EVOLUTION OF THE CLUB TYPES. I1Q 
inclining inward toward the lower edge. This shoulder in 3362 is set 
up by a strong flange, is not apparent on the upper edge, is sharply 
angled on the lower edge, and faces the next unit with a square-cut 
face. In 2478, a piece far less well executed, this shoulder appears on 
both edges of the shaft and continuously around it; its forward aspect 
is not so distinctly vertical to the blade, but meets it withaslant. In 
3361 the shoulder is reduced to an obscure swelling of the shaft-end, 
tumbling home with a rounded aspect toward the blade. This piece 
is of very crude workmanship, carries an obscure extra unit which does 
not appear in the better-executed pieces, and exhibits several puzzling 
variations. 
Between the shoulder of the shaft-end and the blade is a second unit 
set angularly with the extent of the shaft. This is clearly a socket 
designed to hold the blade in one function and in another to attach it 
to the shaft. 
This socket suggestion is not only the sole possible interpretation of 
the structural form, but we find it most interestingly confirmed by a 
similar Melanesian form in the Admiralty Islands, remote in space but 
upon a well-established line of migration. From a paper by H. N. 
Moseley (Journal of the Anthropological Institute, May 1877) is 
extracted the following description: 
“The obsidian lance-heads are secured in a socket of wood attached to the 
end of the shaft by means of a cement and by being bound round with fine 
twine. ‘The socket is hollowed out in a separate piece of wood, and in order 
to facilitate the scooping out process two slots are usually cut in the faces 
of the socket. The shaft of the lance is spliced into a V-shaped slot in the 
lower part of the socket piece. A rounded strengthening piece is retained 
in the socket piece between the actual socket and the narrowed part of it in 
which the slot for the shaft is cut. A very hard and solid gum is used to bed 
the lance-head in its socket and the shaft in its slot, and to mass together the 
turns of fine twine which secure the whole. In some lances the entire socket- 
piece and the turns of binding twine are concealed by an even thick layer of 
the gum, whilst in others the gum is used more sparingly and the turns of 
twine and the wood of the socket-piece are exposed to view. In the former 
class of lances ornamentation is effected by patterns being incised in the layer 
of gum, and these have no Coix lachryma seeds attached to them. In the 
latter class the upper turns of twine are arranged in diagonals, etc., separating 
the ornamental colors, and the actual wood of the socket-pieces is carved and 
colored. The gum employed is probably the same as is used for caulking the 
canoe seams, which is obtained from a brown ovoid fruit about the size of a 
goose’s egg. ‘The efficiency of the fixation of the stone head of the lance 
evidently depends mainly on this gum. The wood of which the socket- 
pieces are made is hard when dry and old, but probably much softer when 
cut in the fresh condition. . . . The socket pieces of the lance-heads 
are elaborately decorated. Some lances have a lozenge-shaped perforation 
in the socket-piece beneath the head.”’ 
Parkinson (p. 354) ascribes the gum to a source in the nuts of the 
Parinarium laurinum. 
