Previous X-ray and ultraviolet spectroscopy suggested that the Fe/O abundance ratio in NGC 1068 may be abnormally high .
We have tested this suggestion by measuring and modeling the ASCA spectrum of NGC 1068 .
We have measured some 15 X-ray lines , to an accuracy better than a factor 2 , and modeled the continuum in two different ways .
The first assumes that the hard X-ray continuum is the reflection of the nuclear source by two , extended , photoionized gas components ; a warm ( T \sim 1.5 \times 10 ^ { 5 } K ) gas and a hot ( T \sim 3 \times 10 ^ { 6 } K ) gas .
All the observed emission lines are produced in this gas and there is an additional , extended 0.6–3 keV pure continuum component .
The model is similar to the one proposed by Marshall et al . ( 1993 ) .
The second model is a combination of a hard reflected continuum with a soft thermal plasma component .
The calculations show that the emission lines in the photoionized gas model are in very good agreement with the observed ones assuming solar metallicity for all elements except for iron , which is more than twice solar , and oxygen , which is less than 0.25 solar .
Models with solar oxygen are possible if the 0.5–1 keV continuum is weaker but they do not explain the magnesium and silicon lines .
The thermal model fit requires extremely low metallicity ( 0.04 solar ) for all elements .
We discuss these findings and compare them with the ASCA spectra of recently observed starburst galaxies .
We argue that the apparent low metallicity of starburst galaxies , as well as of the extended nuclear source in NGC 1068 , are inconsistent with galaxy chemical evolution .
The explanation for this apparent anomaly is still unknown and may involve non-thermal continuum mechanisms and , in some cases , depletion onto grains .
Given the strong H-like and He-like lines , as well as the prominent Fe-L emission features , the origin of the soft X-ray lines in this source is more likely photoionized gas .
We compare our model with the recent Iwasawa et al . ( 1997 ) paper .
We also show that fluorescence lines of low-Z element lines in AGN is likely to be larger than previously assumed .