I agree with your interpretation.  I think the second paragraph of that section describes what the configuration should be before testing, and the third paragraph describes what it should be after testing.

Hg

Stacey D

I agree with your interpretation.   The lab should have had a knurlerized plunger.  However, this movement from center happens often, even if you do have good bending equipment.   Sometimes the problem happens because the lab doesn't etch the specimen to determine where the center of the weld is.  I have a Fischer bend tester, and it has roller sides and a knurlerized plunger.  Still, when I get a specimen where plates having two different tensile strengths are used, the specimen sometimes likes to move rather than stay centered and bend where I placed it. 

There is also the problem of bending unevenly rather than around a given radius.  I usually walk away if one side doesn't bend in the weld / HAZ area and the second side bends very tightly.  To me, the requirement that the convex side be stretched to a certain percentage (found in the ASTM "Percent of Elongation" specification) is the governing requirement.  If I do not see degree of stretch that on the straighter side, I say that the test is invalid, and refuse to sign the pieces off.   This is one of those times when you need a wrap around bending fixture.  However, I had an A572 Gr. 50 test plate welded to a piece of AISI 4140.  When it was bent in the wrap around machine, it began to bend on the 4140 side, then sprung back almost straight when the weld cleared the rolling mandrel, leaving the specimen bent mostly on the A572 side.   I spoke the Engineer / Customer, and he said  to do a longitudinal bend, out of the balance of the test coupon, (which to my mind didn't show anything at all.)

Most of the time I've found the problem when I have had to use the customer's shop made bending fixture.  Nothing can be done but cut another coupon out of the balance of the coupon, or have the welder retake the test.