well, all this is to do with *binning*!
This what AMD and intel does while making the CPUs. If you think they manufacture batches of different processor speeds each, it isnt. Actually, every processor starts to be the fastest. There is a range of speeds a batch aims for, say 1 GHz. Now when the wafer is fabricated and each die is cut, it is tested for maximum gate switching speed, and allowing a safety margin of 10~30% they are stamped accordingly and sent out. For example, a chip found running at 1200 MHz is labelled as 1 GHz, one running at 1000 is labelled as 800. Sometimes even a 1 GHz capable die is labelled to a much lower value, to maintain the ASP (Average Selling Price) of the processor. if you get one of this you are lucky, but still they gurantee it would run at the rated speed, at what highest speed it would run is not known until one actually tries it. And it does depend on each piece, even if it is from the same batch and stepping!
You can get these speed ranges from the stepping charts publiched by the manufacturers. What you can derive from those is that there is certain range of speeds for the same stepping. That means the slowest processor in that range has a potential to run at the highest speed! Although it might not reach that high levels, it would certainly run much faster. Faster processor have less and less overclockability because they are already approaching the max limit of that range.
Your Pentium-II 350 happenes to be the slowest in the breed of 0.25 micron processor, capable of running at much higher speeds. Duron 950 is already close to 1 GHz limit around which its speeds stayed for a long time before a 1.1G Duron came around. Probabely thats the reason why the Duron dint overclock to your expectations.
girish
<font color=red>Nothing is fool-proof. Fools are Ingenious!</font color=red>