Transformation matrix in header of new-fits file

[Anisha Vadher]

I have a question about the new-fits file for an image of a cluster I have taken.

In order to transform from RA/DEC of a star in my cluster to pixel coordinates of an image I have used the transformation matrix and the off-sets in the header of the new fits file. However in order to get the correct coordinates in pixels I have to multiply the horizontal elements of the matrix by a factor of 1.85 (using trial and error). This factor changes for different images that I have taken. Could I have some insight into why this is and where this factor comes from?

I couldn't anything about it in the documentation.

Thanks

[Dustin Lang]

Could this be the cosine(Dec) term?  The CD matrix elements are in "isotropic" degrees; to convert one degree of isotropic degrees in the RA direction to degrees in RA, you need to divide by cosine(Dec).

cheers,
--dustin