French term (edited): fl�che au m�tre

Although David's of course right in all normal senses of 'flèche', I feel that here they are more likely talking about the various ways in which the product itself may be 'out of true' in a linear sense.

The three things they are describing are, I believe:

curvature (overall)

deviation from straightness (any cause) over any one measured metre length

and waviness

The point about specifying 'flèche par mètre' is that it gives a better correlation with the significance of the error; it's like specifying a floor as being 'flat within 1 mm per metre' = no more than 1 mm deviation when moving 1 m from a given point in any direction.

So you might have a length of steel that has, say, an overall tendency to curve, but over any single metre length measured, the deviation from straight might be insignificantly small. Conversely, you might have a length that appears straight as a die from one end to the other, but is as bent as a fiddler's elbow in between.

Unfortunately, although I believe I am right about the parameters being described, I fear I am far less confident about these actual terms I've used to describe them. But I hope at least this explanation may help get you on the right track, and that someone more knowledgeable will check in with the authoritative answer on the actual terms themselves