Models for metal-poor stars with different initial abundances of C, N, O, Mg, and Si. II. Application to the colour-magnitude diagrams of the globular clusters 47 Tuc, NGC 6362, M 5, M 3, M 55, and M92

Stellar models for -2.5 ≤ [Fe/H] ≤-0.5 that have been computed for variations in the C:N:O abundance ratio (for two different values of [CNO/Fe]) are compared with HST Wide Field Camera 3 (WFC3) observations of the globular clusters (GCs) 47 Tuc, NGC 6362, M 5 M 3, M 55, and M 92. The bolometric corrections (BCs) used to transpose the models to the observed planes are based on new MARCS synthetic spectra that incorporate improved treatments of molecules that involve atoms of C, N, and O. On the assumption of well-supported distance moduli and reddenings, isochrones for [O/Fe] = 0.6 and [m/Fe] = 0.4 for the other α elements, which are favoured by binary stars in GCs, generally reproduce the main features of observed colour-magnitude diagrams (CMDs) to within ∼0.03 mag. In particular, they appear to match the spreads in the observed (MF336W - MF438W)0 colours that are spanned by CN-weak and CN-strong stars along the lower giant branch quite well, but not the bluest giants, which are suspected to be N-poor ([N/Fe] -0.5). Both the absolute (MF438W - MF606W)0 colours and the variations in these colours at a given MF606W magnitude on the giant branch are difficult to explain unless the reddest stars are C-rich ([C/Fe] +0.5). Allowing for moderate He abundance variations ($delta , Y sim 0.05$) improves the fits to the observations.