Finally some time to get back to this...
PE makes a good point about the amount of gold or silver being based on the average diameter of the halide crystal.
The Neblette's book I referenced above discusses how the amount of Au and S are related to the crystal habit as well. It says that for AgBr crystals, cubic crystals are more sensitive to the amount of gold, while the octahedral crystals are more sensitive to sulfur concentration.
It also says that octahedral emulsions gain about 0.7 log E in speed for sulfur-only but only about 0.4 log E for gold only. The reverse was true for cubes - they gain about 0.9 log E in speed for gold-only and about 0.4 log E for sulfur only.
The graphs they reproduce also show this effect - for the AgBr octahedral, the optimum Au/S amount was about 1.6 mg/mole Ag sulfur and the speed kept increasing as the concentration of gold was increased. The graph shows an optimum ratio at about 2.0 mg gold/mole Ag and 1.6 mg sulfur/mole Ag. However, as was just mentioned, the speed does keep increasing as the gold concentration increases as it goes off the graph at 2.25 mg Au/mole Ag.
For AgBr cubes, there is a maximum speed gain for the ration of gold concentrations from 1.0 to 2.0 mg Au/mole Ag and 1.4 to 3.0 mg sulfur/mole Ag.
So I guess you could use the chart PE shows to figure out which crystal habit you are making to determine what ratio of Au and S to use.
Also, PE - with all your talk of pAg lately, perhaps you should write your handle as "pE"...