Please read the g77 manual and its references about floating point and IEEE arithmetic conformance. It even had physicist input -- computational scientists need to understand the issues. Please don't follow up to all these lists. I doubt debian-sparc, where I'm reading it, cares.