I sure there is a simple explaination for this, but it just hasn't dawned on me yet.
I am running an impurities assay, a long(35min) normal phase gradient with hexane and THF. The assay works very well, except for the fact that I must make at least two injections before the retention times settle in and become reproducible. Doesn't matter how long I equilibrate at the begining. The column seems to need those two injections. I've played with the gradient times at the scarifice of my resolution to see if the lenght of the gradient was an issue. Nope, still need two injections and then it settles down to run. Anybody got an idea why?

See my comment "priming ...." from Feb. 25, 2000, especially look at the article on which I commented.