I don't know for sure what to cross to. Maybe an actual partridge chicken? Since e^b is supposed to be recessive to everything else you are working with.
I am not sure the Penciling does require e^b (partridge).
I'm going by this:
https://kippenjungle.nl/kruising.html
I can make a "penciled" bird by adding Pg/Pg to e+/e+, or to e^b/e^b, or to e+/e^b.
If you know the exact parents of each chick, you may be able to sort it out by making a list of all the matings and what e-locus gene got passed on each time. It might be faster than doing test matings and trying to interpret the results. It sounds to me like the setup for a big logic puzzle:
Each chicken has exactly two alleles at the e locus, one from each parent, and can only pass one to each chick.
The Barred Rocks and the Black Australorps "should" have been pure for E/E.
The Buff Orpingtons "should" have been pure for E^Wh/E^Wh.
Db is called Dark Brown because of the effect on chick down (on the adult feathers it acts as a Columbian-like restrictor.)
Of course your original chickens might have had other genes than what is "supposed" to be there, but a careful check of the whole lineage might help figure it out.
I'm not sure about pictures, but I find some text descriptions here:
http://www.edelras.nl/chickengenetics/mutations1.html#gen_mut_eumelrest
I see descriptions of Db on various e alleles, and some paragraphs further down I find a discussion of "undercolor" that says:
"Wheaten has a cream feather undercolor. e+ and eb have a gray feather undercolor even in the presence of Co."
(There are more details, but I only quoted a short bit.)
So you could look at the undercolor on your chickens to try to figure out the e locus situation.