George,
I did not mean to start a controversy over this and apologize if it appears that way.
It is true that the baud rate generator mode will run at 1/2 the oscillator. It is also true that the generator can be clocked externally, however the data sheet is not explicit in stating the maximum external rate input to P1.0 in this mode. If you refer to Figure 5 you can see that the internal oscillator is divided by 2 when selected as the clock source as you have stated. If the external pin P1.0 is selected it drives TH2 and TL2 in the same manner without the divide by two. By looking at this figure you could conclude that any frequency could drive the timer but we know this is not true.
So, some mechanism must be used to determine when a transition has occurred on P1.0. In view of the lack of explicit detail of the external input pin, I defer to someone who may have used the device as you have described.
I can check this with a breadboard I have and if I have the time in the next couple of weeks I will try to verify the maximum external clock rate on P1.0 when T2 is configured as a baud rate generator.
Respectfully,
Allen