round upon themselves, inclosing
the woof threads and holding them quite firmly, so that the open mesh
is well preserved even when much strained. Fabrics of this character
have been employed by the ancient potters of a very extended region,
including nearly all the Atlantic States. There are also many varieties
of this form, of fabric resulting from differences in the size and
spacing of the threads. These differences are well brought out in the
series of illustrations that follow.

In regard to the manufacture of this particular fabric, I am unable to
arrive at any very definite conclusion. As demonstrated by Miss Osgood,
it may be knitted by hand, the threads of the warp being fixed at one
end and the woof at both by wrapping about pegs set in a drawing board
or frame, as shown in the diagram, Fig. 74.

[Illustration: Fig. 74.--Diagram showing the method of weaving
Form 2.]

The combination is extremely difficult to produce by mechanical means,
and must have been beyond the reach of any primitive loom. I have
prepared a diagram, Fig. 75, which, shows very clearly the arrangement
of threads, and illustrates a possible method of supporting the warp
while the woof is carried across. As each thread of the woof is laid in
place, the threads of the warp can be thrown to the opposite support,
a turn or half twist being made at each exchange. The work could be done
equally well by beginning at the top and working downward. For the sake
of clearness I have drawn but one pair of the warp threads.

[Illustration: Fig. 75.--Theoretic device for working the twist.]

Fig. 76 illustrates a characteristic example of this class obtained from
a fragment of pottery from the great mound at Sevierville, Tenn.

[Illustration: Fig. 76.--From fragment of mound pottery, Tennessee.]

The impression is quite perfect. The cords are somewhat uneven, and seem
to have been only moderately well twisted. They were probably made of
some vegetable fiber. It will be observed that the threads of the woof
are placed at regular intervals, while those of the web are irregularly
placed. It is interesting to notice that in one case the warp has not
been doubled, the single thread having, as a consequence, exactly the
same relation to the opposing series as corresponding threads in the
first form of fabric presented. The impression, of which this is only a
part, indicates that the cloth was considerably distorted when applied
to the soft clay. T