Except in belts, girths, and perhaps occasionally in very narrow blankets, the shuttle is never passed through the whole width of the warp at once, but only through a space which does not exceed the length of the batten; for it is by means of the batten, which is rarely more than 3 feet long, that the shed is opened.

[Illustration:  PL.  XXXV.—­WEAVING OF DIAMOND-SHAPED DIAGONALS.]

Suppose the woman begins by weaving in the lower shed.  She draws apportion of the healds towards her, and with them the anterior threads of the shed; by this motion she opens the shed about 1 inch, which is not sufficient for the easy passage of the woof.  She inserts her batten edgewise into this opening and then turns it half around on its long axis, so that its broad surfaces lie horizontally; in this way the shed is opened to the extent of the width of the batten—­about 3 inches; next the weft is passed through.  In fig. 42 the batten is shown lying edgewise (its broad surfaces vertical), as it appears when just inserted into the shed, and the weft, which has been passed through only a portion of the shed, is seen hanging out with its end on the ground.  In Plate XXXV the batten is shown in the second position described, with the shed open to the fullest extent necessary, and the weaver is represented in the act of passing the shuttle through.  When the weft is in, it is shoved down into its proper position by means of the reed-fork, and then the batten, restored to its first position (edgewise), is brought down with firm blows on the weft.  It is by the vigorous use of the batten that the Navajo serapes are rendered water-proof.  In Plate XXXVIII the weaver is seen bringing down this instrument “in the manner and for the purpose described,” as the letters patent say.

When the lower shed has received its thread of weft the weaver opens the upper shed.  This is done by releasing the healds and shoving the shed-rod down until it comes in contact with the healds; this opens the upper shed down to the web.  Then the weft is inserted and the batten and reed-fork used as before.  Thus she goes on with each shed alternately until the web is finished.

It is, of course, desirable, at least in handsome blankets of intricate pattern, to have both ends uniform even if the figure be a little faulty in the center.  To accomplish this some of the best weavers depend on a careful estimate of the length of each figure before they begin, and weave continuously in one direction; but the majority weave a little portion of the upper end before they finish the middle.  Sometimes this is done by weaving from above downwards; at other times it is done by turning the loom upside down and working from below upwards in the ordinary manner.  In Fig. 49, which represents one of the very finest results of Navajo work, by the best weaver in the tribe, it will be seen that exact uniformity in the ends has not been attained.  The figure was of such a nature that the blanket had to be woven in one direction only.