Abstract
The steady advance in the sophistication of large scale scientific computing has led in the past few years to the need for h~gh quahty mathematical software collected into the form of standardized computer libraries. This paper reports the experiences of a relatively infant, private project to construct such a standard mathematical subroutine library. The coverage is not intended to be either a comprehensive or general dmcusslon of the construction of mathematical software librames, but rather to share some of the considerations and certffication results of a specific library project.