Turns out I was wrong, as usual. The article compares
grinding the notch to broaching. also, the variation is slight, and probably not statstically significant.
https://books.google.com/books?id=Pqcesk8iZzMC&pg=PA92&lpg=PA92&dq=charpy,+grinding,+broaching&source=bl&ots=AMTL40Vdis&sig=A4T6D9lBNmfviZ6C-bZDxTEkUDQ&hl=en&sa=X&ei=loWtVMauMIr7yATurIDAAg&ved=0CEgQ6AEwAw#v=onepage&q=charpy%2C%20grinding%2C%20broaching&f=falseAnd some infor from NIST
5.3.2 Machining requirements
The dimensional requirements for NIST verification
specimens, given in Table 6, meet or exceed the
ASTM E 23 specifications. This minimizes variations
in impact energy due to physical variations in the
specimens. Also, the notch centering and the length
tolerance for NIST specimens are equivalent to the ISO Standard 164, which permits the specimens
to be used in impact machines with end-centering devices. The NIST requirement for
surface finish is also equivalent to the ISO 164 requirement. All of these dimensional requirements
can be met with standard machining practices.
Specimen notches are form ground on a surface grinder
(machining with a fly cutter or multitooth
cutter is not permitted). To avoid "burning" or cold working the material at the base of the
notch, the next to the last cut is required to remove more than 0.25 mm and less than 0.38 mm
and the final cut must not remove more than 0.12 mm. When the specimens are finished and
ready for shipment, they are given a protective coating of oil.