that still doesn't make sense to me.
If you want to get an idea how much an average tomato weighs, put 100 tomatoes on a scale, then divide by 100.
Yes, if you know that the data coming off is at full speed (no pauses) then measuring a full char or more chars will give you more accuracy.
You could send a MB of data and time it with a stop watch.
At 1200-8-n-1 it should take 2 hours, 25 minutes, 38 seconds.
Or I could set up a gated timer, send one SYN character, measure the time between the first two transitions to the nearest 21 nanoseconds (PIC processor with 16MHz clock and x3 PLL). At 1200 baud it takes a few milliseconds and gets the figure to better than 0.01%. Normally that's enough to settle on a standard baud rate (1200,2400,4800,9600,etc) but if the rate is truly unknown then send a few SYN characters and get the average. Measuring 2 hours, 25 minutes, 38 seconds to the nearest second is less accurate.
Like most things there are many ways to solve a problem. Your way works. My way works. I didn't think your way made sense but now I see it is a way of solving the problem. Thanks for the clarification.