We present estimates for the iron content of the stellar and diffused components of elliptical galaxies, as derived respectively from integrated optical spectra and from ASCA X-ray observations. A macroscopic discrepancy emerges between the expected iron abundances in the hot interstellar medium (ISM) and what is indicated by the X-ray observations, especially when allowance is made for the current iron enrichment by Type Ia supernovae. This strong discrepancy, that in some extreme instances may be as large as a factor of similar to 20, calls into question our current understanding of supernova enrichment and chemical evolution of galaxies. We discuss several astrophysical implications of the inferred low iron abundances in the ISM, including the chemical evolution of galaxies and clusters of galaxies, the evolution of gas flows in elliptical galaxies, and the heating of the intracluster medium. Some of the consequences appear hard to accept, and in the attempt to avoid some of the difficulties we explore ways of hiding or diluting iron in the ISM of ellipticals. None of these possibilities appears astrophysically plausible, and we alternatively raise the question of the reliability of iron L line diagnostic tools that are currently used to infer abundances from X-ray spectra. Various thin-plasma emission models are shown to give iron abundances that may differ significantly, especially at low temperatures (kT less than or similar to 1 keV), when the iron L complex is dominated by iron ions with still many bound electrons. From a collection of ASCA and other X-ray observatory. data, it is shown that current thin-plasma codes tend to give very low iron abundances when the temperature of the objects is below similar to 1 keV. Such objects include various types of binary stars, supernova remnants, starburst galaxies, and AGNs, with the case of galaxy groups being especially well documented. We conclude that, besides rethinking the chemical evolution of galaxies, one should also consider the possibility that existing thin-plasma models may incorporate inaccurate atomic physics for the ions responsible for the iron L complex.
