EVOLUTION OF THE CLUB TYPES. III 
considered specifically with others along the same line. The expan- 
sion of the shaft in this piece (Plate VIII, fig. e) consists of artfully 
carved flanges alternating with the flanges of the head and providing 
a hexagonal section. ‘This alternation of shaft-flanges is observed, 
though less carefully executed, in the rootstocks (Plate V, figs. 5 and 6), 
3175, 3100, 2482, and 3782 a. ; 
We have seen that metamorphism is clearly recognizable in the ula 
of the ball-head; that with no great difficulty it may be traced through 
the wheel-heads. The patterned type can only be regarded as a 
secondary evolution after the stone idea has passed from knowledge; 
decoration has now overwhelmed structural detail. Yet we find the 
retention of a head, which at base is flanged, and of the distal ring and 
knob without alteration. 
In the discussion of the ula we have included the discussion of the 
single unit of the rootstocks (Plate V), which suggests evolution on a 
stone-head type. ‘This is the carefully worked distal projection in 
the flange-shaped species. In 5 out of the 7 pieces in which the 
head is characterized by persistence of the rootlets we find the same 
suggestion of distal projection, in certain instances very formally 
worked out, this being particularly apparent in figure 3, and in 4 less 
completely done. ‘This is not a structural necessity of the timber 
source of any of these rootstocks; it seems to have been carved in the 
flanged-head pieces from a stone original, thence to have been extended 
to the sapling motive by unthinking imitation. 
We have next to consider the types in which the shaft exhibits a 
curve at or near the head. In this group we assemble as to this one 
detail the pandanus and the lipped clubs, and therewith we associate 
the axe-bit club, although the curve in this case is established rather 
through the element to be identified as socket than by the shaft 
itself. As has been already noted in the consideration of the sickle 
club of Niué, we are to seek the nearest relative of this type in Arossi 
of San Cristoval, in the Sclomons, and to find more distant kin in 
New Guinea. Niué, at the remotest eastern limit of Nuclear Poly- 
nesia, may serve to establish for us the fact that a culture of the curved 
club has passed through this province. With this important support 
we feel justified in recognizing in the curved shafts of Fiji an inter- 
mediate locus of the type. 
In the pandanus the critical details are the curve of the shaft in 
immediate proximity to the head, the head of several (5 to 8) rows of 
rather carefully cut knobs or spines, the distal plate usually with a 
limiting raised ring, the distal cone. We figure this type in full in 
Plate II, figure d, and in head detail of 3 pieces in Plate VI. 
Although the head receives in this type such a high degree of detail 
as to warrant the Fijians in interpreting it as based upon the motive 
of the pandanus compound fruit, which I have already shown to be 
