Berry (1968:112-113) recognised theoretical difficulties with the system These are exaggerated by

Pietrusewsky's inclusion of certain traits violating R. Berry's original principles and not included in Berry and Berry (1967), notably tympanic thickening, occiput ridge (= transverse occipital torus) and malar tuberosity, which are continuous and not discrete or quasicontinuous

traits. Likewise certain of Pietrusewsky's inclusions depend for their expression on complex developmental interactions involving many genes e.g. rocker jaws

(Houghton 1977) or the sharpness of the subnasal (lower narial) margin which appears to be strongly related to

bimaxillary breadth and the degree of prognathism (Krogstadt 1973:52). A more general problem is that some traits used are recognised as racial 'markers' (e.g. inca bones

(= ossicle at the lamdba) for Mongoloids (Coon 1962:459) and palatine torus for Australoids (Larnach and Macintosh 1966:82)), other traits act patchily as racial markers

(e.g. metopism which occurs most frequently amongst Caucasoids and Bolivian Indians (Larnach and Macintosh 1966:16; Arensburg, Goldstein and Nathan 1977) and

mylohyoid bridges (used by Pietrusewsky) which occurs most frequently amongst Khoisans and Amerindians/Eskimos

(Lundy 1980:47)), and still other traits discriminate

between populations only at the level of microdifferentiation (this has been shown for auditory exostoses (Larnach and

Macintosh 1970:31-32; see also Pietrusewsky 1974:110)

and very likely will be found true of other of the traits). Unless the pattern of interpopulation distribution of the individual traits is known beforehand, the generated

statistical distances will be difficult to interpret, particularly at the level of macrodifferentiation. Pietrusewsky (1974:30,32; 1976:67-68) and Kellock and

Parsons (1970b:235-239) have difficulty deriving meaningful answers from their results .

The other is the system of Larnach and

Macintosh (1966,1970,1971) which explicitly concentrates on continuous traits, specifically selects those traits showing marked racial contrasts (here Australian Aborigines cf. northeast Asians and Caucasoids) in expression, and scores the grade expressions of these traits (called 'race discrimination characters') such that overall morphological racial contrasts are reflected by the member specimens' total scores (here Australian Aborigines score high,

northeast Asians and Caucasoids score low). (The Larnach and Freedman (1964) and Larnach and Macintosh (1971)

sexing system, for the Aboriginal cranium and mandible respectively, follow the same numerical discrimination

approach based on 'sex discrimination characters'). Recently Dr J. Kamminga (now on the Division of Prehistory, La Trobe University, Melbourne) observed the expression of the

cranial race discrimination characters for a number, of circum-Pacific series and prehistoric southeast Asian specimens, which (Oriental-Pacific) data he loaned to the author for the present study. The author applied the

Larnach and Macintosh system to a series of Solomon Islands crania (chapter 7) and Melanesian mandibles (chapter 8) so that

a fair body of data is available in the recent Oriental- Pacific context.

Application of the cranial race and sex

discrimination characters is facilitated by a set of standard casts, supplied by the Department of Anatomy, Sydney

University, which designate the grade boundaries of most of the

characters not metrically defined (these were used by

Kamminga and the present writer). This greatly

diminishes interobserver variability though borderline expressions are open to differing interpretations.

However casts are not available for those race

discrimination characters graded 'absent', 'trace' or

'distinct' , no doubt based on the belief that such traits

will tend to be either absent or present, with a 'trace' grade for those (few) cases where the trait is just

slightly expressed ( 'quasi-discrete1 traits if you like).

While this may be generally true for the cranial series observed by Larnach, there is no guarantee that it

applies for southeast Asian and Palae-Melanesian series, where expressions intermediate (here 'trace') between

modal Australoid and northeast Asian expressions may be confused with either 'absent' or 'distinct' expressions in proportion to the lack of sharpness between absent and

distinct expressions. For instance the contrast between

a smooth frontal and one showing some median ridging, or a 'smooth' palate and one showing some torus development, is ill-defined and interobserver comparability more open

to guestion. On the other hand the distinction between

a smooth and a gabled vault, or a sharp and a rounded molar orbital border, is more discrete and interobserver comparability thereby enhanced.

Greater initial difficulties faced the interpretation of Larnach and Macintosh's (1971) mandibular characters,

as most of these are truly continuous traits and no casts

are available. At the time of data collection the

whereabouts of L a r n a c h 's original data were unknown so his grades could not be checked directly against the original

s p ecimens. My approach to the problem was to take 20 of

the eastern Australian Aboriginal mandibles given an overall score by Larnach and Macintosh (1971:32), rank them

according to their expression for each of the race

discrimination characters, then assign grades which appeared to correspond with L a r n a c h 's grades on the basis of the

variability apparent. A preliminary total score for each mandible could then be computed following Larnach and

Macintosh's (1971:29) method of scoring. By shifting the

grade boundaries appropriately, a total score either identical or very close to Larnach and Macintosh's was arrived at, hence what may be taken to be the original

range of expression for their race discrimination characters

was closely approached. By further assuming that the 20

mandibles chosen were a representative sample of all the Australian Aboriginal mandibles they studied, grade

boundaries were also established for the important trait

torus lateralis superior (see chapter 8). The sex

discrimination characters sulcus intertoralis and

anterior marginal tubercle could also have been analysed in the same way, but were overlooked; omission is not serious as sexual dimorphism should be apparent even if the grade boundaries are slightly shifted, and present modifications to the Larnach and Macintosh sex

discrimination system diminished the importance of these traits (Appendix C).

Some interobserver variability is part and parcel of the Larnach and Macintosh system, as it is for all morphological traits except the most discrete (e.g.

supernumerary foramina and cranial bones). There are two

telling advantages of the Larnach and Macintosh approach

for the present study. First each cranium or mandible

retains its identity as a single specimen so that a single specimen may be compared with the ranges for cranial series. Second, the technigue is explicitly designed to determine those traits important at the level of macrodifferentiation within the Oriental-Pacific context, for the interpretation of

specimens of unknown race. One does not end up with a

'result' which must then by converted into some kind of biological interpretation.

One endocranial trait, the type of middle meningeal artery, was researched in the present study; indeed middle meningeal artery type is the cranial trait

4 3.

most frequently observable on the LB material.

Giuffrida-Ruggeri (1912:402) recognised five types of reasonably common patterns; Adachi (1928), summarised in Rothman (1937:426), simplified Giuffrida-Ruggeri's scheme by ignoring the point at which anterior and

posterior branches split. Three types remained, Type I and Type II where the middle branch arises from the

anterior and posterior branch respectively, and Type III where middle branches arise from both anterior and

posterior branches.

Now middle branches of some size very frequently extend from both anterior and posterior branches, and the decision as to one or two middle

branches involves drawing a typological boundary in what is really a trait of continuous variation. (Though

Types I and II cannot be confused, Type III may be