~ 
7 
process continues until there is only one cluster. At that 
linkage, the two most distant members were 'Crafting' and 
'‘Aesthetics', which were 2.62 judgement units apart. 
Size of Recall Exemplars 
of the exemplars from both recall tasks, 38 were 
non-material and could not be categorized for size. Table 
10 shows how the 2362 categorized exemplars were distributed 
across the two recall tasks and the four size categories. 
The effect of recall task on exemplar size was statistically 
significant [chi square (3,N=2362) = 255, p < 01]. 
Exemplars of owned things tended to be smaller than 
exemplars of things not owned (Kendall's tau = -0.34, p < 
.01). About 9% of the variance between the recall tasks was 
accounted for by the size measure (eta = .33). .Again, to 
examine the influence of territorial items on the size 
effect, all items in the territorial categories were 
deleted. The size effect was reduced, but was still 
signiticant [chi-square(3,N=1662) = 163, p < .01], 
accounting for about 6.5% of the recall variance. For the 
recall of owned things alone, the size measure had a 
significant relationship [chi-square(3,N=1189) = 16.7, P < 
.01], as it did for the recall of things not owned 
[chi-square (3,N=1173) = 453, p < .01]. With items in 
territorial categories deleted, the size effect remained 
significant for both the owned things [chi-square(3,N=951) =