Using the code autostructure , extensive calculations of inner-shell atomic data have been made for the chemical elements He , C , N , O , Ne , Na , Mg , Al , Si , S , Ar , Ca , Cr , Mn , Fe and Ni .
The results are used to obtain up-dated opacities from the Opacity Project , OP .
A number of other improvements on earlier work have also been included .

Rosseland-mean opacities from OP are compared with those from OPAL .
Differences of 5 to 10 % occur .
OP gives the ‘ Z -bump ’ , at \log ( T ) \simeq 5.2 , to be shifted to slightly higher temperatures .
The opacities from OP , as functions of temperature and density , are smoother than those from OPAL .

The accuracy of the integrations used to obtain mean opacities can depend on the frequency-mesh used .
Tests involving variation of the numbers of frequency points show that for typical chemical mixtures the OP integrations are numerically correct to within 0.1 % .

The accuracy of the interpolations used to obtain mean opacities for any required values of temperature and density depend on the temperature–density meshes used .
Extensive tests show that , for all cases of practical interest , the OP interpolations give results correct to better than 1 % .

Prior to a number of recent investigations which have indicated a need for downward revisions in the solar abundances of oxygen and other elements , there was good agreement between properties of the sun deduced from helioseismology and from stellar evolution models calculated using OPAL opacities .
The revisions destroy that agreement .
In a recent paper Bahcall et al .
argue that the agreement would be restored if opacities for the regions of the sun with 2 \times 10 ^ { 6 } \lesssim T \lesssim 5 \times 10 ^ { 6 } K ( 0.7 to 0.4 R _ { \sun } ) were larger than those given by OPAL by about 10 % .
In the region concerned , the present results from OP do not differ from those of OPAL by more than 2.5 % .