In this paper we first emphasize why it is important to know the successive
zonal harmonics of the Sun's figure with high accuracy: mainly fundamental
astrometry, helioseismology. planetary motions and relativistic effects. T
hen we briefly comment why the Sun appears oblate, going back to primitive
definitions in order to underline some discrepancies in theories and to emp
hasize again the relevant hypotheses. We propose a new theoretical approach
entirely based on an expansion in terms of Legendre's functions, including
the differential rotation of the Sun at the surface. This permits linking
the two first spherical harmonic coefficients (J(2) and J(4)) with the geom
etric parameters that can be measured on the Sun (equatorial and polar radi
i). We emphasize the difficulties in inferring gravitational oblateness fro
m visual measurements of the geometric oblateness, and more generally a dyn
amical flattening. Results are given for different observed rotational laws
. It is shown that the surface oblateness is surely upper bounded by 11 mil
liarcsecond. As a consequence of the observed surface and sub-surface diffe
rential rotation laws, we deduce a measure of the two first gravitational h
armonics, the quadrupole and the octopole moment of the Sun: J(2) = -(6.13/-2.52) x 10(-7) if all observed data are taken into account, and respectiv
ely, J(2) = -(6.84 +/- 3.75) x 10-7 if only sunspot data are considered, an
d J(2) = -(3.49 +/- 1.86) x 10(-7) in the case of helioseismic data alone.
The value deduced from all available data for the octopole is: J(4) = (2.8
+/- 2.1) x 10(-12). These values are compared to some others found in the l
iterature.