In this paper we have presented new broad-band infrared colors and CO indices for early-type galaxies and globular clusters. We have compared our data with recent stellar synthesis models of bright ellipticals and can conclude that the only models which are consistent with the present observations are those with rich giant branches and relatively flat main-sequence luminosity functions. A detailed quantitative comparison is made difficult, however, because of the sensitivity of the predicted infrared colors to the numbers and luminosities of M giants, combined with the observational uncertainties in these latter quantities.
We have examined the dependence of the various colors on absolute galaxian luminosity and on measuring aperture size. The broad-band colors tend to redden with increasing luminosity and with decreasing aperture size. Since we find evidence for the infrared colors and the CO index to be redder and stronger, respectively, in globular clusters of high mean metallicity as opposed to those of low metallicity, it is reasonable to conclude that the changes observed in the integrated light of galaxies also have, as their underlying cause, changes in the mean metallicity of the stellar population. A change in the metallicity acts both directly via a change in the CO strength for stars of the same effective temperature and indirectly via a change in the relative populations of the principal sequences of a color-magnitude diagram. Of particular note is the evidence that the relative changes of the U - V, V - J, and J - K colors within galaxies differ from those between galaxies. It may be possible to understand this result qualitatively if some fraction of the changes in the optical parameters arises from changes in CN blanketing rather than from changes in [Fe/H] alone (Peterson 1976a, b). This may, mean that U - V cannot be used as a pure indicator of the mean metallicity of a composite system. The color changes in systems consisting of metal-rich and metal-poor stars may also be important in understanding the interplay of U - V and V - K. Stellar synthesis models of a complexity greater than any heretofore reported in the literature, and which include a certain amount of chemical evolution, will be required to quantitatively interpret these and other recent photometric observations of galaxies.
This program would not have been possible without the generous allotment of observing time made available to us, particularly at the Hale Observatories, or without the collaboration of Eric Becklin and Gerry Neugebauer. We are grateful to Allan Sandage for allowing us to have access to his compilation of published optical photometry and for his continued interest in this work. We acknowledge helpful conversations with Beatrice Tinsley and Robert O'Connell. Special assistance was provided by d. Walker and G. Livet. An anonymous referee made several critical comments on an earlier version of this paper. This work was supported in part by NSF grant AST 74-18555 A2 and NASA grant NGL 05-002-207.