EVOLUTION OF THE CLUB TYPES. 123 
when it was sought to extend them to another unit. The obsidian 
prototype is the only one in all my acquaintance with South Sea 
motives of design and with the handicraft in which they find expres- 
sion, which accounts for all these units. 
There remain for our consideration the pieces with saw-teeth on the 
cutting-edges. ‘There appears to be a series associable by the number 
of edges carrying the teeth. A uniserial type is represented in 
ANSP 15744 (Plate IV, fig. 6), a horned club with large and dis- 
tinct teeth on the edge opposite the horn. That this type may become 
biserial appears on a club in Kramer’s possession (Samoa, II, 216 0). 
The talavalu clubs are all biserial (Plate III, d, e, f). Then follows a 
triserial piece (fig. a) and a quadriserial piece (fig. g); in each of these 
we find an additional element in an alternating series of much smaller 
teeth of a purely decorative purpose. The type engages somewhat 
with certain of the banded lapalapa, for Kramer figures (Samoa, II, 
216 c) a form which has developed its multiple angular banding effec- 
tively into teeth and exhibits a satisfactory evolution from the museum 
piece 2273. We must regard all the toothed clubs as metamorphs 
upon the weapon of the sawfish. We find in the museum one of these 
fish appendages which has been rived longitudinally, edges slightly 
rounded for grip—a very effective implement of bodily harm any- 
where within the limits of mayhem and murder. ‘The uniserial and 
the multiserial clubs of this type represent various arrangements of 
such rived saws; the biserial represent the saw in its natural order. 
(Parkinson, 420-5, figures a mounted saw from Wuvulu and Aua.) 
But when we pass beyond the simplicity of this identification of the 
saw-teeth we find difficulties in its adjustment to the shaft. In the 
talavalu three of the pieces (figs. d and e) end in a strongly marked 
pyramidion, and 2272 has a still more remarkable terminal in the form 
of a square plate. But the natural end of the saw is slightly curved 
in the arc of a circle and there is no increase in thickness which might 
suggest the median expansion of the wooden pyramidion. In the 
pieces figured as a, f, and g, the end of the head is distinctly cupped, 
quite in the opposite sense from the pyramidion. In the biserial clubs, 
except figure /, there is a distinct shoulder on the shaft out of which 
the blade arises, and in figures a and g we find at this point distinct 
plates respectively triangular and quadrangular. It seems best to 
regard the use of this motive as already in a secondary stage and con- 
_ ditioned by added ornament, which is not in nature associated with 
the saw. ‘Thisis most distinctly the case of the horned club ANSP 
15744. ‘This is readily comprehensible as the addition of the saw 
metamorph upon a type of club already established in some other 
motive, and in ANSP 14522 we are able to discover this type with- 
out the saw addition. 
