In the following, as in AR, we discuss the dielectronic recombination rates by isosequence and we explicitly declare when we have chosen different data from them.

All the available data were fitted, when
needed, with the following formula:

(7) |

where *T* and *E*_{i} are given in eV and *c*_{i} in .The coefficients *c*_{i} and *E*_{i} are given in Table 1.

*H-like - *
From C to Ni we used the data of Mattioli 1988(hereafter M) and Karim & Bhalla (1988)which for the same
elements are in good agreement.
For the elements not explicitly calculated we interpolated the DR rates along
the isosequence.
For Be IV and B V we used the data of Pindzola & Badnell (1992) that are
a factor 2 less then those from previous calculations of Shore (1969)
with the corrected values by Burgess & Tworkowski (1976).
Moreover we divided by a factor 2 also the DR coefficients for HeII and Li III
by Shore (1969).

*He-like -*
We fitted with our formula the calculations of
Chen (1986a)
that are in very good agreement with Nielsen (1986) and Karim & Bhalla
(1989).
For C V we adopt Chen's (1988b) calculation in which the Coster-Kroning
transitions are taken into account that are negligible for higher Z
elements. For BeIII and B IV we adopted Pindzola & Badnell (1992).

*Li-like -*
We adopt the DR calculations of Chen (1991). The total DR rate
coefficient for the 11 lithium-like ions were calculated using the distorted
wave technique and the multi-configuration Dirac-Fock method. The
discrepancies with non-relativistic calculations of Roszman (1987)
are more relevant for highly
charged medium- and high-Z ions.
For the other
ions we have interpolated
along the isosequence. For BeII and B III we have adopted
the DR calculations of Pindzola & Badnell (1992).

*Be-like -*
We adopt Badnell's (1987) calculations
which for high Z agree well with Chen & Crasemann (1988). For O V and B II
we use, respectively, the data of Badnell & Pindzola (1989) and
Pindzola & Badnell (1992).

*B-like -*
Recent calculations of Nahar (1995) and errata from Nahar (1996a),
in the framework of a unified treatment of
electron-ion recombination, show good agreement with low-T DR rates by
Nussbaumer & Storey (1983), and high-T DR rates by Jacobs et al. (1979)
for the ions from C to Al, and for Si and S.
As for Fe, we use the data of AR.
For most of the other ions calculations from SV, LM and M are available.
For the other ions
we used the Burgess (1965) general formula, corrected by Merts et al.
(1976) (hereafter Burgess-Merts formula, BM), using the
most recent data for oscillator strengths and energy transitions.

Iron is the most investigated element so that in the following we always adopt the criteria, except when more recent calculations are available, to normalize the DR calculation rates to the Fe calculations of AR. These criteria can be also justified by the fact that, as pointed out in LM (1991) and Hahn (1991), most of the coefficients show a regular trend along the same isosequence.

*C-like -* *N-like -*
From Ar to Ni we followed the same procedure described for
B-like ions. For the other elements, except for O III for which we used the
data from Badnell & Pindzola (1989), we used the data of Jacobs et al.
(1977a,b, 1979, 1980)
and LM (1991).

*O-like -* *F-like -*
From Ar to Ni we adopt the recent calculations of
Dasgupta & Whitney (1990, 1994)
that for F-like
ions are in good agreement with Chen (1988a)
calculations. For the other elements
we used the calculations of SV.

*Ne-like -*
We use the calculations of Chen (1986b), Romanik (1988)and for Na II the
calculation of SV. For P VI and Cl VIII we adopted the data of LM(1991)
multiplied by a factor 4 to take into to account the results of
Romanik (1988) and Chen (1986b) for adjacient ions.

*Na-like -* *Mg-like -*
As it has been indicated by Zhdanov (1982), Dube & LaGatutta (1987), Dube
et al. (1985),
at high temperature the inner shell transitions of
the form become important, and this effect is more important for
medium and high-Z elements. We follow the procedure of Mattioli (1988)
that for Fe
is in very good agreement with AR. For the other elements we adopt the
calculations of Jacobs and LM taking into account the results obtained for
Mg II and Si IV from LaGattuta & Hahn (1982, 1984). For S VI we adopt
the data of Badnell (1991).

*Al-like -* to *Co-like -*
For all these isosequences, except for Fe, detailed or recent
calculations are
not available so that we used data from SV, LM,
M as well as the general BM formula. Up to the Mn-like isosequence
we renormalize the results to the work of AR.

We mention that recently new calculations have been made on the recombination of Fe II (Nahar et al. 1997) and Fe IV (Nahar 1996b). In the case of Fe II the DR rate is a factor 2 above the AR curve whereas for Fe IV the Nahar results are an order of magnitude below the AR curve although Nahar claims a accuracy. We do not know the cause of the discrepancies but for consistency we prefer to use the AR results. In the last years Teng et al. (1994a-d, 1996) also applied fitting formula for several isoelectronic sequence i.e. H, He, Li, F and Ne. In general our results are within a 10% in agreement with those calculations essentially because we have fitted and/or interpolated the same data, but in some case their fitting formula yield divergences in the dielectronic rates.

Copyright The European Southern Observatory (ESO)