54 RUDDOCK 
Table 3. Values listed by Duerden et al (1987) for the source areas of this study. 
ee 
Locality Awana Te Ahumata Mayor I Green 
nnn 
SiO, 40 41 35 
TiO, 480 — 730 420 — 620 1000 — 1500 
Al,O; 6.03 — 7.73 6.01 — 6.87 4.25 — 5.01 
MnO 220 — 330 210 — 300 730 — 1060 
CaO 0.45 — 1.16 0.37 — 0.62 0.06 — 0.17 
Na,O 1.56 — 3.12 2.48 — 2.74 3.61 — 4.09 
K,O 3.2 — 4.6 3.9 — 4.3 3.1 —3.8 
Trace elements (ppm) 
Rb 151 — 221 192 — 226 141 — 190 
Sr 10 — 70 10 — 30 0 — 40 
Zr 110 — 170 90 — 140 990 — 1200 
Y 10 — 50 20 — 40 80 — 160 
eS 800 OOO 
values in Table 4 for Ba, Sr, V, Cr, Ni and Cu are rather meaningless since they did not 
occur in concentrations above the detection limits for all samples. A further weakness 
of the Euclidean distance method is sensitivity to observation order. However, the 
clustering results were identical regardless of how the data was ordered. 
Discussion 
Although the original qualitative sort is open to error and the quantitative 
sourcing is of small sample size, a single source (Mayor 1) for the green obsidian seems 
valid. The grey grouping, however, is not as easy to substantiate because, while some 
grey source groups can be eliminated as unlikely, a number of possible sources still 
exist in other clusters, as shown by Moore’s (n.d.) chart. This lack of positive sourcing 
is accentuated by the small percentage (2.2%) of grey obsidian analysed in this study. 
Logically a Great Barrier I source would fit, being the closest source of obsidian to this 
site, but, as Seelenfreund-Hirsh (1985) asserts, distance to source may not have always 
been the primary determining factor. However, as Seelenfreund-Hirsh (1985) 
concludes, temporary sites (as site RI1/898 most probably was) tend to have a smaller 
number of sources represented. The inference of only two sources may, therefore, best 
fit within the overall New Zealand situation. 
The change in the relative importance of the two (?) sources with time provides a 
further point for discussion. Over time the total percentage of green obsidian changes 
— from 30.8% in Occupation I to 21.3% in Occupation I (Furey 1986, Table 1). No 
figures were calculated for this study due to the large percentage (33.8) of obsidian in 
the no layer category. Furey’s findings are in line with the work of Leach & de Souza 
(1979), who note the relative reduction, with time, of obsidian from the Mayor I 
