The metallicity of the alpha Centauri system, Z, suffers from uncertai
nities. For this reason, different methods are used to calibrate the s
ystem: calibrations performed in YALE (Edmonds et al. 1992) use a fixe
d value for Z: Z = 0.026 and a convection parameter for each star, whi
le those made in Meudon and Liege (Noels et al. 1991; Neuforge 1993a)
make the hypothesis of a unique convection parameter for the two compo
nents of the system and consider Z as a free parameter. We discuss the
se two techniques, both using models calculated with mixing length con
vection theory, (MLT), and we explain our solution through the behavio
ur of the convection parameter with chemical composition. We also comp
are our results with those of Lydon (1993) and find consistency. With
a precise observational value of Z, of the effective temperature and o
f the luminosities, our results provide a test for the unicity of alph
a, if, in the frame of the same physics, a precise atmosphere treatmen
t can be used and low-temperature opacities are known with sufficient
accuracy. Finally, we perform calibrations with models calculated with
the convection treatment of Canuto & Mazzitelli (1991, 1992), where w
e use LAMBDA = z, z being the distance to the top of the convective en
velope. We avoid thus problems raised by the MLT convection parameter.
In this frame, satisfactory solutions can be found for 0.024 less-tha
n-or-equal-to Z less-than-or-equal-to 0.040.