Cave (Hallam 1972), and the sequence of radiocarbon dates from Devil’s Lair (Dortch 1974; Dortch and Merrilees 1973) pro- vide a radiocarbon dating sequence for south western prehistory which extends from about 25,000 years BP to Modern times. A relatively recent date SUA 364: 6490 ± 145 years BP (R. Gillespie, vers, comm.), has been obtained from a charcoal sample collected during the 1973 excavation of Devil’s Lair, Trench 7 (Dortch 1974). The sample on which this date is based comes from layer G (Dortch 1974, Fig. 3), a sandy deposit containing a quartz flake (B1846), some mussel shell (e.g. B1846) and other faunal remains, and some possible bone artifacts (e.g. B1847). No stone tool forms of great diagnostic value are associated with this date from Devil’s Lair or with the older date (SUA 379) from Northcliffe. Nevertheless it is possible that SUA 364 and SUA 379 relate respectively to an early phase assemblage left by some of the last occupants of Devil’s Lair, and an initial late phase assemblage marking an early period of quarrying and tool manufacture at Northcliffe. If this is true, then the early-late phase transi- tion in south western Australia took place over 6000 years ago. Further investigations at these two sites should provide more data relevant to the age and significance of this transition. Acknowledgements. — I wish to express my sincere thanks to those cited as having given personal com- munications. I acknowledge with thanks the help of Miss Jane Balme who helped prepare text figures, and Dr D Merrilees, Dr. J. P. White and Mr. R. J. Lamport who read the MS. I thank Miss A. McConnell and Mr. W M. McArthur for their help in the field, and I wish to thank Mr. G. Gardner for his co-operation and hospitality. Last I wish to express my gratitude to the Australian Institute of Aboriginal Studies for sponsoring the radiocarbon dates referred to here. References Churchill, D. (1968).— The distribution and prehistory of Eucalyptus diversicolor marginata . . .. and E. calophylla ... in relation to rainfall. The Australian Journal of Botany 16; 125-151. Dcrtch, C. E. (1974).— A twelve thousand year old occu- pation floor in Devil’s Lair, Western Aus- tralia. Mankind 9: 195-205. Dortch. C. E.. and Merrilees, D. (1973).— Human occupa- tion of Devil’s Lair, Western Australia during the Pleistocene. Archaeology and Physical Anthropology in Oceania 8; 89-115. Glover, J. E. (1974). — Petrology of chert artifacts from Devil’s Lair. Western Australia. Journal of the Royal Society of Western Australia 57 : 51-53. Glover, J. E., and Cockbain, A. E. (1971). — Transported Aboriginal artefact material, Perth Basin. Western Australia. 'Nature, Lond. 234: 545- 546. Hallam, S. J. (1972). — An archaeological survey of the Perth area. Western Australia: a progress report on art and artefacts, dates and demo- graphy. Australian Institute of Aboriginal Studies Newsletter 3(5): 11-19. Mulvaney, D. J. (1969).— “T/ie Prehistory of Australia.” Thames and Hudson, London. Pearce. R. H. (1974).— Spatial and temporal distribution of Australian backed blades. Mankind 9: 300-309. Tixier, J. ( 1963) .— Typologie de I’epipaleolithique du Maghreb. Memoire II, Centre de Recherches Anthropologiques. Prehistoriques et Ethno- graphiques, Algiers. White, J. P. (1968). — Fabricators, outils ecailles or scalar cores? Mankind 6: 658-666. Journal of the Royal Society of Western Australia, Vol. 58, Part 2, July, 1975. 63 IV ‘>7*4 ilP ®5>; i INSTRUCTIONS TO AUTHORS Contributions to this Journal should be sent to The Honorary Editor, Royal Society of Western Australia, Western Australian Museum, Perth. Papers are received only froiu or by communication through, Members of the Society. The Council decides whether any contribution will be accepted for publication. All papers accepted must be read either in full or in abstract or be tabled at an ordinary meeting before publication. Papers should be accompanied by a table of contents, on a separate sheet, showing clearly the status of all headings; this will not necessarily be published. Authors should maintain a proper balance between length and substance, and papers longer than 10,000 words would need to be of exceptional importance to be considered for publication. The Abstract (which will probably be read more than any other part of the paper) should not be an expanded title, but should include the main substance of the paper in a con- densed form. Typescripts should be double-spaced on opaque white foolscap paper; the original and one carbon copy should be sent. All Tables, and captions for Figures, should be on separate sheets. Authors are advised to use recent issues of the Journal as a guide to the general format of their papers, includ- ing the preparation of references; journal titles in references may be given in full or may follow any widely used conventional system of abbreviation. Note that all illustrations are Figures, which are numbered in a single sequence. In composite Figures, made up of several photographs or diagrams, each of these should be designated by letter (e.g. Figure 13B). Illustration 5 should include all necessary lettering, and must be suitable for direct photo- graphic reproduction. To avoid unnecessary handling of the original illus- trations, which are usually best prepared between and 2 times the required size, authors are advised to supply extra prints already reduced. Additional printing costs, such as those for folding maps or colour blocks, will normally be charged to authors. It is the responsibility of authors to adhere to the International Rules of Botanical and Zoological Nomenclature. Palaeontological papers must follow the appropriate rules for zoology or botany, and all new stratigraphic names must have been previously approved by the Stratigraphic Nomenclature Com- mittee of the Geological Society of Australia. Thirty reprints are supplied to authors free of charge, up to a maximum of 60 for any one paper. Further reprints may be ordered at cost, provided that orders are submitted with the returned galley proofs. Authors are solely responsible for the accuracy of all information in their papers, and for any opinion they express. Journal Volume 58 of the Royal Society of Western Australia 1975 Part 2 Contents 3. Botany in Western Australia: A survey of progress: 1900-1971. Presidential Address, 1971. By B. J. Grieve. 4. New finds of sand fulgurites from the Perth Basin, Western Australia. By J. E. Glover. 5. Geometric microliths from a dated archaeological deposit near Northclifle, Western Australia. By. C. E. Dortch. Editor: A. J. McComb The Royal Society of Western Australia, Western Australian Museum, Perth 46392/4/75—625 WILLIAM C. BROWN, Government Printer, Western Australia JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA VOLUME 58 PART 3 NOVEMBER, 1975 PRICE TWO DOLLARS REGISTERED FOR POSTING AS A PERIODICAL-CATEGORY B THE President Vice Presidents Past President Joint Hon. Secretaries Hon. Treasurer Hon. Librarian Hon. Editor ROYAL SOCIETY OF WESTERN AUSTRALIA PATRON Her Majesty the Queen COUNCIL 1975-1976 B. E. Balme, D.Sc. P. R. Wycherley, O.B.E., B.Sc., Ph.D., F.L.S. A. J. McComta, M.Sc., Ph.D. ... G. A. Bottomley, B.Sc., Ph.D. - . M. Perry, B.Sc. (Agric.) (Hons.) G. Perry, B.Sc (Hons.) S. J. Curry, M.A. .... A. Neumann, B.A. .... A. J. McComta, M.Sc., Ph.D. P. Atkinson, B.Sc. C. E. Dortch, B.S., M.Phil. B. B. Lamont, B.Sc. (Hons.), Ph.D, L. J. Peet, B.Sc., F.G.S. P. E. Playford, B.Sc., Ph.D. J. C. Taylor, B.Sc., Ph.D., A.R.C.S. A. F. Trendall, B.Sc., Ph.D., A.R.C.S., F.G.S. P. G. Quilty, B.Sc. (Hons.), Ph.D. 6. — Nitrogen oxide levels in suburbs of Perth, Western Australia by G. A. Bottomley^ and F. C. CattelP Manuscript received 22 October 1974; accepted 19 November 1974 Abstract Nitrogen oxide determinations are reported for two residential suburbs southwest of the central city area of Perth, Western Australia. Daily measurements for the period 10 a.m. to noon during 1971, 1972 and 1973 show that the levels of nitrogen oxides in the atmosphere are low in comparison with overseas major urban areas. The day to day values fluctuate considerably but periodogram analysis detects a seven day cycle which reflects local traffic flows with highest values on Fridays and the lowest on Sundays. The thirty day moving average shows a seasonal pattern where average values are low from about August to April and rise to a pronounced peak in May and June (late autumn). Data is given for autumn evenings when local meteorological conditions often result in shallow inversions about sun- set. These inversions are coincident with peak traffic flow and concentrations of as high as 47 pphm have been detected on such even- ings. Some limited monitoring of photo- chemical conversion of NO to NO 2 and produc- tion of ozone has been undertaken. Introduction The residents of Perth, Western Australia, ordinarily enjoy atmospheric quality and clarity which are remarkably good for a city with more than three-quarters of a million inhabi- tants, Partial explanations are adequate mixing and good ventilation (the mean wind speed is 15.6 km/hour) and the location and control of industrial and other pollutant sources. In the City and most residential suburbs the levels of smoke, sulphur dioxide and particulates are modest. However, motor vehicle use has risen sharply in the last decade and the trend con- tinues. Because of the intense and prolonged sunshine (average 2 850 hours annually) pollu- tion from car exhausts and photochemical ef- fects are of public interest. We report here concentrations of oxides of nitrogen at three suburban sites recorded chiefly in 1971, 1972 and the first half of 1973. Relevant geographical information General The City proper and much of the Metropoli- tan residential areas are on the Swan Coastal Plain at elevations less than 70 metres above sea level. Fifteen kilometres to the East the Darling Scarp rises sharply to above 300 metres. Most traffic generated by the business and resi- dential areas is within a radius of fifteen kilo- metres of the City centre. 1 Department of Physical and Inorganic Chemistry, University of Western Australia, Nedlands, West- ern Australia, 6009. 2 Division of Cloud Physics, C.S.I.R.O., Epping, N.S.W. 2121. Recording sites Air samples were taken at about two metres above ground level at three locations (see Fig- ure 1). Figure 1. — Map showing University site (U), Nedlands Site (N) and Peppermint Grove Site (PG) in relation to Perth. University site. — The University of Western Australia campus five kilometres South West of central Perth business area. Most of its forty hectare area is grassed but it includes car parks for about two thousand vehicles. To the North East is four hundred hectares of Kings Park with bush, botanic gardens and open space and to the East and South East there is a broad stretch of the Swan Estuary, hereabouts two or three kilometres wide. On the West the Univer- sity is flanked by residential areas. Nedlands site. — A private residential area about three hundred metres to the South of the University sampling site. Peppermint Grove site. — A private residence ten kilometres South West of the city centre, within one and a half kilometres of the Indian Ocean and close to the open reaches of the lower Swan. The area includes much open space and is not near any recognisable industrial sources though domestic gas and oil appliances are in normal use for water and space heating, as at the other two sites. Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 65 Analytical procedure Table 2 Saltzman’s method (Katz 1968, p. 80) with permanganate oxidation of the nitric oxide was used to determine total nitrogen oxides NOx. Air was usually sampled at about twenty litres per hour and for two hours. The absorbing reagent, N-(l-naphthyl)ethylene-diamine dihy- drochloride with sulphanilic acid in aqueous acetic acid solution was kept in very subdued light during prior storage and until the spectro- photometric determination was completed. Di- rect sunshine degrades the red azo compound. The values are cited as parts per hundred million by volume pphm) of the combined oxides NO plus NO 2 unless a distinction is made in the context. The conversion factor is: 18.8 pphm = 1 microgram metre’^ (25°C) Measurements at the University of Western Australia Two hundred determinations on outside air at the University Site were taken sporadically between June 1970 and June 1971 at various times of the day and night. Table 1 summarises these exploratory results in histogram form. Table 1 General results in preliminary survey at University Site 1970-1971 XOx at Vniversity Site (mean value over period) j>phm 0-0 to 0-99 1 • 0 to 1 • 99 2 • 0 to 2 • 99 3 • 0 to 3 • 99 4 0 to 4-99 5-0 to .5-99 () 0 to 6 -99 7 ■ 0 to 7 • 99 8- 0 to 8-99 9- 0 to 9-99 Above 10-00 (10 -4, 23-6) Number of Observations June 1970 — June 1971 27 63 39 11 16 9 6 4 1 3 181 The low values are evidence for the ability of Perth air to disperse contaminants. Particu- larly in South Westerly air streams in winter and during the regular afternoon sea-breeze (the Fremantle doctor) in summer, values as low as 0.2 pphm have been recorded. However, there are occasional periods of calm air and poor mixing conditions which correlate with higher levels of NOx. Table 2 sets out the highest values recorded June 1970 to June 1971 together with comments on atmospheric condi- tions. We consider it most unlikely that these occasional measurements would have detected the highest levels experienced that year. News- paper reports of the highest value recorded, 23.6 pphm, 8 a.m. to 10 a.m. April 29. 1971, provoked external criticism that emissions from a small oil-fired boiler in the University grounds were responsible, a view the authors do not en- dorse. To eliminate this possible bias sampling was transferred to the Peppermint Grove re:i- dential site. Highest values in preliminary survey at University site 1970-1971 Diue Approx. Time Value Comment NOy pphm. 1970 June 18 4- 6 p.m. 5-4 June 23 8- 9 a.m. 6-7 9-10 a.m. 7-1 July 17... 9-10 a.m. 9-5 July 24. .. 4-8 p.m. 8-2 July 25, ... 4- 8 p.m. 5-4 Aug. 3 .... 6- 8 p.m. 6-0 Aug. 4 .... 9-11 a.m. 9-1 Blue light-scattering material evident be- tween University buildings. 4- 6 p.m. 5-2 6- 9 p.m. 16-4 Sept. 25 8-10 a.m. 5.9 Sept. 29 7- 9 a.m. 7.6 Sept. 30 1971 7-10 a.m. 6-0 April 1 8-10 a.m. 6.0 April 2 4- 6 p.m. 5-2 April 20 8-10 a.m. 6-8 Anticyclone continues for w'eek, Herdsman Lake vegetation on fire two days pre- vious April 21 8-10 a.m. 7-2 April 28 8-10 a.m. 5-8 April 29 8-10 a.m. 23-6 City not visible from 10-Noon 7-8 I’niversity (three miles) at 10 a.m. be- cause of white-out ’ Clears partiallv by 11.00 a.m. April 30 8-10 a.m. 9-0 ‘ Blue haze ’ again .May 1 .... 8-10 a.m. 6-5 May 6 .... 8-10 a.m. 5-8 ‘ Grey-out ’. City just visible at 10.00 a.m. from University. May 7 .... 8-10 a.m. 5-0 5- 7 p.m. 5-8 Peppermint Grove morning series June 1971 to July 1972 Measurements were made 10 a.m. to noon each day for more than a year and the results are set out in Table 3. The daily values fluc- tuate considerably with the weather circum- stances and include many very low values. The maximum values 15.0 and 12.7 pphm recorded respectively on May 12 and June 8 1972, bear- ing in mind the greater distance of the Pepper- mint Grove site from the City, approach com- parability with the University maximum on April 29, 1971. The monthly averages and the maximum values each month show progressive variation in levels of NOx with a pronounced maximum in autumn and early winter months. Figure 2 shows the ‘thirty day moving' average as a function of the mid-date of the averaged sequence. Our hypothesis is that NOx is generated by emissions from vehicular traffic in the morning rush hour and that the recorded levels reflect the inability of the air in each day to disperse that chemical burden. Information supplied by the Bureau of Meteorology for the mean monthly 7 a.m. mixing depth (1966-1969 in- clusive) shows values which range from one hundred and thirty metres to two hundred and ten metres, with the lower values in May- August. The routine radiosonde ascents may Journal of the Royal Society of Western Australia, Vol. 58. Part 3. November, 1975. 66 Table 3 Results of year long survey of morning values at Peppermint Grove 1971 and 1972 pphm . NO 4 - NOa = NO ^ 10 a . m . to noon daily ( 140 a Forrest Street , Peppermint Grove ) DAY June 1971 July Aug . 8 ept . Oct . Nov . Dec . 1971 Jan . 1972 Feb . Mar . Apr . May June Julv 1972 1 6-9 0-5 1-6 0-8 1-2 0-4 0-6 0-2 0-6 0-6 3-4 6-3 1 -6 2 .... 2-5 1-5 0-9 0-9 0-6 0-3 0-6 0-2 0-5 0-6 3-3 1-0 0-7 3 .... 1-7 2-1 1-0 1-1 0-6 0-2 0-6 0-2 0-5 0-6 2-6 1-9 1-1 4 .... 1-0 0-8 11 0-8 0-9 0-2 0-6 0-7 0-5 0-4 1-7 1-0 2-2 5 .... 1-5 3-9 0-8 0-8 0-2 0-6 0-7 0-6 1-1 1-0 2-8 4-6 6 .... 3-3 10 11 0-9 0-2 0-6 0-7 0-7 0-7 11 2-0 0-8 7 .... .... 1-5 4-7 3-5 1-0 0-8 0-3 0-5 0-6 0-7 0-7 1-5 5-0 1-7 8 .... .... 1-0 2-5 1-2 1-8 0-9 0-4 0-5 1-0 0-7 0-9 3-3 12-7 0-8 9 .... .... 1-5 1-7 1-4 1-1 0-8 0-6 0-4 0-8 1-3 0-2 5-0 11-0 0-9 10 .... .... 0-9 7-3 1-3 0-9 0-6 0-5 0-5 0-7 0-8 0-8 1-3 2-6 1-1 11 .... .... 3-8 40 1-5 1-1 1-0 0-7 0-5 0-6 0-7 0-8 0-9 2-0 1-3 2-0 12 .... 2-7 1-3 0-9 11 1-0 0-6 0-7 0-5 0-8 0-2 1-3 15-0 0-8 1-4 13 .. .... 1-5 2-2 0-8 1-1 0-8 0-6 0-5 0-5 0-9 0-2 2-0 1 0 0-4 0-9 14 .... .... 1-1 1-7 1-7 2-0 0-9 0-7 0-7 0-4 1-3 0-2 0-9 0-8 0-4 0-9 15 .. .... 1-6 9. ■ 9. 0-6 1-7 10 0-4 1-0 0-6 0-8 0-2 2-3 1-1 0-7 4-5 16 .... .... 1-1 2-3 0-8 1-9 1-0 0-4 1-1 0-7 0-9 0-5 1-8 3-5 0-4 2-1 17 .... .... 3-8 2-2 0-8 1-4 0-6 0-5 0-7 0-7 0-6 0-3 1-4 4-0 0-5 3-0 18 .... .... 7-2 2-1 1-6 1-4 0-8 0-4 1-0 0-7 0-8 0-3 1-0 1-9 20 1-7 19 .... .... 1-6 2-2 2-0 1-0 1-0 0-4 0-9 0-7 11 0-2 0-9 1 -8 2-0 5-2 20 .... .... 1-2 0-9 1-8 1-1 0-7 0-5 1-0 0-9 11 0-3 1-0 1-2 2-0 4-5 21 .... .... 6-5 0-6 0-9 0-9 0-9 0-9 0-9 0-7 2-5 2-0 2-3 3-0 22 .... .... 2-5 0-8 0-7 0-7 0-5 11 0-8 1-5 0-4 1-7 1-7 0-8 2-0 23 .... .... 6-8 1-0 0-8 0-9 0-5 0-8 1-2 1-3 0-6 1-0 1-9 25 1-6 24 .... .... 3-8 1-4 0-8 0-8 0-6 11 0-6 11 0-4 1-8 2-0 1-8 0-4 25 .... .... 6-5 2-0 1-3 0-9 0-6 0-8 0-6 10 0-5 11 1-7 10 0-4 26 .... .... 6-8 2-2 0-8 0-8 0-8 0-6 0-9 0-3 1-3 1-1 0-8 27 .... .... 4-8 1-9 11 1-1 10 11 0-5 0-9 0-8 0-4 4-8 0-9 1-2 28 .... .... 6-8 1-4 0-7 0-9 0-9 0-5 0-4 0-7 0-5 0-6 7-5 1-7 0-8 29 .... .... 2-1 1-0 0-5 11 0-9 0-6 0-4 0-4 0-6 0-5 3-4 1-0 4-0 30 .... .... 2-7 11 0-9 0-8 0-9 0-4 0-6 0-2 0-5 1-6 2-0 10-3 31 .... 11 0-6 1-0 0-6 0-2 0-5 0-8 Av .... 3-33 2-22 1-34 1-14 0-87 0-62 0-66 0-61 0-81 0-51 1-56 2-36 2-74 1-96 Max . .... 7-2 7-3 3-9 2-0 11 1-2 1-1 1-2 1 • 5 1-3 7-5 15-0 12-7 5-2 Min .... 0-9 0-6 0-5 0-7 0-6 0-4 0-2 0-2 0-2 0-2 0-2 0-8 0-4 0-4 Figure 2.— Morning values of NO^ recorded at Peppermint Grove (dotted line) and at Nedlands (full line) expressed as a thirty day moving average. Journal of the Royal Society of Western Australia, vol. 58, Part 3, November, 1975. 67 not give particularly useful information about the lapse rate below the first hundred metres which is presumably the critical zone for dis- persion of traffic exhausts. Taken in the broad sense the dispersion of the air is rather generally governed by the energy flux from the sun which is very high here in summer. In the winter months the sun is at a relatively low angle and operates for less time before the start of the sampling period. The monthly averages appear to be governed sub- stantially by this main driving force. Daily features of the synoptic weather pattern such as rain storm progressions and anticyclones are obscured by the thirty day averaging procedure but special features will be taken up later. In general high nitrogen oxide levels corres- pond to days of reduced visibility, with sugges- tions of association with a special blue light- scattering quality. This does not require that the source of the nitrogen oxides is also re- sponsible for the impaired visibility but indicates that at these times weather conditions are unfavourable for the dispersion of all pollutants. On days of poor visibilty any photochemical destruction of the nitrogen dioxide would be slower. The entries in Table 3 contain latent infor- mation which supports the car emission origin. It would be reasonable to suppose traffic flow to be substantially less on Sunday mornings in comparison with mid-week mornings. Emission release may therefore be cyclic, of period seven days, and of minimal amplitude on Sundays. The values in Table 3 were subjected by Schus- ter’s method (Whittaker 1944) to a computer- assisted search for periodicities in the range two to thirty-two days in intervals 1/4 days. Part of the periodogram is shown in Figure 3. There is considerable evidence for a seven day component. For seven day periodicity the data can be succinctly expressed as the sum of the values recorded each Monday, each Tuesday, etc., through the week. The totals for fifty- five weeks are: Monday Tuesday Wednesday Thursday Friday Saturday Sunday 70-6 (57 - 3 72-7 90-2 98-3 7 5-2 58 ft a pattern which qualitatively reflects the traffic behaviour in the Metropolitan area. Periodograms for the period March 1, 1972 to June 30, 1972 have also been run on data for twenty-four hours smoke values and sulphur dioxide values in the City. The patterns are not similar to that for the NOx data so it is reasonable to conclude that the sources and dispersion characteristics of NOx, SO 2 and smoke are dissimilar. In particular the seven day frequency component for the SO 2 measure- ments shows a mid-week rather than a Friday maximum. Nedlands series Mornings April 1972 to December 1972 Table 4 shows in summary form measure- ments 10 a.m. to noon at the Nedlands site from March 28, 1972 to June 14, 1973. Figure 2 summarises the trend of these measurements in the form of “thirty day moving averages” and displays the two series together. An interest- ing point is that the thirty day averages for the two sites five kilometres apart for the few weeks when simultaneous measurements were underway are similar in magnitude and in form. A maximum mean concentration of 3 pphm in late autumn and early winter falls from the middle of July to about 1 pphm in September at both sites. In 1971 at Peppermint Grove there was a further decrease to an average Table 4 Results of year-long-survey of morning values at Nedlands pphm. NOx 19 «•?«• to noon daih/ 1972 Mar.* Apr. May June July Aug. Sept. Oct. Nov. Dec. Average ilaxinium Minimum 1-51 1-33 0-7 1-70 8-9 0-4 2-67 191 0-4 2-95 13-2 0-7 2-23 6-4 O-ft 1-38 3-0 0-7 1-18 3-2 0-4 115 2-1 0-7 1-11 40 0-7 0- 97 1- 9 0-5 1973 •Tan. Feb. Mar. Apr. May June* Average Maximum Minimum . .. 0-9ft 2-2 0-2 0- 99 1- 5 0-4 0-98 1 • 5 0-ft 1 -08 2-6 0-3 1-88 5-3 0-7 1-70 3-1 10 * lncomi)lete mouth. Journal of the Royal Society of Western Australia. Vol. 58, Part 3, November. 1975 68 concentration of about 0.7 pphm from November to March but at Nedlands in 1972 the average remained constant at 1 pphm. The relative contributions of the different sites and the different years to this difference cannot be estimated. Periodogram analysis shows a similar fre- quency pattern to that of the Peppermint Grove data and again the totals over sixty three weeks for the weekdays are indicative of traffic origin: Monday Tuesday Wednesday Thursday Friday Saturday Sunday 86-5 85-3 98-e 110-2 122-3 92-1 70-9 Figure 4 shows the daily inter-site comparisons of the 10 a.m. to noon values for a sample period in April and May 1972. The very close comparison between the rise and fall of the values above 2 pphm supports the view that both the long term average and the major observed effects are very probably typical of substantial parts of the Metropolitan area. The correlation between the measurements at both sites for all days of the month can be estimated from the Spearman’s rank correlation coefficient (Fisher 1958). If the days of the month are ranked according to nitrogen oxide values: 6 2d^ r 1.0 n^ - n where i:d^ is the sum of the squares of the rank differences between the two sites and n is the number of days of the month. For April 1972, r - 0.577 and for May, r ^ 0.879, suggesting significant positive correlation. The probability, P, that such a correlation could arise by random sampling from uncorrelated data can be esti- mated from Student’s t test where which gives for April t 3.7 and for May t ^ ^ 9.9. For each month P is less than one in a thousand for a random occurrence. Again we conclude that the same controlling factors for nitrogen oxide concentrations seem to be operative at both sites, local uncorrelated sources can play only a minor part. Results from measurements outside the 10 a.m. to noon period The main sequence of measurements was deliberately taken so as to avoid the traffic peak emissions but some effort has also been made to measure the effects of these. The principal meteorological factor leading to high concentrations of car emissions is the presence of a very shallow mixing layer in the air of the Metropolitan area. Particularly on late autumn afternoons with little or no wind, as sunset approaches inversion conditions set in as the layer of air close to the ground cools by radiation to temperatures substantially below that of air at a few tens or hundreds of metres. The onset of these conditions is often apparent to casual observers in that light smoke from domestic or garden fires is retained at shallow depth with otherwise very clear air above it. The same does not apply to emission from tall chimney stacks where the plumes carry well above this stagnation layer. 15 ( • at 19) Figure 4.— Intercomparison of simultaneous morning values at Peppermint Grove {full lines) and at Nedlands (dots). Journal of the Royal Society of Western Australia, vol. 58, Part 3, November, 1975. 69 Measurements of NOx levels were often pur- posefully taken during such actual or suspected (they are not visually detectable after dark) conditions. Observations outside the regular 10 a.m. to noon period were deliberately biased to times and days of anticipated higher levels and they do not represent an average overview of the early morning or evening conditions. Table 5 lists pertinent but extreme data col- lected at the Nedlands site. Values taken at the same time in Peppermint Grove are broadly similar though the peaks tend to be somewhat lower and later consistent with a Westerly movement of contaminated air from the City centre across Nedlands to the coast at two to three kilometres per hour. Table 5 Notably high measurements recorded in Nedlands 1972 Saturdai/, loth April Above 20 pphm . .. Above 25 pphm .... Peak 27-8 pphm. 1900 to midnight 2000 to midnight . .. 5 hours 4 hours Sunday, 16th April Above 20 pphm Peak 23-8 pphm. 2100 to midnight ... 3 hours Friday, 2S(h April Above 20 pphm Peak 32 • 5 pphm. 1900 to 2100 2 hours Saturday, 'llith April At 19-5 pphm .... 300 to 500 .... 2 hours Wednesday, Srd May Above 20 pphm Above 30 pphm Above 40 pphm Peak 47 pphm. 1800 to 2200 1900 to 2200 1900 to 2100 .... 4 hours 3 hours .... 2 hours Friday, Vlth May Above 20 pphm. .. Peak 30-4 pphm. 900 to 1100 .... 2 hours Tuesday, 2:ird May Above 20 pplim Peak 30-7 pphm. , . 1813 to 2000 .... 2 hours Thursday, 'Ihth May Above 18 pphm l>eak 31 pi)hm. 1200 to 2210 (Data possibly of lower 10 hours eontidence limits) Thursday,\st June Above 30 pphm. Peak 37 • 3 pphm. 1800 to 0100 7 hours Thursday, 29th June Above 20 pphm. . Peak 29 -4 pphm. 1900 to 2300 4 hours The 1972 mid April anticyclone stagnation spells and subsequent Thursday, April 13th, 1972 brought a distinct layer of smoke over the Metropolitan aiea, with the Bureau of Meteorology issuing an air dis- persion alert. Friday, April 14th, was similar with smoke haze extending many kilometres out to sea and persisting well into the afternoon: a dispersion alert was issued for Friday evening to Saturday noon. The smoke is said to have originated in burning-off operations on South West farms. This agricultural burning is com- mon at certain times in Western Australia. Similar controlle d burning of forest land occurs ~ Journal of the Royal Society of Western at selected times; a major discussion of such bush-fire smoke is available (Vines 1971). Satur- day and Sunday were much better optically. By midday Monday winds were affecting the City, clouds were gathering and after a calm period in the evening, wind sprang up again and rain fell early on Tuesday morning. The University wind records showed a forty hour calm starting 6 p.m. on Friday 14th. Alerted by some un- expectedly high values at Nedlands we obtained a considerable number of one hour and two hour duration measurements on April 15th, 16th and 17th. These are shown in Figure 5: similar information at other times in May and June 1972 appear in Table 5. The extreme peak of 47 pphm on Wednesday, May 3rd. 1972 followed the formation at sunset of a very shallow stagnation layer containing smoke from domestic fires under almost calm conditions. The plume from a ship leaving Fre- mantle Harbour persisted virtually intact for twenty minutes. Visual evidence that Perth’s atmosphere can at times be incapable of dealing quickly with at- mospheric contamination is given in Figure 6 which shows a view looking South from the University at 9.45 a.m. and 9.55 a.m on June 7th, 1972. Smoke from a fire at a distance of about eight kilometres has risen to about two hundred metres and has spread horizontally over a distance of about five kilometres. Dis- persion occurred quite suddenly two hours later. Several of the evening events were preceded by an abnormally hot day for the time of the year, but such a feature is not essential. Thurs- day, June 29th was a day with a maximum temperature of 13.4°C, several degrees below average, but again there was a radiation in- version early in the evening to give exceptionally low ground temperature. Friday, May 12th pro- duced quite high values in the morning, yet was associated with morning mist and light fog in Kings Park and slight rain clearing from the South West around 1 p.m. It may be informative to make extremely sim- plified estimates of the amount of nitrogen oxide released. Assume a cylinder of contaminated air five kilometres radius, one hundred metres deep and of average concentration 20 pphm. This contains a volume of 75 x 10^ metre^ or 3 000 X lO*^ moles of gas at one atmosphere pressure. At 20 pphm it will contain 60 000 moles of NO 2 , which is some 3 000 kg. The amount of NOx involved can also be approached from emission estimates. The “Freeway” Bro- chure produced by the Commissioner of Main Roads (1967) shows that Metropolitan Region traffic in an earlier survey amounted to 50 000 trips per hour over a five hour weekday period 3 p.m. to 8 p.m. This is 250 000 trips with an average of say four miles per trip each evening traffic rush period. Taking the Cali- fornian 1971 standard of 4 grams NO per mile gives 6 000 kg. NOx. To the car emission after revision for annual growth we should add of course any NOx from fuel oil and gas consump- tion but only that released at very low level. It then appears that the levels observed are achievable by car exhaust emissions within the Australia, Vol. 58. Part 3. November, 1975. 70 Figure 5. — Intercomparison of simultaneous values at various times for the special circumstances in mid April 1972. Full line Nedlands site, dotted line Peppermint Grove site. Metropolitan area without invoking other sources such as bush fires which do not from other evidence (Vines 1971) appear to contri- bute NOx though they may well produce smoke levels in Perth. NOx levels as indicators of photochemical and other pollutants The NOx values found as maxima in Nedlands compare at the one hour and eight hour level with those recorded as maxima for New Orleans (Zimmer 1965). This is disconcerting as it is most unlikely that the values measured at the particular sites correspond to absolute maxima in the Metropolitan area. The values justify commentary and further experimentation on other forms of car pollutants: carbon monoxide, photochemical oxidant and lead levels. The national primary and secondary ambient air quality standards (Federal Register 1971 ) for nitrogen dioxide prescribed by the Environ- mental Protection Agency of the United States of America is 5 pphm annual arithmetic aver- age. Our measurements even for the total oxides of nitrogen do not approach this figure and nitric oxide, for which there is no standard, pre- dominates over nitrogen dioxide (See Figure 7) at least in the evening air for which the con- centrations are greatest. Although the nitrogen oxide concentrations are not in themselves of serious concern it is possible from them to make inferences of the simultaneous concentrations of carbon mon- oxide. Overseas data for automotive emissions (Hum 1968) and experimental observations from the Continuous Air Monitoring Programs (Zimmer 1965) suggest that the carbon monoxide concentration is typically about forty times the total nitrogen oxide level. If this is true also for Perth, as seems likely, then on four of the seven nights for which concentrations of oxides of nitrogen greater than 20 pphm were measured, the U.S.A. national primary and secondary carbon monoxide air quality standards (Federal Register 1971), 9 pphm maximum eight hour concentration not to be exceeded more than once a year, were violated in some suburbs West of the central Perth area. The levels of 20 pphm NOx recorded several times are certainly sufficient to induce photo- chemical smog given sufficient hydrocarbons, except that the observed incidents occurred in the late evening. Should such levels persist into two or three hours of sunshine we must antici- pate photochemical effects with the charac- teristic lachrymatory effects and with visibility loss. The 8 a.m. to 10 a.m. April 29th, 1971 event was possibly just such a persistent wave. Less nitrogen dioxide than nitric oxide is emitted from combustion processes. In the absence of other contaminants the conversion of nitric oxide (at pphm concentrations) to Journal of the Royal Society of Western Australia. Vol. 58. Part 3, November, 1975. 71 Figure 6. — Smoke from fire South of Swan River on June 7th. 1972. Upper photo 9.45 a.m., lower photo 9.55 a.m. Nedlands site in foreground. nitrogen dioxide is exceedingly slow. However, in the presence of hydrocarbon pollutants and sunlight the conversion is accelerated and typically takes one to two hours. In cities where photochemical smogs occur the nitric oxide peak, formed as a result of the morning traffic, decays and is replaced by a nitrogen dioxide peak. Nitrogen dioxide absorbs light and is removed with the formation of photo- chemical oxidant (mainly ozone) which has a maximum concentration about noon. This sequence although typical of photochemical smog formation is modified by weather condi- tions. Photochemical oxidant concentrations were measured sporadically in 1972 by the neutral buffered potassium iodide method (Katz 1968, p 86). Measurements in Nedlands showed that nitric oxide, nitrogen dioxide and photochemical oxidant concentrations behaved similarly with time to atmospheres in which photochemical smog formation occurs (Tebbens 1968) except that pollutant concentrations are much smaller. For example, as shown in Figure 7, on the 7th June 1972 a morning nitric oxide peak of 5 pphm occurred between 8 a.m. and 9 a.m. a nitrogen dioxide peak of 2.5 pphm occurred between 11 a.m. and noon and an oxidant peak between noon and 1 p.m.. This suggested that photochemical activity was occurring and is consistent with occasional subjective observa- tions by experienced workers of a faint “Los Angeles” smell. Oxidant concentrations were measured routinely on weekdays from October 1972 until the end of December 1972. In October, November and December maximum oxidant concentrations often occurred between 10 a.m. and 11 a.m. in the morning after which time they decreased rapidly and often remained relatively constant throughout the afternoon. The decrease which may be associated with the break up of an in- version or with the appearance of a sea breeze supports the belief that the oxidant is formed at low altitudes. This pattern was not always followed and on 22nd November, 11th December and 16th Decem- ber 1972, concentrations of greater than 6 pphm, which is the unofficial WHO goal (Bilger 1972), were measured. On the 11th December, as shown in Figure 8 an average oxidant con- centration of greater than 8 pphm persisted for about two and a half hours between 1 p.m. and 3.30 p.m. This concentration is greater than that recommended by the U.S. primary and secondary air quality standards for photo- chemical oxidants and which is usually taken as the first indication of a mild photochemical Journal of the Royal Society of Western Australia, Vol. 58. Part 3. November. 1975. 72 Figure 7. — Time dependence of ozone, nitric oxide and nitrogen dioxide on June 7, 1972. Figure 8. — Dependence of oxidant concentration on hour of day, 11 December 1972. smog. This is the first unequivocal evidence for photochemical smog formation in Perth. During 1973 a chemiluminescent detector, based on the reaction of ozone with ethylene was constructed. This detector follows most of the recommendations of the United States Environ- mental Protection Agency (Federal Register 1971), is specific to ozone and has a very fast response. Parallel measurements on a few days with the neutral potassium iodide method and with the chemiluminescent detector suggest that our earlier measurements may have under- estimated the concentration of the photo- chemical oxidant by as much as 20%. Studies were made of the variation with time of the ozone concentration (measured by chemiluminescence) outside the Chemistry De- partment. On 28th August, 1973, for example, the ozone concentration rose from 2.2 pphm at 9 a.m. to a maximum of 6 pphm at noon, re- mained relatively constant to 3 p.m. and decreased to 4 pphm at 5 p.m. Occasional relatively high oxidant concentra- tions were measured on days on which bush fires were burning. However, insufficient mea- surements were made to test whether any posi- tive correlation exists. It is possible that natural organic vapours may participate in the photo- chemical reaction leading to oxidant formation in which case the abnormal release of large amounts of these in bush fires could be important. Topographic factors Since the highest NOx concentrations always seem associated with a shallow stagnant air layer which has accumulated pollution from vehicles and above which is clean air, the ob- served concentrations will depend very sharply on the precise location of the observing station. Accordingly, altitude changes of ten, twenty or fifty metres above ground levels or in buildings may be very significant especially when emis- sions occur over concave land surfaces. Tests should be instituted at such sites as: the Perth foreshore between the Narrows and the Cause- way, Dog Swamp — a classical ‘frost hollow’ traversed by Charles Street, the vicinity of Cot- tesloe railway-station in a valley traversed by Stirling Highway (the fringe of this valley is the location of the Peppermint Grove Sampling site), and at Lake Monger, Perry Lakes, and Herdsman Lake. Sampling sites representative of the whole Metropolitan area may be diffi- cult to select. It was completely beyond our resources to undertake multisite routine measurements, but a preliminary investigation was mounted as follows. About forty members of the University of Western Australia Clerical and Technical Staff Association co-operated in collecting simultan- eous “grab samples” of air at different sites in the Metropolitan area. These grab samples were analysed subsequently. Because of the difficulty in converting nitric oxide to nitrogen dioxide these measurements are not completely quanti- tative and probably relate to the nitrogen diox- Journal of the Royal Society of Western Australia, vol. 58. Part 3, November, 1975. 73 ide component alone. However, they do provide a relative measurement and have shown that at any one time there is a wide variation in concentrations throughout the Metropolitan area. In general low values were found on the outskirts and highest values along a NW to SE corridor through the city. Some localities (e.g. Dog Swamp) consistently recorded high values and in general the ranking of sites in terms of NOx concentrations did not vary much from day to day. For instance, when twenty-four suburban sites were ranked accordingly to nitrogen oxides concentration at 8 a.m. on 27th April and also at 8 a.m. on 4th May, 1972 the rank correlation coefficient was 0.739 corresponding to a value of t -= 5.1 and indicating a value for P of less than 0.1%. The correlation is apparently not as good when the concentrations of nitrogen oxides are lower. The correlation between twenty sites at 8 a.m. on 20th April and 27th April 1972 was described by a value of the rank corre- lation coefficient r of 0.424, a value of t of 2.0 suggesting a value of P of about 6%. Strong correlation between sites often existed even when the measurements were made at dif- ferent times of the day. For twenty-one sub- urban sites the rank correlation coefficient between measurements at 8 a.m. on 27th April and 9 p.m. on 17th April, 1972 was characterised by a value of r = 0.477, corresponding to t ~ 2.4 and P — 4%. The management of Strathern Apartments, Kings Park Avenue, kindly allowed ‘grab sam- ples’ to be taken at ground floor, 14th floor and 23rd floor levels in an attempt to obtain information about vertical distribution of NOx. No successful measurements were made even on days with well deflned shallow stagnation layers. These inquiries should be prosecuted in future with fully automated equipment operat- ing simultaneously at the three levels. General Comment To some extent 1972 was an exceptional year in terms of broad meteorological parameters, but 1973 was not an average year either. Whether the surprisingly high levels of NO^ measured on several days in 1972 were quite abnormal, perhaps the highest ever experienced, is a question which can only be answered by chemical measurements taken over the next several decades. In the absence of extremely de- tailed meteorological information about the sur- face inversions which have occurred in the past history of Perth it would appear unwise to assume that 1972 was a wholly abnormal year with respect to chemical concentrations. The special position of 1972 is perhaps suf- ficently emphasised by the fact that ten air dispersion alerts were issued in 1972 by the Bureau of Meteorology, compared to two in the whole of 1973 and one only in the first nine months of 1974. Irre pective of future events, we contend that the analytical figures obtained in 1972 have merit in drawing attention to the special pollution features of shallow inversions and in providing a datum against which to measure any future year, typical or atypical. Acknowledgements. — Both authors very freely ack- knowledge assistance in terms of information, experi- ence, time and in material ways offered to them by members of the Department of Public Health. West- ern Australia, the Commonwealth Bureau of Meteoro- logy, Perth, several Departments of the University of Western Australia, the Department of Environmental Protection. Western Australia, and many private citi- zens and organisations. The work reoorted here was undertaken in part while Dr. F. C‘. Cattell held tenure of the Western Mining Corporation Fellowship and we are glad pub- licly to recognize generous financial support by the Western Mining Corporation. References Bilger, R. R. (1972).— Motor vehicle emission controls for Australia. Clean Air 6: 44. Federal Register (1971). — 36 (84): 8187. Fisher R A. (195 S) ‘ Statistical Methods for Research Workers”. 13th Ed. Hafner Publishing Co., New York. Hum R W (1968).— Mobile combustion sources. In “Air Pollution”. A. C. Stern ed., Vol. 3, p. 62. Academic Press, New York. Katz, M. (1968).— Inorganic gaseous pollutants. In "Air Pollution”. A. C. Stern ed., Vol. 2, p. 80. Academic Press, New York. Tebbens B D. (1968).— Gaseous pollutants in the air. In "Air Pollution”. A. C. Stern ed., Vol. 1, p. 41. Academic Press, New York. Vines, R. G.. Gibson, L., Hatch, A. B.. King. N. K.. MacArthur. D. A., Packham, D. R. and Taylor, R. J. (1971).— On the nature, prop- erties and behaviour of bush-fire smoke. C.S.I.R.O. Division of Applied Chemistry Technical Paper 4: No. 1. Whittaker, E. T. and Robinson, G. (1944). — “The Cal- cuius of Observations” . 4th Ed. Blackie and Sons, London and Glasgow. Zimmer, C. E. and Larsen, R. I. (1965).— Calculating air quality and its control. J. Air Poll. Con- trol Assn. 15: 565. Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November. 1975. 74 7. — The petrology and probable stratigraphic significance of Aboriginal artifacts from part of south-western Australia by J. E. Glover^ Manuscript received 17 June, 1975; accepted 29 July, 1975 Abstract Aboriginal rock and mineral artifacts found in the southern Northampton Block, Perth Basin, Naturaliste Block, and areas around Albany and Esperance, consist of dolerite. actinolite-rich rock, granite, schist, quartz, quartzite, silcrete, sili- ceous and ferruginous shale and seven varieties of chert. The chert includes novaculite from the Proterozoic Coomberdale Chert, mottled chert probably from the Lower Triassic Kockatea Shale, two varieties of veined epidote-bearing chert from metamorphosed Precambrian strata, opaline and chalcedonic chert from the Middle-Late Eocene Plantagenet Group, similar chert from amygdales of the Neocomian Bunbury Basalt, and fossiliferous cryptocrystalline chert from an unidentified Middle-Late Eocene unit. Artifacts are particularly abundant in the Perth Basin. The fossiliferous cryptocrystalline chert has been found in firmly dated contexts only at Devils Lair, in strata with radiocarbon dates of 19 000 to 12 000 B.P. It is widely distributed in a strip along the western coast and probably came mainly from the west when sea level was significantly lower than at present. Some flakes of fossiliferous cryptocrystalline chert in the southern part of the Perth Basin may have come from Plantagenet Group rocks. Most other arti- facts in the basin were carried west from the Precambrian shield, but some in the north are of local origin. Fulgurite fragments are found in many artifact assemblages, but evidence of their utilization is lacking. Feldspar cleavage flakes are also rather common at some sites. Introduction The Perth Basin and adjacent shield areas in Western Australia contain numerous surface and near-surface accumulations of flaked stone artifacts (primary flakes and cores, utilized flakes and cores, and specific tool types) . This paper describes their petrology, and that of the associated, less abundant ground stone material (axes and grinding stones). Shield areas near the Perth Basin covered in this report include the Naturaliste Block, the south-western part of the Northampton Block, and part of the Pre- cambrian terrain along the southern coast, which is commonly covered by Plantagenet Group rocks. (See Johnstone et al. 1973 for a geological review of the region.) Many of the artifact sites listed in this paper were discovered by the author, or jointly by Mrs. S. J. Hallam and the author. Material from a few other sites was made avail- able by Mrs Hallam, and by the Western Aus- tralian Museum. The lithological composition of artifact as- semblages at the sites was estimated by counting 100 or more flakes. The abundance of artifacts in the Perth Basin is notable, and as those revealed (mainly in sand blow-outs) must rep- 1 Geology Department, University of Western Australia, Nedlands, W.A. 6009. resent a practically insignificant proportion of the total, the volume of transported rock material is truly remarkable. The rock types used include dolerite, actino- lite-rich rocks (rare as artifacts), granite, schist, quartzite, silcrete, siliceous and ferruginous shale, and at least seven varieties of chert, including novaculite, mottled chert, two varieties of veined epidote-bearing chert, two varieties of opaline and chalcedonic chert, and fossiliferous crypto- crystalline chert. Most of the artifacts described come from blown-out areas of sandy country and commer- cial sandpits, and their relative stratigraphic positions and times of accumulation have been established only in a broad way. According to Hallam (1972) sites assumed to be early on typological grounds are almost exclusively scraper assemblages and are very high in fossiliferous chert (the fossiliferous cryptocrystalline chert of this paper). Late assemblages contain many fabricators and are high in quartz. Intermediate sites have a high backed-blade component and are lithologically rather heterogeneous, Hallam also noted that near-coastal sites around Perth are rich in fossiliferous chert. This investigation shows that sites with abundant fossiliferous cryp- tocrystalline chert are concentrated within a belt in the central and northern Perth Basin that extends up to 40 kilometres inland. Flakes are commonly found where the coastal white or grey sand (the Quindalup Dune System of McArthur Sz Bettenay 1960) has been blown away to reveal underlying yellow or yellow- brown sand (Spearwood Dune System) commonly with emergent limestone pipes and pillars. Farther inland they are concentrated in blown- out areas of other sand formations. In commer- cial sandpits, artifacts are normally concentrated in the walls about 2 m below the original dune surface. Patterns of lithologic distribution are fairly clear (Tables 1 and 2). Fossiliferous crypto- crystalline chert is found in almost all near- coastal assemblages between Gregory and Black Point some 700 km to the south. Between Gregory and Eneabba (about 220 km) the proportion of this chert in the assemblages is less than 10 per cent and commonly closer to 1 per cent. From the Eneabba area to Mandurah (300 km), re- ferred to hereafter as the Eneabba-Mandurah belt, fossiliferous cryptocrystalline chert is in- variably present and is abundant (above 10 per cent, locally above 75 per cent). South of Man- durah toward Black Point (about 200 km) the chert apparently composes part of most near- coastal flake assemblages, but the number of Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November. 1975. 75 Figure 1. Map of south-western Western Australia showing flake localities. See Tables 1 and 2 for details about localities. All scales in km. flakes is normally small, and meaningful pro- portions cannot be given. Inland from the coast there is a general but uneven decline in fossili- ferous cryptocrystalline chert: near the north- eastern margin of the Perth Basin it is absent, and near the central eastern margin it ranges from 0-5 per cent. There is insufficient informa- tion to generalize about the south-eastern margin. It should be noted that estimates are difficult at sites on the Precambrian shield, for quartzite flaked by Aborigines is hard to distin- guish from chips formed by weathering of local rock. Mottled chert, feldspar, and siliceous and fer- ruginous shale are restricted to the area north of Dongara, whereas opaline and chalcedonic chert is confined to sites near the south coast. Quart- zite is the most widespread and abundant material. Silcrete and veined epidote-bearing chert are also fairly widespread but are normally minor constituents. Novaculite is uncommon south of the Perth metropolitan area. Man has been in Australia at least 32,000 years (Barbetti & Allen 1972) and perhaps far longer. In Western Australia excavations at Devils Lair, near Augusta, include artifact-bearing strata dating from about 12 000 to about 25 000 BP, and the bottom of the deposit has not been reached. Quartz artifacts are found in strata representing the whole time interval, but fossili- ferous cryptocrystalline chert has so far not been found in strata older than about 19 000 years (Dortch Merrilees 1973; Glover 1974a). This paper is based on the examination of some 18 000 chips and flakes in handspecimen, Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 76 Table 1 Sites at which the proportion of fossiliferous chert is known. Sites are located on 1 : 250.000 Series R 502 maps, and co-ordinates are based on grid references estimated to the nearest huniredth. Co-ordinates refer to centre points of large sites. No. Type of .site Map co-ordinates Flakes counted % fossiliferous chert 1 Sand blow-out, llidlev’s Pool area 22744932 448 4 2 Sand blow-out, 3 km E Florrocks Beach 23444785 1 3 Sand blow-out. 2 km S Walkawav 27164108 376 1 4 Sandy patch. Wondado Springs 28404048 325 0 5 Sand blow-out 0-8 km E Pell Bridge 29453744 301 0-5 6 Sand blow-out E side L. Arromore 29843469 400 1 7 Hoad cutting in sand. Kooringa area 32002760 319 45 8 Sandy soil. X bank Cockleshell Gullv 30372666 215 9 Sand blow-out. X bank Hill lliver 31492481 310 63 10 Sand spoil from dam about 1 km W. Dinner Hill 35772452 154 2-5 11 Sand blow-outs. Pinnacles area 31692063 2107 69 12 Sand near dam, Caro Station 34651927 249 41 13 Hoad cutting in sand, S bank l\Ioore Jliver, Regans Ford 36921652 133 17 14 Blow-outs in yellow sand near mouth of Moore River 34851205 886 86 15 Sand blow-out 2 km XXW Gingin Railway Station 38801260 406 8 16 Sandy area 6 km S Bullsbrook East 40370758 290 8 17 Blow-outs in yellow sand 1 km north Mullaloo Beach 37190695 127 87 18 Gnangara Sandpit .... 38370671 249 20 19 Brambles Sandpit .... .... .... .... .... .... 38410669 193 13 20 Sandy area. SW margin of Lake Gnangara 38550680 684 21 Bell Bros. Sandpit, Gnangara Hoad 39730675 309 22 Ready !Mix Sandpit, Beechboro .... . .. .... 39040592 307 23 23 Hoad cut in sand. Beechboro Hoad .... 39180588 228 29 24 Widgee Road Sandpit, Beechboro .... 39380584 291 16 25 Sand Patch, S side Talbot Wav, Woodlands 37880540 373 71 26 Sand dune, XW of Lake Monger .... 38120517 344 58 27 Red sand. E shore of Lake Monger 38220512 642 44 28 Sandpit. Maida Vale 39880510 347 16 29 Exposed sand, airport runway extension .... 39600492 348 17 30 Rail cutting in sand near Wittenooni Road, Maida Vale 39820496 353 31 Sand blow-out near Bingham Street, Maida Vale 39800490 384 32 Exposed sand. Kewdale .... .... .... .... 39550449 669 32 33 Sandpit, Hardev Road, Cloverdale 39630455 717 23 34 Exposed sand, Xewburn Hoad. Kewdale 39690452 413 19 35 Sand blow-out X side Dowd Street, Kewdale 39550434 318 21 36 Sandpit XE White Street, Orange Grove 40140397 306 G 37 Exposed dune, corner High Hoad and Leach Highway, Riverton 38700377 378 20 38 Exposed sand. Metcalf Hoad. Lynwood .... 39160377 181 14 39 Sand blow-out near Riley Hoad, Xicol Hoad, Lynwood 38980364 317 19 40 Snashall Bros. Sandpit. Bibra Lake area .... 38080308 394 17 41 Coopers Sandpit. Canning ^ ale 39220303 277 16 42 Hot Mix Sandpit. Gosnells ... 40110265 361 G 43 Ready Mix Sandpit, Porrest Hoad, Jandakot 38590273 299 29 44 Calsil Sandpit. Forrest Hoad, Jandakot .... 38780266 118 13 45 MWSS X' DB Sandpit, Lilian Avenue, Armadale 40110266 359 0-*">5 46 Fremwells Sandpit, Hopkinson Road. Cardup .... 39710126 1 589 4 47 Wellard Sandpit. Farmelia 38540114 174 18-5 48 Sand blow-outs, bang’s Farm, Mimdijong 39820092 500 o 49 Sand blow-out. 0-5 km S railway l)ridge. Mundijong .... 40040058 374 7 • 5 50 Sandpit 4-5 km E Stake Hill Bridge, Mandurah 38049805 116 28-5 and microscopic study of 180 thin sections. The study was supplemented by several complete and partial chemical analyses, by X-ray determina- tion of some silica minerals, and by examination of some silcretes with the scanning electron microscope. Colour designations used in the lithologic des- criptions are taken from the Rock-color Chart distributed by the Geological Society of America (Rock-color Chart Committee 1963), and are accompanied by a numerical code from the same source. Lithologies Dolerite Dolerite was used for grinding stones and axe- heads, and is not common as flaked artifacts. The best dolerite for flaking is fine-grained and comes from small dykes or from the chilled margins of large dykes on the Yilgarn Block. Waterworn boulders were a source of some dolerite. Actinolite-rich rocks Actinolitic rocks are rather rare as artifacts, and have so far only been found at Wondado Springs (site 4) and near Howatherra (Site Aa), where they compose less than 5 per cent of the flakes. The finer grained rocks are aphanitic, black (N 1) through greenish black (5G 2/1) to dusky yellow green (5GY 5/2) and are tough, with conchoidal fracture. They are made up mainly of interlocking actinolite grading to tre- molite, with pyrite, magnetite, quartz, and other minerals. Coarser grained rocks tend to be greenish grey (5G6/1) with a less well developed conchoidal fracture, and are made up mainly of andesine and actinolite. They resemble ura- litized dolerite. Journal of the Royal Society of Western Australia, vol. 58. Part 3, November, 1975. 77 Table 2 SiteH at which the proportion of fossiliferous chert is not precisely known. Some sites (Ai-Aiq) may contain numerous fragments weathered from nearby rocks. Other sites Fj) have yielded insufficient flakes, or have not been closely investigated. Museum sites have not been inspected by the author. Sites Si-S^ are near Plantagenet Group rocks and contain dis- tinctive chert flakes. Map co-ordinates as for Table 1, except that sites B^-B^ are located on Sheet 1930 {Edition 1) Series BQll. Site No. Type of site ilap co-ordinates Fossiliferous cryptocrystalline chert* Chert with colloform opal and/or abundant mierocrystalline chalcedony* Ai Sand blow-out 1-4 km W Oakabella Homestead 24824666 r a A, Sand blow-out 1 -5 km S Howatbarra, 400 m S Jloyce Home- 25234603 r a As stead SaiKl blow-out, Howatbarra area, 3 km WSW Boyce Home- 25054596 r a A^ stead Small sand blow-outs in field, W side Highway No. 1. 0-8 km 25214483 r a Ai, S Buller Biver Small sand blowouts. 0-5 km \V locality A., 25144485 a a Ag Sandplain 2 km S Yandanooka near Government Windmill. 35413660 R a A; Bundanoon Sand blowout about 3 km SW Winchester, 0-7 km NW Kilcur 39003096 r a Ag Pool, near mill Sand blowout 4-5 km SSW Gillinjjarra. 0-5 km W hi^iway 40331674 r a As Junction of Avon and Brockman Bivers 41210814 a a A 10 S side Jane Brook, about 1*3 km E Stoneville 41870573 a a Sandpit about 7 km N Piniarra. North Dandalup Boad 38739728 P a H, Small sandpatch 0-4 km SSW Meelup Sprinci, S Dunsborough- LH206816 P a Ba Naturaliste Boad Sand])it E Bussel Highway at iunction with Yelverton Boad LH296653 a 1^4 Sand blowouts, (’owaramup Point LH 128509 r a lis Small sand bk^wouts, S side Margaret Biver Boad 1 km NE IJ1157395 P a Wildcliffe House Sand, N .side Albaiiv-Denmark Boad 3 km W Torbay 56146730 a P s„ Boad cutting in sand, Nannarup Boad 3 km NNW Nannarui) 60386803 a P Ss Sand on laterite. between Dalvup tennis courts and Dalyup 46028345 (1 P S 4 Biver N side Esperance Albany Highway Bailwav cutting in sand. 0-5 km N Shark Lake Siding 48778275 a I> S 5 Sand blowout in dune, 0-25 km N Shark Lake Siding 48748274 H P Sg Sand blowout, 8 side Esperance Boad 3-5 km W .Mt. Edward 50128247 H P m038 Yeagcrui) Dune. Id km 8 Pemberton p a B271 (’leared paddock. Half Moon Farm, 8-4 km SW Kojonup ... Mainly silcrete, no chert B1763 Sand dune. E shore large lake, ,E side main road, 4 km S 8 E Lake P a Vi Banks Hst. near Tambellup Sand blowout about 1 km E Black Point 35417488 Chert from volcanic amygdales present * a = absent, r = rare, p ^ present Granite and Schist Fragments of granite and schist are found in some Perth Basin sites, and the rocks would have been carried westerly from the Yilgarn Block. Most of the material examined, because of its coarse grain size and poor coherence, was probably unsuitable for flaking, and was used mainly for grinders. Quartz Quartz (rock crystal) was used where avail- able, and large clear crystals were not spared. Many chips resemble flaked window or bottle glass, from which they can be distinguished by their anisotropism and lack of bubbles. Rock crystal, though not abundant, is widespread near shear zones on the Yilgarn Block and no specific sites of origin have been determined. Quartzite The term quartzite is used for all polycrystal- line quartz rocks because it is difficult to dis- tinguish fragments derived from such diverse parents as large quartz-rich pods (the quartz blows of prospectors), metaquartzites, and highly silicified orthoquartzites such as those found interbedded with chert in the Proterozoic Moora Group. It is not always easy, in fact, to distinguish quartzite and Coomberdale Chert without microscopic examination. Quartzite was widely utilized for grinding implements and flakes in south-western Australia, and is the main constituent at many sites in the Perth Basin. Source rock is abundant on the Yilgarn Block and on other Precambrian terrains, and the artifacts generally give no precise indica- tion of their place of origin. Some have been broken from waterworn boulders, which would have saved the work of quarrying. In the northern part of the Perth Basin rounded peb- bles and boulders from the Tumblagooda Sand- stone and Mesozoic units seem to have served locally as sources. Feldspar A significant proportion d-28 per cent) of flakes at some sites north of Dongara (sites 2, and A1-A5 inclusive) is made up of cleaved perthite. The flakes are commonly about 3 cm X 2 cm X 1 cm, but they vary in size between wide limits. The overall colour of many flakes is pale yellowish orange (10 YR 8/6) to grey- ish orange (10 YR 7/4) with the host micro- Journal of the Royal Society of Western Australia, Vol. 58. Part 3, November, 1975. 78 dine coloured in shades of brown, pink and grey, and the abundant plagioclase veins in shades of yellow. Feldspar has two perfect cleavages (001) and (010), and reflection from these smooth sur- faces causes the flakes to glisten in the sun. The cleavages meet at an angle of 86® to form straight edges which would be useless for cut- ting. It is not known if the feldspar was used by the Aborigines. Fulgurites Fragments of lechatelierite (silica glass) from sand fulgurites have been found at ten artifact sites in the Perth Basin, and have been de- scribed in detail elsewhere (Glover 1974b, 1975). Sand fulgurites are narrow tubular bodies up to two or three metres long that resemble plant roots in external shape and orientation, and they form from the fusion of sand by lightning. The tubes commonly break into fragments of a few square centimetres in area, and remain as lag in dune blow-outs. Fulgurites that formed in sand blown over former artifact sites can thus be concentrated with the artifacts during subsequent deflation. Some older fulgurite fragments may have been carried to the site by Aborigines, but evidence of handling would be almost impossible to establish. In handspecimen, non-tubular fulgurite frag- ments can be recognized by a smooth, trans- lucent somewhat mammilated appearance on one surface, and a rough opaque aspect, with numerous embedded sand grains, on the other. Microscopically the glass is vesicular with a refractive index close to 1.461. Silcrete Silcrete fragments can be recognized in hand- specimen because they contain quartz grains, commonly in shades of light grey, in a uniform, porcellaneous cement that generally ranges be- tween very pale orange (10 YR 8/2) and pale yellowish orange (10 YR 3/6). The rock has pronounced conchoidal fracture. In thin section most of the silcrete fragments consist essentially of quartz grains in colourless or pale brown cement. A silci^ete from the Northampton area (see Fig. 2B) has a more complex clastic assemblage and contains quartz (some grains with well-defined outgrowths), glauconite, microcline, leucoxene, tourmaline, and siliceous grains of unknown origin. The cement seems to be isotropic under the polariz- ing microscope, but X-ray powder photographs of a typical specimen (Uni. No. 74,628) from site 32 show quartz lines. The refractive in- dex is close to 1.549 ± 0.002, and the specific gravity is indistinguishable from quartz in heavy liquids. Scanning electron microscopy shows that the average grain size is about 1 fxm. It seems therefore that the mineral is quai’tz whose unusual optics are caused by its finely divided state. Silcrete apparently forms in at least two ways (see Hutton et al. 1972) and can form from dif- ferent kinds of rock. Silcretes from different sources are therefore likely to differ texturally and even mineralogically, in rather distinctive ways. Silcrete artifacts in the Perth metropoli- tan area, and at Gingin, resemble outcropping silcrete from the Kojonup area very strongly. ABC Figure 2. — Thin sections of flakes. A. — Siliceous ferruginous shale (Uni. No. 74451/1) from site 2, Horrocks Beach area The curved siliceous bodies of microcrystalline quartz may be algal. Cement consists of hema- tite and goethite. Flake probably derived from Kockatea Shale. Diameter of field 1.3 mm. B.— Silcrete (Uni. No. 74501) from site A-.u Howatharra area. Colourless grains without cleavage are quartz, colourless grains with cleavage are microcline. Dark grain with high relief is tourmaline. Cement is very finely divided quartz. Dia- meter of field 1.3 mm. C. — Mottled ferruginous chert (Uni. No. 74481) from site Ar,. Buller River area. Dark areas contain mixture of hematite, limonite and cryptocrystalline silica. Colourless areas contain radiating chalcedony with cores of microcrystalline quartz. Probably derived from Kockatea Shale. Diameter of Held 1.3 mm. Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 79 Siliceous ferruginous shale The siliceous ferruginous shale flakes are tough, laminated, have conchoidal fracture and range in colour from greyish red (5R 4/2) to moderate brown (5 YR 4/4 to 5 YR 3/4). Micro- scopic examination reveals that up to 15% of some flakes is made up of bodies with strongly curved to fairly straight walls from 0.01 to 0.05 mm thick, composed of microcrystalline quartz, in a cement of hematite and goethite (see Fig. 2A). This rock is found at sites 2 and A 2 and is apparently restricted to the nor- thern part of the Perth Basin. It is the same as that illustrated by Karajas (1969, fig. 24, unpubl. data) from the so-called algal unit of the Triassic Kockatea Shale. Chert Classification According to the hypotheses of primary origin, chert is deposited as silica gel, or forms ^ by partial redistribution, in situ, of silica precipi- tated by organisms such as diatoms and radio- larians. Secondary origin can be demonstrated for many cherts by their field relationships or palimpsest microstructures, and generally in- volves replacement of calcareous rocks. Though there is a considerable literature, ad- vances in chert petrology have been slow (see Dapples 1967) and classification and nomencla- ture have not kept pace with other sedimentary rock groups, notably the carbonates. A descrip- tive nomenclature partly based on that of Williams, Turner and Gilbert (1954) is adopted to distinguish between the varieties of chert artifact described here. The common term flint used by some as a synonym for chert, and by Williams et al. and others for a tough grey or black variety of chert with conchoidal fracture, is unsatisfactorily defined, and is not used. The term chert is taken to include compact sili- ceous rocks composed of opal, chalcedony and cryptocrystalline or microcrystalline quartz, or a mixture of these constituents. Chert can be any colour, and can originate in several ways. Chert composed mainly of microcrystalline quartz is called novaculite in this paper, a term used in parts of the United States and already applied to the Coomberdale Chert by Logan and Chase (1961). Apart from novaculite the descriptive terms opaline and chalcedonic chert, mottled ferruginous chert, veined epidote- bearing chert, and fossiliferous cryptocrystal- line chert, have been used for distinctive local varieties. There is unfortunately disagreement about the size limits indicated by the com- monly used terms “cryptocrystalline” ^ and “microci’ystalline” (see Bissell and Chilingar 1967. Table IV). This paper follows Pettijohn (1957, Table 18) in taking the boundary between the two at 0.01 mm, as it then forms a suitable dividing line between two distinctive varieties of chert found in Western Australia. Crystal size in cryptocrystalline chert is taken to be less than 0.01 mm, whereas microcrystalline grains are defined as being greater than 0.01mm, but too fine for observation with the naked eye. Novaculite Novaculite artifacts are commonly translu- cent in thin chips, and range through shades of light to dark grey, greyish yellow, pink, and red. They are hard, dense and non-porous. Most are banded, but some are intraformational breccias. Artifacts without such structures are difficult to distinguish from quartzite without microscopy. Twelve novaculite artifacts were sectioned and consist essentially of microcrystalline quartz, locally cut by quartz veins up to 0.5 mm thick. The interlocking quartz crystals range in dia- meter mainly between 0.05-0.25 mm, and two of the flakes (Nos. 74527/3, 74536/2) show palimp- sests of dolomite rhombs clearly outlined by iron oxide (see Fig. 2B). The novaculites show macroscopic and micro- scopic features characteristic of the Coomber- dale Chert from which they were certainly de- rived. Novaculite artifacts are far less frequent than artifacts of fossiliferous cryptocrystalline chert at central Perth Basin sites, despite the abundance of strongly outcropping novaculite along many parts of the eastern margin of the basin. Presumably the almost perfect conchoidal fracture of the fossiliferous cryptocrystalline chert made it the more useful of the two. Opaline and chalcedonic chert The opaline and chalcedonic chert artifacts superficially resemble the fossiliferous crypto- crystalline chert artifacts described below, but they have a less developed conchoidal fracture and may be particoloured or mottled. The chal- cedony commonly ranges from shades of light and medium grey (N8 — N5) to pale yellowish brown (10 YR 6/2) and the opal ranges from white (N9) through very pale orange (10 YR 8/2) to more brownish colours where strongly stained. The minerals can be easily distinguished in some artifacts because the chalcedony is translucent whereas the opal is not. Under the microscope (Fig. 30 the chalce- dony is generally microcrystalline, clear, and colourless whereas the opal, which has pro- nounced negative relief, is somewhat cracked and ranges through pale and very pale shades of brown and orange. The opal is colloform in places and commonly includes bodies of radiat- ing-fibrous chalcedony, with quartz cores show- ing undulose extinction. Some of the opal is completely isotropic, but most of it has very low birefringence and fibrous structure under crossed polarisers. X-ray powder photographs of slightly birefringent opal from Esperance show several broadened a-cristobalite lines characteristic of “common” opal, i.e. the opal-CT of Jones & Segnit (1971), which is the same as Mallard’s lussatite (Mallard 1890). Palimpsests of shell fragments, and unsilicified glauconite pellets are found in some opal but are less common in the chalcedony. Flakes of opaline and chalcedonic chert are abundant at sites in the Denmark, Albany and Esperance areas, and are mineralogically and texturally identical with nearby siliceous Plan- tagenet rocks. There is no doubt about the local derivation of these artifacts. Journal of the Royal Society of Western Australia, vol. 58, Part 3, November, 1975. 80 B Figure 3. — Thin sections of flakes. A. — Veined epidote-bearing chert (Uni. No. 74671) from site 15, Gingin area. Veins of microcrystalline quartz cut cloudy groundmass of microcrystalline and cryptocrystalline epidote and silica. The large colourless grains consist of aggregates of interlocking microcrystalline quartz. The rock resembles in texture a meta-volcanic rock or a metamorphosed tuif, and is probably derived from Precambrian terrain. Diameter of field 3.5 mm. B. — Novaculite (Uni. No. 74536/3) from site A 7 , Winchester area. Note the siliceous, rhomb-shaped palimpsests of dolomite, outlines by iron oxide, in cement of microcrystalline quartz. Probably derived from Coomberdale Chert. Diameter of field 0.26 mm. C. — Opaline and chalcedonic chert (Uni. No. 74674) from site S«, Esperance area. The opal (stippled, with cracks) contains bodies of radiating-fibrous chalcedony. An indistinct organic remnant can be seen in the opal toward the top of the field. Derived from Plantagenet Group. Diameter of field 1.3 mm. Farther west, in the Black Point area, sili- ceous amygdales weathered out of the Neocomian Bunbury Basalt are found in artifact assem- blages. The mineralogy of the amygdales is complex and variable, but many contain only colloform common opal (opal-CT), chalcedony and quartz, and chips from these amygdales are practically indistinguishable from palaeontolo- logically barren chips of the Plantagenet chert. Flakes containing zeolites, celadonite or barite can be confidently ascribed volcanic origin. Mottled ferruginous chert Mottled ferruginous chert flakes range in col- our from shades of yellowish brown to purple and are mottled with spots or patches in shades of orange and grey. Microscopic examination (Pig. 20 shows that the mottling is caused by irregularly shaped masses of intimately associ- ated limonite, hematite and cryptocrystalline silica separated by fringes of radiating chalce- dony from aggregates and microcrystalline quartz. Flakes of the rock are fairly numerous at site A, I (10%) but are not normally common and seem to be confined to the northern part of the Perth Basin. Their origin is uncertain, but they may be derived from silicified portions of the Triassic Kockatea Shale. Veined epidote-bearing chert Veined epidote-bearing chert flakes are brown, grey and yellow, with an olive or green cast from disseminated epidote, and are cut by narrow (generally <0.5 mm) quartz veins. Two varieties have been distinguished micro- scopically. One variety contains large, composite grains (up to 0.7 mm in diameter) consisting of interlocking microcrystalline quartz, and rarer large epidote grains in a cloudy groundmass of cryptocrystalline to microcrystalline silica and epidote. Some of the large composite quartz gi’ains have remarkably straight edges, and re- semble silicified phenocrysts in a metamorphosed volcanic rock or crystal fragments in a metamor- phosed tuff. The rocks are microfaulted and are cut by narrow veins of microcrystalline quartz (see Fig. 3A). The second variety consists of poorly sorted, angular, strained quartz grains, microcrystalline quartz aggregates, and micro- cline and epidote grains in a cryptocrystalline to microcrystalline matrix of silica and epidote. These rocks are also veined and microfaulted, and resemble metamorphosed greywackes (see Fig 4A). The origin of both rock types, however, is still rather speculative K Veined epidote-bearing artifacts are widely distributed throughout the Perth Basin and on its margins, but constitute only a very small proportion of material at sites examined. The parent rock almost certainly comes from shield areas. Fossiliferous cryptocrystalline chert Petrology: The surface colour of the fossili- ferous cryptocrystalline chert flakes is generally 1 These textures may be misleading. The flakes also resemble mylonltized Precambrian rock examined after this paper went to press. Journal of the Royal Society of Western Australia. Vol. 58, Part 3. November, 1975. 81 ABC Figure 4. — Thin sections of flakes. A. — Veined epidote-bearing chert (Uni. No. 74699) from site 49, near Mundi- iong. Veins of microcrystalline quartz cut a cloudv matrix of microcrystalline and cryptocrystalline epidote and silica. Large grain with cleavage is microcline, other grains consist of strained quartz, or aggregates of quartz. The rock is probably derived from Precambrian terrain, and resembles meta-greywacke. Diameter oi iield 1.3 mm. B. — Fossiliferous cryptocrystalline chert (Uni No. 74606/1) from site 11, the Pinnacles area. Palimpsests of bryozoan and other fossils in matrix of cryptocrystalline silica. Probably silicifled Eocene limestone from off-shore source. Diameter of field 1.3 mm. C.— Fossiliferous cryptocrystalline chert (Uni. No. 74606/2) from site 11, the Pinnacles area. Palimpsest of small bivalve in cryptocrystalline silica. Note drusy, opaline fringe. Probably silicifled Eocene limestone from off-shore source. Diameter of field 1.3 mm. close to the colour of the surrounding sand, and ranges mainly from white through pale shades of grey, yellow, brown and orange. Common sur- face colours are white (N9), pinkish grey (5 YR 8/1), greyish yellow (5 Y 8/4), pale yellowish brown (10 YR 6/2), light brown (5YR 5/6), very pale orange (10 YR 8/2) and greyish orange (10 YR 7/4). Sections through some flakes show that the outer colour is a patina that gives way to white or pinkish grey (5 YR 8/1) for up to two or three mm, and passes into a core of pale yellowish brown <10 YR 6/2) or a similar colour. Other flakes, however, particularly those taken from white sand, are white or nearly so through- out. Thin-section examination reveals partly to completely silicifled Bryozoa, Foraminifera, sponge remains, small bivalves and echinoderm fragments together with a few unsilicified glau- conite pellets in an essentially cryptocrystalline matrix (Figs. 4B, 40. A little quartz sand and silt is present in some of the chips. The epithet cryptocrystalline is justified because over 50% of the rock is composed of grains less than 0.01 mm in diameter, but it is not uncommon for about 30% of the matrix to consist of chalce- dony grains between 0.01 and 0.04 mm in diam- eter, with somewhat coarser chalcedony filling fossil tests. Many fossils are partly replaced by very finely divided opal, or have a drusy opaline fringe about 0.005 mm thick. In places the fringe is directed outward from the shell surface. The small size and high negative relief of the opaline bodies make them practically opaque undei magnifications of less than XlOO, and where finely divided opal is abundant it resembles cloudy argillaceous material. A few larger opaline bodies attain a length of 0.02 mm, and have pitted surfaces as though etched. A well- developed opaline drusy fringe is shown in Figure 4C. Table 3. Analysis of core and rim of a flake from the Pinnacles area. (from University specimen No. 74606)* Core Rim S102 97.9 98.2 TiO, 0.01 0.01 AlA 0.13 0.19 Fe20;i 0.17 0.17 MnO <0.01 <0.01 MgO 0.03 0.05 CaO 0.11 0.09 K2O 0.04 0.04 Na^O 0.01 0.01 P2O5 0.20 0.13 Loss on ignition 1.6 1.2 100.2 100.1 * Analyst Supervise-Sheen Laboratories Pty. Ltd. XRF except P2O.5 by colorimetry, Si02 by difference. Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 82 Most of the chert is evidently silicified fossili- ferous limestone, and the texture suggests that it was originally biomicrite, according to the classification of Folk (1962). Remnants of cores commonly retain an uneven, roughly curved sur- face from which silicified Bryozoa project up to a millimetre or more. It appears from the gene- ral curvature and from the size of the larger cores, that the flakes were generally struck from chert nodules about the size of a man’s fist. A few artifacts contain abundant spicules, indi- cating a spongolite precursor. A typical flake from the Pinnacles was analysed (Table 3) and shows almost complete silicification. The pale yellowish brown (10 YR 6/2) core and the white (N9) rim with its surface patina of very pale yellowish orange (10 YR 8/4) are not signifi- cantly different in composition. The fossiliferous cryptocrystalline chert breaks fairly easily with smooth conchoidal frac- ture into sharp -edged flakes, which doubtless accounts for its widespread use. However, some chert fragments from sites of white sand have slightly rough surfaces, rather soft blunt edges, and are porous or even friable, apparently be- cause of the removal of calcium carbonate, which can form up to 25% of incompletely silici- fled artifacts. Many flakes in white sand have probably disintegrated completely in this way. An Eocene age is shown for the chert by the Bryozoa, and is confirmed by the presence in two chips of the Middle-Late Eocene foraminifer Maslinella chapmani Glaessner and Wade (Glover & Cockbain 1971). Distribution and Source: Fossiliferous crypto- crystalline chert artifacts are found throughout the Perth Basin, and on the Northampton Block to the north and the Naturaliste Block to the south-west, a north-south distance of about 700 km. They are also found on the southern part of the Precambrian shield at a site near Tambellup. It has already been pointed out that the sites with the highest proportion of these chert artifacts are invariably near the western coast. Chert artifacts are known from sites fre- quented up to the ethnographic present (e.g. Lake Monger, see Pig. 1) and from strata about 12 000 to about 19 000 years old (Devils Lair, see Dortch & Merrilees 1973). The “Middle-Late” Eocene dating of the fos- siliferous cryptocrystalline chert is clearly crucial for no rocks of that age are known to outcrop in the Perth Basin (see for example Quilty 1974 a, b). There are several known units of requisite age elsewhere, namely Giralia Cal- carenite, Merlinleigh Sandstone and Pindilya Formation in the Carnarvon Basin, Norseman Limestone on the southern Yilgarn Block, Plan- tagenet Group in the Albany-Esperance area, and Toolinna Limestone and Wilson Bluff Lime- stone in the Eucla Basin (see the correlation chart of Playford & Cope 1971). The concentration of chert-rich sites near the western coast should be considered in the light of eustatic variation in sea level, as was well appre- ciated by Hallam (1974, p. 80). Recent data indicate that when the oldest known Western Australian chert flakes were being manufactured (about 19 000 years BP), world sea levels were 85-90 m (i.e. about 50 fathoms) lower than now (Morner 1971). These data may be conservative, if applied to Western Australia coasts, accord- ing to new interpretations of global isostacy (see Chappell 1974). They imply that 19 000 years ago the coast would have been about 40 km west of Perth and about 90 km west of Gerald- ton and Bunbury, and the Perth Basin would have had roughly twice its present exposed area. Many artifact sites were probably covered by advancing seas during deglaciation, and there are probably numerous chert-rich sites off-shore. It is reasonable to hypothesize a westerly off- shore source for the chert artifacts in silicified Late Eocene limestone. North-trending wedges of such limestone could well overlie the Late Pal- aeocene-Early Eocene Kings Park Formation, having survived later bevelling. Rock in these wedges could be exposed locally on the sea-floor, perhaps as windows through Pleistocene lime- stone. Reasons for these suppositions are set out in the three numbered paragraphs below. (1) The ratio of chert to other rock types at artifact sites tends to increase markedly from east to west. (2) It is unlikely that the remarkably large volume of chert at sites in the Eneabba- Mandurah belt would have been carried up to 500 or 600 km from the Carnarvon Basin or Albany- Esperance area. It is more likely to have come from western sources distant a few kilometres, or some tens of kilometres at most. (3) Late Eocene rocks have been found in WAPET’s Challenger No. 1 offshore well, about 60 kilometres west of Mandurah, between depths of 510 m and 590 m. These are the only known Late Eocene rocks in the Perth Basin, and have not been recorded from other off-shore wells. No wells have tested the rocks within 20 kilometres of the western coast opposite the Eneabba- Mandurah belt, where Late Eocene rocks, up- dip from the intersection in Challenger No. 1, could crop out on the sea floor. Possible landward sources for the fossiliferous cryptocrystalline chert artifacts will now be considered. Of the Carnarvon Basin formations, Merlinleigh Sandstone and Pindilya Formation are lithologically unsuitable. The Giralia Cal- carenite is fossiliferous, glauconitic and locally silicified but contains large Foraminifera absent from the fossliferous chert artifacts. The low concentration of chert flakes at near-coastal sites north of Eneabba, and the increased abund- ance of chert to the south, farther from the Carnarvon Basin, seem at first sight to support the idea that the Giralia Calcarenite was not the source of the chert. It is not unlikely, how- ever, that some intermediate sites rich in chert are hidden off shore. The soundest reasons for doubting that the Giralia Calcarenite is the source of the abundant chert in the central Perth Basin are the absence of large Foramini- Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 83 fera in chert specimens examined, and the phy- sical difficulty of transporting large volumes of chert over distances of up to 600 km \ The nearest possible sources of fossiliferous cryptocrystalline chert known in southern West- ern Australia are at least 500 km from the chert-rich artifact sites of the Eneabba-Man- durah belt. Units that should be considei*ed are the Plantagenet Group and Norseman Limestone which patchily cover Precambrian basement rocks, the Toolinna Limestone farther east, and the Wilson Bluff Limestone exposed at Wilson Bluff near Eucla and in caves nearby. Chert artifacts from Wilson Bluff have been recorded near Eucla, and their use apparently continued into the ethnographic present. Geologically similar material has been quarried at Koonalda Cave, one hundred kilometres east of Eucla, which is shown to have been in use between 20 000 and 13 000 years ago (Polach et al. 1968). Chert from the Wilson Bluff Limestone closely resembles chert used for artifacts in the Eneabba-Mandurah belt, but is quite unlikely to have been carried 1 200 km to the area. Chert flakes at sites around Esperance, Albany and Denmark are mainly composed of colloform opal and microcrystalline chalcedony, or of chal- cedony alone, and are derived from local Plan- tagenet outcrops. No chert with colloform opal has been found in the Eneabba-Mandurah belt. Some Plantagenet chert from outcrop in the Northcliffe area kindly given to the writer by Mr. C. E. Dortch of the Western Australian Museum is composed of microcrystalline chal- cedony. This resembles the Eneabba-Mandurah material more closely, but is generally coarser grained, and contains fewer microfossils. Two cryptocrystalline chert flakes from the Tambellup area borrowed from the Museum (Museum No. B1763) have blurred textures under the microscope, but are fossiliferous and look rather like the Eneabba-Mandurah mate- rial. Two flakes from the Pemberton area (Museum No. B1038) are largely cryptocrystal- line, but their unusual palimpsest texture, pos- sibly derived from pelletal limestone, has not been precisely matched in the Eneabba-Man- durah material. The flakes from these two sites are of uncertain origin, but their most likely provenance is within a nearby area of Plantag- enet rocks that are non-opaline and cryptocrys- talline. It therefore seems that some Plantagenet rocks not yet examined resemble the Perth Basin flakes in lithology: if so, they could ac- count for some of the chert flakes in the south of the basin. It is uncertain how far north such chert may have been carried, but material from the Plantagenet Group can hardly account for the distribution or volume of chert in the cen- tral part of the basin. It should be added that silicifled Norseman rocks seem commonly to be 1 Note added in proof. Dr. P. E. Playford, Geological Survey of Western Australia, reports (pers. comm.) finding bryozoal chert artifacts (including one large core 7x9x5cm) on Tamala Station, about 20km south of Shark Bay. These resemble the fossili- ferou cryptocrystalline chert in handspecimen. Their significance is yet to be evaluated. fairly opaline, and therefore appear unsuitable as a source. Silicifled Toolinna Limestone, far- ther east, has not been examined. The source of the chert in the Perth Basin therefore remains rather enigmatic, but on bal- ance the abundance and distribution of artifacts in the Eneabba-Mandurah belt favours the hypothesis of offshore origin for them. It also accords with Hallam’.s suggestion based on typology, that chert-rich sites in the Perth area were the earliest (Hallam 1972). After submerg- ence of their source rocks the Aborigines would have been forced to use material mainly from the east, though some chert from old sites would have been re-used. The present position of sea level was probably attained by about 6 000 BP (Thom & Chappell 1975), and fresh chert would therefore have been unavailable by that time. Conclusions The full ethnographic significance of the arti- facts will have to await further typological, stratigraphic and petrologic studies. The search for Middle-Late Eocene rocks in future off-shore wells Or in dredged material is especially im- portant. At present, the data demonstrate local derivation of material near the Northampton Block and in the Albany-Esperance area, and east-west transportation of Precambrian rocks throughout most of the Perth Basin. Superim- posed on this model is a probable pattern of west-east transportation in the Eneabba-Man- durah area, of fossiliferous cryptocrystalline chert from sites now submerged. In the southern part of the Perth Basin there may be chert from both the west and Plantagenet rocks in the east. The conclusions are set out individually below. (1) The flakes of mottled chert, and siliceous and ferruginous shale at northern sites were locally derived. (2) The fossiliferous opaline and chalcedonic chert on the south coast was derived locally from the Plantagenet Group. Some of the non- fossiliferous material came from amygdales in the nearby Bunbury Basalt. (3) The quartz, quartzite, dolerite, granite, schist, novaculite and veined epidote-bearing chert in the central Perth Basin came from east of the Darling Scarp. Each rock probably came from a number of localities. River boulders were a significant source of quartzite and dolerite. (4) The fossiliferous cryptocrystalline chert found in the Eneabba-Mandurah belt probably came from an off-shore source in the west. Chert in the southern Perth Basin may have come partly from Plantagenet rocks to the east. (5) The presence of fossiliferous cryptocrys- talline chert in strata aged from about 12 000 years BP to about 19 000 years BP in the Devils Lair, and the remarkable abundance of similar material at some central Perth Basin sites, indicates extensive and persistent exploitation of the source rocks. (6) Further petrological work on silcrete arti- facts may throw additional light on the patterns of transportation. Journal of the Royal Society of Western Australia. Vol. 58. Part 3, November, 1975. 84 (7) Fulgurite fragments are concentrated with artifacts in blowouts by deflation. There is no evidence that fulgurites were worked by the Aborigines, Acknowledgements. — Mrs. S. J. Hallam, Department of Anthropology, University of Western Australia, con- structively criticized an early draft of this paper, and supplied some material for examination. The Museum of Western Australia also supplied material for exami- nation. West Australian Petroleum Pty. Ltd. kindly released information about Late Eocene rocks encoun- tered in their Challanger No. 1 well. References Barbetti, M., and Allen. H. (1972). — Prehistoric man at Lake Mungo, Australia, by 32 000 years BP. Nature, Land., 240 (5375) : 46-48. Bissell, H. J., and Chilingar G. V. (1967). — Classifica- tion of sedimentary carbonate rocks. In “Carbonate Rocks”. Developments in Sedi- mentology 9A: 87-168. Elsevier, Amsterdam. Chappell, J. (1974). — Late Quaternary glacio-hydro- isostacy on a layered earth. Quaternary Research, 4: 405-429. Dapples, E. C. (1967). — Silica as an agent in diagenesis. In “Diagenesis in Sediments” . Develop- ments in Sedimentology 8 (Eds. Larsen, G., and Chilingar, G. V.), 323-342. Elsevier, Amsterdam. Dortch, C. E., and Merrilees D. (1973). — Human occu- pation of Devil’s Lair, Western Australia during the Pleistocene. Archaeol. Phys. Anthrop. in Oceania, 8: 89-115. Folk, R. L. (1962). — Spectral subdivision of limestone types. In Ham W. E. (Ed.) “Classification of Carbonate Rocks”, 62-84. Am. Ass. Petrol. Geol., Tulsa. Glover. J. E. (1974a). — Petrology of chert artifacts from Devils Lair, Western Australia. J. R. Soc. West. Aust, 57: 51-53. Glover, J. E. (1974b). — Sand fulgurites from Western Australia. J. R. Soc. West. Aust., 57: 97- 104. Glover, J. E. (1975). — New finds of sand fulgurites from the Perth Basin, Western Australia. J. R. Soc. West. Aust. (in press). Glover, J. E. and Cockbain, A. E., (1971). — Transported Aboriginal artefact material, Perth Basin, Western Australia. Nature, Lond., 234 (5331): 545-546. Hallam, S. J. (1972). — An archaeological survey of the Perth area. Western Australia: a progress report on art and artefacts, dates and demography. Aust. Inst. Aboriginal Studies Newsletter, 3 (5): 11-19. Hallam, S. J. (1974). — Excavation in the Orchestra Shell Cave, Wanneroo Western Australia. Part I. Ethnographic and environmental background. Archaeol. and Phys. Anthrop. in Oceania., 9: 66-84. Hutton, J. T., Twidale, C. R., Milnes, A. R., and Rosser, H, (1972). — Composition and genesis of silcretes and silcrete skins from the Beda Valley, Southern Arcoona Plateau, South Australia. J. Geol. Soc. Aust., 19: 31-39. Johnstone, M. H., Lowry, D. C., and Quilty, P. G. (1973). — The geology of southwestern Aus- tralia— a review. J. R. Soc. West. Aust., 56: 5-15. Jones, J. B., and Segnit, E. R. (1971).— The nature of opal I. Nomenclature and constituent phases. J. Geol. Soc. Aust., 18: 57-68. Karajas J. (1969). — A geological investigation of an area between Mt. Minchin and the Bowes River, Northampton District, Western Aus- tralia. Hons. Thesis, Univ. West. Ausr. (unpubl.) Logan, B. W., and Chase, R. L. (1961).— The strati- graphy of the Moora Group, Western Aus- tralia. J. R. Soc. West. Aust., 44: 14-31. McArthur, W. M., and Bettenay, E. (I960).— The devel- opment and distribution of the soils of the Swan Coastal Plain, Western Australia. C.S.I.R.O. Soil Pubis: 16. Mallard, M. E. (1890. — Sur la lussatite, nouvelle variete minerals cristalisee de silice. Bull. Soc. fr. Miner Cristallogr., 13: 63-66. Morner, N.A. (1971). — Eustatic changes during the last 20 000 years and a method of separating the isostatic and eustatic factors in an uplifted area. Palaeogeogr. Palaeoclimatol., Palaeoe- col., 9: 153-181. Pettijohn, F. J. (1957). — “Sedimentary Rocks” (2nd Ed.). Harper, New York. Playford, P. E., & Cope, R. N. (1971).— The Phanerozoic stratigraphy of Western Australia. A corre- lation chart in two parts. Annu. Rep. Min. Dept. West Aust. for 1970, 32-33. Polach, H. A., Golson, J., Lovering, J. F. & Stipp, J. J. (1968). — A.N.U. Radiocarbon Date List II. Koonalda cave series, Nullarbor Plain, South Australia. Radiocarbon, 10, 189-191. Quilty, P. G. (1974a), — Cainozoic stratigraphy in the Perth area J.R. Soc. West. Aust., 57, 16-31. Quilty, P. G. (1974b). — Tertiary stratigraphy of Western Australia. J. Geol. Soc. Aust., 21, 301-318. Rock-color Chart Committee (1963). — “Rock-color Chart”. Distrib. Geol. Soc. Am., New York. Thom. B. G., & Chappell, J. (1975). — Holocene sea levels relative to Australia. Search, 6, 90-93. Williams, H., Turner, F. J. & Gilbert, C. M. (1954). — “Petrography” . Freeman, San Francisco. Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 85 8. — The biology and taxonomy of the cardinalfish, Sphaeramia orbicularis (Pisces; Apogonidae) by Gerald R. Allen^ Manusript received and accepted 29 July 1975 Abstract A Study of the biology of Sphaeramia orbicularis (Cuvier) at the Palau Archipelago, Western Caroline Islands was conducted from November 1971 to March 1972. In order to clarify the taxonomic position of S. orbicularis, a re- view of the genus Sphaeramia is included in which an additional taxon (S. nematoptera) is recognised. The two species differ with regards to ecology, colour pattern, and counts for the gill rakers and soft anal rays. S. orbicularis were usually encountered in aggregations which number from a few to about 30 individuals. They prefer shallow water, usually near the shoreline, in which shade and shelter are pro- vided by rocks, mangrove trees, or man-made constructions. The species is a carnivore which feeds on insects and a variety of small benthic and planktonic animals. Spawning and court- ship activity occurred throughout the study period. Individuals spawned at average intervals which ranged from 19 to 33 days. The males incubate the egg masses orally. The estimated number of eggs per egg mass from three incub- ating males ranged from 6 100 to 11 700. The eggs hatch in approximately eight days at temperatures between 27 and 30°C. The fry are presumably pelagic and make their first appearance inshore at a total length of about 10.0 mm. The average growth rate for postlarval juveniles ranged from 3.3 to 6.4 mm per month. Introduction The present study was conducted while the author was stationed at the Palau Archipelago, Western Caroline Islands as a biologist for the Marine Resources Division of the U.S. Trust Territory Government. The author’s residence (Fig. 1) was built on stilts over the sea, situated on the edge of a tranquil lagoon, bordered by the jungle-covered slopes of Malakal Island. Realis- ing the tremendous potential of this setting I initiated a search for a species of fish which would readily lend itself to an intensive short- term biological investigation. The perfect sub- ject was found in Sphaeramia orbicularis (Cuvier), a small species of cardinalfish (family Apogonidae). There was a permanent colony of approximately 60 individuals living directly under the house. These fish are particularly interesting from a behavioural standpoint as the males exhibit the unusual habit of oral egg incuba- tion, a trait thus far recorded in only three families of marine fishes (Apogonidae, Ariidae, and Opisthognathidae) . The study period ex- tended from November 1971 to March 1972, during which time information was gathered on taxonomy, ecology, behaviour, reproductive biology, and growth. I Department of Fishes, Western Atistralian Museum, Perth. W.A. 6000. Figure 1. — The study area was located under the author’s house at Malakal Island, Palau Archipelago (photo- graphed at high tide). Materials and Methods Approximately two observation sessions per day of about -h hour duration each were con- ducted during the study period. Standard skin- diving equipment (without SCUBA) was em- ployed. Notes were usually recorded on a plastic sheet, but on some occasions they were taken by my wife or son, who sat nearby, as I dictated the observations. The study population was composed of 33 adults (18 males and 15 females) and usually 20 to 31 subadults and juveniles. Recognition of individual fish and the study of their growth was facilitated by clipping the basal elements of cer- tain fin rays. These incisions, if properly exe- cuted, inhibited normal fin regeneration and could be detected throughout the study period. A combination of different clips made it possible to differentiate fish of similar size. The growth subjects were periodically recaptured, measured, and released. Juveniles and small subadults were easily collected with dipnets, but quinaldine (a chemical anaesthetic) was necessary for the capture of larger individuals. Specimens were collected for stomach content analysis with a small Hawaiian-sling multiprong spear. The stomach contents were observed with a binocular dissecting microscope. The number of eggs in the egg masses of several incubating males was esti- mated by teasing the eggs from the mass, then counting the number of eggs which constituted a volume of 1.0 ml. The volume of the entire egg mass was then determined and multiplied by the number of eggs per ml. Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 86 Taxonomy There has been some confusion concerning the synonymy of Sphaeramia orbicularis. For example, Weber and de Beaufort (1929), under their description of Apogon nematopterus, stated “this species is only a variety of A. orbicularis in which some rays of the soft dorsal are filamen- tous. The transverse band on the body is broader than in that species.” Similarly, Munro (1967) suggested that nematopterus individuals were probably females of orbicularis. Fortunately, at Palau I had the opportunity to compare both forms, which were found to be clearly distinct. In order to clarify the systematic position of S. orbicularis a brief review of the genus Sphaera- mia and a key to the species is presented below (see Fraser, 1972, for an analysis of generic characters) . Sphaeramia Fowler and Bean Sphaeramia Fowler and Bean, 1930: 29 (type species, Apogon nematoptera Bleeker, by original designa- tion). KEY TO SPECIES la. Dark bar at middle of body about one scale wide; first few soft dor- sal rays not produced into elon- gate filaments; gill rakers on first arch 24 to 27, soft anal rays 9 S. orbicularis lb. Dark bar at middle of body 3^ to 4 scales wide; first few soft dorsal rays produced into elongate fila- ments; gill rakers on first arch 32 to 37; soft anal rays 10 S. nematoptera Sphaeramia orbicularis (Cuvier) Figs. 2, 3, 5, 6, and 7 Apogon orbicularis Cuvier. 1828: 155 (type locality, Java). Apogon nigromaculatus Hombron and Jacquinot, 1853: 32 (type locality, New Guinea). Diagnosis. — Dorsal rays usually VI-I, 9J (frac- tion indicates last ray is bifurcate): anal rays II, 9; pectoral rays 12; lateral-line scales 26 in- cluding two scales on hypural); gill rakers on first arch 24 to 27; dark bar at middle of body about one scale wide. Colour in alcohol: ground colour of head and body light tan; reddish-brown bar about one scale wide at level of first dorsal spine extending from dorsal fin base to abdomen; portion of head and body anterior to bar with numerous small spots especially concentrated on opercle and interorbital; posterior part of body with num- erous, irregular shaped reddish-brown spots of variable size, those along the middle of the body largest and forming a broken stripe which ter- minates at the base of the tail; first dorsal fin pale reddish-brown with several dark spots on membranes; proximal half of pelvic fins trans- lucent, distal half and spine dark reddish-brown; second dorsal, anal, and caudal fins generally pale, but membranes dusky; pectoral fins pale (slightly reddish) with diffuse reddish-brown spot at base. Colour in life: ground colour of head and body pale grey with yellowish sheen; numerous small reddish-brown spots on snout, inter-orbital, and occipital; breast and abdomen slightly silvery; reddish-brown bar at middle of body; numerous reddish-brown spots and blotches on posterior portion of body; first dorsal spine reddish-brown, except creamy yellow at tip, remainder of first dorsal fin pale yellow with reddish suffusion; second dorsal, anal, pectoral, and caudal fins generally transparent; basal half of pelvic fins pale yellow, outer half dark reddish-brown except actual fin rays and distal tip of mem- brane between pelvic spine and first ray which are pale yellow. Figure 2. — Sphaeramia orbicularis, 55 mm SL, Palau Archipelago (photo by J. E. Randall). Remarks. — The colour patterns of S. orbi- cularis and S. nematoptera are contrasted in Figs. 2 and 3. In addition, differences in counts for the anal fin and gill rakers are presented in Table 1. At Palau these two species are also ecologically separated. S. orbicularis generally occurs in small aggregations in extremely shal- low water (0.1 to one metre) adjacent to the shoreline. It inhabits crevices and caves, and is frequently encountered among mangrove roots or in the vicinity of mangroves, breakwaters, piers, and wreckage (see section on ecology for further habitat information). S. nematoptera, however, is found away from the shoreline in 1.5 to six metres, usually in areas of rich coral growth. In addition, the two species appear to diffffer with regards to maximum size attained. The largest specimen of .S', orbicularis which was collected at Palau was 89 mm standard length (SL), while the largest S. nematoptera measured 61 mm SL. S. orbicularis has been recorded from the East African coast. Andaman Islands, Singapore, Indonesia, New Guinea, Philippine Islands! Hong Kong, Palau Archipelago, Truk, Ponape’ and the Gilbert Islands. Sphaeramia nematoptera (Bleeker) Apogon nematoptera Bleeker, 1856: 35 (type locality Manado, Celebes). Amia 7iematophora Bleeker, 1873-76: pi 313 fie i (type locality, Celebes). Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 87 Table 1 Soft dorsal ray and gill raker counts for species of Sphaeramia Species Dorsal Hays j 1 Gill Rakers 9 10 25 .6 27 28 29 30 31 32 33 34 35 36 37 S. orbicularis 10 .... 1 1 7 1 .... .... j .... S. nematoptera .... 10 .... .... j .... I .... 1 3 2 3 .... j 1 Figure 3. — Sphaeramia nematoptera, 42 mm SL, Palau Archipelago (photo by J. E. Randall). Diagnosis . — Dorsal rays VI-I, 9-^-; anal rays II, 10; pectoral rays 12; lateral-line scales 26 (in- cluding two scales on hypural); gill rakers on first arch 32 to 37; dark bar at middle of body 3-2- to 4 scales wide. Colour in alcohol: ground colour of head and body light tan to greyish; dark reddish-brown bar 3^- to 4 scales wide at level of first dorsal fin extending from dorsal fin base to abdomen; portion of head and body anterior to bar mostly pale without spots (scales may have dusky edges): interorbital, cheek, and chin brownish; yellowish patch sometimes visible on upper por- tion of opercle; posterior part of body with about 25 to 30 round, reddish-brown spots of nearly uniform size (slightly smaller than pupil) ; first dorsal and pelvic fins dark reddish-brown; second dorsal, anal, and caudal fins generally pale, but membranes dusky; pectoral fins pale (slightly reddish); pectoral base pale. Colour in life; head and anterior portion of body with yellowish sheen, peppered with numerous minute dark spots; reddish-brown bar at middle of body; posterior portion of body whitish to pink with maroon spots; first dorsal fin mostly reddish-brown; but membranes be- tween last two spines whitish; second dorsal, anal, pectoral, and caudal fins generally trans- lucent; pelvic fins reddish-brown suffused with yellow (yellowish suffusion more pronounced on basal portion of fin), edge of fin with narrow whitish margin. Remarks . — This species appears to be re- stricted to the region which includes Indonesia, New Guinea, Philippine Islands, and Palau Archipelago. Ecology of S. orbicularis at the Palau Archipelago Habitat . — The Palau Archipelago is situated in the southwestern corner of the north Pacific Ocean, about 960 km east of the Philippine Islands and approximately the same distance north of New Guinea. The archipelago represents the extreme western end of the Caroline Islands and extends northeastward from about 6°53'N to about 8°06'N or over 112 km at a longitude of about 134°29'E. With the exception of two oceanic atolls at the extreme north tip of the chain, the northern half of the archipelago is volcanic in origin and is dominated by the island of Babelthaup which is about 40 km long and 13 km wide. This island rises to an elevation of about 200 metres and most of the shoreline is bordered by dense mangroves. The southern por- tion of the archipelago is remarkably scenic, characterised by a bewildering maze of limestone ridges and conical islets which extends for approximately 48 km. The main portion of the archipelago is surrounded by a barrier reef and there is a fairly well-developed lagoon in the southwest sector. The limestone islands and islets of the south, including Malakal Island, rise up almost ver- tically from the ocean floor, forming a narrow complexity of canals some with depths greater than 30 metres. The islands are deeply undercut at the high-tide line, as much as two or three metres, forming deep notches above a submarine shelf of variable width. Sphaeramia orbicularis is usually encountered in aggregations which number from a few to about 30 individuals in shallow, sometimes brack- ish water. Along the cosat of Babelthaup and to a lesser extent among the southern islands, it is found living among the submerged roots of mangrove trees. The preferred habitat among the limestone islands of the south appears tcD be the shallow submerged shelf immediately adjac- ent to the undercut shoreline. This area abounds with shady crevices and small caves which serve as diurnal retreats for the species. The shore- line is occasionally penetrated by relatively large caves, some of which are completely sub- merged. The fish are generally found around the Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 88 mouth of these or a short distance inside, but they rarely penetrate the inner depths which are perpetually dark. S. orbicularis is also found in the vicinity of piers, stone breakwaters, bridges, and wreckage. The substratum in these areas usually consists of mud or rock and is above the zone of live coral growth. Juveniles are sometimes found at the surface around piers, boat moorings, or floating debris. In summary, the basic habitat requirements appear to consist of shallow water, usually near the shoreline, in which shade and shelter are provided by rocks, mangrove trees, or man-made constructions. Because of the shallow depth and proximity to the land the habitat is subject to rather extreme fluctuations with regards to water depth and salinity. The daily tidal fluctuation at Palau is approximately 2.0 metres and subsequently S. orbicularis must make nearly constant adjust- ments to achieve optimum conditions of depth and shelter. Heavy rain showers are relatively frequent throughout the year. After a night of continuous heavy rain the salinity was measured with a refractometer at 21 ppt. This measure- ment was taken in the study area in 20 cm depth or at about the same level inhabited by the fish. The study area proper (Fig. 1), which was located next to the boat-yard on Malakal Island, consisted of a rectangular pool-like enclosure with concrete walls on three sides. The pool had an area of approximately 24 square metres and was sometimes used for mooring small boats. It was situated directly under our residence, thus the floor of the house formed a roof which com- pletely covered the area about 2.0 metres above the water surface at high tide. Water depth in the enclosure varied from only a few centimetres at low water to about 2.0 metres. Food and Feeding , — The daily activity pattern of the study population can be summarised as follows: during most of the daylight hours (0900-1630) the fish were relatively inactive and remained in several aggregations directly under the house. They changed their position during the day to accomodate to changing conditions of tide and light, usually remaining in the shade, hovering at depths ranging from about 20 to 120 cm. During late afternoon the fish became more active and began to feed. It was not de- termined if feeding continued throughout the night, but on every occasion when nocturnal observations were made, the fish appeared to be active, ranging up to 20 metres from the house, usually in small groups of about two to 10 individuals. S. orbicularis is a carnivore which feeds on insects and a variety of small benthic and plank- tonic animals. Crabs (mainly portunids and small grapsids and insects from a major portion of the ingested food items. The stomachs of 61 specimens (45 to 85 mm SL) were examined. Most of these were collected in the vicinity of the house, a short distance from the study popula- tion. Twenty-five stomachs were empty; how- ever, seven of these were from incubating males, which do not feed during the brooding period. The results of the stomach content analysis of the 36 specimens containing food is presented in Table 2. The results are given as percentage Table 2 Percentage volume of major groups of food organisms in stomachs of Sphaeramia orbicularis (10 stations ; 36 specimens — 45 to 82 mm SL) Food Volume {%) Crabs 28-8 Insects 22-9 Copcpods .... .... .... .... .... 12-9 Sphaeramia eggs .... .... .... .... . .. .... 5-6 Ostracods 5-0 Polychaetes .... .... .... .... .... .... 4-7 Amphipods .... .... .... .... .... .... 3-9 Zoea 3-7 IJnidentifted .... .... .... .... .... .... 3-4 Sergestids (Lucifer) 31 Unidentified crustacean fragments .... .... .... 1-4 Megalops .... .... .... .... .... .... .... 1-2 Small fishes 1-0 Mysids 0-8 Unidentified shrimps 0-8 Stoinatopod larvae 0-4 Pleagic fish eggs .... .... .... .... .. . .... 0-3 Gastropod fragments .... .... .... . .... 01 volume of the different major groups of food organisms. The percentages were estimated vis- ually for individual fish, and the total was com- puted from all the stations. The stomachs which were full usually contained either one or two small crabs, a small insect, or a variety of minute planktonic organisms. Specimens collected be- tween 1700 and 1900 hours generally contained a high percentage of grapsid crabs and insects (beetles and crickets), while individuals taken later in the evening or during the predawn hours appeared to feed predominately on plankton. Most of the specimens collected during the mid- dle of the day (1100 to 1430) had empty stomachs. The data indicates the fish feed actively at dusk, during the early evening and predawn hours (0500-0630). Unfortunately samples were not taken from 2230 to 0500. S. orbicularis is probably opportunistic as far as food items and time of feeding are concerned. For example, fish in the vicinity of the house actively fed on plankton which was attracted to the household lights during the evening, while a collection made at the same time about one km away in total darkness indicated that crabs were the primary food taken. Resting aggregations exhibiting a minimum of feeding activity could be readily induced to feed on plankton by switching on a 75 watt lightbulb above the study area. Similarly, inactive fish during the middle of the day will eagerly take bits of bread and assorted table scraps. As mentioned previously the fish feed actively at dusk. However, on one occasion, a collection of nine individuals made at this time of day at the edge of a landlocked (except for a narrow submarine cave to the sea) saline lake yielded only empty stomachs. It is interesting to note that insects are a major food item. This fact is clearly understand- able if the habitat is considered. With the ex- ception of fish living in the vicinity of man-made structures the preferred habitat of the species is either adjacent to the undercut shoreline or among mangrove roots. In both localities, the vegatative canopy overhangs the habitat of the fish. In areas of undercut shoreline, the lush growth of jungle stops at the high tide mark, but because of the undercutting phoenomena, Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 89 the vegetation extends a considerable distance out of the water, and at high tide the foliage is frequently adjacent to the sea surface. Thus, insects which fall from the trees or accidentally fly into the water are readily available as food. Reproduction and early development Courtship and spawning took place throughout the study period. Presumed courtship was observed around sunset on several occasions. During this activity pairs of S. orbicularis en- gaged in alternating patterns of chase and dis- play. The most common display exhibited con- sisted of one fish confronting another in either parallel or lateral fashion while flicking the pelvic and first dorsal fin back and forth. The tips of these fins are relatively bright coloured which serves to enhance the display. Occasion- ally the fish chased each other for a distance of several cm and sometimes biting was observed, this activity usually being directed at either the sides or anal region. There was no evidence of sexual dimorphism except for the swollen mouth of egg brooding males (Fig. 4). Spawning was not observed and most likely takes place during the middle of the night. Figure 4. — Male S. orbicularis with egg mass in oral cavity. Spawning occurred at more or less regular intervals with individuals spawning about once or twice monthly. Daily records were kept of the number of incubating males present, which ser- ved as an index of spawning activity. The graph in Fig. 5 represents the typical pattern which was observed during the study period. There appeared to be a correlation between the time of spawning and the phase of the moon. Spawning activity was at its highest peak each month at the period between first-quarter and full moon. Also there was a lesser peak between last- quarter and new moon. It is possible that the fish spawn at these times because of the tidal levels. During both of the monthly spawning peaks the high tide occurred within to two hours of midnight. If spawning occurred during these hours, as it suspected, there may be a requisite for high water. At low tide the fish are more or less driven temporarily away from their favoured habitat and might be reluctant to spawn. The spawning record of several males as indicated by the presence of an egg mass in the oral cavity is indicated in Table 3. The spawn- ing interval for these individuals ranged from 16 to 35 days with an average interval of 19 to 33 days between broods. Table 3 Spawning Record of several male Sphaeramia orbicularis Size (nmi SL) Spawning Dates Average Interval (Days) First Second Third 69 1/12/71 24/12/71 25/1/72 27 69 1/12/71 23/12/71 22/1/72 26 72 27/12/71 12/1/72 5/2/72 20 73 2/12/71 6/1/72 5/2/72 33 74 27/12/71 20/1/72 7/2/72 21 76 1/12/71 23/12/71 8/1/72 19 80 1/12/71 23/12/71 10/1/72 20 The spawning of this species is probably sim- ilar to that exhibited by Apogon imberbis (Lin- naeus), which was observed by Garnaud (1950a and b). According to Breder and Rosen (1966), “Garnaud described the mating as a side-to-side affair in which the female placed her ventral under the male and he placed his anal under the female. In this position they performed t>ie common teleost trembling movements. Accord- ing to Garnaud, such behaviour did not take place when the eggs were finally extruded later. These eggs are cast in a single mass of material bound together by tendrils arising from one pole of the eggs. The male immediately takes them in in his mouth. These observations led Garnaud to postulate that this case was one of intei*nal fertilisation.” However, there is no experimental evidence to prove that internal fertilisation takes place in apogonids. Many authors have stated that the male apogonid always incubates the eggs. However, Ebina (1932) reported that both sexes of Apogon semilineatus exhibit this habit. In the present study an examination of the gonads of many specimens indicated that only male S. orbicularis are oral incubators. Males which are incubating egg masses are easy for an observer to distinguish. The gular region is greatly enlarged which dramatically alters the lower profile of the head (Fig. 4). The incubation period lasts about eight days during which time the male period- ically “juggles” or changes position of the mass in its mouth. As mentioned previously, the incub- ating males do not feed; however, Sphaeramia eggs are occasionally found in the stomachs of incubating males. Presumably, these are acci- dentally swallowed or, as in one case when the Journal of the Royal Society of Western Australia. Vol. 58. Part 3. November, 1975. 90 12-t (/) o H < 00 3 O o QO UJ 00 s 3 10 - 6- J r n IT r -1 L T 1 r 1 \ 1 1 1 « 1 C 03 «e 03 #C o NOVEMBER DECEMBER - 1971 JANUARY - 1972 MOON PHASE AND MONTH Figure 5. — Number of incubating males of 5. orbicularis with relation to moon phase and time of month. entire egg mass was found in the stomach, are ingested during periods of stress (specimen was captured with guinaldine). The egg masses are roughly spherical, although somewhat com- pressed, ranging in diameter from 14 to 20 mm. The estimated number of eggs per egg mass from three incubating males 72 to 89 mm SL, ranged from 6 100 to 11 700. The eggs are pinkish- orange when freshly spawned and as the embryos develop they gradually turn to purplish-brown. The smallest incubating male collected during the study was 69 mm SL; the smallest ripe female was 60 mm SL. The ovaries are elongate, compressed structures which are encapsulated in a silvery coloured sheath. One ovary of a 78 mm SL specimen was 11mm x 26 mm and 7 mm thick. Individual eggs are spherical and measure 0.6 to 0.7 mm in diameter. By approximately 24 hours the embryo has entered the blastula stage (Fig, 6a) and at approximately 40 hours gas- tulation is in progress (Fig. 6b) . Up to this point the egg is characterised by the presence of two to five large oil globules. After about 70 hours the embryo is relatively well formed (Fig. 6c). The unpigmented eyes, otic vesicles, statoliths, and somites are clearly evident. The yolk is bilobed and there is a single large oil globule present. The heart which is positioned just under the head, pulsates at a rate of 120 to 140 beats per minute. The 110 hour embryo (Fig. 6d) is similar to the previous stage except the yolk is further reduced, the eyes exhibit pigmentation, and the median fin folds have made their appearance. By about 160 hours (Fig. 6e) the pectoral fins are evident, the eyes are well formed and pigmentation is apparent in the form of two rows of stellar-like melanophores on the ventral surface. A full term embryo is shown in Fig. 6f. Hatching occurs in approximately eight days at temperatures rang- ing from 27 to 30°C. The newly hatched, transparent fry (Fig. 7a) are apparently pelagic for at least a short period. The pelagic larval stage accounts for the wide geographic distribution of the species. The smallest post-larval forms which were collected inshore measured about 10.0 mm total length Figure 7. — Growth stages of S. orbicularis', a. newly hatched larva, 3.3 mm total length; b. postlarval juven- ile. 10.0 mm total length. Journal of the Royal Society of Western Australia. Vol. 58, Part 3, November, 1975. 91 Figure 6. — Early development of S. orbicularis: a. 24 hoxirs; b. 40 hours; c. 70 hours; d. 110 hours; e. 160 hours. Table 4 Summary of growth data for juvenile Sphaeramia orbicularis Initial .Measurement Size Date FinalMeasiirement Size Date Total Increase Growtli Rate per month (mm SL) (mm SL) (mm) (mm) 12-0 6/12/71 22-0 29/2/72 100 :3-6 120 6/12/71 2:3 0 29/2/72 11-0 40 12 0 6/12/71 28-0 29/2/72 16-0 5-8 i:50 8/11/71 :32 • 0 29/2/72 19-0 51 1:M) 6/12/71 22-0 29/2/72 9-0 :3-:3 i;do 6/12/71 :30 • 0 29/2/72 170 6-2 1:D5 6/12/71 240 29/2/72 10-4 8-8 1:D5 6/12/71 28-0 29/2/72 14-5 5-8 l:3-5 6/12/71 28 • 0 29/2/72 14-5 5-8 140 6/12/71 •M) ■ 0 29/2/72 16-0 5 • 8 140 6/12/71 29 • 0 29/2/72 15-0 5-5 14-5 6/12/71 29 • 0 29/2/72 14-5 5-8 15-0 6/12/71 :30 0 29/2/72 15-0 5 • 5 15-0 6/12/71 :30 0 29/2/72 150 5-5 16 0 6/12/71 :30 0 29/2/72 14-0 5-1 ISO 11/8/71 :36 • 0 29/2/72 18-0 4-8 18-0 11/8/71 41-0 29/2/72 21-5 5-7 19-5 11/8/71 42-0 29/2/72 24 0 6-4 (Fig. 7b). These tend to form small aggregations of about five to 20 individuals. They are found in the same habitat as the adults and feed on current-borne plankton. Growth data were collected for 18 juveniles over a four month period. These data are sum- marised in Table 4. The average growth rate for juveniles ranged from 3.3 to 6.4 mm per month. There was no significant growth recorded for 17 fin-clipped adults, ranging in size from 70 to 89 mm SL, during the four month study period. Acknowledgments . — I would like to express apprecia- tion to my wife, Connie, and son, Tony, for their assistance in recording field notes. Thanks are also due Dr. James P. McVey and Dr. Walter A. Starck II for the loan of equipment. Dr. John E. Randall provided black and white photographs (Figs. 2 and 3). References Bleeker, P. (1856). — Beschrijvingen van nieuwe of weinig bekende vischsoorten van Menado en Makassar grootendeels verzameld op eene reis naar den Mclukschen Archipel in het gevolg van den Gouverneur-Generaal Duy- maer van Twist: Acta Soc. Sci. Indo~Neerl., 1: 1-80. (1873-76 ). — Atlas ichthyologique des Indes Orientales Neerlandaises, public sous les auspices du Gouvernement colonial neer- landais. 7: pi. 279-320. Breder, C. M. Jr. and Rosen, D. E. (1966).— “iWodes of Reproduction in Fishes’'. T.F.H. Publications, Inc., Neptune, New Jersey, 941 pp. Cuvier, G. L. (1828).— (7n Cuvier and Valenciennes)— ‘'Histoire naturelle des Poissons”. Paris. F. G. Levrault, 2: 1-490. Ebina, K. (1932).— Buccal incubation in the two sexes of a percoid fish, Apogon semilineatus T. and S. Journ. Imp. Fish. Inst., Tokyo, 27 (1): 19-21. Fowler, H. W. and Bean, B. A. (1930).— Contributions to the biology of the Philippine Archipelago and adjacent regions. The fishes of the families Amiidae, Chandidae, Duleidae, and Serranidae, obtained by the United States Bureau of Fisheries Steamer “Albatross” in 1907 and 1910, chiefly in the Philippine Islands and adjacent seas. Bull. U. S. Nat. Mus., 100 (10) : 1-334. Fraser, T. H. (1972). — Comparative osteology of the shal- low water cardinalfishes (Perciformes: Apogonidae) with reference to the system- atics and evolution of the family. Ichth. Bull. (J. L. B. Smith Inst, of Ichthyology) No. 34: 105 pp., 44 pis. Garnaud, J. (1950a).— Notes partielles sur la reproduc- tion d'Apogon imberis. La terre et la Vis. Paris, No. 1: 39-42. (1950b). — La reproduction et I’incubation branchiale chex Apogon imberis G. et L. Bull. I’Inst. Oceanogr. Monaco. 47 (977): 1-10. Hombron, M. and Jacquinot, M. H. (1853 ). — Voyage au Pole Sud et dans VOceanie sur les corvettes L’ Astrolabe et La Zelee . . . Zoologie, vol. 3, Reptiles and Poissons, Paris, Gide et J. Baudry: 1-56. Munro, I. S. R. {1961) .—‘‘The fishes of New Guinea”. Dept, of Agriculture, Stock and Fisheries, Port Moresby, New Guinea: 1-650. Weber, M. and de Beaufort, L. F. ( 1929) .— T/re fishes of the I ndo- Australian Archipelago. Vol. 5, Anacanthini, Allotriognathi, Heterosomata, Berycomorphi: families: Kuhliidae, Apogoni- dae, Plesiopidae, Pseudoplesiopidae, Priacan- thidae, Centropomidae. Leiden. E. J. Brill: 1-508. Journal of the Royal Society of Western Australia. Vol. 58. Part 3, November, 1975. 92 9.— Eremophila ramiflora (Myoporaceae), a new species from Western Australia by B. Deir Manuscript received and accepted 29 July 1975 Abstract A new species of Eremophila, E. ramiflora is described. This species has affinities with Eremophila fraseri and is sympatric with E. fraseri in the Agnew area of the Northern Goldfields, Western Australia. Introduction While carrying out resin analyses on plants of Eremophila fraseri F. Muell. in the field the author’s attention was drawn to a previously undescribed species of Eremophila. From a dis- tance the plants of the two species can be readily confused. Apart from material collected by the author only two other collections have been made. Although there appears to be a need for extensive taxonomic work in the genus, publica- tion of the description of this species is justified at this time because its boundaries appear well defined. 1 Botany Department, University of Western Australia, Nedlands, Western Australia 6009. Eremophila ramiflora Dell, sp. nov. Species habitu cum E. frasero optime congruens, sed diflert petiolis subdecurrentibus; fioribus in caulibus et novis et veteribus portatis; sepalis dorsalibus sepalis internis vix latioribus; sepalis post anthesin vix dilatatis; corolla magentea; lobo ventral! corollae longitudine tubum aequanti differt. Compact shrub, 1-3 m high, vegetative parts glandular hairy, resinous; branchlets thick, rigid, with amber-coloured resin droplets, with remnants of deflexed leaf bases ca. 0.5 cm long. Leaves lanceolate, alternate, crowded, spreading, deflexed when mature, 1.2-2.2cm broad, up to 7.5 cm long, slightly keeled and recurved; blades viscid, green, gradually attenuate or cuneate into the petioles which are difficult to distin- guish from the lamina; petioles almost decur- rent; margins entire, undulate or repand, rarely serrulate: apex acute. Flowers axillary, usually solitary or in small groups, borne on previous season’s branchlets and on leafy young branch- lets; peduncles 0.7-1. 2 cm long, spreading slightly turned up under the flower; calyx segments Journal of the Royal Society of Western Australia, Vol. 58. Part 3, November, 1975. 93 slightly imbricate when mature, oblong to spathulate, acute to obtuse, ca. 1.2 cm long when in flower, enlarging to 1.5 cm, rarely to 2.0 cm; sepals green becoming reddish and shiny after flowering, then reticulate and membranous, dor- sal sepal scarcely broader than lateral sepals: corolla 2. 5-3.0 cm long, glandular-hairy with some simple hairs inside towards the base; base yellowish-green; tube dark magenta with small dark spots on the outside, cylindrical, scarcely constricted above the ovary, the upper part linear-oblong, slightly curved and not much dilated: lobes unequal, the 4 upper lobes turned back and acute, dorsal lobes short, lateral lobes separated to almost 1/3 the length of the corolla, Figure 2. — Close up of stem showing flowers on new and old wood. ventral lobe long, recurved separated to the mid- dle of the corolla, apex acute and keeled; stamens exserted, filaments glandular pubescent, anthers with magenta lateral papillae, with few glandular hairs; pistil just exserted, style sparse- ly hairy towards stigma, stigma red; ovary glabrous, green towards the apex, 2- chambered with one pair of ovules to each cell. Drupe woody, slightly compressed, shorter than the calyx, 0.6-0. 8 cm broad, 0.8-1. 0 cm high. Gametic chromosome number n=18. Holotpye . — 127.5 km south of Yandal Home- stead, 47 km north of Leonora on road to Mel- rose Station, Western Australia, 28 September 1973, Dell 1050 < PERTH). Other Specimens. — 25 km west of Carnegie Homestead, 122°45'E, 25°45'S, Chinnock 893 (AD 97347300); 127 km east of Sandstone, Dell 1029; 16, 21, 28.2, 48, 67.1 km north of Leonora on road to Wiluna, Dell 1074, 1076, 1077, 1072, 1075 (UWA); between Agnew and Leonora, Lovett (PERTH). Discussion Characters of Eremophila ramifiora which distinguish it from E. fraseri are the almost de- current broad leaf bases, the flowers arising from both the old and new seasons stems, the sepals not enlarging much after flowering, the dorsal sepal scarcely broader than the internal sepals, and the magenta corolla tube with ven- tral lobe separated to the middle of the corolla. The resin components of the leaves and stems also differ from those of E. fraseri. Figures 1, 2 and 3 show details of habit, flowering shoot and floral characters respectively. E. ramifiora occurs in the Leonora — Agnew region of the Eremean Province extending north-east as far as Lake Carnegie. There are insufficient collections to determine how far east the species extends. Both E. fraseri and E. rami- fiora occur on similar red loams with many stones overlying lateritic hardpans. Near Agnew the species are sympatric, E. fraseri being repre- sented as the tetraploid. The nearest diploid population of E. fraseri is some 400 km further north. E. ramifiora has been seen in flower in July and September. It is likely that this species flowers after heavy rains and like E. fraseri blooms twice in some years. Acknowledgements . — The author is grateful to Mr. B. G. Lay for information on the soils of the Carnegie locality, to Mr. G. J. Keighery for providing the chromo- some count of the type material, and to Dr. N. H. Brittan for providing the Latin diagnosis. References Mueller, F. J. H. ( 1878) .—Frapwenta phytographiae Australiae. xi, 51. Melbourne. Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 94 Figure 3.— Details of floral structure. (A.— developing buds, B, C.— mature flower, D.— sepals at flower fall E — corolla tube cut and spread, F.— fruit with strongly veined calyx segments, G.— variation in calyx segments H. — mature leaves. All drawings to same scale; length of bar 1cm). ’ Journal of the Royal Society of Western Australia, Vol. 58, Part 3, November, 1975. 95 INSTRUCTIONS TO AUTHORS Contributions to this Journal should be sent to The Honorary Editor, Royal Society of Western Australia, Western Australian Museum, Perth. Papers are received only from or by communication through, Members of the Society. The Council decides whether any contribution will be accepted for publication. All papers accepted must be read either in full or in abstract or be tabled at an ordinary meeting before publication. Papers should be accompanied by a table of contents, on a separate sheet, showing clearly the status of all headings; this will not necessarily be published. Authors should maintain a proper balance between length and substance, and papers longer than 10,000 words would need to be of exceptional importance to be considered for publication. The Abstract (which will probably be read more than any other part of the paper) should not be an expanded title, but should include the main substance of the paper in a con- densed form. Typescripts should be double-spaced on opaque white foolscap paper; the original and one carbon copy should be sent. All Tables, and captions for Figures, should be on separate sheets. Authors are advised to use recent issues of the Journal as a guide to the general format of their papers, includ- ing the preparation of references; journal titles in references may be given in full or may follow any widely used conventional system of abbreviation. Note that all illustrations are Figures, which are numbered in a single sequence. In composite Figures, made up of several photographs or diagrams, each of these should be designated by letter (e.g. Figure 13B). Illustrations should include all necessary lettering, and must be suitable for direct photo- graphic reproduction. To avoid unnecessary handling of the original illus- trations, which are usually best prepared between 14 and 2 times the required size, authors are advised to supply extra prints already reduced. Additional printing costs, such as those for folding maps or colour blocks, will normally be charged to authors. It is the responsibility of authors to adhere to the International Rules of Botanical and Zoological Nomenclature. Palaeontological papers must follow the appropriate rules for zoology or botany, and all new stratigraphic names must have been previously approved by the Stratigraphic Nomenclature Com- mittee of the Geological Society of Australia. Thirty reprints are supplied to authors free of charge, up to a maximum of 60 for any one paper. Further reprints may be ordered at cost, provided that orders are submitted with the returned galley proofs. Authors are solely responsible for the accuracy of all information in their papers, and for any opinion they express. Journal of the Royal Society of Western Australia Volume 58 1975 Part 3 Contents 6. Nitrogen oxide levels in suburbs of Perth, Western Australia. By G. A. Bottomley and F. C. Cattell. 7. The petrology and probable stratigraphic significance of Aboriginal arti- facts from part of south-western Australia. By J. E. Glover. 8. The biology and taxonomy of the cardinalfish, Sphaeramia orbicularis (Pisces; Apogonidae). By G. R. Allen. 9. Eremophila ramiflora (Myoporaceae), a new species from Western Aus- tralia. By B. Dell. Editor: A. J. McComb The Royal Society of Western Australia, Western Australian Museum, Perth 49536/7/75—625 WILLIAM C. BROWN, Government Printer, Western Australia JOURNAL OF THE ROYAL SOCIETY OF WESTERN AUSTRALIA VOLUME 58 PART 4 DECEMBER, 1975 PRICE TWO DOLLARS REGISTERED FOR POSTING AS A PERIODICAL-CATEGORY B THE President Vice Presidents Past President Joint Hon. Secretaries Hon. Treasurer Hon. Librarian Hon. Editor ROYAL SOCIETY OF WESTERN AUSTRALIA PATRON Her Majesty the Queen COUNCIL 1975-1976 B. E. Balme, D.Sc. P. R. Wycherley, O.B.E., B.Sc., Ph.D., F.L.S. A. J. McComb, M.Sc., Ph.D. .. G. A. Bottomley, B.Sc., Ph.D. M. Perry, B.Sc. (Agric.) (Hons.) G. Perry, B.Sc. (Hons.) S. J. Curry, M.A. .. A. Neumann, B.A. .... A. J. McComb, M.Sc., Ph.D. P. Atkinson, B.Sc. C. E. Dortch, B.S., M.Phil. B. B. Lament, B.Sc. (Hons.), Ph.D. L. J. Peet, B.Sc., F.G.S. P. E. Playford, B.Sc., Ph.D. J. C. Taylor, B.Sc., Ph.D., A.R.C.S. A. F. Trendall, B.Sc., Ph.D., A.R.C.S., F.G.S. P. G. Quilty, B.Sc. (Hons.), Ph.D. 10. — Mammal remains from the upper levels of a late Pleistocene deposit in Devil’s Lair, Western Australia by A. Baynes/ D. Merrilees^ and Jennifer K. Porter^ Manuscript received 22 October 1974; accepted 29 July 1975 Abstract This paper reports on continuing investiga- tions of the fossil mammal fauna from the sandy deposit in Devil’s Lair, a small cave in the extreme south west of Western Australia. Excavations in 1972 and 1973 produced a sub- stantial sample of bone fragments about 12 000 to about 19 000 years old, and a much smaller sample about 19 000 to about 25 000 years old. Most of these remains appear to be of animals eaten by humans occupying the cave intermittently, perhaps in small groups. Neither Dingo nor any of the large extinct Pleistocene marsupials is included among the 35 mammal species so far detected in the deposit. Between about 19 000 years ago and about 12 000 years ago most species are present at all levels, but at least one trend in the rela- tive abundance of the more common species is suggested, and seems to be clearest about 12 000 years ago. An improved list of the mod- ern mammal fauna of the district is presented and compared with the fossil fauna; there appears to have been a considerable decrease in species diversity since the Pleistocene. Species lost have mainly been those adapted to shrub formations or woodlands, whereas species adapted to forest have persisted. These changes are the culmination of the above trend in relative abundance. Two possible causes are discussed; the effects of the glacioeustatic rise in sea level acting alone by changing the extents of different habitats, or acting in con- junction with local climatic changes such as an increase in effective rainfall. Contents Page Introduction .... .... .... .... .... .... .... 97 The 1972 and 1973 excavations 98 Nature and age of deposit 98 Procedures 99 Stratigraphic reliability of the sample 101 Source of the bone sample 102 Bias in the sample 103 The bone sample as an index of human use of the cave 104 The fossil mammal fauna of Devil’s Lair (tabu- lated in Appendix 1) 105 The modern mammal fauna of the Devil’s Lair district .... .... .... .... .... .... 105 Comparison of modern and fossil mammal faunas from Devil’s Lair 107 Modern distribution patterns and habitat adapta- tions of mammal species 108 Changes in the mammal fauna around Devil’s Lair, and interpretation 110 Conclusions 114 References 115 Appendix 1 — details of stratigraphic distribution of mammal species and other vertebrate taxa 118 Appendix 2 — records forming the basis of the modern mammal fauna of the Devil’s Lair district by A. Baynes 124 Appendix 3 — Investigation of degree of overestima- tion inherent in our methods of obtaining “minimum numbers’’ of individuals 125 1 Western Australian Museum, Francis St.. Perth, 6000. Introduction Kxcavations in Devil’s Lair, a small cave in aeolian calcarenite in the extreme south west of Western Australia, were begun in 1955 by E. L, Lundelius with the object of learning something of the prehistoric mammal fauna of the region. After various unsystematic extensions of these initial excavations, and after realisation that the deposit included an archaeological component, the Western Australian Museum began a series of systematic excavations, to be spread over a number of years. In preparation for this series, a reserve including the cave has been vested in the Museum, and a steel mesh fence erected across the cave mouth. Thus it seems justifiable to leave excavations open from field season to field season and to extend them as opportunity offers. The deposit yields vertebrate remains in abundance as well as substantial numbers of artifacts, so that a few excavators working for some days produce material requiring many months of preparation and study. The intention is to report progress at intervals, leading to a summary in due course. Present indications are that several more years’ work must precede any such summary. This paper presents an analysis of animal re- mains recovered from excavations made in Feb- ruary 1972, and March 1973. A general report on the first of the series of systematic excava- tions, made in December 1970, has been issued by Dortch and Merrilees (1972), and a report dealing mainly with the archaeological compon- ent from the 1972 excavations has been issued by Dortch and Merrilees (1973). Dortch (1974) has discussed archaeological aspects of the 1973 excavations, and Glover (1974) has discussed geological aspects of the material of some of the articrafts recovered. Davies <1968, and Appendix to Dortch and Merrilees, 1973) has described human incisor teeth. Other specialist reports are in preparation. We follow the mammal species concepts and names of Ride (1970), Devil’s Lair lies about 5 km from the sea on the eastern side of a ridge. The vegetation of the region has been described by Smith (1973). The western slope of the ridge, which is exposed to the prevailing westerly winds off the southern Indian Ocean, is covered by an open heath in which Acacia decipiens is the principal species. Near the sea Olearia axillaris and Scaevola spp, are also important components. Where some shelter exists on the top of the ridge open scrub, Journal of the Royal Society of Western Australia, Vol. 58, Part 4. December, 1975. 97 low woodland, or low open forest may occur. Peppermint iAgonis flexuosa) or Jarrah {Eucal- yptus marginata) are usually the principal species in these formations. On the eastern slopes of the higher parts of the ridge open for- est quickly gives way to high open forest in which Karri (E. diversicolor) grows in fairly pure stands. DeviFs Lair lies in the middle of one such belt of Karri forest which stretches for some 20 km from Calgardup Brook in the north to Turner Brook in the south. On the inland (eastern) side of the Karri is an extensive open forest of Jarrah and Marri iE. calophylla) con- tinuing in to the valley of the Blackwood River some 10 km to the east, and for a much greater distance to the north east. Only about 3 km to the east of Devil’s Lair is McLeod Creek along which dense shrub vegetation grows. The 1972 and 1973 Excavations Several trenches were opened in Devil’s Lair during the 1972 and 1973 field seasons, for rea- sons given by Dortch and Merrilees (1973), who show the positions of these trenches in the cave and briefly describe the geological setting. Very few animal remains were recovered from Trenches 3 and 4. This paper is based almost entirely on specimens recovered from Trench 6 and from a complex of trenches, Nos 2, 5, 7 and 8 (7 and 8 with subdivisions) of our field and laboratory records. The relations of Trenches 2, 5, 7 and 8 to one another and their dimensions are shown in Figure 1. Trench 6, oriented similarly to Trench 5 but Ih ni long by 1 m wide, was close to the cave wall about 3 m south west and slightly down slope from Trench 5. All depths have been re- corded below the same arbitrary datum mark on the wall of the cave, and thus may be com- pared directly. The various trenches have been excavated to different depths, which are recorded in the tables in Appendix 1. Trenches 7 and 8 have been excavated only to shallow depths as yet, Trenches 5 and 6 to intermediate depths, while Trench 2 is deepest, but no trench has yet reached the bottom of the deposit, and steel rods driven down from the bottoms of Trenches 6 and 2 show at least a metre thickness of depo- sit below each. Nature and age of the deposit Much of the Devil’s Lair deposit is made up of grains of quartz to some extent coated and lightly cemented by calcite. As a result excava- tion is usually easy, yet vertical excavation walls remain stable for years. The colour of the deposit varies. The sandy portions show various shades of orange and brown, but there are paler col- oured bands and irregular masses which have been more or less strongly lithified by calcar- eous cement, while the uppermost layer is black. Flowstones, stalagmites and other almost en- tirely calcareous materials, crystalline at the macroscopic level, are interbedded with the sandy material. Calcareous solutions drip from the roof of the cave and permeate the deposit during the wetter months of the year at present, and presumably always or usually have done so, and have been responsible for patches and zones of lithification. During the drier months of the year, the sandy portions of the deposit become sufficiently dry to pass easily through our screens, but enough moisture is present at some times of the year to support the growth of tree roots. These appear to be associated in some way not at present understood with bands and pockets made up entirely or partly of gypsum. There are lenticular inclusions of ashy ap- pearance, which we have interpreted as hearths, particularly near the top of the deposit. Char- coal is widely dispersed, usually in small pieces not exceeding a few millimetres across. Frag- mented bone is also widely dispersed through the deposit, and there is a small but significant proportion of rock fragments foreign to the cave. Large and small rock fragments, mainly of cal- crete, much or all of which could be derived within the cave, are abundant in places, and broken stalactites also occur. Detailed analysis of the sediments is in pro- gress, and will be reported in due course by M. L. Shackley, In order to avoid prejudice to later deter- minations of stratigraphic equivalences through- out the cave, we have used field designations for the “natural” stratigraphic divisions or arbit- rary depth determined units of excavation (“spits”) differing from trench to trench or field season to season. These field designations are recorded not only in our field notes but also on specimen labels and in the Museum cata- logues, and to some extent are reproduced in the Appendix 1 tables. However, we have attempted to group and analyse our data in terms of the major strati- graphic divisions recognised by Dortch and Merrilees (1973), and comparison of their strati- graphic sections and names with ours may be made readily even though we have modified the names slightly in some cases. A stratigraphic section revealed in the southwestern walls of the adjacent Trenches 5, 2 and 82 is represented in Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 98 £ o E 3 ■a o V o 1 140 }^^ Trench 2 Trench 5 Trench Sj iflowstone 'complex floor of cave 'day lamina; iflowstone slabs' hearth'^. 'day lamina; Ijl^white patches^ lypsum* 11960±140 ;dark earthy!$^g?^5*i''’™^ •■'layer:^jw%»p>^ gypsum O*;, hearthv .hearth- < reliably represents undisturbed mate- rial. We believe that four quartz and some prob- able calcrete artifacts catalogued as B1545 and B1546 from below this charcoal rich band also came from undisturbed material. Two detached lumps of fiowstone (now in the geological collection of the Western Australian Museum under catalogue numbers 13324, 13325) were found near the western face of Trench 7a in its uppermost part, resting on thick macro- crystalline fiowstone obviously in its undisturbed position (fiowstone D, illustrated in Figure 3 of Dortch 1974). These lumps were upside down. The stratigraphic implication is that they and possibly some or all of the bone fragments and other material from the uppermost western part of Trench 7a are disturbed and of unknown age. We suspect that they represent spoil from the unsystematic excavation of the upper part of Trench 2 (shown by Dortch and Merrilees 1972, Figure 1, as “Small Excavation”), which was made before the present systematic series began in 1970, or that they are spoil from ancient dig- ging. From the presence of an unbroken thin lithi- fied band (labelled B in our field records) in the uppermost eastern part of Trench 7a and ex- tending into Trench 7b and 7c, we are satisfied that most of the dark earthy layer in Trench 7 is undisturbed, as illustrated in Figure 3 of Dortch (1974). Nevertheless, some (probably most) of the animals recorded for dark earthy layer in Trench 7 in Appendix 1 tables represent disturbed material of unknown stratigraphic origin. This distui'bance probably affected dark earthy layer in Trench 5 but was not detected during excavation. Therefore we make no infer- ences from the fauna recorded from dark earthy layer. Source of the bone sample Dortch and Merrilees (1973) and Dortch (1974) show that humans occupied Devil’s Lair at times, and it is possible that much of the bone sample here discussed represents discarded rem- nants of human meals. A small proportion of the bone fragments recovered are charred, possibly in the process of cooking carcases, or possibly from being discarded into camp fires, as men- tioned by Hammond (1933) for modern Abori- gines. Charring has been noted affecting the fol- lowing taxa in Devil’s Lair; the catalogue num- ber in brackets represents one example:— Dasyurus (73.8.294), Isoodon (73.8.359), Pera- meles (73.8.448), Potorous (73.12.173), Bettongia penicillata (73.10.1264), B ? lesueur (73.12 154), PetrogaJx (73.9.1151), Macropus fuliginosus (73.9.783), Setonix (73.12.182), Rattus fuscipes (73.11.60), lizard (73.8.821), snake (74.5.32). Presumably marrow would have been eaten by ancient Aborigines, as by other people (e.g. modern American Indians — Neumann and Di- Salvo 1958). Also it is possible that small bones were attractive as food items to a people accus- tomed to using their teeth as vigorously as de- scribed by Campbell (1939) for modern Abori- gines and by Neumann and DiSalvo (1958). But it is difficult to account for the extreme fragmen- tation of practically all bones in the sample (see Figure 3) unless food preparation methods then in use included systematic pounding of the car- cases before or after cooking, as described by Gould (1968a) for modern inland Aborigines. Tedford (in Gould 1968b) comments on fragmen- tation and charring of bones from an inland archaeological site, suggesting that pounding of food animals is a long standing practice of these inland peoples. Brain (1970) suggests that extreme fragmentation of bone in a deposit, and Gorman (1971) that fragmentation with some charring, is a hallmark of human contri- bution. Lundelius (1966) believed that fpgmentation of bone, presence of predator remains, and bone bearing coprolites together indicated that a bone deposit had been accumulated by that predator, in this case Sarcophilus. Indeed, the name “Devil’s Lair” was conferred on the site because of his belief. Much of the bone from our excava- tions is reminiscent of that recovered from the modern Sarcophilus leavings described by Doug- las, Kendrick and Merrilees (1966). It is possible that while primary fragmenta- tion of our bone sample was the work of Abori- gines, secondary fragmentation resulted from devils working over the leavings of human beings after (or perhaps even during) human occupa- tion of the cave. Devils now live successfully in close contact with human activities in parts of Tasmania (Guiler 1970). They may long have done so, and indeed may have occupied the place later held by dogs in Aboriginal camps. If dogs had been living in Devil’s Lair it seems very pro- bable that some remains would have been de- tected. Thus we cannot be sure of how much of the bone in our sample represents human meals, but from the charring of some of it and the form of bias demonstrated below in the remains of journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 102 Macropus fuliginosus, we are confident that some of it does. Adult M. fuliginosus are too large to be killed by devils; however, some re- mains of this species may have been brought into the cave as carrion. The presence of shell fragments of mussel and the eggs of emus (Dortch 1974) marine shell (Dortch and Merrilees 1972) and the single fish vertebra (73.11.177) which is the basis for the fish entry in Trench 8? in the appendix tables, also point to man as a contributing predator. No other single predator species is likely to have sampled the wide range of habitats suggested by the mammals in the deposit. Possibly other predators, such as Dasyurus or owls contributed to the bone sample, but Devil’s Lair does not show an overwhelming preponder- ance of small (mouse or rat sized) mammals as do some other deposits (e.g. those described by Archer and Baynes 1972) interpreted as accu- mulated by owls or Dasyurus. Our provisional belief is that the bone sample in the main represents the leavings of human beings, but that other predators may have con- tributed remains of their own prey or modified the human refuse. Bias in the sample The bone recovered from Devil's Lair is gene- rally well preserved, yet there is marked imbal- ance in the frequency of occurrence of different parts of the skeleton within most species. For example, we examined Macropus fuligin- osus specimens from Trench 6, assembling all bones attributable to this species, not only those cranial and post cranial elements admitted to our faunal analysis, but also vertebrae, ribs and other elements not so admitted. There were large discrepancies between the numbers of bones and teeth of all kinds actually present and the numbers which would have been pre- sent if all teeth and bones of all M. fuliginosus individuals represented in our sample had been left lying within the confines of Trench 6. Dis- crepancies in numbers of teeth, of bones of the hands and feet, and of vertebrae, are set out in Table 1. There appeared to be similarly large discrep- ancies between numbers actually present and expected numbers of long limb bones, pelvis, and so on for all parts of the skeleton, but be- cause these were all fragmented, the discrepan- cies are not easily quantified and tabulated. For example in the “top” subdivision of brownish earthy layer, at least 7 individuals of M. fulig- inosus were represented. But there were only 2 distal and no proximal fragments of humerus, only 1 fragment of radius, only 2 proximal and no distal fragments of femur and so on. Perhaps the most striking discrepancy was in the distinctive large lower incisor teeth. Only 1 of these (73.8.977, from a young animal) was recovered from the whole 2.27 cubic metres of deposit excavated from Trench 6. Yet, as shown in Appendix 3 at least 17 individuals of M. fuliginosus must have contributed to the sample, and it is very likely that 24 or even more in- dividuals actually contributed. If most of the animals represented were taken into the cave as whole carcases, cooked and eaten on the spot, and the bones discarded, one would expect considerable lateral scattering of the bones of any individual. But if most parts of the cave were used, there should be consider- able overlap of the discarded scattered bones of several individuals of any of the more common species on any occupation floor. In an excavation sampling these old occupation floors, one would expect mixtures of individuals to be represented, but if distribution were random any particular anatomical element, such as the first lumbar vertebra, would be about as well represented as any other element, such as the fourth metatarsal of the right foot. Our examination of M. fuliginosus from Trench 6 suggests that one or more selective processes have operated. Many writers, especially under the stimulus of the “osteodontokeratic” concept of Dart (1957), have discussed such selective processes. Butcher- ing of large carcases at the kill site followed by removal of some but not all the bones back to the campsite in dismembered joints (including the “schlepp effect”) is often cited as biassing Table 1 Macropus fuliginosus m Trench 6 stratigraphic subdivision Minimum number of individuals {.juveniles in brackets) Tooth most abundantly represented, and number present Teeth Bones of hand and foot Vertebrae Total present Expected number Total present Expected number Total present Expected number Dark earthy layer .... 1 (?) 0 0 20 0 104 0 50 Pale band 1 (?) 0 0 20 1 104 0 50 Flecked... lens .... 1 (1) 0 0 24 2 104 1 50 Second dark earthy layer .... 2(1) 2 X KP 10 44 1 208 0 100 Cave pearl and bone layer 2(1) (1 each of several) 8 44 7 208 4 100 Top, brownish earthy layer 7 (5) 7 X LU 30 160 26 728 15 350 Upper middle, brownish earthy layer 2(1) 2 X LP 7 44 1 208 3 100 Lower middle, brownish earthy layer Bottom, brownish earthy layer .... 8(6) 7 X LP 43 184 107 832 52 400 2(2) 2 X RP 8 48 20 208 n 100 Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 103 bone samples, for example by White (1952, 1953a, b, 1955), Innskeep and Hendy (1966), Perkins and Daly (1968), Daly (1969), or Perkins (1969), Dart (1954, 1957), Kitching (1963), Wolberg (1970), and others have stressed selection of some bones for use as tools. Brain (1967a, b, c, 1970), Isaac (1967), Schaller and Lowther (1969) and others stress carnivores acting alone, or carnivores acting on accumulations of bone left by human beings. White (1953c, 1956) describes personal or group habits ana customs biassing bone samples. In the case of Devil’s Lair it seems unlikely that the discrepancy between bones present and bones expected has anything to do with off site butchering methods because most of the animals eaten were small and even a large kangaroo carcase can be carried by a man. But selection of bones for making tools can be inferred. For example it seems likely that the dearth of kan- garoo lower incisors, mentioned above, was due to the removal of these teeth, or even of whole jaws containing them, for use as tools. Dortch and Merrilees (1972) illustrate a wallaby lower incisor with transverse incisions which may have been used to bind it to a stick, or may have resulted from such binding. Kangaroo lower in- cisors may well have been used similarly. The one kangaroo lower incisor found in Trench 6 had a very open root, such as one finds in very young animals, and hence may have been too fragile to serve as a tool. Other bone implements from the deposit are described by Dortch and Merrilees (1973) and Dortch (1974). It is probable also that much bone was eaten by man or by devils and reduced to fragments small enough to pass through our screens. Doug- las, Kendrick and Merrilees (1966) report ob- servations on living devils and their effects upon bone, and the specimens concerned, still pre- served in the Western Australian Museum, in- clude a good deal of finely comminuted bone re- covered from the faeces of the devils observed. Not only devils, but also human beings accus- tomed to using their teeth more vigorously than do most modern people, might fragment and in- gest appreciable quantities of bone. Many of the bone fragments in our sample show rounding and smoothing of what initially must have been jagged fracture edges. Many tooth bearing fragments (e.g. Bettongia penicil- lata specimens 73.8.719, 73.9.51, 73.12.379) show this rounding, and also many small fragments of post cranial bones of many kinds. At present we are uncertain about the smoothing process in- volved. It might be from human handling and use, for example to flesh animal skins or smooth wooden spear shafts. It might be from the trampling of fragments into a sandy floor, as de- scribed by Brain (1967b, c), or from passage through and corrosion in the gut of devils or human beings. Whatever the rounding process, it is reasonable to suggest it converts some pro- portion of the original bone sample into parti- cles below the limit of our recovery methods. The bone sample as an index of human use of the cave. By assuming that all the bone recovered in our excavations came from animals eaten in the cave by humans, and by making the other as- sumptions set out below about the bone sample, we arrive at a very rough estimate of the extent to which the cave might have been used from about 19 000 yr B.P. to about 12 000 yr B.P. To do this, we have combined data from first orange brown earthy layer in Trench 5 and brownish earthy layer in Trench 6. For the purpose of comparing estimates of minimum numbers of individuals of a particular species in one stratigraphic unit with the same species in another, we have made no attempt to correct any overestimation shown in the Appen- dix 1 tables. However, for the purpose of inter- preting our bone sample in terms of the meat it represented, we must compare estimates for dif- ferent species, and therefore have applied cor- rection factors to the numbers obtained from the appendix tables. These correction factors are based on our study of the degree of overestima- tion in Sminthopsis , Pseudocheirus and Macro- pus fuliginosus reported in Appendix 3. We mul- tiply the numbers shown in the appendix tables for Macropus fuliginosus, the only large animal present by L and for the animals of intermedi- ate body size {Dasyurus^ Sarcophilus, Thyla- cinus, Isoodon, Perameles, Trichosurus, Pseudo- cheirus, Potorous, both species of Bettongia, Petrcgale, Macropus eugenii, M. irma and Setonix) by §. We leave unchanged the num- bers shown for all other vertebrate taxa which are of small size. Thus corrected, the numbers of individuals for first orange brown earthy layer in Trench 5 and brownish earthy layer in Trench 6 together are: large kangaroos 27 animals of intermediate size 653 small vertebrate animals 566 To translate these findings into numbers of human meals, we assume that about 1 kg live weight of any animal would constitute one meal of meat for one person (cf. McArthur in Mount- ford 1960; Gould 1967). Each large kangaroo might then represent 20 meals, each animal of intermediate size 2 meals, while 5 small animals might be required for one meal. On these as- sumptions, at least 2 000 meals of meat are re- presented by the bone sample under considera- tion. It is unlikely that the human occupants of the cave lived entirely on vertebrate prey ani- mals. If we assume that half their diet was plant food, remains of which have dissipated or have not been recognised in our excavations, and invertebrate animals, then at least 4 000 meals are represented by our bone sample. The area excavated in Trenches 5 and 6 is probably not more than 1% of the area of cave floor below which a thick deposit might exist. If we assume that the 19 000 to 12 000 yr B.P. section of the deposit is typified in Trenches 5 and 6, then the bone sample from these trenches represents only about 1% of what could be re- Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 104 covered from the cave. Thus at least 4 000 x 100 400 000 meals were eaten in the cave over a period of about 7 000 years. This is an average of about 60 meals per year. A family group of 5 people spending 6 con- secutive days in the cave and eating 2 meals (one of meat) each day in the cave would account for 60 meals. Thus one visit to the cave each year by such a family group would be enough to account for the minimum number of individual prey animals represented by our bone sample. However it is likely that our estimates of minimum numbers of individuals greatly under- estimate the actual number of animals eaten in the cave (cf. Perkins in Matolcsi 1973). Also the tables of Appendix 1 suggest that the rate of occupancy of the cave was far from uniform. It is reasonable to suggest alternatives to this concept of a small family group staying in the cave for a few days at a time once a year. Such a group might have stayed there for longer periods less often and the digging and presum- ably use of the large Pit 2 (described by Dortch and Merrilees 1973) reinforces this suggestion. Or the cave may have served as a meeting place for a larger number of people for a few days even less often. At present we are unable to choose between these or other possible alternatives. Indeed, there are so many unexamined and in some cases un- examinable assumptions underlying our calcula- tions that we conclude merely that our sample indicates intermittent rather than continuous human occupation of the cave between 19 000 and 12 000 years ago. The fossil mammal fauna of Devil’s Lair We have recognised 35 species of mammals among material recovered from the 1970, 1972 and 1973 excavations in Devil’s Lair, and these are listed in Table 2. Descriptions of the animals concerned, together with vernacular names, authors of systematic names, and a selected list of scientific studies are given by Ride (1970). Table 2 Mammal species recorded from the 1970, 1972 and 1973 excavations in Devil’s Lair Carnivorous marsupials Dasyurus geofjroii Phascogale tapoatafa Antechinus fiavipes Sminthopsis murina Sarcophilus harrisii Thylacinus cynocephalus Bandicoots Isoodon obesulus Perameles — species not so far identified Possums Trichosurus vulpecula Pseudocheirus peregrinus Cercartetus concinnus Rat-kangaroos Potorous tridactylus Bettongia penicillata Bettongia lesueur Wallabies and kangaroos Petrogale — species not so far identified Lagorchestes — species not so far identified Macropus eugenii Macropus irma Macropus fuliginosus Setonix hrachyurus Native rats and mice Hydromys chrysogaster Pseudomys albocinereus Pseudomys occidentalis Pseudomys shortridgei Pseudomys praeconis Notomys — species not so far identified Rattus (probably all R. fuscipes) Bats Macroderma gigas Nyctophilus timoriensis Nyctophilus geoffroyi Eptesicus pumilus Chalinolobus gouldii Chalinolobus morio Pipistrellus tasmaniensis Tadarida australis Our estimates of minimum numbers of each mammal species from the 1972 and 1973 exca- vations, and also of other vertebrate taxa, ar- ranged stratigraphically, are tabulated in Appen- dix 1. Dortch and Merrilees (1972) similarly tabulate such estimates for the 1970 excavation. Some species are represented by a few individuals only, others by many individuals. We have attempted to show in Figures 4 and 5 the changes in relative proportions of the species more abundantly represented from about 19 000 to about 12 000 years ago in the 1972 and 1973 excavations. Our sample from about 25 000 to about 19 000 years ago, as shown in Appendix 1, is so small that we have not attempted to show relative proportions of the various species in- volved. The modern mammal fauna of the Devil’s Lair district As with most other localities in Australia, the mammal fauna now living in the Devil’s Lair district is not the same as that present just before the arrival of European man. Drastic changes to habitats, efforts to exterminate na- tive “pests”, and introduction of exotic species have all had profound effects on the mammal populations. Therefore we use the term “modern mammal fauna” to mean that occurring in im- mediately pre-European times. This fauna is the only one meaningful for comparison with the fossil fauna, since it represents the end pro- duct of the prehistory of the area. Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 105 i Radiocarbon date yr B.P. 1 7400 i 350 Figure 4. — Relative proportions of the well represented mammal species in two major stratigraphic divisions in Trench 6, Devil’s Lair. Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 106 Appendix 2 gives details of museum and other records forming the basis of our reconstruction of the modern mammal fauna of the Devil's Lair district. These enable us to refine pre- vious attempts (e.g., Merrilees 1968, Archer and Baynes 1972). We have taken into account the likely ranges from Devil’s Lair covered by predators bringing prey back to the cave; and by “Devil’s Lair district” we mean this predator determined area. Table 3 Mammal species we are confident were present in the modern fauna of the Devil’s Lair district Carnivorous marsupials Dasyurus geoffroii Phascogale tapoatafa Sminthopsis murina Possums Tarsipes spencerae Trichosurus vulpecula Pseudocheirus peregrinus Cercartetus concinnus Rat-kangaroos Bettongia penicillata Wallabies and kangaroos Macropus fuliginosus Setonix brachyurus Native rats and mice Hydromys chrysogaster Rattus fuscipes Table 4 Mammal species probably in the modern fauna of the Devil’s Lair district Carnivorous marsupials Antechinus fiavipes Bandicoots Isoodon obesulus Rat-kangaroos Potorous tridactylus Wallabies and kangaroos Macropus irma Native rats and mice Pseudomys shortridgei Pseudomys praeconis Rattus tunneyi Bats Nyctophilus timoriensis Nyctophilus geoffroyi Eptesicus pumilus Chalinolobus morio Pipistrellus tasmaniensis Carnivorous eutherians Canis familiaris Table 5 Mammal species possibly forming part of the modern fauna of the Devil’s Lair district Wallabies and kangaroos Macropus eugenii Bats Chalinolobus gouldii Tadarida australis We are confident that the species listed in Table 3 were part of the modern fauna, and we think it probable that the species listed in Table 4 were too. Table 5 lists mammals re- corded within 75 km of Devil’s Lair in modern time, but whose inclusion in the modern fauna of the Devil’s Lair district is doubtful for rea- sons discussed in Appendix 2. Comparison of modern and fossil mammal faunas from Devil's Lair By combining the species listed in Tables 3 and 4, we suggest that the mammal fauna of the district just before it was affected by European man probably included 25 species, and it may have included the 3 additional species listed in Table 5. Of this total of 28 species, only Tar- sipes spencerae, Rattus tunneyi and Canis fami- liaris are not so far recorded in the fossil fauna from Devil’s Lair. The fossil fauna listed in Appendix 1 includes 33 species, and 2 additional species were recov- ered from the 1970 excavation, namely Lagor- chestes sp. and Chalinolobus morio, discussed below. Of these 35 fossil species (see Table 2), the 10 species listed in Table 6 appear not to have formed part of the modern fauna of the Devil’s Lair district. Small, frail remains of Tarsipes spencerae may have escaped detection. Although the species is known to live at present in heaths in many localities between Yanchep and Dongara, only one fossil specimen (Archer 1973) has been recognized from any of the numerous caves in this region, many of them containing large ac- cumulations of bone from owl pellet deposits. But we suggest that the absence of Rattus tun- neyi and Canis familiaris from the fossil fauna so far recognized represents a real absence of these two species from the Devil’s Lair district in ancient times. Dortch and Merrilees (1972) suggested that a single canine tooth (70.12.202) found by them in 1970 might have represented a Dingo, but this tooth has been examined subsequently by Professor N. W. G. Macintosh and colleagues, and is considered by them (Macintosh, personal communication) in fact to represent Sarco- philus. No specimen representing Canis was found in the 1972 and 1973 excavations. How- ever, the carnivorous marsupial of size compar- able with Canis namely Thylacinus, is repre- sented only by two isolated teeth. The possi- bility remains that both the marsupial and the Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 107 eutherian carnivoiTs were present, but were diffi- cult to catch or distasteful to eat, and there- fore are not well represented, or not repre- sented at all, in the litter left lying in the cave. However, the Dingo is generally believed to have been introduced into Australia by Abo- rigines migrating in Recent time (Jones 1970), and our evidence is consistent with this view. Archer and Baynes (1972) postulate that Rat- tus tunneyi extended its range along the west- ern coastal plain into the extreme south west of Western Australia, including Devil’s Lair dis- trict, only in Recent time. This is supported by unpublished studies by one of us (J.K.P. ) of a stratified deposit in a cave near Augusta, in which Rattus tunneyi occurs only in the top layers of the deposit, dated on charcoal as 2900 ± 80 radiocarbon years B.P. (date SUA-227; or later. Dortch and Merrilees (1972) report the pres- ence of a single specimen (70.12.1132) tenta- tively ascribed to Lagorchestes sp. Unless the specimen is aberrant, we concur in this identi- fication. Among the specimens simply listed as “bats” by Dortch and Merrilees (1972), we have recognized three as ChaJAnolohus morio. The three specimens (70.12.776, 70.12.911 and 71.1.223) come from three different levels m Trench Al. Neither Lagorchestes sp. nor Chali- nolohus morio was recognized in the material excavated in 1972 and 1973. Even though some species of mammals pres- ent as fossils were not present in the modern fauna of the Devil’s Lair district, ail are pres- ent in the modern fauna of Australia as a whole (including Tasmania), and only two of these species seem to have disappeared from the Aus- tralian mainland. These are Thylacinus cyno- cephaJ.us and Sarcophilus harrisii, both of which are known to have survived on the mainland until late Recent time (Partridge 1967, Archer and Baynes 1972). Table 6 Mammal species not forming part of the modern fauna of the Devil’s Lair district, hut recorded from the 1970, 1972 and 1973 excavations in Devil’s Lair Carnivorous marsupials Sarcophilus harrisii Thylacinus cynocephalus Bandicoots Perameles sp. Rat-kangaroos Bettongia lesueur Wallabies and kangaroos Petrogale sp. Lagorchestes sp. Native rats and mice Pseudomys alhocinereus Pseudomys occidentalis Notomys sp. Bats Macroderma gigas Thus it seems that 2 or possibly 3 mammal species have arrived in the Devil’s Lair district, and at least 10 species have disappeared (see Table 6) since the fossiliferous parts of the de- posit were sealed by thick flowstone about 12 000 years ago. Modern distribution patterns and habitat adaptations of mammal species On the basis of general patterns revealed by unpublished studies of modern mammal distri- butions in south western Australia made by one of us (A.B.), we have divided the species listed in Table 2 into two categories. These are “forest mammals” and “non-forest mammals”. The forest mammals, listed in Table 7, are de- fined as those species whose modern distribu- tions include the “forest block”, whicn is the area mapped by Gardner (1944, Plate X) as sclerophyllous forest and mesophytic forest. It includes the formations mapped by Smith (1973) in the Busselton-Augusta region as high open forest and open forest. Sarcophilus harrisii and Thylacinus cynocephalus are regarded as forest mammals on the basis of their modern Tasmanian distributions, even though they ap- parently did not persist in Western Australia into modern times. The non-forest mammals, listed in Table 8, are defined as those species recorded from Devil’s Lair whose distributions do not include the forest block. The distri- bution studies suggest that the natural eastern margin of the forest block, where it merged into woodland or shrub formations (before most of the area was cleared for agriculture), repre- sents an important demarcation of modern faunas. The ranges of most forest mammals extended beyond this line, but it appears to have represented a barrier to non-forest mam- mals in immediately pre-European times. No thorough investigation has been made of differences between the mammal faunas of the Karri high open forest and the Jarrah-Marri open forest. It is possible that fewer mammal species occur naturally in the Karri forest, but in the absence of data we assume that all the forest mammals occurred in both. Species which are restricted in range of dis- tribution or habitat are the most useful indi- cators of environmental conditions. The forest mammal with the most restricted distribution is Potorous tridactylus. In Western Australia, where it may now be extinct, P. tridactylus ap- pears to have been limited to the area of the extreme south west in which some rainfall oc- curs throughout the summer. There it prob- ably inhabited densely vegetated watercourses and wet heaths. Setonix brachyurus also finds densely vegetated forest gullies an optimum habitat: but it has a wider distributional range, from Gingin Brook in the north (Roe 1971) to the Hunter River east of Bremer Bay. The latter locality represents an eastward extension of the previously known range. It is based upon a dentary (M10519), considered to repre- sent a modern animal, which was picked up in 1970 by W. H. Butler. The distribution cf Journal of the Royal Society of Western Australia. Vol. 58, Part 4. December, 1975. 108 Setonix hrachyurus within the northern part of the forest block appears to have been restricted to the western areas; its range probably did not reach the eastern margin. The distribution of Rattus fuscipes in Western Australia is unlike that of any other mammal. It occurs in the southern part of the forest block, principally along river systems, and also in coastal dune systems and heaths from Jurien Bay round to Israelite Bay. There appears to be no reliable evidence for its occurrence in the northern areas of the forest olock. Both of the specimens from that area cited by Taylor and Horner (1973), Western Australian Museum M5815 and C.S.I.R.O. CM859, have been exam- ined by one of us (A.B.) and are considered to be Rattus rattus. Near Devil’s Lair Rattus fuscipes has been collected live both in regen- erating Peppermint open scrub and Karri high open forest. It is possible that the ranges of the three forest mammals Macropus eugenii, Chalinolobus gouldii, and Tadarida australis, listed in Table 3, did not cover the extreme south west of the forest block. In addition little is known of the habitat requirements or full ranges of the other bat species included in Table 7. Some uncer- tainty also exists in the cases of Sarcophilus harrisii and Thylacinus cynocephalus . Table 7 Forest viammals recorded from the Devil’s Lair deposit Carnivorous marsupials ■'Dasyurus geoffroii Phascogale tapoatafa ^Antechinus fiavipes ^Sminthopsis murina * Sarcophilus harrisii Thylacinus cynocephalus Bandicoots *Isoodon ohesulus Possums *Trichosurus vulpecula *Pseudocheirus peregrinus *Cercartetus concinnus Rat-kangaroos *Potorous tridactylus *Bettongia penicillata Wallabies and kangaroos *Macropus eugenii *M. irma fuliginosus *Setonix hrachyurus Native rats and mice Hydromys chrysogaster *Rattus fuscipes Bats Nyctophilus timoriensis N. geoffroyi Eptesicus pumilus Chalinolobus gouldii C. morio Pipistrellus tasmaniensis Tadarida australis ♦Well represented through the deposit and included in Figures 4 and 5. Records indicate that the other 13 forest mammals, not discussed above, occurred right through the forest block. Of these Phascogale tapoatafa, Antechinus fiavipes, and Pseudo- cheirus peregrinus also occurred in the high rainfall woodlands. Hydromys chrysogaster , the Water Rat, is a specialized species normally only found in or adjacent to bodies of water. The remainder have been recorded from a broad spectrum of habitat types, and some range far into the dry inland of Australia; e.g., Dasyurus geoffroii and Trichosurus vulpecula both formerly occurred widely in desert areas. Table 8 Non-forest mammals recorded from the Devil’s Lair deposit Bandicoots *Perameles sp. Rat-kangaroos *Bettongia lesueur Wallabies and kangaroos *Petrogale sp. Lagorchestes sp. Native rats and mice ■^'Pseudomys albocinereus P. occidentalis *P. shortridgei P. praeconis *Notomys sp. Bats Macroderma gigas ♦ Well represented through the deposit and included in Figures 4 and 5. Since we are unable to identify some of the non-forest mammals listed in Table 8 beyond generic level, we here discuss the species most likely on biogeographical grounds to be repre- sented in Devil’s Lair — Perameles bougainville, Petrogale penicillata, Lagorchestes hirsutus and Notomys mitchellii. Six non-forest mammals share a common distributional characteristic, that is they have been recorded along the inland side of the forest block. These are Perameles bougaimnlle, Bettongia lesueur, Pseudomys albocinereus, P. occidentalis, P. shortridgei and Notomys mitch- ellii. Of the other four species, Petrogale peni- cillata and Lagorchestes hirsutus are recorded from just north east of the forest block. Macro- derma gigas from the coastal plain north west of the forest block, and Pseudomys praeconis from just north of the forest block. The modern distributions of all five species of native rats and mice, and possibly Bettongia lesueur, also included the northern part of the western coastal plain. Although only two of the non-forest mammals (P. shortridgei and P. praeconis) extended south on this plain along the western side of the forest block in modern time, we suggest below that all Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 109 our non-forest mammals occupied the southern part of the western coastal plain at some time during the accumulation of the Devil's Lair deposit. The south western populations of Petrogale, restricted in modern time to rocky hills inland from the forest block, are probably conspecific with other populations occurring in arid locali- ties such as the Macdonnell Ranges (Ride 1970), but we are not sure that any is conspecific with the Devil's Lair population. Most Petrogale species now seem to inhabit rock outcrops which shield them from extremes of temperature. Macroderma gigas ranged over much of desert and tropical Western Australia (Douglas 1967), but its remains, presumably of Recent age, are widespread in caves between Yanchep and Don- gara. In one deposit it was present throughout the Recent (unpublished data from Hastings Cave, A.B.) associated with a mammal fauna characteristic of heath habitats (Lundelius 1960, with the qualifications of Archer and Baynes 1972). Sand appears to be a substrate requirement of Pseudomys albocinereus and most species of Notomys. While P. albocinereus occurs mainly in coastal saind plains carrying open heath, Notomys mitchellii occurs not only in such habi- tats but also on inland sand plains carrying open heath and other shrub formations. Pere- meles bougainville probably also occurred in sand plains in south western Australia (Ride 1970). Although Pseudomys occidentalis remains are known from the surface in caves in the coastal sand plains, modern specimens have mainly been collected in areas which have a loam or gravelly sand substrate. Insufficient information is available to deter- mine the habitats occupied by Bettongia lesueur, Largorchestes hirsutus, Pseudomys shortridgei and P. vraeconis in the south west of Aus- tralia. Changes in the mammal fauna around Devil's Lair, and interpretation We have attempted to quantify information on the mammals from our excavations, but have shown that the results must be viewed with caution. We have not applied statistical tests. In order to assess changes in the mammal fauna, we distinguish between the small sample available for the period 25 000 to 19 000 years ago and the larger sample available for the period 19 000 to 12 000 years ago, and between species which are represented only in small numbers widely spaced in the deposit, and those which are present at most levels (even if only in small numbers) , The relative abundances of the well repre- sented species for the later period in Trench 6 and in the Trench 2, 5, 7 and 8 complex are shown as bar diagrams in Figures 4 and 5 respectively. The stratigraphic groupings shown in these figures seem to us broad enough to minimize stratigraphic uncertainties in our sampling procedures and to allow comparisons to be made between Trench 6 and Trenches 2, 5, 7 and 8, We infer real changes in relative abundance where species show similar trends from level to level in Trench 6 and in Trenches 2, 5, 7 and 8. We make no such inferences where trends are dissimilar. We have compared the change in relative abundance of each species from one stratigraphic level to the next higher level in Trench 6 with the corresponding change in Trenches 2, 5, 7 and 8. The numbers of species synchronously in- creasing, decreasing and changing in opposite ways are set out in Table 9. There is no case of a species remaining unchanged. Our com- parisons were made using Figures 4 and 5. The substantial proportion of species chang- ing in opposite ways at all levels (right hand column of Table 9) suggests that much of the observed change is random. However, there appears to be some decrease in randomness to- wards the top of the deposit. This is explained by the trend shown in the central column of Table 9. More species decline in the upper levels of the deposit. Figures 4 and 5 show that, between the top subdivision of brownish/first orange brown earthy layer and (cave earth and) flowstone complex, 5 non-forest mammals iPerameles sp.. Bettong.a lesueur, Pseudomys albocinereus, P. shortridgei, and iVofomi/s sp.) and 4 forest mam- Table 9 Numbers of species changing in relaUi'e abundance in the. same wag in Trench (> and in Trenches 2. 5, 7 and 8 from stratigraphic level to next higher stratigraphic level— data of Figures 4 and 5, dealing with 21 species well represented through the deposit. stratigraphic level No. of species increasing in relative abundance Xo. of species decreasing in relative abundance Xo. of species showing opposing trends, or changes in one trench hut not in the other (Cave earth and) flowstone complex 1 c 5 9 7 c o -top 1 o — ■f> ^ -upper middle 5 6 10 x: H \ es * r- -lower middle J 1 1 7 5 9 ^ P -bottom J 12 . _ . Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 110 mals {Dasyurus geoffroii^ Sminthopsis murina, Bettongia penicillata and Macropus eugenii) de- cline. These forest mammals are all species recorded from a broad spectrum of habitats. At the same time, the 2 species (Potorous tridac- tylus and Setonix brachyurus) most restricted within forest show their greatest increase. Of the 21 species shown in Figures 4 and 5, 15 species (* in Table 7) are forest and 6 species (* in Table 8) are non-forest mammals. We next analyse the forest and non-forest mammals which change in relative abundance from level to level in the same way in Trench 6 as in Trenches 2, 5, 7 and 8. The results (in Table 10) suggest that these forest mammals show no discernible trend from about 19 000 years ago (the time of deposition of the bottom subdivision of brownish/first orange brown earthy layer) to about 12 000 years ago (the time of deposition of cave earth and fiowstone complex). On the other hand, the non-forest mammals analysed at first increase and then decrease in relative abundance during this period. We have no data from Devil’s Lair on trends in relative abundance of mammal species be- tween about 12 000 years ago and modern time, but we have shown above that 10 species (Table 6) which were present about 12 000 years ago were not present in modern time, whereas only 2 (or at most 3) species which were present in modern time were not present in the Devil’s Lair deposit. This net impoverishment of the mam- mal fauna of the district is mainly of the non- forest mammal component. The disappearance of the 2 forest mammals (Sarcophilus harrisii and Thylacinus cynocephalus) may be causally related to the appearance of Canis familiaris. Thus we suggest that most of the non-forest mammals began to decline in terminal Pleis- tocene time, that the decline was marked by about 12 000 years ago, and that it continued subsequently until at some time not at present known, these species became extinct locally. On the other hand, the forest mammals appear to have fluctuated in relative abundance, but to have persisted into modern time. At present forest habitats predominate over non-forest habitats in the Devil’s Lair district. As described in the Introduction, Devil’s Lair is now surrounded by Karri forest which gives way within 2 km to the east to very extensive Jarrah- Marri forest, and within 2 km to the west to low tree and shrub formations. For our purposes we divide the vegation into a “forest zone” occu- pied by the forest mammals and a “non-forest zone” to which the non-forest mammals are re- stricted. Our forest zone comprises the Karri high open forest and Jarrah-Marri open forest mapped by Smith (1973). The main formation in our non-forest zone is Smith’s Acacia open heath, but it includes small areas of Peppermint open scrub, Banksia open woodland. Pepper- mint low woodland, Jarrah low open forest, and Peppermint low open forest. The boundary be- tween Karri high open forest (Figure 6) and Acacia open heath (Figure 7) is now very sharp- ly demarcated near Devil’s Lair by a crest which shelters the Karri forest from the prevailing winds. We include two low tree formations in our non-forest zone because the same plant com- munities at times pass from shrub formations to these low tree formations as they approach climax, and they probably continue to harbour non-forest mammals. This applies particularly to communities in which Peppermint is an im- portant component. On the other hand, the high open and open forest which we include in our forest zone are tree formations from very early stages of their cycles. The faunal changes described above suggest there was a diminution or contraction away from Devil’s Lair of the non-forest zone which began to affect the fauna in terminal Pleistocene time and continued into the Recent. It is tempt- ing to suggest climatic change as the main cause of this inferred vegetational change, but glaciceustatic rise in sea level must also be con- sidered as a contributing or indeed even as the main cause. There seems to have been a lag between maxi- mum glaciation and maximum glacioeustatic fall in sea level, and the extent of both lag and fall have been variously estimated, even for the last glaciation (Milliman and Emery 1968, Guil- Table 10 Comparison of numbers of species of forest with non-forest mammals which show the same trends in relative abundance in Trench 6 and Trenches 2, 5, 7 and S from one stratigraphic level to the next higher level — data from Figures 4 and 5. Stratigraphic level . u-C -upper middle J u. I \ C5 I li -lower middle J 1 \ O 6- -bottom j Species showing same trend in Trench 6 as in Trenches 2, 5, 7, 8. (see Table 9) Forest mammals increasing in relative abundance Forest mammals decreasing in relative abundance Xon-forest mammals increasing in relative abundance Xon-forest mammals decreasing in relative abundance 14 5 4 0 5 11 5 :i 0 3 12 5 4 2 1 9 2 2 5 0 Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. Ill Figure 6. — Karri growing on the eastern side of the ridge, near Devil’s Lair. The density has been reduced by felling'. cher 1969, Jongsma 1970, Richards 1970, Morner 1971, Gill 1971, Walcott 1972, Andrews 1973. Hopkins 1973 and many others). The subject ia controversial and we here assume that at some time after 19 000 yr B.P. the sea west of Devil’s Lair fell to 100 m or more below its present level, and then began to rise, reaching about 40 m below its present level by 12 000 yr B.P., and reaching its present level at some time in the Recent. We assume also that soundings in Hamelin Bay (Archdeacon 1878) reveal a topo- graphy which has not changed markedly since 19 000 yr B.P., whether submerged or not. On these assumptions, the sea coast would have lain more than 20 km west of its present position soon after 19 000 yr B.P., and more than 6 km west about 12 000 yr B.P., but has been close to its present position for several thousand years. The boundary between forest and non-forest zones may have occupied more or less its present position near Devil’s Lair at the time of lowest sea level, perhaps 16 000 years ago. This would have resulted in a non-forest zone some 30 km wide between it and the coast rather than the 3 km of the present day. In this case, human hunters or other preda- tors operating from Devil’s Lair within a radius of about 10 km and hunting mammals impar- tially in forest and non-forest zones would have obtained forest and non-forest mammals in com- parable proportions about 16 000 years ago. As sea level rose from 16 000 yr. B.P. to, say, 13 000 yr. B.P., the non-forest zone may have been nar- rowed, but may still have exceeded the hunting range of predators operating from Devil’s Lair, so that the proportions of mammals obtained by them from different zones would remain un- changed. Figure 7. — Acacia open heath on the seaward slope of the ridge west of Devil’s Lair. But by about 12 000 yr. B.P. the encroaching sea may have reduced the non-forest zone to less than 10 km width. Then the Devil’s Lair hunters would have brought back to the cave a smaller proportion of non-forest than of forest mammals. Subsequently with continued en- croachment of the sea, the proportion of non- forest mammals would have fallen further. On this view, the critical time at which the extent of the non-forest zone available to hunters fell below their hunting range would have coin- cided with the end of accumulation of first orange brown earthy layer in Trench 5, for example. By the time that flowstone complex accumulated, forest mammals would have formed a larger proportion of the hunters’ prey. This may have been a time of very rapid rise in sea level (Cullen 1967). Such a view fits our data, but it may seem un- likely that the boundary between forest and non-forest zones would have been stable. This boundary probably results from dynamic bal- ance of a number of determinants such as amount and distribution of rainfall, windiness, etc., which would have been subject to consider- able fluctuation between 19 000 and 12 000 years ago. Churchill <1968) in his study of Karri, Jarrah and Marri extending back over much of Recent time has shown that rainfall characteristics rather than temperature deter- mine the distribution of these species. Another determinant (Smith 1973), probably not subject to fluctuation, is topographic — shelter from the prevailing winds. By analogy with other parts of the world, it would appear likely that temperature, quantity of rainfall, distribution of rainfall, windiness, evaporation rates and other climatic determi- Journal of the Royal Society of Western Australia, Vol. 58, Part 4. December, 1975. 112 nants of plant growth varied from 19 000 to 12 000 years ago, even for a middle latitude site with strong maritime influence like Devil’s Lair. Prom consideration of the studies of Van Andel, Heath, Moore and McGeary 1967, Galloway 1967, Damuth and Fairbridge 1970, Howard and Hope 1970, Derbyshire 1971, Adam 1971, Bowler and Hamada 1971, Quinn 1971, Bowler, Thorne and Polach 1972, Colinvaux 1972, Costin 1972, Klein 1972, Loffler 1972, Mercer 1972, Van der Hammen 1972, Webster and Streten in Walker 1972, Bon- atti and Gartner 1973, Goudie, Allchin and Hegde 1973, Kershaw 1973, Martin 1973, Van Geel and Van der Hammen 1973, and others, we suggest that temperature and quantity of rainfall may have increased, windiness decreased, and evaporation rates changed little from about 19 000 to about 12 000 years ago in the Devil’s Lair district. In such circumstances, the forest boundary might have lain more or less in its present position, and our model outlined above (of human or other hunters based on Devil's Lair finding the proportion of non-forest zone available to the decreasing relative to forest) might be tenable. The model above requires postulation of mini- mum environmental difference between late Pleistocene and present day conditions around Devil’s Lair. But our data are also consistent with more substantial differences. For example it is possible that there was no Karri and all the forest mammals lived in open iorests, or even that the non-forest zone surrounded Devil’s Lair. However, we are able to suggest that a forest barrier separated Devil’s Lair from the wide variety of vegetation formations on the inland side of the forest block. While our forest mam- mals include all the species recorded in modern time from the forest block, except Myrrnecohiiis fasciatus and Tachyglossus acvJ,eatus, our non- forest mammals appear to be a biassed sample. There are about 46 species of mammals recorded from habitats near the forest block. Of these 22 species inhabit both forest and non-forest habitats and are included in our ’‘forest mam- mals”. Of the remaining 24 species, only lO are recorded from Devil’s Lair and hence in- cluded in our “non-forest mammals”. We suggest that this reflects the types of habi- tats ^ through which our non-forest mammal species had to pass to invade the Devil’s Lair district. There appears to be a predominance of species characteristic of open heath habitats in our non-forest mammals. Such a bias would be consistent with species spreading south along a wider western coastal plain carrying pre- dominantly shrub formations, round pei’sistent forest acting as a barrier to inland popula- tions. Our sparse data on poorly represented species, and on the fauna of the period 25 000 to 19 000 years ago do not permit us to make many inferences. The minimum number of ver- tebrate individuals available to represent the period from about 25 000 to about 19 000 years ago is only 217 and the number of mammal species included is 21. (See Appe ndix 1 tables for Journal of the Royal Society of Western Trench 2 and the lowest part of Trench 5.) These 21 species represent both forest and non- forest zones, and we can infer therefore that both zones lay within reach of hunters based on Devil’s Lair. Hence they could procure the same kinds of mammal prey as their succes- sors, but it remains to be seen whether the pro- portions differed greatly. Two Pseudomys occidentalis individuals were recovered from this small sample from the low- est levels in our excavations, whereas only a single tooth represents this species in the upper levels. This tooth, although it was found in Trench 6, could have come from old material dug out of Pit 2 in ancient times and distributed about what was then the cave floor. P. occiden- talis may have disappeared early from the Devil’s Lair district. However, little should be inferred from absence of any of the sparsely represented species. Archer and Baynes (1972) commented on the presence of Hydromys chrysogaster and Nyctophilus timoriensis in their deposit at Turner Brook (south of Devil’s Lair) and the absence of these two species from the 1970 and previous excavations in Devil’s Lair. They suggested that these two species might be found as excavation proceeded in Devil’s Lair, and so it has proved. The virtual absence of Pseudomys occidentalis from the abundant re- mains so far recovered from upper levels in the Devil’s Lair deposit may be just such another sampling accident. Bats are only sparsely present in the Devil’s Lair deposit. None occasions any surprise with the exception of Macroderma gigas, previously recorded by Cook (1960) and now confirmed by us. Douglas (1967) discusses possible reasons for major contraction in the former range of M. gigas, among them that it might find difficulty in exploiting even abundant prey in a forested area. Its presence in Devil’s Lair may provide further evidence of a more extensive non-forest zone than now, or the few animals represented might be storm blown vagrants which sheltered in the cave but perished from cold or hunger without establishing a viable local colony. Between the time of sealing with thick flow- stone of the richly fossiliferous parts of the Devil’s Lair deposit and modern time, various non-forest mammals disappeared. We have at present little understanding of the reasons for these disappearances, and little knowledge of their timing. Perhaps it is a simple matter of further shrinkage of the non-forest zone and expansion of the forest zone in Recent time as sea level continued to rise and/or annual or summer rainfall increased. It is possible that non-forest mammals might have been able to maintain viable populations in the Devil s Lair district while a broad coastal corridor permitted contact with the main popu- lations further north. When this link became attenuated as the corridor contracted to- wards its present dimensions, they may have been broken into isolated colonies which dwin- dled and eventually disappeared. The late ar- rival of Rattus tunneyi (see above) by a narrow coastal corridor may not be inconsistent with ustralia. Vol. 58, Part 4. December, 1975. 113 this concept. R. tunneyi is probably preadapted to success in this situation because it can maintain large populations in pockets of vege- tation among mobile sand dunes at beach edges (e.g. specimens M8750-60 collected in 1970 at False Entrance, Shark Bay, by A.B.). Conclusions Our scanty data on the period 25 000 to 19 000 years ago, and our fuller data on the period 19 000 to 12 000 years ago, suggest that the mammal fauna in the Devil’s Lair district remained approximately stable in the sense of neither gaining nor losing many species. De- spite the growth of glaciers and ice sheets to a maximum and their subsequent decline in higher latitudes than that of Devil’s Lair, de- spite any causative or consequent changes in climate on a world wide scale which might have had repercussions at Devil’s Lair, and despite the corresponding changes in sea level, a suffi- cient variety of habitats was present to sup- port a more diverse mammal fauna than that present just before the arrival of European man. A similar degree of stability is reported by Flood (1973) for an inland site in eastern Victoria from before 18 000 to about 8 000 years ago. In the case of our coastal site, stability may have begun to break down about 12 000 years ago. Within this framework of late Pleistocene stability in species composition of the mammal fauna around Devil’s Lair, fluctuations in rela- tive proportions of the species probably occur- red. Our stratigraphic control, understanding of the processes by which the deposit accumu- lated, and framework of radiometric dates are not yet sufficiently precise to define such fluc- tuations in detail. However our data do suggest a decline in mammals not known to inhabit forest This trend appears to have been well established about 12 000 years ago and may have continued into the Recent. At all events, most of this group of mammals were locally extinct before European man arrived. To account for the changes we have observed it is not necessary to postulate any major dif- ference between the observable present and inferred past vegetational boundaries near Devil’s Lair, and hence any marked climatic difference between late Pleistocene and modern time. But equally our data are consistent with climatic changes of some kinds, such as in- crease in total rainfall or in summer rainfall, in which case our findings would match in trend those reviewed by Fairbridge <1972). We are unable at present to separate climatic and non-climatic effects. As Calaby (in Mulvaney and Golson 1971), Freeman (1973) and others have pointed out, it is not easy to make palaeo- environmental inferences from mammal re- mains. Lower vertebrates, land snails, pollen or other biota more stringently controlled by cli- mate may be more useful. Our climatic model is based on analogy with distant regions and like other such models may be misleading (Ver- stappen 1970, Galloway in Mulvaney and Gol- son 1971). All we can claim is that our climatic and sea level model is consistent with our data. The first lists of mammal species present in the Devil’s Lair deposit (Lundelius 1960, 1966; Cook 1960) were improved and extended by Dortch and Merrilees (1972), and we have been able to add to the list, perhaps most notably by the inclusion of Pseudomys occidentalis. Despite these additions to the faunal list from Devil’s Lair, there are some notable absences : Canis familiaris on one hand, and species of Sthenu- rus, Protemnodon, and other large extinct mar- supials on the other. We suggest that the Dingo did not arrive in the district until after flow- stones sealed the richly fossiliferous^ deposit about 12 000 years ago, and perhaps Sthenurus, Protemnodon and other such taxa had become locally extinct before 25 000 years ago. Since we know that a considerable depth of deposit remains to be excavated, we think it possible that these extinct taxa may yet be found. If the assessment of Jones (1973) is sound and there was a major episode of ex- tinction “a geologically short time before 30 000 B.P.”, this possibility seems more likely. On the other hand, if Aboriginal occupation of south western Australia is an ancient one, and if the arrival of the Aborigines is causally re- lated to the extinction of Sthenurus, Protemno- don and the like (Merrilees 1968, 1973), then Jones’ “geologically short time” might be tens or even hundreds of thousands of years. It re- mains to be seen whether the Devil’s Lair de- posit is deep enough to contain such extinct taxa. We suggest that human occupation of the cave was sporadic and, if groups larger than single families were involved, very infrequent. Peterson (in Mulvaney and Golson 1971), Tay- lor 1972, and others have commented on the archaeological consequences of such situations, and in the case of Devil’s Lair, with its low average rate of sedimentation, we cannot ex- pect often to find the record of a single visit by a family group, which usually is “below the threshold of archaeological visibility”. We may have to be content with the mixed traces of several or many visits unless we can refine our stratigraphic understanding very greatly. Thus it may be difficult to detect any seasonal rhythm in the use of the cave unless this was very regular and of long standing, and it may be difficult even to fit Devil’s Lair into any such pattern as that mentioned by Wright (1971 — following Hale and Flannery); we have as yet made no attempt to examine our data from these points of view. Acknowledgements —Much of the preparatory sorting and identification on which this sLudy is based was made possible by a grant (E67/16743) from the Austra- lian Research Grants Committee. We had help In the field from those persons named by Dortch and Mer- rilees (1973), and in both field and laboratory from Miss S. Sofoulis and Miss H. Powell. Miss M. L. Shackley, Miss J. Henry, Dr D. J. Kitchener and Messrs C. E. Dortch and G. W. Kendrick gave use advice on special points. Mrs R. Henderson prepared the figures for publication and Mrs M. Konstantinu and Mrs G. E. Handley typed the manuscript and tables. We are grateful for this help, and finally to Dr D. J. Kitchener for his critical appraisal of the whole paper. Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 114 References Adam, D. P. (1971). — Palaeoclimatic significance of the stagnation of the Cariaco Trench, feature (London) 232: 469. Archdeacon, W. E. (1878). — Admiralty chart 1034 “Cape Naturaliste lo King George Sound”. With small corrections up to 1960. (London, Ad- miralty). Archer, M. (1973). — Phascolarctos (Marsupialia, Vow- batoidea) and an associated fossil fauna from Koala Cave near Yanchep, Western Australia. Helictite 10; 49-59. Archer, M. and Baynes, A. (1972). — Prehistoric mam- mal faunas from two small caves in the extreme south-west of Western Australia. Journal of the Royal Society of Western Australia 55: 80-89. Andrews, P. B. (1973). — Late Quaternary continental shelf sediments off Otago Peninsula, New Zealand. New Zealand Journal of Geology and Geophysics 16: 793-830. Bokonyi, S. (1970). — A new method for the determina- tion of the number of individuals in ani- mal bone material. American Journal of Anthropology 74: 291-292. Bonatti, E. and Gartner, S. (1973). — Caribbean climate during Pleistocene ice ages. Nature (London) 244: 563-565. Bowler, J. M. and Hamada, T. (1971). — Late Quaternary stratigraphy and radiocarbon chronology of water level fluctuations in Lake Keilambete, Victoria. Nature (London) 232: 330-332. Bowler, J. M., Thorne, A. G. and Polach, H. A. (1972).— Pleistocene man in Australia: age and sig- nificance of the Mungo skeleton. Nature (London) 240: 48-50. Brain, C, K. (1967a). — Hottentot food remains and their bearing on the interpretation of fossil bone assemblages. Scientific Papers of the Namib Desert Research Station 32. Brain, C. K. (1967b). — Bone weathering and the problem of bone pseudo-tools. The South African Journal of Science. 63: 97-99. Brain, C. K. (1967?-c.). — New light on old bones. Samah 9 (1): 22-27. Brain, C. K. (1970). — New finds at the Swartkrans aus- tralopithecine site. Nature (London) 225: 1112-1119. Campbell, T. D. (1939). — Food, food values and food habits of the Australian Aborigines in rela- to their dental conditions. The Australian Journal of Dentistry 43: 1-15, 45-55, 73-87, 141-156, 177-198. Chaplin, R. E. (1971). — “The study of animal hones from archaeological sites”. (Seminar Press, London). Churchill. D. M. (1968). — The distribution and prehis- tory of Eucalyptus diversicolor F. Mueil., E. marginata Donn ex Sm. and E. calophylla R. Br. in relation to rainfall. Australian Journal of Botany 16: 125-151. Colinvaux, P. A. (1972). — Climate and the Galapagos Islands. Nature ("London) 240: 17-20. Cook, D. L. (1960). — Some mammal remains found in caves near Margaret River. The Western Aus- tralian Naturalist 7: 107-108. Costin, A. B. (1972). — Carbon-14 dates from the Snowy Mountains area, south-eastern Australia, and their interpretation. Quaternary Re- search 2: 579-590. Cullen, D. J. (1967). — Submarine evidence from New Zealand of a rapid rise in sea level about 11 000 years B.P. Palaeogeography , Paleo- climatology, Palaeoecology 3: 289-298. Daly, P. (1969). — Approaches to faunal analysis in archaeology. American Antiquity 34 (2): 146- 153. Damuth, J. E. and Fairbridge, R. W. (1970). — Equatorial Atlantic deep-sea aricosic sands and ice-age aridity in tropical South America. Geological Society of America Bulletin 81 : 189-206. Dart, R. A. (1954). — The predatory transition from ape to man. International Anthropological and Linguistic Review (Miami, Florida) 1: 201- 218. Dart, R. A. (1957). — The osteodontokeratic culture of Australopithecus prometheus. Transvaal Museum Memoir 10. Davies, P. L. (1968).— An 8 000 to 12 000 years old human tooth from Western Australia. Archaeology and Physical Anthropology in Oceania 3 : 33-40. Derbyshire, E. (1971). — A synoptic approach to the at- mospheric circulation of the last Glacial maximum in southeastern Australia. Palaeo- geography, Palaeoclimatology, Palaeoecology 10: 103-124. Dortch, C. E. (1974). — A twelve thousand year old occu- pation floor in Devil’s Lair, Western Aus- tralia. Mankind 9: 195-205. Dortch, C. E. and Merrilees, D. (1972). — A salvage excavation in Devil’s Lair Western Aus- tralia. Journal of the Royal Society of IVesf- er7i Australia 54: 103-113. Dortch, C. E. and Merrilees, D. (1973). — Human occupa- tion of Devil’s Lair, Western Australia dur- ing the Pleistocene. Archaeology and Physi- cal. Anthropology in Oceania 8: 89-115. Douglas, A. M. (1967). — The natural history of the Ghost Bat, Macroderma gigas (Microchirop- tera, Megadermatidae) , in Western Austra- lia. The Western Australian Naturalist 10: 125-138. Douglas, A. M., Kendrick, G. W. and Merrilees, D. (1966). — A fossil bone deposit near Perth, Western Australia, interpreted as a carni- vore’s den after feeding tests on living Sar- cophilus (Marsupialia, Dasyuridae) Journal of the Royal Society of Western Australia 49: 88-90. Flood, J. (1973). — Pleistocene human occupation and extinct fauna in Cloggs Cave, Buchan, south-east Australia. Nature (London) 246* 303. Freeman, L. G. (1973). — The significance of mammalian faunas from Paleolithic occupations in Cantabrian Spain. American Antiquity 38- 3-44. Galloway, R. W. (1967). — Dating of shore features at Lake George, New South Wales. The Aus- tralian Journal of Science 29: 477. Gardner, C. A. (1944). — The vegetation of Western Aus- tralia. Journal of the Royal Society of West- ern Australia 28: xi-lxxxvii. Gill, E. D. (1971). — The far-reaching effects of Quatern- ary sea level changes on the fiat continent of Australia. Proceedings of the Royal Society of Victoria 84: 189-205. Glover, J, E. (1974).— Petrology of chert artifacts from Devil’s Lair, Western Australia. Journal of the Royal Society of Western Australia 57: 51-53. Gorman, C. (1971). — The Hoabinhian and after- sub- sistence patterns in South-east Asia dur- ing the late Pleistocene and early Recent periods. World Archaeology 2: 300-320. Goudie, A. S., Allchin, B. and Hegde, K. T. M. (1973). The former extensions of the Great Indian Sand Desert. The Geographical Journal 139- 243-257. Gould, R. A. (1967). — Notes on hunting, butchering and sharing of game among the Ngatatjara and their neighbours in the West Australian Desert. Kroeber Anthropological Society Papers 36: 41-66. Gould, R. A (1968a).— Chipping stones in the outback Natural History (New York) 77 (2): 42-48. Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 115 Gould. R. A. (1968b).— Preliminary report on excava- tions at Puntutjarpa rock shelter, near the Warburton Ranges. Western Australia. Archaeology and Physical Anthropology in Oceania 3: 161-185. Guilcher. A. (1969) .—Pleistocene and Holocene sea level changes. Earth Science Reviews 5; 69-97. Guiler. E. R. (1970).— Observations on the Tasmanian Devil. Sarcophilus harrisii (Marsupialia: Dasyuridae). Australian Journal of Zoology 18: 49-70. Hammond. J. E. (1933?).— “Winjan’s people. The story of the South-West Australian Aborigines”. (Imperial Printing Company, Pertn.) Hopkins, D. M. (1973).— Sea level history in Beringia during the past 250 000 years. Quaternary Research 3; 520-540. Milliman, J. D. and Emery. K. O. (1968).— Sea levels during the past 35 000 years. Science 162; 1121-1123. Morner. N-A. ( 1971) .—Eusta tic changes during the last 20 000 years and a method of separating the isostatic and eustatic factors in an uplifted area. Palaeogeography , Palaeocliviatology , Palaeoecology 9: 153-181. Mountford, C. P. ed. (1960 ). — “Records of the American- Australian scientific expedition to Arnhem Land. 2. Anthropology and nutrition.” (Mel- bourne University Press). Mulvaney. D. J. and Golson, J. ed. (1971 ). — “Aboriginal man and environment in Australia” (Aus- tralian National University Press, Canberra.) Neumann, H. H. and DiSalvo, N. A. (1958). — Caries in Indians of the Mexican Cordillera, the Peruvian Andes and at the Ama:2on head- waters. British Dental Journal 104: 13-16. Howard. T. M. and Hope. G. S. (1970).— The present and past occurrence of beech (Nothofagus cunninghamii Oerst.) at Wilsons promon- tory Victoria, Australia. Proceedings of the Royal Society of Victoria 83: 199-209. Innskeep, R. R. and Hendy, B. Q. (1966). — An interest- ing association of bones from the Elands- fontein fossil site. Actas del V Congreso Panafricano de Prehistoria y de Estudio del Cuaternario 2: 109-124. Isaac, G. L. (1967). — Towards the interpretation of oc- cupation debris: some experiments and ob- servations. Kroeber Anthropological Society Papers 37: 31-57. Jongsma, D. (1970). — Eustatic sea level changes in the Arafura Sea. Nature (London) 228: 150-151. Jones, R. (1970).— Tasmanian Aborigines and dogs. Mankind 7: 256-271. Jones, R. (1973).— Emerging picture of Pleistocene Aus- tralians. Nature (London) 246: 278-281. Kershaw, P (1973).— Quaternary history of rain-forests in Australia. Wildlife in Australia 10: 82-83. Kitchino- J W (1963).— ‘‘Bone, tooth and horn tools of Palaeolithic man” (Manchester Univer- sity Press). Sub-titled “An account of the osteodontokeratic discoveries in Pin Hole Cave, Derbyshire”. Klein R G. (1972).— The Late Quaternary mammalian fauna of Nelson Bay Cave (Cape Province, South Africa): its implication for mega- faunal extinctions and environmental and cultural change. Quaternary Research 2 : 135- 142. Ldffler E. ( 1972) .—Pleistocene glaciation in Papua and New Guinea. Zeitschrift fiir Geomorphologie 13: 32-58. Lundelius, E. L. (I960).— Post Pleistocene faunal suc- cession in Western Australia and its cli- matic interpretation. Report of the Inter- national Geological Congress, XXI Sessiori, Norden 1960 Part IV: 142-153. Lundelius, E. L. (1966). — Marsupial carnivore dens in Australian caves. Studies in Speleology 1: 174-180. Martin, H. A. (1973).— Paly nology and historical ecology of some cave excavations in the Australian Nullarbor. Australian Journal of Botany 21: 283-316. Matolcsi, J. ed. (1973 ). — “Domestikationsforschung und Geschichte der Haustiere” (Akademiai Kiado, Budapest.) Mercer. J. H. (1972). — Chilean glacial chronology 20 000 to 11 000 carbon-14 years ago: some global comparisons. Science 176: 1118-1120. Merrilees, D. (1968). — Man the destroyer: late Quatern- ary changes in the Australian marsupial fauna. Journal of the Royal Society of West- ern Australia 51: 1-24. Merrilees. D. (1973). — Early human occupation of south- western Australia. Journal of the Royal Society of Western Australia. 56: 44-46. Partridge, J. (1967). — A 3 300 year old Thylacine (Mar- supialia, Thylacinidae) from the Nullarbor Plain. Western Australia. Journal of the Royal Society of Western Australia 50: 57- 59. Perkins, D. (1969). — Fauna of Catal Huyuk: evidence for early cattle domestication in Anatolia. Science 164: 177-179. Perkins. D. and Daly, P. (1968). — A hunters’ village in Neolithic Turkey. Scientific American 219 (5): 96-106. Perry. D. H. (1971). — Observations of a young naturalist in the lower Blackwood valley in the year 1919. The Western Australian Naturalist 12: 46-47. Quinn. W. H. (1971). — Late Quaternary meteorological and oceanographic developments in the equatorial Pacific. Nature (London) 229: 330-331. Richards, H. G. (1970). — Changes in shoreline during the past million years. Proceedings of the American Philosophical Society 114 : 198-204. Ride, W. D. L. (1970 ). — “A guide to the native mam- mals of Australia” . (Oxford University Press, Melbourne). Roe, R. (1971). — Trial excavation in a small cave. Gin- gin. The Western Australian Naturalist 11: 183-184. Schaller, G. B. and Lowther, G. R. (1969). — The rele- vance of carnivore behaviour to the study of early hominids. South western Journal of Anthropology 25: 307-341. Shortridge, G. C. (1910). — An account of the geographi- cal distribution of the marsupials and mono- tremes of south-west Australia having special reference to the specimens collected during the Balston Expedition of 1904-1907. Proceedings of the Zoological Society of London i909: 803-848. Shortridge, G. C. (1936). — Field notes (hitherto unpub- lished) on Western Australian mammals south of the Tropic of Capricorn (exclusive of Marsupialia and Monotremata) and re- cords of specimens collected during the Balston Expeditions (November 1904 to June 1907). Proceedings of the Zoological Society of London 1936: 743-749. Smith, F. G. (1973 ). — “Vegetation map of Busselton and Augusta” (Western Australian Department of Agriculture). Taylor, W. W. (1972). — The hunter-gatherer nomads of northern Mexico : a comparison of the archival and archaeological records. World Archaeology 4: 167-178. Taylor, J. M. and Horner, B. E. (1973). — Results of the Archbold Expeditions no. 98. Systeinics of native Australian Rattus (Rodentia Muri- dae). Bulletin of the American Museum of Natural History 150 (1): 1-130. Thomas, O. (1888 ). — “Catalogue of the Marsupialia and Monotremata in the collection of the Brit- ish Museum (Natural History)” (Trustees. British Museum (Natural History) London). Journal of the Royal Society of Western Australia, Vol. 58. Part 4. December. 1975. 116 Uerpmann, H-P. (1973). — Animal bone finds and eco- nomic archaeology : a critical study of “osteo-archaeological” method. World Archaeology 4: 307-322. Van Andel, T. H., Heath, G. R., Moore. T. C. and McGeary, D. F. R. (1967). — Late Quaternary history, climate and oceanography of the Timor Sea, northwestern Australia. American Journal of Science 265: 737-758. Van Geel, B. and Van der Hammen. T. (1973).— Upper Quaternary vegetational and climatic sequence of the Fuguene area (Eastern Cor- dillera, Colombia). Palaeogeography, Palaeo- cLimatology, Palaeoecology 14: 9-92. Van der Hammen, T. (1972).— Changes in vegetation and climate in the Amazon Basin and sur- rounding areas during the Pleistocene. Geo- logie en Mijnhouw 51: 641-643. Verstappen, H. Th. (1970). — Aeolian geomorphology of the Thar Desert and palaeoclimates. Ziet~ schrift fur Geomorphologie Supplement- band 10: 104-120. Walcott, R. I. (1972). — Past sea levels, eustasy and de- formation of the earth. Quaternary Research 2: 1-14. Walker, D. ed. (1972). — “Bridge and barrier: the natural and cultural history of Torres Strait” (Australian National University, Canberra, Publication BG/3). White, T. E. (1952). — Observations on the butchering technique of some aboriginal peoples. No. 1. American Antiquity 17: 337-338. White, T. E. (1953a). — Observations on the butchering technique of some aboriginal peoples. No. 2. American Antiquity 19: 160-164. White, T. E. (1953b). — Observations on the butchering technique of some aboriginal peoples. Nos. 3, 4, 5 and 6. American Antiquity 19. 254- 264. White, T. E. (1953c). — A method of calculating the dietary percentage of various food animals utilized by aboriginal peoples. American Antiquity 18: 396-398. White, T. E. (1955). — Observations on the butchering technique of some aboriginal peoples. Nos. 7, 8 and 9. American Antiquity 21: 170-178. White, T. E. (1956), — The study of osteological materials in the plains. American Antiquity 41- 401- 404. Wolberg, D. L. (1970). — The hypothesized osteodonto- keratic culture of the Australopithecinae : a look at the evidence and the opinions. Current Anthropology 11: 23-30, with com- ments and reply, 31-37, Wright, G. A. (1971). — Origins of food production in south-western Asia: a survey of ideas. Cur- rent Anthropology 12 : 447-470 Journal of the Royal Society of Western Australia. Vol, 58, Part 4, December, 1975. 117 APPENDIX I Numbers of Individuals of momma! species ond otner vertebrofe foxo TRENCH 6 Sfrofigraphic divisions ond subdivisions with rodlocorbon dote - yr B.P. o u S o V “O «I Floor of cove First dork earthy loycr (Levels 1,2) Cove eorth and flowstone complex Brownish eorthy loyer (Level 9) - Pole bond (Level 3) - Flecked lens (Levels 4,5) - Mixture including ports of subdivisions obove ond below - Second dork eorthy loyer (Level 6) - Mixture including ports of subdivisions obove ond below - Cove pearl ond bone loyer Interfingering with flowstone (Levels 7,8) 54 54- 77 77- 87 87-c. 97 1+1? 97-c. 107 1 107-C.109 2 4+1? - top (including Heorth 1) - upper middle - lower middle - bottom 1) C.109- 131 2 1 3 2 131- 141 5 2+1? Totol 7 5+1? 2 141- 151 5 2+1? 1 151- 161 3 161- 171 6 1? 2 1 Total 14 2+2? 2 2 171- 181 5 2 181- 191 4 191- 201 4+1? 2 201- 211 4 1 1 211- 221 5 1 2 221- 231 3 1 1 2 231- 241 5 1 1 7400 - 350 241- 251 5 2+1? 1 251- 261 5 1 1 Totol 40+1? 2 1 11+1? 5 261- 271 3 3 1 pit in SE corner of trench) 271 -c .290 2 1 3 Totol 5 4 4 ■ Brownish eorthy layer. Level 9) c .290 Unexcovoted deposit ot least 400 3 1 1 6 4 14 1 I 1 2 8 1 2 2 2 2 2 1 9 1 I I 1 1 3 3 1? 1 9 3 1? 1 2? 4 11 12 3 20 27 7 27 10 + 2 ? 2 ? 1 8 2 8 9 2 16 3 8 2 4 5 1 1 7 4 1 16 4 12 14 1 3 23 7 1? 7 4 5 4 1 6 5 5 3 3 3 5 3 6+1? 4 5 3 1 1 6 2 1? 18+1? 11 13 10 2 1 17 10 7 2 3 5 1? 1 4 4+1? 5 1 3 3 3 4 7 4 6 4 1 3 9 5+1? 5 3 4 2 1? 2 6 3+1? 7 2 5 7 1+1? 2 7 4+1? 4 2 4+1? 3 1 7 8 8 3 7 8 5 12 6 7 2 4 8 3 7+1? 2+2? 6 1+1? 3+1? 5 2 7 3 56 20+1? 39+2? 44 4+2? 17 .62+1? 39+6? 5 1 5 6 2 1 9 2 7 1 6 7 2 5 2 12 2 11 13 2 3 14 4 Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975 118 recovered from excovoMons mode in Devil's Loir In 1972 ond 1973 TRENCH 6 I? 1 2 1+1? 1 1 5 4 1 3 1 1 24 1 3 2 3 25 7 6+1? 7 5 57 3 3 4 6 9 1 5+1? 1+1 ? 1? 2 4 1 1 1 8+1? 4+1? 4+1? 8 13 1 2 1 6+1? I 2 3 1 4 2+1? 2 2 6 1 2 2 3 16+1? 1 3+1? 6 7 3 ■1 3+1? 4 1? 1 2 5 2 1 5 1+1? 4 1 1? 2 1 1 1 ? 2 7 3 3 3 2 1 1? 1+1? 2 3 1 6+1? 2 1 5 3 2 1 7 1 5 5 1 3+1? 2 2 3 I 38+3? 3+1? 9+2? 28 18+2? 9 6+] ? 5 2? 7 1 3 2 3 4 1? 1 8 2+2? 3 11 1+1 ? 1 4 5 5 11 16 16 5 7 9 1 2 2 5 1 12 1 12 1 3 9 9 26 47 49 1 4 8 10 4 3 4 14 5 7 12 24 3 5 1 15 1? 2 2 2 10 5+1? 7 1 27 1 7 1? 2 1 1 3 6 1? 2 3 8 2+1? 6 2+1? 15 2 4+1? 6 4 11 15 3 4 7 12 1 1 4 7 7+3? 20 35+2? 78 1 5 3 9 3 14 16 1 8 17 25 1 1 2 2 2 12 0.30 40 54- 77 3 1 77- 87 12 2 P 87-c . 97 4 1 P IS 5 P 97-c .107 3 1 P 17 3 P 107-c .109 57 13 5 340 0.45 720 4 8 P 1? C.109- 131 4 30 P 131- 141 8 38 2 1? 209 0.48 435 2 28 P Ml- 151 3 12 P 151- 161 3 23 P 161- 171 8 63 3 242 0.45 540 3 11 P 171- 181 5 P 181- 191 4 6 P 191- 201 4 13 201- 211 5 13 P 211- 221 2 7 P 1 221- 231 3 10 P 231- 241 2 7 P 241- 251 2 6 P 1 251- 261 25 78 8 2 661 1.35 490 2 9 P 261- 271 1 6 P 271 -c. 290 3 15 2 148 0.24 620 Journal of the Royal Society of Western Australia. Vol. 58, Part 4, December, 1975 119 APPENDIX 1 (conHnued) Number of indlvlduols of mammal species ond other vertebrofe ta>io Otrotlgraphlc divisions and subdivisions with rodiocorbon dotes ■ yr B. P. S “ -2 ii o> c * 5 Q Z Floor of cove Dark eorthy layer Flowstone complex 12050 i 140 67- 78 1 3 3 1 4 5 9 1 4 5 Mixture including ports of divisions above ond below 2 1 1 3 1 3 1 2+1? 1? - top 11960 t 140 78- 95 3 2 3 3 1 3 4 1 95- 101 2 1 3 1 3 3 2 2 Total 5 1 2 3 6 2 6 7 1 3 7 4 - upper middle 101- 111 1 2 1 2 1 ? First orange brown 111- 121 2 3 1 1 1 2 2 eoHhy Total 3 5 , 2 3 1? 7 3 - lower middle 121- 141 5+1? 2 8 4 5 6 1 141- 161 2 4 1 3 1 6 2+1? 161- 171 1 1 1 3 2 1 8 3 171- 181 1 1 1 1 3 1 2 2 Totol 9+1? 2 14 6 10 14 1 3 24 13+1? * bottom 19000 ± 250 181- 201 201-C.212 2 1 1 3 1 1 2 1 1 3 1 1 2 1 1 1 Totol 3 1 4 1 3 4 2 3 2 Laminofed layer C.212 Light earthy layer 19250 ± 900 212- 220 220-C.231 1 2 1 1 1 1 1 1 1 • 1? 3 2 2 1 2 1? Borided loyer C.231 Second oronge brown eorthy loyer 231- 245 1 3 1 1 1 2 1 (Bottom of excavation, but not of Second oronge brown eorthy loyer) 245 TRENCH 2 .'Upper port of deposit removed prior to 1972) (Pit 2 fill, lower port only) First - bottom oronge brown earthy loyer 184- 199 1 199- 209 1 209- 219 1 219-C.223 1 1? 3 2 1 1 1 1 1 1 1 1 I 1 1 2 2 2 1 1 Total 3 1+1? 6 2 4 3 1 6 2 Lithified bond with chorcool C.223 Light eorthy loyer 223- 237 1 1 1 1 1 Second orange brown eorthy loyer 20400 i 1000 237-C.267 1 1 2 2 1 2 1 1 Lithified loyer, prominently lominoted C.267- 296 1 2 1 1 2 2 1 Third oronge brown eorthy layer 296- 316 1 Banded layer 3I6-C.339 1 3 2 2 2 1 1 Chorcool rich bond Fourth oronge brown earthy loyer 24600 - 800 C.339 340- 346 1 1? 1 1 2 (Bottom of exCQvction, but possibly not of Fourth oronge brown eorthy layer) Unexcovoled deposit 346 ot leost 460 Journal of the Royal Society of Western Australia. Vol. 58. Part 4, December, 1975 120 Petrogole no> ipeciftcoll/ idenMfied recovered from excovoHons mode in Devil's Loir in 1972 ond 1973 TRENCH 5 a Z 1? 1 1 1 2 1 3 3 5 1 9.0 0.17 170 56- 67 1 1+1? 2 6 2 5 4 7 8 4 P 67 0.18 370 67- 78 6 4+1? 2 4 16+1? 2 3 l+I? 4 1+1? I I I 1 1 2 1 2 79 0.28 355 49 0.24 205 174 0.66 265 78- 95 95- 101 101 - 111 111 - 121 121- 141 141- 161 161- 171 171- 181 181- 201 201-C.212 37 0.30 1 20 212 - 220 220 -c. 231 19 0.15 130 1 1+1? 1 2 1 2 18 0,11 160 231- 245 TRENCH 2 I ' P 14 0.02 700 1 1 1 1 1 1 1 1 1 1? 1 1 1 1 2 1 1 P P 184- 199 199- 209 209- 219 219-C.223 4 1 2 2 1? 2 1 5 2 48 0.25 190 1 1 1 2 P 10 0.07 140 223- 237 2 1 1 1? 2 2 2 1 22 0.15 150 237 -c. 267 I 1 1 1 3 9 11 1 1 25 0.13 190 C.267- 296 1? 1 1 1 1 1 2 4 1 9 0.09 100 296- 316 1 1 1 5 1 3 2 6 5 17 2 2 36 0.10 360 316-C.339 2 1 16 0.03 5 30 340 - 346 Journal of the Royal Society of Western Australia. Vol. 58, Part 4, December, 1975 121 APPENDIX 1 (continued) Nui.iOtfs of individuals of mommol species ond other vertebrote tows TRENCH 7 Strotigraphic divisions and subdivisions 0 8 a 0 c 3 0 a « u 0 o U O i> 9r 0 3 u 3 C D) V i 3 c *3 c 3 X cilloto 3 •> 0. 5 0 3 E 0 c >. u 3 o 3 > 3 0 V u 0 c 3 V ole nus a. 3 3 « 0 c 3 3 V oJ 0 o 8* V 0 a. IS u « c < 0 "c ‘i IZ SL 0 o 0 Thylocii o ■8 0 Peromel 0 u 8 -o 3 4> £ 0 o V u i 0 0 fi. Bettonq o c 0 (NW corner of Trench 7c) Floor of cave Dork eorthy Ioyer (A,B,C, A under C,E, A under E) 69 69 -c. 79 5 7+1? 4 1 9 1 6+1? 7+1? 4 6+1? 6 4 Mixture Including ports of divisions obove ond below 2 1 2 1 — 79- 87 1+2? 1? 3 3 1 3 3 1 - G,H,1 87-c. 90 2 4 7 10 8+1? 5 7 1 Flowstone complex - J - K (including Occupotlon Floor 1, Dortch 1974) c. 90 c. 91 2 4 6 6 1 6 1 11 1 2 3 5+3? 1 1+2? - L - Mixture including parts of subdivisions obove 2 1? 5 3 9 7 5+1? 3 5+1? 3+2? 1+1? Totol 7+2? 1? 13 10 1? 25 1 27 25+2? 5 16+1? 19+5? 4+3? Mixture including ports of divisions obove ond below 1 1 2 3 3 1? 3 - fop - Heorth 2 - M (including oronge ond brown mottled portions) 92- 93 93- no 2 9 2 2 1 7 11 1 1+1? 3 23 8 3 16 5 16 2 2 5 4+1? 3 26 2+1? 9+1? 1? First - Hearth z 2 1? 1 oronge brown - HP 1 1? eorthy Ioyer - "below HP" - MM(pale frioble coorse groined Ioyer) 110-C.115 3 2+2? 1? 2 2 4 2 3 4 1 1 1 5 2 4 ! - Heorth y - Sub MM 2 1? 2 2 1 1 1 1 2 1 Totol 20 2 3 22+4? 4+2? 38 13 25 31 6 12+1? 40+1? 17+3? (Bottom of excovotion but not of First oronge brown earthy Ioyer) C.115 TRENCH 8 'NW corner of Trench 8^) Floor of cove 79 1 Dork earthy Ioyer (A,C, A under C,E, A under E) 79-c. 109 Mixture including ports of divisions obove ond below 3 7+2? 5 1 6 1 5+1? 5 4 5 4 1+1? Flowstone - F (gypsum) c.109- no 1 1 1 1 1 3 1 4 2 complex - K (including Occupotlon Floor 1, Dortch 1974) 1 10- ? Totol 2 8 2 5 2 5 4 3 4 5 2 - 3 1 4 3 2 4 3 - top Hearth 2 (TRENCH 8- only) First orange brown eorthy ioyer - M - MM (pole Frioble coarse groined Ioyer) - Heorth y - Mixture including ports of subdivisions obov* - Heorth y plus Sub MM - Sub MM Totol - Mixture of subdivisions obove ond below (NW corner 116- 122 122-, 128 128- 129 18 19+2? 1 ? I 1 1 4 1 ? 1 ? 6+1? 3+1? 12 1? - upper middle - N - O - P - Q C.136 136- 149 149- 151 151- 161 'Bottom of excovation but not of First ororsge brown eorthy Ioyer) Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975. 122 recovered from excovofionj mode In Devil's Loir in 1972 ond 1973 TRENCH 7 (NW corner of Trench 7c) 6 1 1 3 8 5 3 1+1? 16 12 29 1 1+1? 11 7+1? P 135 0.19 710 69 69 -c. 79 1 1 1 2 2 2 P 2 1 1? 1 4 2 3 2 2 P 79- 87 4 1+1? 4 14 1 6 10 14 1 1? 6 6 P 1? 87-c. 90 1 2 1 3 7+1? 4 4 5 13 1? 11 7 P c. 90 3 2 2? 3 7 1? 9 8 16 1 10 8 P c. 91 10 5 2+4? 11 32+1? 5 1? 19 25 46 1 1? 1? 1 29 24 4 1? 351 0.37 950 1 2 1 2 1 1 3 4 4 2 P 1 1 1+1? 2 4 4 2 6 6 14 1 1? 1 1 1? 9 3 92- 93 10 8+1? 2+1? 8 9 12 3+1? 11+1? 22 22 54 1 1 19 36 P 93- no 1 1 1 1 1 1 2 1 1? 1 2 4 P 2 1 1 1 3 4 3 6 8 16 1 2 19 110-C.115 1 1 1 3 3 1 1 7 1 7 1+1? 1 1 1 4 3 4 2 12 2 11 18+1? 12+1? 6+2? 13 18 27 3+1? 22+2? 39 40 105 1 1? 2 1 2 1? 35 83 3 549 0.58 950 TRENCH 8 (NW corner of Trench 8^ 1? 1 3 5 5 5 2 21 79 0.31 70 79-C.109 3+2? 3 1+1? 2 8 5 1 2+1? 19 25 34 1 1 1 9 5 P 2 1 1 3 11? 3 4 6 2 1 P ' o i . 1 1 1 2 2 6 2 1 1 7 14 18 1? 1+2? 1 7 4 P 1 1 no- ? 3 1 3 3 9 2 2 1? 1 10 18 24 1? 1+2? 1 9 5 2 1 1 108 0.05 2000 2 1 1 2 3 3 3 3 7 15 24 3 6 9 P ■ (NW corner 2 1 ? 1 8 2 4 5+1? 8 24 9 1? 19 26 46 98 1 1 6 22 36 1 2 2 3 1 2 1 2 1 1 3 2 7 1 36 1 1 1 1? 1 3 1 1? 2 4 10 1 44 1 1? 1 1 2 1 1 2 2 6 4 24 5 2+1? 4 2+1? 1 7 3 1? 2 7 8 23 6 104 319 0.29 1100 210 0.18 1200 116- 122 122- 128 128- 129 129 -c. 136 c :. 136 136-” ■ 149 149- 151 151- 161 161 Journal of the Royal Society of Western Australia, Vol. 58, Part 4, December, 1975 123 Appendix 2 Records forming: the basis for the modern mammal fauna of the Devil’s Lair district by A. Baynes The evidence presented here depends upon unpublished research on the distributions of most south westeim mammal species. Sonie have not yet been fully covered, and their status is correspondingly less certain. Unless otherwise stated all specimens re- ferred to are in the collections of the Western Australian Museum. Those in the modern mammal collection are distinguished by a simple number, usually with the prefix M; ver- tebrate palaeontological collection specimens quoted have numbers divided by two fullstops, and no letter prefix. Mammals listed in Table 3, confidently included in the modern fauna Dasyurus geoffroii. Three specimens (M1839, M1852, M1861) in the modern mammal collection of the W.A. Museum were obtained in 1934 from Forest ° N.E. of Devil’s Lair. Two specimens (M1824, M1825) sent in from Forest Grove in the same year, another (M1717) from Karridale 7 km south in the previous year were all discarded after registration. Phascogale tapoatafa. The closest specimen was ob- obtained from Forest Grove in 1933. Other smgle records from within 20 km of Devil s Lair are M2032 from Forest Grove in 1936 (discarded), M2270 from Witchcliffe in 1938, M2711 from East Witchcliffe in 1946 (discarded), M5090 from Kudardup in 1962, and M7944 from Augusta in 1967. In addition eight P. tapoatafa have been recorded from Margaret River; those for which the accession number as well as the year is in brackets were discarded: M1588 (1931), (M2707, 1946), (M2911, 1952), M4028 (c. 1960), M4293 (c. 1960), M4534 (c. 1960), M4569 (1961), M6901 (1966). Sminthopsis murina. A specimen (M206) was ob- tained in the “Caves District” in 1915, and three others (M1171, M1642. and M2059) were sent in from Forest Grove in 1929, 1932 and 1936 respectively. Tarsipes spencerae. A specimen (M1250) was sent in to the W.A. Museum in 1930 from Group 75 which is probably east of Forest Grove rather than near its postal address at that time, which was Karridale. Either way the specimen probably originated within 10 km of Devil’s Lair. Another specimen (M2397) obtained from Karridale in 1940 was not kept. Trichosurus vulpecula. A specimen (M112) from near Mammoth Cave was registered in 1914 but dis- carded; another (M6583) was collected at the same locality in 1965. A specimen (M224) was obtained from Margaret River in 1915. Psetidocheirus peregrinus. Only a single modern specimen (M5835) has been found near Devil’s Lair. It consists of a jaw picked up on the surface from near Lake Cave, about 8 km N. of Devil’s Lair, in 1963. Shortridge (1910) reported that the species was fairly plentiful near Margaret River at the time of the Ralston Expedition of 1904-1907. He collected sev- eral at Burnside. Remains of P. peregrinus are also moderately common on the surface in other caves near Devil’s Lair. Cercartetus concinnus. This species is represented as a modern mammal in the immediate area by a single record, M1715 from Forest Grove, sent to the West- ern Australian Museum in 1933 but subsequently lost. Another (M2217) was obtained from Margaret River in 1937. Its remains are found on the surface of cave deposits in the region. Also it was represented in a small cave deposit at Turner Brook, 14 S. of Devil’s Lair which was radiocarbon dated at 430 ± 160 years B.P. (Archer and Baynes 1972). Bettongia penicillata. A series of specimens (M17- M23) were obtained near Mammoth Cave in 1912. Of these M17 and M20 are still represented by specimens in the W.A. Museum. Two (M1084, M1086) were ob- tained from Karridale in 1928, and another two (M1340, M1351) were sent in from Karridale in 1930. M1351 was discarded. Macropus fuliginosus. A modern skull (M2364) was obtained from Calgardup Brook in 1939 and another (M2389) sent in to the W.A. Museum in the same year from Witchcliffe. Remains of the species are com- mon on the floors of many other caves in the area around Devil’s Lair. Kangaroos of this species were seen frequently during field work at Devil’s Lair in 1973. Setonix brachyurus. Modern specimens were sent in from Karridale as follows: M1121, M1125 (1929), M1402 (1931), and M1765 (1933). The catalogue indicates that another (Ml 15) merely labelled as from the Mar- garet River district, was collected near Mammoth Cave in 1914. Specimen 8924 originated from Margaret River in 1907. Surface specimens of the species are common in caves throughout the Cape Leeuwin-Cape Naturaliste region. Setonix brachyurus was collected at Augusta by John Gilbert (Thomas 1888). Hydromys chrysogaster. A specimen (M1685) sent in from Forest Grove in 1933 was discarded^ but three others (M6576, M6580-1) were collected alive in Mam- moth Cave in 1965. Specimen M7 was sent in from the Margaret River district in 1912, and M221 was ob- tained from the same area in 1915 but discarded. Rattus fuscipes. This species is today common in the immediate vicinity of Devil’s Lair, e.g. M8166. There is a modern specimen (44.2.15.36) from Augusta registered in the British Museum (Natural History) In 1844 (Taylor and Horner 1973): this was probably collected by John Gilbert. Mammals listed in Table 4, probably part of the modern fauna Eight of the twelve species in this group are rep- resented by specimens collected alive in the general area. Antechinus fiavipes. A specimen (M2037) was obtained in 1936 from Rosa Brook about 22 km N.N.E. of Devils Lair. Also species is abundant in the fossil fauna from the small deposit at Turner Brook (Archer and Baynes 1972). A. fiavipes remains are present on the surface in several caves near Devil’s Lair. Isoodon obesulus. Five specimens have reached the W A Museum from the Margaret River district at various times: M226 (1915), M4522 (1959), M4466 (1960) M7580 (1966), M7626 (1967). The species was also col- lected near Margaret River by Shortridge (1910). Its remains are abundant on the surface in a number of caves near Devil’s Lair, and it was present in the small deposit at Turner Brook (Archer and Baynes 1972). Potorous tridactylus. Although this species has not been collected as a live animal near Devil’s Lair its remains are very common on the surface in caves in the area. It seems likely that it was present as a member of the modern fauna around Devil’s Lair. Information obtained by Shortridge (1910) from Abo- rigines also suggests that it was present in the Mar- garet River district during the last century. Macropus irma. In some ways this species repre- sents the greatest problem of interpretation because specimens collected alive in the region may falsely suggest the species to have been a member of the modern fauna around Devil’s Lair. A specimen (M458) was obtained from Augusta in 1920, and another (M8335) along the Brockman Highway between Augusta and Margaret River in 1968. Perry (1971) includes M. irma in a list of species frequently seen alorig the lower Blackwood River in 1919. Against this evidence is the fact that although remains of the species are known from caves near Devil’s Lair the specimens nave all undergone some chemical alteration, suggesting age. Most originate from caves known to contain material of considerable antiquity. It is possible that M. irma reinvaded the area near Devil’s Lair after the first felling of the forests last century. The three rodent species in Table 4 were all Pres- ent in the deposit at Turner Brook (Archer and Baynes 1972). Both Pseudomys shortridgei and Rattus tunneyi were abundant in the deposit; Pseudomys praeconis was only represented by 3 specimens in a total of about 650. However, this low relative abun- dance of P. praeconis in a cave deposit fauna is typical of this species in the southern part of its range (A. Baynes, unpublished observations). Nyctophilus timoriensis. The modern specimens col- lected nearest to Devil’s Lair were obtained about 60km away: M36 from Nannup in 1913, and M1247 and M1248 from Wonnerup in 1929. However, the journal of the Royal Society of Western Australia, Vol. 58. Part 4, December. 1975. 124 total number of modern specimens in the collection is quite small. The species is included in Table 4 on the record of its presence in the deposit at Turner Brook (Archer and Baynes 1972). Many bat species are among the mammals not yet fully covered in the survey of distributions, but the modern mammal collection of the Western Australian Museum includes specimens of four species collected near Devil’s Lair. One specimen with its year cf capture is quoted for each: Nyctophilus geoffroyi M6584 near Strongs Cave just to the west of Devil’s Lair in 1965, Eptesicus pumilus M4183 Boranup about i km east of Devil’s Lair in 1961, Chalinolohus morio M3788 in Mammoth Cave 10 km north of Devil’s Lair in 1959, and Pipistrellus tasmaniensis M4182 from Boranup in 1961. Canis familiaris. Shortridge (1936) collected five specimens of Dingo from Margaret River during the Balston Expeditions of 1904-07 and Perry (1971) in- cludes it in the list of species frequently seen on the Lower Blackwood River in 1919. The W.A. Museum has only a single modern specimen from the Mar- garet River district, M4204 collected in 1958. However, the locality of even this one is suspect. The species is abundantly represented by remains on the surface of deposits in caves near Devil’s Lair. One cave has been named Dingo Cave because so many skulls have been recovered from it. Mammals listed in Table 5, possibly forming part of the modern fauna Macropus eugenii. This species is placed here mainly because of a lack of fossil specimens of young appear- ance from caves near Devil’s Lair. All those known originate from caves which include material of con- siderable antiquity. They are generally encrusted or appear to have undergone chemical alteration. The closest record to Devil’s Lair of Macropus eugenii taken alive is the specimens collected by Shortridge (1910) at Ellensbrook about 30 km to the north. W. D. L. Ride possesses copies of the data on the labels of five of these specimens which are in the British Museum (Natural History). The label attached to the skin of No. 6.9.1.32 bears the following note ‘^Macropus eugenii does not seem to occur to the south of Mar- garet River on the coast (according to natives)”. Ride (pers. comm.). The author of the note was probably Shortridge. There are other specimens from this general area in the W.A. Museum modern mammal collection. One from Cape Naturaliste about 65 km north of Devil’s Lair consists of only a skin to which is attached a Tunney collecting label bearing the number 06 and the locality, but no date. It was probably collected between 1900 and 1910. Three other specimens (12860-2) were collected at Lake Muir about 140 km E. of Devil’s Lair in 1912. Chalinolobus gouldii. The nearest records of mod- ern specimens are from 55 km from Devil’s Lair: M8391 at Busselton in 1969 and M10935 from Darradup in 1973. Tadarida australis. The nearest record is M5420 col- lected in 1963 at Donnelly River Mill, about 75 km E. of Devil’s Lair. Mammals listed in Table 6, not present in the mod- ern fauna of the Devil’s Lair district but recorded in the fossil fauna. It is necessary to consider the records of one of the species included in Table 6. Petrogale sp. It is unlikely that any species of Petrogale was present in the modern fauna of the Devil’s Lair district. Fossil specimens are known from a number of caves near Devil’s Lair, but all originate from deposits which are known to include specimens of considerable age. and most appear to have under- gone some chemical alteration. In addition there is one specimen which must be discussed in detail It is a skull of Petrogale sp.. at present in the verte- brate palaeontological collection (69.6.62), but form- erly in the modern mammal collection (M114). The catalogue indicates that M114 is a skull of Setonix brachyurus from “Mammoth Cave, Margaret River” collected by L. Glauert in April 1914. Specimen M115 is a skin and skull of Setonix brachyurus with the same locality data as M114. Skull 69.6.62, still bear- ing the number M114, retains the remains of dried flesh (e.g. the left eardrum) it also shows starred fractures of the cranium behind the left orbit. On this evidence it seems likely that the specimen was collected as a live animal, probably with a shot gun, and that it is not a fossil. The nearest locality from which living specimens of a Petrogale sp. have been recorded is Dale River near Beverley (Shortridge 1910). This is about 240 km N.N.E. of Mammoth Cave. When this gap is considered in conjuction with the absence of young fossil material from caves near Devil’s Lair, it appears likely that specimen 69.6.62 is the result of misassociation of specimen and data during prepara- tion or storage. Appendix 3 Investigation of degree of overestimation inherent in our methods of obtaining ‘'minimum numbers” of individuals The entries in Appendix 1 represent our esti- mated “minimum number” of individuals for each taxon for each stratigraphic unit specified. In the case of Trench 6, however, excavation was made by 2 cm spits from 109 cm to 271 cm, but for convenience in tabulation, we show the data grouped into 10 cm intervals. Grouping was made by adding the 2 cm spit records for each taxon. In order to investigate the maximum degree of overestimation of “minimum numbers” of indi- viduals inherent in our methods, we chose the 211 cm to 271 cm interval in Trench 6 because it both contained sufficient numbers of specimens of very large and very small animals, and ap- peared to be a stratigraphic continuum. This interval was excavated in horizontal spits 2 cm thick in what we later judged from the sections and from our observations as excavation pro- gressed to be a continuum of gently dipping sedi- ments. Not only were our spits very small but they transgressed the dipping bedding planes, so that different beds might be represented in the same spit. We assembled all the bone material of three species, catalogued or otherwise, which was ad- mitted to our analysis from this 211 to 271 cm interval in Trench 6. The first of these species was one of the smallest sized mammals repre- sented, Sminthopsis murina, the second was a species of intermediate body size, Pseudocheirus peregrinus and the third was the largest species (with the possible exception of man) so far re- corded in the deposit. Macropus fuliginosus. In each case, the number of individuals was esti- mated without regard to depth, as though com- plete vertical mixing had occurred. This was in contrast to the assumption on which the appen- dix tables were prepared, namely that no verti- cal mixing had occurred between 2 cm spits. It is to be expected that the true number of indi- viduals represented in the deposit would lie be- tween these two extremes. The degree of over- estimation in the appendix tables would be expected to be least for Sminthopsis, intermedi- ate for Pseudocheirus and greatest for M. fuliginosus because a small thickness of sedi- ment would completely cover all Sminthopsis bones, most Pseudocheirus bones, but few Macropu^ fuliginosus bones. Vertical mixing would be greatest for this last named species because larger bones would be more likely to be dislodged by scuffling human or other feet, whereas small bones would be more likely to be trodden in. The number of individuals of Sminthopsis murina recorded in the Appendix 1 tables be- tween 211 and 271cm in Trench 6 is 10. When we considered the specimens as a single sample Journal of the Royal Society of Western Australia, Vol. 58, Part 4. December, 1975. 125 irrespective of depth, there proved to be 5 left dentaries, 6 right dentaries, 2 left maxillae and 2 femora. Some of these were incomplete, but no two tooth bearing fragments from the same quarter could have represented the same animal. So on these grounds at least 6 individuals were represented. Appreciable tooth wear was discer- nible on 3 of the left dentaries but not on any of the 6 right dentaries, so that at least 3 aged animals and 6 young animals must have con- tributed to the sample, making 9. Of the 29 spits involved, only two pairs of ad- jacent spits contained S. murina. These pairs and the other 6 spits containing S. murina were separated from one another by at least 2 cm and up to 8 cm. On depth grounds alone at least 8 individuals might be expected. Thus it would appear that our methods did not result in great overestimation of the num- bers of individuals of S. murina. By extrapola- tion to other depth ranges and other very small species, we suggest that the numbers in the appendix tables are not greatly overestimated. The specimens of S. murina involved in this lest were 73.8.638, 746, 802. 892, 927, 958, 969; 73.9.24, 41, 37: 73.12.309, 310. The number of individuals of Pseudocheirus peregrinus recorded in the Appendix 1 tables between 211 and 271 cm in Trench 6 is 37. By assembling the specimens irrespective of depth, we could demonstrate the presence of only 21 individuals, the key anatomical structure in this case being left dentary fragments with third molars or their alveoli. Thus our methods may overestimate the numbers of individuals of this, and presumably other animals of intermediate size, to an appreciable extent. In this case we had no guide from depth considerations because all but one of the 29 spits contained P. pere- grinus. The specimens involved were 73.8.584, 585, 602, 625, 653, 654, 671, 713, 732, 751, 781, 782. 808, 809, 826, 855, 856, 876. 877, 895, 909, 910, 930. 931. 948, 962, 972, 988; 73.9.18, 28. 47. 48, 67. 75. 89; 73.12.332, 339, 343. The number of individuals of Macropus fuligi- nosus recorded in the Appendix 1 tables between 211 and 271 cm in Trench 6 is 17. From the specimens assembled as one sample, irrespective of depth, we could demonstrate the presence of only 6 individuals, the key anatomical structure in this case proving to be left upper first incisor teeth. In order to judge the possible effect of excava- ting by 10 cm instead of 2 cm spits, we then assembled the M. fuliginosus specimens from the 211 to 271 cm interval in Trench 6 in groups from 211 to 221 cm, 221 to 231 cm, and so on. If we had in fact excavated Trench 6 in these 10 cm spits, we would have recorded only 9 indi- viduals of M. fuliginosus, thus probably over- estimating, but much less seriously than by ex- cavating as we did in 2 cm spits. The M. fuliginosus specimens involved were 73.8.591, 608, 644. 738. 763, 793, 815, 836, 863, 883, 900, 901 917. 935, 977; 73.12.308, 333, 335, 372. As a further check on the degree of over- estimation of M. fuliginosus, we considered specimens from spits of about 10 cm thickness excavated from Trench 5, between 78 cm and 212 cm. We believe these range through the same major stratigraphic division as the interval from Trench 6 previously considered, though the Trench 5 sample probably represents a larger proportion of this stratigraphic division. There were 9 individuals of M. fuliginosus recorded for the 78 to 212 cm depth interval in Trench 5. When we assembled the specimens as a sample irrespective of depth, we could demonstrate the presence of only 4 separate individuals, with anterior molariform teeth providing the best guide to numbers in this particular sample. The specimens concerned were 73.9.354, 381, 409, 429, 430, 460, 532, 677, 703, 783; 73.12.499. As a final check on overestimation of M. fuliginosus, we assembled all the material from Trench 6, not only of the categories used to as- semble the data for the appendix tables but also any other post cranial material attributable to M. fuliginosus by reason of its size or form. Ignoring all consideration of depth or strati- graphy, we could demonstrate the presence of only 17 individuals of M. fuliginosus, compared with 50 estimated by adding the entries for the various stratigraphic subdivisions, as in Appen- dix 1. By considering separately the major strati- graphic subdivisions of Trench 6, but using the full sample of M. fuliginosus rather than the partial sample on which Appendix 1 is based, and omitting stratigraphically mixed samples, we reached the following numbers: — First dark earthy layer Pale band Flecked . . . lens Second dark earthy layer Cave pearl and bone layer Brownish earthy layer — top Brownish earthy layer — up- per middle Brownish earthy layer — lower middle Brownish earthy layer — bottom estimates of minimum 1 (not included in Ap- pendix 1 tables because based on post cranial elements not admitted to our analysis) 1 (ditto) 1 (ditto) 2 (cf. 1 in Appendix 1 tables) 2 (cf. 3 in Appendix 1 tables) 7 (cf. 8 in Appendix 1 tables) 2 (cf. 6 in Appendix 1 tables) 8 (cf. 28 in Appendix 1 tables) 2 (cf. 3 in Appendix 1 tables) The lower middle subdivision of brownish earthy layer, for example, was excavated in 43 two centimetre spits and 1 four centimetre spit, and minimum numbers for each species esti- mated for each spit. The 28 M. fuliginosus indi- viduals recorded in Appendix 1 for this sub- division represent the sum of the numbers recorded for each spit; this is in marked contrast with only 8 individuals estimated by treating the subdivision as though it were a discrete strati- graphic entity. (There may be some connection between the high degree of overestimation in this stratigraphic subdivision and the low degree of bias imposed by the collecting agency, implied in Table 1, but if so, we have made no study of any such connection.) We conclude that our methods may lead to overestimation, which becomes more likely with increasing size of, the animal concerned and de- creasing thickness of the excavation unit sampled. journal of the Royal Society of Western Australia. Vol. 58. Part 4, December, 1975. 126 INSTRUCTIONS TO AUTHORS Contributions to this Journal should be sent to The Honorary Editor, Royal Society of Western Australia, Western Australian Museum, Perth. Papers are received only from or by communication through, Members of the Society. The Council decides whether any contribution will be accepted for publication. All papers accepted must be read either in full or in abstract or be tabled at an ordinary meeting before publication. Papers should be accompanied by a table of contents, on a separate sheet, showing clearly the status of all headings; this will not necessarily be published. Authors should maintain a proper balance betvfeen length and substance, and papers longer than 10,000 words would need to be of exceptional importance to be considered for publication. The Abstract (which will probably be read more than any other part of the paper) should not be an expanded title, but should include the main substance of the paper in a con- densed form. Typescripts should be double-spaced on opaque white foolscap paper; the original and one carbon copy should be sent. All Tables, and captions for Figures, should be on separate sheets. Authors are advised to use recent issues of the Journal as a guide to the general format of their papers, includ- ing the preparation of references; journal titles in references may be given in full or may follow any widely used conventional system of abbreviation. Note that all illustrations are Figures, which are numbered in a single sequence. In composite Figures, made up of several photographs or diagrams, each of these should be designated by letter (e.g. Figure 13B). Illustrations should include all necessary lettering, and must be suitable for direct photo- graphic reproduction. To avoid unnecessary handling of the original illus- trations, which are usually best prepared between H and 2 times the required size, authors are advised to supply extra prints already reduced. Additional printing costs, such as those for folding maps or colour blocks, will normally be charged to authors. It is the responsibility of authors to adhere to the International Rules of Botanical and Zoological Nomenclature. Palaeontological papers must follow the appropriate rules for zoology or botany, and all new stratigraphic names must have been previously approved by the Stratigraphic Nomenclature Com- mittee of the Geological Society of Australia. Thirty reprints are supplied to authors free of charge, up to a maximum of 60 for any one paper. Further reprints may be ordered at cost, provided that orders are submitted with the returned galley proofs. Authors are solely responsible for the accuracy of all information in their papers, and for any opinion they express. Journal Volume 58 of the Royal Society of Western Australia 1975 Part 4 Contents 10. Mammal remains from the upper levels of a late Pleistocene deposit in Devil’s Lair, Western Australia. By A. Baynes, D. Merrilees and Jennifer K. Porter. Editor: A. J. McComb The Royal Society of Western Australia, Western Australian Museum, Perth WILLIAM C. BROWN, Government Printer, Western Australia 49537/7/75—625