Given the importance of the temperature at the base of the corona for
the modeling of the solar wind, we investigate the range of temperatur
es which have been deduced from remote measurements in coronal holes,
within a heliocentric distance of 1.6 R(s), and the accuracy to which
these temperatures have been inferred. We also present results of an a
nalysis of extreme ultraviolet (EUV) observations made simultaneously
in three spectral lines at the limb in a polar coronal hole, with very
little contamination from quiet region emission along the line of sig
ht. The values most commonly quoted for temperatures below 1.6 R(s) ex
tend from 7 x 10(5) K to 1.6 x 10(6) K. However, by carefully reviewin
g the assumptions that enter into these inferences, we find that, with
in 1.6 R(s), the temperature in a coronal hole is at or below 1.3 x 10
(6) K. Our own analysis yields a temperature range of 7.8-9.3 x 10(5)
K, between 1.02 to 1.07 R(s) for the coronal hole, and a distinctly di
fferent temperature range of 9.4 x 10(5)-1.2 x 10(6) K for the quiet r
egions bordering it. Inhomogeneities within the coronal hole itself co
ntribute to a 14% variation in its inferred temperature. Our results s
how that, from data currently available, temperatures in the inner cor
ona cannot be derived with an accuracy better than 20%, even when negl
ecting many sources of uncertainty such as instrument calibration, lin
e of sight effects, departure from ionization balance, and inaccuracie
s of the atomic data. We also show in this study that the elemental ab
undance, which is one of the parameters that influence the temperature
inference, can in turn be significantly constrained when intensity ra
tios from three spectral lines are used. We find that, within the wide
range of values quoted so far in the literature, the larger values of
the abundance of Olog (A(O)) = 8.96!, Nelog (A(Ne)) = 8.2!, and Mg
log (A(Mg)) = 7.65! relative to hydrogen when log (A(H)) = 12!, yield
a self-consistent temperature.