Oscillator strengths for fine-structure transitions in F-like Na III are ca
lculated for the first time ab initio using the Breit-Pauli opacity project
(OP) R-matrix method. Many transitions are in excellent agreement with a r
ecent CIV3 calculation by McPeake and Hibbert (2000 J. Phys. B: At. Mel. Op
t. Phys. 33 2809), for example our gf-values for 2s(2)2p(5) P-2(o)-P-2(o)-2
s2p(6) S-2 Of 0.357 (J = 3/2-1/2) and 0.177 (J = 1/2-1/2) are within 2% of
CIV3. However, a problematic case is the 2s2p(6) S-2-2s(2)2p(4)(S-1)3p P-2(
o) multiplet: our gf-value is 0.023, but the CIV3 value is 50% greater. No
apparent reason can be found for this discrepancy, although it is noted tha
t different CIV3 runs give different results, whereas the present gf agrees
with the original OP one-which actually agrees with the value we obtain fr
om repeating the CIV3 work. Transitions involving the 2p(4)3d levels tend t
o show bigger disagreements between CIV3 and our gf-values than lower-lying
transitions. The CIV3 work is limited to n > 3 spectroscopic states, and t
he incomplete modelling of the 3d <----> 4s mixing could be a factor in som
e of these discrepancies. A specific case highlighted is the doublet transi
tions of 2p(4)(P-3)3p-2p(4) (P-3)3d. Further work is urged by the atomic st
ructure community to help shed light on these problems.