Tested in its worst direction
On this page11
A woven fabric does not burn the same way in both directions
Warp and weft differ in yarn, twist and density, and the standard treats that as a fact to be measured rather than averaged.
So the procedure starts by finding out which direction is worse
Conduct preliminary trials to determine the quickest burning direction. One specimen is cut from each direction of the fabric, 50 mm by 150 mm, two inches by six.
And then every real specimen comes from that direction
Prepare and test five specimens from the fastest burning direction. The classification is therefore a statement about the fabric at its worst, not about the fabric on average.
Which is a deliberate choice and it is worth naming as one
An average over both directions would produce a number that no part of the fabric actually exhibits. For a fire standard the useful figure is the one describing the direction a flame will in fact take.
The specimen size is fixed and it has a consequence
Fifty millimeters by a hundred and fifty, which is two inches by six. If a specimen that size cannot be cut, the item is not tested.
And the rule names the items it will not test
Narrow fabrics and loose fibrous materials manufactured less than 50 mm in width in either direction shall not be tested, and the examples given are hula skirts, leis, fringe, loose feathers, wigs and hairpieces.
That exclusion is by specimen geometry, not by risk
The rule is not saying a lei is safe. It is saying the test cannot be performed on one, which is a different statement and a more honest one than inventing a result.
The marking instruction is a small thing with a real reason
Identify the fabric direction being careful not to make any identifying marks in the exposed area to be tested. A pen mark is a substance on the specimen and a substance on the specimen is part of what burns.
And one special case cuts the specimen to hit the worst feature
A metallic thread fabric is tested as a plain surface fabric, with the specimens cut so that the metallic thread is parallel to the long dimension and arranged so the test flame impinges on a metallic thread.
So even inside one specimen the flame is aimed at the hazard
Which is the same principle as choosing the direction and the area. Four separate places in this part make the same decision, and the next section is about all of them together.