The Behrens distribution, or sometimes the Behrens-Fisher distribution as related to the Behrens-Fisher Problem, is frequently used for calculating the conservative joint confidence intervals and pvalues for the difference in means between two normal samples of unequal variance.
Code for R for calculating the Behrens distribution is available that makes use of the fact that the Behrens distribution is a linear combination of two t distributions and a contributed R package "tdist" by Viktor Witkovsky [homepage, see 1, 2 and 3 for references].
List of all files in the directory available here.
by Jacob Colvin, contact me at jbcolvin at fastmail dot fm