ADDITIONS AND ORNAMENT. 131 
3 teeth, two of 4, one of 3, the remainder of 4. In figure 6 the selu 
are presented in three opposing pairs separated by a band—one of 5 
teeth, four of 6, and one of 7. 
In connection with figure 4 we have just suggested the possibility 
that the triangle might stand for an imperfect stage of the carving of 
the zigzag. The same holds true of the serration unit; it might repre- 
sent more or less of a zigzag unit in which the angles of one face had 
been carved and the designer had not yet begun to apply his shark- 
tooth burin to the opposite face. Yet there is at least equal proba- 
bility that this unit of design came independently into existence, for the 
carver need but look at the edges of the shark-tooth with which he is 
working and he will find a motive in nature. The units in which we 
find serration consist of teeth, always angular, arising from a base 
whose bottom line is cleanly rectilinear. This angularity is constant; 
if it were an incomplete sennit motive we should look to find the 
tendency toward smooth curves which is clearly apparent in figures 
1 to 4. The serration units occur in opposite-facing pairs separated 
by a bar in figure 6 and in the two wonderfully beautiful figures 33 
and 34. In figure 82 is found a solitary instance of this opposition 
outward in which the septum bar is lacking. Without septum and 
facing in the same direction serration in pair or series is found in 
figures 43, 81, 88, and 89, and appearing in the single unit in figures 
15, 45, 81, 82, and 83. ‘This decoration appears in the field with birds 
in flight in 81, 82, 83, 88, and 89; we may be justified in this connection 
in looking upon it as a sky sign, a cloud derivative. In figure 104, 
diagonally approximated to the conventional feet of a man, occur 
two units of a design very close to the serration; in the upper there are 
6 small rectangular figures dependent from a rectilinear bar, in the 
lower 3 such figures. If this be not a degradation form of serration 
the motive is by no means apparent; the beam-and-billet motive is 
contraindicated, for the reason that in the abundantly thatched archi- 
tecture of Nuclear Polynesia beams and raiters are structural details 
which never appear in plain sight. 
The next unit of design which arises for examination is the lozenge, 
including therewith a few figures obtainable by the same method but 
varying in shape from the roughly quadrangular to irregular polygonal 
forms. At the right of figure 1, in which we find approximated zigzags, 
if we reckon up from the bottom we are able to discern the germ of 5 
lozenges by reason of the fact that in beginning the carving the artist 
had started with his lines in opposite directions and for some little 
space was able to maintain this opposition before being conquered by 
the tendency toward uniformity which has in the end resulted in paral- 
lelism in place of antagonism of the lines. The same is true in figure 2, 
where at least 4 reasonably good lozenges occur at the right of the 
design. In figures 25 and 26 we shall have to recognize the triangle 
